Beruflich Dokumente
Kultur Dokumente
By
Tharmarajah ANAPAYAN
March 2010
Keywords
KEYWORDS
LiteSteel beams, Hollow flange beams, Hollow flange steel beams, Lateral
distortional buckling, Lateral buckling tests, Section moment capacity tests, inelastic
reserve bending moments, Finite element analyses (FEA), Cold-formed steel
structures, Flexural members, Web stiffeners, Transverse web stiffeners.
iii
Keywords
iv
Abstract
ABSTRACT
The LiteSteel Beam (LSB) is a new hollow flange channel section developed by
OneSteel Australian Tube Mills using a patented Dual Electric Resistance Welding
technique. The LSB has a unique geometry consisting of torsionally rigid rectangular
hollow flanges and a relatively slender web. It is commonly used as rafters, floor
joists and bearers and roof beams in residential, industrial and commercial buildings.
It is on average 40% lighter than traditional hot-rolled steel beams of equivalent
performance. The LSB flexural members are subjected to a relatively new Lateral
Distortional Buckling mode, which reduces the member moment capacity. Unlike the
commonly observed lateral torsional buckling of steel beams, lateral distortional
buckling of LSBs is characterised by simultaneous lateral deflection, twist and web
distortion.
Current member moment capacity design rules for lateral distortional buckling in
AS/NZS 4600 (SA, 2005) do not include the effect of section geometry of hollow
flange beams although its effect is considered to be important. Therefore detailed
experimental and finite element analyses (FEA) were carried out to investigate the
lateral distortional buckling behaviour of LSBs including the effect of section
geometry. The results showed that the current design rules in AS/NZS 4600 (SA,
2005) are over-conservative in the inelastic lateral buckling region. New improved
design rules were therefore developed for LSBs based on both FEA and experimental
results. A geometrical parameter (K) defined as the ratio of the flange torsional
rigidity to the major axis flexural rigidity of the web (GJf/EIxweb) was identified as
the critical parameter affecting the lateral distortional buckling of hollow flange
beams. The effect of section geometry was then included in the new design rules
using the new parameter (K). The new design rule developed by including this
parameter was found to be accurate in calculating the member moment capacities of
not only LSBs, but also other types of hollow flange steel beams such as Hollow
Flange Beams (HFBs), Monosymmetric Hollow Flange Beams (MHFBs) and
Rectangular Hollow Flange Beams (RHFBs).
The inelastic reserve bending capacity of LSBs has not been investigated yet
although the section moment capacity tests of LSBs in the past revealed that inelastic
reserve bending capacity is present in LSBs. However, the Australian and American
v
Abstract
cold-formed steel design codes limit them to the first yield moment. Therefore both
experimental and FEA were carried out to investigate the section moment capacity
behaviour of LSBs. A comparison of the section moment capacity results from FEA,
experiments and current cold-formed steel design codes showed that compact and
non-compact LSB sections classified based on AS 4100 (SA, 1998) have some
inelastic reserve capacity while slender LSBs do not have any inelastic reserve
capacity beyond their first yield moment. It was found that Shifferaw and Schafers
(2008) proposed equations and Eurocode 3 Part 1.3 (ECS, 2006) design equations
can be used to include the inelastic bending capacities of compact and non-compact
LSBs in design. As a simple design approach, the section moment capacity of
compact LSB sections can be taken as 1.10 times their first yield moment while it is
the first yield moment for non-compact sections. For slender LSB sections, current
cold-formed steel codes can be used to predict their section moment capacities.
It was believed that the use of transverse web stiffeners could improve the lateral
distortional buckling moment capacities of LSBs. However, currently there are no
design equations to predict the elastic lateral distortional buckling and member
moment capacities of LSBs with web stiffeners under uniform moment conditions.
Therefore, a detailed study was conducted using FEA to simulate both experimental
and ideal conditions of LSB flexural members. It was shown that the use of 3 to 5
mm steel plate stiffeners welded or screwed to the inner faces of the top and bottom
flanges of LSBs at third span points and supports provided an optimum web stiffener
arrangement. Suitable design rules were developed to calculate the improved elastic
buckling and ultimate moment capacities of LSBs with these optimum web
stiffeners. A design rule using the geometrical parameter K was also developed to
improve the accuracy of ultimate moment capacity predictions.
This thesis presents the details and results of the experimental and numerical studies
of the section and member moment capacities of LSBs conducted in this research. It
includes the recommendations made regarding the accuracy of current design rules as
well as the new design rules for lateral distortional buckling. The new design rules
include the effects of section geometry of hollow flange steel beams. This thesis also
developed a method of using web stiffeners to reduce the lateral distortional buckling
effects, and associated design rules to calculate the improved moment capacities.
vi
Publications
PUBLICATIONS
vii
Publications
viii
Table of Contents
TABLE OF CONTENTS
Keywords ...iii
Abstract ...v
Publications vii
Table of Contents ix
List of Figures xv
List of Tables xxv
Statement of Original Authorship xxxi
List of Symbols ..xxxiii
Acknowledgements .xxxv
CHAPTER 1
CHAPTER 2
ix
Table of Contents
x
Table of Contents
CHAPTER 3
CHAPTER 4
xi
Table of Contents
CHAPTER 5
CHAPTER 6
xii
Table of Contents
CHAPTER 7
CHAPTER 8
xiii
Table of Contents
CHAPTER 9
Appendix A.A-1
Appendix B.B-1
Appendix C.C-1
Appendix D.D-1
Appendix E.E-1
References..........................................................................................................R-1
xiv
List of Figures
LIST OF FIGURES
xv
List of Figures
xvi
List of Figures
Figure 2.51: Schematic Diagram for Lateral Buckling Tests of HFBs ................... 2-64
Figure 2.52: Schematic Diagram of Test Rig Including Support System ............... 2-65
Figure 2.53: Lateral Buckling Tsts of RHS Beams ................................................ 2-66
Figure 2.54: Test Arrangement for C- and Z- Section Beams ................................ 2-67
Figure 4.1: Experimental Results of Mahaarachchi and Mahendran (2005a) .......... 4-2
Figure 4.2: LSB Test Specimens............................................................................... 4-3
Figure 4.3: Different Types of Test Methods ........................................................... 4-5
Figure 4.4: Overall View of Test Rig........................................................................ 4-6
Figure 4.5: Support System ....................................................................................... 4-7
Figure 4.6: Flange Twist at Failure of a 250x75x2.5 LSB with 3.5 m Span ............ 4-8
Figure 4.7: Flange Twist Restraint Arrangement of LSBs ....................................... 4-9
Figure 4.8: Loading System .................................................................................... 4-10
Figure 4.9: Data Logger and Load Cells ................................................................. 4-12
Figure 4.10: Wire Displacement Transducers (WDTs) .......................................... 4-12
Figure 4.11: Schematic Diagram of a Typical Test Specimen................................ 4-13
Figure 4.12: Schematic Diagram of Flange Twist Restraints ................................. 4-14
Figure 4.13: Typical Lateral Distortional Buckling Failure.................................... 4-15
xvii
List of Figures
Figure 5.1: Schematic Diagrams of Ideal and Experimental FE Models .................. 5-2
Figure 5.2: Actual and Idealised LSBs...................................................................... 5-4
Figure 5.3: Typical Finite Element Mesh for LSB Models....................................... 5-8
Figure 5.4: Stress-Strain Relationships ..................................................................... 5-9
Figure 5.5: Idealised Simply Supported Boundary Conditions............................... 5-10
Figure 5.6: Boundary Conditions of the Ideal Finite Element Model of LSB ........ 5-11
Figure 5.7: Typical Loading Method for the Ideal Finite Element Model of LSB . 5-12
Figure 5.8: Loading and Boundary Conditions of the Experimental Finite Element
Model of LSB .......................................................................................................... 5-14
Figure 5.9: Loading Plate Twisting in the Experimental FE Model ....................... 5-15
Figure 5.10: Various Plate Elements in Experimental Finite Element Model ........ 5-16
Figure 5.11: Critical Buckling Mode from Elastic Buckling Analysis of Ideal Finite
Element Model ........................................................................................................ 5-17
Figure 5.12: Effect of Imperfection Direction Based on Nonlinear Analysis ......... 5-18
Figure 5.13: Residual Stress Distributions in LSB Sections ................................... 5-19
Figure 5.14: Typical Residual Stresses Distribution for LSB Sections .................. 5-20
Figure 5.15: Elastic Buckling Modes of 200x60x2.0 LSB ..................................... 5-23
Figure 5.15: Elastic Buckling Modes of 200x60x2.0 LSB ..................................... 5-24
Figure 5.16: Ultimate Failure Modes of 200x60x2.0 LSB ..................................... 5-25
Figure 5.17: Comparison of Elastic Buckling Moments ......................................... 5-28
Figure 5.18: Bending Moment vs Vertical Deflection at Mid-Span Curves for
150x45x1.6 LSB (3000 mm Span) .......................................................................... 5-30
xviii
List of Figures
xix
List of Figures
Figure 6.12: Comparison of Experimental Moment Capacities with the Design Curve
based on Equations 6.7 (a) to (c) ............................................................................. 6-24
Figure 6.13: Comparison of FEA and Experimental Moment Capacities with the
Design Curve based on Equations 6.7 (a) to (c) ...................................................... 6-24
Figure 6.14: Comparison of FEA Moment Capacities with the Design Curve based
on Equations 6.10 (a) to (c) ..................................................................................... 6-27
Figure 6.15: Comparison of Experimental Moment Capacities with the Design Curve
based on Equations 6.10 (a) to (c) ........................................................................... 6-28
Figure 6.16: Comparison of Experimental Moment Capacities with the Design Curve
based on Equations 6.11 (a) to (c) ........................................................................... 6-29
Figure 6.17: Comparison of FEA Moment Capacities with the Design Curve based
on Equations 6.11 (a) to (c) ..................................................................................... 6-30
Figure 6.18: Hollow Flange Beams......................................................................... 6-30
Figure 6.19: Comparison of FEA Moment Capacities of HFBs from Avery et al.
(1999b) with Equations 6.7 (a) to (c) ...................................................................... 6-35
Figure 6.20: Non-Dimensional Member Moment Capacity versus Modified
Slenderness d for LSBs .......................................................................................... 6-37
Figure 6.21: Non-Dimensional Member Moment Capacity versus Slenderness for
LSBs ........................................................................................................................ 6-38
Figure 6.22: Moment Capacity Design Curve for LSBs based on a Modified
Slenderness Parameter K1 ..................................................................................... 6-39
Figure 6.23: Moment Capacity Design Curve for LSBs based on a Modified
Slenderness Parameter K2d .................................................................................... 6-40
Figure 6.24: Moment Capacity Design Curve for LSBs based on a Modified
Slenderness Parameter Kd ..................................................................................... 6-42
Figure 6.25: Comparison of Experimental Results with Equation 6.18.................. 6-43
Figure 6.26: Moment Capacities of LSBs with Similar Values of GJf/EIxweb (Set 1) . 6-
44
Figure 6.27: Moment Capacities of LSBs with Similar Values of GJf/EIxweb (Set 2) . 6-
45
Figure 6.28: Moment Capacities of LSBs with Similar Values of GJf/EIxweb (Set 3) . 6-
46
xx
List of Figures
Figure 6.29: Comparison of FEA Moment Capacities of HFBs from Avery et al.
(1999b) with Equation 6.18 .................................................................................... 6-47
Figure 6.30: Comparison of FEA Moment Capacities of Selected HFBs from Avery
et al. (1999b) with Equation 6.18 ............................................................................ 6-47
Figure 6.31: Moment Capacities of HFBs with Similar Values of GJf/EIxweb (Set 1) 6-
49
Figure 6.32: Moment Capacities of HFBs with Similar Values of GJf/EIxweb (Set 2) 6-
49
Figure 6.33: Moment Capacities of New LSBs with Different GJf/EIxweb Values . 6-51
Figure 6.34: Moment Capacities of Hollow Flange Steel Beams with GJf/EIxweb
0.0811 ...................................................................................................................... 6-53
Figure 6.35: Moment Capacities of Hollow Flange Steel Beams with the Modified
Slenderness Parameter K as Defined in Equation 6.19 ........................................... 6-54
Figure 6.36: Moment Capacities of Hollow Flange Steel Beams with the Modified
Slenderness Parameter K as Defined in Equation 6.20 ........................................... 6-54
Figure 6.37: MHFB and RHFB Sections ................................................................ 6-55
Figure 6.38: Comparison of Moment Capacities of Hollow Flange Steel Beams with
Similar Values of GJf/EIxweb ................................................................................... 6-61
xxi
List of Figures
xxii
List of Figures
Figure 8.8: Ideal Finite Element Model with Full Twist Restraint at the Supports
(Including Flanges) and Web Stiffeners ................................................................. 8-10
Figure 8.9: Idealised Simply Supported Boundary Conditions .............................. 8-11
Figure 8.10: Loading Method of Ideal Finite Element Model ................................ 8-12
Figure 8.11: Elastic Lateral Distortional Buckling Failure Modes of LSBs with
Various Stiffener Arrangements ............................................................................. 8-15
Figure 8.11: Elastic Lateral Distortional Buckling Failure Modes of LSBs with
Various Stiffener Arrangements ............................................................................. 8-16
Figure 8.12: LSBs with Web Stiffeners at Different Spacings ............................... 8-18
Figure 8.13: Elastic Lateral Buckling Modes of LSBs ........................................... 8-19
Figure 8.13: Elastic Lateral Buckling Modes of LSBs ........................................... 8-20
Figure 8.14: Elastic Lateral Distortional Buckling of LSB with Web Stiffener ..... 8-24
Figure 8.15: Modw/Mo versus Span for LSBs with Web Stiffeners ......................... 8-28
Figure 8.16: Modw/Mo versus Slenderness for LSBs with Web Stiffeners .............. 8-28
Figure 8.17: Comparison of Modw with Equation 8.2.............................................. 8-29
Figure 8.18: Comparison of Modw with Equation 8.3.............................................. 8-30
Figure 8.19: Lateral Buckling Mode of a 2 m Span 150x45x2.0 LSB from Non-linear
FEA ......................................................................................................................... 8-32
Figure 8.20: Ultimate Moments of LSBs with Web Stiffeners............................... 8-36
Figure 8.21: Comparison of Ultimate Moments of LSBs with and without Web
Stiffeners ................................................................................................................. 8-36
Figure 8.22: Comparison of Ultimate Moments with Equation 8.4........................ 8-37
Figure 8.23: Comparison of Ultimate Moments with Equation 8.5........................ 8-38
Figure 8.24: Comparison of Ultimate Moments with Equation 8.6........................ 8-39
Figure 8.25: Comparison of Ultimate Moments with Equation 8.7........................ 8-40
Figure 8.26: Comparison of Ultimate Moments with Equation 8.8........................ 8-41
xxiii
List of Figures
xxiv
List of Tables
LIST OF TABLES
Table 2.1: Avery et al.s (1999b) Coefficients for Equation 2.32........................... 2-34
Table 2.2: Idealised Simply Supported Boundary Conditions ................................ 2-49
Table 2.3 Nonlinear Analysis Parameters ............................................................... 2-51
Table 2.4: Tensile Coupon Test Results ................................................................. 2-57
xxv
List of Tables
Table 5.3: Elastic Lateral Buckling Moments of Actual and Idealised LSB Sections
................................................................................................................................. ..5-5
Table 5.4: Percentage Differences in Elastic Lateral Buckling Moments of Idealised
and Actual LSBs........................................................................................................ 5-6
Table 5.5: Idealised Simply Supported Boundary Conditions ................................ 5-10
Table 5.6: Membrane Residual Stress Distribution of LSB Sections ..................... 5-20
Table 5.7: Comparison of Elastic Buckling Moments of LSB from FEA, Thin-Wall
and Pi and Trahairs (1997) Equation ..................................................................... 5-26
Table 5.7 (Continued): Comparison of Elastic Buckling Moments of LSB from FEA,
Thin-Wall and Pi and Trahairs (1997) Equation.................................................... 5-27
Table 5.8: Comparison of Experimental and FEA Ultimate Moment Capacities ... 5-30
xxvi
List of Tables
Table 6.10 (continued): Comparison of Avery et al.s (1999b) FEA Results with
Eq.6.7 ...................................................................................................................... 6-34
Table 6.11: Capacity Reduction factors for Eq.6.18 ............................................... 6-43
Table 6.12: Section Properties of LSBs including K .............................................. 6-44
Table 6.13: Section Properties of HFBs including K.............................................. 6-46
Table 6.14: Two New LSBs with Different GJf/EIxweb and K Values .................... 6-50
Table 6.15: FEA Moment Capacity Results of Two New LSBs ............................ 6-50
Table 6.16: Two New LSBs with Higher Values of GJf/EIxweb .............................. 6-52
Table 6.17: FEA Moment Capacity Results of Two New LSBs with Higher Values
of GJf/EIxweb ............................................................................................................ 6-52
Table 6.18: Dimensions of MHFB and RHFB Sections ......................................... 6-56
Table 6.19: Section Properties of MHFBs and RHFBs including K ...................... 6-56
Table 6.20: FEA Results of MHFB Sections without Residual Stresses ................ 6-57
Table 6.21: FEA Results of RHFB Sections without Residual Stresses ................. 6-58
Table 6.22: FEA Results of LSB Sections without Residual Stresses .................... 6-59
xxvii
List of Tables
Table 7.14: Compactness of LSBs Based on AS 4100 and AS/NZS 4600 ............. 7-33
Table 7.15: Section Moment Capacities of LSBs ................................................... 7-33
Table 7.16: Comparison of Ultimate Moment Capacities from FEA and Current
Design Rules ........................................................................................................... 7-35
Table 7.17: Comparison of Ultimate Moment Capacities from FEA and Eurocode 3
Part 1.3 (ECS, 2006 & 1996) .................................................................................. 7-37
Table 7.18: Comparison of Ultimate Moment Capacities from FEA and Shifferaw
and Schafer (2008) .................................................................................................. 7-40
Table 7.19: The Ratios of Mu/My and Section Compactness .................................. 7-41
Table 7.20: Dimensions and Properties of Non-Standard Compact LSBs ............. 7-42
Table 7.21: The Ratios of Mu/My of Some Non-Standard Compact LSBs ............. 7-42
Table 7.22: Comparison of Ultimate Moment Capacities from FEA and Eurocode 3
Part 1.3 (NSAI, 2006) for Non-Standard Compact LSBs ....................................... 7-44
Table 7.23: Comparison of Ultimate Moment Capacities from FEA and Shifferaw
and Schafer (2008) for Non-Standard Compact LSBs ............................................ 7-45
Table 7.24: Average and Maximum Membrane Strains of LSB Sections at Failure
............................................................................................................................. 7-48
xxviii
List of Tables
xxix
List of Tables
xxx
STATEMENT OF ORIGINAL AUTHORSHIP
The work contained in this thesis has not been previously submitted to meet
requirements for an award at this or any other higher education institution. To the
best of my knowledge and belief, the thesis contains no material previously
published or written by another person except where due reference is made.
Tharmarajah Anapayan
Signed: __________________________________________________
Date: __________________________________________________
xxxi
xxxii
List of Symbols
LIST OF SYMBOLS
b = plate width
COV = Coefficient Of Variance
E = Youngs modulus
fcr = elastic critical buckling stress
fu = ultimate tensile strength
fyif = inner flange yield stress
fyof = outer flange yield stress
fyw = web yield stress
G = shear modulus
HFSB = Hollow Flange Steel Beam
HFB = Hollow Flange Beam
Ix = second moment of area about major axis
Iy = second moment of area about minor axis
Iw = warping constant
J = torsional constant
k = buckling co-efficient
K = geometrical parameter
L = span
= slenderness
d = modified slenderness
dw = modified slenderness with web stiffeners
LDB = Lateral Distortional Buckling
LSB = LiteSteel Beam
Mb = member moment capacity
Mc = critical moment
Mo = elatic lateral torsional buckling moment
Mod = elatic lateral distortional buckling moment
Modw = elatic lateral distortional buckling moment with web stiffeners
Mp = plastic moment
MPC = Multiple Point Constraint
xxxiii
List of Symbols
xxxiv
Acknowledgements
ACKNOWLEDGEMENTS
The author wishes to express sincere gratitude to his supervisor, Professor Mahen
Mahendran for his patient guidance, invaluable expertise, rigorous discussions 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. Jung Kwan Seo for his assistance and experience and friendship during his
postdoctoral study at QUT.
Author would like to thank QUT and OneSteel Australian Tube Mills (OATM) and
Australian Research Council (ARC) for providing financial support to this research.
The author would also like to thank Mr. Ross Dempsey, Manager - Research and
Testing, OneSteel Australian Tube Mills for his technical contributions, and his
overall support to the many different phases of this research project. Thanks also to
the School of Urban Development and the Faculty of Built Environment and
Engineering at QUT for providing the necessary facilities and technical support.
Many thanks to the structural laboratory staff members, particularly Mr. Arthur
Powell, Mr. Brian Pelin and Mr. Terry Beach for their assistance with operating the
equipment, fabrication and preparation of test set-up and specimens. Also many
thanks to staffs of high performance computing (HPC) and research support services
for providing necessary facilities and support with high performance computers and
relevant finite element packages. Special thank is given to Mr. Mark Barry for his
great help regarding HPC facilities.
The author wishes to thank Dr. John Papangelis for his assistance with THIN-WALL
program. Special thanks are given to senior postgraduate students, Dr. Yasintha
Bandulaheva and Mr. Win Kurniawan for their support during this research. It is also
important to thank fellow post-graduate students, Mr. Sivapathasunderam Jeyaragan,
Ms. Nirosha Dolamune Kankanamge, Mr. Poologanathan Keerthan, Mr.
Shanmuganathan Gunalan and Mr. Balachandren Baleshan for their support and
xxxv
Acknowledgements
contribution to this research, and other postgraduate students for their friendship at
QUT.
Finally, the author wishes to express his sincere appreciation to his parents and sister,
particularly his mother, for their blessings, providing endless support and
encouragement and beliefs in his abilities.
xxxvi
Introduction
CHAPTER 1
1.0 INTRODUCTION
Cold-formed steel members have been widely used in building applications for over
five decades. Their markets include the secondary cladding and purlin applications as
well as the primary applications as beams and columns of industrial, commercial and
housing systems. The reasons behind the growing popularity of these cold-formed
steel products include their ease of fabrication, high strength to weight ratio and
suitability for a wide range of applications. These advantages can result in more cost-
effective designs compared with hot-rolled steel members, especially in short-span
applications.
Cold-formed steel members can be produced in a wide variety of section profiles, the
most commonly used of are the C- (channels) and the Z- sections. The thickness of
steel most frequently used for these structural members ranges from about 0.4 mm to
6.4 mm.
1-1
Introduction
Although these cold-formed steel members are considered to be more efficient than
hot-rolled steel members, they suffer from many complex buckling modes and their
interactions because they are usually slender sections that are either unsymmetric or
singly symmetric. Therefore an advanced cold-formed section, called the Hollow
Flange Steel Beams (HFSBs), was identified by cold-formed steel researchers,
manufacturers and designers as an alternative and improved section to replace the
conventional cold-formed C- and Z- sections and smaller hot-rolled I- and channel
sections (Dempsey, 1990 and Mahendran and Avery, 1997).
The Hollow Flange Steel beams (HFSB) are a new group of cold-formed steel
sections made of two torsionally rigid closed flanges and a slender web. Such
innovative sections have a unique geometry and light weight compared to traditional
hot-rolled steel members. They are also more efficient structurally than hot-rolled
steel members. Recently, two different types of HFSs such as Hollow Flange Beam
(HFB) and LiteSteel Beam (LSB) have been developed for use in the building and
construction industries. The first HFS manufactured by OneSteel Australian Tube
Mills (OATM) formerly known as Smorgon Steel Tube Mills (SSTM) during early
1990s is the HFB, which was also called as DogBone.
The HFB is a unique cold-formed steel section developed for use as flexural
members. It was manufactured from a single strip of high strength steel (G450 steel
with a minimum guaranteed yield stress of 450 MPa) using electric resistance
1-2
Introduction
welding. The structural efficiency of the HFB due to the torsionally rigid closed
triangular flanges combined with economical fabrication process was the basis of
HFB development.
1-3
Introduction
Figures 1.3 (a) and (b) show the typical cross-section and an isometric view of HFB,
respectively while Table 1.1 presents the details of such HFBs. This doubly
symmetric member has been used as both compression and flexural members.
The HFBs when used as flexural members are subjected to a relatively new Lateral
Distortional Buckling (LDB) mode which reduces their moment capacity. This
caused the researchers to focus on this detrimental effect in the 1990s. It can be seen
in Table 1.1 that the flange width was 90 mm for all the HFBs and other flange
widths could not be manufactured using the existing equipment. The electric welding
process was also found to be somewhat expensive for the manufacturers. Therefore
the HFB production was discontinued in 1997.
The LiteSteel Beam (LSB) is the recently invented hollow flange steel beam
developed by OATM using a patented Dual Electric Resistance Welding (DERW)
technique. The LSB has a unique shape and manufacturing process which provides
an extremely efficient strength to weight ratio. It has potentially wide range of
applications in residential, commercial, and industrial construction, and is on average
40% lighter than traditional hot-rolled structural sections of equivalent bending
strength.
1-4
Introduction
Figure 1.4 shows the typical section of LSBs. The high strength steel material used
for LSBs is DuoSteel grade with a web yield stress of 380 MPa and a flange yield
stress of 450 MPa. Initially it is from a base steel with a yield stress fy of 380 MPa
and a tensile strength fu of 490 MPa. However, the cold-forming process improves
the yield stress and tensile strength of the LSB flanges to 450 MPa and 500 MPa,
respectively (not for web). The mechanical properties of steel used in the design of
LSBs are given in Table 1.2.
Currently there are 13 variations of the LSBs which range from a depth of 125 mm to
300 mm while the width of the hollow flange varies from 45 mm to 75 mm. The
thickness of steel used for the beams ranges from 1.6 mm to 3.0 mm. The LSB is
manufactured in standard lengths of 12 and 14.5 metres. Table 1.3 shows the section
dimensions for the range of commercially available LSB members.
Designation Flange
Mass
d x bf x t Depth
mm mm mm kg/m mm
300 x 75 x 3.0 LSB 14.4 25.0
2.5 LSB 12.1 25.0
300 x 60 x 2.0 LSB 8.71 20.0
250 x 75 x 3.0 LSB 13.3 25.0
2.5 LSB 11.2 25.0
250 x 60 x 2.0 LSB 7.93 20.0
200 x 60 x 2.5 LSB 8.81 20.0
2.0 LSB 7.14 20.0
200 x 45 x 1.6 LSB 4.90 15.0
150 x 45 x 2.0 LSB 5.26 15.0
1.6 LSB 4.27 15.0
125 x 45 x 2.0 LSB 4.87 15.0
1.6 LSB 3.95 15.0
1-5
Introduction
The HFS manufacturing process is illustrated in Figure 1.5. Cleaning and painting is
then performed prior to bundling and stacking. LSB is coated with the AZ+ alloy
coating system while HFB is coated with general water based paint. It provides a
coating thickness of 18-24 microns and protects up to twice the level provided by a
traditional steel tube primer and has resistance to scratching.
1-6
Introduction
Hollow flange steel beams are light weight and most economical cold-formed steel
sections. Even though different types of hollow flange steel beams have been
investigated by researchers in the past, the only such section that is currently
available is the LSB. It has found increasing popularity in residential, industrial and
commercial buildings not only due to their light weight and cost effectiveness, but
also due to their beneficial characteristics of including torsionally rigid flanges
combined with economical fabrication processes. The LSB sections can be used as
flexural members, truss members and studs in a range of building systems. They
have been used in both residential and commercial buildings. Some of the
applications of LSBs are illustrated in Figures 1.6 (a) to (e).
1-7
Introduction
(e) Purlins
Figure 1.6: Applications of LSBs
(http://www.litesteelbeam.com.au)
The LSB is on average 40% lighter than traditional hot-rolled steel beams of
equivalent performance. This is because of the improved structural performance in
terms of load carrying capacity. The LSB can be lifted and carried like a timber beam
and can be easily worked to run services through or fix other materials to it.
The light weight of LSB provides it with a greater ease of constructability and on-site
versatility and limits the necessity of cranes and other heavy lifting equipment. The
beam material also ensures an ease of construction for the builder as standard power
tools can be used to cut, drill and install it. The connection attributes of LSB allowed
the builder to connect the floor bearers directly to the RHS posts and then fix the
floor joists to the bearers using Tek-screws and therefore off-site fabrication is not
required. The LSB is easy to weld like other structural steel beams if required. One
of the key benefits of LSB is its unique profile, with a thin flat web and two hollow
flanges at the top and bottom. The web is easily worked, allowing for cabling and
other services to be run through it.
LSB exhibits several practical advantages over conventional beams such as hot-
rolled steel and timber sections. Some of them are as follows;
1-8
Introduction
LSB is 'termite proof' and non-combustible, i.e., steel is the preferred option
for the substructure.
LSB has smaller deflection at cantilever end when compared with the
equivalent timber beam.
Non shrinking and non creeping at ambient temperatures.
LSB is more economical and durable than timber.
Two men could easily handle and place the longest beam using simple lifting
aids.
Ease of attachment of timber joists and balustrade posts.
Ease of drilling holes on site to permit bolted assembly.
Formwork unneeded.
Economy in transportation and handling.
The use of thin-walled, cold-formed high strength steel products in the building
industry has significantly increased in recent years. This directed researchers to focus
in this area, particularly Hollow Flange Steel Beams (HFSBs), which are the newly
invented sections by OATM. The HFSs include HFBs and LSBs as described in the
last section. Since the HFB is currently not available in the industry, this research
was mainly focus on the LSBs and then the applicability of the outcomes (design
rules and recommendations) was investigated for HFBs.
The HFBs and LSBs when used as flexural members are subjected to a relatively
new Lateral Distortional Buckling (LDB) mode which reduces the member moment
capacity significantly for intermediate spans. Unlike the commonly observed lateral
torsional buckling of steel beams, the lateral distortional buckling of HFSs is
characterised by simultaneous lateral deflection, twist and cross sectional change due
to web distortion (see Figure 1.7). Although some research (Mahaarachchi and
Mahendran, 2005 a-e, Avery et al., 1999 a, b, and 2000 a, b) has been completed on
HFSs, the effect of hollow flanges and the relative rigidity between flanges and web
elements on the lateral distortional buckling behaviour is not fully understood.
1-9
Introduction
Tension
Section Twist
Web Distortion
Flange Lateral Deflection
Compression
The current design rules in AS/NZS 4600 (SA, 2005) for flexural members subject to
lateral distortional buckling were developed by Mahaarachchi and Mahendran
(2005d) based on the lower bound to the results from the numerical and experimental
studies of the currently available LSB sections (13 of them). The effect of LSB
section geometry was not considered despite the fact that the member geometry
influences the lateral distortional buckling behaviour. It is also necessary to develop
accurate design rules that are applicable to other hollow flange steel beams. The
critical geometric parameter that determines the lateral distortional buckling of LSBs
is unknown. In recent times, the manufacturing process of LSBs has been further
improved while a different steel grade has also been introduced. Such changes to
LSBs and their manufacturing process are likely to influence the lateral buckling
moment capacities of LSBs. Therefore it is necessary to verify the adequacy of
current design rules for the available LSBs. It is also important to fully understand
the effect of section geometry including the relative rigidity between hollow flange
and web elements on their lateral distortional buckling behaviour and to include a
suitable parameter in the relevant design rules.
1-10
Introduction
inelastic bending capacity nor included the recently produced LSB sections. Also, a
finite element model to predict the section moment capacity of LSBs has not been
developed. Therefore it is necessary to investigate the presence of inelastic reserve
bending capacity of LSBs based on experimental and finite element analyses.
Past research by Avery and Mahendran (1997) and Mahendran and Avery (1997)
stated that the use of transverse web plate stiffeners effectively eliminated the
detrimental lateral distortional buckling of HFBs. Figure 1.8 shows the web stiffener
(connected to the flanges) arrangement developed by Avery and Mahendran (1997).
However, Kurniawans (2005) investigations on LSBs produced some conflicting
outcomes. His experimental studies based on quarter point loading showed that the
use of web stiffeners did not significantly improve the flexural moment capacity of
LSBs while his finite element analyses based on an ideal finite element model of
LSB with ideal support conditions and a uniform moment improved the lateral
buckling moment capacities. This contradiction should be investigated and the
optimum web stiffener configuration that improves the lateral distortional buckling
moment capacity of LSBs should be determined.
Simple design rules are not available to predict the elastic lateral buckling and
ultimate moments of LSBs with web stiffeners. Therefore it is important to
investigate the effect of web stiffeners on the lateral distortional buckling moment
capacities of LSBs and to develop suitable design rules to predict their elastic lateral
buckling and ultimate moment capacities.
1-11
Introduction
The overall objective of this research is to investigate the member moment capacity
of LSBs with and without web stiffeners subject to lateral distortional and lateral
torsional buckling effects, and the section moment capacity of LSBs including their
inelastic reserve bending capacity so that safe and efficient design guidelines can be
developed for LSB flexural members in relation to their applications in the
construction industry.
3. Use the developed ideal finite element models with ideal support conditions
and a uniform moment in a parametric study to investigate the lateral
distortional and lateral torsional buckling modes of failures and the reduction
of member moment capacity due to lateral distortional buckling of LSBs, and
compare with predicted member moment capacities using the current
AS/NZS 4600 (SA, 2005) design rules. Modify and develop new design rules
if necessary. (Phase 1).
1-12
Introduction
5. Verify the adequacy of the developed member moment capacity design rules
for lateral distortional buckling for other types of hollow flange steel beams
such as hollow flange beams. (Phase 1).
9. Develop suitable finite element models of LSBs with web stiffeners and
undertake a study to determine the optimum web stiffener configuration that
eliminates/reduces the detrimental lateral distortional buckling effects. (Phase
3).
10. Develop appropriate design rules to predict the elastic lateral distortional
buckling and ultimate member moment capacities of LSBs with the optimum
web stiffener configuration based on a detailed parametric study. (Phase 3).
11. Verify the applicability of the geometrical parameter found for unstiffened
LSBs to LSBs with web stiffeners. Develop or modify design equations based
on this geometrical parameter to predict the member moment capacities of
LSBs with web stiffeners. (Phase 3).
Ultimately, the accurate design rules developed for unstiffened and stiffened LSBs
will enhance their structural efficiency, mitigate lateral distortional buckling effects,
allow designers to make use of the increased lateral buckling capacities and the
1-13
Introduction
available inelastic reserve bending capacities, and increase their range of applications
in the construction industry.
The scope and limitations of this research based on the above three phases are as
follows:
1. Lateral distortional buckling behaviour of hollow flange steel beams subject
to a uniform bending moment with shear centre loading was considered.
However, the effects of moment gradient and load height were not
considered.
2. Elastic local buckling effects that may occur in thinner hollow flange steel
beams were not considered nor the interaction effects of local and lateral
buckling modes of failure.
3. Effects of transverse web stiffeners on the lateral distortional buckling
moment capacity of LSBs were considered under a uniform bending moment
with shear centre loading. However, the effects of moment gradient and load
height were not considered.
1-14
Introduction
Chapter 2 presents the relevant literature review to successfully carry out this
research. It describes the buckling behaviour of cold-formed steel
members and hollow flanges sections. Further, it describes the past
research conducted on cold-formed steel members and LSBs subject to
lateral distortional buckling and the current design methods.
Chapter 3 presents the details of tensile coupon tests, residual stress measurements
and imperfection measurements of LSBs.
Chapter 6 presents the details of a parametric study on the lateral buckling of LSBs,
effects of imperfections and residual stresses and the development of
design rules based on the ideal finite element model of LSBs subject to
lateral buckling. This chapter also includes the investigation on the
effects of section geometry of LSBs on the lateral distortional buckling
behaviour and the development suitable design rules based on a suitable
geometrical parameter.
1-15
Introduction
Chapter 8 presents the details of the effect of web stiffeners on the lateral
distortional buckling moment capacity of LSBs based on finite element
analyses. Suitable design rules to calculate the elastic lateral buckling
and ultimate moment capacities of LSBs with web stiffeners are also
presented in this chapter.
Chapter 9 presents the significant findings from this research and the
recommendations for future research.
1-16
Literature Review
CHAPTER 2
2.0 LITERATURE REVIEW
Figure 2.1 shows the cross sections of cold-formed steel sections with different
shapes used in the industry. Cold-formed structural steel members can be classified
into two major types such as individual structural framing members and panel and
decks. The usual shapes of the cold-formed steel used for individual structural
framing are channels (C- sections), Z- sections, angles, hat sections, I- sections, T-
sections and tubular members. Figure 2.2 (a) shows some cold-formed sections used
in structural framing. The major function of this type member is to carry load,
structural strength and stiffness are the main considerations in design. Such sections
can be used as primary framing members in buildings up to six stories in height. In
tall multi-story buildings the main framing is typically of heavy hot-rolled shapes
2-1
Literature Review
and the secondary elements may be of cold-formed steel members such as steel
joists, decks, or panels. In this case the heavy hot-rolled steel shapes and the cold-
formed steel sections supplement each other. Cold-formed sections are also used as
chord and web members of open web steel joists, space frames, arches and storage
racks.
Another category of cold-formed steel sections is shown in Figure 2.2 (b). These
sections are generally used for roof decks, floor decks, wall panels, siding material
and bridge forms. Some deeper panels and decks are cold formed with web
stiffeners. Steel panels and decks not only provide structural strength to carry loads,
but they also provide a surface on which flooring, roofing, or concrete fill can be
applied. They can also provide space for electrical conduits, or they can be perforated
and combined with sound absorption material to form an acoustically conditioned
ceiling.
2-2
Literature Review
Brake forming involves producing one complete fold at a time along the full length
of the section, using a machine called a press brake. Figure 2.4 illustrates the stages
of press braking processes. For sections with several folds, it is necessary to move
the steel plate in the press and to repeat braking operation several times.
Roll forming is the more popular process for producing large quantities of a given
shape. Hollow flange steel beams such as Hollow Flange Beams (HFBs) and
LiteSteel Beams (LSBs) are manufactured by roll forming process. The initial
tooling costs are high, but the subsequent labour content is low. Press braking is
normally used for low-volume production where a variety of shapes are required.
The main disadvantage of roll forming is the time it takes to change rolls for a
different size section.
Stress-Strain Relationship
There are two common types of typical stress-strain curves such as sharp yielding
type and gradual-yielding type. The sharp yielding type is a typical of stress-strain
curve of medium strength cold rolled steel which shows a linear region followed by a
distinct plateau then the strain hardening up to the ultimate tensile strength before
reaching the failure. On the other hand, high strength steel (i.e. G450) does not
exhibit a yield point with a yield plateau. Gradual-yielding occurs after the linear
region.
Ductility
Ductility is defined as the ability of a material to undergo sizable plastic deformation
without fracture. It reduces the harmful effects of stress concentrations and permits
cold-forming of a structural member without impairment of subsequent structural
2-3
Literature Review
behaviour. This is not only important for the cold-forming process, but also to avoid
catastrophic brittle behaviour in structural members. High strength steel has a lower
ductility as a consequence of higher yield stress as shown in Figure 2.5 (b). A
conventional measure of ductility is the percentage permanent elongation after
fracture in a 50 mm gauge length of a standard tension coupon. AS/NZS 4600 (SA,
2005) Clause 1.5.1.4 states the ductility criterion of cold-formed steel.
2-4
Literature Review
The yield stress and the Youngs Modulus vary significantly depending on the
location in the cross section. The yield stress is higher in the rounded corners than in
the flats, but the Youngs Modulus is lower. The stress-strain relationship for the
corners is also different from that for the flats (Put et al., 1999).
Residual Stresses
Residual stresses occur as a result of manufacturing and fabricating processes.
Unlike in hot-rolled members which often have uniform membrane residual stresses,
the thin-walled members have uneven distribution and higher flexural stresses due to
the cold forming. The residual stress causes premature yielding than is expected if it
is neglected, and it reduces the member stiffness. Hancock et al. (2001) stated that
the increased residual stress is one of the factors that cause rapid fracture.
Current Australian cold-formed steel Standard AS/NZS 4600 (SA, 2005) was based
mainly on the latest AISI Specification. Standards Australia published AS 4100 (SA,
1998) for steel structures which is the latest version and has been most suitable for
2-5
Literature Review
hot-rolled members. British Standard BS 5950 Part 5 (BSI, 1998) specifies guidance
for the design of cold-formed structural steel works. Eurocode 3 (ECS, 1996) and
Canadian Standards are the other international standards providing design guidance
for cold-formed steel structures. The Direct Strength Method is the new design
method for cold-formed steel members and is adopted in the supplement to the North
American specification (AISI, 2004). It is also included in Section 7 of AS/NZS
4600 (SA, 2005). A Direct Strength Method Design Guide (CF06-1) was published
by AISI (2006).
The cross section deformations associated with each of the three buckling modes are
illustrated in Figure 2.7. Local buckling involves distortion of the cross section with
only rotation occurring at interior fold lines of the section. Distortional buckling
involves distortion of the cross section with rotation and translation occurring at
interior fold lines. Lateral torsional buckling excludes distortion of the cross section;
however, translation and rotation of the entire cross section occur. Local buckling
occurs in short span members while lateral torsional buckling occurs in long span
members and distortional buckling in beams with intermediate spans. Figures 2.8 (a)
to (d) show the different modes of buckling failure that occur in channel sections.
2-6
Literature Review
(a) Local (b) Distortional (c) Lateral Distortional (d) Lateral Torsional
Past investigations have revealed two distinctive distortional buckling modes that are
commonly observed in cold-formed steel members namely flange distortional
buckling and lateral distortional buckling. Flange distortional buckling involves
rotation of a flange and lip about the flange/web junction of a C- section or Z-
section while lateral distortional buckling involves transverse bending of vertical
web (see Figures 2.9 (a) and (b)). Flange distortional buckling is most likely to occur
in the open thin-walled sections such as C- and Z- sections while lateral distortional
buckling is the most likely in hollow flange steel beams where the high torsional
rigidity of the tubular compression flange prevents it from twisting during lateral
displacement (Pi and Trahair, 1997). Clause 3.3.3.3 of AS/NZS 4600 (SA, 2005)
gives a comprehensive review of distortional buckling.
2-7
Literature Review
The flange distortional and lateral buckling of cold-formed C- and Z- section steel
members has been extensively investigated. Lau and Hancock (1987) presented
distortional buckling formulae for channel columns while Kwon and Hancock (1991)
proposed design equations for channel section columns undergoing local and
distortional buckling. Hancock et al. (1994) provided design strength curves for thin-
walled C- sections undergoing distortional buckling. Zhao et al. (1995) carried out
lateral buckling tests on rectangular hollow section beams and proposed design
formulations for member moment capacity while Pi and Trahair (1995) developed
lateral buckling strength equations for rectangular hollow sections. Rogers and
Schuster (1997) investigated the distortional buckling of cold-formed steel C-
sections in bending while Hancock (1997) provided a design method for distortional
buckling of C- section flexural members. Pi et al. (1998) investigated the lateral
buckling strength of channel section beams while Put et al. (1999) conducted lateral
buckling tests on channel beams. Put et al. (1998) carried out lateral buckling tests on
Z- beams while Pi et al. (1999) provided lateral buckling strength formula for Z
sections. Lecce and Rasmussen (2005) carried out experimental investigation on
distortional buckling of stainless steel channel sections and Yu and Schafer (2006)
investigated the distortional buckling behaviour of C- and Z- sections by a series of
experiments. It has been found that the cold-formed hollow flange steel beams
severely suffer from lateral distortional buckling due to torsionally rigid flanges with
slender web (Dempsey, 1990) as mentioned earlier.
capacity and stiffness of the beam are the most important criteria. A brief review of
buckling behaviour and design aspects of HFSs such as Hollow Flange Beams
(HFBs) and LiteSteel Beams (LSBs) is presented in this section. Figures 2.10 (a) and
(b) show the typical cross sections of HFB and LSB, respectively.
2
k 2 E t
f cr = (2.1)
12(1 2 ) b
Where k, b and t are the plate local buckling coefficient, plate width and plate
thickness, respectively while E is the Youngs modulus and is Poissons ratio. The
plate local buckling coefficient (k) depends upon the support conditions. Depending
on the restraint conditions along the longitudinal boundaries and the type of loading,
the plate local buckling coefficient (k) takes different values. A plate element is
defined as slender if the elastic critical local buckling stress (fcr) calculated using
Equation 2.1 is less than the material yield stress (fy). A slender section will buckle
locally before the squash load (Py) or the yield moment (My) is reached. Although
local buckling occurs at a stress level lower than the yield stress of steel, it does not
necessarily represent the failure of members. The failure is governed by post-
buckling strength which is generally much higher than the local buckling strength.
The theoretical analysis of post-buckling and failure of plates is extremely difficult,
and generally requires a computer analysis to achieve an accurate solution (Hancock
2-9
Literature Review
et al., 2001). If the elastic critical buckling stress exceeds the yield stress fy, the
compression element will buckle in the inelastic range (Yu, 2000).
The buckling behaviour of HFBs was investigated by Dempsey (1990) using a finite
strip buckling analysis program BFINST6. His buckling analysis has shown that
the buckling coefficients (k) are generally equal to or greater than 4.0 for flange
element and the web element, thus verifying that the flange and web elements are
adequately stiffened. Figure 2.11 shows the buckling stresses over a wide range of
half-wavelengths. Local buckling occurs in the top compression flange at a half-
wavelength of approximately the flat width of the compression element (Point A).
Both of the flange return and the compression portion of the web do not experience
local buckling because the stresses are lower and are not uniform and their flat width
to thickness ratio (b/t) is much smaller.
Figure 2.11: Different Buckling Modes and Stresses of HFB Subject to Bending
(Dempsey, 1990)
2-10
Literature Review
(Mahaarachchi and Mahendran, 2005b). Figure 2.12 shows the typical local buckling
mode of LSBs obtained from finite element buckling analysis.
Tension
Compression
Figure 2.12: Local Buckling Mode of LSB Sections
Compression
Tension
Figure 2.13: Lateral Distortional Buckling Mode of LSB Sections
2-11
Literature Review
Pi and Trahair (1997) stated that the survey of research information on HFBs
indicated that there is no simple formulation for predicting the effect of lateral
distortional buckling on the lateral buckling of HFBs. On this basis, they attempted
to find a simple but sufficiently accurate closed form solution for the effects of web
distortion on the elastic lateral buckling of simply supported HFBs in uniform
bending. They also attempted to develop an advanced theoretical method for
predicting the effects of stress-strain curve, residual stresses and geometrical
imperfections on the strengths of HFBs that fail by lateral-distortional buckling.
The equation for flexural torsional buckling moment resistance Mo (see Equation
2.20) was modified by Pi and Trahair (1997) by introducing an effective torsional
rigidity GJe (see Equation 2.22) in place of the nominal torsional rigidity (GJ) to
calculate the lateral distortional buckling moment resistance Mod (see Equation 2.21).
2-12
Literature Review
The elastic lateral distortional buckling moments, predicted by Thin-wall (MTW) and
obtained from Equation 2.21 (Mod) were compared with flexural torsional buckling
moment (Mo) for two HFB sections by Pi and Trahair (1997), and are shown in
Figure 2.14. It can be seen that the approximate values Mod are in close agreement
with the accurate Thin-wall values MTW, and also these lateral-distortional buckling
values are significantly lower than the flexural-torsional buckling moments Mo.
2-13
Literature Review
decreases with the span and increases with the minor axis flexural rigidity (EIy), the
torsional rigidity (GJ) and the warping rigidity (EIw). For a simply supported beam in
uniform bending, the moment Mo at elastic buckling (Timoshenko and Gere, 1961) is
given by Equation 2.2.
2 EI y 2 EI w
M0 = GJ + (2.2)
L2 L2
Extensive research has shown that the bending moment distribution has a very
significant effect on the elastic buckling resistance (Trahair, 1993) and that uniform
bending is the worst case. Trahair (1995) illustrated the influence of elastic lateral
buckling on the strengths of beams and is shown in Figure 2.16, in which
Msx
= (2.3)
Mo
is a modified slenderness and Msx is the nominal major axis section capacity, as
governed by yielding and local buckling effects.
At low slenderness, the strengths of compact beams rise above the major axis section
capacity Msx = Mpx due to strain hardening effects. The strengths of intermediate
slenderness beams lie on a transition from the section capacity Msx to the elastic
buckling resistance Mo. At high slenderness, the beam strengths are close to the
elastic buckling strengths, but may rise above them due to pre-buckling and post-
buckling strengthening effects which are unaccounted for in the determination of
elastic lateral buckling moment.
Mo
Mo
2-14
Literature Review
Many practical beams are of intermediate slenderness and they fail before the section
capacity Msx or the elastic buckling resistance Mo can be reached. This failure is due
to premature yielding resulting from initial imperfections and twist and residual
stresses. Web distortion also reduces the elastic buckling resistance. As stated in the
earlier sections, cold-formed hollow flange steel beams are more prone to lateral
distortional buckling. Some parameters which influence the beam strength are given
below.
Past research on cold-formed channel sections (Pi et al., 1998), Z-sections (Pi et al.,
1999), rectangular hollow sections (Pi and Trahair, 1995) and HFBs (Pi and Trahair,
1997) incorporated the effects of pre-buckling deflections while the research on
LSBs (Mahaarachchi and Mahendran, 2005d) did not consider them.
2-15
Literature Review
only realised for very slender beams, this post-buckling behaviour causes the beam
strength to remain above the minor axis section capacity Msy, even when this is
greater than the elastic buckling resistance.
Past research identified that the lateral buckling strength of cold-formed channel
section beams (Pi et al., 1998) and Z- section beams (Pi et al., 1999) with web
distortion are lower than those without web distortion. However, the effects of web
distortion are small for these beams as they have low torsional rigidities.
Lateral buckling strengths of HFBs with web distortion are lower than those without
web distortion for beams of intermediate slenderness, but the effects of web
distortion on the strength of beams with very high and very low slenderness are small
(Pi and Trahair, 1997). Findings by Mahaarachchi and Mahendran (2005a) on LSBs
also support this fact.
2-16
Literature Review
Dekker and Kemp (1998) developed a simplified theoretical model for I-sections
undergoing lateral distortional buckling. Trahair and Hancock (2004) developed a
simple advanced method to design steel members against out-of-plane failure. They
included the effects of high moment, residual stresses and geometrical imperfections
on yielding. In this study, it was suggested that the nominal beam strength may be
obtained by making an inelastic lateral buckling analysis using inelastic moduli (EI =
2-17
Literature Review
IME, GI = IMG) which are reduced below their elastic values to allow for the effects
of initial imperfections and twists and residual stresses. The design procedure to
calculate the beam nominal strength derived by Trahair and Hancock (2004) is given
next.
Inelastic lateral buckling moment MI under uniform moment is equal to the nominal
strength Mb and is given in Equation 2.5.
2IM EI y 2 IM EI w
MI = IM GJ + (2.5)
L2 L2
Where
2 2
1 M M
IM = 0.9 while 1.0 (2.6)
1.2 Mpx Mpx
Where, M is the bending moment at the cross section and Mpx is the major axis full
plastic moment. These formulations are suitable for hot-rolled compact I-section
beams but are not suitable for cold-formed steel beams as they have different initial
imperfections, residual stresses and cross sections.
Pi et al. (1998) stated that the lateral buckling strengths of cold-formed channel
section beams with positive twist rotations were higher than those of the beams with
2-18
Literature Review
negative twist rotations. This explains that the negative initial imperfection and twist
are more unfavourable to the lateral buckling strength and the positive initial
imperfection and twist are more desirable. Research on cold-formed Z- sections by Pi
et al. (1999) also supported this.
The magnitudes of the initial imperfection and twist vary randomly and should be
treated probabilistically (Schafer and Pekoz, 1998). However, many researchers
considered consistent magnitudes. Pi and Trahair (1997) found that the central initial
imperfection and twist Uso and so that are consistent with AS 4100 (SA, 1990) for
cold-formed hollow flange beams are given by Equations 2.6 (a) and (b).
Mpx
= slenderness,
Mod
Ny = column elastic buckling load about the minor axis (Equation 2.7)
2 EIy
Ny = (2.7)
L2
Pi and Trahair (1997) investigated the effects of initial geometric imperfection and
twist of HFBs. They found that the strength increases as the initial imperfection
decreases, but the differences are small for beams with very low or very high
slenderness. This means the effects of initial geometric imperfections for beams with
2-19
Literature Review
Mahaarachchi and Mahendran (2005e) measured the initial imperfection and twist of
LSBs and concluded that the local plate imperfections were within the
manufacturers fabrication tolerance limit while the overall member imperfections
were less than the AS 4100 recommended limit of Span/1000 (SA, 1998). However,
it is necessary to measure the initial imperfections of the test beams prior to testing in
order to obtain accurate results.
For hot-rolled steel members, the residual stresses are mainly due to uneven cooling
after hot-rolling. Past researchers (Yang et al., 1952 and Tebedge et al. 1973) found
that the magnitude of maximum residual stress in hot-rolled steel sections made of
moderate strength steels is approximately equal to 30% of the yield stress of the
material and the residual stresses are assumed to be uniformly distributed through the
plate thickness, i.e. no bending residual stresses.
2-20
Literature Review
The inelastic buckling moment varies markedly with both magnitude and the
distribution of the residual stresses. The moment at which inelastic buckling initiates
depends on the magnitude of the residual compressive stresses at the flange tips,
where yielding causes significant reductions in the effective rigidities. The flange tip
stresses are comparatively high in hot-rolled beams, especially those with high ratios
of flange to web area, and so inelastic buckling is initiated comparatively early in
these beams. The residual stresses in hot-rolled beams decrease away from the flange
tips, and so the extent of yielding increases and the effective rigidities steadily
decrease as the applied moment increases. Because of this, the inelastic buckling
moment decreases in an approximately linear fashion as the slenderness increases
(Trahair and Bradford, 1991).
For cold-formed steel members, the residual stresses are mainly caused by the cold-
forming process and thus are quite different from those of hot-rolled sections. In fact,
for cold-formed sections, flexural residual stresses are considered the most important
component and these stresses can be as high as 50% of the material yield stress
(Schafer and Pekoz, 1998). Pi et al. (1998) indicated that residual stresses reduced
the lateral buckling strength of cold-formed channel section beams particularly for
intermediate slenderness. Pi et al. (1999) also demonstrated the same fact for cold-
formed Z- sections and Pi and Trahair (1997) concluded this for HFBs.
Mahaarachchi and Mahendran (2005c) indicated the reduction on the lateral buckling
strength of LSBs was about 8%. Therefore, this effect can not be neglected.
M
M M
M
(a) (b) (c) (d)
2-21
Literature Review
Figure 2.19 (a) shows a uniform bending moment along the beam span caused by
equal and opposite end moments. This is the most severe case, for which yielding is
constant along the beam so that the resistance to lateral buckling is reduced
everywhere. Less severe case is the case of an unbraced beam with a load
concentrated in the centre (Figure 2.19 (b)), for which yielding is confined to a small
central portion of the beam, so that any reductions in the section properties are
limited to this region. Uniformly distributed load also has a similar effect. Even less
severe cases are those of beams with unequal end moments M and M (Figure 2.19
(c)), where yielding is confined to small portions near the supports. The least critical
case is that of equal end moments that bend the beam in double curvature (Figure
2.19 (d)), for which the moment gradient is steepest and the regions of yielding are
most limited.
Put et al. (1999) plotted the member capacity curves for cold-formed channel beams
for different types of moment gradients (Figure 2.20). They also made a comparison
of the design methods of AS 4100 (SA, 1990) for hot-rolled beams and AS/NZS
4600 (SA, 1996) for cold-formed beams and suggested that the single design curve
of AS/NZS 4600 was unduly optimistic for near uniform bending, and unnecessarily
conservative for high moment gradients. The lateral distortional buckling strengths
increase with moment modification factor m towards the linear elastic buckling
curve according to AS 4100 (SA, 1990) predictions. Further, they stated that AS
4100 (SA, 1990) is more accurate, because it makes allowance for the effects of the
moment distributions on the inelastic buckling resistance.
2-22
Literature Review
Pi and Trahair (1997) also provided member capacity curves for HFBs with different
types of moment distribution according to AS 4100 (SA, 1990) predictions.
However, the use of AS 4100 (SA, 1990) for cold-formed steel members is
questionable as cold-formed steel members have different material and mechanical
properties compared to hot-rolled steel members. This research is mainly concerned
with AS/NZS 4600 (SA, 2005) to develop design curves for LSBs and other cold-
formed hollow flange steel beams. It is possible to use appropriate moment
modification factors m to modify the buckling moments. Recently, Kurniawan and
Mahendran (2009b) investigated the moment gradient effects on the lateral buckling
strength of LSBs. They found that the moment modification factor is approximately
equal to 1.0 for quarter point loading, which would be useful in the lateral buckling
tests of LSBs.
Trahair (1993) presented an equation to calculate the elastic lateral torsional buckling
moment of a simply supported beam with load height effects.
Pi and Trahair (1997) modified that equation to calculate the elastic lateral
distortional buckling moment of HFBs and is given below.
Mqd 0.4myqPy
2
0.4myqPy
= m 1 + + (2.8)
Mod Mod Mod
2-23
Literature Review
Where yq is the load height, Py is the transverse load and Mqd is the maximum elastic
buckling moment including the load height effects. Pi and Trahair (1997) concluded
that this equation provides more accurate predictions than when using the effective
length for top flange loading (i.e. effective length, Le = 1.4 L for top flange loading
based on AS 4100 (SA, 1998)). Recently, Kurniawan and Mahendran (2009a)
investigated the effects of load heights on the lateral buckling strength of LSBs.
2.5.9 Warping
Warping occurs when the twisting of a member results in the cross-section distorting
out-of-plane along the direction of the members longitudinal axis. Most of the cold-
formed members except closed tubular sections have cross-sections which tend to
warp when subjected to torsion. When a thin-walled member is restrained at any
particular cross-section, a complex distribution of longitudinal warping stresses is
developed. These stresses act in conjunction with those due to St. Venant torsion to
resist the applied torque.
In practical structures beam ends are connected by web cleats, web stiffeners and
additional web plates. These may induce flange end-restraining moments that oppose
the warping deformations and modify the elastic lateral buckling resistance of the
beam. It is important to determine the end warping restraint stiffness to predict the
elastic lateral distortional buckling resistance accurately. Pi and Trahair (2000)
investigated the effects of distortion and warping at beam support and proposed some
simple formulations to calculate the lateral distortional buckling moments which
allow for both torsional and warping rigidity reductions in uniform bending. They
also provided the warping restraint stiffnesses of some commonly used end-support-
conditions.
2-24
Literature Review
Xiao et al. (2004) investigated the effects of warping stress on the lateral torsional
buckling of cold-formed zed-purlins. They developed an analytical model based on
energy method to analyse the elastic lateral torsional buckling of zed-purlins subject
to partial lateral restraint from the metal sheeting under a uniformly distributed uplift
load. They considered various loading positions and concluded that the influence of
warping stress is less when the load is acting at the web central line, hence increases
the critical load.
2-25
Literature Review
Ms = fy Ze (2.9)
Where, fy is the nominal yield stress. The effective section modulus (Ze) used to
calculate the section moment capacity (Ms) is governed by the compactness of the
sections individual plate equations and is given by Equations 2.10 to 2.13:
The element with the greatest ratio e/ey is to be used for calculating the effective
section modulus (Ze). The plate element slenderness (e) as in Clause 5.5.2 is given
by Equation 2.14:
2-26
Literature Review
b fy
e = (2.14)
t 250
Where, b is the clear width of the element outstand from the face or between the
faces of the supporting plate element and t is the element thickness. The effective
section of the compact element (Zc) is given by Equation 2.15:
Mb = msMs Ms (2.16)
Mb = msdMs Ms (2.17)
The slenderness reduction factors (s and sd) based on AS 4100 (SA, 1998) and Pi
and Trahair (1997), respectively, are given by Equations 2.18 and 2.19, respectively.
Bradford (1992) reported that the relationship between distortional buckling
strength, yielding and elastic distortional buckling is the same as that between the
lateral buckling strength, yielding and elastic lateral buckling. This implies that if
the elastic buckling moment (Mo) is replaced by the elastic buckling moment in the
distortional mode (Mod) then the method outlined by AS 4100 (SA, 1998) is
adequate for the design of hollow flange steel beams subject to lateral distortional
buckling.
2-27
Literature Review
Ms 2 Ms
s = 0.6 +3 1.0 (2.18)
Mod Mod
Ms 2 Ms
sd = 0.6 + 2.8 1.0 (2.19)
Mod Mod
The elastic lateral torsional buckling moment (Mo) of section with equal flanges for
simply supported beam under uniform bending as in AS 4100 (SA, 1998) Clause
5.6.1.1 is:
2 EI y 2 EI w
M0 = GJ + (2.20)
L2 L2
2 EIy 2 EIw
Mod = GJe + (2.21)
L2 L2
Et 3 L2
2GJf
0.91 2 d 1
GJe = (2.22)
Et 3 L2
2GJf +
0.91 2
d 1
Where, GJf is the flange torsional rigidity and d1 is the clear web depth.
2-28
Literature Review
experimental data. The design equation was particularly over-conservative in the low
slenderness region. Both methods tended to slightly underestimate the capacities at
the region of high slenderness, i e. conservative.
Ms = fy Ze (2.23)
2-29
Literature Review
1.052 b f *
= (2.24)
k t E
The procedure to find out the effective widths of uniformly compressed stiffened
elements for capacity calculations is given in Clause 2.2.1.2 of AS/NZS 4600 (SA,
2005).
For 0.673 : be = b (2.25a)
For > 0.673: be = b (2.25b)
0.22
1
= 1.0 (2.26)
The procedure to find out the effective widths of stiffened elements with stress
gradient for capacity calculations is given in Clause 2.2.3.2 of AS/NZS 4600 (SA,
2005).
be
be1 = (2.27)
3
be
For -0.236 : be2 = (2.28a)
2
For > -0.236 : be2 = be be1 (2.28b)
Where
be = effective width from Equation 2.25 with f1* substituted for f* and
with k determined as follows;
k = 4 + 2(1- )3 + 2(1- )
f 2*
= *
f1
f 2*
= web stresses calculated on the basis of the effective section
f 1*
2-30
Literature Review
b be
be
be
(a) Actual Element (b) Effective Element and Stress on Effective Element
2-31
Literature Review
bending of a vertical web with lateral displacement of the compression flange while
the same clause from the latest version of AS/NZS 4600 (SA, 2005) outlines the
procedure developed by Mahaarachchi and Mahendran (2005d). Member moment
capacity Mb is given in Equation 2.29:
Ze
Mb = Mc (2.29)
Z
Where, Z is the full section modulus, Mc is the critical moment and Ze is the effective
section modulus. The critical moment Mc as in AS/NZS 4600 (SA, 1996) is given by
Equations 2.30a and b:
2 d
For d < 1.414: Mc = My 1 (2.30a)
4
1
For d 1.414: Mc = My 2 (2.30b)
d
Where, d = member slenderness (Equation 2.31)
My = first yield moment
My
d = (2.31)
Mod
The elastic lateral distortional buckling moment Mod can be determined using
Equations 2.21 and 2.22 or a buckling analysis program such as THINWALL. A
comparison of the AS/NZS 4600 (SA, 1996) design member moment capacities and
the experimental and FEA results of LSBs made by Mahaarachchi and Mahendran
(2005d) is shown in Figure 2.24.
Mahaarachchi and Mahendran (2005d) divided the ultimate moments from the
experiments by a moment modification factor of 1.09 to account for the non-uniform
moment distribution caused by the quarter-point loading. However, recent research
(Kurniawan and Mahendran, 2009b) on the effects of moment gradient on the lateral
distortional buckling of LSBs revealed that the moment distribution factor is closer
to 1.0 for quarter point loading. With reference to Figure 2.24, Mahaarachchi and
Mahendran (2005d) stated that the design rule of AS/NZS 4600 (SA, 1996) for
lateral distortional buckling was not suitable as it was quite conservative for beams
with low slenderness while being unconservative for intermediate slenderness
(inelastic buckling region). Further, they stated that the web distortion significantly
2-32
Literature Review
reduced the ultimate moment capacities and are not accounted in AS/NZS 4600 (SA,
1996) predictions.
1.40
Experimental
FEA
AS 4600 (SA, 1996)
1.20
1.00
0.80
Mu/My
0.60
0.40
0.20
0.00
0.00 0.50 1.00 1.50 2.00 2.50 3.00
Member Slenderness
Alternative member capacity equations have been proposed by Trahair (1997). The
accuracy of these equations for the design of HFB flexural members was investigated
and lateral distortional buckling design curves were produced at QUT by Avery et al.
(1999b). Design curves for HFBs were derived using the finite element analysis
results of Avery et al. (1999a), which was verified against the lateral distortional
buckling test results of Mahendran and Doan (1999).
Ms
d = (2.33)
M od
The suitable coefficients (a, b, c and n) were established using the least square
method by Avery et al. (1999b). Values of a = 1.0, b = 0.0, c = 0.424, and n = 1.196
were found to minimise the total error for the Trahairs (1997) design equations.
2-33
Literature Review
Even though this approach is more accurate for the HFB section range it is very
complicated and requires different design curves for each thickness of HFB.
Mahaarachchi and Mahendran (2005d) used the coefficients defined by Avery et al.
(1999b) to predict the moment capacities of LSBs and compared them with their
FEA results. They concluded that the predictions based on Avery et al.s (1999b)
method were very similar to AS/NZS 4600 (SA, 1996) predictions and were not
suitable for LSB sections. It is quite conservative for beams of low slenderness while
being unconservative for beams of intermediate slenderness (inelastic buckling
region).
1.4
125x45x1.6 LSB
1.2 125x45x2.0 LSB
150x45x1.6 LSB
150x45x2.0 LSB
1.0
200x45x1.6 LSB
200x60x2.0 LSB
Mu/Ms, Mb/Ms
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Figure 2.25: Comparison of FEA Results with Avery et al.s (1999b) Predictions
(Mahaarachchi and Mahendran, 2005d)
2-34
Literature Review
125x45x1.6 LSB
1.40
125x45x2.0 LSB
150x45x1.6 LSB
150x45x2.0 LSB
1.20
200x45x1.6 LSB
200x60x2.0 LSB
200x60x2.5 LSB
1.00 250x60x2.0 LSB
250x75x2.5 LSB
M u /M y, M b /M y
250x75x3.0 LSB
0.80 300x60x2.0 LSB
300x75x2.5 LSB
300x75x3.0 LSB
0.60 AS4600 (1996)
Equation 2.32
15
Exp 125x45x1.6 LSB
0.40 Exp 125x45x2.0 LSB
Exp 150X45X16 LSB
Exp 150x45x2.0 LSB
0.20 Exp 200x45x1.6 LSB
Exp 200x60x2.5 LSB
Exp 250x75x2.5 LSB
0.00 Exp 250x75x3.0 LSB
0.00 0.50 0.63 1.00 1.50 1.59 2.00 2.50
Exp 250x60x2.0 LSB
Exp 300X60X20 LSB
d Exp 300x75x2.5 LSB
Exp 300x75x3.0 LSB
Figure 2.26: Comparisons of New Design Rules, FEA and Experiments (=0.85)
(Mahaarachchi and Mahendran, 2005d)
2-35
Literature Review
1.400
125x45x1.6 LSB
1.200 125x45x2.0 LSB
150x45x1.6 LSB
150x45x2.0 LSB
1.000 200x45x1.6 LSB
200x60x2.0 LSB
200x60x2.5 LSB
Mu/My, Mb/My
0.200
0.000
0.000 0.500 1.000 1.500 2.000 2.500
Figure 2.27: Comparisons of New Design Rules, FEA and Experiments (=0.90)
(Mahaarachchi and Mahendran (2005d)
2-36
Literature Review
Elastic buckling analysis results are directly integrated into DSM. This provides a
general method of designing cold-formed steel members and creates the potential for
much broader extensions than the traditional specifications with limited applicability.
However, limitations of the direct strength method (as implemented in AISI 2004)
are given in the direct strength method design guide (CF06-1) and some of them are
presented here;
overly conservative if very slender elements are used.
shift in the neutral axis is ignored.
empirical method calibrated only to work for cross-sections previously
investigated.
Section 7 of AS/NZS 4600 (SA, 2005) gives the direct strength method provisions
for cold-formed steel beams subject to bending with separate provisions for local
2-37
Literature Review
Clause 7.2.2.2 of AS/NZS 4600 (SA, 2005) gives design procedure to determine the
nominal member moment capacity (Mbe) for lateral torsional buckling and is given in
Equations 2.36 (a), (b) and (c).
10 10 My
For 2.78 My Mo 0.56 My: Mbe = My 1 (2.36b)
9
36Mo
My= Zf fy (2.37)
Where, Mo is the elastic lateral torsional buckling moment, Zf is the full section
modulus and fy is the yield stress.
As in Clause 7.2.2.3 of AS/NZS 4600 (SA, 2005) the nominal member moment
capacity (Mbl) for local buckling is given in Equations 2.38 (a) and (b).
M be
l = (2.39)
M ol
2-38
Literature Review
Mod Mod
0.5 0.5
My
d = (2.42)
Mod
The elastic distortional buckling moment (Mod) is given by:
It should be noted that the lateral distortional buckling is not included in these
provisions. Also, these design procedures are based on advanced buckling analyses
of cold-formed C- and Z- sections and the applicability of these predictions to hollow
flange steel beams such as HFBs and LSBs are questionable as their sectional
properties are different. However, their applicability should be investigated to
develop such type of design procedures for hollow flange steel beams.
Avery and Mahendran (1997) investigated the lateral buckling capacity of Hollow
Flange Beams (HFBs) with web stiffeners using finite element analysis while
Mahendran and Avery (1997) carried out large scale experiments to validate the
results from finite element analysis and to improve the stiffener configuration. They
found that 5 mm web plate stiffeners welded to the flanges on both sides of the web
at third points of the span were adequate to eliminate the lateral distortional buckling
2-39
Literature Review
Plate, both
Plate, both sides.
sides.
The cross stiffeners, box stiffeners and rectangular hollow section stiffeners only
provided a slightly higher strength increment than the transverse web stiffeners. This
was because the effect of a stiffener was mostly due to the constraints provided,
which were independent of the stiffener size. Their FEA revealed that a web stiffener
welded to flange only (Type D) was just as effective as a fully welded web stiffener
(Type A), while the web stiffener welded to the web only (Type E) had a slight
improvement (Figure 2.29). They stated that this was because the majority of the
strength increment was provided by tying together the rotational degrees of freedom
2-40
Literature Review
of the flanges, forcing the section to remain undistorted. This constraint is more
effective than a prevention of web distortion by welding the stiffener to the web only.
The effect of web stiffener thickness (5, 10, 15 and 20 mm) on the strength increment
was small. This is because the section properties of the stiffeners are less significant
than the nature of constraint they provided. Therefore they recommended that a plate
with a 5 mm (or larger) thickness can be used as a web stiffener. They also studied
the location of stiffeners at mid-span (1 stiffener), third points (2 stiffeners) and
quarter points (3 stiffeners) by using FEA. The use of three stiffeners or more did not
provide significantly greater strength than the strength obtained from two stiffeners.
They recommended using stiffeners at third points of the span which usually provide
an optimum compromise between the cost of fabrication and strength obtained.
5 mm Plate,
both sides, RHS, both
welded to web sides, welded 5 mm Plate,
and flanges to web and one side,
flanges welded to web
and flanges
5 mm Plate,
5 mm Plate, 5 mm Plate, one side,
both sides, both sides, welded to
welded to web welded to flanges only
only flanges only
In the experimental investigation, a screw fastened connection for the web stiffener
to the HFB flanges was developed as shown in Figure 2.30. The web stiffeners were
bent to fit the inclined flanges. These special stiffeners were compared with the
stiffener welded to the flanges and it was found that those stiffeners improve the
buckling capacity in a similar manner to those welded to flanges. This means
advantages in ease of installation and reduced cost. Furthermore, the screw fastening
did not introduce any geometrical imperfections and residual stresses. They
2-41
Literature Review
2-42
Literature Review
2-43
Literature Review
Member capacity (Mb) of LSBs undergoing lateral distortional buckling and lateral
torsional buckling can be calculated using appropriate clauses of AS/NZS 4600 (SA,
2005). Clause 3.3.3.3(b) of AS/NZS 4600 (SA, 2005) provides formulations to
determine Mb of beams undergoing lateral distortional buckling, which are presented
in the earlier section of this chapter (Equations 2.29 and 2.35a, b and c). Clause
3.3.3.2.1 of AS/NZS 4600 (SA, 2005) provides formulations to determine Mb of
beams undergoing lateral torsional buckling and is given by Equation 2.44 (This is
also given in Direct Strength Method as Equation 2.36).
1
For b 1.336: Mc = My 2 (2.44c)
b
2-44
Literature Review
The connection type of the stiffener to the beams also was found to have some
effects on the capacity increase. Kurniawan (2005) stated that the TEK screw
fasteners were somewhat little loose due to the action of tying together the rotational
degrees of flange freedom. In particular, this occurred in the experiments using
stiffener types B, C, D and E because the screws were fastened at the flange corners
thus subjected to pull out action (tension). A screw has a lower resistance to tension
than to shear action. The web stiffener and the threaded rods were fastened and
bolted to the inner flanges, respectively hence the screws and bolts were subjected to
a shear action. He suspected this might be one of the reasons why other stiffener
types gave lower strength improvement than the web stiffener and threaded rods.
Ultimately, Kurniawan (2005) concluded that none of the proposed stiffener types
could effectively improve the LiteSteel beam bending strength while the lateral
buckling strengths of Hollow Flange beams, with similar section characteristic to
LSBs, were improved significantly by simply introducing two web stiffeners (Avery
and Mahendran, 1997). This led Kurniawan (2005) to undertake an advanced
computer analysis of finite element modelling in order to verify the experimental
investigations and to further observe the behaviour of stiffened LSBs.
while the ideal model incorporated ideal constraints, nominal imperfections and a
uniform bending moment within the span. Residual stresses were not considered. The
analysis only considered stiffener Type A (web stiffener) as it was (except threaded
rods) found to be the best option. Threaded rods were not included due to the
complexity in modelling.
Elastic buckling analysis results showed that the elastic lateral distortional buckling
was almost eliminated when five or more web stiffeners were used. However, this is
not an efficient and practical solution.
The effects of section properties of web stiffeners are less significant than the nature
of the constraint they provided as suggested by Avery and Mahendran (1997).
Kurniawan (2005) also stated that there is a small strength increment (about 1%) for
the beam stiffened with 10 mm thickness of web stiffeners. A fully connected web
stiffener to the beam (ie. both flanges and web) was also found to have only a
marginal improvement compared to that connected to the flanges only, because the
majority improvement was provided by tying together the flanges.
Kurniawans (2005) ideal finite element model gave a contradictory result in which
lateral distortional buckling was effectively reduced by just using two web stiffeners.
For 3 m span LSBs, the use of two web stiffeners enhanced the elastic buckling
capacity by 24%, three times higher than that predicted by experimental finite
element model.
Kurniawan (2005) undertook finite element analysis of HFBs with quarter point
loading to check whether the loading conditions might affect the number of web
stiffeners as Avery and Mahendran (1997) incorporated overhang method loading
2-46
Literature Review
and concluded that two web stiffeners would adequately eliminate the lateral
distortional buckling. However, Kurniawan (2005) did not obtain strength increase
for HFB as much as Avery and Mahendran (1997) achieved. It should be noted that
Kurniawan (2005) considered only one HFB (30090HFB2.8) with 3 m span and
further investigation is required to conclude this fact.
Finally, Kurniawan (2005) used the same finite element model which was used by
Maharachchi and Mahendran (2005c). However, it was found that this finite element
model had some deficiencies such as inadequate lateral restraint, improper
application of end moments and loads and the direction of initial imperfections.
Experiments also found to be limited to only one LSB section. Therefore detailed
finite element analyses using both experimental model with quarter point loading and
ideal model of LSBs should be undertaken. Web plate stiffeners with 5 mm thickness
would be adequate. Further, large scale experimental analyses are also required to
validate the FEA results.
The finite element analysis process involves three major phases such as pre-
processing, solution and post processing. The purpose of pre-processing is to develop
an appropriate finite element mesh, assign suitable material properties, and apply
boundary conditions in the form of restraints and loads. Governing equations are
assembled into a matrix form and are solved numerically in the solution phase.
The assembly process depends not only on the type of analysis (e.g. static or
dynamic), but also on the models element types and properties, material properties
and boundary conditions. Post-processing begins with a thorough check for problems
2-47
Literature Review
that may have occurred during the solution stage. Most solvers provide a log file,
which should be searched for warning or error messages, and which will also provide
a quantitative measure of how well behaved the numerical procedures were during
solution.
Mahaarachchi and Mahendran (2005c) used shell elements (S4R5) to model the
LiteSteel beams. This element is a thin, shear flexible, isoparametric quadrilateral
shell with four nodes and five degrees of freedom per node, utilising reduced
integration and bilinear interpolation schemes. R3D4 rigid body elements and stiff
beam elements were used to create the ideal pinned member end restraints and
loading for both models. The R3D4 element is a rigid quadrilateral element with four
nodes and three translational degrees.
Using some convergence studies, Mahaarachchi and Mahendran (2005c) selected the
element sizes to be 5 mm in width and 10 mm in length in the longitudinal direction.
The support conditions were simulated to provide the required idealised simply
supported conditions where the following requirements were satisfied.
2-48
Literature Review
T1 T2 T3 R1 R2 R3
One end Yes No No No Yes Yes
Other end No No No No Yes Yes
Mid span No Yes Yes Yes No No
Table 2.2 shows the boundary conditions of the simply supported beam. The presence
of symmetry allowed them to model only half the span which would reduce the
analysing time. T and R represent the translation and rotation, respectively and the
subscripts (1, 2, and 3) represent the direction. Field Yes means that it is free to
move in that direction. Figure 2.37 illustrates the global axes selected to input the
boundary conditions for the analysis.
Y, 2 Y, 2
X, 1 Z, 3 L/2
Z, 3 X, 1
Figure 2.35 shows the load and boundary conditions of the ideal finite element model
developed by Mahaarachchi and Mahendran (2005c). An elastic strip width of 20
mm was included adjacent to the pinned end of all the sections in order to eliminate
the undesirable stress concentrations by concentrated loading. They used rigid
beam type Multiple Point Constraint (MPC) elements to spread the concentrated
moment evenly to the web and flanges at the shear centre of the cross section. Single
tie MPC and explicit MPCs with different degrees of freedom were applied to link
2-49
Literature Review
the rigid beam elements and the elastic strip. Explicit MPC (UX, RY and RZ)
denoted that the degrees of freedom of X-translation (Code UX), Y-rotation (Code
RY) and Z-rotation (Code RZ) are linked.
Y
X, 1
20 mm Elastic
Z, 3
Figure 2.35: Ideal Finite Element Model
(Mahaarachchi and Mahendran, 2005c)
Figures 2.36 (a) and (b) show the details of the experimental finite element model
developed by Mahaarachchi and Mahendran (2005c). Even though both overhang
method and quarter point loading were adopted in the experiments, only the quarter
point loading method was simulated. They used single point constraints and
concentrated nodal forces and applied loads as closely as possible. The experimental
specimens included a 70 mm width rigid plate at each support to prevent distortion
and twisting of the cross section. They modelled these stiffened plates as rigid body
using R3D4 elements. In ABAQUS (HKS, 2003) a rigid body is a collection of
nodes and elements whose motion is governed by the motion of a single node, known
as the rigid body reference node. Therefore they applied simply supported boundary
conditions to the node at the shear centre in order to provide an ideal pinned support.
They modelled the steel plate and bolts using rigid beam elements to simulate the
experimental set up.
Elastic modulus E of 200,000 MPa and Poissons ratio of 0.3 were adopted. They
used a nominal yield stress fy of 380 MPa for web and 450 MPa for flanges in the
ideal model while average measured yield stresses were used in the experimental
model.
2-50
Literature Review
Mahaarachchi and Mahendran (2005c) carried out both elastic buckling and
nonlinear static analyses. Elastic buckling analyses were used to obtain the
eigenvectors for the inclusion of geometric imperfections. Nonlinear static analysis
included the effects of large deformation, material yielding, imperfection and
residual stresses. Table 2.3 shows the parameters considered by Mahaarachchi and
Mahendran (2005c) in their nonlinear static finite element analysis. They explicitly
modelled the measured out-of-straightness and twist imperfections in the
experimental model while an assumed value of L/1000 was used in the ideal model.
They applied positive imperfections to the beams which would have given higher
lateral buckling strengths compared with negative imperfections (Pi et al., 1998).
Hence the negative imperfections should be considered in the future research of
LSBs.
Table 2.3 Nonlinear Analysis Parameters
(Mahaarachchi and Mahendran, 2005c)
Parameter Assigned Value
Max. no. of load increments Between 50 and 250
Initial increment size 0.001
Min. increment size 1x10-7
No. of integration points per shell 9
Automatic increment reduction Enabled
Large displacements Enabled
2-51
Literature Review
2-52
Literature Review
2-53
Literature Review
2-54
Literature Review
Flexural capacity of HFBs was investigated by Avery et al. (2000) using finite
element analyses. From their investigations, they discovered that the elastic lateral
distortional buckling moment and the ultimate capacities of HFBs can be accurately
predicted from their finite element analyses and therefore used them in the
development of design curves and suitable design procedures. Their study involved
two models, namely experimental and ideal models. The ABAQUS S4R5 shell
elements were employed to create the mesh and the R3D4 rigid body elements were
used to model the pinned end conditions. The loads and boundary conditions, as used
by Zhao et al. (1995) in the study of lateral buckling of cold-formed RHS beams,
were used in these models to provide idealized simply supported boundary
conditions and a uniform bending moment was applied. However, they have not been
able to eliminate the warping restraints due to the overhang in the experimental
models. The models incorporated all the significant effects that might influence the
ultimate capacity of HFBs, including material inelasticity, local buckling, member
instability, web distortion, residual stresses and initial geometric imperfections. They
explicitly modelled the measured imperfections in the experimental models, with
magnitudes as measured by Mahendran and Doan (1999). A nominal imperfection
magnitude of L/1000 was used in the ideal model. Residual stresses were modelled
using Doan and Mahendrans (1996) residual stress model which was based on the
measured residual stresses.
Avery and Mahendran (1997) studied the lateral distortional buckling behaviour of
hollow flange beams with web stiffeners using finite element analysis. The finite
2-55
Literature Review
element analysis program, MSC/NASTRAN was used for this study and the
quadrilateral shell elements (QUAD4) were used in this finite element modelling.
Triangular shell elements (TRIA3) were used to model the stiffeners. Only half of
the beam was modelled by making use of the symmetry of geometry and loading
conditions about the centre plane of the span. They developed both ideal and
experimental models. The ideal model consisted of typical idealised simply
supported boundary conditions and overhang method loading provided that the
cantilever span was fully restrained against lateral buckling. Also, no warping
restraints were allowed at the support. The experimental model was similar to the
ideal model which simulated the actual experimental set up. However, warping
restraints induced by overhang length were not eliminated.
They compared both the elastic buckling and non-linear analysis results from
experimental and ideal finite element model and concluded that the difference is
insignificant except short spans. For short span members, an interactive buckling
mode between the main span and the cantilever was observed in the experimental
model. Further, they used their ideal finite element model for parametric studies
using elastic buckling analysis.
Mahaarachchi and Mahendran (2005a, b, and e) carried out lateral buckling tests,
section capacity tests and tensile coupon tests on LSBs. They also measured the
initial imperfections and residual stresses for LSBs.
2-56
Literature Review
of AS 1391 (SA, 1991). Strain gauges were attached to the tensile test coupons in
order to measure the strain at fracture including uniform elongation. Some coupons
had strain gauges on both sides and those measurements were used to obtain the
Youngs modulus of elasticity, and the stress-strain curves. Percentage elongations
and strains were also determined by measuring the longitudinal displacement of
sixteen 5 mm transverse gauge lines over the constant width section of the tensile
coupon.
The test results showed that the yield stresses exceeded the nominal flange yield
stress of 450 MPa and the web nominal yield stress of 380 MPa due to the heavy
cold-working involved in the making of LSB sections. Table 2.4 shows the average
measured yield stresses and ultimate stresses of the flange and the web of several
LSBs.
Figure 2.42: Typical Stress-Strain Curves of the Base Steel used in LSB Sections
(Mahaarachchi and Mahendran, 2005e)
2-57
Literature Review
Figure 2.42 shows the typical stress-strain curves of steel used in LSB sections. The
lack of yield plateau in the stress-strain curves of flange specimens indicating the
higher level of cold-working in the flanges. The web and flange yield stresses varied
depending on the thickness and LSB section (Mahaarachchi and Mahendran, 2005e).
The most commonly used destructive method is the sectioning method which is
considered to be the most reliable method of measuring residual stresses.
Mahaarachchi and Mahendran (2005e) used the sectioning method using electrical
strain gauges to measure the longitudinal and transverse residual stresses of LSBs.
However, since early tests showed the absence of transverse residual stresses, most
of the tests included only the longitudinal residual stress measurements. Electrical
strain gauges were used on both the inside and outside surfaces of the flanges. Due to
the difficulties of attaching the strain gauges on the inside corner, the measured
residual strains were extrapolated to the corners.
Figure 2.43 show some pictures of sectioning carried out by Mahaarachchi and
Mahendran (2005e) while the idealised residual stress distribution models for LSBs
are given in earlier section as Figure 2.37.
2-58
Literature Review
Travelator
Table Laser Sensor
2-59
Literature Review
Tests results showed that the first yield of LSB sections occurred at about 0.75-0.82
of the theoretical first yield moment. Most of the specimens exhibited a flange local
buckling failure. Mahaarachchi and Mahendran (2005b) compared the test results
with predicted section capacities using AS 4100 (SA, 1998) and AS/NZS 4600 (SA,
1996). They found that all the experimental results exceeded the section moment
capacities predicted by both AS 4100 and AS/NZS 4600. AS 4100 predictions were
13% lower than the experimental results while AS/NZS 4600 predictions were lower
than the tested failure moments by 18%.
T-Shape Stiffeners
Plate Stiffeners
2-60
Literature Review
Mahaarachchi and Mahendran (2005a)s support system was based on the support
systems used by Zhao et al. (1995), Put et al. (1999) and Mahendran and Doan
(1999) with some modifications. The support conditions provided restraint against
in-plane and out-of-plane deflections and longitudinal twisting, and allowed major
and minor axis rotations. One of the support system was designed as a roller. Figure
2.47 illustrates the support system used.
Mahaarachchi and Mahendran (2005a) reviewed the loading methods, the over-hang
loading method used by Zhao et al. (1995) and Mahendran and Doan (1999) and the
quarter point loading method used by Put et al. (1999), and proposed these two
2-61
Literature Review
methods with improved modifications. In the overhang method, the loads were
applied to the test beam at a distance of 1 m from the support in the upward direction
and this provided a uniform bending moment within the entire span whereas the
quarter point loading method provided a uniform bending moment only between the
points of load applications. Even though the overhang loading method is to be
preferred as it provides a uniform moment within the entire span, the overhang
component of the test beam provides a warping restraint to the test beam.
Mahaarachchi and Mahendrans (2005a) test results showed that the overhang
loading method gave higher test capacity results by about 12% due to the effect of
warping restraints and hence they used the quarter point loading method. The loading
system was designed to prevent the possible restraints to the displacement and
rotations of the test beam using a special wheel system (Figure 2.48 (a)). Loads were
applied through the shear centre of the test beam to avoid any torsional and load
height effects to the test beam (Figure 2.48 (b)). The test results showed that the
lateral distortional buckling mode was most severe for intermediate spans. The
support conditions provided by Mahaarachchi and Mahendran (2005a) are likely to
allow local flange twist during the application of loading although the ideal simply
supported boundary conditions do not allow such twisting. Further experiments are
needed to identify the occurrence of such local flange twist. It is necessary to
restraint this twisting if it occurs during the experiments.
Loading at Shear
Wheels Centre
Smooth Tracks
2-62
Literature Review
(a) Stiffener Type A (b) Stiffener Type B (c) Stiffener Type C (d) Stiffener Type D
Additional
Thin Plate
2-63
Literature Review
The experimental set-up used in this study consisted of a simply supported boundary
condition. Two load-controlled hydraulic jacks, located on the overhangs were used
to apply the loads and the web stiffeners at the support prevented any local bearing
failure of the bottom flange.
Mahendran and Doan (1999) conducted lateral buckling tests on HFBs. A purpose-
built test rig was used in this study to obtain the bending capacity of HFBs under
uniform bending moment. The test rig included a support system and a loading
system, which were attached to an external frame consisting of a main girder and
four columns as shown in the schematic diagram in Figure 2.52. The support system
was designed to ensure that the test beam had simply supported end conditions,
whereas the loading system was designed in such a way that no restraints were
induced as the beam deformed during loading.
2-64
Literature Review
Two vertical loads were applied at the end of two overhangs to produce a uniform
bending moment within the span of the specimen. The simply supported end
conditions of the span were simulated in a similar way to that of Zhao et al. (1995)
used for the Rectangular Hollow Sections (RHS) but were modified to suit the HFBs.
However, warping restraints induced by overhang of the beam could not be
eliminated in this system. The loading system included two hydraulic jacks instead of
gravity loading system used by Zhao et al. (1995). They were operated under load
control to ensure that the same load was applied and hence identical bending
moments were provided at the ends of the single span. Loading was applied at the
shear centre.
2-65
Literature Review
Zhao et al. (1995) stated that the support system used in their study (see Figure 2.53
(b)) was similar to that used by Trahair (1969) in his elastic lateral buckling tests of
aluminium I-beams and later by Papangelis and Trahair (1987) in their flexural
torsional buckling tests of arches.
Warping displacements were not prevented except by the adjacent cantilever lengths.
The restraint to warping provided by the cantilever lengths can be considered
minimal because significant warping does not occur in tubular sections, compared
with I-sections. Hollow flange steel beams such as LSBs and HFBs considered in
this research are open steel beams and hence they are expected to induce significant
amount of warping displacements compared to RHS beams. The loading system
included gravity loads being applied by suspending lead blocks on a platform which
was supported by hangers. Zhao et al. (1995) cited that the loading system used in
their study was similar to those used by Cherry (1960) and Hancock (1975), where
the vertical load applied acted through the centroid of the section and no restraints
were applied against out-of-plane movement at the loading point.
2-66
Literature Review
(a) Test Arrangement for C- and (b) Support System for Z- Section
Z-Section Beams
2-67
Literature Review
imperfections and residual stresses of cold-formed steel sections are different from
those of hot-rolled steel members mainly due to the manufacturing process which
involves cold-forming. The mechanical properties of cold-formed steel members are
affected by cold forming specifically in the regions of bends. The resulting changes
in material properties must be included in the design of cold-formed steel members.
Hollow flange steel beams such as HFBs and LSBs are some innovative cold-formed
steel members used in the building industry due to their superior structural and
construction efficiencies. However, the manufacturing of HFBs was discontinued in
1997 because of its relatively expensive electric welding process and lack of
equipment facilities.
Past research identified that hollow flange steel beams suffer from lateral distortional
buckling when used as flexural members. Lateral distortional buckling is
characterised by simultaneous lateral displacement, twist and web distortion which is
severe for intermediate spans and reduces the member moment capacity. AS 4100
(SA, 1998) does not provide any design procedure to determine the lateral
distortional buckling moment capacity of hollow flange steel beams while AS/NZS
4600 (SA, 1996) provides some design formulae. The equation given in AS 4100
(SA, 1998) to predict the lateral torsional buckling moment capacity was modified by
Pi and Trahair (1997) to predict the lateral distortional buckling moment capacity of
HFBs. However, it was not validated by experiments. Avery et al. (2000) then
modified Trahairs (1997) equation to predict the lateral distortional buckling
moment capacity of HFBs by applying suitable coefficients for each thickness of
HFBs. But the applicability of these equations to predict the lateral distortional
buckling moment of LSBs is questionable.
2-68
Literature Review
and the type of loading in this study found to have some unacceptable stress
distribution on the LSB cross section at failure. Therefore it is important to develop
an accurate ideal finite element model to predict the lateral distortional buckling
moment capacity of LSBs. Also, the existing design rules do not account for the
section geometry of LSBs and can only be used for the existing 13 LSB sections.
Therefore it is important to investigate the effect of section geometry on the lateral
distortional buckling moments of hollow flange steel beams and to develop a
comprehensive design procedure for all types of hollow flange steel beams (HFBs
and LSBs).
Mahaarachchi and Mahendran (2005a) carried out large scale experiments on LSBs
with quarter point and overhang type loading and found that the overhang loading
method increased the ultimate moments of LSBs due to warping. Therefore, they
used quarter point loading method and incorporated a moment modification factor of
1.09 to account for non-uniform bending moment along the span. However, this was
taken from AS 4100 (SA, 1998) which was developed for hot-rolled steel members
and the applicability of this to cold-formed steel members is questionable. Recent
research (Kurniawan and Mahendran, 2009b) on the effects of moment gradient on
the lateral distortional buckling capacity of LSBs revealed that the moment
distribution factor is closer to 1.0 for quarter point loading. Hence, this results can be
used in the experimental investigation.
Mahaarahchi Mahendran (2005b) carried out section moment capacity tests on LSBs
and their results revealed that some of the LSBs have inelastic reserve bending
capacity beyond their first yield moment. However, they did not develop finite
element models nor developed suitable design rules for LSBs to predict their section
moment capacities with inelastic reserve bending capacities.
Web stiffeners were used with both HFBs and LSBs to eliminate the detrimental
effect of lateral distortional buckling. The results showed that the use of two web
2-69
Literature Review
stiffeners at third points of span were able to eliminate the lateral distortional
buckling of HFBs (Avery and Mahendran, 1997) while LSBs (Kurniawan, 2005)
required large number of web stiffeners. Overhang type loading method was used in
the experimental investigation of HFBs while quarter point loading was adopted for
LSBs. These different loading conditions and geometries of HFBs (doubly
symmetric with triangular hollow flanges) and LSBs (singly symmetric with
rectangular hollow flanges) would be some of the factors in determining the required
number of web stiffeners. It is important to investigate the effects of web stiffeners
on the lateral distortional buckling of LSBs.
2-70
Material Properties, Residual Stresses and Imperfections of LSBs
CHAPTER 3
3.1 Introduction
The unique cold-forming and dual electric resistance welding process of LiteSteel
Beam (LSB) sections introduce considerable differences to the stress-strain curves,
residual stresses and initial geometric imperfections compared to the conventional
hot-rolled and cold-formed steel sections. Although Mahaarachchi and Mahendran
(2005e) measured the mechanical properties, residual stresses and initial geometric
imperfections of LSB sections the quality of the manufacturing process in relation to
cold-forming and electric resistance welding has been improved over the last three
years. Currently the LSB sections are made from a single strip of G60 galvanized
steel while the earlier LSB sections were manufactured from a single steel strip of TF
380 coil. The coating system has also been changed from EnviroKote water-based
primer paint protective coating system to AZ+ alloy coating system. Therefore it is
important to determine the accurate mechanical properties of these LSBs that can be
used in the finite element analyses of LSBs. The mechanical properties of LSBs were
measured using tensile coupon tests.
The residual stresses due to combined cold-forming and electrical resistance welding
process can have a significant effect on the behaviour and strength of LSB sections.
Unlike other cold-formed steel sections the LSB sections have both flexural and
membrane residual stresses due to cold-forming and welding processes, respectively.
Maharaachchi and Mahendran (2005e) measured the residual stresses of LSBs and
developed approximate residual stress models for both flexural and membrane
residual stresses. However, it was found that the sum of the membrane residual
forces was not equal to zero in their model. Further, their model had more
compressive stresses at the welding point despite the common belief that welding
creates tensile membrane residual stresses. In order to investigate these issues a
residual stress test was carried out for the web element of a 150x45x1.6 LSB section.
3-1
Material Properties, Residual Stresses and Imperfections of LSBs
This chapter presents the details of tensile coupon tests, residual stress and geometric
imperfection measurement tests of LSBs, and the results.
The structural behaviour of LSBs depends on the mechanical properties of the steel
used, which are required in the finite element analyses of LSBs. Therefore tensile
tests of steel coupons taken from LSBs were conducted to determine the required
important mechanical properties based on the procedure specified in the Australian
Standard AS 1391 (SA, 2007).
Fifty four coupons were fabricated in the workshop at the Queensland University of
Technology. They were taken from the LSB sections used in the section and member
moment capacity tests. The coupons were cut in the longitudinal direction from
various locations of the beam, namely the outside flange, inside flange and web.
The coupon size and shape are important variables that can affect the behaviour of
tensile coupons. Previous research of G550 steel (Rogers and Hancock, 1997)
consisted of tensile testing of coupons with large grips. They have shown that uneven
gripping of these coupons might have occurred due to misalignment and improper
lubrication of the test machine grips, i.e. only one side of the grip portion of the
coupon was secured, resulting in either a slip of the test specimen as the load was
increased or eccentric tensile forces. This can cause a reduction in the true capacity
of the test specimens and influence their stress-strain behaviour. It is necessary to
3-2
Material Properties, Residual Stresses and Imperfections of LSBs
design tensile coupons which limit the possibility of an eccentric connection between
the test machine grips and the test coupon. Therefore tensile coupon dimensions were
chosen in accordance with the recommendations of AS 1391 (SA, 2007), and are
shown in Figure 3.1. It is considered that the dimensions selected here do not unduly
affect the stress-strain behaviour of tensile coupons.
The coupons were immersed in a 1:1 diluted hydrochloric acid basin for about 45
minutes to remove the coating. The surface of the tensile coupons was then cleaned
with fine grade emery paper and then with an acetone solution. A 2 mm strain gauge
was attached to the tensile coupons using CN Cyanoacrylate adhesive and a 25 mm
gauge length extensometer was used in the middle of the specimen to measure the
strain more accurately.
15 mm
25 mm
R20
40 mm 40 mm
120 mm
The tensile coupon tests were carried out using the Tinius Olsen testing machine in
the QUT Structural Testing Laboratory. A displacement control method was used to
3-3
Material Properties, Residual Stresses and Imperfections of LSBs
apply the tension load at a rate of 0.5 mm per minute. Special jaw systems that have
the ability to translate and rotate in-plane were attached to the test machine cross-
heads. This system of jaws minimised the presence of any end eccentricities due to
any misalignment of the grips and hence eliminated specimen twisting and bending
that usually occurs when the grips are tightened (see Figure 3.2). The tension force
was applied until a fracture occurred in the specimen. The measurements of load,
extensometer and strain gauge were recorded automatically at a fast rate (every
second) using a data acquisition system attached to a personal computer.
(b) Extensometer
Test results derived based on the measured thicknesses are summarised in Table 3.1
while the typical stress-strain curves for the web and flange elements are given in
Figure 3.3. Test results show that the measured yield stresses exceed the nominal
flange yield stress of 450 MPa and the nominal web yield stress of 380 MPa. The
web and flange yield stresses varied depending on the thickness and LSB section.
Further details of tensile coupon tests including stress-strain curves can be found in
Appendix A.1.
3-4
Material Properties, Residual Stresses and Imperfections of LSBs
600.0
500.0
400.0
Stress (N/mm )
2
Inside Flange
200.0
100.0
0.0
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00 11.00
% Strain
Dimensions
(mm) Youngs
Test fy fu
Location Modulus fu/fy
Specimen Thick (MPa) (MPa)
Width (GPa)
-ness
Outside Flange 2.22 12.03 557.7 592.9 193 1.06
300x60x2.0
Inside Flange 2.02 12.03 496.3 534.2 210 1.08
LSB
Web 1.98 12.02 447.1 524.2 195 1.17
Outside Flange 2.90 12.02 552.2 592.8 188 1.07
250x75x2.5
Inside Flange 2.60 12.02 502.2 536.4 197 1.07
LSB
Web 2.54 12.03 446.0 515.4 208 1.16
Outside Flange 1.79 12.07 536.9 587.1 213 1.09
200x45x1.6
Inside Flange 1.66 12.01 491.3 542.6 233 1.10
LSB
Web 1.61 12.02 456.6 537.2 230 1.18
Outside Flange 2.22 12.01 537.6 582.3 213 1.08
150x45x2.0
Inside Flange 2.02 12.02 491.8 532.4 213 1.08
LSB
Web 1.97 12.01 437.1 516.4 208 1.18
Outside Flange 1.77 11.97 557.8 604.4 210 1.08
150x45x1.6
Inside Flange 1.63 11.97 487.5 549.2 205 1.13
LSB
Web 1.58 11.96 455.1 539.8 220 1.19
Outside Flange 2.16 11.88 544.1 582.2 195 1.07
125x45x2.0
Inside Flange 1.97 11.90 493.4 539.3 206 1.09
LSB
Web 1.94 11.89 444.4 532.3 227 1.20
fy Yield Stress, fu Ultimate Stress
3-5
Material Properties, Residual Stresses and Imperfections of LSBs
Table 3.1 presents the average values from three stress-strain curves for each
location. The ultimate stress (fu) was the maximum stress obtained from the stress-
strain curve while the yield stress (fy) was calculated based on the 0.2% proof stress.
These values are within 3% of the results from the manufacturers (OATM). The
measured yield and ultimate stresses are compared with the results from
Mahaarachchi and Mahendran (2005e) in Table 3.2. It is clearly seen that most of the
stresses for the new LSBs are higher than that of Mahaarachchi and Mahendran
(2005e). Although some of the results show about 20% increment when compared to
Mahaarachchi and Mahendrans (2005e) results, on average the yield and ultimate
stresses were increased by about 8% and 3%, respectively. It should be noted that the
results for the new LSBs should be used in the numerical modelling of LSBs in this
research.
Mahaarachchi and
Current Current/MM
Test Mahendran (2005e)
Location
Specimen fy fu
f (MPa) fu (Mpa) fy fu
(MPa) (MPa) y
Outside Flange 557.7 592.9 568 635 0.98 0.93
300x60x2.0
Inside Flange 496.3 534.2 492 557 1.01 0.96
LSB
Web 447.1 524.2 452 537 0.99 0.98
Outside Flange 552.2 592.8 525 582 1.05 1.02
250x75x2.5
Inside Flange 502.2 536.4 478 547 1.05 0.98
LSB
Web 446.0 515.4 420 531 1.06 0.97
Outside Flange 536.9 587.1 478 530 1.12 1.11
200x45x1.6
Inside Flange 491.3 542.6 442 506 1.11 1.07
LSB
Web 456.6 537.2 381 494 1.20 1.09
Outside Flange 537.6 582.3 498 547 1.08 1.06
150x45x2.0
Inside Flange 491.8 532.4 451 508 1.09 1.05
LSB
Web 437.1 516.4 373 507 1.17 1.02
Outside Flange 557.8 604.4 540 576 1.03 1.05
150x45x1.6
Inside Flange 487.5 549.2 483 519 1.01 1.06
LSB
Web 455.1 539.8 430 523 1.06 1.03
Outside Flange 544.1 582.2 503 547 1.08 1.06
125x45x2.0
Inside Flange 493.4 539.3 455 508 1.08 1.06
LSB
Web 444.4 532.3 377 496 1.18 1.07
Mean 1.08 1.03
fy Yield Stress, fu Ultimate Stress, MM Mahaarachchi and Mahendran (2005e)
3-6
Material Properties, Residual Stresses and Imperfections of LSBs
Residual stresses are important in the design of LSBs because the combined cold
roll-forming and dual electric resistance welding process used in the manufacturing
of LSBs is likely to cause higher initial residual stresses. These residual stresses
could reduce the bending moment capacities of LSBs. Therefore the use of a residual
stress model with accurate distribution and magnitudes of residual stresses is
important in any advanced numerical analysis of LSBs. Generally there are three
types of experimental methods used in residual stress measurements, namely,
destructive, semi-destructive and non-destructive methods. The most commonly used
destructive method is the sectioning method, which was used by Mahaarachchi and
Mahendran (2005e). They developed separate models for both flexural and
membrane residual stresses. However, the improvements to the LSB manufacturing
process over the last three years are likely to have modified the residual stress
distributions in LSBs. Also the membrane residual stress distribution developed by
Mahaarachchi and Mahendran (2005e) revealed some unbalanced forces in the LSB
cross-section. Therefore a residual stress test was carried out for the web element of a
150x45x1.6 LSB section.
6.5 mm
2
Inner 22 mm
Surface 1 Outer Surface
31 mm
0 Neutral Axis
28 mm
3
24 mm
5 mm4
3-7
Material Properties, Residual Stresses and Imperfections of LSBs
The strain gauge locations are numbered as 0, 1, 2, 3 and 4 as shown in Figure 3.4.
The strain gauges were covered by a paper tape to avoid any damage during the
cutting process. They were connected to a data acquisition system which enabled the
recording of strains after each sectioning.
(a) First Transverse Cut (b) Specimen Ready for 2nd Transverse Cut
(c) Second Transverse Cut (d) Specimen Ready for Longitudinal Cut
(e) Longitudinal Cut (f) Small Pieces of LSB Specimen at the end
3-8
Material Properties, Residual Stresses and Imperfections of LSBs
The specimen was first cut across the cross-section on either side of the installed
strain gauges using a band saw as shown in Figure 3.5 (a). A 100 mm specimen was
obtained after the first transverse cut on either side of strain gauges. The second
transverse cut was then carried out as shown in Figure 3.5 (c). Figure 3.5 (d) shows
the 50 mm specimen after two transverse cuts. The longitudinal sectioning was then
continued as shown in Figure 3.5 (e) to separate the strain gauges. Final strain gauge
readings were recorded after one hour and 24 hours of the completion of the
sectioning to allow for temperature effects. Figure 3.5 (f) shows the small pieces of
the LSB specimen at the end of sectioning.
3.3.2 Results
All the strains measured from the sectioning method are the strains released due to
sectioning. Final released strains were calculated from the difference between the
strain gauge reading before cutting and those obtained after 24 hours following
cutting. Figure 3.6 shows the measured released strains on both the outer and inner
surfaces of the web element. A positive value indicates tensile strain whereas a
negative value indicates compressive strain.
600
400
Strain (Micro Strain)
200
0
-60 -40 -20 0 20 40 60
-200
-600
Distance from the Web Neutral Axis (mm)
3-9
Material Properties, Residual Stresses and Imperfections of LSBs
These released strains were converted to the released stresses using a Youngs
modulus value of 200,000 MPa. It can be assumed that all the residual stresses are
released during the sectioning process used in this research. Therefore the residual
stresses are equal to the calculated released stresses in magnitude but with an
opposite sign, i.e, released tensile stresses are equal to compressive residual stresses
and vice versa.
The residual stresses were then used to calculate the membrane and flexural residual
stresses using the following formulae.
Figure 3.7 shows the calculated membrane and flexural residual stresses. The stresses
at the corners were extrapolated and further simplifications are made to determine
the membrane residual stress distribution shown in Figure 3.8.
80
60
40
20
Stress (MPa)
0
-60 -40 -20 0 20 40 60
-20
-40
-60
Membrane Stress
Flexural Stress
-80
Distance from the Web Neutral Axis (mm)
Figure 3.7: Measured Stresses along the Web Element of a 150x45x1.6 LSB
3-10
Material Properties, Residual Stresses and Imperfections of LSBs
The residual stresses are expressed as a ratio of the virgin plates yield stress of 380
MPa. The membrane residual stresses are then compared with those of Mahaarachchi
and Mahendran (2005e) in Figure 3.8.
-0.23fy
0.03fy -0.15fy
-0.41fy
0.11fy 0.03fy
0.12fy
0.60fy
0.11fy
0.03fy
-0.41fy
0.03fy
-0.15fy
-0.23fy
The new membrane residual stress along the web element shows a significant
reduction compared to those measured by Mahaarachchi and Mahendran (2005e).
However, the reduced values may not be used in numerical analyses as they were
derived from only one test. But the results confirm that there are compressive
membrane residual stresses at the web-flange junction, which is one of the main
objectives of this test.
1.07fy
0.24fy
0.24fy 0.41fy
0.38fy 0.8fy
0.15fy
0.24fy
0.38fy 0.8fy
0.24fy 0.41fy
0.24fy
1.07fy
3-11
Material Properties, Residual Stresses and Imperfections of LSBs
Figure 3.9 shows the comparison of flexural residual stresses between the current
research and that of Mahaarachchi and Mahendran (2005e). The residual stress along
the web element from this research reveals a noticeable reduction compared to those
measured by Mahaarachchi and Mahendran (2005e). Therefore it can be concluded
that both membrane and flexural residual stresses have been reduced in the new
LSBs due to possible improvements to the manufacturing process of LSBs.
However, the reduced value may not be used in numerical modelling as it was
obtained from only one test.
In order to determine the need to use accurate residual stresses in the numerical
analyses of LSBs, the individual effects of flexural and membrane residual stresses
on the flexural moment capacities of LSBs were thoroughly investigated by using
numerical analyses of LSBs. It was found that the flexural residual stresses play a
significant role in the reduction of moment capacities while the membrane residual
stresses have lesser effects. Further, the web membrane residual stresses were found
to have only a limited effect when compared to the flange membrane residual
stresses on the moment capacities of LSBs. More details can be found in Section
6.3.2 of Chapter 6 and Seo et al. (2008).
Based on the findings from the residual stress test and the numerical analyses of Seo
et al. (2008), it was decided not to use the reduced residual stress values obtained
from this research. Instead the flexural residual stress distribution of Mahaarachchi
and Mahendran (2005e) was retained while their membrane residual stress
distribution was modified for the web element and the left flange. The web
membrane residual stress values were changed from 0.60fy and -0.41fy to 0.50fy and -
0.50fy at the middle of the web and web-flange junction, respectively. The left flange
membrane residual stress value was slightly modified from 0.23fy to 0.2567fy for
150x45x1.6 LSB. Figure 3.10 shows the new membrane residual stress distribution
recommended for use in the numerical analyses of 150x45x1.6 LSB.
3-12
Material Properties, Residual Stresses and Imperfections of LSBs
-0.2567fy
0.03fy
-0.50fy
0.03fy
0.11fy
0.50fy
0.11fy
0.03fy
-0.50fy
0.03fy
-0.2567fy
Different values are used for other different LSB sections in order to make the net
membrane force in the cross-section to zero. Table 3.3 presents the proposed
membrane residual stress values for all the LSBs. It should be noted that no change
has been made to the flexural residual stresses proposed by Mahaarachchi and
Mahendran (2005e).
3-13
Material Properties, Residual Stresses and Imperfections of LSBs
The magnitudes of initial geometric imperfections were measured for some test
specimens using the imperfection measuring equipment specially designed and built
at the Queensland University of Technology (see Figure 3.11). The imperfection
measuring equipment included a levelled table with guided rails with an accuracy of
0.01 mm, a laser sensor, travelator to move the sensor and a data logger.
Measurements were taken along three lines in the longitudinal direction of the
specimen at 100 mm intervals. They were made to determine the initial crookedness
(lack of straightness) and twist along the web and both flanges of the LSBs.
Travelator
Table
3-14
Material Properties, Residual Stresses and Imperfections of LSBs
Typical imperfections measured along the length of a 4 m long 200x45x1.6 LSB are
shown in Figure 3.12. Global imperfections were measured at different locations of
the web and flange of the specimen and the average of those measurements are
presented in Figure 3.12 (a). Local imperfections were measured along the cross-
section of the web at quarter points and the mid span of the beam and are presented
in Figure 3.12 (b).
2.0
1.5
1.0
Imperfections (mm)
0.5
0.0
0 500 1000 1500 2000 2500 3000 3500
-0.5
Along the Web
-1.0
Span Length (mm)
1.2
1.0
0.8
0.6
Imperfections (mm)
0.4
Quarter Point 1
Mid Span
0.2
Quarter Point 2
0.0
-100 -50 0 50 100
-0.2
-0.4
Distance from Neutral Axis (mm)
3-15
Material Properties, Residual Stresses and Imperfections of LSBs
Measured imperfections of other LSB sections are presented in Appendix A.2. The
results show that the measured local plate imperfections are within the
manufacturers fabrication tolerance limit of depth or width/150 while the overall
member imperfections are less than the recommended limit of span/1000 (SA, 1998).
This demonstrates that the unique manufacturing process of LSB does not lead to
geometric imperfections that exceed the currently accepted fabrication tolerances.
Mahaarachchi and Mahendran (2005e) also measured these imperfections for a large
number of LSBs and confirmed this observation. The measured geometric
imperfection values and distribution can be used in the numerical modelling of LSBs
to improve its accuracy of simulating the structural behaviour of LSBs.
3.5 Conclusions
This chapter has presented the details of a series of tests to determine the mechanical
properties, residual stresses and initial geometric imperfections of the current LSB
sections manufactured using a unique cold-formed and dual electric resistance
welding process. A series of tensile tests was carried out based on coupons taken
from the web, outside flange and inside flange elements of six LSBs, and the
mechanical properties were compared with those of Mahaarachchi and Mahendran
(2005e). It was found that both the yield stress and ultimate tensile strength have
increased due to the improved manufacturing process over the last three years.
Residual stresses were measured along the web element of a 150x45x1.6 LSB
section, and it was found that the improved manufacturing process has reduced the
level of residual stresses in LSBs. The results also showed that there are compressive
membrane forces at the web-flange junction. The membrane residual stress
distribution was slightly modified so that the net membrane force in the LSB cross-
section is zero. Initial geometric imperfections of some LSB members showed that
the LSB imperfections are within the currently accepted fabrication tolerances.
3-16
Lateral Buckling Tests of LSB Sections
CHAPTER 4
4.0 LATERAL BUCKLING TESTS OF LSB SECTIONS
4.1 Introduction
Mahaarachchi and Mahendran (2005a) did not consider short span LSBs in their
lateral bucking tests. In their tests, both the top and bottom flanges of LSBs were
kept free at the supports which would have reduced the moment capacities due to the
occurrence of flange twist despite the fact that the idealised simply supported
boundary conditions do not allow such twisting. This effect can be minimised or
eliminated by plotting these results on the non-dimensional capacity curve using the
appropriate elastic lateral distortional buckling moments from numerical analyses.
Considering the new developments in LSB sections and their manufacturing and the
limitations in Mahaarachchi and Mahendrans (2005a) series of tests as described
above, it was decided to undertake another series of lateral buckling test. In this
study a total of 12 tests was undertaken using the currently available LSBs. Some of
the tests considered the same spans used by Mahaarachchi and Mahendran (2005a)
for comparison purposes while other tests considered shorter spans subject to lateral
distortional buckling to obtain more points on the non-dimensional member moment
4-1
Lateral Buckling Tests of LSB Sections
capacity curve. The LSBs with experimental failure moments below the AS/NZS
4600 (SA, 2005) member moment capacity curve were also chosen in this
experimental study (see Figure 4.1). The experimental results will be used to validate
the finite element analyses and to investigate the effects of section geometry on the
lateral distortional buckling moment capacities of LSBs, which is one of the major
objectives of this research. This chapter presents the details of this experimental
study and the results.
1.20
AS/NZS 4600 (SA, 2005)
300x75x3.0 LSB
300x75x2.5 LSB
1.00 300x60x2.0 LSB
Dimensionless Moment Capacity, M u / My
250x75x3.0 LSB
250x75x2.5 LSB
0.80 250x60x2.0 LSB
200x60x2.5 LSB
200x60x2.0 LSB
200x45x1.6 LSB
0.60 150x45x2.0 LSB
150x45x1.6 LSB
125x45x2.0 LSB
0.40 A 125x45x1.6 LSB
B
C D
0.20
E F G
H
0.00
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2
Slendreness, d
Six LSBs with different section geometries were considered in the experimental
study with the beam span ranging from 1200 mm to 4000 mm. The quarter point
loading method was considered for all the tests. Figure 4.1 shows the dimensionless
plot of Mahaarachchi and Mahendrans (2005a) experimental failure moments with
the existing member capacity design curve from AS/NZS 4600 (SA, 2005) based on
measured LSB dimensions and thicknesses. They conducted some tests using the
overhang loading method, but used the quarter point loading for most of the tests.
This figure shows only the results from the quarter point loading tests of the
currently available 13 LSB sections although they tested some LSB sections, which
are not currently available.
4-2
Lateral Buckling Tests of LSB Sections
In Figure 4.1, points A and B correspond to 150x45x2.0 LSB and 150x45x1.6 LSB
with a span of 1.2 m, respectively. However, it was suspected that the shear buckling
between the support and the loading point might have reduced the capacity of these
beams even though Mahaarachchi and Mahendran (2005a) reported a lateral
distortional buckling failure. Current research considered a 150x45x2.0 LSB with 2
m span, and a 150x45x1.6 LSB with 1.8 m span, which are not likely to fail by shear
buckling based on preliminary finite element analyses. However, a 150x45x1.6 LSB
with 1.2 m span was also considered to check whether the beam fails by shear
buckling. Points C and D correspond to 250x75x2.5 LSB, points E and G correspond
to 200x45x1.6 LSB and points F and H correspond to 300x60x2.0 LSB with spans of
3 m and 4 m, respectively. Same tests were repeated with modified boundary
conditions as these failure moments are below the AS/NZS 4600 (SA, 2005) member
capacity curve. Full details of the test beam specimens are presented in Table 4.1
whereas Figure 4.2 illustrates the typical LSB test specimens.
riw ro
ro
Beam depth (d), flange width (bf) and the flange depth (df) were measured for each
test beam before testing while the thicknesses of LSB plate elements were carefully
measured using a micrometer. Accurate thickness of each plate element is important
to obtain the elastic lateral distortional buckling moment as a small change in
thickness would cause a significant change in the buckling capacities. The
4-3
Lateral Buckling Tests of LSB Sections
thicknesses measured in the tensile coupon tests are likely to be more accurate as the
thickness of the coating in the test beam would affect the actual measurements. Table
4.1 presents the details of test specimens including the measured LSB dimensions
and the base metal thicknesses from tensile coupon tests. It can be seen that the outer
flange thickness (tof) is larger than the nominal thickness while the web thickness (tw)
is smaller than the nominal thickness due to the variation in cold-working within the
section. The measurements of small corners were not taken as it was difficult to
measure them, and it was decided to use the nominal corner dimensions provided by
the manufacturers, i.e. outer radius ro is equal to two times the thickness (2t) and the
inner radius riw is equal to 3 mm.
Thickness, t Flange
LSB Section Span d df bf Twist
No tof tif tw
Restraint
1 250x75x2.5LSB 3500 2.90 2.60 2.54 251.0 75.0 25.5 No
2 300x60x2.0LSB 4000 2.22 2.02 1.98 302.0 60.0 20.5 No
3 200x45x1.6LSB 4000 1.79 1.66 1.61 201.0 45.0 14.8 No
4 300x60x2.0LSB 3000 2.22 2.02 1.98 299.0 60.0 20.0 No
5 200x45x1.6LSB 3000 1.79 1.66 1.61 201.0 45.0 14.9 Yes
6 150x45x1.6LSB 3000 1.75 1.62 1.58 150.0 46.0 15.1 Yes
7 150x45x2.0LSB 3000 2.22 2.05 1.96 150.0 45.0 15.0 Yes
8 200x45x1.6LSB 2000 1.79 1.66 1.61 200.0 45.0 14.9 Yes
9 150x45x2.0LSB 2000 2.22 2.02 1.97 151.0 45.0 14.9 Yes
10 150x45x1.6LSB 1800 1.77 1.63 1.58 150.0 46.0 14.6 Yes
11 125x45x2.0LSB 1200 2.16 1.97 1.94 125.0 45.0 14.6 Yes
12 150x45x1.6LSB 1200 1.77 1.63 1.58 150.5 45.5 14.6 Yes
Note: All dimensions are in mm.
tof outer flange thickness, tif inner flange thickness, tw web thickness.
4-4
Lateral Buckling Tests of LSB Sections
method, loads are applied at quarter points within the span and a uniform bending
moment will be produced between the loading points (see Figure 4.3 (b)).
P P
Overhang Span, L
Overhang
Bending Moment
Bending Moment
Zhao et al. (1995) and Mahendran and Doan (1999) carried out lateral buckling tests
of cold-formed RHS beams and hollow flange beams, respectively, using the
overhang loading method. Put et al. (1998) used the quarter point loading method in
their investigation. Mahaarachchi and Mahendran (2005a) conducted lateral buckling
tests of LSBs using both the overhang and the quarter point loading methods in their
preliminary investigation in order to determine the most suitable method. They found
that the overhang loading system could cause undesirable warping effect due to the
overhang component of the test beam. Therefore they used the quarter point loading
system in most of their tests and considered a moment modification factor of 1.09 to
allow for the non-uniformity of bending moment within the span. Pokharel and
4-5
Lateral Buckling Tests of LSB Sections
Mahendran (2006) also used the quarter point loading method to investigate the
bending moment capacities of LSBs with holes. Kurniawan and Mahendran (2009b)
found that the moment modification factor is much closer to 1.0 for LSBs under
quarter point loading. Hence the quarter point loading system was considered in this
research to study the lateral distortional buckling behaviour of LSBs.
A carefully designed special test rig was used to simulate a uniform moment between
the quarter points of LSB members. This test rig included special support conditions
that prevented the in-plane and out-of-plane deflections and twisting rotation without
restraining in-plane and out-of-plane rotations and warping displacements. Also it
was capable of applying the load through the shear centre of the mono-symmetric
LSB sections with no twisting and lateral restraints to the test beam.
Figure 4.4 shows the overall view of the test rig used in the lateral distortional
buckling tests that consisted of a support system and a loading system, which were
attached to an external frame consisting of two main beams and four columns. The
wheel system facilitates loading jacks to move laterally in both directions (along and
across the beam) without creating any restraints.
Test Beam
Figure 4.4: Overall View of Test Rig
4-6
Lateral Buckling Tests of LSB Sections
The support system should ensure that the test beam is simply supported in-plane and
out-of-plane at the supports. The support system used in this experimental program
was developed by Mahaarachchi and Mahendran (2005a), which is similar to that
used by Zhao et al. (1995), Put et al. (1999) and Mahendran and Doan (1999), but
with some modifications. The support conditions provided fixity against in-plane
vertical deflections, out-of-plane deflections and twist rotations, but allowed major
and minor axis rotations. This means that the test beam could rotate freely about its
in-plane horizontal axis and vertical axis at the support, but did not twist.
Box Frame
Side Guide
and Running
LSB Section Track
Ball Bearing
Clamping
Plate
Figure 4.5 shows the movable support at one end of the beam. The support of the
other end was the same except that the side bearing was prevented from rolling along
the running track by horizontal stops. The in-plane vertical movements and lateral
movements were prevented by the running tracks and side guides. The box-frame
4-7
Lateral Buckling Tests of LSB Sections
with the side bearing allowed the test beam to rotate about its major axis while the
top and bottom bearings allowed minor axis rotation and differential flange rotations
(about the minor axis) associated with warping displacement rotations.
The two supports were aligned to ensure that the vertical deflections remained in the
same plane. The test beam was connected to the support system by using 4 M10 bolts
and a 10 mm thick clamping plate. This plate was used to prevent web crippling and
twisting of the section at the supports.
Figure 4.6: Flange Twist at Failure of a 250x75x2.5 LSB with 3.5 m Span
4-8
Lateral Buckling Tests of LSB Sections
Stiffener to
eliminate
flange twist
A loading system was designed in order to apply two vertical loads at the quarter
points through the shear centre, which would produce a uniform bending moment
between the loading points. A gravity loading system was used by other researchers
in the past (Zhao et al., 1995 and Put et al., 1998) to investigate the lateral buckling
of simply supported beams. The gravity loads were applied by suspending a lead
block on a platform that was supported by hangers through the centroid of the
section. However, this method was considered tedious and labour intensive and
could not load the beam continuously. Mahendran and Doan (1999) used an
improved loading system in their lateral buckling tests of hollow flange beams
(HFB) where they applied the vertical loads using a hydraulic jack system operated
under load control. Mahaarachchi and Mahendran (2005a) indicated that the
hydraulic loading system used by Mahendran and Doan (1999) was also not the most
suitable method as it restrained the lateral movement of the test beam, did not allow
the continuation of loading into the post-ultimate load range and the whole loading
set-up was prone to damage. To eliminate the problems associated with this
hydraulic loading system, Mahaarachchi and Mahendran (2005a) improved and
4-9
Lateral Buckling Tests of LSB Sections
developed a new hydraulic loading system. In their system, two hydraulic rams were
connected to a specially designed wheel system on one end and to a load cell on the
other end. The load cell was then connected to a universal joint and then to the
loading point (shear centre) of the test beam. This system was operated under
displacement control with identical loads being applied at both loading points.
Load Cell
Universal Joint
Loading arm
Load position
adjusting bolt
P P
Load 2
Load 1
Support 1
Support 2
4-10
Lateral Buckling Tests of LSB Sections
4-11
Lateral Buckling Tests of LSB Sections
WDTs
All the load and displacement measurement units were connected to the data
acquisition (EDCAR) system which recorded the measurements automatically at
intervals of one second (s). The EDCAR unit included a HP3497A DATA
4-12
Lateral Buckling Tests of LSB Sections
acquisition unit, a HP3498A extender and a PC. Calibration factors of the load cells
and wire displacement transducers (WDT) were determined and input to the EDCAR
unit before each test.
Test specimens were cut 150 mm longer than their intended span since the
connection assembly needed an extra 75 mm at each support. Holes with 12 mm
diameter were drilled on the web at the loading and support positions to
accommodate M10 bolts as shown in Figure 4.11.
Steel plate stiffeners of 6 mm thickness were welded between the inner faces of the
flanges at the middle of each support of the test beam using a tack weld as shown in
Figure 4.12. The widths of these plate stiffeners were 10 mm less than the flange
width in order to accommodate the web plates at the supports as shown in Figure
4.12 (b). Deflection measuring points were marked before the beam was positioned
and clamped to the test rig. The test beam was inserted within the box frame and the
clamping plates were bolted to the test beam (see Figure 4.5). These plates were used
to avoid web crippling and twisting of the section at the supports. The width of the
support plates (web plate) was 75 mm for the beams with the depth of 200 mm or
less, and 100 mm for the beams with the depth more than 200 mm. The depth of
these plates was chosen to cover the full web. The bolt spacing used for 75 mm width
support plates was 45 mm (vertical) x 45 mm (horizontal) while they were 160 mm
4-13
Lateral Buckling Tests of LSB Sections
Tack weld
Web Plate
(a) Front View (b) Cross Sectional View
Note: all the dimensions are in mm.
The calibration factors of all the measuring devices were entered in the EDCAR unit.
A small load was applied first to allow the loading and support systems to settle on
wheels and bearings evenly. The measuring system was then initialized with zero
values. A trial load of 10% of the expected ultimate capacity was applied and
released in order to remove any slackness in the system and to ensure functionality.
The load was then applied gradually and smoothly using a manually operated
hydraulic pump (Figure 4.8 (b)) while the test data was recorded continuously at one
second intervals. The applied load started to drop off when the test beam buckled
out-of-plane. The loading was continued until the test beam failed by out-of-plane
buckling. The loading was also continued after failure in order to obtain the
4-14
Lateral Buckling Tests of LSB Sections
unloading characteristics of the test beam. A typical LSB specimen after failure is
shown in Figure 4.13.
Load Load
Twelve tests were conducted on LSB sections to investigate their lateral distortional
buckling behaviour and ultimate moment capacity. The first four tests did not include
the flange twist restraints while the remaining eight tests included them as shown in
Table 4.1. All the test beams except 150x45x1.6 LSB with a span of 1.2 m failed due
to lateral distortional buckling (Figure 4.14). Some local web buckling was observed
after the ultimate load was reached (Figure 4.15). Shear buckling was observed
between the loading points and the support for 150x45x1.6 LSB with a 1.2 m span.
4-15
Lateral Buckling Tests of LSB Sections
Figure 4.15 shows different views of local web buckling, which was observed just
after the ultimate failure. In the initial loading stages, the top and bottom flanges
slowly started to move laterally until its elastic buckling moment was reached. As the
applied moment reached its ultimate capacity, the lateral deflection of the top and
4-16
Lateral Buckling Tests of LSB Sections
bottom flanges increased rapidly and the beam collapsed by lateral distortional
buckling modes as shown in Figure 4.14. At the beginning of the tests, lateral
deflections were negative but they gradually changed to positive. Flange twist was
not observed in test beams with plate stiffeners at the supports (see Figure 4.16).
As expected, 150x45x1.6 LSB with 1.2 m span exhibited a shear buckling failure and
hence it did not reach the expected lateral distortional buckling capacity. Figure 4.17
shows the shear buckling failure of this test beam.
Shear
Buckling
Figure 4.17: Shear Buckling Failure of 150x45x1.6 LSB with 1.2 m Span
4-17
Lateral Buckling Tests of LSB Sections
During the test, both the applied quarter point loads (P) were measured and recorded
in kN using the EDCAR unit. The average of these two quarter point loads was then
used to calculate the applied uniform moment (M) using the following formula:
M = PL/4 (4.1)
35
30
25
Moment (kNm)
20
at Mid Span
Under the Load
15
10
0
0 10 20 30 40 50 60 70 80 90
Vertical Deflection (mm)
35
30
Bottom Flange(Compression)
Moment (kNm)
20
15
10
0
-10 10 30 50 70 90 110 130 150
Lateral Deflection at Mid Span (mm)
4-18
Lateral Buckling Tests of LSB Sections
Figures 4.18 (a) and (b) show the variation of applied moment vs vertical and lateral
deflections of a 250x75x2.5 LSB with 3.5 m span. All the other moment vs
deflection curves for the tested beams are given in Appendix B. Table 4.2
summarises the ultimate failure moments (Mu) and the type of failure for the tested
beams, and compares them with the corresponding results from Mahaarachchi and
Mahendran (2005a). It is clearly seen that all the test beams except 300x60x2.0 LSB
have higher lateral buckling moment capacities than that of Mahaarachchi and
Mahendran (2005a). This improvement is considered to be due to the improved
manufacturing process of LSBs and the resulting reduced residual stress effects, the
use of an higher strength steel and the improved simply supported boundary
conditions with flange twist restraints.
The ultimate failure moments obtained in this study together with those from
Mahaarachchi and Mahendrans (2005a) test will be used in the comparison with
predicted moment capacities from the current design methods. Although
Mahaarachchi and Mahendrans (2005a) ultimate moments were slightly less than
those from this study, plotting them under a non-dimensional moment capacity curve
format would minimise or eliminate the effects of these differences.
4-19
Lateral Buckling Tests of LSB Sections
Thickness, t (mm)
Span d bf df Mu Failure
LSB Sections
(mm) tof tif tw (mm) (mm) (mm) (kNm) Mode
300x75x3.0LSB 4000 3.18 3.18 2.84 300 75.31 25.17 40.05 LDB
300x75x2.5LSB 4000 2.87 2.87 2.51 300 75.24 25.05 35.31 LDB
300x60x2.0LSB 4000 2.15 2.15 1.98 300 60.28 19.97 16.94 LDB
300x60x2.0LSB 3000 2.15 2.15 1.98 300 60.28 19.97 19.74 LDB, LB
250x75x3.0LSB 4000 3.08 3.08 2.77 250 76.35 25.22 33.35 LDB
250x75x2.5LSB 4000 2.79 2.79 2.48 250 75.98 24.92 28.37 LDB
250x75x2.5LSB 3000 2.79 2.79 2.48 250 75.98 24.92 29.85 LDB
250x60x2.0LSB 4000 2.09 2.09 1.96 250 60.47 20.12 17.28 LDB
250x60x2.0LSB 3000 2.09 2.09 1.96 250 60.47 20.12 18.25 LDB, LB
200x60x2.5LSB 4000 2.58 2.58 2.34 200 60.23 19.95 17.18 LTB
200x60x2.5LSB 3500 2.58 2.58 2.34 200 60.23 19.95 16.78 LTB
200x60x2.0LSB 4000 2.03 2.03 1.85 200 60.15 20.31 12.98 LTB
200x60x2.0LSB 3500 2.03 2.03 1.85 200 60.15 20.31 12.4 LDB
200x45x1.6LSB 4000 1.56 1.56 1.48 200 45.05 14.98 5.66 LTB
200x45x1.6LSB 3000 1.56 1.56 1.48 200 45.05 14.98 6.18 LDB
150x45x2.0LSB 3000 2.11 2.11 1.89 150 44.95 14.73 8.44 LDB
150x45x2.0LSB 2400 2.11 2.11 1.89 150 44.95 14.73 8.26 LDB
150x45x2.0LSB 2000 2.11 2.11 1.89 150 44.95 14.73 9.03 LDB
150x45x1.6LSB 3000 1.60 1.60 1.60 150 45.12 14.89 6.56 LDB
150x45x1.6LSB 2400 1.60 1.60 1.60 150 45.12 14.89 7.01 LDB
150x45x1.6LSB 2000 1.60 1.60 1.60 150 45.12 14.89 7.21 LDB
125x45x2.0LSB 3500 1.98 1.98 1.98 125 45.10 14.93 7.88 LTB
125x45x2.0LSB 2300 1.98 1.98 1.98 125 45.10 14.93 8.41 LTB
125x45x2.0LSB 2000 1.98 1.98 1.98 125 45.10 14.93 8.45 LDB
125x45x2.0LSB 1600 1.98 1.98 1.98 125 45.10 14.93 8.55 LDB
125x45x1.6LSB 3500 1.62 1.62 1.62 125 45.07 14.95 6.69 LTB
125x45x1.6LSB 2300 1.62 1.62 1.62 125 45.07 14.95 7.11 LTB
125x45x1.6LSB 2000 1.62 1.62 1.62 125 45.07 14.95 7.55 LDB
125x45x1.6LSB 1600 1.62 1.62 1.62 125 45.07 14.95 7.51 LDB
Note: LDB Lateral Distortional Buckling, LTB Lateral Torsional Buckling, LB
Local Buckling, tof outer flange thickness, tof inner flange thickness, tof
web thickness.
4-20
Lateral Buckling Tests of LSB Sections
Table 4.3 shows the lateral buckling test details and the results from Mahaarachchi
and Mahendran (2005a). Although they conducted lateral buckling tests with both
the overhang and the quarter point loading methods, only the results from the latter
were considered due to several reasons including the undesirable warping effects
associated with the overhang loading method. Among their results from the quarter
point loading, the results for 150x45x2.0 LSB and 150x45x1.6 LSB sections with 1.2
m span (Points A and B in Figure 4.1) were not considered as their capacities were
reduced by premature shear buckling failures. Since Kurniawan and Mahendran
(2009b) showed that the moment modification factor for quarter point loading of
LSB flexural members can be taken as 1.0 the ultimate moment capacities from tests
will not be reduced as was done by Mahaarachchi and Mahendran (2005a).
The ultimate moment capacity results from this research and those from
Mahaarachchi and Mahendran (2005a) were plotted within the framework of non-
dimensional moment capacity curves and compared with the member moment
capacity predictions from the current design method given in AS/NZS 4600 (SA,
2005).
Clause 3.3.3.3 (b) of AS/NZS 4600 (SA, 2005) outlines the design rules for members
subject to bending under distortional buckling that involves transverse bending of a
vertical web with lateral displacements of the compression flange, which were
developed by Mahaarachchi and Mahendran (2005d). In this case, the nominal
member moment capacity Mb is given by Equation 4.2.
M
M b = Ze c (4.2)
Z
For hollow flange beams, it is appropriate to determine the effective section modulus
(Ze) at a stress corresponding to Mc/Z, where Mc is the critical moment as defined in
Equations 4.3 (a) to (d) and Z is the full elastic section modulus.
4-21
Lateral Buckling Tests of LSB Sections
0.59
For 0.59 < d < 1.70: Mc = My (4.3b)
d
1
For d 1.70: Mc = My 2 (4.3c)
d
d = My / Mod ) (4.3d)
Where Mod is the elastic lateral distortional moment and My is the first yield moment.
Mod can be determined from an elastic buckling analysis program or by using
available buckling formulae while My is given by
My = Z f y (4.4)
Calculations of Mod and My are quite important when plotting the experimental
points in the non-dimensional member capacity curve as a minor change in the
calculation could move the points to another location in the plot. Pi and Trahairs
(1997) Mod equations (Eqs.2.21 and 2.22 in Chapter 2) are considered to be accurate
and have been used in the design capacity tables of LSBs. However, these equations
are valid only for a constant thickness throughout the cross section while the beams
used in the experimental study have different thicknesses. Hence Pi and Trahairs
(1997) equations become unhelpful in this situation. The only option is to use a
software that allows for different thicknesses of plate elements in the cross section. A
well established finite strip analysis program Thin-Wall is considered to be capable
of using different thicknesses in obtaining the elastic lateral distortional buckling
moments Mod of tested sections and spans. It also has the capability to include the
corners of LSB cross-section. Being able to simulate the varying plate thickness and
corners makes Thin-Wall the most suitable in this case. Further, Thin-wall assumes
idealised simply supported boundary conditions at the supports while most of the
experiments in this research also considered the same. Thus Thin-Wall is more
suitable to obtain the Mod of tested LSBs in this research. Mahaarachchi and
Mahendran (2005a) also assumed the same despite the fact their test beams were not
4-22
Lateral Buckling Tests of LSB Sections
fully prevented from flange twist. Hence their experimental Mod values can be
slightly less than those calculated using Thin-Wall. Since there were no other simple
options, Thin-Wall was used to calculate the elastic lateral distortional buckling
moments of all the tested beams, which included the measured dimensions including
thicknesses and nominal corners.
The elastic section modulus Z of tested beams was also calculated using Thin-Wall
using the measured LSB dimensions and thicknesses. The first yield moment My was
then determined by using the measured yield stresses of tested beams. Corners of the
tested beams were not measured and it is not accurate to calculate Mod and My
without corners as the ultimate moment capacities of the tested beam had corners.
Therefore, as decided earlier, the nominal dimensions of corners were used rather
than obtaining those properties without corners. Table 4.4 gives the calculated
section properties, the elastic buckling and yield moment capacities, and the ultimate
moment capacities of the tests from this research while Table 4.5 gives the same for
the tests of Mahaarachchi and Mahendran (2005a).
Table 4.4: Measured Properties and Capacities of LSBs Used in the Current
Lateral Buckling Tests
Span fy Z My Mod Mu
LSB Section d Mu/My
(mm) (MPa) (x103mm3) (kNm) (kNm) (kNm)
300x60x2.0LSB 3000 557.7 101.70 56.72 22.82 1.58 18.09 0.32
300x60x2.0LSB 4000 557.7 103.10 57.50 18.68 1.75 17.17 0.30
250x75x2.5LSB 3500 552.2 119.00 65.71 47.83 1.17 34.13 0.52
200x45x1.6LSB 2000 536.9 39.60 21.26 11.54 1.36 10.72 0.50
200x45x1.6LSB 3000 536.9 39.95 21.45 8.54 1.58 9.24 0.43
200x45x1.6LSB 4000 536.9 39.95 21.45 6.87 1.77 5.92 0.28
150x45x2.0LSB 2000 537.6 32.58 17.52 14.33 1.11 10.76 0.61
150x45x2.0LSB 3000 537.6 32.46 17.45 10.54 1.29 9.87 0.57
150x45x1.6LSB 1200 557.8 26.79 14.94 16.65 0.95 9.29* 0.62
150x45x1.6LSB 1800 557.8 26.79 14.94 11.89 1.12 9.30 0.62
150x45x1.6LSB 3000 557.8 26.53 14.80 8.67 1.31 8.27 0.56
125x45x2.0LSB 1200 544.1 24.56 13.36 19.92 0.82 10.83 0.81
*Shear Failure.
4-23
Lateral Buckling Tests of LSB Sections
Table 4.5: Measured Properties and Capacities of LSBs Used in the Lateral
Buckling Tests of Mahaarachchi and Mahendran (2005a)
Span fy Z My Mod Mu
LSB Section d Mu/My
(mm) (MPa) (x10 mm3)
3
(kNm) (kNm) (kNm)
300x75x3.0LSB 4000 528 173.90 91.82 52.11 1.33 40.05 0.44
300x75x2.5LSB 4000 511 157.90 80.69 45.00 1.34 35.31 0.44
300x60x2.0LSB 4000 568 104.00 59.07 18.95 1.77 16.94 0.29
300x60x2.0LSB 3000 568 104.00 59.07 23.53 1.58 19.74 0.33
250x75x3.0LSB 4000 506 132.80 67.20 51.46 1.14 33.35 0.50
250x75x2.5LSB 4000 525 120.90 63.47 44.54 1.19 28.37 0.45
250x75x2.5LSB 3000 525 120.90 63.47 53.77 1.09 29.85 0.47
250x60x2.0LSB 4000 580 79.12 45.89 18.63 1.57 17.28 0.38
250x60x2.0LSB 3000 580 79.12 45.89 22.66 1.42 18.25 0.40
200x60x2.5LSB 4000 496 70.34 34.89 22.90 1.23 17.18 0.49
200x60x2.5LSB 3500 496 70.34 34.89 25.34 1.17 16.78 0.48
200x60x2.0LSB 4000 473 56.17 26.57 17.68 1.23 12.98 0.49
200x60x2.0LSB 3500 473 56.17 26.57 19.28 1.17 12.40 0.47
200x45x1.6LSB 4000 478 36.14 17.27 6.29 1.66 5.66 0.33
200x45x1.6LSB 3000 478 36.14 17.27 7.73 1.49 6.18 0.36
150x45x2.0LSB 3000 498 32.01 15.94 10.26 1.25 8.44 0.53
150x45x2.0LSB 2400 498 32.01 15.94 12.22 1.14 8.26 0.52
150x45x2.0LSB 2000 498 32.01 15.94 14.00 1.07 9.03 0.57
150x45x1.6LSB 3000 540 25.12 13.56 8.19 1.29 6.56 0.48
150x45x1.6LSB 2400 540 25.12 13.56 9.61 1.19 7.01 0.52
150x45x1.6LSB 2000 540 25.12 13.56 10.89 1.12 7.21 0.53
125x45x2.0LSB 3500 503 23.73 11.94 8.87 1.16 7.88 0.66
125x45x2.0LSB 2300 503 23.73 11.94 12.58 0.97 8.41 0.70
125x45x2.0LSB 2000 503 23.73 11.94 14.01 0.92 8.45 0.71
125x45x2.0LSB 1600 503 23.73 11.94 16.50 0.85 8.55 0.72
125x45x1.6LSB 3500 549 19.71 10.82 7.33 1.22 6.69 0.62
125x45x1.6LSB 2300 549 19.71 10.82 10.09 1.04 7.11 0.66
125x45x1.6LSB 2000 549 19.71 10.82 11.09 0.99 7.55 0.70
125x45x1.6LSB 1600 549 19.71 10.82 12.85 0.92 7.51 0.69
4-24
Lateral Buckling Tests of LSB Sections
1.20
1.00
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20
Slenderness, d
All the ultimate moment capacity and slenderness results were non-dimensionalised
for the purpose of comparison and are plotted in Figure 4.19. The test beam capacity
Mu and the AS/NZS 4600 (SA, 2005) moment capacities Mb are plotted as Mu/My
and Mb/My on the vertical axis whereas the non-dimensional member slenderness d
(=(My/Mod)1/2) is plotted on the horizontal axis. The first yield moment My and d
were calculated using Equations 4.4 and 4.3 (d), respectively. All of these values are
given in Tables 4.4 and 4.5.
Figure 4.19 shows the comparison of experimental results with the current AS/NZS
4600 (SA, 2005) design curve. Most of the experimental results from this research
were found to be higher than AS/NZS 4600 (SA, 2005) predictions. Table 4.6 shows
the ratio of Mu/My for Mahaarachchi and Mahendrans (2005a) test results to the
predictions from AS/NZS 4600 (SA, 2005) design rules. It has a maximum value of
1.30 and a minimum value of 0.87. Only a few test results were less than the
predictions from the current design rule. Table 4.7 shows the ratio of Mu/My for the
test results from this research to the predictions from AS/NZS 4600 (SA, 2005). It
has a maximum value of 1.24 and a minimum value of 0.85. It is clearly seen that the
test results are on average about 13% higher than the predictions from the current
4-25
Lateral Buckling Tests of LSB Sections
design rule. Overall, the current design curve in AS/NZS 4600 (SA, 2005) is quite
conservative compared to the test results from both this research and Mahaarachchi
and Mahendran (2005a).
4-26
Lateral Buckling Tests of LSB Sections
In this section, all the test ultimate moment capacities were compared with the
current design rule predictions in a non-dimensionalised format by using the
measured LSB dimensions, thicknesses and yield stresses and Thin-Wall software to
calculate the correct values of Mod and My required for non-dimensionalisation. This
approach allowed the comparison of all the test results in the same plot despite the
differences in thicknesses and yield stresses while also allowing for the effect of
corners to be included. Including the effects of corners and varying thickness and
yield stress was considered important when comparing with test results. However, in
some lateral buckling tests in this research and all of Mahaarachchi and Mahendrans
(2005a) tests the twisting of flanges was not fully eliminated at the support. This
could have lead to slightly reduce Mod and thus lower values in the tests, but non-
dimensional plots would have eliminated this effect. Since the exact Mod values
could not be measured during tests they were calculated using Thin-Wall and used in
the calculations of non-dimensional slenderness (d). However, Thin-Wall assumes
twist restraint of the entire cross-section at the support, and thus it would have given
higher Mod values for the tested beams without flange twist restraint. This implies
4-27
Lateral Buckling Tests of LSB Sections
slightly lower d values and hence higher predictions from AS/NZS 4600 (SA,
2005), leading to slightly lower ratios of (EXP) / (AS/NZS 4600) capacities. In
summary, the calculated ratios of ultimate moment capacities from experiments and
AS/NZS 4600 (SA, 2005) shown in Tables 4.6 and 4.7 and Figure 4.19 are likely to
be slightly higher if the effects of flange twist restraint were included via exact Mod
values. However, since the mean values of this ratio are already 1.03 and 1.13 in
Tables 4.6 and 4.7, it is concluded that the current design rule is quite conservative
compared to the experimental results from Mahaarachchi and Mahendran (2005a)
and this research.
Table 4.8: Effect of Flange Twist Restraint from Finite Element Analysis
Table 4.8 shows the ratios of Mod and Mu for some test beams with and without
flange twist restraint (FTR) from preliminary finite element analysis while Figure
4.20 shows the typical elastic buckling failure modes of a test beam with and without
4-28
Lateral Buckling Tests of LSB Sections
flange twist restraint. The Mod values are on average about 10% higher for the beams
with flange twist restraint while the Mu values are on average about 8% higher.
However the non-dimensional slenderness d (= (My/Mod)1/2) is reduced by only 5%
by considering the beams with flange twist restraint.
A capacity reduction factor () was calculated based on the test results in Tables 4.6
and 4.7 using the recommended AISI procedure (AISI, 2007). For an overall mean
and COV values of 1.05 and 0.11 based on 40 tests in Table 4.6 and 4.7 a capacity
reduction factor of. 0.91 was determined. This is greater than the recommended
capacity reduction factor of 0.90 in AS/NZS 4600 (SA, 2005). Hence, it confirms
that the current AS/NZS 4600 (SA, 2005) deign rule is conservative.
This study has also shown that despite improved manufacturing process and the use
of higher strength steel the same design curve can be used conservatively. It should
be noted that the current design method provided in AS/NZS 4600 (SA, 2005) was
developed by Mahaarachchi and Mahendran (2005d) based on the lower bound of
FEA and experimental results. Therefore it may be possible to improve the current
design curve. However, it can not be achieved by using only the test results as there
can be several shortcomings and limitations. The test results will be now used to
validate a finite element model of LSBs, which will be followed by a detailed
parametric study. Design curve will then be modified based on the finite element
analytical and experimental results.
4.6 Conclusions
This chapter has described the lateral buckling tests carried out to investigate the
lateral distortional buckling behaviour and member moment capacities of LSB
sections. A total of 12 tests were carried out with the beam span ranging from 1200
mm to 4000 mm, which included compact, non-compact and slender LSB sections.
The quarter point loading method was used and all the test beams failed in lateral
distortional buckling except 150x45x1.6 LSB with 1200 mm which exhibited a shear
buckling failure between the support and the loading positions. The test moment
capacity results from this research and Mahaarachchi and Mahendran (2005a) were
4-29
Lateral Buckling Tests of LSB Sections
compared with the predictions from the current design rules in AS/NZS 4600 (SA,
2005). It was found that the test moment capacity results from this research were on
average about 13% higher than the AS/NZS 4600 (SA, 2005) predictions while those
of Mahaarachchi and Mahendran (2005a) were on average about 3% higher than the
predictions from AS/NZS 4600 (SA, 2005). The use of accurate Mod values for some
test beams without flange twist restraint would have given higher ratios of test
capacity to AS/NZS 4600 prediction. Therefore it is considered that the current
AS/NZS 4600 (SA, 2005) design rule for lateral distortional buckling is considered
to be conservative for LSBs. Further research using finite element analyses is likely
to develop improved design rules for LSBs.
4-30
Finite Element Modelling of LSBs
CHAPTER 5
5.1 Introduction
Two types of finite element models were considered in this research, namely the
ideal finite element models and the experimental finite element models. The purpose
and the description of these models are given next.
5-1
Finite Element Modelling of LSBs
The ideal finite element models (Figure 5.1a) These models incorporated
ideal constraints such as idealised simply supported boundary conditions and
a uniform bending moment throughout the span, nominal dimensions, yield
stresses, geometric imperfections and residual stresses. These idealised
conditions usually simulate the worst case, and hence they are commonly
adopted in the development of design curves as well as in the parametric
studies into the effects of section geometry of LSB on its member moment
capacity.
The experimental finite element models (Figure 5.1b) These models were
developed with the objective of simulating the actual test members physical
geometry, loads, constraints, mechanical properties, geometric imperfections
and residual stresses as closely as possible. They were used for the
comparison with experimental test results of LSBs subjected to quarter point
loads reported in Chapter 4. This comparison was intended to establish the
validity of the finite element model for explicit modelling of initial
geometric imperfections, residual stresses, lateral distortional buckling
deformations, and associated material yielding in non-linear static analyses.
Although this does not directly verify the suitability of the ideal finite
element model for its use in the development of design curves, this approach
is reasonably acceptable as the ideal conditions are simply a theoretical
assumption and are difficult to simulate in the real experiments.
Span/2
Span/4
Span/2
(b) Experimental Model
5-2
Finite Element Modelling of LSBs
Both ideal and experimental finite element models did not consider the corners
although the actual LSB sections have corners as shown in Table 5.1. All the LSB
sections have an inner radius of 3 mm at the web-flange junction while the outer
radius equals to two times the thickness. However, including the corners in finite
element modelling would be cumbersome in relation to the application of mechanical
properties, geometric imperfections and residual stresses. Table 5.1 presents the
nominal dimensions and the yield stresses of both flange and web elements of LSBs.
fy
d d1 bf df t ro riw
LSB Section Flange Web
mm mm mm mm mm mm mm MPa MPa
300x75x3.0 LSB 300 244 75 25 3.0 6.0 3.0 450 380
300x75x2.5 LSB 300 244 75 25 2.5 5.0 3.0 450 380
300x60x2.0 LSB 300 254 60 20 2.0 4.0 3.0 450 380
250x75x3.0 LSB 250 194 75 25 3.0 6.0 3.0 450 380
250x75x2.5 LSB 250 194 75 25 2.5 5.0 3.0 450 380
250x60x2.0 LSB 250 204 60 20 2.0 4.0 3.0 450 380
200x60x2.5 LSB 200 154 60 20 2.5 5.0 3.0 450 380
200x60x2.0 LSB 200 154 60 20 2.0 4.0 3.0 450 380
200x45x1.6 LSB 200 164 45 15 1.6 3.2 3.0 450 380
150x45x2.0 LSB 150 114 45 15 2.0 4.0 3.0 450 380
150x45x1.6 LSB 150 114 45 15 1.6 3.2 3.0 450 380
125x45x2.0 LSB 125 89 45 15 2.0 4.0 3.0 450 380
125x45x1.6 LSB 125 89 45 15 1.6 3.2 3.0 450 380
Note: ddepth, d1clear web depth, bf flange width, df flange depth, t thickness,
ro outer corner radius, riw inner corner radius, fy yield stress
5-3
Finite Element Modelling of LSBs
df
d1
d
ri ro
ro
df
ro ro
bf
Elastic section modulii of actual and idealised LSB sections which were calculated
using a well established finite strip analysis program, Thin-Wall, and the results are
compared in Table 5.2. It is seen that the idealised LSBs over-estimated the elastic
5-4
Finite Element Modelling of LSBs
section modulus by 2.79% on average. Elastic lateral buckling moments were also
compared for both actual and idealised LSB sections using Thin-Wall. Table 5.3
presents the elastic lateral buckling moments of these LSB sections.
Table 5.3: Elastic Lateral Buckling Moments of Actual and Idealised LSB
Sections
Span (mm)
LSB Section 1500 2000 3000 4000
Actual Ideal Actual Ideal Actual Ideal Actual Ideal
300x75x3.0 LSB 138.5 145.9 94.5 98.7 64.3 66.4 51.6 53.1
300x75x2.5 LSB 114.8 119.9 76.1 79.0 50.7 52.1 41.1 42.1
300x60x2.0 LSB 51.2 53.4 34.0 35.3 22.6 23.2 18.2 18.6
250x75x3.0 LSB 119.7 125.6 87.1 90.6 63.0 65.0 51.2 52.6
250x75x2.5 LSB 97.5 101.6 69.0 71.4 49.7 51.0 41.1 42.0
250x60x2.0 LSB 43.7 45.5 30.9 32.0 22.1 22.6 18.1 18.4
200x60x2.5 LSB 50.0 52.1 38.9 40.3 28.8 29.7 23.1 23.7
200x60x2.0 LSB 37.8 39.2 29.0 29.9 22.0 22.5 18.1 18.4
200x45x1.6 LSB 13.4 13.8 10.9 11.2 8.2 8.4 6.6 6.7
150x45x2.0 LSB 17.8 18.5 14.3 14.8 10.4 10.7 8.1 8.3
150x45x1.6 LSB 13.4 13.8 10.9 11.2 8.2 8.4 6.6 6.7
125x45x2.0 LSB 17.5 18.1 14.2 14.6 10.2 10.5 8.0 8.2
125x45x1.6 LSB 13.2 13.5 10.9 11.2 8.2 8.4 6.5 6.6
Percentage differences of the elastic lateral buckling moments of actual and idealised
LSB sections are presented in Table 5.4. The use of idealised LSB section led to an
5-5
Finite Element Modelling of LSBs
Based on the results reported in Tables 5.2 and 5.4 on the effects of corners on the
elastic section modulus and the elastic lateral buckling moments of LSBs, it can be
concluded that the effect of corners is small and that it is adequate to use the
idealised LSB section in finite element modelling. It was found that the effect of
corners in the other sectional properties of LSBs such as torsional constant (J) and
the second moment of area (I) were similar to that of elastic section modulus (Z) and
hence the details are not provided. Most importantly, the moment capacity results
from the finite element analyses of idealised LSB sections will be non-
dimensionalised before they are used in the development of design rules and/or in
drawing important conclusions. This implies clearly that such small differences in
section properties and buckling moment capacities with the use of idealised LSB
sections without corners will not influence the final design rules or recommendations
of this research.
5-6
Finite Element Modelling of LSBs
Both ideal and experimental finite element models were developed using MD
PATRAN (MSC Software, 2008) pre-processing facilities while they were analysed
using ABAQUS (HKS, 2007). MD/PATRAN (MSC Software, 2008) post-processing
facilities were then used to view the results from the ABAQUS analyses.
Shell elements are generally used to model thin-walled structures. ABAQUS (HKS,
2007) includes general purpose shell elements as well as elements that are
specifically formulated to analyse thick and thin shell problems. The general
purpose shell elements provide robust and accurate solutions in most applications
and have the capability of providing sufficient degrees of freedom. Therefore, local
buckling deformations and spread of plasticity could be explicitly modelled. The
shell element in ABAQUS (HKS, 2007) called S4R5 was used to develop the LSB
model. This element is thin, shear flexible, isometric quadrilateral shell with four
nodes and five degrees of freedom per node, utilizing reduced integration and
bilinear interpolation scheme.
One of the most important aspects of finite element modelling is to identify a suitable
mesh size for the accurate modelling of structural response. Finer meshes are
generally preferred to obtain accurate predictions although there is no general
guidelines on the required mesh density, which depends on the type of structure and
analysis involved (Ashraf et al., 2006). But finer meshes make the whole process
more expensive in terms of computational time. A compromise is therefore required
between the level of accuracy and the cost of a solution. Convergence study by
Mahaarachchi and Mahendran (2005c) indicated that a minimum mesh size density
comprising of 5 mm 10 mm elements was required to represent accurate residual
stress distributions, spread of plasticity, and local buckling deformations of LSB
sections. Element widths equal to or less than 5 mm and a length of 10 mm were
selected as the suitable mesh size for the entire cross-section. Nine integration points
through the thickness of the elements were used to model the distribution of flexural
residual stresses in the LSB sections and the spread of plasticity through the
thickness of the shell elements. Kurniawan (2007) also used the same mesh size and
5-7
Finite Element Modelling of LSBs
the number of integration points in his models of LSBs in order to investigate the
moment distribution and load height effects on the moment capacity of LSBs.
Hollow Flanges
Web
Figure 5.3 shows the typical finite element mesh of the LSB used in this research.
The accuracy of the model and appropriateness of the finite element mesh density
and number of integration points was justified by the results of the verification
analyses presented in Section 5.3.3.
The ABAQUS classical metal plasticity model was used in the analysis. This model
implements the von Mises yield surface to define isotropic yielding, and associated
plastic flow theory. This assumption is generally acceptable for most calculations
with metals. The ideal models included the nominal web and flange yield stresses of
380 and 450 MPa, respectively. These yield stresses are the minimum specified
values for the range of LSB sections (Dempsey, 2001). The yield stresses of web,
outside flange and inside flange were also measured using tensile coupon tests (see
Table 3.1 in Chapter 3 for details), and these measured yield stresses were used in the
experimental models of LSBs.
The application of loads and boundary conditions of ideal and experimental finite
element models have similarities and contrasts. For simplicity, the boundary
conditions of these two models are described separately in the following sections.
5-9
Finite Element Modelling of LSBs
T1 T2 T3 R1 R2 R3
One end Yes No No No Yes Yes
Other end No No No No Yes Yes
Mid span No Yes Yes Yes No No
Table 5.5 shows the boundary conditions of ideal simply supported beams. The
presence of symmetry allowed the use of only half the span, which would reduce the
analysis time. In Table 5.5, T and R represent the translation and rotation,
respectively and the subscripts (1, 2, and 3) represent the direction while field Yes
means that it is free to move in that direction. Figure 5.5 illustrates the global axes
selected to input the boundary conditions for the analysis.
Y, 2 Y, 2
X, 1 Z, 3 L/2
Z, 3 X, 1
Figure 5.6 shows the boundary conditions used in the ideal finite element model
considered in this research. The pin support at one end was modelled by using a
Single Point Constraint (SPC) of 234 applied to all the nodes at the end.
Symmetrical boundary condition of SPC 156 was applied to the mid-span of LSB
as only the half span was modelled due to the presence of symmetry conditions.
5-10
Finite Element Modelling of LSBs
Symmetric
Plane
SPC 156
SPC 234
Figure 5.6: Boundary Conditions of the Ideal Finite Element Model of LSB
To simulate a uniform end moment across the section, linear forces were applied at
every node of the beam end, where the nodes above the middle of the web were
subject to tensile forces while the nodes below the middle of the web were subject to
compressive forces. The force at the middle of the web was zero and was linearly
increased within the cross section as shown in Figure 5.7. A tensile force of 1000 N
and a compressive force of 1000 N were applied at the nodes on the top and bottom
faces of LSB cross section. Figure 5.7 shows the applied loads on each node for
200x45x1.6 LSB section. The simulated moment due to the applied loads at each
node can be calculated by multiplying the load at each node by the distance of the
corresponding node to the middle of the web. The total moment is the arithmetic sum
of the above individual moments. Sample calculations can be found in Appendix C.1.
The loading and boundary conditions used in their ideal finite element model used in
5-11
Finite Element Modelling of LSBs
this research are similar to that used by Kurniawan (2007) in his research on the
moment distribution effects of LSBs.
Figure 5.7: Typical Loading Method for the Ideal Finite Element Model of LSB
In the experimental study (Chapter 4), a quarter point loading was applied at the
shear centre with the Idealised simply supported boundary conditions as mentioned
in the above section. This was carefully simulated in the experimental finite element
5-12
Finite Element Modelling of LSBs
model by applying the loads and accurate boundary conditions at the shear centre. A
single point constraint (SPC) and concentrated nodal forces were used in this model
to simulate the experimental boundary conditions and applied loads as closely as
possible. The presence of symmetry permitted modelling of only half the span.
Experimental specimens included a 75 mm width rigid plate at each support, which
was connected to the web of the LSB specimen by using four bolts to prevent
distortion and twisting of the cross-section. These stiffening plates were simulated
using thick shell elements and the web mechanical properties. Shell elements of 10
mm thickness were considered to be appropriate to predict the experimental failure
moments and load-deflection curves for most of the test beams. However, the use of
10 mm thick shell elements simulated the action of support plates for 150x45x2.0
LSB and 125x45x2.0 LSB resulted in highly rigid model, which over-predicted the
moment capacities. Hence 5 mm thick shell elements were used as support plates in
the experimental finite element models of those LSBs. Although past research
(Mahaarachchi and Mahendran, 2005c) considered Rigid Body R3D4 elements to
simulate the support plates, the preliminary finite element analysis showed that these
rigid body elements gave increased elastic buckling moments and ultimate failure
moments. Hence the R3D4 elements were not used in the current experimental finite
element models. Simply supported boundary conditions were applied to the node at
the shear centre in order to provide an ideal pinned support, which was connected to
the support plates using four rigid beam MPCs to simulate the bolt connections as
used in the experiments. In the experimental test set-up, a concentrated load was
applied at the shear centre at the quarter point of the span using a steel plate. The
steel plate was connected to the beam web by using three bolts along the beam
centreline. Same loading arrangement was implemented in the experimental finite
element model using a concentrated nodal load applied at the cross-section shear
centre while simulating the bolts using rigid beam MPCs as shown in Figure 5.8.
Thicker shell elements (10 mm) with the mechanical properties of web elements
were used to represent the loading plate.
5-13
Finite Element Modelling of LSBs
Symmetric Plane
In the preliminary finite element analysis of experimental model the loading plate
twisted significantly and the LSB members deformed in the negative direction
although all the tested LSBs failed in the positive direction and the load plate
twisting was not possible. Therefore, a twist restraint of SPC 4 was provided at the
loading point together with the point load at the shear centre.
5-14
Finite Element Modelling of LSBs
Figure 5.9 shows the typical loading plate twisting observed in the experimental
finite element models of LSBs at ultimate failure. This is not possible and was
eliminated by restraining the degree of freedom 4 (rotation about the longitudinal
axis).
Steel stiffener plates of 6 mm thickness were welded to the inner surfaces of the
flanges at each support as flange twist restraints in the experimental testing except
the first four tests. These plates were modelled in the experimental finite element
model using the elastic perfect plastic material model and a yield stress of 300 MPa.
Welding process was not simulated as this was a tack weld and the effects are
negligible. Also the maximum bending moment occurred at mid-span while the
moment at the support was zero. Therefore a small change in the residual stress and
imperfection due to welding (if any) would not create any adverse effects on the
ultimate failure of the beam. Figure 5.10 identifies the various plate elements with
different mechanical material properties as defined in ABAQUS. Measured
dimensions and yield stresses were used for both the web and flange elements of
LSBs.
It should be noted that the first four tests did not include the stiffener plates
mentioned above and hence their experiment models did not include the flange twist
restraints.
5-15
Finite Element Modelling of LSBs
Outside Flange
Inside Flange
Web
Support Plate
The magnitude and direction of geometric imperfection are some of the important
parameters which reduce the buckling moment capacity of a beam. This should be
considered in finite element modelling as real beams are not perfectly straight. Based
on the geometric imperfection measurements reported in Mahaarachchi and
Mahendran (2005e) and Chapter 3, and the fabrication LSB tolerance limit, a value
of L/1000 was considered conservatively as the overall geometric imperfection in
both the ideal and experimental finite element models of LSBs. A value of depth or
width/150 was considered as the local plate imperfection. However, local plate
imperfection was not considered in the finite element models subject to lateral
buckling as there was no local buckling. The critical imperfection shape was
introduced by ABAQUS *IMPERFECTION option with the lateral distortional
buckling eigenvector obtained from an elastic buckling analysis, and therefore
included lateral displacement, twist rotation, and cross section distortion. Detail
coding to input the imperfection is given as follows.
5-16
Finite Element Modelling of LSBs
For example, if a file with the name 300x75x3.0LSB3.0 (3000 mm span) is analysed
for elastic buckling analysis and the results show that the critical buckling mode is
Mode 2, then the imperfection coding is as follows.
Deformed LSB
Tension
Midspan
Compression
Figure 5.11: Critical Buckling Mode from Elastic Buckling Analysis of Ideal
Finite Element Model
Figure 5.11 shows the critical buckling mode from elastic buckling analysis.
Preliminary investigations revealed that the imperfection direction such as positive
and negative influenced the ultimate moment capacity of LSBs subject to lateral
buckling. Figures 5.12 (a) and (b) show the non-linear analysis results for positive
and negative imperfection shapes.
5-17
Finite Element Modelling of LSBs
The residual stress is an important parameter in the flexural strength of steel beams
as this can cause premature yielding, and reduce their bending strength. Both flexural
and membrane residual stresses were used in both ideal and experimental finite
element models. Figure 5.13 shows the residual stress distribution used in the
numerical modelling which include the flexural residual stress distribution used for
all the LSB sections and the membrane residual stress distribution for 200x45x1.6
LSB. Table 5.6 presents the values of membrane residual stresses for the available 13
LSB sections. Further details of how this table was developed are presented in
Chapter 3.
5-18
Finite Element Modelling of LSBs
1.07fy
0.24fy -0.2567fy
0.24fy 0.41fy 0.03fy
-0.50fy
0.11fy 0.03fy
0.38fy 0.8fy
0.24fy 0.50fy
0.38fy 0.8fy
0.11fy
0.03fy
0.24fy 0.41fy -0.50fy
0.03fy
0.24fy -0.2567fy
1.07fy
(b) Membrane Residual Stress
(a) Flexural Residual for 200x45x1.6LSB
Stress
The residual stresses distributions shown in Figure 5.13 and Table 5.6 were modelled
using the ABAQUS *INITIAL CONDITIONS option, with TYPE = STRESS,
USER. The user defined initial stresses were created using the SIGINI Fortran user
subroutine. A subroutine defining the residual stress distribution for a beam section
is provided in Appendix C.2. This subroutine defines the local components of the
initial stress as a function of the global coordinates. The flexural residual stress is
also a function of the integration point number through the thickness. As the global
coordinates were used to define the local stress components, it was necessary to
allow for member imperfections in the calculations. Equations with the member
length as a variable and constant deformation factors obtained from the buckling
analysis were used to represent the imperfection of top and bottom flanges
approximately.
5-19
Finite Element Modelling of LSBs
In both the ideal and experimental finite element models of LSBs, the initial stresses
were applied in a *STATIC step with no loading and the standard model boundary
conditions to allow equilibration of the initial stress field before starting the response
history. The contours of residual stress after equilibration in a typical ideal finite
element model are shown in Figure 5.14. However, the equilibration of the initial
stress may require additional deformation to bring the model into equilibrium due to
5-20
Finite Element Modelling of LSBs
the unbalanced stress. Past research (Yuan, 2004) in the finite element modelling of
hot-rolled I sections considered an additional force field in the *STATIC step to
reverse this extra initial deformations. This force field was the reaction forces
obtained from a preliminary finite element analysis with all the nodes fixed in the x,
y and z translation degrees of freedom. But this technique was found inappropriate
because the force field remains in the subsequent non-linear analysis step which
provides further restraint to the section. Nevertheless, this initial deformation effect
was considered to be insignificant in the analysis, thus no attempt was made in this
research to eliminate this. In summary, the application of residual stress in this
research was similar to that used by Kurniawan (2007), which was successfully
implemented for LSBs subject to lateral buckling.
Two methods of analysis, elastic buckling and nonlinear static analyses, were used in
this study. Elastic buckling analyses were carried out first and were used to obtain
the eigenvectors for the inclusion of geometric imperfections. Nonlinear static
analysis, including the effects of large deformation and material yielding, was
adopted to investigate the behaviour of LiteSteel beam sections up to failure.
ABAQUS uses the Newton-Raphson method to solve the non-linear equilibrium
equations. The RIKS method in ABAQUS was also included in the nonlinear
analysis. It is generally used to predict geometrically unstable nonlinear collapse of
structures. The RIKS method uses the load magnitude as an additional unknown and
solves simultaneously for loads and displacements. Therefore, another quantity
should be used to measure the progress of the solution. For this purpose, ABAQUS
uses variable arc-length constraint to trace the instability problems associated with
nonlinear buckling of beams. The parameters used for non-linear static analyses are
as follows:
5-21
Finite Element Modelling of LSBs
The finite element models of LSBs were developed in MD/PATRAN and submitted
to ABAQUS (Version 6.7-1) for analysis. Following is a summary of the steps
involved in the finite element modelling of LSBs used in this research.
1. Define geometric surfaces for the flanges and the web.
2. Mesh those surfaces using shell S4R5 elements.
3. Define the support and mid-span boundary conditions and the loads.
(including the MPC rigid beam in the experimental model to simulate the
bolts).
4. Define and assign the mechanical properties for web and flanges (support,
load and stiffener plates for experimental model).
5. Define buckling analysis parameters and run bifurcation buckling analysis
using ABAQUS.
6. Obtain the critical buckling eigenvector and the required maximum
deformation factors for member equations to be included in the residual
stress input subroutine.
7. Prepare the residual stress input subroutine.
8. Define the non-linear static analysis parameters.
9. Using ABAQUS, run a non-linear static analysis, which consists of two
load steps:
Equilibration STATIC step with the standard boundary
conditions, initial geometric imperfection and residual stress input
subroutine.
Ultimate load factor step with the applied moment or load.
It is necessary to verify the accuracy of the developed finite element models prior to
their use in the development of member capacity curves and design
recommendations. This was achieved by conducting two series of comparisons. The
first series, presented in Section 5.3.2, involved comparison of the elastic lateral
distortional buckling moments obtained using the ideal finite element model with the
corresponding moment solutions obtained from the established finite strip analysis
program, Thin-Wall and the predictions from Pi and Trahair (1997). The second
series of comparisons, presented in Section 5.3.3, involved the use of the
5-22
Finite Element Modelling of LSBs
A series of elastic buckling analyses was conducted using the ideal finite element
model of LSBs developed in this research. The results showed that the LSB exhibited
three distinct buckling modes, namely local buckling for short span members, lateral
distortional buckling (LDB) for intermediate span members and lateral torsional
buckling (LTB) for long span members. It was confirmed that the LSBs with
intermediate spans commonly ranging from 1500 mm to 6000 mm (smaller sections
exhibited LDB at 750 mm and the LTB started at about 5000 mm) are prone to
lateral distortional buckling. It was found that the level of web distortion in lateral
distortional buckling varied as a function of beam slenderness, where increasing
beam slenderness reduced the web distortion and thus approached lateral torsional
buckling.
5-23
Finite Element Modelling of LSBs
No web
Distortion
Figures 5.15 (a) to (d) show these three buckling modes of 200x60x2.0 LSB section
obtained from the elastic buckling analysis of its ideal finite element model. Figures
5.16 (a) and (b) show the ultimate failure modes of 200x60x2.0 LSB section obtained
from the non-linear static analysis of its ideal finite element model. For 500 mm
span, yielding occurred before local buckling at ultimate failure and the ultimate
capacity is its section moment capacity. More details of the finite element model of
5-24
Finite Element Modelling of LSBs
LSBs subject to local buckling will be discussed later in Chapter 7. When the beams
are subject to positive imperfection the lateral distortional and lateral torsional
buckling failure modes at ultimate failure were similar to those exhibited in elastic
buckling analyses while a failure mode as shown in Figure 5.16 (b) was observed for
the beams with a negative imperfection.
Elastic buckling moment results from the ideal finite element model were compared
with the corresponding elastic buckling solutions obtained from Thin-Wall and Pi
and Trahairs (1997) equation for elastic lateral distortional buckling moment
(Eqs.2.21 and 2.22 in Chapter 2). Pi and Trahairs (1997) equation has been verified
and adopted in the design capacity tables for LSBs. The comparison was intended to
verify the accuracy of the finite element type, mesh density, boundary conditions and
5-25
Finite Element Modelling of LSBs
loading method used in the finite element model for LSBs. The elastic buckling
moments from FEA, Thin-Wall and Pi and Trahairs (1997) equation and the
percentage differences from the comparisons are summarised in Table 5.7.
Table 5.7: Comparison of Elastic Buckling Moments of LSB from FEA, Thin-
Wall and Pi and Trahairs (1997) Equation
300x75x3.0 LSB 300x75x2.5 LSB
% Difference % Difference
Span Elastic Buckling Elastic Buckling
Compared Compared
(mm) Moment Moment
with FEA with FEA
FEA PT TW PT TW FEA PT TW PT TW
1000 183.8* 317.4 185.2 - 0.75% 107.8* 267.3 108.9 - 0.98%
1500 144.6 155.0 145.9 7.2% 0.93% 107.5* 127.6 108.9 - 1.34%
2000 97.9 102.8 98.7 5.0% 0.88% 78.4 82.4 79.0 5.16% 0.79%
3000 65.7 67.8 66.4 3.3% 1.13% 51.6 53.4 52.1 3.38% 0.93%
4000 52.4 53.7 53.1 2.6% 1.39% 41.6 42.7 42.1 2.74% 1.18%
6000 38.0 38.8 38.7 2.0% 1.71% 31.1 31.7 31.5 2.20% 1.55%
8000 29.7 30.2 30.3 1.7% 1.88% 24.7 25.2 25.2 1.95% 1.78%
10000 24.3 24.7 24.8 1.6% 2.02% 20.4 20.8 20.8 1.81% 1.91%
300x60x2.0 LSB 250x75x3.0 LSB
1000 44.0* 117.7 44.4 - 0.75% 223.0* 255.0 225.2 - 1.01%
1500 43.9* 56.5 44.4 - 1.02% 124.4 132.2 125.6 6.31% 1.01%
2000 35.0 36.7 35.3 4.75% 0.74% 89.7 93.6 90.6 4.28% 1.03%
3000 23.0 23.8 23.2 3.46% 0.96% 64.1 65.9 65.0 2.79% 1.31%
4000 18.4 18.9 18.6 2.90% 1.20% 51.8 52.9 52.6 2.25% 1.56%
6000 13.5 13.8 13.7 2.33% 1.63% 37.5 38.1 38.2 1.82% 1.84%
8000 10.7 10.9 10.9 2.06% 1.88% 29.1 29.6 29.7 1.63% 1.96%
10000 8.8 8.9 8.9 1.87% 1.95% 23.8 24.1 24.2 1.54% 2.02%
250x75x2.5 LSB 250x60x2.0 LSB
1000 130.9* 213.4 132.5 - 1.22% 52.9* 94.5 53.3 - 0.70%
1500 100.6 107.1 101.6 6.45% 0.95% 45.1 47.7 45.5 5.85% 0.80%
2000 70.8 73.9 71.42 4.36% 0.89% 31.7 33.0 32.0 4.29% 0.82%
3000 50.4 51.8 50.97 2.82% 1.11% 22.4 23.1 22.6 3.14% 1.12%
4000 41.4 42.4 41.99 2.32% 1.35% 18.2 18.7 18.4 2.64% 1.37%
6000 30.9 31.5 31.39 1.94% 1.68% 13.4 13.7 13.6 2.14% 1.72%
8000 24.4 24.8 24.86 1.81% 1.89% 10.5 10.7 10.7 1.86% 1.81%
10000 20.1 20.4 20.46 1.70% 1.94% 8.6 8.8 8.8 1.75% 1.97%
* Subject to local buckling, TW Thin-Wall, PT Pi and Trahairs (1997) Eq.
5-26
Finite Element Modelling of LSBs
5-27
Finite Element Modelling of LSBs
The comparison of elastic buckling moments from the three methods is also shown in
Figure 5.17, where the elastic buckling moments are plotted against the span. The
results from only four LSBs were plotted for clarity. The comparison shows that
FEA results agree well with the results from both Thin-Wall and Pi and Trahairs
(1997) equation, where the average difference is about 1.5% and 2.9%, respectively.
While Pi and Trahairs (1997) equation gives an approximate solution, the small
difference with Thin-Wall may be due to a very fine mesh density used in the finite
element model. Most of the bigger LSB sections exhibited local buckling at 1000
mm span as shown in Table 5.7. The local buckling moments from FEA agreed very
well with Thin-Wall results, where the percentage difference is less than 1% on
average. Pi and Trahairs (1997) equation only provides solutions for lateral
distortional buckling, thus its short span results can not be compared with the FEA
results.
200
300x75x3.0 LSB - FEA
180 300x75x3.0 LSB - Pi and Trahair's Eq.
300x75x3.0 LSB - Thin-Wall
160 300x60x2.0 LSB - FEA
Elastic Buckling Moment, (kNm)
40
20
0
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
Span (mm)
Based on Table 5.7 and Figure 5.17, it can be confirmed that the ideal finite element
model developed in this research accurately predicts the elastic lateral distortional
buckling moments of all the LSB sections for a range of member slenderness. Since
the FEA results under-estimated the elastic buckling moments by 1.5% and 2.9% on
average when compared with predictions from Thin-Wall and Pi and Trahairs
(1997) equation, respectively, this FEA model can be conservatively used in this
5-28
Finite Element Modelling of LSBs
research for the development of design curves and parametric studies. Further, these
comparisons verify the suitability and accuracy of the element type, mesh density,
geometry, boundary conditions and the method used to generate the required uniform
bending moment distribution.
It is important to validate the finite element model for non-linear analyses prior to
using it to develop the member capacity curves for LSB sections subject to a uniform
bending moment. This was achieved by comparing the non-linear static analysis
results with the results obtained from experimental tests carried out in this research.
This comparison was intended to establish the validity of the shell element model for
explicit modelling of initial geometric imperfections and residual stresses, lateral
distortional and local buckling deformations, and the associated material yielding.
The accuracy of the residual stress models, local imperfection magnitudes, and the
finite element mesh density will also be established.
Table 5.8 compares the ultimate moment capacity results from experiments with the
non-linear static analyses using the experimental finite element model. A comparison
of FEA and experimental test results is also provided in the form of bending moment
versus vertical deflection curves in Figures 5.18 to 20 for different LSB sections.
These figures compare the measured experimental in-plane deflection to the
corresponding deflections predicted by the finite element analyses with residual
stresses and geometrical imperfections. The vertical deflection was taken at the
centre of the web at mid-span. As seen in these figures, bending moment versus
vertical deflection curves of finite element analyses agreed reasonably well with the
experimental curves. The reason for the observed small difference may be due to the
use of thicker shell elements for load and support plates in the finite element
analyses. Also, the vertical deflections from the experiments were not exactly
measured vertically due to the lateral deflection of beams during the experiments.
It should be noted that the first four tests did not consider the flange twist restraints
(FTR) while the corresponding experimental finite element models also did not
5-29
Finite Element Modelling of LSBs
consider them. Since the test of 150x45x1.6 LSB with 1200 mm span failed by shear
buckling it was not modelled numerically for comparison purposes.
9.0
8.0
7.0
6.0
EXP
Moment (kNm)
5.0 FEA
4.0
3.0
2.0
1.0
0.0
0 10 20 30 40 50 60 70
Vertical Deflection (mm)
5-30
Finite Element Modelling of LSBs
7.0
6.0
5.0
Moment (kNm)
4.0
EXP
3.0 FEA
2.0
1.0
0.0
0 10 20 30 40 50 60 70
Vertical Deflection (mm)
20.0
18.0
16.0
14.0
Moment (kNm)
12.0 EXP
FEA
10.0
8.0
6.0
4.0
2.0
0.0
0 10 20 30 40 50 60 70
Vertical Deflection (mm)
5-31
Finite Element Modelling of LSBs
10.0
9.0
8.0
7.0
Moment (kNm)
6.0
EXP - Top Flange (Tension)
5.0 EXP - Bottom Flange (Compression)
2.0
1.0
0.0
-10 0 10 20 30 40 50 60 70 80 90
Lateral Deflection (mm)
7.0
6.0
5.0
4.0
EXP - Bottom Flange (Compression)
FEA - Top Flange
3.0
FEA - Bottom Flange
2.0
1.0
0.0
-10 10 30 50 70 90 110 130 150
Lateral Deflection (mm)
5-32
Finite Element Modelling of LSBs
12.0
10.0
8.0
Moment (kNm)
2.0
0.0
-10 10 30 50 70 90 110 130 150
Lateral Deflection mm)
Typical bending moment vs. lateral deflection curves are provided in Figures 5.21 to
23. These figures compare the measured experimental out-of-plane deflection at mid-
span for both the top and bottom flanges with the corresponding deflections
predicted by the finite element analyses. Appendix C.3 presents the remaining
moment versus deflection graphs for other LSB sections and spans. As seen in these
figures, the bending moment versus lateral deflection curves from finite element
analyses deviate slightly from the experimental curves. This could be due to the use
of twist restraint (SPC 4) at the loading point in finite element analyses while the
experimental lateral deflections could not have been measured exactly horizontal due
to the vertical deflection of test beams. The hydraulic jacks could also have imposed
a lateral restraint to the beam sections during testing. Even though the loading system
was designed to avoid any lateral restraints, there could have been friction in the
bearings. This was not measured and no attempts were made to include the friction
effects. However, it is considered that such lateral restraints had minimal effect on
the buckling moments of the tested LSB members.
5-33
Finite Element Modelling of LSBs
The comparisons provided in Table 5.8 and Figures 5.18 to 23 demonstrate that the
experimental finite element model predicts the ultimate failure moment accurately.
The mean ratio of the ultimate moment capacities from the finite element analyses
and experiments was 0.99 with a COV of 0.047. This result suggests that the
developed finite element model is accurate, considering the possible approximations
in the finite element models and limitations in the experimental measurements as
described above.
The finite element model provided reasonable comparisons with all the experimental
results. It is therefore reasonable to assume that the experimental comparisons
presented in this section establish the validity of the shell element model for explicit
modelling of initial geometric imperfections and residual stresses, lateral distortional
buckling deformations, and the associated material yielding. The suitability of the
residual stress model, geometric imperfection magnitudes, and the finite element
mesh density has also been verified.
5.4 Conclusions
This section has described the details of ideal and experimental finite element models
developed for the investigation into the behaviour and capacity of LSB flexural
members. The models accurately predicted both the elastic lateral distortional buckling
moments and the non-linear ultimate moment capacities of LiteSteel beam members
subject to pure bending. The models accounted for all the significant behavioural
effects including material inelasticity, lateral distortional buckling deformations,
member instability, web distortion, residual stresses, and geometric imperfections.
The models were validated by the comparison of elastic buckling and ultimate moment
capacity results with corresponding results from an established finite strip analysis
program Thin-Wall and Pi and Trahairs (1997) equation, and experimental test
results, respectively. The validated model can be used for the development of design
curves for LSB flexural members.
5-34
Parametric Studies and Design Rule Development
CHAPTER 6
6.1 Introduction
Other types of hollow flange steel beams including the Hollow Flange Beam (HFB)
were also considered in this chapter to investigate the applicability of the developed
design rules to other hollow flange steel beams. Effects of geometric imperfections
and residual stresses on the member moment capacities of LSBs are also presented in
this chapter.
6-1
Parametric Studies and Design Rule Development
The results presented in Chapter 5 confirmed that the experimental finite element
model of LSBs could accurately simulate the observed experimental behaviour of
LSBs in the lateral buckling tests whereas the ideal finite element model was
validated by a comparison of elastic lateral distortional buckling moments of LSBs
with the well established finite strip analysis program Thin-wall and Pi and Trahairs
(1997) equations (Eq.2.21 and 2.22 in Chapter 2). In the parametric study the ideal
finite element model was used to analyse the lateral distortional buckling behaviour
of 13 LSBs with spans varying from 1 to 10 m and develop an extensive data base of
member moment capacities for the purpose of developing improved design rules.
This model accurately represents a simply supported LSB section subject to a
uniform bending moment, with idealised boundary conditions including no warping
restraints, rotational restraints, or cross-section distortion at the supports. Appropriate
initial geometric imperfections, residual stresses, buckling deformations, cross-
section distortion, material characteristics and spread of plasticity effects were
explicitly modelled. Nominal dimensions, thicknesses, yield stresses and material
properties provided by the LSB manufacturers were used in the numerical analyses.
The corners were not considered in the finite element analyses. However, the use of
non-dimensional moment capacity plots in the development of design rules was
considered to eliminate any effect due to the approximation of not including the
corners. The finite element models of LSBs were developed using MD/ PATRAN
(MSC, 2008) pre-processing facility and analysed using finite element solver
ABAQUS 6.7 (HKS, 2007) while MD/PATRAN (MSC, 2008) post-processing
facility was used to view the results from numerical analyses.
A significant amount of time and effort was required in creating the models in the
pre-processing phase, which included the geometry creation, mesh application and
the application of loads and boundary conditions. Therefore, PATRAN database
journal file containing instructions for the pre-processor was used to automatically
generate a model. Variables such as geometry, finite element mesh, loads, boundary
conditions, material properties and analysis input parameters were then automatically
created by rebuilding the journal file. This simple method was able to generate a
large number of models without creating each model separately. The created
6-2
Parametric Studies and Design Rule Development
ABAQUS input files were analysed using the bifurcation buckling solution sequence
to obtain the elastic buckling eigenvectors. The global geometric imperfections and
residual stresses were then included in the nonlinear analysis model, and the analysis
was continued using the nonlinear static solution sequence.
Table 6.1 presents the nominal dimensions of 13 LSBs considered in the parametric
study. The thickness range of LSB sections considered in this study was 1.6 mm to 3
mm, while the ranges of section depth and flange width were 125 mm to 300 mm and
45 mm to 75 mm, respectively. The hollow flange width to depth ratio is 3 for all the
LSB sections. Following nominal mechanical properties provided by the LSB
manufacturers were used:
Clear
Flange Flange Thick-
Depth Depth
Width Depth -ness
LSB Sections of Web
d d1 bf df t
(mm) (mm) (mm) (mm) (mm)
300x75x3.0 LSB 300 250 75 25 3.0
300x75x2.5 LSB 300 250 75 25 2.5
300x60x2.0 LSB 300 260 60 20 2.0
250x75x3.0 LSB 250 200 75 25 3.0 d d1
250x75x2.5 LSB 250 200 75 25 2.5
250x60x2.0 LSB 250 210 60 20 2.0
200x60x2.5 LSB 200 160 60 20 2.5
200x60x2.0 LSB 200 160 60 20 2.0
200x45x1.6 LSB 200 170 45 15 1.6
df
150x45x2.0 LSB 150 120 45 15 2.0
150x45x1.6 LSB 150 120 45 15 1.6 bf
125x45x2.0 LSB 125 95 45 15 2.0
125x45x1.6 LSB 125 95 45 15 1.6
6-3
Parametric Studies and Design Rule Development
Based on AS/NZS 4600 (SA, 2005) design rules for local buckling, the web and
flange plate elements of all the LSB sections are fully effective if corners are
included. In this case, the effective section modulus is equal to the full section
modulus when the maximum compressive flange stress is taken as its yield stress.
However, if corners are not included as assumed in the finite element models, the
effective section moduli of five LSB sections are about 2% less than their full section
moduli. Local buckling could therefore occur in the case of these slender LSB
sections with short spans. However, the ultimate moment capacity results of such
cases were not considered in this research as it was focussed on lateral buckling only.
Appendix D.1 provides the sample calculations for the effective section moduli of
LSB sections.
Due to the presence of torsionally rigid rectangular hollow flanges and a relatively
slender web, the dominant failure mode of LSBs is lateral distortional buckling. The
lateral distortional buckling behaviour of LSBs made of thin and high strength steel
is complicated and is dependent on a number of parameters including section
geometry. Initial geometric imperfection direction and residual stresses are also
considered critical for the lateral distortional buckling capacities of LSBs. The
effects of imperfection direction and residual stresses on the ultimate moment
capacities of LSBs are presented in this section while the effects of section geometry
will be presented later in this chapter.
Initial geometric imperfections are present in real beams and their magnitude and
direction influence the moment capacities of LSBs subject to lateral distortional
buckling. Past research has shown that the presence of initial geometric imperfection
reduces the capacity of LSBs. However, the effect of the direction of initial
geometric imperfection of LSBs on the moment capacity is not well understood.
Since LSBs are mono-symmetric sections their imperfection direction is likely to
have a significant effect on their moment capacity (Kurniawan and Mahendran,
2009b). Figure 6.1 shows the positive and negative imperfection directions of LSBs.
6-4
Parametric Studies and Design Rule Development
Tension
Compression
Tension
Compression
Both positive and negative initial geometric imperfections were included in the finite
element analyses and it was found that the negative imperfection was critical as the
ultimate moments were lower for the beams with negative imperfections than in the
case with positive imperfections. As stated in Chapter 5 a magnitude of L/1000 was
used as initial geometric imperfection in all the finite element models of LSBs. Table
6.2 presents the ultimate moment capacities of LSBs with positive and negative
imperfections. The residual stresses were not used in these models.
6-5
Parametric Studies and Design Rule Development
Table 6.2 presents the ultimate moment capacities (in kNm) of five LSB sections
with positive and negative geometric imperfections and the ratios of ultimate
moment capacities of LSBs with negative and positive imperfections. The ultimate
moment capacities are also plotted against their span in Figure 6.2. It can be seen that
the ultimate moment capacities of LSBs with negative imperfections are always
lower than that of LSBs with positive imperfections. The ratios of ultimate moment
capacities with negative and positive imperfections given in Table 6.2 show that the
effect of negative imperfection is higher for small spans compared with large spans
while it is smaller for intermediate spans. However, this effect appears to increase
again for very large spans.
6-6
Parametric Studies and Design Rule Development
90.0
300x75x3.0 LSB - Pos IMP
300x75x3.0 LSB - Neg IMP
80.0
300x60x2.0 LSB - Pos IMP
30.0
20.0
10.0
0.0
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
Span (mm)
6-7
Parametric Studies and Design Rule Development
300x75x3.0 LSB
Span with RS Effect of RS
w.o. RS
(mm) F M F + M F/w.o RS M/w.o RS (F + M)/w.o RS
1500 69.79 62.87 67.04 60.84 0.90 0.96 0.87
3000 53.56 49.15 50.10 46.29 0.92 0.94 0.86
4000 46.41 43.30 43.42 40.92 0.93 0.94 0.88
6000 35.55 34.24 33.88 32.45 0.96 0.95 0.91
8000 28.99 28.15 27.44 26.60 0.97 0.95 0.92
10000 24.34 23.74 23.14 22.55 0.98 0.95 0.93
300x60x2.0 LSB
Span with RS Effect of RS
w.o. RS
(mm) F M F + M F/w.o RS M/w.o RS (F + M)/w.o RS
1750 31.20 27.48 28.82 25.62 0.88 0.92 0.82
3000 20.25 19.32 18.70 17.77 0.95 0.92 0.88
4000 17.05 16.32 15.70 14.98 0.96 0.92 0.88
6000 12.91 12.60 11.98 11.57 0.98 0.93 0.90
8000 10.43 10.23 9.63 9.44 0.98 0.92 0.90
10000 8.92 8.74 8.32 8.13 0.98 0.93 0.91
200x60x2.5 LSB
Span with RS Effect of RS
w.o RS
(mm) F M F + M F/w.o RS M/w.o RS (F + M)/w.o RS
1500 28.92 27.02 27.85 26.25 0.93 0.96 0.91
3000 24.41 22.69 22.92 21.68 0.93 0.94 0.89
4000 21.20 19.84 19.96 18.95 0.94 0.94 0.89
6000 16.04 15.38 15.32 14.73 0.96 0.96 0.92
8000 12.83 12.47 12.29 11.94 0.97 0.96 0.93
10000 10.87 10.57 10.45 10.16 0.97 0.96 0.93
200x45x1.6 LSB
Span with RS Effect of RS
w.o RS
(mm) F M F + M F/w.o RS M/w.o RS (F + M)/w.o RS
1500 11.43 10.49 10.74 9.85 0.92 0.94 0.86
3000 7.69 7.34 7.19 6.85 0.96 0.94 0.89
4000 6.36 6.16 5.96 5.76 0.97 0.94 0.91
6000 4.70 4.61 4.44 4.33 0.98 0.94 0.92
8000 3.88 3.80 3.68 3.60 0.98 0.95 0.93
150x45x1.6 LSB
Span with RS Effect of RS
w.o RS
(mm) F M F + M F/w.o RS M/w.o RS (F + M)/w.o RS
1000 10.60 9.70 10.13 9.44 0.92 0.96 0.89
2000 8.94 8.24 8.41 7.87 0.92 0.94 0.88
4000 6.28 6.01 5.98 5.71 0.96 0.95 0.91
6000 4.72 4.55 4.49 4.35 0.96 0.95 0.92
8000 3.79 3.72 3.65 3.55 0.98 0.96 0.94
w.o. RS without Residual Stress, F Flexural RS, M Membrane RS
6-8
Parametric Studies and Design Rule Development
Effects of flexural residual stresses, membrane residual stresses and both residual
stresses are presented in the last three columns of Table 6.3. It can be seen that the
flexural residual stresses significantly reduced the moment capacities of LSBs in the
case of intermediate spans while their effect is not significant in the case of large
spans. In contrast, the effect of membrane residual stresses is low for LSBs with
intermediate spans whereas it is significant in the case of large spans. The effect of
combined flexural and membrane residual stresses is found to be almost the addition
of the individual effects of flexural and membrane residual stresses. Since both
flexural and membrane residual stresses are present in LSBs, they should be
considered in the analysis and design of LSBs. Overall effect of residual stresses is
higher for LSBs with intermediate spans than with large spans.
75.0
300x75x3.0 LSB - without RS
70.0
300x75x3.0 LSB - with Flexural RS
55.0
50.0
45.0
40.0
35.0
30.0
25.0
20.0
1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000
Span (mm)
Figure 6.3 shows the variation of moment capacities of 300x75x3.0 LSBs without
residual stresses, with flexural residual stresses, membrane residual stresses and both
residual stresses as a function of span. The moment capacity curves of other LSBs
are presented in Appendix D.2.
6-9
Parametric Studies and Design Rule Development
The moment capacity results shown in Table 6.3 and Figure 6.3 are now plotted in a
non-dimensionalised format of moment capacity (Mu/My) vs member slenderness
(d), where Mu is the ultimate moment, My is the first yield moment and d =
(My/Mod)1/2.
1.00
300x75x3.0 LSB - without RS
0.90 300x75x3.0 LSB - with RS
Dimensionless Moment Capacity, Mu/My
0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
0.50 0.60 0.70 0.80 0.90 1.00 1.10 1.20 1.30 1.40 1.50 1.60 1.70 1.80 1.90 2.00 2.10 2.20 2.30 2.40 2.50
Slenderness, d
1.00
300x60x2.0 LSB - without RS
0.90 300x60x2.0 LSB - with RS
Dimensionless Moment Capacity, Mu/My
0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
0.50 0.60 0.70 0.80 0.90 1.00 1.10 1.20 1.30 1.40 1.50 1.60 1.70 1.80 1.90 2.00 2.10 2.20 2.30 2.40 2.50
Slenderness, d
6-10
Parametric Studies and Design Rule Development
1.00
200x60x2.5 LSB - without RS
0.90
200x60x2.5 LSB - with RS
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
0.50 0.60 0.70 0.80 0.90 1.00 1.10 1.20 1.30 1.40 1.50 1.60 1.70 1.80 1.90 2.00 2.10 2.20 2.30 2.40 2.50
Slenderness, d
1.00
200x45x1.6 LSB - without RS
0.90 200x45x1.6 LSB - with RS
Dimensionless Moment Capacity, Mu/My
0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
0.50 0.60 0.70 0.80 0.90 1.00 1.10 1.20 1.30 1.40 1.50 1.60 1.70 1.80 1.90 2.00 2.10 2.20 2.30 2.40 2.50
Slenderness, d
6-11
Parametric Studies and Design Rule Development
1.00
150x45x1.6 LSB - without RS
0.90 150x45x1.6 LSB - with RS
Dimensionless Moment Capacity, Mu/My
0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
0.50 0.60 0.70 0.80 0.90 1.00 1.10 1.20 1.30 1.40 1.50 1.60 1.70 1.80 1.90 2.00 2.10 2.20 2.30 2.40 2.50
Slenderness, d
Figures 6.4 to 6.8 compare the ultimate moments of five LSBs with and without
residual stresses. A similar pattern can be observed in all of them where the residual
stress effect is significant for LSBs with intermediate slenderness and this effect is
reduced with increasing slenderness. For LSBs with intermediate slenderness (from
0.50 to 1.10), there is about 16% reduction due to residual stresses. For LSBs with
high slenderness (from 1.11 to 1.70), it is about 10% while for those with very high
slenderness (above 1.71), it is about 8%. The results also showed that the effect of
residual stress is significant for slender LSBs in comparison with compact LSBs.
Here, 300x60x2.0 LSB and 200x45x1.6 LSB are slender beams based on AS/NZS
4600 design rules (see Table D.2 of Appendix D.1).
The ultimate moment capacities of 13 LSBs shown in Table 6.1 with varying spans
were obtained by using the ideal finite element model with the critical negative
geometric imperfections and residual stresses. Figure 6.9 and Table 6.4 present the
ultimate moment capacity results of LSBs subject to lateral distortional and lateral
torsional buckling as a function of span.
6-12
Parametric Studies and Design Rule Development
70
300x75x3.0 LSB
300x75x2.5 LSB
60 300x60x2.0 LSB
250x75x3.0 LSB
250x75x2.5 LSB
50 250x60x2.0 LSB
Moment Capacity, Mu (kNm)
200x60x2.5 LSB
200x60x2.0 LSB
40 200x45x1.6 LSB
150x45x2.0 LSB
150x45x1.6 LSB
30 125x45x2.0 LSB
125x45x1.6 LSB
20
10
0
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000
Span (mm)
In Figure 6.9 the ultimate moment capacities of LSBs subject to lateral distortional
and lateral torsional buckling are shown. The ultimate moment capacities of LSBs
subject to local buckling effects were not considered in this research as it is focused
on lateral buckling effects. The next chapter will provide the design and details of
LSBs subject to local buckling effects including the section capacity tests, finite
element modelling and design.
6-13
Parametric Studies and Design Rule Development
6-14
Parametric Studies and Design Rule Development
As seen from the results of LSBs with intermediate slenderness, the moment capacity
is reduced below the first yield moment due to the interaction of yielding and
buckling effects. This inelastic lateral distortional buckling capacity is influenced by
residual stress distributions and initial geometric imperfections. For LSBs with high
slenderness, the ultimate moment capacity can be predicted approximately by the
elastic lateral distortional buckling moment Mod. This indicates that the effects of
yielding, residual stresses and initial geometric imperfections are very small for
slender beams. Figure 6.9 shows that the moment capacity curves of LSBs with
identical flange properties merge with increasing span. For example,
It was also observed that for LSBs with very high slenderness, the ultimate moment
capacity exceeded the elastic lateral distortional buckling moment. This is due to the
effects of pre-buckling deflections, which transform a straight beam into a negative
arch and thus increases its moment capacity (Trahair, 1993), in particular for small
beams such as the 150 and 125 LSBs with spans more than 8000 mm. However, this
effect can be ignored as it is unlikely for very slender beams to be used without any
lateral restraint.
In this section, the ultimate moment capacities of LSBs obtained from finite element
analyses are compared with the predictions from the current design rules in AS/NZS
4600 (SA, 2005). Clause 3.3.3.3(b) of AS/NZS 4600 (SA, 2005) presents the design
rules for members subject to bending under distortional buckling that involves
transverse bending of a vertical web with lateral displacement of the compression
flange. The member moment capacity, Mb, is given by Equation 6.1:
Ze
Mb = Mc (6.1)
Z
6-15
Parametric Studies and Design Rule Development
where
Z = full section modulus
Mc = critical moment
Ze = effective section modulus
My
d = (6.3)
Mod
The elastic lateral distortional buckling moment Mod can be calculated using Pi and
Trahairs (1997) equations as provided in Equations 6.4 and 6.5 or an elastic
buckling analysis program such as Thin-Wall or CUFSM.
2 EIy 2 EIw
Mod = GJe + (6.4)
L2 L2
where
EIy = minor axis flexural rigidity
EIw = warping rigidity
GJe = effective torsional rigidity
L = span
6-16
Parametric Studies and Design Rule Development
Et 3 L2
2GJf
0.91 2 d 1
GJe = (6.5)
Et 3 L2
2GJf +
0.91 2
d 1
where
GJf = flange torsional rigidity
t = nominal thickness
d1 = clear depth of the web
1.20
AS 4600 (2005)
300x75x3.0 LSB
300x75x2.5 LSB
1.00 300x60x2.0 LSB
250x75x3.0 LSB
250x75x2.5 LSB
0.80 250x60x2.0 LSB
200x60x2.5 LSB
Mu/My, Mb/My
200x60x2.0 LSB
0.60 200x45x1.6 LSB
150x45x2.0 LSB
150x45x1.6 LSB
125x45x2.0 LSB
0.40
125x45x1.6 LSB
0.20
0.59 1.70
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40
Slenderness, d
Figure 6.10: Comparison of Moment Capacity Results from FEA with AS/NZS
4600 (SA, 2005) Design Curve
Figure 6.10 compares the member moment capacities from finite element analyses
with the AS/NZS 4600 (SA, 2005) design curve based on the above equations. The
ultimate moment capacities (Mu) and the elastic lateral distortional buckling
moments (Mod) were obtained from finite element analyses while the first yield
moments (My) were obtained by using Equation 6.6 where the elastic section
modulus (Z) was calculated as shown in Appendix D.1.
My = Z f y (6.6)
6-17
Parametric Studies and Design Rule Development
As stated in the earlier chapters the corners of LSBs were ignored in the calculation
of section properties (Z) since the finite element models also did not include the
corners. The nominal flange yield stress of 450 MPa was used to calculate the first
yield moment My.
Figure 6.10 clearly shows that almost all the finite element analysis data points are
above the current design curve for intermediate slenderness (inelastic lateral buckling
region). Experimental study (Chapter 4) also showed that the moment capacities of
LSBs were higher than the predictions from the current design rule in the inelastic
region. Table 6.5 compares the member moment capacities from finite element
analyses and AS/NZS 4600 (SA, 2005) for each LSB and span. The mean and COV
values of the ratio of member moment capacities from FEA and AS/NZS 4600 (SA,
2005) are 1.08 and 0.088 for LSBs within the inelastic lateral buckling region (0.59 <
d < 1.70). For these calculated mean and COV values, a capacity reduction factor
() of 0.96 was determined using the recommended AISI procedure (AISI, 2007).
This is greater than the recommended capacity reduction factor of 0.90 in AS/NZS
4600 (SA, 2005) for flexural members and hence confirms that the current AS/NZS
4600 (SA, 2005) design rule is conservative in the inelastic lateral buckling region.
6-18
Parametric Studies and Design Rule Development
Table 6.5: Comparison of Moment Capacities from FEA and AS/NZS 4600 (SA,
2005)
FEA M u / My Ratio
Span Mod My
LSB Section d Mu AS/NZS FEA/(AS/NZS
(mm) (kNm) (kNm) FEA
(kNm) 4600 4600)
1500 144.55 77.24 0.73 60.84 0.79 0.81 0.98
2000 97.87 77.24 0.89 54.52 0.71 0.66 1.06
3000 65.69 77.24 1.08 46.29 0.60 0.54 1.10
300x75x3.0
4000 52.37 77.24 1.21 40.92 0.53 0.49 1.09
LSB
6000 38.00 77.24 1.43 32.45 0.42 0.41 1.02
8000 29.71 77.24 1.61 26.60 0.34 0.37 0.94
10000 24.29 77.24 1.78 22.55 0.29 0.31 0.93*
1750 94.02 64.79 0.83 47.26 0.73 0.71 1.03
2000 78.39 64.79 0.91 44.41 0.69 0.65 1.06
3000 51.62 64.79 1.12 36.91 0.57 0.53 1.08
300x75x2.5
4000 41.59 64.79 1.25 32.62 0.50 0.47 1.07
LSB
6000 31.05 64.79 1.44 26.43 0.41 0.41 1.00
8000 24.72 64.79 1.62 22.02 0.34 0.36 0.93
10000 20.43 64.79 1.78 18.81 0.29 0.32 0.92*
1750 41.99 45.17 1.04 25.62 0.57 0.57 1.00
2000 35.04 45.17 1.14 23.45 0.52 0.52 1.00
3000 22.99 45.17 1.40 17.77 0.39 0.42 0.93
300x60x2.0
4000 18.36 45.17 1.57 14.98 0.33 0.38 0.88
LSB
6000 13.50 45.17 1.83 11.57 0.26 0.30 0.86*
8000 10.65 45.17 2.06 9.43 0.21 0.24 0.89*
10000 8.76 45.17 2.27 8.13 0.18 0.19 0.93*
1250 160.82 60.06 0.61 53.63 0.89 0.97 0.92
1500 124.35 60.06 0.69 51.49 0.86 0.85 1.01
2000 89.72 60.06 0.82 48.24 0.80 0.72 1.11
250x75x3.0 3000 64.12 60.06 0.97 43.59 0.73 0.61 1.19
LSB 4000 51.78 60.06 1.08 39.41 0.66 0.55 1.20
6000 37.46 60.06 1.27 31.97 0.53 0.47 1.14
8000 29.14 60.06 1.44 26.40 0.44 0.41 1.07
10000 23.75 60.06 1.59 22.49 0.37 0.37 1.01
1500 100.64 50.38 0.71 41.82 0.83 0.83 1.00
2000 70.79 50.38 0.84 38.39 0.76 0.70 1.09
3000 50.41 50.38 1.00 34.50 0.68 0.59 1.16
250x75x2.5
4000 41.43 50.38 1.10 31.72 0.63 0.54 1.18
LSB
6000 30.87 50.38 1.28 26.24 0.52 0.46 1.13
8000 24.40 50.38 1.44 22.07 0.44 0.41 1.07
10000 20.07 50.38 1.58 18.92 0.38 0.37 1.01
*outside the inelastic lateral buckling region (elastic buckling region)
6-19
Parametric Studies and Design Rule Development
6-20
Parametric Studies and Design Rule Development
FEA M u / My Ratio
LSB Span Mod My
d Mu FEA/(AS/NZS
Sections (mm) (kNm) (kNm) AS/NZS
(kNm) FEA 4600)
4600
750 38.84 14.35 0.61 13.53 0.94 0.97 0.97
1000 26.74 14.35 0.73 12.13 0.85 0.81 1.05
1250 21.31 14.35 0.82 11.63 0.81 0.72 1.13
1500 18.21 14.35 0.89 11.16 0.78 0.66 1.17
150x45x2.0 2000 14.52 14.35 0.99 10.23 0.71 0.59 1.20
LSB 3000 10.48 14.35 1.17 8.53 0.59 0.50 1.18
4000 8.17 14.35 1.33 7.13 0.50 0.45 1.12
6000 5.62 14.35 1.60 5.36 0.37 0.37 1.01
8000 4.26 14.35 1.83 4.37 0.30 0.30 1.02*
10000 3.43 14.35 2.05 4.00 0.28 0.24 1.17*
750 30.57 11.58 0.62 10.10 0.87 0.96 0.91
1000 20.38 11.58 0.75 9.44 0.82 0.78 1.04
1250 15.96 11.58 0.85 8.87 0.77 0.69 1.11
1500 13.61 11.58 0.92 8.47 0.73 0.64 1.14
150x45x1.6 2000 11.02 11.58 1.02 7.87 0.68 0.58 1.18
LSB 3000 8.24 11.58 1.19 6.71 0.58 0.50 1.16
4000 6.56 11.58 1.33 5.71 0.49 0.44 1.11
6000 4.60 11.58 1.59 4.35 0.38 0.37 1.01
8000 3.52 11.58 1.81 3.55 0.31 0.30 1.01*
10000 2.84 11.58 2.02 3.20 0.28 0.25 1.12*
750 33.65 11.15 0.58 10.81 0.97 1.00 0.97*
1000 24.72 11.15 0.67 10.58 0.95 0.88 1.08
1250 20.47 11.15 0.74 10.37 0.93 0.80 1.16
1500 17.82 11.15 0.79 10.14 0.91 0.75 1.22
125x45x2.0 2000 14.35 11.15 0.88 9.46 0.85 0.67 1.27
LSB 3000 10.31 11.15 1.04 8.13 0.73 0.57 1.29
4000 7.99 11.15 1.18 6.93 0.62 0.50 1.24
6000 5.47 11.15 1.43 5.26 0.47 0.41 1.14
8000 4.14 11.15 1.64 4.27 0.38 0.36 1.07
10000 3.33 11.15 1.83 3.78 0.34 0.30 1.14*
*outside the inelastic lateral buckling region (elastic buckling region)
6-21
Parametric Studies and Design Rule Development
FEA M u / My Ratio
LSB Span Mod My
d Mu AS/NZS FEA/(AS/NZS
Sections (mm) (kNm) (kNm) FEA
(kNm) 4600 4600)
The comparison of FEA and experimental member moment capacity results with the
predictions from the current design rules in AS/NZS 4600 (SA, 2005) in the last
section and Chapter 4, respectively, showed that the current design rule is
conservative in the inelastic buckling region while it is adequate in the elastic
buckling region. Therefore the member moment capacity results from experiments
and finite element analyses were combined and used to improve the current design
equations. Experimental and finite element analyses reveal the presence of at least
three buckling modes for LSB flexural members, namely, local, lateral distortional
and lateral torsional buckling. Current design rules consider three distinct regions
such as local buckling/yielding, inelastic and elastic buckling regions, which
correspond to the above buckling modes. Since the current design rule accurately
predicts the moment capacities of LSBs in the elastic buckling region (mostly subject
to lateral torsional buckling), a new design rule was developed for the inelastic
lateral distortional buckling region. The new design equation was established by
solving for minimum total error for all the 13 LSB sections and spans considered
6-22
Parametric Studies and Design Rule Development
here. This was achieved by minimising the square of the difference between the non-
dimensionalised moment capacity results (Mu/My) from FEA and experiments and
that predicted by the new equation (Mb/My). The new design rule for the inelastic
buckling region is given by Equation 6.7(b) and Figure 6.11 compares the design
curve based on this equation with the current AS/NZS 4600 (SA, 2005) design curve
and FEA results. Figures 6.12 and 6.13 show the comparison of the design curve
based on Eqs.6.7 (a) to (c) with experimental results and a combination of
experimental and FEA results, respectively.
1
For d 1.74: Mc = My 2 (6.7c)
d
1.20
AS/NZS 4600 (2005)
Equation 6.7
300x75x3.0 LSB
1.00 300x75x2.5 LSB
300x60x2.0 LSB
250x75x3.0 LSB
0.80 250x75x2.5 LSB
250x60x2.0 LSB
200x60x2.5 LSB
Mu/My, Mb/My
200x60x2.0 LSB
0.60 200x45x1.6 LSB
150x45x2.0 LSB
150x45x1.6 LSB
0.40 125x45x2.0 LSB
125x45x1.6 LSB
0.20
05 17
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Slenderness, d
Figure 6.11: Comparison of FEA Moment Capacities with the Design Curve
based on Equations 6.7 (a) to (c)
6-23
Parametric Studies and Design Rule Development
Equation 6.7
1.20 EXP MM - 300x75x3.0LSB
EXP MM - 300x75x2.5LSB
EXP MM - 300x60x2.0LSB
EXP MM - 250x75x3.0LSB
1.00 EXP MM - 250x75x2.5LSB
EXP MM - 250x60x2.0LSB
EXP MM - 200x60x2.5LSB
EXP MM - 200x60x2.0LSB
EXP MM - 200x45x1.6LSB
0.80 EXP MM - 150x45x2.0LSB
EXP MM - 150x45x1.6LSB
EXP MM - 125x45x2.0LSB
Mb/My, Mu/My
EXP MM - 125x45x1.6LSB
0.60 This Research - 300x60x2.0LSB
This Research - 250x75x2.5LSB
This Research - 200x45x1.6LSB
This Research - 150x45x2.0LSB
This Research - 150x45x1.6LSB
0.40 This Research - 125x45x2.0LSB
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40
Slenderness, d
1.20
Eq.6.7
EXP This Research
EXP MM
1.00
FEA
0.80
Mu/My, Mb/My
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Slenderness, d
Figure 6.13: Comparison of FEA and Experimental Moment Capacities with the
Design Curve based on Equations 6.7 (a) to (c)
The mean and COV values of the ratios of ultimate member moment capacities from
FEA, experiments and FEA and experiments, and Eq.6.7 (b) were calculated, ie.
FEA / Eq.6.7 (b), Test / Eq.6.7 (b) and (FEA + Test) / Eq.6.7 (b), and are presented
in Table 6.6. The corresponding capacity reduction factors () were also determined
using the AISI procedure (AISI, 2007), and are included in Table 6.6. The AISI
procedure is described next.
6-24
Parametric Studies and Design Rule Development
Using the values of common parameters given above, Equation 6.8 leads to Equation
6.9.
0.0566 + CpVp 2
= 1.672 Pme 2.5 (6.9)
Vp, Pm and Cp values have to be determined from experiments or analyses. The use
of FEA provides a large number of moment capacity results (about 110 results in this
research) and hence finite element analyses are increasingly used for the
development of design rules. However, experimental results are also needed to
demonstrate that the developed design rules are accurate. In this research the
6-25
Parametric Studies and Design Rule Development
developed design rules were calibrated using FEA and experimental moment
capacity results separately and in combination as shown in Table 6.6.
As seen in Table 6.6, a capacity reduction factor of 0.92 was obtained for FEA
results, which is greater than the recommended factor of 0.90 in AS/NZS 4600 (SA,
2005). However, it was 0.86 for experiments, which is less than the recommended
value. This is because of comparatively low mean and high COV values of the ratios
of experimental to predicted moment capacities. As shown in Figure 6.12, most of
the experimental data points of Mahaarachchi and Mahendran (2005a) were below
the developed design curves which caused the reduction of the mean value. This may
be due to the approximate elastic lateral distortional buckling moments (Mod) used in
plotting the data points. The Mod value was calculated using Thin-Wall for the tests
of Mahaarachchi and Mahendran (2005a). Thin-Wall assumes ideal supported
conditions (i.e. no flange twist) although local flange twist was not fully eliminated
in their tests. The use of accurate Mod values for these tests would have given higher
ratios of test/predicted moment capacities and hence a greater capacity reduction
factor as discussed in Chapter 4. Nevertheless, the capacity reduction factor was 0.90
when both finite element analysis and experimental results were considered, and this
is considered adequate. Therefore it is recommended that the developed design
equation (Eq.6.7 (b)) can be used in the design of LSBs with a capacity reduction
factor of 0.90 within the guidelines of AS/NZS 4600 (SA, 2005).
Although Equation 6.7 (b) was developed for LSBs subject to lateral buckling with
an acceptable capacity reduction factor of 0.90 by considering both finite element
analysis and experimental results, attempts were also made to develop design rules
by considering FEA results and experimental results separately. They were also
developed with an acceptable capacity reduction factor of 0.9. Equation 6.10 was
6-26
Parametric Studies and Design Rule Development
developed by considering FEA moment capacity results only. Figure 6.14 compares
the FEA results with the developed design equation. The mean, COV and capacity
reduction factor for the moment capacity ratios of FEA / Eq.6.10 (b), EXP / Eq.6.10
(b) and (FEA + EXP) / Eq.6.10 (b) are given in Table 6.7.
1
For d 1.74: Mc = My 2 (6.10c)
d
1.20
AS/NZS 4600 (2005)
Equation 6.10
1.00 FEA
0.80
Mu/My, Mb/My
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Slenderness, d
Figure 6.14: Comparison of FEA Moment Capacities with the Design Curve
based on Equations 6.10 (a) to (c)
6-27
Parametric Studies and Design Rule Development
when compared with experimental results. Hence a lower capacity reduction factor
of 0.84 has to be used. However, an acceptable capacity reduction factor of 0.89 was
obtained when both FEA and experimental results were compared with this equation.
1.20
Equation 6.10
EXP This Research
1.00 EXP MM
0.80
Mu/My, Mb/My
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Slenderness, d
When only the experimental results were considered, the following equations were
developed.
1
For d 1.74: Mc = My 2 (6.11c)
d
Figure 6.16 compares the experimental results with the above design equation based
on experimental results only. The mean, COV and capacity reduction factor for the
moment capacity ratios of FEA / Eq.6.11 (b), EXP / Eq.6.11 (b) and (FEA + EXP) /
Eq.6.11 (b) are given in Table 6.8.
6-28
Parametric Studies and Design Rule Development
1.20
Equation 6.11
EXP This Research
1.00 EXP MM
0.80
Mu/My, Mb/My
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Slenderness, d
6-29
Parametric Studies and Design Rule Development
1.20
Equation 6.11
FEA
1.00
0.80
Mu/My, Mb/My
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Slenderness, d
Figure 6.17: Comparison of FEA Moment Capacities with the Design Curve
based on Equations 6.11 (a) to (c)
Hollow Flange Beam (HFB) shown in Figure 6.18 is the first hollow flange section
developed by the LSB manufacturers. Table 6.9 shows the nominal dimensions of
HFBs while their nominal flange and web yield stresses are 550 MPa and 475 MPa,
respectively. Although the HFB is currently not available in the building industry it
is considered in this research to investigate the applicability of the developed lateral
distortional buckling design rules for other hollow flange steel beams.
6-30
Parametric Studies and Design Rule Development
Avery et al. (1999a, 1999b and 2000) investigated the lateral buckling behaviour of
HFB sections shown in Table 6.9 except 20090HFB28 using finite element analyses,
and developed suitable design procedures. Their FEA results of ultimate moment Mu
and elastic lateral distortional buckling moment Mod are given in Table 6.10. The
first yield moments My of these HFBs with corners were calculated by using their Z
values obtained from Thin-Wall. Since Thin-Wall includes HFBs and LSBs with
corners among its standard sections, it was used here instead of using the basic
principles as for LSBs without corners. Preliminary calculations showed that the
HFB sections with their thickness less than or equal to 2.8 mm are likely to have
local buckling effects in their flanges in the case of short spans. However, comparing
the first yield moments and the ultimate moments, and the chosen spans chosen for
each HFB in Table 6.10, it is clear that Avery et al. (1999a, b, 2000) has only
considered the FEA moment capacities of HFB members subject to lateral
distortional and lateral torsional buckling. Therefore there is no need to allow for any
local buckling effects as a result of the reduced Z values of these HFB sections. The
ultimate moments are compared with the predictions from the design rule developed
for LSBs (Equation 6.7) in Table 6.10 and Figure 6.19. Only the moment capacities
in the inelastic lateral buckling region were considered in the calculation of mean and
COV of the ratios of member moment capacities from FEA and Eq.6.7 (b). The mean
and COV values of this ratio were 1.09 and 0.047 that gave a capacity reduction
factor of 0.98 based on the AISI (2007) procedure. This indicates that Eq.6.7 (b) is
very conservative for HFBs.
6-31
Parametric Studies and Design Rule Development
Table 6.10: Comparison of Avery et al.s (1999b) FEA Results with Eq.6.7
Span Mod My FEA Mu/My Ratio
HFB Section d
(mm) (kNm) (kNm) Mu (kNm) FEA Eq.6.7 FEA/Eq.6.7
1500 194.76 207.13 1.03 141.15 0.68 0.63 1.08
2000 125.56 207.13 1.28 107.39 0.52 0.49 1.06
2500 94.87 207.13 1.48 86.77 0.42 0.41 1.03
3000 78.45 207.13 1.62 74.17 0.36 0.36 1.00
45090HFB38
4000 60.91 207.13 1.84 58.48 0.28 0.29 0.96*
5000 50.81 207.13 2.02 49.64 0.24 0.25 0.98*
6000 43.76 207.13 2.18 43.29 0.21 0.21 0.99*
8000 34.24 207.13 2.46 34.62 0.17 0.17 1.01*
1500 179.13 175.29 0.99 126.01 0.72 0.66 1.09
2000 119.08 175.29 1.21 99.62 0.57 0.53 1.08
2500 92.48 175.29 1.38 82.94 0.47 0.45 1.05
3000 78.87 175.29 1.49 72.36 0.41 0.40 1.02
40090HFB38 4000 61.37 175.29 1.69 58.22 0.33 0.34 0.97
5000 51.35 175.29 1.85 49.67 0.28 0.29 0.97*
6000 44.20 175.29 1.99 43.36 0.25 0.25 0.98*
8000 34.49 175.29 2.25 34.61 0.20 0.20 1.00*
11000 26.89 175.29 2.55 26.89 0.15 0.15 1.00*
1000 316.85 145.26 0.68 132.96 0.92 0.89 1.03
1500 164.76 145.26 0.94 111.38 0.77 0.69 1.11
2000 114.08 145.26 1.13 92.34 0.64 0.57 1.11
2500 91.25 145.26 1.26 78.93 0.54 0.50 1.08
3000 78.09 145.26 1.36 70.20 0.48 0.45 1.06
35090HFB38
4000 62.22 145.26 1.53 58.28 0.40 0.39 1.03
5000 52.07 145.26 1.67 49.91 0.34 0.35 0.99
6000 44.73 145.26 1.80 43.55 0.30 0.31 0.97*
8000 34.76 145.26 2.04 34.69 0.24 0.24 1.00*
11000 25.89 145.26 2.37 26.84 0.18 0.18 1.04*
1000 278.84 116.99 0.65 112.11 0.96 0.91 1.05
1500 152.79 116.99 0.88 97.23 0.83 0.73 1.13
2000 111.22 116.99 1.03 85.15 0.73 0.63 1.15
2500 91.51 116.99 1.13 76.22 0.65 0.57 1.14
3000 79.29 116.99 1.21 69.26 0.59 0.53 1.13
30090HFB38 4000 63.47 116.99 1.36 58.43 0.50 0.46 1.09
5000 52.98 116.99 1.49 50.23 0.43 0.41 1.06
6000 45.36 116.99 1.61 43.84 0.37 0.36 1.03
8000 35.07 116.99 1.83 34.83 0.30 0.30 0.99*
11000 26.02 116.99 2.12 26.72 0.23 0.22 1.03*
14000 20.63 116.99 2.38 22.20 0.19 0.18 1.08*
*outside the inelastic lateral buckling region (elastic lateral buckling region)
6-32
Parametric Studies and Design Rule Development
Table 6.10 (continued): Comparison of Avery et al.s (1999b) FEA Results with
Eq.6.7
Mu/My
HFB Span Mod My FEA Ratio
d
Sections (mm) (kNm) (kNm) Mu (kNm) FEA Eq.6.7 FEA/Eq.6.7
6-33
Parametric Studies and Design Rule Development
Table 6.10 (continued): Comparison of Avery et al.s (1999b) FEA Results with
Eq.6.7
Mu/My
HFB Span Mod My FEA Ratio
d
Sections (mm) (kNm) (kNm) Mu (kNm) FEA Eq.6.7 FEA/Eq.6.7
1500 78.58 55.94 0.84 44.48 0.80 0.76 1.05
2000 57.03 55.94 0.99 40.57 0.73 0.66 1.11
2500 47.79 55.94 1.08 37.75 0.67 0.60 1.13
3000 42.51 55.94 1.15 35.24 0.63 0.56 1.12
4000 35.83 55.94 1.25 31.67 0.57 0.51 1.12
25090HFB23 5000 31.11 55.94 1.34 28.64 0.51 0.46 1.10
6000 27.40 55.94 1.43 25.85 0.46 0.43 1.08
8000 21.93 55.94 1.60 21.54 0.39 0.37 1.05
11000 16.70 55.94 1.83 16.97 0.30 0.30 1.02*
14000 13.40 55.94 2.04 14.16 0.25 0.24 1.06*
18000 10.57 55.94 2.30 11.99 0.21 0.19 1.13*
1500 74.00 41.05 0.74 36.19 0.88 0.83 1.06
2000 59.11 41.05 0.83 35.01 0.85 0.76 1.12
2500 51.49 41.05 0.89 34.08 0.83 0.72 1.15
3000 46.31 41.05 0.94 33.26 0.81 0.69 1.18
4000 38.75 41.05 1.03 31.16 0.76 0.63 1.20
5000 33.16 41.05 1.11 28.74 0.70 0.58 1.20
20090HFB23
6000 28.83 41.05 1.19 26.25 0.64 0.54 1.19
8000 22.67 41.05 1.35 21.85 0.53 0.46 1.15
11000 17.02 41.05 1.55 17.23 0.42 0.38 1.10
14000 13.56 41.05 1.74 14.21 0.35 0.33 1.05*
18000 10.65 41.05 1.96 11.84 0.29 0.26 1.11*
25000 7.80 41.05 2.29 9.20 0.22 0.19 1.18*
Mean 1.09
COV 0.047
*outside the inelastic lateral buckling region (elastic lateral buckling region); not
considered in the calculation of Mean and COV values.
6-34
Parametric Studies and Design Rule Development
1.20
Equation 6.7
45090HFB38
40090HFB38
1.00
35090HFB38
30090HFB38
30090HFB33
0.80
30090HFB28
25090HFB28
Mu/My, Mb/My
25090HFB23
0.60 20090HFB23
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Slenderness, d
Figure 6.19: Comparison of FEA Moment Capacities of HFBs from Avery et al.
(1999b) with Equations 6.7 (a) to (c)
As seen in Table 6.10 and Figure 6.19, Equation 6.7 (b) is conservative in the
inelastic lateral buckling region. However, Avery et al. (1999b) did not include the
membrane residual stresses in their finite element analyses of HFBs despite the fact
that the manufacturing process of HFBs also involved an electric welding process
similar to that used for LSBs. This welding process would have created some
membrane residual stresses in HFBs, which would have reduced their moment
capacities. The effects of membrane residual stress of LSBs were investigated in this
research and are presented in Table 6.3. It was found that on average the membrane
residual stress reduced the lateral buckling moment capacities of LSBs by about 6%.
It is believed that the effect of welding process in the making of HFBs is similar to
that of LSBs and hence the effects of membrane residual stresses on the moment
capacities of HFBs could also be similar to that of LSBs. Therefore, the mean FEA to
predicted moment capacity ratio is likely to be reduced by about 6% if the membrane
residual stress is considered in the analysis of HFBs. The capacity reduction factor
will also be reduced from 0.98. Therefore, no attempts were made to modify Eq.6.7
(b) or to develop a new design rule to predict the moment capacities of HFBs subject
to lateral distortional buckling.
6-35
Parametric Studies and Design Rule Development
2 EIy 2 EIw
Mo = GJ + (6.12)
L2 L2
where
EIy = minor axis flexural rigidity
EIw = warping rigidity
GJ = torsional rigidity
L = span
When plotting the non-dimensional member capacity curves for beams subject to
lateral distortional buckling, a modified member slenderness parameter
{d=(My/Mod)1/2} is used instead of {=(My/Mo)1/2}. This procedure was used in the
earlier sections of this chapter. However, a closer look at the finite element analysis
data points for LSBs plotted in the non-dimensional moment capacity versus
modified slenderness d format as shown in Figure 6.20 reveals that the points are
scattered to some extent. Although suitable design rules in the form of Eqs.6.7 (a) to
(c) have been developed based on these data points through a process of minimising
the total error, it underestimates the member moment capacities of some LSB
6-36
Parametric Studies and Design Rule Development
sections (compact sections) while overestimating them for other LSB sections
(slender sections) as shown in Figures 6.11 and 6.13. If the member capacity
equations are developed based on less scattered data, this shortcoming will be
eliminated and their accuracy will be equally good for all the LSB sections.
Following sections discuss the use of other modified member slenderness parameters
to achieve this.
1.20
1.00
0.80
Mu/My
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Slenderness, d
The FEA member moment capacities of LSBs are plotted in the non-dimensional
member capacity versus slenderness format in Figure 6.21 where the slenderness
My
was based on lateral torsional buckling moment Mo (= ) . This approach does not
Mo
include the effect of web distortion observed with lateral distortional buckling and
hence leads to more scattered data as shown in Figure 6.21. The elastic lateral
torsional buckling moments (Mo) were calculated based on Eq.6.12 and are presented
in Appendix D.3. Comparison of data points in Figures 6.20 and 6.21 clearly
demonstrates this. The use of a modified member slenderness parameter, d
My
(= ) , in Figure 6.20 considers the effects due to web distortion and hence
Mod
reduces the scatter among the data points. However, further improvements are
6-37
Parametric Studies and Design Rule Development
1.20
1.00
0.80
Mu/My
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
1/2
Slenderness, (MY/Mo)
Trahair (1995a) introduced a geometrical parameter Et3L2 / GJfd1 in the equation for
effective torsional rigidity GJe (Eq.6.5) and stated that it was a measure of the
relative magnitude of the flexural rigidity of the web in comparison with the torsional
rigidity of the flanges in the investigation of elastic lateral distortional buckling of
hollow flange beams. An attempt was therefore made to determine a geometrical
parameter K1 in terms of Et3L2 / GJfd1 that can be used to modify the slenderness
parameter as K1 instead of d in order to reduce the scatter of FEA data points.
When the structural parameter K1 defined by Equation 6.13 was developed and used,
it was found that the scatter of data points was reduced as seen in Figure 6.22.
1
K1 = (6.13)
1/ 4
Et L 3 2
0.3 + 0.35
GJfd 1
where K1 is a factor determined based on several trial and error attempts to reduce
the scatter of data points. The member moment capacity design rules in this case are
given by the following equations.
6-38
Parametric Studies and Design Rule Development
1.20
FEA
Equation 6.14
1.00
0.80
Mu/My, Mb/My
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Modified Slenderness, K1
Figure 6.22: Moment Capacity Design Curve for LSBs based on a Modified
Slenderness Parameter K1
For the design rules developed (Eq.6.14b) the mean and COV of the ratio of FEA to
predicted moment capacities were 1.00 and 0.069 with a capacity reduction factor of
0.90. However, a comparison of FEA data points in Figures 6.20 and 6.22 reveals
that the scatter has not been reduced much when d was replaced by K1. Therefore
an attempt was made to plot FEA data points in the Mu/My vs K2d format, where
My
d = and K2 is a geometrical parameter defined in terms of Et3L2/GJfd1 by
Mod
Equation 6.15.
1
K2 = (6.15)
0.15
Et L 3 2
0.5 + 0.5
GJ f d 1
6-39
Parametric Studies and Design Rule Development
1.20
FEA
Equation 6.16
1.00
0.80
Mu/My, Mb/My
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Modified Slenderness, K2d
Figure 6.23: Moment Capacity Design Curve for LSBs based on a Modified
Slenderness Parameter K2d
New design rules were developed for the FEA data points plotted in the new format.
For these design rules given by Eq.6.16, the mean and COV of the ratio of FEA to
predicted moment capacities were 1.00 and 0.048 with a capacity reduction factor of
0.90.
Comparison of Figures 6.20, 6.22 and 6.23 show that the plot in the Mu/My vs K2d
format has the least scatter of data points among them. However, it is to be noted that
the parameter Et3L2/GJfd1 has already been included in the Mod equation via GJe
My
equation in d = . This implies that the parameter Et3L2/GJfd1 has been used
Mod
twice by considering K2d to include the effects of web distortion in LSBs. This does
not appear to be appropriate. The following equations clearly explain this fact.
6-40
Parametric Studies and Design Rule Development
2 EIy 2 EIw
Mod = GJe + ,
L2 L2
Et 3 L2 Et 3 L2
2GJf 2
0 .91 2
d 1
, Divide by GJf 0 .91 2
d 1
GJe = GJe =
Et 3 L2 Et 3 L2
2GJf + 2 +
0.91 d 1 0.91 GJfd 1
2 2
Apart from this, the calculation of K2 for each span is not a simple task for designers.
Therefore, from a design point of view, the use of Equation 6.16 with K2d was not
considered suitable. Several other parameters such as depth/thickness,
width/thickness, depth/width, slenderness of plate elements, torsional rigidity and
flexural rigidity were considered to determine a simple geometrical parameter which
would reduce the scatter of FEA data points of LSBs. Finally it was found that the
use of a new K parameter defined as a function of the ratio of torsional rigidity of the
flanges to the major axis flexural rigidity of web (GJf/EIxweb) considerably reduced
the scatter of FEA data points. The new parameter K is defined by Equation 6.17.
Figure 6.24 shows the FEA data points plotted in the non-dimensional moment
capacity (Mu/My) versus modified slenderness (Kd) format.
1
K= (6.17)
GJf
0.85 +
EIxweb
where
GJf = torsional rigidity of the flange
EIxweb = major axis flexural rigidity of the web
6-41
Parametric Studies and Design Rule Development
1.20
300x75x3.0 LSB
300x75x2.5 LSB
300x60x2.0 LSB
1.00
250x75x3.0 LSB
250x75x2.5 LSB
250x60x2.0 LSB
0.80 200x60x2.5 LSB
200x60x2.0 LSB
Mu/My , Mb/My
200x45x1.6 LSB
0.60 150x45x2.0 LSB
150x45x1.6 LSB
125x45x2.0 LSB
0.40 125x45x1.6 LSB
Equation 6.18
6.11
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Modified Slenderness, Kd
Figure 6.24: Moment Capacity Design Curve for LSBs based on a Modified
Slenderness Parameter Kd
Based on the FEA moment capacity results plotted in Figure 6.24, new design rules
were developed as given by Equation 6.18.
Comparison of Figures 6.20, 6.23 and 6.24 reveal that the moment capacity plot in
the new Mu/My versus Kd format in Figure 6.24 has little scatter among the data
points. Therefore Equations 6.18 (a) and (b) are considered to be accurate and
recommended as alternative improved equations to Eqs.6.7 (a) to (c). Since the
horizontal axis was changed to the modified slenderness (Kd) the elastic buckling
region as provided in Equation 6.7 (c) does not have any meaning. Hence there are
only two regions in Equations 6.18 (a) and (b), namely, the local buckling/yielding
region (Kd 0.52) and the lateral buckling region (Kd > 0.52).
6-42
Parametric Studies and Design Rule Development
1.20 8
Equation 6.11
MM EXP - 300x75x3.0LSB
MM EXP - 300x75x2.5LSB
MM EXP - 300x60x2.0LSB
MM EXP - 250x75x3.0LSB
1.00 MM EXP - 250x75x2.5LSB
MM EXP - 250x60x2.0LSB
MM EXP - 200x60x2.5LSB
MM EXP - 200x60x2.0LSB
MM EXP - 200x45x1.6LSB
0.80 MM EXP - 150x45x2.0LSB
MM EXP - 150x45x1.6LSB
Mb/My, Mu/My
MM EXP - 125x45x2.0LSB
MM EXP - 125x45x1.6LSB
0.60 This Research - 300x60x2.0LSB
This Research - 250x75x2.5LSB
This Research - 200x45x1.6LSB
This Research - 150x45x2.0LSB
This Research - 150x45x1.6LSB
0.40 This Research - 125x45x2.0LSB
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40
Modified Slenderness, Kd
Figure 6.25 compares the experimental moment capacity results with Equation 6.18
based on the new geometrical parameter K. To plot the experimental data points,
torsional rigidity of the flange and the major axis flexural rigidity of the web were
calculated based on the measured dimensions of the tested beams and the parameter
K was calculated as shown in Appendix D.4. The mean, COV and capacity reduction
factor for the ratios of FEA / Eq.6.18 (b), EXP / Eq.6.18 (b) and (FEA + EXP) /
Eq.6.18 (b) are given in Table 6.11. The capacity reduction factor obtained in the
case of FEA and FEA+EXP were 0.90 and 0.89 and thus confirm the adequacy of the
new design rules. However, it was only 0.83 when only the experimental moment
capacity results were considered. As explained in Section 4.5 of Chapter 4, the use of
accurate Mod values for the tests of Mahaarachchi and Mahendran (2005a) will
eliminate the approximation in the evaluation of test results and thus increase the
capacity reduction factor.
6-43
Parametric Studies and Design Rule Development
Table 6.12 presents the values of torsional rigidity of flange (GJf) and major axis
flexural rigidity of web (EIxweb) for all the 13 LSB sections considered in this
research. It also includes the values of the important geometrical parameter K used in
the development of Eqs.6.18 (a) and (b). As seen in Table 6.12, each LSB section has
unique values of GJf/EIxweb and K. Figures 6.26 to 6.28 show the plots of FEA
moment capacities in the non-dimensional moment capacity format of Mu/My versus
slenderness d in an attempt to study the effect of K and GJf/EIxweb on the moment
capacity curves of LSBs.
1.20
Equation 6.7
300x75x3.0 LSB
300x75x2.5 LSB
1.00
0.80
Mu/My, Mb/My
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Slenderness, d
6-44
Parametric Studies and Design Rule Development
1.20
Equation 6.7
250x75x3.0 LSB
250x75x2.5 LSB
1.00 200x60x2.5 LSB
200x60x2.0 LSB
150x45x2.0 LSB
0.80 150x45x1.6 LSB
Mu/My, Mb/My
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Slenderness, d
Figure 6.26 shows the FEA moment capacity data points of 300x75x3.0 LSB and
300x75x2.5 LSB, which have similar GJf/EIxweb values of 0.0164 and 0.0172 (see
Table 6.12). It is clearly seen that the data points follow the same trend. Similarly in
Figure 6.27, the FEA moment capacity data points of six LSBs with different
dimensions but with similar values of GJf/EIxweb (about 0.03 as seen in Table 6.12)
are plotted, which show the same trend. Figure 6.28 is another example where two
LSBs (125x45x2.0 LSB and 125x45x1.6 LSB) with similar values of GJf/EIxweb
(about 0.06) follow the same trend. Hence it is concluded that as demonstrated by
Figures 6.26 to 28 the chosen parameters GJf/EIxweb and K are appropriate in
reducing the scatter of FEA data points of LSBs. Another important finding was that
LSBs with high values of GJf/EIxweb plotted above the design curve and those with
low values plotted below the design curve (Figure 6.28). In other words, non-
dimensional moment capacity ratios (Mu/My) increase with the parameter GJf/EIxweb
for a given slenderness. As shown earlier, this scatter can be significantly reduced by
using the K factor which includes the parameter GJf/EIxweb.
6-45
Parametric Studies and Design Rule Development
1.20
Equation 6.7
300x60x2.0 LSB
1.00 200x45x1.6 LSB
125x45x2.0 LSB
125x45x1.6 LSB
0.80
Mu/My, Mb/My
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Slenderness, d
An attempt was then made to verify the applicability of Equation 6.18 based on the
new K factor for HFB sections. For this purpose, the torsional rigidity of triangular
hollow flanges and the major axis flexural rigidity of web were calculated as for
LSBs. However, the corners of HFBs can not be ignored as they are large compared
to the corners in LSBs. Therefore, Thin-wall was used to calculate the torsional
constant of triangular hollow flange (Jf) with corners. Since Avery et al.s (1999b)
finite element analyses also considered the corners of HFBs, the same configuration
of HFBs was used in the calculation of parameter K. Table 6.13 shows the values of
torsional rigidity of flange, major axis flexural rigidity of web and the parameter K
for HFBs with corners.
6-46
Parametric Studies and Design Rule Development
1.20
Equation 6.18
6.11
45090HFB38
40090HFB38
1.00
35090HFB38
30090HFB38
30090HFB33
0.80
30090HFB28
25090HFB28
Mu/My, Mb/My
25090HFB23
0.60 20090HFB23
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Modified Slenderness, Kd
Figure 6.29: Comparison of FEA Moment Capacities of HFBs from Avery et al.
(1999b) with Equation 6.18
Figure 6.29 shows the comparison of Avery et al.s (1999b) ultimate moment
capacities of HFBs from their finite element analyses with Equations 6.18 (a) and
(b). It shows that the moment capacity data points are more scattered than in the case
of LSBs (compare with Figure 6.24). A closer look at Figure 6.29 shows that the data
points of 20090HFB23, 25090HFB23 and 25090HFB28 caused this scatter. Hence
Figure 6.30 was plotted without considering these three HFBs.
1.20
Equation 6.18
6.11
45090HFB38
1.00
40090HFB38
35090HFB38
0.80
30090HFB38
Mu/My, Mb/My
30090HFB33
0.60 30090HFB28
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Modified Slenderness, Kd
6-47
Parametric Studies and Design Rule Development
As seen in Figure 6.30, the FEA moment capacity data points of HFBs have only a
small scatter when the data points of 20090HFB23, 25090HFB23 and 25090HFB28
were excluded. It was found that the values of GJf/EIxweb for these three HFBs were
much higher than those of the remaining HFBs as shown in Table 6.13. This ratio
was about 0.106 for 25090HFB28 and 25090HFB23 and 0.305 for 20090HFB23
while the remaining HFBs have a ratio less than 0.0485. The ratios of GJf/EIxweb for
LSBs varied from 0.0078 to 0.0667 and the moment capacity results of LSBs agree
well with Equation 6.18 as seen in Figure 6.24.
Based on the results of LSBs and HFBs, it can be observed that the moment capacity
results of LSBs and HFBs with GJf/EIxweb values in the range of 0.0078 and 0.0667
agree well with Equations 6.18 (a) and (b) while those with a ratio of 0.1058 and
higher do not agree. This indicates that there is a need to define suitable lower and
upper limits of this ratio when Equation 6.18 can be used for LSBs and HFBs. The
upper limit is expected to be between 0.0667 and 0.1058 based on the above
observations. However, in order to determine the lower limit, further finite element
analyses are needed. Finite element analyses showed that the level of web distortion
is very small for the beams with high values of GJf/EIxweb. For example, 125x45x2.0
LSB (0.0625) and 125x451.6 LSB (0.0667) have high values of GJf/EIxweb among the
13 LSBs and their finite element analysis data points were seen to plot well above the
Mu/My versus d design curve defined by Eq.6.7 as seen in Figure 6.11 or 6.28. They
are small sections with smaller web depth and the tendency to fail by web distortion
is low. Similarly, 20090HFB23 (0.3054), 25090HFB23 (0.1058) and 25090HFB28
(0.1060) have high values of GJf/EIxweb among the nine HFBs and their FEA moment
capacity data points were also seen to plot well above the design curve defined by
Eq.6.7 as seen in Figure 6.19. These beams are also beams with a smaller web depth
and web distortion is less likely to occur. The plot of HFB data points with similar
values of GJf/EIxweb were compared in Figures 6.31 and 6.32 to validate the
applicability of this parameter. These figures show that HFBs with similar values of
GJf/EIxweb follow the same trend in Mu/My versus d plots where 30090HFB38,
30090HFB33 and 30090HFB28 have GJf/EIxweb values of 0.0459, 0.0472 and 0.0485,
respectively, while those of 25090HFB28 and 25090HFB23 are about 0.106.
Therefore it can be concluded that the use of parameter GJf/EIxweb is appropriate in
6-48
Parametric Studies and Design Rule Development
the investigation into the effects of section geometry of hollow flange steel beams
subject to lateral distortional buckling.
1.20
Equation 6.7
30090HFB38
1.00 30090HFB33
30090HFB28
0.80
Mu/My, Mb/My
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Slenderness, d
1.20
Equation 6.7
25090HFB28
1.00 25090HFB23
0.80
Mu/My, Mb/My
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Slenderness, d
6-49
Parametric Studies and Design Rule Development
Table 6.14: Two New LSBs with Different GJf/EIxweb and K Values
LSB
Span Mod (kNm) My (kNm) d Mu (kNm) Mu/My
Sections
750 134.95 65.61 0.70 51.36 0.78
1000 84.90 65.61 0.88 43.84 0.67
1500 46.76 65.61 1.18 31.77 0.48
300x45x3.6
2000 32.41 65.61 1.42 24.14 0.37
LSB
3000 20.46 65.61 1.79 16.20 0.25
4000 15.09 65.61 2.08 12.28 0.19
5000 11.99 65.61 2.34 9.84 0.15
6-50
Parametric Studies and Design Rule Development
1.20
Equation 6.18
Available LSBs
1.00 300x45x3.6LSB
135x50x1.6LSB
0.80
Mu/My, Mb/My
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40
Modified Slenderness, Kd
Figure 6.33: Moment Capacities of New LSBs with Different GJf/EIxweb Values
Figure 6.33 shows a comparison of the FEA moment capacity data points of the new
LSBs as plotted in the non-dimensional member moment capacity Mu/My versus
modified slenderness Kd format. As seen in this figure, the FEA moment capacities
of LSBs with GJf/EIxweb values in the range of 0.0021 and 0.0811 agree well with the
design curve based on Equation 6.18. Based on the comparisons in Figure 6.29, the
moment capacities of HFBs with a GJf/EIxweb value of 0.1058 did not comply with
Equation 6.18. Further, it is obvious that the value of K decreases with increasing
values of GJf/EIxweb and this is the reason why the FEA data points of HFBs with
higher values of GJf/EIxweb plot below the developed design curve. Therefore, it was
decided to define a lower limit for the values of K instead of an upper limit for
GJf/EIxweb. For lower values of GJf/EIxweb, the K values become higher which will
bring the FEA data points towards right in the Mu/My versus Kd plots as for
300x45x3.6 LSB in Figure 6.33, i e. conservative. Therefore an upper limit may not
be essential for K.
An attempt was made to plot the FEA data points of all the HFBs with the GJf/EIxweb
values of greater than 0.0811 in the Mu/My versus d format to observe the variations
in the plot. For this purpose two more non-standard LSB sections with higher values
of GJf/EIxweb were considered in the finite element analyses. They are 125x45x1.8
LSB and 125x47x2.4 LSB with GJf/EIxweb values of 0.1123 and 0.2065, respectively.
6-51
Parametric Studies and Design Rule Development
The dimensions and the FEA results of these LSBs are given in Tables 6.16 and 6.17,
respectively. Figure 6.34 shows the moment capacities of these HFBs and LSBs with
GJf/EIxweb values greater than 0.0811.
Table 6.17: FEA Moment Capacity Results of Two New LSBs with Higher
Values of GJf/EIxweb
6-52
Parametric Studies and Design Rule Development
1.20
Equation 6.7
125x47x2.4 LSB (0.2065)
125x45x1.8 LSB (0.1123)
1.00 135x50x1.6 LSB (0.0811)
25090HFB28 (0.1060)
25090HFB23 (0.1058)
0.80 20090HFB23 (0.3054)
Mu/My, Mb/My
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Slenderness, d
As seen in Figure 6.34, the FEA moment capacity data points of different hollow
flange steel beams with different values of GJf/EIxweb followed a similar trend
irrespective of the GJf/EIxweb values (when GJf/EIxweb values are greater than 0.0811).
As seen in Figure 6.34, Eq.6.7 (b) is very conservative for these sections and hence
the use of Eq.6.18 (b) based on the modified slenderness Kd will be useful. For the
applicability of Eqs.6.18 (a) and (b) it is therefore reasonable to define a single K
value for the hollow flange steel beams with the values of GJf/EIxweb greater than
0.0811. Based on this, a value of 0.8812 was defined as the lower limit for K based
on the limiting GJf/EIxweb value of 0.0811. Equation 6.19 defines the K value with its
lower limit. The FEA moment capacities of the above mentioned hollow flange steel
beams were plotted in the Mu/My versus Kd format with the same K value of 0.88 in
Figure 6.35.
1
K= 0.88 (6.19)
GJf
0.85 +
EIxweb
6-53
Parametric Studies and Design Rule Development
1.20
Equation 6.18
Available LSBs
125x45x1.8 LSB
1.00 135x50x1.6 LSB
125x47x2.4 LSB
25090HFB28
25090HFB23
0.80 20090HFB23
Mu/My, Mb/My
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40
Modified Slenderness, Kd
Figure 6.35: Moment Capacities of Hollow Flange Steel Beams with the
Modified Slenderness Parameter K as Defined in Equation 6.19
As seen in Figure 6.35, Equation 6.18 with a lower limit of 0.88 for K gives
reasonable approximations for both LSBs and HFBs. However, some of the FEA
data points of HFBs with high slenderness were below the design curve predicted by
Eq.6.18. Therefore the lower limit of K was increased to 0.90 as shown next.
1
K= 0.90 (6.20)
GJf
0.85 +
EIxweb
1.20
Equation 6.18
Available LSBs
125x45x1.8 LSB
1.00 135x50x1.6 LSB
125x47x2.4 LSB
25090HFB28
25090HFB23
0.80 20090HFB23
Mu/My, Mb/My
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40
Modified Slenderness, Kd
Figure 6.36: Moment Capacities of Hollow Flange Steel Beams with the
Modified Slenderness Parameter K as Defined in Equation 6.20
6-54
Parametric Studies and Design Rule Development
Figure 6.36 shows that Equation 6.18 with a lower limit for K of 0.90 as defined in
Eq. 6.20 is accurate for all the hollow flange steel beams such as HFBs and LSBs.
Hence it is recommended that Equations 6.18 and 6.20 can be used to accurately
predict the member moment capacities of hollow flange steel beams with varying
section geometries.
Although hollow flange beams and LiteSteel beams are the common shapes
introduced by the industry in recent times, other types of hollow flange steel beams
such as Monosymmetric Hollow Flange Beam (MHFB) and Rectangular Hollow
Flange Beam (RHFB) shown in Figure 6.37 are likely to be introduced by the
industry in the future. Three MHFBs and four RHFBs with different GJf/EIxweb
values were therefore considered in this research to investigate the applicability of
the developed design rules for lateral distortional buckling. Their dimensions are
given in Table 6.18. The dimensions of RHFB sections are similar to those of LSBs.
The section slenderness characteristics and the yield stresses of RHFBs and MHFBs
are similar to that of LSBs and HFBs used in this research.
bf bf
df
df
D D
6-55
Parametric Studies and Design Rule Development
The corners were not considered in the analyses of MHFBs and RHFBs as for LSBs.
The torsional constant of flange Jf and the major axis flexural rigidity of the web
EIxweb, and the two important geometrical parameters GJf/EIxweb and K were calculated
for these sections. The relevant calculations are presented in Appendix D.5 while
Table 6.19 presents the results.
Both elastic and nonlinear finite element analyses were carried out for the beams
shown in Table 6.18. Nonlinear finite element analyses were undertaken using the
critical negative imperfection of L/1000, but without any residual stresses as their
residual stresses are not known. Flange and web yield stresses of 550 MPa and 475
MPa, respectively, were considered for MHFBs as they were the nominal yield
stresses of HFBs while the LSB flange and web yield stresses of 450 MPa and 380
6-56
Parametric Studies and Design Rule Development
MPa, respectively were considered in the finite element analyses of RHFBs. It was
observed that the ultimate moments were critical with negative imperfections for
singly symmetrical sections such as LSBs and MHFBs while both positive and
negative imperfections gave the same results for doubly symmetric sections such as
HFBs and RHFBs.
Tables 6.20 and 6.21 present the ultimate moments, elastic lateral buckling moments
and the first yield moments of MHFBs and RHFBs, respectively. The ultimate FEA
moments of all the available 13 LSBs without residual stresses are provided in Table
6.22. The FEA moment capacity results given in these tables are used to plot Figures
6.38 (a) to (d) for different types of hollow flange steel beams but with similar values
of GJf/EIxweb.
6-57
Parametric Studies and Design Rule Development
6-58
Parametric Studies and Design Rule Development
6-59
Parametric Studies and Design Rule Development
1.20
Equation 6.7
3006020RHFB (0.0078)
0.80
Mu/My, Mb/My
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Slenderness, d
1.20
Equation 6.7
250x60x2.0 LSB (0.0149)
200x45x1.6 LSB (0.0116)
1.00
200x45x1.6 RHFB (0.0116)
0.80
MU/My, Mb/My
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Slenderness, d
6-60
Parametric Studies and Design Rule Development
1.20
Equation 6.7
18079MHFB32 (0.0337)
1.00 250x75x2.5 LSB (0.0336)
200x60x2.0 LSB (0.0336)
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Slenderness, d
1.20
Equation 6.7
125x45x2.0 LSB (0.0625)
125x45x1.6 LSB (0.0667)
1.00
21090MHFB38 (0.0665)
125x45x2.0 RHFB (0.0625)
0.80
Mu/My, Mb/My
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Slenderness, d
As seen in these figures, the FEA moment capacity data points of different hollow
flange steel beams follow the same trend provided they have similar GJf/EIxweb (or K)
values. However, in this case, the ultimate moment capacities of these beams were
obtained without considering residual stresses. If the effect of residual stress
6-61
Parametric Studies and Design Rule Development
variation of the hollow flange steel beams is similar, then it can be concluded that the
use of Equation 6.18 with parameter K as defined in Eq.6.20 is suitable for all the
hollow flange steel beams such as LSB, HFB, MHFB and RHFB.
6.9 Conclusions
This chapter has presented the details of a detailed parametric study into the lateral
distortional buckling behaviour of hollow flange steel beams such as LSBs, HFBs
and their variations (MHFBs and RHFBs)). The effects of initial geometric
imperfections and residual stresses were presented first for LSBs. The comparison of
ultimate moment capacities of LSBs from finite element analyses and experimental
studies with the current design rules in AS/NZS 4600 (SA, 2005) showed that the
current design rule was conservative by about 8% in the inelastic buckling region.
New improved design rules were developed for the LSBs based on both FEA and
experimental results, which can be used to predict the moment capacities of LSBs
with a capacity reduction factor of 0.90. The applicability of the developed design
rule was investigated for HFBs and it was found that the design rule developed for
monosymmetric LSBs was very conservative as HFB is a doubly symmetric section.
A geometrical parameter defined as the ratio of flange torsional rigidity to the major
axis flexural rigidity of the web (GJf/EIxweb) was found to be a critical parameter that
reduced the scatter in the FEA data points of hollow flange steel beams in the non-
dimensionalised moment capacity plots based on Mu/My versus d. New design rules
were developed by using a modified slenderness parameter Kd where K was
determined as a function of GJf/EIxweb. The new design rules based on the modified
slenderness parameter Kd were found to be accurate in calculating the moment
capacities of not only LSBs and HFBs but also other types of hollow flange steel
beams such as MHFBs and RHFBs if their residual stress variations are similar to
that of LSBs and HFBs. The developed design rules in this research can be used in
the design of hollow flange steel beams subject to uniform bending while appropriate
moment modification factors developed by Kurniawan and Mahendran (2009b) can
be used for other types of loadings.
6-62
Section Moment Capacity of LSB
CHAPTER 7
7.1 Introduction
The LiteSteel Beams (LSBs) with intermediate and long spans are subjected to
lateral distortional and lateral torsional buckling, respectively, while short span LSBs
exhibit local buckling. Earlier chapters provided the details of experiments and finite
element analyses of LSBs subject to lateral buckling effects and the development of
accurate member moment capacity design rules. In the developed design rules, the
section moment capacity of LSBs was limited to the first yield moment. However,
Mahaarachchi and Mahendrans (2005b) section moment capacity tests revealed that
the moment capacities of compact and non-compact LSB sections could be higher
than their first yield moments. Therefore an attempt was made to investigate the
inelastic reserve capacity of compact and non-compact LSBs. Section moment
capacity tests were carried out on selected compact, non-compact and slender LSB
sections while finite element analyses were conducted for all 13 LSB sections. Table
7.1 presents the section classification for LSBs with corners, which was determined
in accordance with AS 4100 (SA, 1998). The relevant calculations are given in
Appendix E.1. In Table 7.1, S denotes slender sections while NC and C represent
non-compact and compact sections.
7-1
Section Moment Capacity of LSB
For each LSB test beam, the following dimensions, beam depth (d), flange width (bf)
and the flange depth (df) and the thicknesses of LSB plate elements, were carefully
measured using a vernier calliper and a micrometer. Accurate thickness of each plate
element is important to obtain the elastic lateral distortional buckling moment as a
small change in thickness will cause significant changes to the buckling capacities of
LSBs. The LSB plate thicknesses were also accurately measured in the tensile
coupon tests after removing the coating. Table 7.2 presents the details of test
specimens including the measured LSB dimensions and the base metal thicknesses
from tensile coupon tests. The measurements of small corners were not taken as it
was difficult to measure them. Hence the nominal corner dimensions provided by the
manufacturers were used, i.e. the outer radius ro is equal to twice the thickness (2t)
and the inner radius riw is equal to 3 mm. It should be noted that the flange yield
stresses of 300x75x3.0 LSB and 250x60x2.0 LSB were based on the tensile test
results provided by the LSB manufacturers. Table 7.3 presents the measured yield
stresses of outer flange, inner flange and web elements of LSBs.
7-2
Section Moment Capacity of LSB
Test No. LSB Sections fyof (MPa) fyif (MPa) fyw (MPa)
1 150x45x1.6 LSB 557.8 487.5 455.1
2 200x45x1.6 LSB 536.9 491.3 456.6
3 150x45x2.0 LSB 537.6 491.8 437.1
4 250x75x2.5 LSB 552.2 502.2 446.0
5 300x75x3.0 LSB 497.8* 481.5* 440.1*
6 250x60x2.0 LSB 523.0* 473.0* 429.9*
7 300x60x2.0 LSB 557.7 496.3 447.1
Note: fyof outer flange yield stress, fyif inner flange yield stress, fyw web yield
stress.
* from LSB manufacturers (OATM).
Loading
Spreader beam
Load Cell
Rollers
Load transfer plate T-Stiffeners
LSB
T-Stiffeners
7-3
Section Moment Capacity of LSB
T shaped steel plate stiffeners were used to connect the LSBs back to back and to
support and transfer the loads. Steel plates with the same height as the LSB web
element were also attached to the test beam on both sides of the web to avoid any
relative movement between LSBs. All the plates and T-stiffeners were connected to
the web of the test specimens by using 18 mm diameter bolts with a vertical spacing
of 100 mm symmetrically from the centreline and a horizontal spacing of 45 mm.
However, the vertical bolt spacing was limited to 45 mm for 200x45x1.6 LSB,
150x45x2.0 LSB and 150x45x1.6 LSB due to their smaller web depths. Figure 7.2
shows the overall view of the test set-up.
7-4
Section Moment Capacity of LSB
Hydraulic Ram
Load Cell
Spreader
Beam
Back to Back
Test Beam
Lateral
Restraint
T-Shaped
Stiffeners
Test specimens were supported on half rounds placed upon ball bearing as shown in
Figure 7.3. The bottom surfaces of the half rounds and alloy balls were machine
ground and polished to a high degree of smoothness, and smooth ball bearing
surfaces were lubricated to further facilitate the sliding of the half rounds on the ball
bearing when the beam deflected under load. The ends of the test beam were free to
rotate upon the half rounds. Thus it was considered that simply supported conditions
were simulated accurately at the end supports.
7-5
Section Moment Capacity of LSB
The simply supported LSB specimens were tested by loading them symmetrically at
two points through a spreader beam that was loaded centrally by a hydraulic ram and
an Enerpac electrical hydraulic pump. This four-point bending arrangement
provided a central region of uniform bending moment and zero shear force. A 500
kN Phillips load cell connected to the EDCAR data acquisition system was used to
measure the applied load. During the tests, the vertical deflections were measured
using displacement transducers located at the top and bottom flanges of the specimen
at mid-span and loading points. The EDCAR data acquisition system was used to
record the applied load and all the deflections until the failure of specimen. Figure
7.4 shows the details of load application and deflection measurement technique used
in the experiments.
Loading arm
Displacement Transducers
The applied bending moment was calculated by multiplying the measured applied
load with the distance between the support and the loading point. Typical moment
versus deflection curves are shown in Figures 7.5 to 7.7. The deflections measured at
the loading points and the mid-spans are denoted as Load-point and Mid-span in
these figures. Other curves are presented in Appendix E.2.
7-6
Section Moment Capacity of LSB
22
Load-point
20
Mid-span
18
16
14
Moment (kNm)
12
10
0
0 5 10 15 20 25 30 35 40
Deflection (mm)
100
90
80
Load-point
70 Mid-span
60
Moment (kNm)
50
40
30
20
10
0
0 5 10 15 20 25 30 35 40 45
Deflection (mm)
7-7
Section Moment Capacity of LSB
22
20
18
16
Load-point
14 Mid-span
12
Moment (kNm)
10
0
0 5 10 15 20 25 30
Deflection (mm)
All the specimens failed by local buckling of the top compression flange at mid-span
near the maximum load that was followed by a rapid unloading and increased
deflection. Local web buckling was also observed soon after flange local buckling.
Elastic buckling was not observed in any test as the flange elements of all the LSBs
were either compact or non-compact as shown in Table 7.1. Figures 7.5 to 7.7 show
the moment versus deflection curves for compact, non-compact and slender LSBs,
respectively. For the compact section (150x45x2.0 LSB), the vertical deflection
increased linearly with moment until the ultimate moment, which was followed by a
long horizontal plateau as shown in Figure 7.5. This was as expected for a compact
section. For non-compact section (300x75x3.0 LSB), the plateau was not significant
as shown in Figure 7.6 while for slender section (200x45x1.6 LSB), the load dropped
suddenly with increasing deflection after failure as shown in Figure 7.7. The good
agreement with the expected moment versus deflection curves for compact, non-
compact and slender LSB sections confirms the accuracy of the experimental
investigation in relation to loading method and measurements. Typical local buckling
failures observed in the tests are shown in Figures 7.8 to 7.10. Local buckling failure
occurred within the mid-span of the test beam in most cases.
7-8
Section Moment Capacity of LSB
T - Stiffeners
7-9
Section Moment Capacity of LSB
T - Stiffeners
There was no sudden unloading associated with lateral deflection while no specimen
failed due to insufficient material ductility. Although the failure modes of tested
specimens were similar, there were some differences in the way the failure occurred.
For compact LSB sections, large flange deformations and yielding occurred at
moments closer to the failure moment. For non-compact sections, yielding and large
flange deformations appeared to occur earlier while for slender sections, local web
buckling occurred, which was followed by large flange deformations and yielding.
There was no welding failure in this series of tests although Mahaarachchi and
Mahendran (2005b) reported a weld failure in the test of 200x45x1.6 LSB. This
confirms that the welding strength of the new LSB sections is structurally adequate.
The ultimate moments of tested specimens are given in Table 7.5 while the next
section compares these results with the section moment capacities predicted by the
current design rules.
Table 7.5: Ultimate Moments of LSBs
Ultimate Moment
Test No. LSB Sections
Mu (kNm)
1 150x45x1.6 LSB 16.18
2 200x45x1.6 LSB 20.88
3 150x45x2.0 LSB 20.20
4 250x75x2.5 LSB 70.68
5 300x75x3.0 LSB 93.00
6 250x60x2.0 LSB 42.12
7 300x60x2.0 LSB 53.36
7-10
Section Moment Capacity of LSB
The section moment capacity (Ms) of cold-formed steel section is usually based on
the initiation of yielding in the extreme compression fibre in the Australian cold-
formed steel standard AS/NZS 4600 (SA, 2005). The inelastic reserve capacity is
allowed subject to restrictive conditions as discussed in Section 7.4. Effects of local
buckling are accounted for by using the effective widths (be) of slender elements in
compression in the calculation of effective section modulus (Ze). The plate element
slenderness is a function of the applied stress (f *) as shown in Equation 7.2. This
accounts for the reduction in the strength due to local buckling effects with
increasing member slenderness. Clause 3.3.2 of AS/NZS 4600 (SA, 2005) provides
the equation for section moment capacity (Ms) as given in Equation 7.1, where fy is
the yield stress.
Ms = fy Ze (7.1)
1.052 b f *
= (7.2)
k t E
where
= plate slenderness
k = plate buckling coefficient
b = flat width of element excluding radii
t = thickness of the uniformly compressed stiffened elements
f* = design stress in the compression element
7-11
Section Moment Capacity of LSB
0.22
1
= 1.0 (7.4)
The section moment capacities of tested LSBs were calculated based on the AS/NZS
4600 (SA, 2005) design method described above and without considering any
inelastic reserve capacity. It is noted that the design provisions in NAS (AISI, 2004)
are identical to those of AS/NZS 4600 and hence the comparisons and findings are
the same for both codes. It was found that the effective width is equal to the actual
width for all the elements of 13 LSBs sections when their corners are included. The
full section modulus (Z) can therefore be used to calculate the section moment
capacity. Therefore the section moment capacity Ms is equal to the first yield
moment My for all the currently available 13 LSBs when their corners are included.
Sample effective width calculations are given in Appendix D.1. The elastic section
modulus (Z) of tested beams was calculated by using Thin-Wall based on their
measured dimensions. Although the corners of LSBs were not measured, the nominal
corners were used in the calculations. Measured outer flange yield stress was
considered as fy in Equation 7.1. Table 7.6 presents the section moment capacities
(Ms) based on AS/NZS 4600 (SA 2005), ultimate moments from experiments (Mu)
and their ratios (Mu / Ms). It must be noted that in this case Ms is equal to My for all
LSBs. It can be seen that the ratio of Mu/Ms (or Mu/My) for 150x45x2.0 LSB is 1.15,
which is a compact section, while it is about 1.0 for slender sections except
300x60x2.0 LSB. For non-compact sections this ratio is 1.08 on average. This
confirms that there is some inelastic reserve moment capacity for compact and non-
compact sections while slender sections do not have it. The reason why this ratio is
less than unity for 300x60x2.0 LSB is not known. However, this is the most slender
section with very deep web element among the available 13 LSBs. It is likely the
prediction of AS/NZS 4600 (SA, 2005) is high in this case of slender LSB. Figure
7.11 shows the failure mode of 300x60x2.0 LSB which exhibits flange and web local
buckling due to the presence of a slender web element.
7-12
Section Moment Capacity of LSB
Table 7.6: Section Moment Capacities from Tests and AS/NZS 4600 (SA, 2005)
Section moment capacity test results from Mahaarachchi and Mahendran (2005b)
were also considered in this research as the current tests did not include all the
available LSBs. Measured dimensions and yield stresses of Mahaarachchi and
Mahendran (2005b) were used to calculate the elastic section modulus (Z) and the
section moment capacities of LSBs. As decided earlier, nominal corners were
included in the calculations. Tables 7.7 to 7.9 give the details of the tests conducted
by Mahaarachchi and Mahendran (2005b) and the results.
7-13
Section Moment Capacity of LSB
7-14
Section Moment Capacity of LSB
As seen in Table 7.9, the Mu/Ms ratios of compact sections are 1.20 on average
except 200x60x2.5 LSB, which has a value of 1.50 while non-compact sections have
a ratio of 1.15 on average. Slender sections have a ratio of 1.0 on average. It can be
seen that the ratio of Mu/Ms for 300x60x2.0 LSB from Mahaarachchi and
Mahendrans (2005b) test was 0.89, which compares well with the ratio of 0.93 from
the experiments of this research. This confirms the lower section moment capacity
ratio observed with the slender 300x60x2.0 LSB section. The Mu/Ms ratio is below 1
for 300x60x2.0 LSB, which indicates that it could not reach the first yield moment.
This is as predicted by AS 4100 (SA, 1998) for this slender section according to its
classification. However, AS/NZS 4600 (SA, 2005) predicted that 300x60x2.0 LSB
section will reach its first yield moment and hence Ms is equal to My and the Mu/Ms
ratio becomes less than 1.0 for 300x60x2.0 LSB. This implies that AS/NZS 4600
(SA, 2005) is unconservative in predicting the section moment capacities of some
slender LSBs. However, experimental results alone are not sufficient to confirm this.
Based on the section moment capacity test results from Mahaarachchi and
Mahendran (2005b) and this research, it is concluded that only compact and non-
compact LSB sections have inelastic reserve capacity. Since it is not accurate to
develop inelastic reserve capacity design rules based on experimental results alone,
numerical studies were also conducted for all the available 13 LSBs. The following
7-15
Section Moment Capacity of LSB
section provides the details of finite element analyses of LSBs to determine their
section moment capacities.
Although two types of finite element models, ideal and experimental finite element
models, were considered in the investigation of lateral buckling capacities of LSBs as
described in Chapter 5, only the experimental finite element model was considered in
the investigation of the section moment capacities of LSBs. This experimental model
included the actual experimental conditions with measured dimensions and yield
stresses. The results from the experimental finite element models were compared
with test results to validate the finite element models in relation to the element type,
mesh size, initial geometrical imperfections, residual stress, local buckling
deformation and material yielding. Following the validation of the model, idealised
simply supported boundary conditions, nominal dimensions and yield stresses were
applied to this model and analyses were conducted using this model in order to
develop suitable design rules. Details of the experimental finite element models used
this study are described next.
A total of seven section moment capacity tests were carried out in this research and
all of them were modelled using MSC/ PATRAN (MSC Software, 2008) pre-
processing facilities while ABAQUS (HKS, 2007) was used to analyse the models.
MSC/PATRAN (MSC Software, 2008) post-processing facilities were then used to
view the results of ABAQUS analyses. The shell element in ABAQUS (HKS, 2007)
called S4R5 was used to develop the LSB model as in the previous models of LSBs
subject to lateral buckling described in Chapter 5. A mesh size of 5 mm x 10 mm
was selected to be appropriate, i e. 5 mm along the cross section and 10 mm along
the longitudinal direction.
7-16
Section Moment Capacity of LSB
a b/2
Span/2
Figure 7.12 shows the schematic diagram of the experimental finite element model.
In the experimental study, two LSBs were connected back to back and the load was
applied through the spreader beam at mid-span (see Figures 7.1 and 7.2). This type of
arrangement was considered to eliminate the twisting of test beams, which is
considered to be equivalent to the shear centre loading. Such back to back beam
testing is commonly used in section moment capacity tests. However, it is not
necessary to model the actual experimental set-up including two LSBs connected
back to back and the spreader beam. A simplified model of single LSB loaded at its
shear centre was considered to be appropriate as it will simulate the actual
experimental conditions. Therefore, the experimental finite element model shown in
Figure 7.13 was considered to be appropriate in the validation of numerical analyses.
Only half the span was modelled due to the symmetrical nature of loading and
boundary conditions of the test set-up. The material model, mechanical properties
and boundary conditions were the same as for the experimental finite element models
of LSBs described in Chapter 5.
7-17
Section Moment Capacity of LSB
Lateral Restraints
(SPC 345)
Idealised simply supported boundary conditions and a point load were applied at the
shear centre as shown in Figures 7.13 (a) and (b). In the experiments, only the top
flange (compression) was laterally restrained as shown in Figure 7.2. This was
simulated in the finite element model by using the boundary condition of SPC 345 at
7-18
Section Moment Capacity of LSB
all the nodes on the outer face of the flanges where lateral deflection (SPC 3),
twisting (SPC 4) and minor axis rotation (SPC 5) were locked as shown in Figure
7.13 (c). Although the experiments included lateral restraints in the top flange only,
the finite element models included such restraints for both top and bottom flanges as
the preliminary finite element analyses revealed some lateral displacements of
bottom flange (tension flange) due to numerical instability.
The web side plates that were used to connect the LSBs and T-Stiffeners were
modelled as rigid body by using R3D4 elements. In ABAQUS (HKS, 2007) a rigid
body is a collection of nodes and elements whose motion is governed by the motion
of a single node, known as the rigid body reference node. The motion of the rigid
body can be prescribed by applying boundary conditions at the rigid body reference
node. So the simply supported boundary conditions and the load were applied on the
corresponding rigid body reference nodes at the shear centre. Figure 7.14 identifies
the various plate elements with different mechanical material properties as defined in
ABAQUS. The elastic perfect plastic material model with measured yield stresses
was considered in this study.
Outside Flange
Inside Flange
Web
effects. Usually, the initial buckling mode from the elastic buckling analyses is
considered to be critical and this buckling mode is used to apply the initial geometric
imperfections. However, most of the initial elastic buckling modes of LSBs from the
finite element models exhibited a local buckling failure between the support and the
load while higher modes revealed a local buckling failure at mid-span as seen in
Figure 7.15 (a). Therefore, the initial geometric imperfections were applied based on
the higher modes of the elastic buckling analyses for those beams. Both membrane
and flexural residual stresses were also included in the finite element analyses. These
residual stress distributions were the same as used in the LSB models subject to
lateral buckling described in Chapter 5. In the SIGINI Fortran user subroutine (see
Appendix C), the lateral deflections at the top and bottom flanges were set to zero as
there were no lateral displacements.
Figure 7.15: Failure Modes from Finite Element Analyses of 150x45x2.0 LSB
7-20
Section Moment Capacity of LSB
Figure 7.16: Failure Modes from Finite Element Analyses of 300x60x2.0 LSB
Both elastic and non-linear static analyses were carried out for the developed LSB
models. Figures 7.15 and 7.16 show the failure modes of 150x45x2.0 LSB and
300x60x2.0 LSB, respectively, from both elastic and non-linear static analyses. The
failure shape obtained from non-linear static analysis was similar to that exhibited in
the experiment. It can be seen that the ultimate failure mode of 300x60x2.0 LSB
exhibits a web local buckling (Figure 7.16 (b)) while that of 150x45x2.0 LSB
exhibits only yielding (Figure 7.15 (b)). This agreed with the experimental failure
modes of these slender (300x60x2.0 LSB) and compact (150x45x2.0 LSB) sections.
As described above, significant level of web local buckling could be the reason for
the reduced ratio of Mu/Ms for the slender 300x60x2.0 LSB section. It should be
noted that the local buckling failure in these beams occurred at mid-span within the
loading points during the experiments and the corresponding finite element model
was able to simulate the local buckling failure and the location where it occurred.
Further, Figure 7.17 shows the typical post-ultimate failure mode obtained from
7-21
Section Moment Capacity of LSB
finite element analyses. Flange and web local buckling with yielding as seen in this
figure agreed well with the failure mode in the experiments (Figure 7.9). Thus
Figures 7.15 to 7.17 confirm that the developed finite element model accurately
predicts the failure modes of LSBs.
Table 7.10 compares the ultimate moment capacity results from the non-linear finite
element analyses and experiments undertaken in this research. A comparison of FEA
and experimental test results is also provided in the form of bending moment versus
vertical deflection curves in Figures 7.18 and 7.19 for different LSB sections. Other
curves are presented in Appendix E.3. These figures compare the measured
experimental vertical deflections at loading point and mid-span with the
corresponding deflections predicted by FEA. The good agreement between the
results from experiments and finite element analyses indicates that the developed
finite element model is accurate.
7-22
Section Moment Capacity of LSB
18
16
14
Load-point:Exp
12
Mid-span:Exp
Moment (kNm)
Load-point:FEA
10
Mid-span:FEA
0
0 5 10 15 20 25 30
Deflection (mm)
22
20
18
16
14
Moment (kNm)
Load-point:Exp
12 Mid-span:Exp
Load-point:FEA
10 Mid-span:FEA
0
0 5 10 15 20 25 30
Deflection (mm)
The comparisons provided in Table 7.10 and Figures 7.18 and 7.19 demonstrate that
the experimental finite element model predicts the ultimate failure moments of LSBs
accurately. The mean ratio of the ultimate moment capacities from experiments and
finite element analyses was 1.02 with a COV of 0.018. This result suggests that the
model is accurate, considering the possible sources of error caused by unavoidable
differences between the experimental test and finite element model.
7-23
Section Moment Capacity of LSB
An attempt was made to numerically model the section capacity tests carried out by
Mahaarachchi and Mahendran (2005b) based on the experimental finite element
model developed here. The measured dimensions and yield stresses in Tables 7.7 and
7.8 were used in these finite element models. Table 7.11 compares the ultimate
moment capacities from experiments and finite element analyses.
As seen in Table 7.11, most of the FEA ultimate moments are less than the
experimental failure moments. But it does not mean that the finite element model
was inadequate in predicting the ultimate moments of Mahaarachchi and
Mahendrans (2005b) tests as the same model accurately predicted the ultimate
moments of the tests carried out in this research. The main difference between the
two series of tests is in relation to the yield stresses of test specimens. Therefore, it is
suspected that the measured yield stresses provided in Mahaarachchi and Mahendran
(2005b) were not accurate for their section moment capacity test specimens. Also it
appears that the ratio of the outer flange yield stress to web yield stress (fyof/fyw) has
a significant influence on their ultimate moments. When the yield stresses of web,
inner flange and outer flange were taken as 450, 460 and 528 MPa in the finite
element analyses instead of the reported values in Table 7.8 (431, 438 and 528 MPa),
7-24
Section Moment Capacity of LSB
the ultimate moment of 300x75x3.0 LSB was 91.6 kNm (increased from 89.5 kNm).
Such small changes to even the yield stresses of web and inside flange elements
appear to lead to increased ultimate moments. Hence it is possible that the difference
in FEA and experimental ultimate moments might have been caused by differences
in the yield stresses used in the analyses. It is noted that the mean and COV of the
ratio of ultimate moments from experiments and FEA are 1.08 and 0.051,
respectively, even with the yield stresses reported in Table 7.8.
Finite element analyses reported so far included an elastic perfect plastic material
model. This could have lead to the under-estimation of the ultimate moment
capacities of compact LSB sections. Therefore the measured stress-strain
relationships for outside and inside flanges and web elements shown in Figures 7.20
(a) to (c) were used in the non-linear analyses of one of the compact sections,
150x45x2.0 LSB. Measured stress-strain relationships using tensile coupon tests
were simplified as shown in Figures 7.20 (a) to (c), and the true stresses and strains
were also plotted in these figures. Relevant calculations are presented in Appendix
E.4. The ultimate moment capacity from this analysis is 19.0 kNm, which is only 1%
higher than the corresponding value of 18.8 kNm obtained using an elastic perfect
plastic model. This is possibly due to the smaller strain hardening modulus of flange
elements as seen in Figures 7.20 (a) and (b). This result confirms that it is adequate
to use an elastic perfect plastic material model for LSBs.
7-25
Section Moment Capacity of LSB
700.0
600.0
500.0
Engineering
True
Stress, MPa
400.0
300.0
200.0
100.0
0.0
0.0000 0.0050 0.0100 0.0150 0.0200 0.0250 0.0300 0.0350 0.0400 0.0450
Strain
700
600
500
Engineering
True
Stress, MPa
400
300
200
100
0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
Strain
7-26
Section Moment Capacity of LSB
600
500
Engineering
True
400
Stress, MPa
300
200
100
0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09
Strain
(c) Web
Figure 7.20: Stress-Strain Curves of 150x45x2.0 LSB
It is necessary to obtain the section moment capacities of all the available 13 LSBs
with nominal dimensions and yield stresses in order to investigate the inelastic
reserve capacity of each LSB. This can be achieved by using the non-linear analyses
based on the validated finite element model. Next section provides these details.
It was believed that the ideal finite element model used to develop the design curves
for LSBs subject to lateral buckling as provided in Chapter 5 can be used with
reduced spans to obtain the section moment capacities of LSBs. Therefore, an
attempt was made to develop the ideal finite element models of LSBs with nominal
dimensions and yield stresses excluding corners as for the ideal models of LSBs
subject to lateral buckling. However, in this case, the top and bottom flanges were
laterally restrained by using SPC 345 as for the finite element models used in this
chapter to resist any lateral buckling deformations.
Preliminary finite element analyses without residual stresses revealed that the ratios
of ultimate moment (Mu) from FEA to the section moment capacity Ms based on
AS/NZS 4600 (SA, 2005) did not exceed 1.06 even for compact LSBs although they
were about 1.15 from the experiments. If the residual stresses were included, the
7-27
Section Moment Capacity of LSB
ultimate moment ratios would have been less than 1.06. Table 7.12 presents the
ultimate moments from the ideal finite element models without residual stresses for
LSBs with a span of 500 mm. The section moment capacity Ms was calculated by
using AS/NZS 4600 (SA, 2005) without corners as shown in Appendix D.1. Here,
250x75x2.5 LSB is a non-compact section and 200x45x1.6 LSB is a slender section
while 150x45x2.0 LSB is a compact section based on AS 4100 (SA, 1998).
Table 7.12: Ultimate Moments from the Ideal Finite Element Model
As seen in Table 7.12, the ratios of Mu/Ms from the ideal finite element model for
250x75x2.5 LSB and 200x45x1.6 LSB agreed reasonably well with those from
experiments. However, the ratio of Mu/Ms from the ideal finite element model was
only 1.06 while it was 1.15 from the experiment for the compact LSB section. Based
on this, it was concluded that the ideal finite element model developed was not able
to predict the section moment capacity of LSBs. Several attempts were made to
investigate the reason for this and to create an appropriate finite element model.
Finally, it was found that the method used to create a uniform bending moment along
the span based on linearly varying tension and compression nodal forces has not
allowed the section to exceed its yield moment.
The longitudinal stresses across the cross-section of LSBs are shown in Figures 7.21
(a) to (c). These figures also show the stress variations when experimental finite
element models were used with nominal dimensions and yield stresses for the three
LSBs chosen. The stress variations across the section of 150x45x2.0 LSB based on
the experimental finite element model with nominal dimensions and yield stresses
were considered to be appropriate as this shows that most of the web and flange
sections away from the neural axis have yielded, i e. showing section plastification.
However, the stress variation from the ideal finite element model reveals that most of
the web element has not yielded. The stress variation of 250x75x2.5 LSBs also
7-28
Section Moment Capacity of LSB
confirms this. For 200x45x1.6 LSB, the stress variation is similar in both ideal and
experimental models as this is a slender section without any inelastic reserve moment
capacity. Figure 7.21(d) shows the longitudinal stress variation along the length of
150x45x2.0 LSB at ultimate failure based on the experimental finite element model
with nominal dimensions. It can be seen that the stresses are similar along the
longitudinal axis at midspan between the loading points (right side of loading plate in
Figure 7.21 (d)). This also confirms the absence of any torsional moment in LSB
flanges and thus also the accuracy of applying the loads through the shear centre of
LSBs. The stress variations of other LSBs are similar to that of Figure 7.21 (d).
80
60
40
Distance across Section, (mm)
20
0
-500 -450 -400 -350 -300 -250 -200 -150 -100 -50 0 50 100 150 200 250 300 350 400 450 500
-20
Ideal Model
-60
-80
Stress, (MPa)
140
120
100
80
Distance across the Section, (mm)
60
40
20
0
-500 -450 -400 -350 -300 -250 -200 -150 -100 -50 0 50 100 150 200 250 300 350 400 450 500
-20
-100
-120
-140
Stress, (MPa)
Figure 7.21: Stress Variation across the Cross-section of LSB from FEA
7-29
Section Moment Capacity of LSB
120
100
80
60
Distance across the Section, (mm)
40
20
0
-500 -450 -400 -350 -300 -250 -200 -150 -100 -50 0 50 100 150 200 250 300 350 400 450 500
-20
-100
-120
Stress, (MPa)
Figure 7.21: Stress Variation across the Cross-section of LSB from FEA
Based on Table 7.12 and Figures 7.21 (a) to (c), it was decided to use the
experimental finite element model with nominal dimensions and yield stresses to
obtain the section moment capacities of LSBs. All the available 13 LSBs were
analysed using this experimental finite element model with nominal dimensions and
yield stresses. Initial geometric imperfections and residual stresses were also
included. For the tested LSBs, the experimental finite element models were simply
modified by replacing the measured dimensions and yield stresses with nominal
dimensions and yield stresses. Elastic perfect plastic material model with nominal
yield stress was used as the effect of including strain hardening gave only a very
small increase to the moment capacity (<1%). The ultimate moments and the
7-30
Section Moment Capacity of LSB
dimensions a and b of the LSB models used in the analyses are presented in
Table 7.13. Here, a is the distance between the support and loading point while b
is the distance between the loads (see Figures 7.1 and 7.12). The effect of residual
stresses is also presented in this table and was found to be very small with an average
reduction of 2%. It is important to compare the ultimate moments obtained from
finite element analyses with the predictions from the current design rules. Next
section provides these details.
The procedure to calculate the section moment capacity of steel sections is provided
in AS 4100 (SA, 1998), Eurocode 3 Part 1.3 (ECS, 1996 & 2006) and AS/NZS 4600
(SA, 2005). The AS/NZS 4600 (SA, 2005) procedure is given in Section 7.2.3 of this
chapter, which limits the section moment capacity to the first yield moment.
The nominal section moment capacity (Ms) is defined in AS 4100 (SA, 1998) as
follows:
Ms = fy Ze (7.5)
7-31
Section Moment Capacity of LSB
The effective section modulus (Ze) allows for the effects of local buckling. The
section moment capacity (Ms) of a section is governed by the compactness of its
plate elements and is given by Equations 7.6 (a) to (d):
ey
For e > ey (Slender Sections, Flange elements) Ze = Z (7.6d)
e
The element with the greatest ratio e/ey is to be used in calculating the effective
section modulus (Ze). The plate element slenderness (e) is given by Equation 7.7:
b fy
e = (7.7)
t 250
where, b is the clear width of the element outstand from the face or between the faces
of the supporting plate element and t is the thickness.
The effective section modulus of the compact element (Zc) is given by Equation 7.8:
The section moment capacities and the plate slenderness values of LSBs were
calculated based on the above AS 4100 (SA, 1998) procedure. The corners were not
included and the centreline dimensions were used as assumed in finite element
analyses. Sample calculations are presented in Appendix E.5. Based on AS/NZS
4600 (SA, 2005) it was found that some of the LSBs have ineffective horizontal
7-32
Section Moment Capacity of LSB
flange elements (ie. slender elements) when corners were not included. Hence their
section moment capacities are slightly less than their first yield moments. Table 7.14
gives the section compactness of LSBs based on AS/NZS 4600 (SA, 2005) and AS
4100 (SA, 1998). Based on AS/NZS 4600 (SA, 2005), there are only five slender
LSBs as indicated by bold S while the slenderness values of the other three LSBs
denoted as S are very close to the limiting value of 0.673 (refer Appendix D.1 for
further details).
7-33
Section Moment Capacity of LSB
Table 7.15 gives the first yield moment My, section moment capacity Ms based on
AS 4100 (SA, 1998) and AS/NZS 4600 (SA, 2005), and the ratios of Ms/My. As seen
in this table, the section moment capacity predictions of AS 4100 (SA, 1998) for
compact and non-compact LSB sections are more than the first yield moments (i e.,
Ms/My >1). However, these ratios are less than one for slender sections as expected.
Based on AS/NZS 4600 (SA, 2005) predictions, these ratios are unity for compact
sections, but they are 0.98 for slender sections.
Table 7.16 compares the FEA ultimate moment capacities with the section moment
capacity predictions of AS 4100 (SA, 1998) and AS/NZS 4600 (SA, 2005). It can be
seen that the predictions of AS/NZS 4600 (SA, 2005) are conservative by about 10%
for compact and non-compact sections. This is because in the AS/NZS 4600 (SA,
2005) calculations the inelastic reserve capacity was not used. Clause 3.3.2.3 of
AS/NZS 4600 (SA, 2005) states that the inelastic reserve capacity may be used
subject to four conditions. Two of them are: the effect of cold-forming is not
included in determining the yield stress (fy); the ratio of the depth of the compressed
portion of the web to its thickness does not exceed the slenderness ratio 1 defined as
1.11/(fy/E)1/2. Currently available LSBs do not satisfy both these conditions. In
relation to the first condition, if an increased yield stress (fya) based on Clause 1.5.1.2
has already been used to include the effect of cold-forming, the inelastic reserve
capacity cannot be used. However, Clause 1.5.1.2 refers to strength increase resulting
from cold-forming in relation to cold-working of the corners (bends) of cold-formed
sections. The section moment capacities of LSBs are currently not based on the basic
yield strength of 380 MPa (yield strength of parent steel plate). Instead it is based on
a higher flange yield stress of 450 MPa that includes the benefit of significant cold-
working of hollow flange elements, and not due to that of corners as stated in Clause
1.5.1.2. The rectangular and square hollow sections (RHS and SHS) are
manufactured using a very similar method to that of LSBs, and their inelastic reserve
bending capacities are calculated using AS 4100 (SA, 1998) based on the increased
yield stress enhanced by the cold-working of their flange elements. Hence it is
possible to use the available inelastic reserve capacity of LSBs although it does not
satisfy the first condition of Clause 3.3.2.3. In relation to the second condition
relating to web slenderness, Clause 3.3.2.3 appears to be quite restrictive as 1 value
for LSB sections is only about 23.4 (when a yield stress of 450 MPa is used). Other
7-34
Section Moment Capacity of LSB
design codes such as Eurocode 3 Part 1.3 (ECS, 2006) does not have such limits. In
summary although the compact and non-compact LSB sections have inelastic reserve
bending capacities as shown by both experiments and FEA, it is not possible to take
advantage of them using the current AS/NZS 4600 (SA, 2005) design rules.
In the case of some slender sections, AS/NZS 4600 (SA, 2005) predictions are
reasonable, but appear to be slightly less than the ultimate moments from FEA. The
AS 4100 (SA, 1998) predictions are less than both FEA and AS/NZS 4600 (SA,
2005) moment capacities for slender sections, implying that AS 4100 (SA, 1998)
design rules are more conservative for slender sections. However, the predictions of
AS 4100 (SA, 1998) are higher than the moment capacities from FEA for some of
the non-compact and compact sections. In general, AS 4100 (SA, 1998) design rules
appear to predict the available inelastic reserve capacity of non-compact and compact
LSBs reasonably well, considering the observation that FEA predictions are less than
the corresponding experimental capacities.
Table 7.16: Comparison of Ultimate Moment Capacities from FEA and Current
Design Rules
From the above comparisons of section moment capacities of LSBs, AS 4100 (SA,
1998) design rules are more suited for predicting the inelastic reserve bending
7-35
Section Moment Capacity of LSB
capacity of LSBs. However, in principle, they cannot be used for LSBs as they are
cold-formed sections. Hence the European cold-formed steel structures standard
Eurocode 3 Part 1.3 (ECS, 2006 & 1996) was used to calculate the section moment
capacities of LSBs as they do not include a conservative limit for web slenderness.
If Weff = Wel ,
M c , Rd = f yb Wel + 4(W pl Wel )1 e max e 0 / M 0 W pl f yb / M 0
_ _
(7.10)
where Weff, Wel and Wpl are the effective section modulus, gross elastic section
modulus and the plastic section modulus, respectively.
_
max is taken as the slenderness of the element which correspond to the largest value
_ _ _
of e e 0 . The plate element slenderness, p is defined in EN 1993-1-5.
_ _ _
For double supported plane elements e = p and e 0 = 0.5 + 0.25 0.055(3 + )
_ _ _
For outstand elements e = p and e 0 = 0.673 .
_ _ _
For stiffened elements e = p and e 0 = 0.65 .
Table 7.17 presents the section moment capacities (Ms) of LSBs determined using
the above rules. Relevant calculations are presented in Appendix E.6.
7-36
Section Moment Capacity of LSB
The comparison of ultimate moments from FEA and Eurocode 3 Part 1.3 (ECS,
2006) in Table 7.17 shows that the section moments capacities of compact and some
of the non-compact LSBs are predicted well by Eurocode 3 Part 1.3 (ECS, 2006).
However, it is conservative for some non-compact LSBs such as 250x75x2.5 LSB,
200x60x2.0 LSB, 150x45x1.6 LSB and 125x45x1.6 LSB. The horizontal flange
elements of these LSBs were found to be slender as shown in Table E.7 of Appendix
E.6. This is the reason for the underestimation of the section moment capacities of
these LSBs. Further, Eurocode 3 Part 1.3 (ECS, 2006) was conservative for slender
sections. This implies that Eurocode 3 Part 1.3 (ECS, 2006) design rules are more
conservative in the case of slender and some non-compact sections. The FEA
ultimate moments of some compact and non-compact sections indicated by bold
letters in Table 7.17 agreed well with Eurocode 3 Part 1.3 design rules with a mean
of 1.00 and COV of 0.020.
In the above calculations and discussions, a higher yield stress of 450 MPa was used
for LSB outer flanges. However, Eurocode 3 Part 1.3 (ECS, 2006) design rules only
allow the basic yield strength, fyb, to be used. This is similar to the AS/NZS 4600
(SA, 2005) design rues which do not allow the use of an enhanced yield stress due to
cold-forming when calculating the inelastic reserve bending capacity. However,
7-37
Section Moment Capacity of LSB
based on the reasons discussed earlier by comparing LSBs with RHS and SHS
sections, it is proposed that Eurocode 3 Part 1.3 (ECS, 2006) design rules are used to
predict the inelastic reserve capacities of compact and non-compact LSB sections
using the enhanced yield stress of flanges (450 MPa).
The inelastic reserve bending strength of cold-formed steel sections was first
investigated by Reck et. al (1975). Their test results showed that cold-formed steel
sections did not achieve higher inelastic bending capacities like the hot-rolled
sections due to the inability of cold-formed sections to sustain high compressive
strains. The ratio of the compressive strain to yield strain (Cy) was found to be a
function of the compressive flanges width to thickness ratio (b/t). Yener and Pekoz
(1983, 1985) developed design rules to determine the inelastic bending capacity
based on the recommended ratio of compressive strain to yield strain (Cy) as a
function of the b/t ratio of compression elements. These design rules are adopted in
the NAS (AISI, 2004) and AS/NZS 4600 (SA, 2005). Recently Shifferaw and
Schafer (2008) investigated the inelastic bending capacity of conventional open cold-
formed steel members such as C and Z-section beams and proposed suitable design
rules under the direct strength method format. They state that inelastic reserve
bending capacity is available in cold-formed steel beams.
AS/NZS 4600 (SA, 2005) and NAS (AISI, 2004) design rules allow the calculation
of inelastic bending capacity based on the calculated maximum compressive strains.
However, in the last section, the inelastic reserve bending capacities of LSBs were
not calculated based on these rules, as they did not meet the two conditions including
the limit on web slenderness. In their report, Shifferaw and Schafer (2008) state that
the presence of reduced inelastic bending capacity in cold-formed steel beams in
comparison to hot-rolled steel beams is due to higher web to flange area,
unsymmetric sections resulting in first yield occurring in the tension flange and the
inability of cold-formed steel sections to sustain high compressive strains. However,
LSBs despite being cold-formed, do not have the above shortcomings as they are not
7-38
Section Moment Capacity of LSB
0.544
ly
If l < ly , Cy = (7.11)
l
where, ly = 0.776 and l = M y M crl . Here, Mcrl is the elastic local buckling
moment.
They then developed the following equation for the inelastic bending capacity (Mn)
that lie between the yield moment (My) and plastic moment (Mp) capacities.
2.42
Mn M y 1
=1 for Cy >1 (7.12)
Mp My Cy
Shifferaw and Schafer (2008) also gave relevant equations for the inelastic bending
capacity as a function of slenderness using the direct strength method (DSM) format
by combining Eqs. 7.11 and 7.12.
1.32
M n = M y + (M p M y )1 l
If l < ly ,
(7.13)
ly
The DSM based design equation above was used to predict the inelastic reserve
capacity of compact and non-compact LSBs, and the results are shown in Table 7.18
7-39
Section Moment Capacity of LSB
and compared with FEA ultimate moments. This comparison provides mean and
COV values of 1.03 and 0.006 for compact and non-compact LSB sections. This
indicates a good agreement with the proposed equations of Shifferaw and Schafer
(2008). In comparison with Eurocode 3 Part 1.3 (ECS, 2006) design equations, these
equations predicted the section moment capacities of all the compact and non-
compact LSBs more accurately. Therefore it is recommended Shifferaw and
Schafers (2008) design equations are used to predict the inelastic reserve capacities
of compact and non-compact LSBs while the section moment capacities of slender
LSBs can be predicted by AS/NZS 4600 (SA, 2005), NAS (AISI, 2004) and
Eurocode 3 Part 1.3 (ECS, 2006).
7-40
Section Moment Capacity of LSB
Table 7.19 provides the ratios of ultimate FEA moments to the first yield moments of
LSBs without corners. It is quite easy to calculate the actual section moment capacity
of each LSB using the ratios of Mu/My. Table 7.19 also provides the section
compactness of LSBs based on AS 4100 (SA, 1998) in both cases of with and
without corners.
Section Compactness
LSB Sections
Mu/My Without Corners With Corners
300x75x3.0 LSB 1.05 NC NC
300x75x2.5 LSB 1.03 S S
300x60x2.0 LSB 0.98 S S
250x75x3.0 LSB 1.09 NC C
250x75x2.5 LSB 1.07 NC NC
250x60x2.0 LSB 1.02 S S
200x60x2.5 LSB 1.10 NC C
200x60x2.0 LSB 1.07 NC NC
200x45x1.6 LSB 1.02 S S
150x45x2.0 LSB 1.10 C C
150x45x1.6 LSB 1.08 NC NC
125x45x2.0 LSB 1.13 C C
125x45x1.6 LSB 1.11 NC NC
As seen in Table 7.19, compact and non-compact sections exhibited some amount of
inelastic reserve capacity, i e. about 9% on average with a maximum of 13%, while
slender sections do not have any inelastic reserve capacity. However, the ratio of
Ms/My has the values of about 1.18 for compact sections (Table 7.15) such as
200x60x2.5 LSB, 150x45x2.0 LSB and 120x45x2.0 LSB although the FEA results
give only about 1.10 to 1.13. It was found that the shape factor (S/Z) for all the
available LSBs without corners was 1.18 and the relevant calculations are provided
in Appendix E.5. Hence the achievable maximum inelastic reserve moment capacity
is 18% of My, which is only for compact sections. Since there are only two compact
sections in the available 13 LSBs, some non-standard LSBs were created and
analysed using finite element analyses. For this purpose, some slender and non-
compact LSBs were converted to compact sections by increasing their thicknesses
7-41
Section Moment Capacity of LSB
and the details of these non-standard LSBs are presented in Table 7.20. The section
moduli of these beams can be calculated by using Thin-Wall or the procedure shown
in Appendix D.1. The flange and web yield stresses were taken as 450 MPa and 380
MPa as for standard LSBs available in the industry.
Clear
Flange Flange Section
Depth Depth Thickness
Width Depth Modulus
LSB Sections of Web
D d1 bf df T Z
(mm) (mm) (mm) (mm) (mm) (10 mm3)
3
Mp FEA Mu
LSB Sections My (kNm) Mu/My
(kNm) (kNm)
300x75x3.9 LSB 99.24 117.28 107.0 1.08
250x75x3.3 LSB 65.80 77.76 72.5 1.10
200x45x3.0 LSB 31.34 37.08 34.1 1.09
150x45x3.0 LSB 21.05 24.79 23.8 1.13
Table 7.21 presents the first yield moment, the ultimate moment capacity from finite
element analyses and the ratios of Mu/My for these non-standard LSBs. As seen in
this table, the ratios of Mu/My are in the range of 1.08 to 1.13 for these compact LSB
sections. This may be due to the inability of finite element analysis to simulate the
true inelastic bending capacity of steel beams. Mahaarachchi and Mahendrans
(2005b) test results have shown the presence of full plastic moment capacity for
compact LSB sections, ie. many LSBs reached about 1.20 My, which is the plastic
moment capacity for LSBs with corners. Past research (Greiner, 2001) has shown
that finite element modelling based on mechanical properties derived from tensile
coupon tests is unable to capture the full plastic moment capacity of steel beams. The
section moment capacity equations based on Eurocode 3 Part 1.3 (ECS, 2006) and
Shifferaw and Schafer (2008) were able to predict the section moment capacities of
7-42
Section Moment Capacity of LSB
compact and non-compact LSB sections based on FEA. However, this may be due to
the development of these equations based on FEA results. It is possible for compact
cold-formed steel sections such as LSBs to develop their full plastic moment
capacities (Mp). Mahaarachchi and Mahendrans (2005b) test results show that
compact LSB sections are capable of reaching their full plastic moment capacities in
addition to the use of an enhanced flange yield stress of 450 MPa due to cold-
working. However, as a safer conservative approach, it is recommended to use the
results of Mu/My provided in Table 7.19 based on FEA results or Equations 7.9 and
7.10 based on Eurocode 3 Part 1.3 (ECS, 2006) or Equation 7.13 based on Shifferaw
and Schafer (2008). Hence the maximum amount of inelastic reserve capacity for
compact LSB sections is 1.13 My.
Although FEA was not able to predict the plastic moment capacities of four non-
standard compact LSBs (Table 7.21), it was believed that FEA would predict the full
plastic moment capacities of very compact LSBs with thicker plate elements.
Therefore two such sections, 150x45x4.0 LSB and 150x45x5.0 LSB, were analysed.
However, the ratios of Mu/My for these highly compact LSBs were found to be 1.12
in comparison to their Mp/My ratio of 1.17. It was then found that the use of different
yield stresses for flange and web elements also influenced the ultimate moment
capacities of LSBs obtained from FEA. Hence when the same LSB sections
(150x45x4.0 and 150x45x5.0 LSB) were analysed using the same flange and web
yield stress of 450 MPa, FEA predicted their plastic moment capacities. However,
FEA was not able to predict the plastic moment capacity of 150x45x2.0 LSB
although it is classified as compact based on AS 4100 (SA, 1998). Appendix E.7
provides the details of FEA and the results.
An attempt was then made to obtain the section moment capacities of conventional
hot-rolled I- and C-sections. For this purpose, 150UB14.0 and 150PFC17.7 sections
were modelled without corners using the same flange and web yield stress of 320
MPa as shown in their design capacity tables (AISC, 1994). It was found that FEA
was able to predict the full plastic moment capacity of the doubly symmetric
150UB14.0 section, but not that of the monosymmetric 150PFC17.7. The ratio of
Mu/My was only 1.14 for the mono-symmetric hot- rolled PFC section in comparison
to its Mp/My ratio of 1.19 (see Appendix E.7). This contradicts the moment capacities
7-43
Section Moment Capacity of LSB
given in AISC (1994), which recommends the full plastic moment capacity for
150PFC17.7.
Based on the results in Appendix E.7, it can be concluded that conventional finite
element analyses may not able to predict the full plastic moment capacities of
compact mono-symmetric steel sections unless they are made of very thick plate
elements (with small b/t ratios). Further experiments are needed to confirm these
observations.
A comparison of FEA ultimate moments with the predictions from Eurocode 3 Part
1.3 (ECS, 2006) is presented in Table 7.22. As seen in Table 7.22, Eurocode 3 Part
1.3 (ECS, 2006) was not be able predict the section moment capacities of non-
standard compact LSBs from FEA except 250x75x3.3 LSB despite the fact it well
predicted the section moment capacities of standard compact and non-compact LSBs.
It was found that Eurocode 3 Part 1.3 (ECS, 2006) design rules predicted the full
plastic moment capacities for those non-standard LSBs except 250x75x3.3 LSB as
shown in Appendix E.6. This implies that the current Eurocode 3 Part 1.3 (ECS,
2006) design rules allow the full plastic moment capacities for highly compact cold-
formed sections including LSBs. The FEA do not predict the full plastic moment
capacities for these sections and hence resulted in the disagreement between their
results in Table 7.22.
7-44
Section Moment Capacity of LSB
2006) design rules, Shifferaw and Schafers equation does not predict the full plastic
moment capacities and hence appear to be conservative for these very compact
sections. Therefore, it is recommended to use Shifferaw and Schafers (2008) design
rule (Equation 7.13) to predict the section moment capacities of compact and non-
compact LSBs.
The above discussion appears to indicate that compact LiteSteel beams with
torsionally rigid flanges and no free edges have inelastic reserve moment capacity
despite the fact they are cold-formed sections. This inelastic reserve capacity can be
calculated based on Shifferaw and Schafers and Eurocode 3 part 1.3 (ECS, 2006)
design equations. Some current cold-formed steel codes such as AS/NZS 4600 (SA,
2005) and NAS 2007 (AISI, 2007) have restrictions based on slenderness as there is
a concern about the excessive strains at failure that may also lead to fracture in the
section. It is believed that the maximum longitudinal strain at failure should not
exceed three times the yield strain to avoid material fracture (max < 3y). Also, the
inelastic reserve moment that could be achieved will also depend on the maximum
strain at failure. Therefore, an attempt was made to obtain the membrane strain
variation across the LSB cross-sections including the maximum strain. The strain
variations across 150x45x3.0 LSB at the ultimate moment are shown in Figure 7.22.
Strains of the nodes at the edge of the cross section at mid-span were used in plotting
this figure. However, it was found that the strain varied along the flange element as
shown in Figure 7.23.
7-45
Section Moment Capacity of LSB
80
60
40
Distance across Section, (mm)
20
0
-7.0 -6.0 -5.0 -4.0 -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0
-20
-40
-60
-80
Strain (10-3)
As seen in Figure 7.23, the strain values increased from 6.25 x 10-3 to 17.5 x 10-3.
Therefore, the maximum strain (max) on this LSB at failure is 17.5 x 10-3 at the outer
most corner. The yield strain (y) of LSB can be calculated as follows.
7-46
Section Moment Capacity of LSB
In this case, max > 3y for 150x45x3.0 LSB, hence material fracture may occur.
Figure 7.24 shows the membrane strain variation across the section and along one
element at mid-span as a fringe result from finite element analyses. Maximum
longitudinal strains for all the available 13 LSBs and 4 non-standard LSBs were
obtained at ultimate failure from FEA. Table 7.24 presents the maximum membrane
strains and Mu/My ratios of these LSBs. The average membrane strains along the
horizontal flange elements are also included in Table 7.24.
7-47
Section Moment Capacity of LSB
As seen in Table 7.24, the results for 125x45x2.0 LSB and 150x45x3.0 LSB show
that their maximum compressive strains at the ultimate failure are greater than 3y
(6.75 x 10-3). This indicates that these compact LSBs may fail by material fracture
before they reach their higher inelastic moment capacities. However, Mahaarachchi
and Mahendrans (2005b) tests showed that there was no material fracture during the
tests of many compact sections, which reached their full plastic moment capacities
(1.20 My). Also the FEA results showed that the compression flanges yielded first
and that the maximum compressive strains were higher than the maximum tensile
strains in all cases (13 available and 4 non-standard compact LSBs). Experiments
also revealed that the failure was due to inelastic buckling of compression flanges.
Table 7.24 results show that higher Mu/My ratios are achieved for the most compact
sections. Since the slenderness values of compression plate elements of these
sections are small, they are able to reach higher compressive strains at failure and
hence higher moment capacities. Further experiments and numerical modelling are
7-48
Section Moment Capacity of LSB
required before full plastic moment capacities are adopted for compact LSB sections.
It is recommended that the section moment capacity equation (Equation 7.13) based
on Shifferaw and Schafer (2008) is used. Eurocode 3 Part 1.3 design rules can also
be used for this purpose.
7.7 Conclusions
This chapter has presented the details of an experimental investigation, finite element
analyses and a parametric study on the section moment capacities of LSBs. Four
point bending tests were carried out for seven LSBs. Experimental ultimate moment
capacities from this research and Mahaarachchi and Mahendran (2005b) were
compared with those predicted by the current design rules for section moment
capacity based on AS/NZS 4600 (SA, 2005). Appropriate finite element models were
developed and validated using the experimental results. The validated finite element
models with nominal dimensions and yield stresses were used to obtain the section
moment capacities of all the available 13 LSBs. A comparison of the section moment
capacity results from finite element analyses, experiments and design codes showed
that compact and non-compact LSBs based on AS 4100 (SA, 1998) have some
inelastic reserve capacity while slender LSBs do not have any inelastic reserve
capacity beyond their first yield moment. This chapter has presented the section
moment capacities of LSBs based on experiments, finite element analyses and the
current steel design codes and has made some suitable recommendations.
Although the currently available LSBs exceed the slenderness limits and other
conditions of AS/NZS 4600 (SA, 2005) and NAS (AISI, 2007) for inelastic reserve
bending capacity, considerable inelastic bending capacities exist for LSBs as evident
from experiments and finite element analyses of LSBs and should be included in
design. For this purpose, it is recommended that the inelastic bending capacity
7-49
Section Moment Capacity of LSB
equations developed by Shifferaw and Schafer (2008) are used to predict these
capacities for non-compact and compact LSBs. Eurocode 3 Part 1.3 (ECS, 2006)
design equations can also be used. As a simple design approach, it is also possible to
use the ultimate moment capacity of compact LSB sections as 1.10 times their first
yield moment while it is the first yield moment for non-compact sections. For slender
LSB sections, current cold-formed steel codes can be used to predict their section
moment capacities.
7-50
Effect of Web Stiffeners on the Lateral Buckling of LSBs
CHAPTER 8
8.1 Introduction
The LiteSteel Beams (LSBs) with intermediate and long spans are subjected to
lateral distortional and lateral torsional buckling, respectively. Lateral distortional
buckling occurs due to the presence of torsionally rigid rectangular flanges and a
relatively slender web. Simultaneous lateral displacement, section twist and web
distortion occur during this lateral distortional buckling of LSB as seen in Figure 8.1.
Lateral distortional buckling significantly reduces the flexural moment capacity of
LSBs with intermediate spans as shown in Chapter 6. Such moment capacity
reduction can be eliminated if the observed web distortion in LSBs is eliminated or
reduced. Past research (Avery and Mahendran, 1997, Mahendran and Avery, 1997)
has shown that the use of web stiffeners reduces web distortion and hence improves
the flexural moment capacity of hollow flange steel beams such as Hollow Flange
Beams (HFB) (see Figure 8.2). Avery and Mahendran (1997) stated that web
stiffeners act to prevent web distortion by coupling the rotational degrees of freedom
of the top and bottom flanges of HFBs. Hence they found that simple plate stiffeners
welded or screw fastened to only the top and bottom flanges were able to reduce the
web distortion and thus improve the lateral buckling moment capacities of HFBs.
(a) Experiments
8-1
Effect of Web Stiffeners on the Lateral Buckling of LSBs
Section Twist
Web Distortion
Lateral
Displacement
(b) FEA
Figure 8.2: Use of Web Stiffeners in HFBs (Mahendran and Avery, 1997)
Although Avery and Mahendran (1997) and Mahendran and Avery (1997) showed
that the use of web stiffeners significantly improved the flexural moment capacity of
HFBs using both large scale experiments and finite element analyses, Kurniawans
(2005) investigations on LSBs produced some conflicting outcomes. His
experimental studies based on quarter point loading showed that the use of web
stiffeners did not significantly improve the flexural moment capacity of LSBs while
his finite element analyses based on an ideal finite element model of LSB with ideal
support conditions and a uniform moment gave improved buckling moment
capacities. It is unlikely that this conflicting outcome was caused by the difference in
moment distributions, in which case, the experimental studies should have given a
8-2
Effect of Web Stiffeners on the Lateral Buckling of LSBs
higher elastic buckling capacity. Instead it may be due to the lack of flange twist
restraints at the supports in his lateral buckling experiments (see Figure 8.3 (a)). The
use of web side plates alone was unable to provide the required flange twist restraint
(FTR).
(a) Support without FTR (b) Local Flange Twist (c) Support with FTR
As described in Chapter 4, this leads to local flange twists at the supports as shown in
Figure 8.3 (b) and thus does not produce the ideal simply supported boundary
conditions in which the entire section has full twist restraint. Transverse web
stiffeners were provided at the supports for most of the lateral buckling tests
described in Chapter 4 in order to provide full twist restraint at the supports as shown
in Figure 8.3 (c). The difference between the support conditions in Kurniawans
(2005) experimental and finite element analyses might have caused the moment
capacity differences observed by him. Further, Kurniawans (2005) finite element
analyses were limited to elastic buckling analyses while his experimental study was
also limited to one LSB section. Therefore a thorough investigation is required to
investigate the effect of web stiffeners on the lateral distortional buckling and
ultimate strength behaviour of LSBs. For this purpose the validated finite element
models of LSBs developed in Chapter 5 were used by including the required web
stiffeners at the supports and appropriate locations within the span. It is important to
investigate the reasons for the conflicting outcomes of Kurnaiawan (2005) and then
to determine the most suitable and cost-effective type, size and spacing of the
8-3
Effect of Web Stiffeners on the Lateral Buckling of LSBs
required web stiffeners that will provide the improved lateral distortional buckling
capacities for LSBs. This was first undertaken using a series of elastic buckling
analyses. Both elastic and non-linear lateral buckling analyses were then undertaken
for the chosen web stiffener arrangement and suitable design rules were also
developed. This chapter presents the details of this investigation and the results.
Avery and Mahendran (1997) and Mahendran and Avery (1997) found that the use of
5 mm thick steel plate stiffeners screwed or welded to the top and bottom flanges of
HFBs (Hollow Flange Beams) at third points within the span (Figure 8.2) was the
most optimum arrangement to improve the lateral buckling moment capacities based
on their experimental and finite element analyses. Kurniawan (2005) also found that
the use of 5 mm steel plate stiffeners at third points within the span improved the
lateral buckling moment capacity of LSBs based on his finite element analyses. They
considered various types of web stiffeners such as angle sections, threaded rod
fasteners, square hollow sections, LSBs and rectangular hollow sections (see Figure
8.4) and concluded that steel plates screwed or welded to the hollow flanges
provided the most simple and cost-effective web stiffener arrangement. Further they
reported that the use of other types of web stiffeners and arrangements did not
increase the lateral buckling moment capacities of hollow flange steel beams with
compared to their cost. Therefore 5 mm steel plate web stiffeners were considered in
this research. However, the use of this plate web stiffener with LSBs as used in the
experiments of Kurniawan (2005) must be investigated first since he stated that the
use of web stiffeners did not significantly improve the buckling moment capacity of
LSB. As mentioned earlier in this chapter, the use of additional web stiffeners at the
supports to provide the required flange twist restraint is likely to eliminate this
problem.
8-4
Effect of Web Stiffeners on the Lateral Buckling of LSBs
Figure 8.4: Types of Web Stiffeners Used by Avery and Mahendran (1997) and
Kurniawan (2005)
8-5
Effect of Web Stiffeners on the Lateral Buckling of LSBs
Figure 8.4: Types of Web Stiffeners Used by Avery and Mahendran (1997) and
Kurniawan (2005)
Two types of finite element models were used in this research, namely, ideal and
experimental finite element models as shown in Figure 8.5. Ideal models of LSBs
were based on ideal simply support conditions and a uniform moment. Ideal simply
supported boundary conditions were implemented by fixing the vertical and lateral
deflections and twist of the section at the supports. Experimental finite element
models were used to simulate the LSBs as used in the lateral buckling experiments
with quarter point loading. Chapter 5 provides the details of these models of LSBs
8-6
Effect of Web Stiffeners on the Lateral Buckling of LSBs
without web stiffeners that have been validated using the experimental results in
Chapter 4. These validated models were modified by including the required web
stiffeners in this research. Nominal dimensions of LSBs were used in the analyses.
Steel plates with 5 mm thickness and a yield stress of 300 MPa were considered at
the supports and one third points of the beam span. The plate stiffeners at the
supports provided the required flange twist restraint. Figure 8.6 shows the
experimental finite element model of LSB with web stiffeners.
Span/2
Span/4
Span/2
(b) Experimental Model
Figure 8.6: Experimental Finite Element Model of LSB with Web Stiffeners
8-7
Effect of Web Stiffeners on the Lateral Buckling of LSBs
Web Side
Plate
Loading Plate
Web stiffener
providing
flange twist
restraint Web Stiffener
Mid-Span Boundary
5 mm Gap Condition, SPC 156
MPC
Web Stiffener
Support Boundary
Condition, SPC 234
Figure 8.6: Experimental Finite Element Model of LSB with Web Stiffeners
Shell elements of 5 mm width and 10 mm length were used as shown in Figure 8.6
(a). Figure 8.6 (b) shows the various plates used in the experimental finite element
model with web stiffeners at supports and suitable locations within the span. It also
includes the usual web side plates used at the supports. Figure 8.6 (c) shows the cross
sectional view of LSB with web stiffeners, which includes the support and mid-span
boundary conditions and loading. The loading and boundary conditions were the
same as used in the experimental finite element models of LSBs described in Chapter
5. In the finite element models of stiffened LSBs, the web stiffeners were connected
to the inner flange surface by a process of equivalencing the nodes of the web
stiffener plate and the nodes of the inner surface of the flange so that the web
8-8
Effect of Web Stiffeners on the Lateral Buckling of LSBs
stiffener plate and the flange can act as an integral member. The welding process was
not modelled as it was decided to recommend a tack weld and the effects of this
welding on the residual stresses were considered to be negligible. A 5 mm gap was
provided between the stiffener and the web element as it is not practical to provide
the stiffener next to the web due to the corners present in LSBs.
Figure 8.7: Experimental FE Model with Web Stiffeners and Flange Twist
Restraints
8-9
Effect of Web Stiffeners on the Lateral Buckling of LSBs
Figure 8.7: Experimental FE Model with Web Stiffeners and Flange Twist
Restraints
Web Stiffener
Figure 8.8: Ideal Finite Element Model with Full Twist Restraint at the
Supports (Including Flanges) and Web Stiffeners
8-10
Effect of Web Stiffeners on the Lateral Buckling of LSBs
Figure 8.8 shows the typical ideal finite element model of LSBs with web stiffeners.
It was used in the parametric study after resolving the conflicting outcomes from
Kurniawan (2005). Hence flange twist restraints at the supports were not modelled
explicitly using web stiffeners as shown in Figures 8.7 (b) and (d). Instead they were
included in the models via idealised simply supported boundary conditions which
provide full twist restraint to the entire section at the supports as shown in Figure 8.8.
The idealised boundary conditions at the support and the boundary condition of
symmetric plane are presented in Table 8.1. The presence of symmetry allowed the
use of only half the span, which reduced the analysis time. In Table 8.1, T and R
represent the translation and rotation, respectively and the subscripts (1, 2, and 3)
represent the direction while field Yes means that it is free to move in that
direction. Figure 8.9 illustrates the global axes selected to input the boundary
conditions for the analysis. The section twist was restrained by fixing the X axis
rotation (SPC 4) while the vertical and lateral displacements were also fixed at all the
nodes of the end-span of LSB as shown in Figure 8.8. Therefore, additional web
stiffener plates are not needed at the support as the idealised simply supported
boundary condition provides the required flange twist restraint and eliminate the
local flange twist (Figure 8.3 (b)).
T1 T2 T3 R1 R2 R3
One end Yes No No No Yes Yes
Other end No No No No Yes Yes
Mid span No Yes Yes Yes No No
Y, 2 Y, 2
X, 1 Z, 3 L/2
Z, 3 X, 1
8-11
Effect of Web Stiffeners on the Lateral Buckling of LSBs
To simulate a uniform end moment across the section, linear forces were applied at
every node of the beam end, where the nodes above the middle of the web were
subject to tensile forces while the nodes below the middle of the web were subject to
compressive forces. The force at the middle of the web was zero and was linearly
increased within the cross section as shown in Figure 8.10. A tensile force of 1000 N
and a compressive force of 1000 N were applied at the nodes on the top and bottom
faces of LSB cross section. This loading method was the same as that used in the
finite element models of LSBs without web stiffeners, as described in Chapter 5.
8-12
Effect of Web Stiffeners on the Lateral Buckling of LSBs
8.2.2 Results
Table 8.2 presents the elastic lateral distortional buckling moments of LSBs with
varying web stiffener arrangements from both experimental and ideal finite element
models described in the previous section. It includes the elastic lateral distortional
buckling moments (Mod) from the ideal finite element model (as given in Chapter 6),
the elastic torsional buckling moment Mo calculated using Eq.8.1, and the values of
Mod with various arrangements of web stiffeners as obtained from the experimental
finite element models.
The elastic lateral torsional buckling moment Mo can be calculated by using the
following equation.
2 EIy 2 EIw
Mo = GJ + (8.1)
L2 L2
where
EIy = minor axis flexural rigidity
EIw = warping rigidity
GJ = torsional rigidity
L = span
Table 8.2: Elastic Lateral Distortional Buckling Moments of LSBs with Web
Stiffeners
Mod Mod (kNm)
Mo
LSB Span (kNm) Experimental Model
(kNm)
Sections (mm) Ideal No WS WSs WSTP WS
Eq. 8.1
Model (a) (b) (c) (d)
300x60x2.0 3000 33.95 22.99 22.05 23.80 25.02 29.06
LSB 4000 24.66 18.36 17.55 19.46 19.66 22.81
200x45x1.6 3000 10.68 8.33 8.14 8.95 9.00 10.18
LSB 4000 7.89 6.67 6.43 7.05 6.89 7.63
150x45x2.0 2000 18.35 14.52 12.15 14.30 13.01 15.81
LSB 3000 11.96 10.48 9.01 10.42 9.39 10.93
WSs Web stiffeners at the supports providing flange twist restraint,
WSTP Web stiffeners at third points within the span,
WS Web stiffeners at the supports and at third span points.
8-13
Effect of Web Stiffeners on the Lateral Buckling of LSBs
Table 8.3: Effect of Web Stiffener Arrangements on the Results of Mod from
Experimental Finite Element Models
Span WSs / No WSTP / No WS / WS / No
LSB Sections
(mm) WS WS WSs WS
300x60x2.0 3000 1.08 1.13 1.22 1.32
LSB 4000 1.11 1.12 1.17 1.30
200x45x1.6 3000 1.10 1.11 1.14 1.25
LSB 4000 1.10 1.07 1.08 1.19
150x45x2.0 2000 1.18 1.07 1.11 1.30
LSB 3000 1.16 1.04 1.05 1.21
WSs Web stiffeners at the supports providing flange twist restraint,
WSTP Web stiffeners at third points within the span,
WS Web stiffeners at the supports and at third span points.
As seen in Tables 8.2 and 8.3, the elastic lateral distortional buckling moments (Mod)
were increased by 19 to 32% for the LSBs and spans considered here when web
stiffeners were used at the third points within the span and the supports. However,
when the web stiffeners were used only at third span points of span without any
stiffeners at the supports (without flange twist restraint at the support), the
improvement to elastic lateral distortional buckling moment was not significant. This
difference is small for 200x45x1.6 LSB with 4 m span and 150x45x2.0 LSB with 3
m span because these LSBs exhibit lateral torsional buckling with very small web
distortion for these spans. The results in the Tables 8.2 and 8.3 clearly demonstrate
the need to use web stiffeners at both the supports an third span points.
Figures 8.11 (a) to (d) show the elastic lateral distortional buckling modes obtained
for 200x45x1.6 LSB section from finite element analyses based on the experimental
finite element model with various configurations of web stiffeners. A comparison of
Figures 8.11 (a) and (b) clearly demonstrates that the use of web stiffeners at the
supports significantly reduced the local flange twist at the support. Comparison of
Figures 8.11 (a) and (c) shows that web distortion was reduced when web stiffeners
were used at third span points. Although the use of web stiffeners at third span points
reduced the web distortion the use of web stiffeners at the supports is also important
as this further improved the moment capacities by avoiding local flange twist as
shown in Figure 8.11 (d).
8-14
Effect of Web Stiffeners on the Lateral Buckling of LSBs
Web Distortion
Figure 8.11: Elastic Lateral Distortional Buckling Failure Modes of LSBs with
Various Stiffener Arrangements
8-15
Effect of Web Stiffeners on the Lateral Buckling of LSBs
Figure 8.11: Elastic Lateral Distortional Buckling Failure Modes of LSBs with
Various Stiffener Arrangements
8-16
Effect of Web Stiffeners on the Lateral Buckling of LSBs
Based on these finite element elastic buckling analyses (Tables 8.2 and 8.3 and
Figure 8.11), it is concluded that the use of web stiffeners at every third point within
the span can effectively improve the lateral distortional buckling moment capacity of
LSBs provided web stiffeners are also used at the supports. This simulates the
idealised simply supported boundary conditions with full twist restraint. These
results also provide the explanation why Kurniawans (2005) experimental and
numerical analyses gave conflicting outcomes in relation to the buckling capacity
improvements due to web stiffeners.
Having confirmed the effectiveness of using web stiffeners in improving the lateral
distortional buckling moment capacities of LSBs, it is now necessary to investigate
the optimum size and spacing of the required plate web stiffeners. For this purpose, a
series of elastic buckling analyses was conducted using the ideal finite element
model to investigate the lateral distortional buckling moment capacities of LSBs as a
function of web stiffener thickness and spacing. The following section provides the
details of these analyses and the results.
It was decided to use steel plates welded to the inner faces of top and bottom flanges
as web stiffeners. However, the location or the number and spacing of web stiffeners,
which effectively improve the lateral distortional buckling moment capacity, has to
be determined. It is obvious that the moment capacity will increase with increasing
number of web stiffeners. However, the cost will also increase with it. Therefore a
series of elastic buckling analyses was undertaken for LSBs with 5 mm thick plate
stiffeners at varying spacings of span/2, span/3 and span/4 as shown in Figures 8.12
(a) to (c) using the ideal finite element model and the results are presented in Table
8.4.
It should be noted that there is no need of steel plates at the support for these ideal
finite element models as the idealised simply supported boundary conditions provide
the required restraints against flange twist as mentioned earlier in this chapter.
8-17
Effect of Web Stiffeners on the Lateral Buckling of LSBs
(a) Span/2
Web Stiffeners (5 mm thick)
(b) Span/3
(c) Span/4
Table 8.4: Effect of Web Stiffener Spacing on the Elastic Distorional Buckling
Moments of LSBs in kNm
8-18
Effect of Web Stiffeners on the Lateral Buckling of LSBs
As seen in Table 8.4, the elastic lateral distortional buckling moments (Mod) increase
with decreased stiffener spacing. The ratio of Mod values for span/3 and span/2 was
about 1.05 for intermediate spans while they were about 1.02 for span/4 and span/3.
This indicates that the degree of improvement to Mod is not significant when the web
stiffener spacing was reduced from span/3 to span/4. An additional web stiffener thus
only provides about 2% increase in Mod. Therefore span/3 was considered to be the
optimum web stiffener spacing based on both member capacity and cost. Avery and
Mahendran (1997) also made a similar recommendation based on their elastic
buckling studies of HFBs with web stiffeners.
(a) Span/2
(b) Span/3
8-19
Effect of Web Stiffeners on the Lateral Buckling of LSBs
(c) Span/4
Figure 8.13: Elastic Lateral Buckling Modes of LSBs
Figures 8.13 (a) to (c) show the elastic lateral distortional buckling modes of a
300x75x3.0 LSB with 4 m span with various web stiffener spacings. They show that
the level of web distortion was decreased with increasing number of web stiffeners.
By considering both the cost and the capacity improvement, web stiffener spacing of
span/3 was considered to be adequate. Table 8.4 clearly demonstrates this as the
maximum improvement is only about 3% when the web stiffeners are used at a
spacing of span/4 when compared to that of span/3.
For the chosen web stiffener spacing, it is important to investigate the effects of
different sizes (thicknesses) of web stiffeners and to determine the optimum size.
Four thicknesses of 3 mm, 4 mm, 5 mm and 10 mm were considered in this
investigation. Table 8.5 presents the elastic lateral distortional buckling moments
(Mod) of various web stiffener sizes for all the available 13 LSBs. The buckling
moment capacities increase with increasing thickness of web stiffeners for
intermediate spans while this increment is very small for long spans for which web
distortion is small. Buckling moment capacity improvement was about 1.4 % for
intermediate spans when 5 mm web stiffeners were replaced with 10 mm web
stiffeners. Therefore, the use of 10 mm web stiffeners cannot be justified.
8-20
Effect of Web Stiffeners on the Lateral Buckling of LSBs
Table 8.5: Effect of Web Stiffener Sizes on the Elastic Lateral Distortional
Buckling Moments of LSBs
8-21
Effect of Web Stiffeners on the Lateral Buckling of LSBs
Table 8.5 (cont.): Effect of Web Stiffener Sizes on the Elastic Lateral
Distortional Buckling Moments of LSBs
Modw (kNm) % Increase
LSB Span 4 mm 5 mm 10 mm
Sections (mm) 3 mm 4 mm 5 mm 10 mm vs vs vs
3 mm 4 mm 5 mm
1000 113.70 114.33 114.81 116.63 0.55 0.42 1.59
2000 50.43 50.61 50.75 51.27 0.36 0.28 1.02
3000 33.14 33.21 33.27 33.49 0.21 0.18 0.66
200x60x2.5
4000 24.91 24.94 24.96 25.08 0.12 0.08 0.48
LSB
6000 16.81 16.81 16.82 16.86 0.00 0.06 0.24
8000 12.73 12.73 12.74 12.76 0.00 0.08 0.16
10000 10.25 10.25 10.25 10.27 0.00 0.00 0.20
1500 56.33 56.59 56.79 57.59 0.46 0.35 1.41
2000 41.11 41.27 41.38 41.85 0.39 0.27 1.14
3000 27.09 27.16 27.21 27.41 0.26 0.18 0.74
200x60x2.0
4000 20.38 20.41 20.43 20.54 0.15 0.10 0.54
LSB
6000 13.78 13.79 13.79 13.84 0.07 0.00 0.36
8000 10.47 10.48 10.48 10.50 0.10 0.00 0.19
10000 8.46 8.46 8.46 8.48 0.00 0.00 0.24
1500 20.33 20.42 20.49 20.89 0.44 0.34 1.95
2000 14.73 14.78 14.82 15.05 0.34 0.27 1.55
3000 9.65 9.67 9.69 9.79 0.21 0.21 1.03
200x45x1.6
4000 7.24 7.25 7.26 7.31 0.14 0.14 0.69
LSB
6000 4.88 4.89 4.89 4.91 0.20 0.00 0.41
8000 3.70 3.70 3.70 3.72 0.00 0.00 0.54
10000 2.98 2.98 2.98 3.00 0.00 0.00 0.67
750 50.70 50.99 51.22 52.34 0.57 0.45 2.19
1000 35.41 35.59 35.74 36.44 0.51 0.42 1.96
1500 22.47 22.57 22.61 22.93 0.45 0.18 1.42
2000 16.65 16.69 16.73 16.90 0.24 0.24 1.02
150x45x2.0
3000 11.10 11.12 11.13 11.20 0.18 0.09 0.63
LSB
4000 8.39 8.39 8.40 8.44 0.00 0.12 0.48
6000 5.66 5.67 5.67 5.68 0.18 0.00 0.18
8000 4.28 4.28 4.28 4.29 0.00 0.00 0.23
10000 3.38 3.38 3.38 3.39 0.00 0.00 0.30
1000 28.80 28.94 29.07 29.71 0.49 0.45 2.20
1500 18.35 18.42 18.47 18.77 0.38 0.27 1.62
2000 13.63 13.67 13.70 13.86 0.29 0.22 1.17
150x45x1.6 3000 9.11 9.12 9.14 9.21 0.11 0.22 0.77
LSB 4000 6.89 6.90 6.91 6.94 0.15 0.14 0.43
6000 4.68 4.68 4.68 4.70 0.00 0.00 0.43
8000 3.55 3.55 3.55 3.56 0.00 0.00 0.28
10000 2.86 2.86 2.86 2.86 0.00 0.00 0.00
8-22
Effect of Web Stiffeners on the Lateral Buckling of LSBs
Table 8.5 (cont.): Effect of Web Stiffener Sizes on the Elastic Lateral
Distortional Buckling Moments of LSBs
Modw (kNm) % Increase
LSB Span 4 mm 5 mm 10 mm
Sections (mm) 3 mm 4 mm 5 mm 10 mm vs vs vs
3 mm 4 mm 5 mm
750 45.78 46.04 46.25 47.25 0.57 0.46 2.16
1000 32.84 33.00 33.13 33.72 0.49 0.39 1.78
2000 15.98 16.02 16.04 16.18 0.25 0.12 0.87
125x45x2.0 3000 10.74 10.75 10.76 10.82 0.09 0.09 0.56
LSB 4000 8.14 8.14 8.15 8.18 0.00 0.12 0.37
6000 5.50 5.50 5.50 5.51 0.00 0.00 0.18
8000 4.15 4.15 4.15 4.16 0.00 0.00 0.24
10000 3.33 3.33 3.33 3.34 0.00 0.00 0.30
1000 26.74 26.88 26.99 27.53 0.52 0.41 2.00
2000 13.11 13.14 13.16 13.29 0.23 0.15 0.99
3000 8.82 8.83 8.84 8.90 0.11 0.11 0.68
125x45x1.6
4000 6.70 6.70 6.71 6.74 0.00 0.15 0.45
LSB
6000 4.55 4.55 4.56 4.57 0.00 0.22 0.22
8000 3.45 3.45 3.45 3.46 0.00 0.00 0.29
10000 2.77 2.77 2.77 2.78 0.00 0.00 0.36
Based on these elastic buckling analyses of LSBs with web stiffeners, it was decided
to use 5 mm thick plate web stiffeners, welded to the inner faces of top and bottom
flanges at third points within the span and supports as the optimum web stiffener
configuration. It was then decided to obtain the elastic lateral distortional buckling
moments and the nonlinear ultimate moments of all the 13 LSBs using the ideal
finite element model shown in Figure 8.8. The following section presents the details
of the elastic buckling analyses of LSBs with web stiffeners.
The ideal finite element model of LSBs with the optimum web stiffener arrangement
was considered in these elastic buckling analyses. As explained earlier, web
stiffeners were not explicitly modelled at the supports, instead they were simulated
via idealised simply supported conditions that provided full twist restraint. The
lateral distortional buckling mode obtained from these analyses revealed reduced
web distortion when compared to that of LSBs without web stiffeners. Figure 8.14
8-23
Effect of Web Stiffeners on the Lateral Buckling of LSBs
shows the elastic lateral distortional buckling mode of 1500 mm span 150x45x2.0
LSB with web stiffeners. It shows that the web distortion is small.
Web Stiffener
Figure 8.14: Elastic Lateral Distortional Buckling of LSB with Web Stiffener
Some LSBs exhibited a local buckling failure mode in the case of some intermediate
spans despite the fact they exhibited a lateral distortional buckling mode without web
stiffeners. This demonstrates that lateral distortional buckling is delayed for some
LSBs by using web stiffeners. Table 8.6 presents the elastic lateral torsional buckling
moments (Mo), elastic lateral distortional buckling moments without web stiffeners
(Mod) and the elastic lateral distortional buckling moments (Modw) of LSBs with web
stiffeners. It also compares the ratios of these buckling moments.
8-24
Effect of Web Stiffeners on the Lateral Buckling of LSBs
8-25
Effect of Web Stiffeners on the Lateral Buckling of LSBs
8-26
Effect of Web Stiffeners on the Lateral Buckling of LSBs
As seen in Table 8.6, the use of web stiffeners improved the elastic lateral
distortional buckling moment capacities. The ratio of Modw/Mod reduces with
increasing span while the ratio of Modw/Mo increases. The elastic lateral distortional
buckling moment of stiffnened LSBs (Modw) approaches the elastic lateral torsional
buckling moment (Mo) with increasing spans. The ratio of Modw/Mo of 0.97 reveals
this fact. It is important to develop a relationship between Mo and Modw in order to
8-27
Effect of Web Stiffeners on the Lateral Buckling of LSBs
calculate Modw without undertaking any finite element analyses. For this purpose, the
ratio of Modw/Mo was plotted against span in Figure 8.15 while Figure 8.16 shows the
variation of this ratio with non-dimensional slenderness = (My/Mo)1/2 .
1.00
0.90
300x75x3.0 LSB
0.80 300x75x2.5 LSB
300x60x2.0 LSB
0.70 250x75x3.0 LSB
250x75x2.5 LSB
0.60
250x60x2.0 LSB
Modw/Mo
200x60x2.5 LSB
0.50
200x60x2.0 LSB
0.40 200x45x1.6 LSB
150x45x2.0 LSB
0.30 150x45x1.6 LSB
125x45x2.0 LSB
0.20 125x45x1.6 LSB
0.10
0.00
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000
Span, (mm)
Figure 8.15: Modw/Mo versus Span for LSBs with Web Stiffeners
1.00
0.90
300x75x3.0 LSB
0.80 300x75x2.5 LSB
300x60x2.0 LSB
0.70 250x75x3.0 LSB
250x75x2.5 LSB
0.60 250x60x2.0 LSB
Modw/Mo
200x60x2.5 LSB
0.50
200x60x2.0 LSB
200x45x1.6 LSB
0.40
150x45x2.0 LSB
0.30 150x45x1.6 LSB
125x45x2.0 LSB
0.20 125x45x1.6 LSB
0.10
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40
Slenderness ()
Figure 8.16: Modw/Mo versus Slenderness for LSBs with Web Stiffeners
The first yield moment My was calculated for LSBs without corners as described in
Appendix D.1 and the values are presented in Chapter 6. However, Table 8.7
presents the first yield moments (My) of all the available 13 LSBs.
8-28
Effect of Web Stiffeners on the Lateral Buckling of LSBs
My
LSB Sections
(kNm)
300x75x3.0 LSB 77.24
300x75x2.5 LSB 64.79
300x60x2.0 LSB 45.17
250x75x3.0 LSB 60.06
250x75x2.5 LSB 50.38
250x60x2.0 LSB 35.10
200x60x2.5 LSB 31.98
200x60x2.0 LSB 25.79
200x45x1.6 LSB 17.23
150x45x2.0 LSB 14.35
150x45x1.6 LSB 11.58
125x45x2.0 LSB 11.15
125x45x1.6 LSB 9.00
Based on the variation of FEA data points in Figure 8.16, two possible equations
were developed. Equation 8.2 was a linear equation while Equation 8.3 was a second
order polynomial equation and the relevant curves are shown in Figures 8.17 and
8.18.
1.0
0.9
FEA
0.8 Equation 8.2
0.7
0.6
Modw/Mo
0.5
0.4
0.3
0.2
0.1
0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4
Slenderness ()
8-29
Effect of Web Stiffeners on the Lateral Buckling of LSBs
1.0
0.9
FEA
0.8
Equation 8.3
0.7
0.6
Modw/Mo
0.5
0.4
0.3
0.2
0.1
0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4
Slenderness ()
Figures 8.17 and 8.18 present the comparison of FEA elastic lateral distortional
buckling moments of LSBs with web stiffeners (Modw) with the developed design
equations as shown next.
where, = (My/Mo)1/2
The ratios of FEA to predicted buckling moment ratios were obtained and the mean
and COV values were calculated for both equations. Equation 8.2 has a mean FEA to
predicted value of 1.00 and a COV of 0.017 while those for Equation 8.3 are 1.00
and 0.015. This indicates that the developed equations are accurate to predict the
lateral distortional buckling moments of LSBs with the chosen web stiffener
configuration in this research. Although Equation 8.2 is considered to be a simple
equation, it over-estimates the buckling moments at higher slenderness values (i e,
> 2). Therefore, it is recommended to avoid using Equation 8.2 for LSBs with high
slenderness values while Equation 8.3 is reasonable for any slenderness values.
8-30
Effect of Web Stiffeners on the Lateral Buckling of LSBs
8-31
Effect of Web Stiffeners on the Lateral Buckling of LSBs
8-32
Effect of Web Stiffeners on the Lateral Buckling of LSBs
Table 8.8: Comparison of Ultimate Moments with and without Web Stiffeners
Span Mu (kNm)
LSB Sections % increase
(mm) without WS with WS
2000 54.52 65.37 19.91
3000 46.29 54.76 18.30
4000 40.92 46.29 13.12
300x75x3.0 LSB
6000 32.45 34.60 6.62
8000 26.60 27.56 3.59
10000 22.55 23.02 2.10
3000 36.91 46.19 25.16
4000 32.62 39.05 19.71
300x75x2.5 LSB
6000 26.43 29.17 10.36
8000 22.02 23.33 5.95
10000 18.81 19.52 3.80
3000 17.77 22.62 27.32
300x60x2.0 LSB 4000 14.98 17.98 20.00
6000 11.57 12.81 10.71
8000 9.43 10.02 6.24
1250 53.63 60.51 12.82
1500 51.49 59.30 15.16
2000 48.24 56.42 16.96
3000 43.59 49.54 13.65
250x75x3.0 LSB
4000 39.41 43.03 9.20
6000 31.97 33.18 3.78
8000 26.40 27.05 2.46
10000 22.49 22.77 1.24
2000 38.39 47.20 22.95
3000 34.50 41.36 19.89
4000 31.72 36.17 14.04
250x75x2.5 LSB
6000 26.24 28.10 7.07
8000 22.07 22.81 3.36
10000 18.92 19.29 1.96
2000 20.91 26.53 26.92
3000 17.29 21.07 21.86
4000 15.04 17.29 14.97
250x60x2.0 LSB
6000 11.82 12.62 6.80
8000 9.65 10.05 4.17
10000 8.28 8.52 2.91
WS Web Stiffeners
8-33
Effect of Web Stiffeners on the Lateral Buckling of LSBs
Table 8.8 (continued): Comparison of Ultimate Moments with and without Web
Stiffeners (WS)
Mu (kNm)
LSB Sections Span (mm) % increase
without WS with WS
1000 28.45 31.89 12.11
1250 27.08 31.30 15.57
1500 26.25 30.29 15.39
2000 24.41 27.97 14.60
200x60x2.5 LSB 3000 21.68 23.64 9.04
4000 18.95 20.01 5.64
6000 14.73 15.03 2.02
8000 11.94 12.12 1.49
10000 10.16 10.27 1.17
1500 20.08 24.46 21.83
2000 18.54 22.74 22.68
3000 16.64 19.19 15.30
200x60x2.0 LSB
4000 14.75 16.23 10.04
6000 11.79 12.26 4.02
8000 9.71 9.89 1.83
1500 9.85 12.37 25.55
2000 8.43 10.49 24.56
200x45x1.6 LSB 3000 6.85 7.88 15.11
4000 5.76 6.26 8.55
6000 4.33 4.51 4.21
1000 12.13 13.83 14.01
1250 11.63 13.23 13.75
1500 11.16 12.56 12.54
2000 10.23 11.16 9.12
150x45x2.0 LSB 3000 8.53 8.93 4.69
4000 7.13 7.33 2.80
6000 5.36 5.40 0.62
8000 4.37 4.40 0.76
10000 4.00 4.00 0.00
1000 9.44 11.16 18.25
1250 8.87 10.70 20.60
1500 8.47 10.13 19.61
2000 7.87 9.10 15.61
150x45x1.6 LSB 3000 6.71 7.28 8.42
4000 5.71 5.98 4.65
6000 4.35 4.45 2.29
8000 3.55 3.62 1.87
10000 3.20 3.24 1.35
WS Web Stiffeners
8-34
Effect of Web Stiffeners on the Lateral Buckling of LSBs
Table 8.8 (continued): Comparison of Ultimate Moments with and without Web
Stiffeners (WS)
Span Mu (kNm)
LSB Sections % increase
(mm) without WS with WS
750 10.81 11.49 6.27
1000 10.58 11.46 8.37
1250 10.37 11.23 8.29
1500 10.14 10.81 6.68
2000 9.46 9.82 3.86
125x45x2.0 LSB
3000 8.13 8.31 2.24
4000 6.93 7.04 1.50
6000 5.26 5.29 0.50
8000 4.27 4.27 0.00
10000 3.78 3.78 0.00
1000 8.21 9.17 11.71
1250 7.93 9.07 14.43
1500 7.82 8.78 12.29
2000 7.38 8.11 9.86
125x45x1.6 LSB 3000 6.47 6.78 4.82
4000 5.61 5.74 2.31
6000 4.31 4.37 1.20
8000 3.51 3.53 0.74
10000 3.07 3.09 0.85
WS Web Stiffeners
As seen in Table 8.8, the increase in the non-linear lateral buckling moment capacity
of LSBs is high for intermediate spans (up to 27%) while it is small for long spans.
This is as expected since lateral torsional buckling is the dominant buckling mode for
long span members. Thus web distortion is reduced for long spans and hence the
increase in the non-linear moment capacity is also small. Figure 8.20 shows the
variation of ultimate moment capacities of LSBs with web stiffeners while Figure
8.21 shows a comparison of ultimate moment capacities of some LSBs with and
without web stiffeners. The increase in the ultimate moments of LSBs due to the use
of web stiffeners with its span is evident in Figure 8.21. As mentioned above the
moment capacity improvement is high for LSBs with intermediate spans.
8-35
Effect of Web Stiffeners on the Lateral Buckling of LSBs
70
300x75x3.0 LSB
300x75x2.5 LSB
300x60x2.0 LSB
60 250x75x3.0 LSB
Ultimate Lateral Buckling Moments (kNm)
250x75x2.5 LSB
250x60x2.0 LSB
50 200x60x2.5 LSB
200x60x2.0 LSB
200x45x1.6 LSB
40 150x45x2.0 LSB
150x45x1.6 LSB
125x45x2.0 LSB
30 125x45x1.6 LSB
20
10
0
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
Span (mm)
70
300x75x3.0 LSB with WS
300x75x3.0 LSB without WS
60
250x75x2.5 LSB with WS
Ultimate Lateral Buckling Moments (kNm)
30
20
10
0
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
Span (mm)
Figure 8.21: Comparison of Ultimate Moments of LSBs with and without Web
Stiffeners
8-36
Effect of Web Stiffeners on the Lateral Buckling of LSBs
The ultimate moments from FEA were non-dimensionalised and compared with the
design curve for the LSBs without web stiffeners developed in Chapter 6. Figure
8.22 shows the comparison of these FEA results with Equation 8.4 developed in
Chapter 6 as Equation 6.7.
1
For d 1.74: Mc = My 2 (8.4c)
d
My
where, d =
Mod
1.20
Equation 8.4
FEA
1.00
0.80
Mu/My, Mb/My
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Slenderness, d, dw
Figure 8.22 compares the ultimate moments with Equation 8.4 where the slenderness
dw is given by the following equation.
My
dw =
Modw
Here, the elastic lateral distortional buckling moments of LSBs with web stiffeners
Modw were used instead of Mod.
8-37
Effect of Web Stiffeners on the Lateral Buckling of LSBs
Since the FEA data points are well above the design curve based on Equation 8.4 in
the intermediate slenderness region, it was considered that the use of equations
developed for the back to back LSBs may be more suitable. The design equations for
the ultimate moment capacities of back to back LSBs were developed by Jeyaragan
and Mahendran (2009) and are given next.
1
For d 1.80: Mc = My 2 (8.5c)
d
1.20
FEA
Equation 8.5
1.00
0.80
Mu/My, Mb/My
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Slenderness, d, dw
8-38
Effect of Web Stiffeners on the Lateral Buckling of LSBs
Figure 8.23 shows that most of the FEA data points are below the design curve based
on Equation 8.5. The mean of the ratio of ultimate moment capacities from FEA and
this equation was found to be 0.97 and the associated COV was 0.049 with a capacity
reduction factor of 0.88. This indicates that Equation 8.5 is also not suitable to
predict the ultimate moment capacities of stiffened LSBs subject to lateral buckling.
Therefore a new design equation (Eq. 8.6) was developed by solving for minimum
total error for the available FEA ultimate moments (92 results).
1
For dw 1.70: Mc = My 2 (8.6c)
dw
1.20
Equation 8.6
FEA
1.00
0.80
Mu/My, Mb/My
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Slenderness, dw
The mean and COV of the ratio of ultimate moment capacities from FEA and
Equation 8.6 were calculated to be 1.02 and 0.050, respectively for the inelastic
region. The capacity reduction factor in this case was found to be 0.92, which is
slightly higher than the recommended value of 0.90. Since no experiments have been
carried out in this study, it is appropriate to accept a slightly higher capacity
reduction factor of 0.92.
8-39
Effect of Web Stiffeners on the Lateral Buckling of LSBs
A geometrical parameter was found to reduce the scatter of the FEA data points of
the ultimate moment capcacities of LSBs without web stiffeners as described in
Chapter 6. An attempt was also made to determine the applicability of the same
geometrical parameter in the investigation of LSBs with web stiffeners. Figure 8.25
shows the comparison of the FEA ultimate moments of LSBs with web stiffeners and
the new design equation with geometrical parameter as presented in Chapter 6 (Eq.
6.18).
1
K= (8.8)
GJf
0.85 +
EIxweb
where
GJf = torsional rigidity of the flange
EIxweb = major axis flexural rigidity of the web
1.20
FEA
Equation 8.7
1.00
0.80
Mu/My, Mb/My
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Slenderness, Kdw
8-40
Effect of Web Stiffeners on the Lateral Buckling of LSBs
As seen in Figure 8.25, Equation 8.7 is conservative at the beginning of the inelastic
region. The mean and COV of the ultimate moment capacities from FEA and this
equation were found to be 1.03 and 0.038, respectively, with a capacity reduction
factor of 0.93. Therefore, a new design equation was developed by solving for
minimum total error for the available FEA ultimate moments as shown next.
1.20
FEA
Equation 8.8
1.00
0.80
Mu/My, Mb/My
0.60
0.40
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60
Slenderness, Kdw
Equation 8.8 predicted the ultimate moment capacities accurately as reflected by the
mean and COV of the ratio of ultimate moment capacities from FEA and Equation
8.8, which were 1.02 and 0.043, respectively. The capacity reduction factor was
found to be 0.92, which is slightly greater than the recommended value of 0.90.
8-41
Effect of Web Stiffeners on the Lateral Buckling of LSBs
8.6 Conclusions
This chapter has presented the details of an investigation on the effects of web
stiffeners on the lateral distortional buckling moment behaviour and capacity of
LSBs. Various types of web stiffener configurations including their size and spacing
were considered using a series of elastic buckling analyses. It was found that 5 mm
thick steel plates welded to the inner surfaces of the top and bottom flanges at the
beam supports and at third points within the span considerably improved the lateral
distortional buckling moment capacities of LSBs. It was found that this improvement
was achieved when web stiffeners are also used at the supports, thus providing flange
twist restraints to the entire section including its flanges.
The use of web stiffeners reduced the level of web distortion considerably and thus
allowed the LSB members to achieve at least 85% of lateral torsional buckling
capacity for short and intermediate spans, but reached about 97% of lateral torsional
buckling capacity for long spans. The web stiffeners can also be screw-fixed instead
of welding to the inner faces of top and bottom flanges of LSBs. Thinner web
stiffeners (3 mm or 4 mm) can also be considered to be equally effective for thinner
and smaller LSBs. Suitable equations were developed to calculate the elastic lateral
distortional buckling moments of LSBs with the above mentioned web stiffener
configurations.
8-42
Effect of Web Stiffeners on the Lateral Buckling of LSBs
The ultimate moment capacities of LSBs with web stiffeners were compared with the
developed design rules for single and back to back LSBs without web stiffeners.
Since they were not suitable, a new design rule was developed to accurately predict
the ultimate moment capacities of LSBs with web stiffeners subject to lateral
buckling.
It was found that the use of a geometrical parameter K significantly reduced the
scatter in the FEA data points where this parameter K was the same as that used in
the lateral buckling investigation of LSBs without web stiffeners. A new design rule
with the geometrical parameter K was also developed to accurately predict the
ultimate moment capacities of LSBs with web stiffeners.
8-43
Conclusions and Recommendations
CHAPTER 9
This thesis has described a detailed investigation into the flexural behaviour of
LiteSteel Beams (LSBs) based on experimental and finite element analyses. This
investigation included three phases. In the first phase the member moment capacity
of LSBs and other types of hollow flange sections such as Hollow Flange Beams
(HFBs), Monosymmetric Hollow Flange Beams (MHFBs) and Rectangular Hollow
Flange Beams (RHFBs) subject to lateral distortional buckling was investigated
while in the second phase, the section moment capacity of LSBs subject to local
buckling effects including the inelastic reserve moment capacity was investigated. In
the third phase the use of web stiffeners was investigated in order to improve the
lateral distortional buckling moment capacities of LSBs.
The LSB flexural members are subjected to a relatively new lateral distortional
buckling mode, which reduces their member moment capacities. A detailed
investigation into the flexural behaviour of LSBs and their member moment
capacities was undertaken in the first phase of this research using experimental and
finite element analyses. It included 12 lateral buckling tests of LSBs using a quarter
point loading arrangement, finite element modelling of tested LSBs, and a detailed
parametric study to develop suitable design rules. Numerical studies in this phase
entail the development of two finite element models, namely, experimental and ideal
models to simulate the flexural behaviour of LSBs including their lateral buckling
characteristics. A general purpose finite element analysis program ABAQUS Version
6.7 (HKS, 2007) and MSC PATRAN (PATRAN, 2008) were used in this study.
Experimental finite element models were used to simulate the tested LSBs and to
validate the models by a comparison of experimental and finite element analysis
results while ideal finite element models were used to develop member moment
capacity data under uniform moment conditions that were used to propose suitable
design rules. Chapters 4, 5 and 6 presented the details of Phase one of this research
project and the results.
9-1
Conclusions and Recommendations
In Phase two of this research, the section moment capacity of LSBs was investigated
based on experiments and finite element analyses. It included four point bending
tests of seven LSBs, numerical simulations and a detailed parametric study of section
moment capacity of LSBs including inelastic reserve capacity of LSBs. Suitable
experimental finite element models were developed and their results were compared
with experimental results for validation purposes. The validated model was then used
in the parametric study. The results from the parametric study and experiments were
used to review the available design rules for section moment capacity including the
presence of inelastic reserve bending capacity. Chapter 7 presented the details of
Phase 2 of this research project and the results.
Phase three of this research included an investigation on the effects of web stiffeners
on the lateral distortional buckling moment behaviour and capacity of LSBs. This
included finite element analyses of LSBs with different configurations of web
stiffeners in order to develop an optimum web stiffener configuration, and a detailed
parametric study to develop suitable design rules with optimum web stiffener
configuration. Chapter 8 presented the details of Phase 3 of this research project and
the results.
Chapter 3 of this thesis presented the details of tensile coupon tests and residual
stress and geometric imperfection measurements of LSBs used in the experimental
study. Chapter 4 presented the details of lateral buckling tests of LSBs. It includes a
comparison of the experimental results with the member moment capacity
predictions from AS/NZS 4600 (SA, 2005). Chapter 5 presented the details of the
finite element analyses of LSBs subject to lateral buckling including validation of
finite element models. The details of the detailed parametric study of LSBs subject to
lateral distortional buckling and the development of design rules are presented in
Chapter 6. The effect of section geometry and the applicability of developed design
rules for other types of hollow flange sections are also included in this chapter.
Chapter 7 presented the details of section moment capacity of LSBs including the
inelastic reserve moment capacity based on experiments and finite element analyses.
Chapter 8 presented the details of the effect of web stiffeners on the lateral
distortional buckling strength of LSBs. It includes the details of the optimum web
9-2
Conclusions and Recommendations
The most important outcomes obtained from this research are as follows.
Significantly improved understanding and knowledge of the flexural
behaviour of LSBs including their section and member moment capacities.
Development of accurate member capacity design rules for LSBs and other
hollow flange sections based on AS/NZS 4600 (SA, 2005). Some of them
included the effect of section geometry on the lateral distortional buckling
1
moment capacity using a geometrical parameter K ( = ).
GJf
0.85 +
EIxweb
Where, (GJf) is the flange torsional rigidity and (EIxweb) is the major axis
flexural rigidity of web. The new design rules can be used to accurately
predict the member moment capacity of LSBs with a capacity reduction
factor of () 0.90.
Assessment of the current design rules for the section moment capacity of
LSBs.
Confirmation of the presence of inelastic reserve moment capacity for
compact LSBs and suitable design rules to include it in design.
Development of an optimum web stiffener configuration to improve the
lateral distortional buckling moment capacity of LSBs and associated design
rules to predict the improved elastic buckling and ultimate member moment
capacities of LSBs with optimum web stiffeners.
The measured initial geometric imperfections were well below the fabrication
tolerance limits for flexural members.
Residual stresses of LSBs were measured and the current membrane residual
stress distributions of LSBs proposed by Mahaarachchi and Mahendran
(2005e) were improved.
9-3
Conclusions and Recommendations
The approximation of LSBs round corners with right angle corners in the
finite element modelling has negligible effect on the section properties as
well as the elastic lateral buckling moments.
Finite element analyses using the experimental finite element models
developed in this research well predicted the ultimate moments and load-
deflection curves obtained from experiments.
Ideal finite element models developed in this research were able to predict the
elastic lateral buckling moments of LSBs with the moments from a well
established finite strip analysis program Thin-Wall and the elastic lateral
distortional buckling moments from Pi and Trahairs (1997) equation with an
average deviation of 1.5% and 2.9%, respectively.
The developed finite element model was able to capture both the elastic and
non-linear ultimate strength behaviour of LSBs.
9-4
Conclusions and Recommendations
9-5
Conclusions and Recommendations
9-6
Conclusions and Recommendations
The web stiffeners can also be screw-fixed instead of welding to the inner
faces of top and bottom flanges of LSBs.
Suitable design equations were developed to calculate the elastic lateral
distortional buckling moments of LSBs with the recommended optimum web
stiffener configurations.
It was found that the developed design rules for single and back to back LSBs
without web stiffeners were not suitable to predict the lateral buckling
moments of LSBs with web stiffeners.
A new design rule was developed to accurately predict the ultimate moment
capacities of LSBs with web stiffeners subject to lateral buckling. This can be
used with a capacity reduction factor () of 0.90.
The use of a geometrical parameter K significantly reduced the scatter in the
FEA data points where this parameter K was the same as that used in the
lateral buckling investigation of LSBs without web stiffeners.
A new design rule with the geometrical parameter K was also developed to
more accurately predict the ultimate moment capacities of LSBs with web
stiffeners.
The developed design equations for LSBs with 5 mm thick steel plate
stiffeners welded to the inner faces of the flanges of LSBs at third span points
and supports is recommended to be equally applicable when thinner web
stiffeners (3 mm or 4 mm) are used for thinner and smaller LSBs.
It is recommended that the following research projects are undertaken in the future to
advance the knowledge in this field.
9-7
Conclusions and Recommendations
9-8
Appendix A
APPENDIX A
A.1: Stress-Strain Curves for LSB Section Material
700.0
600.0
557.7
500.0
Stress (N/mm2)
200.0
100.0
0.0
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50
% Strain
600.0
500.0
496.3
400.0
Extensometer
Stress (N/mm2)
E=220GPa
0.2% Proof Stress
300.0
200.0
100.0
0.0
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50
% Strain
A-1
Appendix A
600.0
500.0
447.1
400.0
Extensometer
Stress (N/mm2)
E=190GPa
300.0 0.2% Proof Stress
200.0
100.0
0.0
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00
% Strain
600.0
500.0
400.0
Web
Stress (N/mm2)
Outside Flange
E=210GPa
300.0 Inside Flange
200.0
100.0
0.0
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00
% Strain
A-2
Appendix A
600.0
500.0
400.0
Stress (N/mm2)
Web
Outside Flange
300.0
Inside Flange
E=200GPa
200.0
100.0
0.0
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00
% Strain
600.0
500.0
400.0
Inside Flange
Stress (N/mm2)
Outside Flange
Web
300.0 E=215GPa
200.0
100.0
0.0
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00 11.00
% Strain
A-3
Appendix A
700.0
600.0
500.0
Outside flange
Stress (MPa)
200.0
100.0
0.0
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00
Strain (%)
700
600
500
Stress (MPa)
400
Outside flange
Inside flange
300
Web
E=200GPa
200
100
0
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00 11.00
Strain (%)
A-4
Appendix A
1.8
5
1.6
4
Neutral Axis
3
1.4 2
1
1.2
Imperfection (mm)
1.0
0.8
0.6
0.4
Along 1
Along 2
0.2 Along 3
Along 4
Along 5
0.0
0 500 1000 1500 2000 2500 3000 3500
Span Length (mm)
Figure A.9: Global Geometric Imperfection along the Web for 250x75x2.0 LSB
1.8
1 2
1.5
1.2
0.9
0.6
Imperfections (mm)
0.3
0.0
0 500 1000 1500 2000 2500 3000 3500
-0.3
-0.6
Flange - Average
-0.9
Web - Average
Web Middle
-1.5
Web Top
-1.8 Flange Location 1
Span Length (mm)
Flange Location 2
A-5
Appendix B
APPENDIX B:
B.1: Moment vs Deflection Curves for the Tested LSB Specimens
20.0
18.0
16.0
14.0
Moment (kNm)
4 m Span, at MidSpan
8.0
4 m Span, under the Load
6.0
4.0
2.0
0.0
0 10 20 30 40 50 60 70 80
Vertical Deflection (mm)
20.0
18.0
16.0
14.0
3 m Span, at Top Flange (Tension)
Moment (kNm)
12.0
3 m Span, at Bottom Flange (Compression)
10.0
4 m Span, at Top Flange (Tension)
8.0
4 m Span, at Bottom Flange (Compression)
6.0
4.0
2.0
0.0
-10 10 30 50 70 90 110 130 150
Lateral Deflection at Mid Span (mm)
B-1
Appendix B
12.0
10.0
8.0
Moment (kNm)
6.0
2 m Span, MidSpan
4.0 2 m Span, under the Load
3 m Span, at MidSpan
3 m Span, under the Load
2.0 4 m Span, at MidSpan
4 m Span, under the Load
0.0
0 10 20 30 40 50 60 70 80
Vertical Deflection (mm)
12.0
10.0
8.0
Moment (kNm)
6.0
0.0
-10 10 30 50 70 90 110 130 150
Lateral Deflection at Mid Span (mm)
B-2
Appendix B
12.0
10.0
8.0
Moment (kNm)
6.0
2.0 m Span, at MidSpan
2.0 m Span, under the Load
4.0 3.0 m Span, at MidSpan
3.0 m Span, under the Load
2.0
0.0
0 10 20 30 40 50 60 70 80 90 100
Vertical Deflection (mm)
12.0
10.0
8.0
Moment (kNm)
6.0
0.0
-10 10 30 50 70 90 110 130 150
Lateral Deflection at Mid Span(mm)
B-3
Appendix B
10.0
9.0
8.0
7.0
Moment (kNm)
6.0
5.0
0.0
0 10 20 30 40 50 60
Vertical Deflection (mm)
10.0
9.0
8.0
7.0
Moment (kNm)
6.0
5.0
0.0
-10 10 30 50 70 90 110 130
Lateral Deflection at Mid Span (mm)
B-4
Appendix B
12.0
10.0
8.0
Moment (kNm)
6.0
4.0
MidSpan
under the Load
2.0
0.0
0 5 10 15 20 25 30 35
Vertical Deflection (mm)
12
10
8
Moment (kNm)
4
Top Flange (Tension)
0
-2 0 2 4 6 8 10 12 14 16 18
Lateral Deflection at Mid Span (mm)
B-5
Appendix C
APPENDIX C:
C.1: Sample Calculation of Application of End Moment in the Ideal FE Model
of LSB
Nodal loads were applied at each node to create a uniform end moment across the
section of LSB. Sample calculations for an ideal model of 200x45x1.6 LSB are given
next.
Step 1
Find the linear equation of load across the cross section.
-1000 N
198.4 mm
99.2 mm Centreline
Y
0 mm
1000 N
The equation is
Load = -10.08064516x + 1000 (C.1)
This equation was used as a spatial function in MD/PATRAN and applied to all
the nodes at the end support of LSB.
C-1
Appendix C
1500
1000
y = -10.08064516x + 1000
500
Load
0
0 50 100 150 200 250
-500
-1000
-1500
Step 2
Calculate the applied moment.
Find the load information from the input file to the elastic buckling analysis.
-F
Lever
arm
F
Y
Y1
Note: It is important to have the finite element mesh so that there is a node at the
middle of the web. Then the loads will be created symmetrically about the centreline
of the web.
C-2
Appendix C
Table C.2 shows the calculation of applied moment for a 200x45x1.6 LSB ideal FE
model. Here, the node number and the corresponding loads are extracted from the
input file of the elastic buckling analysis. Distance along the Y direction can be
calculated using the equation we obtained before.
For example,
Node 1
Load = 864.92 N
From equation, Load = -10.08064516*Distance along Y direction + 1000
Hence, Distance along Y direction = (1000-864.92)/10.08064516
= 13.40 mm
Lever arm/2 = 99.2-13.4
= 85.80 mm
Applied Moment = 2 * load * (lever arm/2)
= 2 * 864.92 * 85.8 * 10-6 kNm
= 0.1484 kNm
C-3
Appendix C
Step 3
Application of this applied moment.
The critical buckling load factor obtained from the elastic buckling analysis and the
maximum load factor obtained from the non-linear static analysis should be
multiplied by this applied moment to determine the actual values.
C-4
Appendix C
C-5
Appendix C
ENDIF
C
C FLEXURAL RESIDUAL STRESS
IF((NOEL.GE.9151.).AND.(NOEL.LE.10500.)) THEN
sigmaout=(0.24*Fy+0.83*Fy*IMPTUF/BF)*(1.-2.*(nipt-
ipt)/(nipt-1.))
ELSEIF((NOEL.GE.1801.).AND.(NOEL.LE.3150.)) THEN
sigmaout=(0.24*Fy+0.83*Fy*IMPBLF/BF)*(1.-2.*(nipt-
ipt)/(nipt-1.))
ELSEIF((NOEL.GE.10951.).AND.(NOEL.LE.12300.)) THEN
sigmaout=(0.38*Fy+0.42*Fy*IMPTLF/BF)*(1.-2.*(nipt-
ipt)/(nipt-1.))
ELSEIF((NOEL.GE.1.).AND.(NOEL.LE.1350.)) THEN
sigmaout=(0.38*Fy+0.42*Fy*IMPBUF/BF)*(1.-2.*(nipt-
ipt)/(nipt-1.))
ELSEIF(((NOEL.GE.10501.).AND.(NOEL.LE.10950.)).OR.
& ((NOEL.GE.1351.).AND.(NOEL.LE.1800.))) THEN
sigmaout=(0.41*Fy)*(1.-2.*(nipt-ipt)/(nipt-1.))
ELSEIF(((NOEL.GE.8701.).AND.(NOEL.LE.9150.)).OR.
& ((NOEL.GE.3151.).AND.(NOEL.LE.3600.))) THEN
sigmaout=(0.24*Fy)*(1.-2.*(nipt-ipt)/(nipt-1.))
ELSEIF((NOEL.GE.3601.).AND.(NOEL.LE.8700.)) THEN
sigmaout=(0.24*Fy)*(1.-2.*(nipt-ipt)/(nipt-1.))
ENDIF
C
C MEMBRANE RESIDUAL STRESS
IF((NOEL.GE.3601.).AND.(NOEL.LE.8700.)) THEN
IF((Y.GE.13.4).AND.(Y.LE.99.2)) THEN
MEMB=(0.01166*Y-0.65618)*Fy
ELSEIF((Y.GE.99.2).AND.(Y.LE.185)) THEN
MEMB=(-0.01166*Y+1.65618)*Fy
ENDIF
ELSEIF((NOEL.GE.10951.).AND.(NOEL.LE.12300.)) THEN
MEMB=0.11*Fy-0.08*Fy*IMPTLF/BF
ELSEIF((NOEL.GE.1.).AND.(NOEL.LE.1350.)) THEN
MEMB=0.11*Fy-0.08*Fy*IMPBUF/BF
ELSEIF(((NOEL.GE.10501.).AND.(NOEL.LE.10950.)).OR.
& ((NOEL.GE.1351.).AND.(NOEL.LE.1800.))) THEN
MEMB=0.03*Fy
ELSEIF(((NOEL.GE.8701.).AND.(NOEL.LE.9150.)).OR.
& ((NOEL.GE.3151.).AND.(NOEL.LE.3600.))) THEN
MEMB=-0.2567*Fy
ENDIF
SIGMA(1)=sigmaout+MEMB
c
SIGMA(2)=0
SIGMA(3)=0
C
RETURN
END
C-6
Appendix C
40.0
35.0
30.0
25.0
Moment (kNm)
EXP
20.0 FEA
15.0
10.0
5.0
0.0
-10 0 10 20 30 40 50 60 70
Vertical Deflection (mm)
45.0
40.0
35.0
30.0
Moment (kNm)
25.0
EXP - Top Flange (Tension)
20.0 EXP - Bottom Flange(Compression)
FEA - Top Flange
15.0 FEA - Bottom Flange
10.0
5.0
0.0
-10 10 30 50 70 90
Lateral Deflection (mm)
C-7
Appendix C
20.0
18.0
16.0
14.0
12.0
EXP - Bottom Flange (Compression)
10.0
FEA - Top Flange
6.0
4.0
2.0
0.0
-10 10 30 50 70 90 110 130 150
Lateral Deflection (mm)
20.0
18.0
16.0
14.0
Moment (kNm)
12.0
EXP
10.0
FEA
8.0
6.0
4.0
2.0
0.0
0 5 10 15 20 25 30
Vertical Deflection (mm)
C-8
Appendix C
20.0
18.0
16.0
14.0
EXP - Top Flange (Tension)
Moment (kNm)
6.0
4.0
2.0
0.0
-10 10 30 50 70 90
Lateral Deflection (mm)
10.0
9.0
8.0
7.0
Moment (kNm)
6.0
EXP
5.0 FEA
4.0
3.0
2.0
1.0
0.0
0 5 10 15 20 25 30 35 40 45 50
Vertical Deflection (mm)
C-9
Appendix C
10.0
9.0
8.0
7.0
Moment (kNm)
3.0
2.0
1.0
0.0
-10 10 30 50 70 90 110 130
Lateral Deflection (mm)
12.0
10.0
8.0
EXP
Moment (kNm)
FEA
6.0
4.0
2.0
0.0
0 5 10 15 20 25
Vertical Deflection (mm)
C-10
Appendix C
12.0
10.0
8.0
Moment (kNm)
6.0
EXP - Top Flange (Tension)
EXP - Bottom Flange (Compression)
4.0 FEA - Top Flange
FEA - Bottom Flange
2.0
0.0
-10 0 10 20 30 40 50
Lateral Deflection (mm)
12.0
10.0
8.0
Moment (kNm)
6.0
EXP
FEA
4.0
2.0
0.0
0 10 20 30 40 50 60 70 80 90
Vertical Deflection (mm)
C-11
Appendix C
12.0
10.0
8.0
Moment (kNm)
6.0
EXP
FEA
4.0
2.0
0.0
0 5 10 15 20 25 30 35 40 45 50
Vertical Deflection (mm)
14
12
10
Moment (kNm)
0
-10 0 10 20 30 40 50 60 70 80 90
Lateral Deflection (mm)
C-12
Appendix C
9.0
8.0
7.0
6.0
Moment (kNm)
2.0
1.0
0.0
-10 0 10 20 30 40 50 60
Lateral Deflection (mm)
10.0
9.0
8.0
7.0
Moment (kNm)
6.0
5.0
4.0 EXP
FEA
3.0
2.0
1.0
0.0
0 5 10 15 20 25 30 35 40 45 50
Vertical Deflection (mm)
C-13
Appendix C
12.0
10.0
8.0
Moment (kNm)
6.0
EXP
4.0 FEA
2.0
0.0
0 5 10 15 20 25 30 35
Vertical Deflection (mm)
12.0
10.0
8.0
Moment (kNm)
6.0
EXP - Top Flange (Tension)
EXP - Bottom Flange (Compression)
4.0 FEA - Top Flange
FEA - Bottom Flange
2.0
0.0
-2 0 2 4 6 8 10 12 14 16 18
Lateral Deflection (mm)
C-14
Appendix D
APPENDIX D:
D.1: Sample Calculations of Moment Capacities
Sample Calculations Based on AS/NZS 4600 (SA, 2005) Design Rules
300x75x3.0 LSB section based on nominal dimensions and yield stresses without
corners. Centreline dimensions are used.
t = 3 mm
d = 300 mm
d1 = 250 mm
d1 df = 25 mm
d
bf = 75 mm
fyf = 450 MPa
df fyw = 380 MPa
bf
Ms = Ze fy
tw = 3 mm
d1 = 250 + 3 = 253 mm
d1 / tw = 253/3 = 84.33 < 200 Clause 2.1.3.4a
D-1
Appendix D
1.052 b f*
= Clause 2.2.1.2 (4)
k t E
1.052 72 450
=
4 3 200000
22 f2 = 383.33 MPa
f2
f1 = 450 MPa
*
148.5 f2
= *
Clause 2.2.3.2 (5)
f1
383.33
= = 0.852
450
k = 4 + 2(1-)3 + 2(1 ) Clause 2.2.3.2 (4)
k = 4.30
D-2
Appendix D
1.052 b f*
=
k t E
1.052 22 450
=
4.30 3 200000
1.052 b f*
=
k t E
Since 300x75x3.0 LSB is fully effective, its effective section modulus Ze is equal to
its full section modulus Z. Therefore the section moment capacity Ms is equal to the
first yield moment My of this LSB. However, some other LSBs are not fully effective
as their flanges are not fully effective when the corners are not considered in the
calculations. Tables D.1 and D.2 give the details of calculations leading to their
effective widths.
D-4
Appendix D
As seen in Table D.2, the flange slenderness values () of eight LSB sections are
greater than 0.673 and hence these sections are considered to be slender according to
AS/NZS 4600 (SA, 2005). However, the flange slenderness values of three LSBs
(200x45x1.6 LSB, 150x45x1.6 LSB and 125x45x1.6 LSB) are closer to the limiting
value of 0.673. Hence only five LSBs marked by bold italic letters in the table
should be considered as slender. The effective width be was calculated using the
procedure mentioned above (be = b).
D-5
Appendix D
As seen in Table D.3, be1+be2 = be for all 13 LSBs. Therefore, the vertical flange
elements are all fully effective.
As seen in Table D.4, be1+be2 > be/2 (compression portion of web) for all 13 LSBs.
Therefore, the web elements are all fully effective.
In summary, when the LSBs are considered without corners (as used in finite
element modelling), their horizontal flange elements are not fully effective for five
LSBs as shown in Table D.2. Therefore the effective section moduli of these five
LSBs are less than their full section moduli. However, it must be noted that the
flange elements of LSBs are fully effective if their corners are included as shown in
the LSB manufacturers manuals (OATM, 2008). The procedure to calculate the
effective section moduli of these LSBs is given next.
D-6
Appendix D
The elastic section moduli of all the available 13 LSBs are calculated by using the
basic principles. The effective widths of the horizontal compression flange elements
calculated in Table D.2 are used here. Tables D.5 and D.6 provide the details of the
effective second moment of area calculations for 300x75x3.0 LSB and 300x75x2.5
LSB.
3 5
1
d d1
2 df 4 y=0
bf
7
D-7
Appendix D
Effective second moment of area about the major axis, Ie = 21.43 * 106 + 4.060 * 106
= 25.49 * 106 mm4
Effective elastic section modulus, Ze = Ie / y = 25.49 * 106 / (297-148.50)
= 171.65 *103 mm3
This is very close to the value obtained from Thin-wall (171.7*103 mm3). This
confirms the accuracy of calculations used here.
Here, y is the distance from the neutral axis to the extreme fibre of compression
flange (top flange).
For 300x75x3.0 LSB, My = Ms = Zfy = Zefy = 171.65 * 103 * 450 * 10-6 = 77.24 kNm
Centroid, y1 = 147.54 mm
Effective second moment of area about major axis, Ie = 21.151 * 106 mm4
Effective elastic section modulus, Ze = 21.151 * 106 / (297.50-147.54)
= 141.04 *103 mm3
D-8
Appendix D
Four non-standard LSBs were considered in the parametric study of this research.
They are
1. 300x45x3.6 LSB Ze = Z = 145.79 x103 mm3
2. 135x50x1.6 LSB Ze = 22.92 x103 mm3 and Z = 23.78 x103 mm3
3. 125x45x1.8 LSB Ze = Z = 22.07 x103 mm3
4. 127x45x2.4 LSB Ze = Z = 29.10 x103 mm3
The effective widths of these LSBs were calculated by using the same AS/NZS 4600
(SA, 2005) procedure used with the 13 standard LSBs and it was found that the
effective widths are equal to the actual widths except in the case of 135x50x1.6
LSBs. The flange flat element of this LSB has a slenderness value of 0.755 (>0.673)
and hence its effective section modulus will be less than the full section modulus.
D-9
Appendix D
32.0
300x60x2.0 LSB - without RS
30.0
300x60x2.0 LSB - with Flexural RS
28.0
300x60x2.0 LSB - with Membrane RS
26.0
Ultimate Moment Capacities (kNm)
22.0
20.0
18.0
16.0
14.0
12.0
10.0
8.0
6.0
1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000
Span (mm)
30.0
200x60x2.5 LSB - without RS
28.0 200x60x2.5 LSB - with Flexural RS
200x60x2.5 LSB - with Membrane RS
26.0
200x60x2.5 LSB - with Flexural + Membrane RS
Ultimate Moment Capacities (kNm)
24.0
22.0
20.0
18.0
16.0
14.0
12.0
10.0
8.0
1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000
Span (mm)
D-10
Appendix D
12.0
200x45x1.6 LSB - without RS
9.0
8.0
7.0
6.0
5.0
4.0
3.0
1000 2000 3000 4000 5000 6000 7000 8000 9000
Span (mm)
11.0
150x45x1.6 LSB - without RS
8.0
7.0
6.0
5.0
4.0
3.0
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
Span (mm)
D-11
Appendix D
Elastic lateral torsional buckling moments of LSBs were calculated using Eq.6.12.
D-12
Appendix D
750 44.75
1000 31.48
1250 24.37
1500 19.93
125x45x1.6 2000 14.67
LSB 3000 9.64
4000 7.19
6000 4.78
8000 3.58
10000 2.86
D-13
Appendix D
1
K=
GJf
0.85 +
EIxweb
where,
GJf = torsional rigidity of the flange
EIxweb = major axis flexural rigidity of the web
b = 75 mm
d = 25 mm
d t = 3.0 mm
2
4 Am
From the basic principles, Torsional Constant, Jf =
S
t
where,
Am = area enclosed by the median line of the cross section.
S = perimeter along the median line of the section.
4 * {(b t ) * (d t )}2
Jf =
2{(b t ) + (d t )} / t
D-14
Appendix D
d1 = 250 mm
t
3
td
From basic principles, Ixweb = 1
12
3 * 2503
=
12
= 3906.25 * 103 mm4
Elastic modulus, E = 200 000 MPa
Flexural rigidity of web, EIxweb = 200 000 * 3906.25 * 103
= 781 250 * 106 Nmm2
GJ f 12812 *106
Hence =
EI xweb 781250 * 106
= 0.0164
1
Therefore, K=
GJf
0.85 +
EIxweb
1
K=
0.85 + 0.0164
K = 1.0224
D-15
Appendix D
b = 85 mm
a d = 25 mm
t = 3.6 mm
a = 88.60 mm
d
t
2
4 Am
From the basic principles, Torsional Constant, Jf =
S
t
where,
Am = area enclosed by the median line of the cross section.
S = perimeter along the median line of the section.
2
1
4 * (b t ) * (d t )
Jf = 2
[(b t ) + (d t ) + (a t )]/ t
2
1
4 * (85 3.6) * (25 3.6)
Jf = 2
[(85 3.6) + (25 3.6) + (88.60 3.6)]/ 3.6
D-16
Appendix D
The calculation method for the major axis flexural rigidity is the same for any hollow
flange sections as they all have a rectangular web element.
Therefore, depth of the web, d1 = 215 mm, t = 3.6 mm
3
td1
Ixweb =
12
3.6 * 2153
=
12
= 2981.5125 * 103 mm4
Elastic modulus, E = 200 000 MPa
Flexural rigidity of web, EIxweb = 200 000 * 2981.5125* 103
= 596 302.5 * 106 Nmm2
GJ f 4653.4 *106
Hence, =
EI xweb 596302.5 *106
= 0.0078
1
Therefore, K=
GJf
0.85 +
EIxweb
1
K=
0.85 + 0.0078
K = 1.0657
Note: Calculation of GJf and EIxweb for RHFB is similar to that of LSBs as the flange
and web dimensions are the same.
D-17
Appendix E
APPENDIX E:
E.1: Section Compactness of LSBs with Corners Based on AS 4100 (SA, 1998)
Top Flange Element Web Element Overall
d d1 bf df t ro riw Compact- Compact- Compact-
LSB Sections fy (MPa) b ef b ew -ness
(mm) (mm) (mm) (mm) (mm) (mm) (mm) Flange Web -ness -ness
300x75x3.0 LSB 300 244 75 25 3.0 6.0 3.0 450 380 63.0 28.17 C 244 100.3 NC NC
300x75x2.5 LSB 300 244 75 25 2.5 5.0 3.0 450 380 67.0 35.96 NC 244 120.3 S S
300x60x2.0 LSB 300 254 60 20 2.0 4.0 3.0 450 380 53.0 35.55 NC 254 156.6 S S
250x75x3.0 LSB 250 194 75 25 3.0 6.0 3.0 450 380 66.0 29.52 C 194 79.73 C C
250x75x2.5 LSB 250 194 75 25 2.5 5.0 3.0 450 380 67.0 35.96 NC 194 95.67 NC NC
250x60x2.0 LSB 250 204 60 20 2.0 4.0 3.0 450 380 53.0 35.55 NC 204 125.8 S S
200x60x2.5 LSB 200 154 60 20 2.5 5.0 3.0 450 380 52.0 27.91 C 154 75.95 C C
200x60x2.0 LSB 200 154 60 20 2.0 4.0 3.0 450 380 53.0 35.55 NC 154 94.93 NC NC
200x45x1.6 LSB 200 164 45 15 1.6 3.2 3.0 450 380 38.8 32.53 NC 164 126.4 S S
150x45x2.0 LSB 150 114 45 15 2.0 4.0 3.0 450 380 38.0 25.49 C 114 70.27 C C
150x45x1.6 LSB 150 114 45 15 1.6 3.2 3.0 450 380 38.8 32.53 NC 114 87.84 NC NC
125x45x2.0 LSB 125 89 45 15 2.0 4.0 3.0 450 380 38.0 25.49 C 89 54.86 C C
125x45x1.6 LSB 125 89 45 15 1.6 3.2 3.0 450 380 38.8 32.53 NC 89 68.58 C NC
Note:
E-1
Appendix E
55
Load-point
50 Mid-span
45
40
35
Moment (kNm)
30
25
20
15
10
0
0 5 10 15 20 25 30 35 40
Deflection (mm)
75
Load-point
70
Mid-span
65
60
55
50
Moment (kNm)
45
40
35
30
25
20
15
10
5
0
0 5 10 15 20 25 30 35 40
Deflection (mm)
E-2
Appendix E
45
Mid-span
40
35
30
Moment (kNm)
25
20
15
10
0
0 5 10 15 20 25 30 35
Deflection (mm)
18
Load-point
16 Mid-span
14
12
Moment (kNm)
10
0
0 5 10 15 20 25 30
Deflection (mm)
E-3
Appendix E
100
90
80
70
Moment (kNm)
60
Load-point:Exp
50
Mid-span:Exp
Load-point:FEA
40
Mid-span:FEA
30
20
10
0
0 5 10 15 20 25 30 35 40 45
Deflection (mm)
55
Load-point:Exp
50 Mid-span:Exp
Load-point:FEA
45
Mid-span:FEA
40
35
Moment (kNm)
30
25
20
15
10
0
0 5 10 15 20 25 30 35 40
Deflection (mm)
E-4
Appendix E
75
Load-point:Exp
70
Mid-span:Exp
65 Load-point:FEA
60 Mid-span:FEA
55
50
Moment (kNm)
45
40
35
30
25
20
15
10
5
0
0 5 10 15 20 25 30 35 40
Deflection (mm)
45
40
35
30
Mid-span:Exp
Moment (kNm)
25 Load-point:FEA
Mid-span:FEA
20
15
10
0
0 5 10 15 20 25 30 35
Deflection (mm)
E-5
Appendix E
22
20
18
16
14
Moment (kNm)
12
10
Load-point:Exp
8
Mid-span:Exp
6 Load-point:FEA
Mid-span:FEA
4
0
0 5 10 15 20 25 30 35 40
Deflection (mm)
E-6
Appendix E
Figures E.10 (a) to (c) show the measured stress-strain curves for 150x45x2.0 LSBs
from tensile coupon tests.
600.0
500.0
400.0
Strain Gauge
Stress (MPa)
Extenso Meter
Tangent
300.0
0.2% proof stress
200.0
100.0
0.0
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50
% Strain
600.0
500.0
Strain Gauge
Extenso Meter
400.0 Tangent
0.2% Proof Stress
Stress (MPa)
300.0
200.0
100.0
0.0
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00
% Strain
E-7
Appendix E
600.0
500.0
Strain Gauge
Extenso Meter
400.0 Tangent
0.2% Proof Stress
Stress (MPa)
300.0
200.0
100.0
0.0
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00
% Strain
(c) Web
Based on the above figures, some important stress and strain values (engineering) are
considered in plotting the true stress-strain curve. Table E.1 shows these values
Table E.1: Measured (Engineering) Stress and Strain Values of 150x45x2.0 LSB
These values are converted to true stress, strain values by using following equations.
E-8
Appendix E
Table E.2 shows the true stress and strain values which were used in the finite
element analyses.
The following yield stresses are calculated based on the measured stress-strain
curves.
Elastic section modulus of 150x45x2.0 LSB was found to be 31.89 x 103 mm3. This
was calculated with nominal dimensions without corners.
Therefore, the first yield moment, My = fy x Z = 539.3 x 31.89 x 103 x 10-6
= 17.20 kNm
Since this LSB is compact, section moment capacity Ms = My = 17.20 kNm
E-9
Appendix E
Appendix E.1 provided the plate slenderness and the section compactness of LSBs
with corners. This section provides the section compactness and the section moment
capacities of LSBs without corners. Centreline dimensions are used in all the
calculations.
t = 3 mm
d = 300 mm
d1 = 250 mm
d d1 df = 25 mm
bf = 75 mm
fyf = 450 MPa
df
fyw = 380 MPa
bf
b fy
Plate Slenderness, e =
t 250
For Flange
b fy
ef =
t 250
72 450
ef =
3 250
= 32.20
Assuming lightly welded cold-formed steel
ep = 30, ey = 40
ep < ef < ey
flange is non-compact
E-10
Appendix E
for Web
b fy
ew =
t 250
253 380
ew =
3 250
= 103.97
Assuming lightly welded cold-formed steel
ep = 82, ey = 115
ep < ef < ey
web is non-compact
Overall section is non-compact.
Full elastic section modulus Z = 171.65 * 103 mm3 (Thin-wall or Appendix D.1)
Plastic section modulus (S)
S = ((d1 + t)2/4) * t + 4 * t * (df - t)(d-df)/2 + t (bf t)(d - t) + t (bf t) (d1 + t)
= (2532/4) * 3 + 4 * 3 * (25-3)(300-25)/2 + 3(75-3)(300-3) + 3(75-3)(250 + 3)
= 203.11 * 103 mm3
Zc = min (1.5Z, S)
Zc = min (1.5 * 171.65 * 103, 203.11 * 103)
Zc = 203.11 * 103 mm3
Since the LSB is non-compact,
ey s
Ze = Z + (Zc Z )
ey ep
115 103.97
Ze = 171.65 *103 + (203.11 171.65) *10
3
115 82
Similarly, the effective section modulus and the section moment capacities of other
LSBs were calculated based on the above procedure and the results are presented in
Table E.3.
E-11
Appendix E
When calculating the effective section modulus of LSBs with slender web and non-
compact flange, the section modulus of web element is factored by the ratio of (ey/
ew)2.
When calculating the effective section modulus of LSBs with non-compact web and
flanges, the effective section modulus values were calculated by using both ew and
ef values separately and the lower section modulus was selected. Table E.5 presents
E-12
Appendix E
the details of full section modulus Z, plastic section modulus S, critical section
modulus Zc and the effective section modulus Ze. The section moment capacity Ms is
also given in this table.
Overall
Z S S/Z Zc Ze Ms
LSB Sections Compact-
(103 mm3) (103 mm3) (103 mm3) (103 mm3) (kNm)
-ness
300x75x3.0 LSB NC 171.65 203.11 1.18 203.11 182.16 81.97
300x75x2.5 LSB S 143.98 170.47 1.18 170.47 140.67 63.30
300x60x2.0 LSB S 100.38 119.44 1.19 119.44 90.46 40.71
250x75x3.0 LSB NC 133.47 157.81 1.18 157.81 152.45 68.60
250x75x2.5 LSB NC 111.96 132.50 1.18 132.50 114.20 51.39
250x60x2.0 LSB S 78.00 92.39 1.18 92.39 75.10 33.80
200x60x2.5 LSB NC 71.07 84.00 1.18 84.00 82.89 37.30
200x60x2.0 LSB NC 57.32 67.84 1.18 67.84 58.47 26.31
200x45x1.6 LSB S 38.29 45.40 1.19 45.40 36.63 16.48
150x45x2.0 LSB C 31.89 37.68 1.18 37.68 37.68 16.96
150x45x1.6 LSB NC 25.74 30.45 1.18 30.45 27.44 12.35
125x45x2.0 LSB C 24.77 29.34 1.18 29.34 29.34 13.21
125x45x1.6 LSB NC 19.99 23.73 1.19 23.73 21.34 9.60
It can be noted that the shape factor (S/Z) for all the 13 LSBs and non-standard LSBs
are about 1.18.
E-13
Appendix E
The procedure to calculate the slenderness of each plate element of 300x75x3.0 LSB
is presented here. It should be noted that centreline dimensions are used.
t = 3 mm
d = 300 mm
d1 = 250 mm
d d1 df = 25 mm
bf = 75 mm
fyf = 450 MPa
df
fyw = 380 MPa
bf
_
p
bp
=
( )
12 1 f yb
2 0.5
( )
72 12 1 0.3 450
=
2 0.5
22 f2 = 383.33 MPa
f2
f1 = 450 MPa
*
148.5 f2
= *
Table 4.1
f1
383.33
= = 0.852
450
E-14
Appendix E
8.2
K= Table 4.1
1.05 +
K = 4.31
_
p
bp
=
(
12 1 f yb
2
) 0.5
( )
22 12 1 0.3 450
=
2 0.5
2
= = 1 k = 23.9 (Table 4.1)
1
_
p
bp
=
( )
12 1 f yb
2 0.5
( )
253 12 1 0.3 383.33
=
2 0.5
It should be noted that these slenderness values are the same as obtained based on
AS/NZS 4600 (SA, 2005) procedures as shown in Appendix D.
Now the section moment capacity of LSBs can be calculated based on Cl 6.1.4.1 of
Eurocode 3 Part 1.3 (ECS, 2006).
If the effective section modulus Weff is less than the gross elastic section modulus
Wel
M c.Rd = Weff fyb / MO (E.1)
If the effective section modulus Weff is equal to the gross elastic section modulus Wel
M c.Rd = f yb Wel + (W pl Wel )* 4 * 1 e max / MO
eo (E.2)
E-15
Appendix E
Table E.7 presents the slenderness values of compression horizontal flange element
of all the available 13 LSBs and 4 non-standard LSBs and their section moment
capacities. It should be noted that the plate elements are considered to be slender if p
> 0.673.
The section moment capacities of these LSBs were calculated based on these
slenderness values and Eq.E.1 (slender sections) and E.2 (compact sections).
It should be noted that for 300x75x3.3 LSB, 200x45x3.0 LSB and 150x45x3.0 LSB
the section moment capacity Ms is equal to their plastic moments.
For example,
300x75x3.3 LSB, S = 117.28 * 1000 / 450 = 260.62 * 103 mm3 (see Table E.6).
E-16
Appendix E
E.7: Predicting the Plastic Moment Capacity of Compact Steel Beams Using
Finite Element Analyses
It was found that finite element analyses did not predict the full plastic moment
capacity of compact LSBs although experimental results showed that some compact
LSBs reached their full plastic moment capacities. Therefore an attempt was made to
investigate this by using finite element analyses (FEA). Table E.8 shows the details
of this investigation and the results.
Table E.8 shows the first yield moment (My) and the plastic moment (Mp) of four
compact LSB sections as well as their ultimate moments (Mu) from FEA. It was
found that using the same yield stress for web and flange elements of LSBs in FEA
gave improved ultimate moment capacities. It can be seen that Mu/My ratio increased
from 1.10 to 1.13 when the thickness was increased from 2 mm to 3 mm while the
ratio was similar (1.12) for the LSBs with 3 to 5 mm thicknesses when the nominal
flange (450 MPa) and web (380 MPa) yield stresses were used. However, when the
flange and web yield stresses were taken to be the same at 450 MPa, higher Mu/My
ratios of 1.16 were obtained, i e. very close to their Mp/My ratios of 1.17. However,
the Mu/My ratio was 1.13 for 150x45x2.0 LSB section in comparison to its Mp/My
ratio of 1.18. These results in Table E.7 indicate that FEA are not able to predict the
full plastic moment capacities when a smaller yield stress was used for the web
element. They also show that FEA are only able to predict the full plastic moment
capacities of thicker and very compact sections.
E-17
Appendix E
An attempt was then made to obtain the section moment capacities of conventional
hot-rolled I- and C-sections. For this purpose, 150UB14.0 and 150PFC17.7 sections
were modelled without corners using the same flange and web yield stress of 320
MPa as shown in their design capacity tables (AISC, 1994). Table E.9 shows the
details of these sections and the results.
As seen in Table E.9, FEA was able to predict the full plastic moment capacity of
compact hot-rolled I-sections (150UB14.0). However, the ratio of Mu/My was only
1.14 for the mono-symmetric hot- rolled PFC section in comparison to its Mp/My
ratio of 1.19. This contradicts the moment capacities given in AISC (1994), which
recommends the full plastic moment capacity for 150PFC17.7.
Based on the results in Tables E.8 and E.9, it can be concluded that conventional
finite element analyses may not able to predict the full plastic moment capacities of
compact mono-symmetric steel sections unless they are made of very thick plate
elements (with small b/t ratios). Further experiments are needed to confirm these
observations.
E-18
References
REFERENCES
AISC (1994), Design Capacity Tables for Structural Steel, Australian Institute of
Steel Construction, Second Edition, Vol. 1, Sydney, Australia.
Akay, H. V., Johnson, C. P. and Will, K. M. (1977), Lateral and Local Buckling of
Beams and Frames, Journal of Structural Engineering, ASCE, Vol. 103, No. 9,
pp.1821-1832.
American Iron and Steel Institute (AISI) (2004), North American Specification for
the Design of Cold-Formed Steel Structural Members, Washington, D.C., USA.
American Iron and Steel Institute (AISI) (2006), Direct Strength Method Design
Guide CF06-1, North American Specification for the Design of Cold-Formed Steel
Structural Members, Washington DC, USA.
American Iron and Steel Institute (AISI) (2007), North American Specification for
the Design of Cold-Formed Steel Structural Members, Washington, D.C., USA.
R-1
References
Doan, V. and Mahendran, M. (1996), A Residual Stress Model for Hollow Flange
Beams, Research Report 96-39, Physical Infrastructure Centre, Queensland
University of Technology, Brisbane, Australia.
Eurocode 3 Part 1.3 (ECS, 1996), Design of Steel Structures, Part 1.3: General Rules
Supplementary Rules for Cold-Formed Members and Sheeting, British Standard
Institution, London, UK.
R-2
References
Eurocode 3 Part 1.3 (ECS, 2006), Design of Steel Structures Part 1.3: General
Rules Supplementary Rules for Cold-Formed Thin Gauge Members and Sheeting,
British Standard Institution, London, UK.
Hancock, G. J., Kwon, Y. B. and Bernard, E. S. (1994), Strength Design Curves for
Thin-Walled Sections Undergoing Distortional Buckling, Journal of Construction
Steel Research, Vol. 3, pp.169-186.
Hibbit Karisson and Sorensen (HKS), (2007), ABAQUS Users Manual, Pawtucket,
RI., USA.
Kurniawan, C.W. (2007), Lateral Buckling Behaviour and Design of the new
LiteSteel Beam, ME Thesis, QUT, Brisbane, Australia.
OATM (2008), Design Capacity Tables for LiteSteel Beams, OneSteel Australian
Tube Mills, Brisbane, Australia.
R-5
References
Pi, Y.-L. and Trahair, N. S. (1992a), Prebuckling Deflections and Lateral Buckling
Theory, Journal of Structural Engineering, ASCE, Vol. 118, No. 11, pp. 2949-
2966.
Pi, Y.-L. and Trahair, N. S. (1992b), Prebuckling Deflections and Lateral Buckling
Applications, Journal of Structural Engineering, ASCE, Vol. 118, No. 11, pp.
2967-2985.
Pi, Y.-L. and Trahair, N. S. (1997), Lateral Distortional Buckling of Hollow Flange
Beams, Journal of Structural Engineering, ASCE, Vol. 123, No. 6, pp.695-702.
Pi, Y. L., Put, B. M. and Trahair, N. S. (1998), Lateral Buckling Strengths of Cold-
Formed Channel Section Beams, Journal of Structural Engineering, ASCE,
Vol.124, No.10, pp.1182-1191.
Pi, Y.-L., Put, B. M. and Trahair, N. S. (1999), Lateral Buckling Strengths of Cold-
Formed Z-Section Beams, Thin-Walled Structures, Vol. 34, pp.65-93.
R-6
References
Pi, Y.-L., and Trahair, N. S. (2000), Distortion and Warping at Beam Supports,
Journal of Structural Engineering, ASCE, Vol.126, No.11, pp.1279-1287.
Put, M. B., Pi, Y. L. and Trahair, N. S. (1999), Lateral Buckling Tests on Cold-
Formed Channel Beams, Journal of Structural Engineering, ASCE, Vol.125, No.5,
pp. 532-539.
Put, M. B., Pi, Y. L. and Trahair, N. S. (1998), Lateral Buckling Tests on Cold-
Formed Z-Beams, Research Report No. R775, Department of Civil Engineering,
University of Sydney, Sydney, Australia.
Roger, C.A. and Hancock, G.J. (1996), Ductility of G550 Sheet Steels in Tension
Elongation Measurements and Perforated Tests, Research Report R735, Dept. of
Civil Engineering, University of Sydney, Sydney, Australia.
Schafer, B.W. (2006), Direct Strength Method Design Guide, American Iron and
Steel Institute, Washington, D.C., USA.
R-7
References
Standards Australia (SA, 2007), AS 1391, Methods for Tensile Testing of Metals,
Sydney, Australia.
Takabatake, H. (1988), Lateral Buckling of I Beams with Web Stiffeners and Batten
Plates, International Journal of Solid Structures, Vol. 24, No. 10, pp.1003-1019.
R-8
References
Xiao-ting, C., Long-yuan Li, and Roger, K. (2004), The Effects of Warping Stress
on the Lateral Torsional Buckling of Cold-Formed Zed-Purlins, Journal of Applied
Mechanics, ASME, Vol. 71, pp.742-744.
Yang, C. H., Beedle, L. S. and Johnston, B. G. (1952), Residual Stress and the
Yield Strength of Steel Beams, Welding Journal, Vol. 31, No. 4, pp. 205-229.
Yu, W. W. (2000), Cold-Formed Steel Design, 3rd edition, John Wiley & Sons
Inc., New York, USA.
R-10