Weld Effective Lengths For Hollow Structural Section Connections PDF

WELD EFFECTIVE LENGTHS FOR
HOLLOW STRUCTURAL SECTION CONNECTIONS
by
KYLE J. TOUSIGNANT
A thesis submitted in conformity with the

requirements for the degree of Doctor of Philosophy
Department of Civil Engineering
University of Toronto
© Copyright by Kyle J. Tousignant 2017
WELD EFFECTIVE LENGTHS FOR
HOLLOW STRUCTURAL SECTION CONNECTIONS
Kyle J. Tousignant
Doctor of Philosophy
Department of Civil Engineering
University of Toronto
2017
ABSTRACT
An experimental and finite element (FE) research program was conducted to investigate the “fit-for-
purpose” design of welds in hollow structural section (HSS) connections using weld effective lengths, and the
applicability of the North American directional strength-increase factor for fillet welds to HSS. Weld
effective lengths for circular hollow section (CHS) connections are studied for the first time.
Experimental testing of a 10-m span, simply-supported, rectangular hollow section (RHS) Warren truss,
with fillet- and groove-welded overlapped K-connections, was performed by applying a single quasi-static
point load to truss panel points to cause sequential rupture of nine test welds to overlapping branches. The
weld in-situ strength, effect of key connection parameters, and safety (reliability) index of AISC formulae for
weld design in RHS overlapped K-connections were determined. Modifications to the AISC formulae and a
simplified method for modelling overlap-jointed RHS trusses under service load conditions are proposed.
The North American directional strength-increase factor for fillet welds to HSS (covering both RHS and
CHS) was then studied through 33 experimental tests and 73 non-linear FE analyses of weld-critical HSS-to-
rigid plate connections. The safety index was determined for prominent North American design codes, and a
recommendation to prohibit the use of the directional strength-increase factor for the design of fillet welds to
all HSS was made. More accurate design formulae for fillet welds to HSS are proposed.
Twelve large-scale experiments and 256 non-linear FE tests were then conducted to determine weld
effective lengths for CHS-to-CHS X- (or Cross-) connections, which are shown to decrease as branch-to-
chord width ratio, chord wall slenderness, and branch-to-chord thickness ratio increase. North American fillet
weld design formulae were found to already provide adequate reliability provided that the directional
strength-increase factor is not used. A new fillet weld design approach for CHS X-connections, using weld
effective lengths, is proposed.
ii
ACKNOWLEDGEMENTS
First and foremost, I would like to thank my supervisor, Professor Jeffrey A. Packer, for his support,
guidance and dedication to this project and my professional development. I would like to also thank the
technical staff at the University of Toronto Structural Testing Facility: Mr. R. Basset, Mr. J. MacDonald, Mr.
M. Sun, Mr. G. Buzzeo, Mr. B. Cook, Mr. M. Fiss, and Mr. A McClenaghan.
Financial support for this project was provided by the American Institute of Steel Construction (AISC),
the Natural Sciences and Engineering Research Council of Canada (NSERC), and the Steel Structures
Education Foundation (SSEF). Hollow structural sections were donated by Atlas Tube, Harrow, Ontario, and
fabrication services were donated by Walters Inc., Hamilton, Ontario.
Additional thanks are extended to Kubes Steel, Hamilton, Ontario, for fabricating the CHS-to-rigid plate
connection specimens, and to my colleagues Mr. C. Ritchie and Mr. P. Oatway for their ongoing interest in
this work. The laboratory contributions of Dr. G. S. Frater, Dr. M. Sun, Mr. P. Oatway, Ms. J. Lu, and Mr. F.
Wei are gratefully acknowledged.
iii
TABLE OF CONTENTS
Abstract .................................................................................................................................................. ii
Acknowledgements ............................................................................................................................................ iii
Table of Contents ............................................................................................................................................... iv
List of Tables ................................................................................................................................................. ix
List of Figures ............................................................................................................................................... xiii
Abbreviations and Symbols ............................................................................................................................. xix
Chapter 1: Introduction ............................................................................................................................... 1
1.1. Hollow Structural Sections ....................................................................................................... 1
1.2. Weld Effective Lengths ............................................................................................................ 2
1.3. Research Program Goal and Overview ..................................................................................... 3
Chapter 2: Weld Effective Lengths in RHS Overlapped K-Connections ................................................... 6
2.1. Introduction ............................................................................................................................... 6
2.2. Experimentation ........................................................................................................................ 7
2.2.1. Scope............................................................................................................................ 7
2.2.2. Truss Design ................................................................................................................ 7
2.3. Geometric and Mechanical Properties .................................................................................... 11
2.4. Instrumentation and Loading Strategy .................................................................................... 13
2.4.1. Instrumentation .......................................................................................................... 13
2.4.2. Loading Strategy ........................................................................................................ 14
2.5. Results ..................................................................................................................................... 16
2.6. Evaluation of AISC 360 (2010) .............................................................................................. 20
2.6.1. AISC 360 (2010) Provisions for Weld Effective Lengths in RHS Overlapped
K-Connections ........................................................................................................... 20
2.6.2. Safety Level Inherent in AISC 360 (2010) ................................................................ 22
2.7. Recommendation .................................................................................................................... 23
2.7.1. Background ................................................................................................................ 23
2.7.2. Proposal ..................................................................................................................... 23
2.7.3. Safety Level Inherent in Recommendation................................................................ 25
2.7.4. Comments .................................................................................................................. 25
2.8. Summary ................................................................................................................................. 27
iv
Table of Contents v
2.8.1. Footnote ..................................................................................................................... 27

Chapter 3: Analysis of RHS Trusses with Overlapped K-Connections ................................................... 28
3.1. Introduction ............................................................................................................................. 28
3.1.1. Recommendations for RHS Truss Analysis .............................................................. 28
3.1.2. Previous Research ...................................................................................................... 29
3.2. Experimentation ...................................................................................................................... 30
3.3. Results ..................................................................................................................................... 32
3.4. Evaluation of Frame Analysis Models .................................................................................... 35
3.4.1. Comparison of Experimental and Theoretical Truss Behaviour ................................ 36
3.5. Summary ................................................................................................................................. 40
Chapter 4: Fillet Welds in HSS-to-Rigid Plate Connections: Experimentation ....................................... 41
4.1. Introduction ............................................................................................................................. 41
4.1.1. Weld Design Philosophies for HSS Connections ...................................................... 41
4.1.2. Effect of Loading Angle on Fillet Weld Behaviour................................................... 42
4.2. Design Specifications.............................................................................................................. 46
4.2.1. ANSI/AISC 360 (2016) ............................................................................................. 46
4.2.2. CAN/CSA S16 (2001) ............................................................................................... 47
4.2.3. CAN/CSA S16 (2014) ............................................................................................... 47
4.2.4. EN1993-1-8 (2005) .................................................................................................... 48
4.3. Experimentation ...................................................................................................................... 50
4.3.1. Geometric Properties ................................................................................................. 50
4.3.2. Mechanical Properties................................................................................................ 54
4.3.3. Instrumentation .......................................................................................................... 54
4.3.4. Connection Tests ....................................................................................................... 56
4.4. Results and Analysis ............................................................................................................... 57
4.4.1. ANSI/AISC 360 (2016) ............................................................................................. 58
4.4.2. CAN/CSA S16 (2014) ............................................................................................... 59
4.4.3. CAN/CSA S16 (2001) ............................................................................................... 59
4.4.4. EN1993-1-8 (2005) .................................................................................................... 60
4.4.5. Evaluation of Directional Strength-Increase Factor .................................................. 63
4.4.6. Isolation of CHS-to-rigid plate Test Results.............................................................. 63
4.5. Influence of Weld Size............................................................................................................ 64
4.6. Summary ................................................................................................................................. 65
4.6.1. Footnote ..................................................................................................................... 65
Chapter 5: Fillet Welds in HSS-to-Rigid Plate Connections: Finite Element Modelling......................... 66
5.1. Introduction ............................................................................................................................. 66
Weld Effective Lengths for Hollow Structural Section Connections

Table of Contents vi
5.2. Previous Experimentation ....................................................................................................... 67

5.3. Finite Element Models ............................................................................................................ 67
5.3.1. General ....................................................................................................................... 68
5.3.2. Material Modelling .................................................................................................... 70
5.3.3. Model Sensitivity Study............................................................................................. 73
5.3.4. Model Fracture Criterion ........................................................................................... 75
5.4. Finite Element Models Evaluated Against Experiments ........................................................ 76
5.4.1. Ultimate Loads and Failure Mode ............................................................................. 76
5.4.2. Load-Displacement Response.................................................................................... 78
5.5. Finite Element Parametric Study ............................................................................................ 79
5.5.1. Results and Analysis .................................................................................................. 81
5.6. Proposed Alternate Model for Fillet Weld Strength ............................................................... 86
5.7. Reliability Analysis of FE Results for Design Codes ............................................................. 88
5.8. Fillet Weld Size to Develop HSS Branch Yield Strength ....................................................... 90
5.9. Summary ................................................................................................................................. 92
Chapter 6: Fillet Weld Effective Lengths in CHS X-Connections: Experimentation .............................. 93
6.1. Introduction ............................................................................................................................. 93
6.2. Test Specimen Preparation and Material Testing ................................................................... 94
6.2.1. Connection Geometric Considerations ...................................................................... 96
6.2.2. Geometric Properties of the As-Laid Welds .............................................................. 98
6.2.3. Mechanical Properties of the As-Laid Welds .......................................................... 103
6.3. Connection Tests and Instrumentation .................................................................................. 103
6.4. Discussion of Results ............................................................................................................ 106
6.4.1. Applied Load Versus Deformation Response.......................................................... 106
6.4.2. Non-Uniform Strain Distributions Adjacent to the Weld ........................................ 108
6.5. Evaluation of AWS D1.1 (2015)........................................................................................... 109
6.5.1. Existing Provisions for Weld Effective Lengths in CHS X-Connections ............... 109
6.5.2. Safety Level Inherent in AWS D1.1 (2015) ............................................................ 110
6.5.3. Comparison to AISC 360 (2016) and CSA S16 (2014) ........................................... 112
6.5.4. Evaluation of AWS D1.1 (2015) Total Weld Length Approximations ................... 114
6.6. Evaluation of Procedures to Determine Weld Effective Lengths from Tests ....................... 116
6.7. Summary ............................................................................................................................... 119
Chapter 7: Fillet Weld Effective Lengths in CHS X-Connections: Finite Element Modelling Study ... 120
7.1. Introduction ........................................................................................................................... 120
7.2. Fillet Weld Design for CHS X-Connections......................................................................... 121
7.2.1. AWS D1.1 (2015) .................................................................................................... 121

Table of Contents vii
7.2.2. CIDECT (Wardenier et al. 2008) ............................................................................. 122

7.2.3. Caulkins (1968) and Marshall (1992) ...................................................................... 123
7.3. Finite Element Modelling ..................................................................................................... 124
7.3.1. Material Properties................................................................................................... 126
7.3.2. Model Sensitivity Study........................................................................................... 128
7.4. Finite Element Models Evaluated Against Experimental Results ........................................ 132
7.5. Finite Element Parametric Study .......................................................................................... 135
7.5.1. Details of Parametric Models .................................................................................. 136
7.6. Results and Evaluation of the Parametric Study ................................................................... 137
7.6.1. Effect of Chord Slenderness and Branch-to-Chord Diameter Ratio ........................ 139
7.6.2. Effect of Branch Inclination Angle .......................................................................... 140
7.6.3. Effect of Branch-to-Chord Thickness Ratio ............................................................ 141
7.7. Regression Analysis .............................................................................................................. 141
7.8. Evaluation of Design Methods .............................................................................................. 143
7.8.1. Fit-For-Purpose Design Methods for Fillet Welds .................................................. 143
7.9. Summary ............................................................................................................................... 146
Chapter 8: Conclusions ........................................................................................................................... 147
8.1. Conclusions and Impacts of Design Recommendations ....................................................... 147
8.2. Recommendations for Future Research ................................................................................ 148
References .............................................................................................................................................. 150
Weld Effective Lengths in RHS Overlapped K-Connections ............................................... 159
A.1. Truss Fabrication Drawings .................................................................................................. 159
A.2. Measured Weld Dimensions for Fillet and PJP Welds in RHS Overlapped K-
Connections........................................................................................................................... 168
A.3. Tensile Coupon Test Results for RHS and Weld Metal ....................................................... 173
A.4. Photographs of Truss Supports and Point Load Device........................................................ 174
A.5. Photographs of Test Connections and Failure Modes........................................................... 175
A.6. Variations in Strain Around the Branch Perimeter for Additional RHS Overlapped K-
Connections........................................................................................................................... 177
Analysis of RHS Trusses with Overlapped K-Connections ................................................. 180
B.1. Measurements and Section Properties for RHS Truss Members .......................................... 180
B.2. Section Property Formulae According to EN10219-2 (2006) .............................................. 181
B.3. Truss Element Designations and LVDT Photographs .......................................................... 183
B.4. Panel Point Deflection Data .................................................................................................. 184
B.5. Axial Load Data .................................................................................................................... 187
B.6. Bending Moment Data for Test 4.2 ...................................................................................... 189
Fillet Welds in HSS-to-Rigid Plate Connections: Experimentation ..................................... 191

Table of Contents viii
C.1. Fillet Weld Resistance According to EN1993-1-8 (2005) for a Skewed-T Fillet Weld ....... 191
C.2. CHS-to-Rigid Plate Fabrication Drawings ........................................................................... 193
C.3. Weld Measurements for CHS-to-Rigid plate Connections ................................................... 197
C.4. Theoretical Throat Dimension of a Skewed-T Fillet Weld ................................................... 204
C.5. Material Property Test Results .............................................................................................. 206
C.6. Photographs of Experimental Tests and Failure Modes for CHS-to-Rigid Plate Tests ........ 207
C.7. Examples of Weld Strength Calculation Method for Fillet-Welded CHS-to-Rigid Plate
Connections........................................................................................................................... 212
C.8. Approach to Calculating Loading Angle for CHS-to-Plate Connections ............................. 214
C.9. Template Length Formula for CHS-to-Plate Connections.................................................... 216
C.10. Summary of Predicted Weld Strengths for HSS-to-Rigid Plate Connections ...................... 217
Fillet Welds in HSS-to-Rigid Plate Connections: Finite Element Modelling....................... 219
D.1. Supplementary Figures ......................................................................................................... 219
D.2. Summary of Actual and Predicted Strengths According to Code Design Methods for
FE CHS- and RHS-to-Rigid Plate Tests ............................................................................... 223
Fillet Weld Effective Lengths in CHS X-Connections: Experimentation ............................ 225
E.1. CHS-to-CHS X-Connection Fabrication Drawings .............................................................. 225
E.2. Template Length Formula for CHS-to-CHS Connections .................................................... 233
E.3. Measured Fillet Weld Dimensions for CHS-to-CHS X-Connections................................... 236
E.4. 3D Solidworks Weld Profiles for All Test Welds................................................................. 243
E.5. Post-Rupture Macroetch Measurements ............................................................................... 246
E.6. Material Property Test Results.............................................................................................. 247
E.7. Photographs of Experimental Tests and Failure Modes ....................................................... 248
E.8. Summary of Actual and Predicted Strengths According to Code Design Methods for
FE CHS X-Connection Tests ................................................................................................ 253
Fillet Weld Effective Lengths in CHS X-Connections: Finite Element Modelling .............. 254
F.1. Complete Parametric Study Results ...................................................................................... 254
F.2. Correlation Plots and Reliability Analysis Parameters for Code Design Methods ............... 259

LIST OF TABLES
Table 2.1. Measured properties of nine RHS overlapped K- (test) connections ................................................ 8
Table 2.2. Average effective weld throat thickness for individual weld elements ........................................... 12
Table 2.3. All-weld-metal tensile coupon test results for RHS truss tests ....................................................... 13
Table 2.4. Actual and predicted nominal weld strength for each test connection (actual-to-predicted
ratio shown in parentheses) ......................................................................................................... 19
Table 2.5. Mean actual-to-predicted electrode strength (ρM) and its associated variation (VM) amongst
typical weld metal ....................................................................................................................... 23
Table 3.1. Measured and calculated section properties for the RHS truss members ....................................... 31
Table 3.2. Mean percentage errors for theoretical (predicted) axial forces ...................................................... 37
Table 3.3. Comparison of experimental and theoretical (predicted) maximum truss deflections .................... 39
Table 4.1. Comparison of fillet weld effective throats required to develop the yield resistance of a
connected RHS branch member wall (updated from McFadden et al. 2013) ............................. 42
Table 4.2. Measured geometric properties, failure loads, and failure modes for θ = 90° RHS-to-rigid
plate specimens ........................................................................................................................... 51
Table 4.3. Measured geometric properties, failure loads, and failure modes for θ = 60° RHS-to-rigid
plate specimens ........................................................................................................................... 52
Table 4.4. Measured geometric properties, failure loads, and failure modes for θ = 90° and θ = 60°
CHS-to-rigid plate specimens ..................................................................................................... 52
Table 4.5. Measured material properties for all 33 HSS-to-rigid plate connections ........................................ 54
Table 5.1. FE mesh sensitivity study results .................................................................................................... 74
Table 5.2. Comparison of experimental and FE results for HSS-to-rigid plate tests ....................................... 77
Table 5.3. Non-dimensional parameters and FE parametric study results (PFE/AwFEXX and failure mode)
for RHS-to-rigid plate connections ............................................................................................. 81
Table 5.4. Non-dimensional parameters and FE parametric study results (PFE/AwFEXX and failure mode)
for CHS-to-rigid plate connections ............................................................................................. 81
Table 5.5. Reliability analysis parameters with respect to 21 weld-critical RHS-to-rigid plate FE test
results with θ = 90°...................................................................................................................... 89
Table 5.6. Reliability analysis parameters with respect to 20 weld-critical CHS-to-rigid plate FE test
results θ = 90°.............................................................................................................................. 89
Table 6.1. Measured properties of 12 CHS X- (test) connections.................................................................... 95
Table 6.2. Weld dimensions and predicted fracture loads for test joints according to existing AWS
D1.1-15 provisions for weld effective lengths in CHS X-connections ..................................... 102
Table 6.3. All-weld-metal tensile coupon test results for CHS X-connection tests ....................................... 103
ix
List of Tables x
Table 6.4. Residual chord deformation (at start of test) and chord deformation at rupture for all 12 tests.... 107
Table 6.5. Reliability analysis parameters for 12 CHS X-Connections ......................................................... 111
Table 6.6. Weld effective length ratios for CHS X-connections determined using three possible
methods ..................................................................................................................................... 117
Table 7.1. Geometric properties of CHS X- (test) connections and comparison of experimental and FE
results ........................................................................................................................................ 124
Table 7.2. Mesh sensitivity study results for test 102-273-90a ...................................................................... 129
Table 7.3. Effect of X-connection model size (scalability) ............................................................................ 132
Table 7.4. Yield stress and ultimate stress of materials used in the parametric models................................. 137
Table 7.5. Reliability analysis parameters for proposed procedure with ϕ = 0.80, 0.75 and 0.67 ................. 144
Table A.1. List of truss fabrication drawings ............................................................................................... 159

Table A.2. Weld measurements for test K-90-0.50a .................................................................................... 168
Table A.3. Weld measurements for test K-60-0.50 ...................................................................................... 169
Table A.6. Weld measurements for test K-90-0.50b .................................................................................... 170
Table B.1. Measured RHS section properties .............................................................................................. 180
Table B.2. Calculated RHS section properties ............................................................................................. 180
Table B.3. Panel point deflections in test 4.1 (applied load = 483.3 kN to panel point 4) ........................... 184
Table B.4. Panel point deflections in test 4.2 (applied load = 489.6 kN to panel point 4) ........................... 184
Table B.5. Panel point deflections in test 4.3 (applied load = 485.3 kN to panel point 4)........................... 184
Table B.6. Panel point deflections in test 10 (applied load = 490.4 kN applied to panel point 10) ............. 185
Table B.7. Panel point deflections in test 6.1 (applied load = 496.5 kN applied to panel point 6) .............. 185
Table B.8. Panel point deflections in test 8 (applied load = 497.4 kN applied to panel point 8) ................. 185
Table B.14. Axial forces by truss model in test 4.1 ....................................................................................... 187
Table B.15. Mean percentage errors in axial force predictions in test 4.1 ..................................................... 187
Table B.16. Axial forces by truss model in test 4.2 ....................................................................................... 188

List of Tables xi
Table B.17. Mean percentage errors in axial force predictions in test 4.2 ..................................................... 188
Table C.1. List of CHS-to-rigid plate test specimen fabrication drawings .................................................. 193
Table C.2. Weld measurements for CHS-to-rigid plate test no. 22 ............................................................. 197
Table C.14. Application of component approach to CSA S16-01 with and without the sinθ factor for
test no. 33 .................................................................................................................................. 212
Table C.15. Application of component approach to AISC 360 (2016), CSA S16 (2014), and EN1993-
1-8 (2005) for test no. 33 ........................................................................................................... 213
Table C.16. Summary of actual and predicted strengths for all experimental RHS- and CHS-to-rigid
plate connections according to AISC 360-16, CSA S16-14, and CSA S16-01 ......................... 217
Table C.17. Summary of actual and predicted strengths for RHS- and CHS-to-rigid plate connections
according to EN1993-1-8:2005 Directional Method ................................................................. 218
Table D.1. Summary of actual and predicted nominal strengths for 21 FE RHS-to-rigid plate
connections according to AISC 360-16, CSA S16-14, and EN1993-1-8:2005 Directional
and Simplified Methods (θ = 90° connections) ......................................................................... 223
Table D.2. Summary of actual and predicted nominal strengths for 20 FE CHS-to-rigid plate
connections according to AISC 360-16, CSA S16-14, EN1993-1-8:2005 Directional and
Simplified Methods (θ = 90° connections)................................................................................ 224
Table E.1. List of CHS-to-CHS X-connection fabrication drawings ........................................................... 225
Table E.2. Weld measurements for 102-273-90a......................................................................................... 236
Table E.3. Weld measurements for 102-273-90b ........................................................................................ 237

List of Tables xii

Table E.14. Fillet weld macroetch measurements for CHS X-connections ................................................... 246
Table E.15. Comparison of average weld throat dimensions obtained from post-rupture macro-etch
examinations and external measurements ................................................................................. 246
Table E.16. Summary of actual and predicted strengths for all experimental CHS X-connections
according to AWS D1.1-15, AISC 360-16, and CSA S16-14 ................................................... 253
Table F.1. Parametric study results .............................................................................................................. 254
Table F.2. Reliability analysis parameters for code design methods ........................................................... 259

LIST OF FIGURES
Fig. 1.1. Applications of HSS ........................................................................................................................... 1

Fig. 2.1. Elevation of the truss, dimensions and joint designations .................................................................. 9
Fig. 2.2. Terminology for RHS overlapped K-connections and weld details (including labelling
convention) for test joints ............................................................................................................ 10
Fig. 2.3. Typical connection detail drawing (shown for test K-30-0.50a or K-30-0.50b) .............................. 11
Fig. 2.4. Typical measurement procedure and mould profile ......................................................................... 12
Fig. 2.5. Spacing of strain gauges around the branch footprint adjacent to the welded connection ............... 14
Fig. 2.6. Laboratory testing arrangement for full-scale RHS overlapped K-connection experiments............ 15
Fig. 2.7. Typical weld fractures (and instrumentation) ................................................................................... 16
Fig. 2.8. Typical MTS load versus branch load magnitude relationship (shown for test K-30-0.50a) ........... 17
Fig. 2.9. Typical distribution of normal strain around branch perimeter for specimens with Ov = 30%,
= 0.50 and B/t = 27.5 ................................................................................................................ 18
= 0.50 and B/t = 27.5 ................................................................................................................ 18
= 0.50 and B/t = 27.5 ................................................................................................................ 19
Fig. 2.12. Weld effective length dimensions when 50% ≤ Ov .......................................................................... 21
Fig. 2.13. Correlation with all test results for AISC 360-10 provisions ........................................................... 24
Fig. 2.14. Correlation with all test results for modified AISC 360-10 provisions ............................................ 24
Fig. 2.15. Correlation with all test results without using weld effective length rules....................................... 24
Fig. 2.16. Correlation with current test results for AISC 360-10 provisions .................................................... 25
Fig. 2.17. Correlation with current test results for modified AISC 360-10 provisions .................................... 25
Fig. 2.18. Correlation with current test results without using weld effective length rules ............................... 26
Fig. 3.1. RHS (a) overlapped and (b) gapped K-connections, adapted from Packer & Henderson
(1997) .......................................................................................................................................... 28
Fig. 3.2. Truss centreline dimensions and e values at connections................................................................. 30
Fig. 3.3. SG and LVDT locations ................................................................................................................... 31
Fig. 3.4. Overall view of the testing arrangement .......................................................................................... 32
Fig. 3.5. Potential measured in-plane strain distribution for truss members .................................................. 32
Fig. 3.6. Experimental (a) axial force and (b) bending moment distributions for test 4.1 .............................. 34
Fig. 3.7. Assumptions in models (a) P, (b) PR, (c) R, and (d) RR with respect to noding at interior
joints, and at corner joints ........................................................................................................... 35
xiii
List of Figures xiv
Fig. 3.8. Comparison of experimental and theoretical bending moment distributions for test 4.1 ................. 38
Fig. 4.1. Fillet-welded (a) lap-splice, (b) cruciform, and (c) HSS-to-rigid plate test specimens .................... 43
Fig. 4.2. Stress components in the plane of the weld effective throat ............................................................ 49
Fig. 4.3. Connection specimens (with RHS or CHS members) ...................................................................... 50
Fig. 4.4. Examples of fillet weld throat measurements from the macroetch examinations ............................ 53
Fig. 4.5. Typical load-strain curves from four strain gauges on one side of RHS (Group B strain
gauges, shown for test no. 5) ....................................................................................................... 55
Fig. 4.6. Instrumentation and test set-up ......................................................................................................... 56
Fig. 4.7. RHS- and CHS-to-rigid plate specimens at failure .......................................................................... 57
Fig. 4.8. Comparison of actual and nominal strengths per AISC 360-16 without the sinθ factor .................. 58
Fig. 4.9. Comparison of actual and nominal strengths per AISC 360-16 with the sinθ factor ....................... 58
Fig. 4.10. Comparison of actual and nominal strengths ................................................................................... 59
Fig. 4.11. Comparison of actual and nominal strengths per CSA S16-14 with the sinθ factor ........................ 59
Fig. 4.12. Comparison of actual and nominal strengths per CSA S16-01 without the sinθ factor ................... 60
Fig. 4.13. Comparison of actual and nominal strengths per CSA S16-01 with the sinθ factor ........................ 60
Fig. 4.14. Comparison of actual strengths and design strengths per EN1993-1-8 (2005): Directional
Method ........................................................................................................................................ 62
Fig. 4.15. Effect of weld size on actual-to-nominal strength ratio per AISC 360-16 without the sinθ
factor............................................................................................................................................ 64
Fig. 4.16. Effect of branch slenderness on actual-to-nominal strength ratio per AISC 360-16 without
the sinθ factor .............................................................................................................................. 64
Fig. 5.1. HSS-to-rigid plate connection specimens (with RHS or CHS members) ........................................ 66
Fig. 5.2. FE CHS-to-rigid plate connection model with θ = 60° .................................................................... 68
Fig. 5.3. FE RHS-to-rigid plate connection model with θ = 60° .................................................................... 69
Fig. 5.4. Effective weld dimensions in FE models (shown for a model with θ = 90°) ................................... 69
Fig. 5.5. Post-necking generated T- T curve using the Ling (1996) procedure (shown for CHS 127.4 ×
11.55 TC) .................................................................................................................................... 71
Fig. 5.6. FE CHS TC (a) geometry and meshing, and (b) comparison of deformed shape and results at
fracture ........................................................................................................................................ 72
Fig. 5.7. Comparison of numerical and experimental - curves (shown for CHS 127.4 × 11.55 TC).......... 72
Fig. 5.8. Mesh arrangements shown on a 1/4 θ = 90° FE CHS-to-rigid plate connection.............................. 73
Fig. 5.9. Load-displacement curves for CHS test no. 26 with fine, medium and coarse mesh, and four
CHS through-thickness elements ................................................................................................ 74
Fig. 5.10. Progression of fracture in FE RHS-to-rigid plate test no. 13 ........................................................... 78
Fig. 5.11. Comparison of RHS-to-rigid plate FE and experimental load-displacement responses .................. 79
Fig. 5.12. Comparison of CHS-to-rigid plate FE and experimental load-displacement responses .................. 79
Fig. 5.13. Engineering stress-strain curves for the cold-formed HSS, end plate, and weld metal used in
the FE parametric study............................................................................................................... 80

List of Figures xv
Fig. 5.14. Effect of the ratio tw/tb on fillet weld strength in θ = 90° HSS-to-rigid plate connections: (a)
for RHS branches; (b) for CHS branches .................................................................................... 83
Fig. 5.15. Examples of branch rotation in fillet-welded joints to RHS ............................................................ 84
Fig. 5.16. Effect of the ratios Bb/tb and Db/tb on fillet weld strength in θ = 90° HSS-to-rigid plate
connections: (a) for RHS branches; (b) for CHS branches ......................................................... 84
Fig. 5.17. Effect of θ on fillet weld strength in HSS-to-rigid plate connections .............................................. 85
Fig. 5.18. Evaluation of design equations against all 90° (FE and experimental) HSS-to-rigid plate test
results: (a) for RHS branches; (b) for CHS branches .................................................................. 88
Fig. 6.1. Variation of X-connection stress distribution ................................................................................... 94
Fig. 6.2. Connection layout and nomenclature ............................................................................................... 96
Fig. 6.3. Local dihedral angle curves for test joints, with subtended angle measured from the crown
heel .............................................................................................................................................. 97
Fig. 6.4. Fit-up of branch to chord after profiling and tack welding .............................................................. 98
Fig. 6.5. Vector calculus method used to determine weld lengths ............................................................... 100
Fig. 6.6. 3D Solidworks models of weld profile and weld dimensions ........................................................ 101
Fig. 6.7. Typical testing arrangement (shown for test 127-273-90a)............................................................ 104
Fig. 6.8. Strain gauges near weld toe (and weld fracture) in test 127-273-90a ............................................ 105
Fig. 6.9. Typical weld fractures .................................................................................................................... 106
Fig. 6.10. Typical load versus chord deformation relationships ..................................................................... 107
Fig. 6.11. Procedure for calculation of load versus deformation for second welds tested ............................. 107
Fig. 6.12. Typical strain distributions adjacent to test weld (θ = 90° connections) ........................................ 108
Fig. 6.13. Typical strain distributions adjacent to test weld (θ = 60° connections) ........................................ 108
Fig. 6.14. Effect causing compressive strains at the crown (ρ = 0° and 180° points) (θ = 90°
connections) .............................................................................................................................. 109
Fig. 6.15. Correlation of existing AWS D1.1-15 provisions with the test results, with weld effective
lengths ....................................................................................................................................... 112
Fig. 6.16. Correlation of existing AWS D1.1-15 provisions (excluding weld effective lengths) and
AISC 360-16 provisions with test results .................................................................................. 112
Fig. 6.17. Correlation of CSA S16-14 provisions with test results................................................................. 113
Fig. 6.18. Comparison of lw/πDb using Eq. (6.6) (AWS 2015) and the vector-calculus method .................... 114
Fig. 6.19. Comparison of lw/πDb using Eq. (6.7) (AWS 2015) and the vector-calculus method .................... 115
Fig. 7.1. CHS X-connection general configuration and non-dimensional parameters ................................. 120
Fig. 7.2. Load transfer efficiency across the weld of a θ = 90° CHS-to-CHS T-connection, adapted
from Marshall (1992) ................................................................................................................ 123
Fig. 7.3. FE CHS-to-CHS X-connection geometry, mesh layout, and boundary conditions with θ =
90° ............................................................................................................................................. 125
Fig. 7.4. FE CHS-to-CHS X-connection geometry, mesh layout, and boundary conditions with θ <
90° (θ = 60° shown) .................................................................................................................. 126
Fig. 7.5. Comparison of typical experimental (solid line) and FE (dashed line) stress-strain curves........... 127

List of Figures xvi
Fig. 7.6. Mesh layouts used in the mesh sensitivity study ............................................................................ 128
Fig. 7.7. Comparison of FE and experimental mesh sensitivity parameters for test 102-273-90a ............... 129
Fig. 7.8. Ratios of Paꞌ/PFE and a/ FE for the training set for different values of ef.weld ................................ 131
Fig. 7.9. Comparison of experimental and FE load-deformation curves ...................................................... 133
Fig. 7.10. Correlation of experimental and FE results .................................................................................... 133
Fig. 7.11. Comparison of typical experimental and FE longitudinal strain distributions adjacent to the
weld ........................................................................................................................................... 134
Fig. 7.12. Compatible values of and θ to keep Ѱ between 60° and 120° along the entire weld length,
determined using methodology by Luyties & Post (1988) ........................................................ 136
Fig. 7.13. Engineering stress-strain curves for materials used in the parametric models ............................... 137
Fig. 7.14. Load-deformation curves for typical FE tests ................................................................................ 138
Fig. 7.15. Effect of chord slenderness and branch-to-chord diameter ratio on effective length ..................... 139
Fig. 7.16. Effect of the product (D/t) on effective length ............................................................................. 140
Fig. 7.17. Effect of branch inclination angle on effective length ................................................................... 140
Fig. 7.18. Effect of branch-to-chord thickness ratio on effective length ........................................................ 141
Fig. 7.19. Weld effective lengths in CHS-to-CHS X-connections with 0.10 ≤ ≤ 0.50 and 10 ≤ D/t ≤
50 according to Eq. (7.14) ......................................................................................................... 142
Fig. 7.20. Correlation of proposed design method with all test results .......................................................... 144
Fig. A.1. Truss elevation, joint layout and typical connection detail (drawing T-1) .................................... 160
Fig. A.2. Web members 11-1, 1-4, 4-5 and 5-8 cut details (according to drawing T-1) .............................. 161
Fig. A.3. Web members 8-9, 9-7, 7-6 and 6-3 cut details (according to drawing T-1) ................................ 162
Fig. A.4. Web member 3-2 and 2-12 cut details (according to drawing T-1)............................................... 163
Fig. A.5. Upper reinforcement details .......................................................................................................... 164
Fig. A.6. Lower reinforcement details .......................................................................................................... 165
Fig. A.7. Weld-critical test joints 1, 2, 5, 6 and 9 (according to drawing T-1) ............................................. 166
Fig. A.8. Weld-critical test joints 3, 4, 7 and 8 (according to drawing T-1) ................................................. 167
Fig. A.9. Weld dimensions for (a) fillet weld elements and (b) PJP weld elements .................................... 168
Fig. A.10. TC test results for RHS branch and chord materials ..................................................................... 173
Fig. A.11. All-weld-metal TC test results (E71T-1C electrode)..................................................................... 173
Fig. A.12. Truss load point, reaction, and out-of-plane support ................................................................... 174
Fig. A.13. Weld-critical overlapped K-connections with instrumentation immediately prior to testing........ 175
Fig. A.14. Weld-critical overlapped K-connections with instrumentation immediately after testing ............ 176
Fig. A.15. Distribution of normal strain around branch perimeter for test K-90-0.50a .................................. 177
Fig. A.16. Distribution of normal strain around branch perimeter for test K-30-0.50b ................................. 177
Fig. A.17. Distribution of normal strain around branch perimeter for test K-90-0.71 ................................... 178
Fig. A.18. Distribution of normal strain around branch perimeter for test K-60-0.71a .................................. 178

List of Figures xvii
Fig. A.19. Distribution of normal strain around branch perimeter for test K-60-0.71b ................................. 179
Fig. A.20. Distribution of normal strain around branch perimeter for test K-30-0.71 ................................... 179
Fig. B.1. Truss member and panel point designations used in Appendix B ................................................. 183
Fig. B.2. LVDTs transverse to the chord longitudinal axis (a) along the chord and (b) at the support ........ 183
Fig. B.3. Experimental bending moment distributions for test 4.2 ............................................................... 189
Fig. B.4. Comparison of experimental and theoretical bending moment distributions for test 4.2 .............. 190
Fig. C.1. Stress components in the plane of the weld effective throat for Ѱ = θ = 90° ................................ 191
Fig. C.2. Ninety degree CHS-to-rigid plate test specimens .......................................................................... 194
Fig. C.3. Sixty degree CHS-to-rigid plate test specimens ............................................................................ 195
Fig. C.4. All-weld-metal TC test specimen and trial weld specimens for CHS-to-rigid plate tests ............. 196
Fig. C.5. Skewed-T fillet weld profile dimensions....................................................................................... 204
Fig. C.6. TC test results for CHS branch materials ...................................................................................... 206
Fig. C.7. All-weld-metal TC test results (E71T-1C-H8 electrode) .............................................................. 206
Fig. C.8. TC test results for end-plate material (heat no. 4506365) ............................................................. 206
Fig. C.9. Photographs of test 22 ................................................................................................................... 207
Fig. C.13. Photographs of tests 26 and 27 ...................................................................................................... 209
Fig. D.1. FE CHS-to-rigid plate connection model with θ = 90° ................................................................. 219
Fig. D.2. FE RHS-to-rigid plate connection model with θ = 90° ................................................................. 220
Fig. D.3. Graphical comparison of FE versus predicted ultimate load (showing FE failure modes) ........... 220
Fig. D.4. Non-uniform FE strain distributions adjacent to the weld in RHS-to-rigid plate tests.................. 221
Fig. D.5. Comparison of Pa/Py and PFE/Py to tw/tb for FE and experimental HSS-to-rigid plate tests .......... 222
Fig. E.1. Test 127-406-90a/b specimen ........................................................................................................ 226
Fig. E.7. All-weld-metal TC test specimen, trial weld specimens, and parts list for CHS X-connection
tests ............................................................................................................................................ 232
Fig. E.8. CHS-to-CHS template geometry view no. 1 ................................................................................. 233

List of Figures xviii
Fig. E.9. CHS-to-CHS template geometry view no. 2 ................................................................................ 233

Fig. E.10. CHS-to-CHS template geometry view no. 3 ................................................................................. 233
Fig. E.11. 3D Solidworks models of weld profile for 102-series tests with θ = 90° ...................................... 243
Fig. E.12. 3D Solidworks models of weld profile for 127-series tests with θ = 90° ...................................... 244
Fig. E.13. 3D Solidworks models of weld profile for tests with θ = 90° ....................................................... 245
Fig. E.14. TC test results for CHS chord and branch materials...................................................................... 247
Fig. E.15. All-weld-metal TC test results (E71T-1C electrode) ..................................................................... 247
Fig. E.16. Photographs of test 102-273-90a ................................................................................................... 248
Fig. E.17. Photographs of test 102-273-90b and 102-406-90a ....................................................................... 248
Fig. E.18. Photographs of test 102-406-90b ................................................................................................... 249
Fig. E.19. Photographs of tests 127-273-90a and 127-273-90b ..................................................................... 249
Fig. E.21. Photograph of test 127-406-90b .................................................................................................... 250
Fig. E.23. Photograph of test 102-406-60b .................................................................................................... 251
Fig. E.24. Photographs of tests 127-406-60a and 127-406-60b ..................................................................... 252
Fig. F.1. Correlation of existing AWS D1.1-15 provisions with all test results, with weld effective
lengths ....................................................................................................................................... 259
Fig. F.2. Correlation of existing AWS D1.1-15 provisions (excluding weld effective lengths) and
AISC 360-16 provisions with all test results ............................................................................. 260
Fig. F.3. Correlation of CSA S16-14 provisions with all test results ........................................................... 260

ABBREVIATIONS AND SYMBOLS
AISC = American Institute of Steel Construction

AWS = American Welding Society
CSA = Canadian Standards Association
CIDECT = International Committee for the Development and Study of Tubular Construction
CISC = Canadian Institute of Steel Construction
CEN = European Committee for Standardization
IIW = International Institute of Welding
ISO = International Organization for Standardization
CHS = circular hollow section

COV = coefficient of variation
FCAW = flux-cored arc welding
FE = finite element
HSS = hollow structural section
LRFD = load and resistance factor design
LSD = limit states design
LVDT = linearly varying differential transformer
NDT = non-destructive test
PTFE = polytetrafluoroethylene
RHS = rectangular hollow section
SG = strain gauge
TC = tensile coupon
A = cross-sectional area of the HSS chord

Ab = cross-sectional area of the HSS branch
Ag = area of the complement of the outside RHS corner
Am = area of fusion face between weld and base metal
Aw = effective throat area of weld (= twlw)
A = area of the complement of the inside RHS corner
B = overall width of the RHS chord, measured normal to the plane of the connection
Bb = overall width of the RHS branch, measured normal to the plane of the connection
Bbi = overall width of the overlapping RHS branch
Bbj = overall width of the overlapped RHS branch
C = regression constant
D = overall diameter of the CHS chord
xix
Abbreviations and Symbols xx
Db = overall diameter of the CHS branch

E = Young’s modulus
FEXX = electrode ultimate strength
Fnw = nominal strength of the weld metal per unit area
Fu = ultimate stress of the HSS
Fup = ultimate stress of the plate
Fy = yield stress of the HSS chord
Fyb = yield stress of the HSS branch
Fybi = yield stress of the overlapping RHS branch
Fybj = yield stress of the overlapped RHS branch
Fyp = yield stress of the plate
Fyw = yield stress of weld metal
H = overall height of the RHS chord, measured in the plane of the connection
Hb = overall height of the RHS branch, measured in the plane of the connection
Hbi = overall depth of the overlapping RHS branch
Hbj = overall depth of the overlapped RHS branch
I = moment of inertia of the HSS
Ig = moment of inertia of the outside RHS corner regions
I = moment of inertia of the inside RHS corner regions
Ka = weld length factor
Me = experimental bending moment in the truss, measured with SGs
Mw = strength reduction factor in CSA S16-14 to allow for the variation in deformation
capacity of weld elements with different orientations
Ov = overlap (%) = (q/p × 100) %
P = applied force
P3%D,a = actual (experimental) load at connection plastification limit
P3%D,FE = finite element load at connection plastification limit
PFE = failure load in the finite element model
PFE* = FE load at experimental weld rupture displacement.
Pa = actual (experimental) ultimate load
Paꞌ = greatest load sustained by the weld
Pe = experimental axial force in the truss, measured with SGs
𝑃⃗𝑖 = position vector at point i along the branch-chord intersection
Pnw = nominal/predicted weld fracture load
Pnwl = total nominal strength of longitudinally loaded fillet welds
Pnwt = total nominal strength of transversely loaded fillet welds (without sinθ factor
applied)
Pnw = nominal strength of the weld
Pr = required weld strength using LRFD load combinations
Py = yield load of the HSS branch
Qw = shear strength of weld per unit length
S = elastic section modulus of the HSS
⃗
𝑉 = approximation to the weld length between two points along the weld

Abbreviations and Symbols xxi
VG = coefficient of variation of ρG
VM = coefficient of variation of ρM
VP = coefficient of variation of ρP
VR = coefficient of variation of ρR
Vrw = design resistance of the fillet weld
Xu = electrode ultimate strength per CSA S16
Z = plastic section modulus of the HSS
a = distance from the chord face of the SGs measured along the RHS web centreline;
regression constant
b = distance from the heel of the SGs measured along the RHS chord centreline;
regression constant
beoi = effective length of the weld to the chord
beov = effective length of the weld to the overlapped branch
d = greatest perpendicular dimension measured from a line flush to the base metal
surface to the weld surface
e = eccentricity in a truss connection, positive being away from the branches;
loading eccentricity of the fillet weld
g = gap between web members at a truss panel point (ignoring welds) on the chord face
hg = length of outside RHS corner radius contour
h = length of inside RHS corner radius contour
i = subscript/term used to identify the overlapping branch member; subscript/term used
to identify weld elements; symbol denoting ρ or ρ + Δρ
j = subscript/term used to identify the overlapped branch member
kFE = finite element initial stiffness
ka = actual (experimental) initial stiffness
l = length of the chord
lb = web (or branch) member length between node points; length of the branch
le = weld effective length
le,i = effective length of weld to the overlapping branch
lh = weld leg measured along the plate; weld leg measured along the HSS chord
lt = template length
lv = weld leg measured along the HSS branch
lw = total weld length
p = projected length of the overlapping branch on the connecting face of the chord
q = overlap length, measured along the connecting face of the chord beneath the region
of overlap of the branches
r = radius of gyration of the HSS
ro = average outside corner radius of RHS
ri = average inside corner radius of RHS
t = wall thickness of the HSS chord member
tb = wall thickness of the HSS branch member
tbi = wall thickness of the overlapping RHS branch member
tbj = wall thickness of the overlapped RHS branch member

Abbreviations and Symbols xxii
tp = plate thickness
tw = weld effective throat dimension
w = weighting factor
x = distance between the collar and the weld toe along the branch
Δmax = maximum truss deflection
Δρ = subtended angle increment
α = chord length parameter (=2l/D)
αR = coefficient of separation (taken to be 0.55)
= ratio of overall branch width to chord width for RHS; ratio of overall branch
diameter to chord diameter for CHS; theoretical angle between the plane of the
effective throat and the fusion face corresponding to lv
w = correlation factor for fillet welds per EN1993-1-8
+
= safety index
= theoretical angle between the plane of the effective throat and the fusion face
corresponding to lh; half diameter-to-thickness ratio of the chord (=D/2t)
M2 = partial safety factor of 1.25 for the resistance of weld in EN1003-1-8:2005
= chord deformation
FE = finite element chord deformation at rupture
a = actual (experimental) chord deformation at rupture
= engineering strain
T = true strain
Tꞌ = true strain at the start of coupon necking
avg = average strain
bottom = in-plane strain on the bottom face of the RHS
e = equivalent (von Mises) strain
ef = equivalent (von Mises) strain at rupture for failure criterion
ef,plate = equivalent (von Mises) strain at rupture for plate failure criterion
ef,weld = equivalent (von Mises) strain at rupture for weld failure criterion
rup = strain at rupture
top = in-plane strain on the top face of the RHS
u = ultimate engineering strain
y = yield strain
θ = angle of loading measured from the weld longitudinal axis for fillet weld strength
calculation (in degrees); branch inclination angle
θ1 = angle of loading (in degrees) of the weld element under consideration
θ2 = angle of loading (in degrees) of the weld element in the joint that is nearest to 90˚
θi = included angle between the overlapping branch and chord
θj = included angle between the overlapped branch and chord
ρ = subtended angle around the branch, measured from heel
ρG = mean ratio of the measured-to-nominal weld throat area
ρM = mean ratio of the measured-to-nominal electrode ultimate strength
ρP = mean test-to-predicted capacity ratio
ρR = bias coefficient for the resistance

Abbreviations and Symbols xxiii
= engineering stress
T = true stress
Tꞌ = true stress at the start of coupon necking
u = ultimate engineering stress
𝜎⊥ = normal stress perpendicular to the plane of the throat
𝜎∥ = normal stress parallel to the axis of the weld
= branch-to-chord thickness ratio
𝜏⊥ = shear stress (in the plane of the throat) perpendicular to the axis of the weld
𝜏∥ = shear stress (in the plane of the throat) parallel to the axis of the weld
ϕ = resistance factor (associated with the LRFD or LSD method)
ϕ+ = adjustment factor for +
Ψ = local dihedral angle (angle between the base metal fusion faces)

Chapter 1: INTRODUCTION
1.1. HOLLOW STRUCTURAL SECTIONS

Hollow structural section (HSS) production began in 1952, in Corby, England, and today accounts for a
large share of the steel market in Europe, Asia, and North America. HSS are often specified for their aesthetic
characteristics (Fig. 1.1), but they are also technically superior to other steel sections in many applications.
Higher strength-to-weight ratio, torsional stiffness, and lower surface area, are just a few examples of the
technical advantages that can be achieved with HSS. Over the last 60 years, the International Committee for
the Development and Study of Tubular Construction (CIDECT) has sponsored research programs to
investigate HSS member stability, fire protection, composite construction, and connection static and fatigue
behaviour, with the results of these programs being incorporated into national and international codes,
standards and guidelines, including The American Institute of Steel Construction (AISC) Specification for
Structural Steel Buildings (AISC 2010), Canadian Institute of Steel Construction (CISC) Guide (Packer &
Henderson 1997) and Eurocode 3: Design of Steel Structures (CEN 2005). Research on tubular structures is
ongoing, with particular emphasis on blast resistance, high strain-rate loading, manufacturing, and
connections.
(a) Retractable stadium roof at the Rogers Centre in (b) Beijing National Stadium in Beijing, China
Toronto, Canada
Fig. 1.1. Applications of HSS
1
Chapter 1: Introduction 2
A simple and cost-effective way to make connections between HSS members is to weld them directly
together. Directly welded HSS connections prevent the need for additional material, such as gusset plates,
and time-intensive fabrication. For these so-called standard truss connections, smaller HSS branch members
are connected to larger HSS chord members by means of a weld around the perimeter of the intersection of
the two members. These welds experience non-uniform loading as a result of the HSS walls being non-
uniformly flexible. Stiffer elements in the walls, such as corners in a rectangular hollow section (RHS) chord,
attract more load. The stresses in the weld at the location of these stiffer elements are thus higher than the
average stress in the weld, which can lead to premature rupture. In many cases, conventional design equations
for predicting the strength of such welds to HSS have been questioned (Davies & Packer 1982; Frater &
Packer 1992a, 1992b; Packer & Cassidy 1995; McFadden & Packer 2014; Tousignant & Packer 2014).
In 1989, the International Institute of Welding (IIW) recommended that welds to HSS branches be
designed to develop the yield capacity of the member, so that they are capable of resisting, by default, any
arrangement of loads in the member. This remains today as one of two acceptable design methods for welds
between HSS. The second method, recognized by the International Organization for Standardization (ISO
2013), uses weld effective lengths to explicitly take into account the weld non-uniform loading.
1.2. WELD EFFECTIVE LENGTHS

Over the past 30 years, substantial research efforts have focused on establishing a design method for
welds in RHS-to-RHS connections, which are loaded in a highly non-uniform manner. The weld effective
length (or property) concept was developed from this research to permit the design of welds in HSS
connections for actual (design) branch member forces (as opposed to the member yield capacity or the
connection capacity). Recommendations for weld effective lengths in RHS-to-RHS T-, Y-, X-, and gapped
K-connections were made by Frater & Packer (1992a, 1992b) and Packer & Cassidy (1995), and adopted by
AISC in a new Chapter K4 (Welds of Plates and Branches to RHS) of the 2010 Specification for Structural
Steel Buildings (AISC 360-10). AWS D1.1-15 Structural Welding Code – Steel (AWS 2015) also gives weld
effective lengths for some hollow section connections. Since the release of AISC 360-10, research on weld
effective lengths in moment-loaded RHS-to-RHS T-connections has also been conducted (McFadden &
Packer 2014). A detailed review of all prior research on weld effective lengths for RHS connections can be
found in Section 2.2 of Tousignant & Packer (2014).
At the time of AISC 360-10’s publication, recommendations for weld effective lengths in RHS-to-RHS
overlapped K-connections did not exist, and proposed formulae, agreed upon by the AISC 360 Connections
technical committee (TC6), were published in Chapter K4 of the same document. While thought to be
conservative, there was a paucity of experimental evidence to support the validity of these formulae.
Furthermore, since the addition of AISC 360-10 Chapter K4, there has been a prominent question about
the need for weld effective lengths (or properties) for connections between circular hollow sections (CHS).

No supporting research on weld-critical joints between CHS has hitherto been performed and, as a result,
most specifications are reluctant to provide a “fit-for-purpose” approach to weld design for CHS connections.
Both validation of the AISC 360-10 formulae for weld effective lengths in RHS-to-RHS overlapped K-
connections, and a weld design procedure for CHS-to-CHS connections based on actual branch member
forces, are needed.
1.3. RESEARCH PROGRAM GOAL AND OVERVIEW

The goal of the research presented herein is threefold:
1. to complete the recommendations for weld effective lengths in RHS truss connections by
experimentally evaluating the proposed formulae for weld effective lengths in RHS-to-RHS
overlapped K-connections;
2. to evaluate the extent to which fillet welds to RHS and CHS branch members behave differently with
regard to the weld effective length phenomenon, and to evaluate the safety of conventional design
equations in this case; and
3. to determine the effect of non-uniform flexibility on the strength of welds in CHS-to-CHS
connections, and develop an accurate (yet safe) design procedure for fillet welds in such connections.
This research first investigates the validity of the AISC 360-10 formulae for weld effective lengths in
RHS-to-RHS overlapped K-connections (Chapter 2), by testing one large-scale, 10-m span, simply-
supported, Warren truss. Nine overlapped connections, within the truss, were designed to be weld-critical and
sequentially failed by producing an axial force distribution with a point load, applied quasi-statically, to
strategic panel points. The structural reliability (or safety index) of the AISC (2010) specification formulae
for the effective length of welds in RHS overlapped K-connections was determined. The results indicate that
these provisions are conservative. Hence, a modification to the current requirement that limits the effective
width of transverse weld elements is proposed. The proposal establishes a more economical and yet still safe
weld design method for welds in such connections.
Elastic tests were also performed on the truss, using the same loading scheme, to obtain experimental
axial force, bending moment, and truss deflection patterns (Chapter 3). A total of 53 axial forces and bending
moments were measured in the chord and web members during two tests, and deflections at panel points
along the bottom chord were measured during a broader series of nine tests. The theoretical behaviour given
by practical frame analysis models, consisting of: (a) all joints pinned and concentric; (b) all joints rigid and
concentric; (c) pin-ended webs connected eccentrically to continuous chords; and (d) rigid-ended webs
connected eccentrically to continuous chords, is compared to the experimental results. Recommendations are
thus made for the analysis of welded RHS trusses with overlapped K-connections.
This research then examines the structural reliability of the (1.00+0.50sin1.5θ) directional strength-
increase factor for fillet welds used in North America (Chapter 4). Contemporary design procedures for fillet

welds to HSS in several prominent design codes are first discussed. Then, the structural reliability associated
with the (1.00+0.50sin1.5θ) factor is examined through a series of laboratory tests on fillet-welded connections
between HSS and rigid end plates. A total of 33 connection specimens, with the welds designed to be the
critical elements, was tested to failure by application of axial tension loading to the HSS members. The
experimentally obtained weld strengths were compared to the predicted nominal strengths. It was found that
the use of the directional strength-increase factor leads to unsafe strength predictions, particularly for large
weld sizes. Hence, it is determined that restrictions on the application of the (1.00+0.50sin1.5θ) factor for
welds to HSS members need to be considered. It was also found that Eurocode 3 fillet weld design provisions
give conservative strength predictions.
To extend these results, Chapter 5 presents a finite element (FE) investigation on the behaviour of fillet-
welded HSS-to-rigid plate connections. The FE models are validated by comparison of load-displacement
response and fracture load with results from the previous 33 experimental tests, and a parametric study is
performed. Seventy-three FE tests are therein conducted to evaluate the effect of weld size, HSS branch wall
slenderness and branch inclination angle, on fillet weld strength. The inherent problem with single-sided fillet
welds to a tension-loaded element is illustrated. A reliability analysis is then performed, and it is determined
that the (1.00 + 0.50sin1.5θ) is unsafe for both RHS and CHS connections, when considered separately,
especially for connections with large welds. Hence, an alternative yet safe method for estimating the strength
of fillet welds to HSS, based on weld size, is proposed. An expression for the fillet weld size required to
develop the yield strength of a 90° HSS branch member is derived.
Chapter 6 presents the first-ever laboratory-based test program conducted to assess the performance of
fillet welds in connections to a CHS chord. Six large-scale, fillet-welded CHS-to-CHS X-connections,
designed to be weld-critical, were tested under quasi-static tension applied to the ends of each branch.
Fracture of 12 test welds (two per connection) was obtained. By using the carefully measured mechanical and
geometric properties of the fillet welds and CHS members, the safety index of the existing AWS (2015)
specification provisions for weld effective lengths in CHS-to-CHS X-connections was determined. For the
range of parameters studied, the AWS formulae are found to be very conservative, hence a comprehensive
parametric FE modelling study was conducted to develop more liberal recommendations (Chapter 7).
In Chapter 7, non-linear FE models of CHS-to-CHS X-connections, with weld fracture, are validated by
comparison of spot strains, load-deformation response, and fracture load with the previous 12 experimental
tests. Two hundred and fifty-six weld-critical connections, with varied branch-to-chord diameter ratio, chord
wall slenderness, branch inclination angle, and branch-to-chord thickness ratio, are then analysed under
quasi-static branch tension. The effect of these parameters on fillet weld strength is illustrated, and the safety
index of North American specification provisions for weld effective lengths is confirmed. An alternative
method for estimating fillet weld strength, with specific weld effective lengths, is proposed.
Recommendations for a new design approach that meets the minimum target safety index in North America
are made.

Chapter 8 presents a summary of this research, its impact on engineering design practice, and
recommendations for future research.

Chapter 2: WELD EFFECTIVE LENGTHS IN RHS
OVERLAPPED K-CONNECTIONS
2.1. INTRODUCTION
It is well known that the differences in relative stiffness of rectangular hollow structural section (HSS)
walls cause non-uniform load transfer along lines of welds at a branch connection. Historically, international
design recommendations have required that these welds be designed to develop the yield strength of the
member, such that they may resist any arrangement of loads in the branch. This requirement is almost
exclusively based on old recommendations from the International Institute of Welding (IIW 1989). Designing
welds to branches to develop the yield strength of the member is justifiable in situations when there is low
confidence in the design forces, or if plastic stress redistribution is required in the connection (Packer et al.
2009). This design method is not always merited, and its requirement for large weld sizes is excessively
conservative in many situations.
Extensive laboratory tests have been performed at the University of Toronto on welds in both isolated
rectangular hollow section (RHS) connections and complete trusses (Frater & Packer 1992a, 1992b; Packer &
Cassidy 1995; McFadden et al. 2013; McFadden & Packer 2014) which have led to the development and
international recognition (IIW 2012; ISO 2013) of a more modern design approach based on actual branch
member forces to achieve more appropriate and economical weld sizes. This so-called “fit-for-purpose”
approach makes use of effective weld properties to account for the non-uniform loading of the weld
perimeter.
In the 2010 edition of AISC Specification for Structural Steel Buildings (AISC 2010), a detailed design
method considering effective weld properties for predominantly statically loaded RHS-to-RHS connections
was given in Section K4: Welds of Plates and Branches to Rectangular HSS. Table K4.1 Effective Weld
Properties for Connections to Rectangular HSS contained formulae to determine the effective length of welds
for axially-loaded RHS connections and the effective elastic section modulus of welds subject to bending.
The design methods in Table K4.1 for welds in axially-loaded T-, Y-, X- and gapped K-connections were
based on experimental data from full-scale tests on connections in which failure occurred by shear rupture
along a plane through the weld (herein called “weld-critical connections”) (Frater & Packer 1992a, 1992b;
Packer & Cassidy 1995). However, at the time that AISC 360-10 was published, no such data was available
6
Chapter 2: Weld Effective Lengths in RHS Overlapped K-Connections 7
to substantiate the design methods given, in the same document, for welds in unreinforced RHS moment T-
connections and axially-loaded RHS overlapped K-connections.
In order to evaluate the adequacy of these design methods, AISC initiated a two-phase study at the
University of Toronto. The first phase of the study investigated the strength and behaviour of welds in
unreinforced RHS moment T-connections. The results of this phase have been published by McFadden &
Packer (2014). Phase two of the study is presented herein.
2.2. EXPERIMENTATION
2.2.1. SCOPE
An experimental program was developed to test large-scale RHS overlapped K-connections in order to
verify, or adjust, the weld effective length rules defined by Equations K4-10 to K4-12 in Table K4.1 of AISC
360 (2010). Nine overlapped, 60° K-connections within one large-scale, 10-m span, simply-supported
Warren truss, were designed to be weld-critical under the application of tension to the overlapping branch.
Key parameters, such as the branch member overlap (Ov), the branch-to-chord width ratio ( -ratio) and the
chord wall slenderness (B/t), were investigated, and varied within the Limits of Applicability of Section K2.3
of the Specification (AISC 2010). The non-uniform distribution of normal strain in the branch, near the
connection, was measured with strain gauges oriented along the longitudinal axis of the member at uniform
spacing around its perimeter, and the weld strength was obtained directly from strain gauges in the constant
stress region of the branch. To induce weld rupture, a single point load was applied to various truss panel
points in a quasi-static manner. The loading strategy was carefully planned to accentuate the force in the
critical web member(s) and resulted in all nine joints failing by shear rupture along a plane through the weld.
2.2.2. TRUSS DESIGN
2.2.2.1. General
Connections were welded using a semi-automatic flux-cored-arc-welding (FCAW) process with full-CO2
shielding gas, and fabricated from members conforming to CSA G40.20/G40.21 Class C (CSA 2013a) and
ASTM A1085 (ASTM 2015a). The experimental test designations, and a summary of the key test parameters
for each joint (Ov, , and B/t), are given in Table 2.1.

Table 2.1. Measured properties of nine RHS overlapped K- (test) connections

No. a Test RHS web member RHS chord member Ov B/t
Bb × Hb × tb Ab b Fyb c B × H × t Ab Fy c
mm× mm × mm mm2 MPa mm× mm × mm mm2 MPa %
1 K-90-0.50a 90
6 K-90-0.50b 90
2 K-60-0.50 127.0 × 127.0 × 7.78 3625 412 254.4 × 254.4 × 9.24 8804 387 60 0.50 27.5
5 K-30-0.50a 30
7 K-30-0.50b 30
3 K-90-0.71 90
4 K-60-0.71a 60
127.0 × 127.0 × 7.78 3625 412 178.7 × 178.7 × 12.53 7776 380 0.71 14.3
9 K-60-0.71b 60
8 K-30-0.71 30
- T2 joint 4† 50
127.8 × 127.8 × 11.81 - 416 204.0 × 204.0 × 11.71 - 359 0.63 17.4
- T2 joint 6† 50
Note: θi = θj = 60°; and Bbj/ tbj = 16.3 for connections 1 - 9.
a
Refer to Fig. 2.1.
b
Cross-sectional areas determined by cutting a prescribed length of HSS, weighing it, and then using a density of 7850
kg/m3 (CISC 2016) to calculate its cross-sectional area.
c
Yield strength of all HSS determined from tensile coupon tests performed according to ASTM A370 (2017).
†
Tests by Frater (1991); failed by a combined mechanism of weld fracture and premature branch yielding.
The amount of overlap was varied, from 30% to 90%, and chord member sections that were selected
produced relatively rigid and flexible connections. Connections were made to an HSS 178 × 178 × 13 that
were more rigid ( = 0.71 and B/t = 14.2), and to an HSS 254 × 254 × 9.5 chord were more flexible ( = 0.50
and B/t = 27.5). Web members (HSS 127 × 127 × 8.0) were specified to minimize the ratio: predicted weld
strength/connection resistance and to also allow, by virtue of matched-width web members, either side of the
truss connection to be designated as the overlapping (or “test”) branch. The latter detail was intended to
support the design of a loading sequence to achieve sequential rupture of welds within the truss (see Section
2.4.2).
Complete truss testing has been the preferred approach for testing welds in K-connections because it
correctly accounts for connection boundary conditions (i.e. member continuity and truss deflection effects)
(Frater & Packer 1992a, 1992b, 1992c). The truss layout and its dimensions are shown in Fig. 2.1. In Fig. 2.1,
load locations for the nine tests are numbered around the truss and connection numbers (and designations) are
shown inside of the truss. The fabrication drawings for the truss, in U.S. customary units (as they were
provided to the fabricator), are provided in Appendix A.1.

Fig. 2.1. Elevation of the truss, dimensions and joint designations
2.2.2.2. Weld Joint Details

The test welds were those to the overlapping branches and each was comprised of three distinct weld
joint details (see Fig. 2.2): a longitudinal 90° fillet-weld detail (side a), a transverse 60° fillet-weld detail
(sides c and d) and a longitudinal partial-joint-penetration (PJP) flare-bevel-groove-weld detail (side a’). The
PJP flare-bevel-groove weld is formed by the butt joint in the matched-width web member connection. In this
region, the deposition of sound weld metal to the bottom of the flare can be hindered by bridging of the weld
puddle between the surfaces of the two branches (Packer & Frater 2005). Thus, the throat of such welds can
be highly variable. It should be noted that correct input for the geometric and mechanical properties of the as-
laid welds is requisite to the following analysis and thus, to establish a more reliable (precise) picture of the
weld throat in this region, a complete penetration (CP) detail was specified (with a ¼ in. root gap and
backing), and subsequently qualified in accordance with Clause 4.11 of AWS D1.1 (2015), for PJP welds.

Fig. 2.2. Terminology for RHS overlapped K-connections and weld details (including labelling convention)
for test joints
All critical test welds (to the overlapping branches at the connections), with the exception of weld
element c, were performed in the horizontal position. Weld element c was performed in the flat position.
Minimum weld sizes, as specified in Table 5.7 and Table 3.5 of AWS D1.1 (2015) and Table J2.4 and Table
J2.3 of AISC 360 (2016) for fillet welds and PJP flare-bevel-groove welds, respectively, were used to ensure
enough heat input during welding to establish a sound weld. The “hidden toe” of the overlapped branch
(beneath c in Fig. 2.2) was always welded to the chord, and the remainder of the welds in the truss were sized
so as to not fail before yielding of the attached branch member. Welding (or not welding) of the hidden toe of
the overlapped (or through) branch is believed to be an issue that affects failure of the welded joint to the
overlapped branch (when it is loaded in tension). Since the experiments herein were designed to examine
welding of the tension-loaded overlapping branch, it was decided to adopt the conservative practice of fully
welding the non-critical through branch. Section 2.7.4 provides further comments on this matter. Fig. 2.3
shows the specified weld sizes and the associated welding symbols in a typical connection detail for a joint
with Ov = 30%.

Fig. 2.3. Typical connection detail drawing (shown for test K-30-0.50a or K-30-0.50b)
During fabrication, the backing bar in detail a’ of K-60-0.50 was pried, about the tack weld (see Fig. 2.2),
away from the inside face of the HSS branch. The member itself was the last one to be fitted into the truss,
and was hammered into place. It is believed that during hammering, catching of the backing bar, which made
contact with the chord, caused it to be pried. This complication was not identified until after welding of the
opposite side (details b and b’) was complete. The resulting “gap” was filled with weld metal and welding of
the test joint proceeded. Based on non-destructive (ultrasound) test results, this was not a cause for rejection;
but it is speculated that the strength of the joint was nevertheless affected by this defect. The joint is
identified in the subsequent analysis (Figs. 2.13 – 2.18) by a red data point.
All of the test welds were ground (long after welding) to reduce the weld throat dimension to below the
minimum sizes specified by AISC and AWS. This was necessary to obtain a weld-fracture failure mode.
Since the code provisions are based on achieving the necessary heat input at the time of welding, and
minimum sizes were at that time provided, the soundness of the welds was likely unaffected.
2.3. GEOMETRIC AND MECHANICAL PROPERTIES

The measured geometric and mechanical properties of the HSS members and the nine test connections
are shown previously in Table 2.1. For the PJP welds, the throat dimension (tw) of the PJP flare-bevel-groove
welds (side a’) was measured according to Eq. (2.1), below:

tw  tbi  d (2.1)
where d = greatest perpendicular dimension measured from a line flush to the overlapping branch member
surface to the weld surface; and tbi = average measured thickness of the overlapping branch member.
For the fillet welds, tw was determined by making a negative mould of each fillet-weld element at
numerous locations along its length, as shown in Fig. 2.4. The mould was cut normal to the axis of the weld
root, scanned, and digitally measured; the effective throat was taken as the minimum distance between the
root and face of the diagrammatic weld. Fig. 2.4 also shows a cut of the mould, and a typical weld throat
measurement. Over 180 weld dimensions were taken for the nine connections (five along each of the four
sides of the connection), and the average measured values for the weld throat dimension are shown in Table
2.2. The individual measurements of weld dimensions for the fillet and PJP welds are provided in Appendix
A.2.
Fig. 2.4. Typical measurement procedure and mould profile
Table 2.2. Average effective weld throat thickness for individual weld elements
Test Measured weld throat dimension, tw
a a’ b b’ c d
mm mm mm mm mm mm
K-90-0.50a 3.45 3.18 3.12 3.58 3.76 4.27
K-90-0.50b 4.60 3.45 3.81 3.66 3.84 4.27
K-60-0.50 2.67 3.81 2.39 3.56 3.86 4.22
K-30-0.50a 3.35 4.60 2.95 3.96 4.34 3.63
K-30-0.50b 3.28 4.29 3.05 4.29 4.01 3.53
K-90-0.71 3.18 3.89 3.18 3.81 3.63 3.84
K-60-0.71a 3.99 3.51 3.86 3.12 3.78 3.86
K-60-0.71b 3.43 3.56 3.23 3.76 3.84 3.96
K-30-0.71 4.57 4.93 3.40 4.78 4.27 3.78
T2 joint 4 † 4.50 4.50 4.39 4.39 7.19 6.71
T2 joint 6 † 7.11 7.11 6.50 6.50 9.09 10.59
Note: b and b’ are analogous to a and a’ (see Fig. 2.2), but on the opposite side of the overlapping branch.
†
Tests by Frater (1991).

Mechanical properties of the as-laid welds were determined by tensile coupon (TC) tests (three total) as
specified by AWS D1.1 (2015). The results are shown in Table 2.3. The average yield stress (by 0.2% strain
offset) was 563 MPa and the average ultimate strength (FEXX) was 619 MPa with 27.5% elongation at rupture.
The measured ultimate strength was 26.3% stronger than the nominal strength of the electrode used (AWS
E71T-1C). Stress-strain curves generated from the all-weld metal TC tests are provided in Appendix A.3.
Table 2.3. All-weld-metal tensile coupon test results for RHS truss tests
All-weld-metal coupon Yield stress Young’s modulus, E FEXX Rupture strain, εrup a
designation
MPa MPa MPa %
[i] 558 202000 629 27.0
[ii] 561 200000 612 26.4
[iii] 567 219000 617 29.2
Average 563 207000 619 27.5
a
Rupture strain determined by re-joining the fractured coupon and measuring: change in gauge length / initial gauge
length
2.4. INSTRUMENTATION AND LOADING STRATEGY

2.4.1. INSTRUMENTATION
The actual (experimental) weld fracture loads (Pa) were obtained from two linear strain gauges (SGs)
located in-plane and at mid-length (in the constant elastic stress region) of the web members on the toe and
heel sides of the member (Mehrota & Govil 1972). The breaking loads were hence calculated according to
Eq. (2.2):
Pa  Ab E avg (2.2)
where Ab = cross-sectional area of the branch, determined by weighing the cross-section; E = elastic modulus
of the RHS, determined by tensile coupon tests in accordance with ASTM A370 (ASTM 2017); and ɛavg =
average strain measured on opposite faces of the RHS.
The non-uniform normal strain distribution around the branch perimeter, adjacent to the weld, was
measured using SGs oriented along the longitudinal axis of the member, and 25 mm away from the weld toe
[in order to avoid the strain concentrations caused by the notch effect (Packer & Cassidy 1995)]. Since the
strain distribution is theoretically symmetric about the y-y axis of the member (for plane-frame behaviour),
SGs were only installed on half of the member (along Hbi and half of Bbi on two sides), with an additional SG
at mid-height on the opposite side to confirm the assumption of symmetry. The SG spacing is shown in Fig.
2.5.

Fig. 2.5. Spacing of strain gauges around the branch footprint adjacent to the welded connection
2.4.2. LOADING STRATEGY

The single point load was applied at a truss panel point by a 2700-kN capacity MTS Universal
Testing Frame, and resulted, by design, in a distribution of member forces which accentuated the load in a
particular, predetermined branch member (ergo the weld to it). Failure was planned to always occur in the test
welds, instead of by some connection, member, or stability failure mode.
After rupture occurred in the intended test weld, the connection was repaired by over-welding the gap
with a new weld designed to develop the yield strength of the member. Hence, the welded connection was no
longer critical. The location of the point load was subsequently altered to cause failure at another joint. Since
the MTS test frame was fixed to the laboratory floor, the truss itself was either translated, rotated 180° and/or
inverted to achieve the new distribution of forces. The roller supports were always located at the ends of the
truss, as seen previously in Fig. 2.1.
The laboratory testing arrangement and a series of typical weld fractures are shown in Figs. 2.6 and
2.7, respectively. Additional photographs of the truss supports and point-load device are provided in
Appendix A.4, and photographs of all connections and their failure modes are provided in Appendix A.5.

Fig. 2.6. Laboratory testing arrangement for full-scale RHS overlapped K-connection experiments

(a) Test K-60-0.71a (b) Test K-30-0.50a
(c) Test K-30-0.50b (d) Test K-90-0.71
Fig. 2.7. Typical weld fractures (and instrumentation)
2.5. RESULTS
All of the test welds failed in a brittle manner, by fracture along a plane through the weld, which occurred
almost simultaneously at all locations around the branch perimeter. Failure was sudden, and accompanied by
a dynamic “jolt” (caused by the release of strain energy) which, in some tests, displaced SGs from the branch
member surface (see Fig. 2.7a).
Fig. 2.8 shows the relationship between the applied MTS load and the load in the branch member
(measured by SGs in the position shown) for test K-30-0.50a. By virtue of a constant slope (indicating a
linear variation in average strain), it can be seen that the member itself remained elastic throughout the entire
load range. The branch load at rupture was hence calculated using Eq. 2.2.

Fig. 2.8. Typical MTS load versus branch load magnitude relationship (shown for test K-30-0.50a)
Figs. 2.9 - 2.11 show the variation in strain measured by 13 different SGs adjacent to the test welds at the
initial unloaded stage and at 50%, 80% and 100% of the weld rupture load for three typical test welds. The
variations in strain for the remaining test welds are provided in Appendix A.6.
For the joints tested, it was found that the magnitude of strain decreased as a function of the distance
from the toe of the connection – believed to be caused by differences in the relative stiffness of the chord ( =
0.50 and B/t = 27.5) and the overlapped branch ( = 1.00 and Bbj/tbj = 16.3), that results in the latter attracting
more load. As Ov increases, this change becomes less pronounced and is accompanied, generally, by a higher
average failure stress in the weld. The magnitude of strain along the branch transverse faces is seen to
decrease towards the mid-wall locations (SGs 1 and 13) except for the final stage of stress redistribution
(Figs. 2.9 and 2.11). This variation is expectedly more pronounced when the branch lands on a flexible chord
( = 0.50, B/t = 27.5), and for low values of Ov. The less sudden change in strain approaching the mid-wall
along the toe (in Figs 2.9 - 2.11) is due to the more uniform transverse stiffness of the overlapped branch ( =
1.00 and B/t = 16.3).
In Fig. 2.9, some of the weld to the heel actually remains in compression for the entire load range (branch
in tension). The strain at the toe in Fig. 2.9 may have been exacerbated by the proximity of the weld to the
hidden toe of the overlapped branch which was itself welded and thus increased the stiffness of the
connection and the weld effective length at this location.

Fig. 2.9. Typical distribution of normal strain around branch perimeter for specimens with Ov = 30%, =
0.50 and B/t = 27.5
0.50 and B/t = 27.5

0.50 and B/t = 27.5
The distributions of strain around the branch members adjacent to the test welds show that longitudinal
welds to overlapped K-connections can be regarded as completely effective at resisting the applied load when
Ov = 60% and 90%. For Ov = 30% (Fig. 2.9), the strain along the longitudinal weld (SGs 4 - 10) can be seen
to be more non-uniform. The transverse welds are always only partially effective and generally become less
effective as the -ratio decreases, as Ov decreases, and as B/t increases. These trends are verified by the actual
rupture loads (Pa), which are given in Table 2.4.
Table 2.4. Actual and predicted nominal weld strength for each test connection (actual-to-predicted ratio
shown in parentheses)
Pnw
kN
Test Pa AISC 360-10 Modified AISC 360-10 Without weld effective
kN lengths
K-90-0.50a 1232 832 (1.48) 876 (1.41) 965 (1.28)
K-90-0.50b 1277 872 (1.46) 916 (1.39) 1010 (1.26)
K-60-0.50 * 596 614 (0.97) 685 (0.87) 894 (0.67)
K-30-0.50a 765 383 (2.00) 454 (1.69) 850 (0.90)
K-30-0.50b 738 378 (1.95) 449 (1.64) 832 (0.89)
K-90-0.71 1139 885 (1.29) 930 (1.22) 1014 (1.12)
K-60-0.71a 974 649 (1.50) 712 (1.37) 894 (1.09)
K-60-0.71b 863 663 (1.30) 725 (1.19) 912 (0.95)
K-30-0.71 1054 463 (2.28) 529 (1.99) 947 (1.11)
T2 joint 4 † 1686 778 (2.17) 854 (1.97) 1143 (1.48)
T2 joint 6 † 1668 1165 (1.43) 1272 (1.31) 1681 (0.99)
Note: Underlined values are strength predictions which exceed the measured strength (i.e. nominally unsafe).
*
Imperfect weld root detail (see 2.2.2.2).
†
Tests by Frater (1991); failed by a combined mechanism of weld fracture and premature branch yielding.

2.6. EVALUATION OF AISC 360 (2010)

2.6.1. AISC 360 (2010) PROVISIONS FOR WELD EFFECTIVE LENGTHS IN RHS
According to the Load and Resistance Factor Design (LRFD) method of AISC 360 (2010), the available
strength of welds to axially-loaded RHS branches (Pnw) is based on the limit state of shear rupture along the
plane of the weld effective throat, according to Eq. (2.3):
Pnw  Fnwtwle (2.3)
where Fnw = nominal strength of weld metal; tw = weld effective throat around the perimeter of the branch;
and le = effective length of fillet and groove welds. An LRFD resistance factor, ϕ, equal to 0.75 and 0.80,
applies for fillet welds and PJP groove welds, respectively.
In Table J2.5 of AISC 360-10, Fnw is specified as 0.60FEXX for both fillet and PJP groove welds. In the
case of the former, it implies that the failure mode is by shear rupture on the effective throat; however, for
PJP groove welds (sides a’ and b’), it is an arbitrary reduction factor that has been in effect since the early
1960s to compensate for the notch effect of the unfused area of the joint and does not imply that the tensile
failure mode is by shear stress on the effective throat (per AISC 360-10 Specification Commentary to
Chapter J). Since a CP detail was provided (see Section 2.2.2.2) in order to establish a high degree of
certainty with respect to the fusion area (and the weld throat dimension) in this region, a more suitable term
of 1.00FEXX has been used herein for Fnw for groove welds.
The formulae for le are given in Table K4.1 of AISC 360 (2010) and are as follows:
• When 25% ≤ Ov < 50%:
 O  H  Ov  
1  v   bi
20v H bi
le,i       beoi  beov (2.4)
50  100   sin i
  100  sin i   j   
• When 50% ≤ Ov < 80%:
 O  H  O  H bi 
le,i  2 1  v   bi   v     beoi  beov (2.5)
 100   sin i  100  sin i   j   
 
• When 80% ≤ Ov ≤ 100%:

 O  H  Ov  H bi 
le,i  2 1  v   bi       Bbi  beov (2.6)


100   sin i  100  sin i   j   
where i = subscript used to refer to the overlapping branch; j = subscript used to refer to the overlapped
branch; Hb = overall height of the branch member measured in the plane of the connection; and θ = included
angle between the branch and the chord (= 60° for all test connections). The weld effective length dimensions
when 50% ≤ Ov are shown in Fig. 2.12.
Fig. 2.12. Weld effective length dimensions when 50% ≤ Ov
The terms beoi and beov are empirically derived from laboratory tests in the 1960s and 1970s (Davies &
Packer 1982) and quantify the effective widths of weld to the branch face, normal (transverse) to the plane of
the connection:
10  Fy t 
beoi    Bbi  Bbi (2.7)
B / t  Fybi tbi 
10  Fybj tbj 
beov    Bbi  Bbi (2.8)
Bbj / tbj  Fybi tbi 
where B = overall width of the RHS chord, normal to the plane of the connection; Bb = overall width of the
RHS branch, normal to the plane of the connection; t = wall thicknesses of the chord; tb = wall thicknesses of
the branch; Fy = yield stress of the chord; and Fyb = yield stress of the branch.
AISC 360 (2010) also limits the values of beoi/2 and beov/2 through a notwithstanding clause, which states,
“When Bbi/B > 0.85 or θi > 50°, beoi/2 shall not exceed 2t and when Bbi/Bbj > 0.85 or (180° – θi – θj) > 50°,
beov/2 shall not exceed 2tbj.” Thus, for the RHS overlapped K-connections tested, the upper limits of beoi = 4t
and beov = 4tbj apply.

2.6.2. SAFETY LEVEL INHERENT IN AISC 360 (2010)

In order to assess whether adequate or excessive safety margins are inherent, one can check to ensure that
a minimum safety index ( +) of 4.0 (as adopted by AISC 360-10 per Chapter B of the Specification
Commentary) is achieved using a reliability analysis in which ϕ is given by Eq. (2.9) (Fisher et al. 1978;
Ravindra & Galambos 1978):
    R exp   R  VR 

(2.9)
where αR = coefficient of separation taken as 0.55 (Ravindra & Galambos 1978); ρR = bias coefficient for
resistance; VR = associated coefficient of variation (COV) of ρR; and ϕ +
= adjustment factor for +
that is
needed when +
≠ 3.0 (Fisher et al. 1978). The bias coefficient for resistance (ρR) and its associate COV (VR)
are given by Eqs. (2.10) and (2.11):
R  M G P (2.10)
VR  VM2  VG2  VP2 (2.11)
where ρM = mean ratio of actual-to-nominal electrode strength; ρG = mean ratio of actual-to-nominal weld
throat area; ρP = mean test-to-predicted capacity ratio (with predicted capacity calculated using actual
measured properties); and VM, VG, and VP = COV of ρM, ρG, and ρP, respectively.
A formula to calculate ϕ + based on the reliability index ( +) was derived by Franchuk et al. (2002):
  0.0062(  )2  0.131   1.338


(2.12)
The mean actual-to-nominal electrode strength (ρM) and its associated COV (VM) were taken from a
database of 708 tests summarized in Table 2.5. The data from recent University of Toronto test programs in
Table 2.5 are average values from tensile coupon tests done by McFadden & Packer (2014), those presented
earlier in this Chapter (recall that the measured electrode strength was 26.3% stronger than the specified
nominal strength), and those presented later in this thesis, in Chapters 4 and 7. The composite mean and COV
of all test data were used for ρM and VM, respectively.
The mean measured-to-nominal weld throat area (ρG) and its associated COV (VG) were taken as 1.03 and
0.10, respectively, as justified by Callele et al. (2009). These factors account for the fact that larger weld
throats are typically provided by convexity of the weld face.

The mean test-to-predicted capacity ratio (ρP) was taken as the average over all tests of actual
(experimental) strength divided by the predicted strength, calculated using the measured values of tw, the
weld lengths, and FEXX.
Table 2.5. Mean actual-to-predicted electrode strength (ρM) and its associated variation (VM) amongst typical
weld metal
Study Number of tests ρM VM
Lesik & Kennedy (1990) 672 1.12 0.077
Callele et al. (2009) 32 1.15 0.080
Recent University of Toronto test programs (including
4 1.21 0.039
current work) a
Composite/total values: 708 1.12 0.121
a
nominal electrode strength is assumed to be 490 MPa for all electrodes tested.
In the evaluation that follows, correlation plots are produced using the measured ultimate weld strengths
(rupture loads) from the nine tests (including test K-60-0.50, with the imperfect root detail), and the results
from two similar connection tests that were conducted at the University of Toronto (the details of which
appear at the bottom of Table 2.1) (Frater 1991).
The implied resistance factor, ϕ, is equal to 0.883 for the AISC (2010) specification provisions and is
larger than the necessary resistance factors for fillet welds and PJP groove welds (0.75 and 0.80, respectively)
indicating an excessive level of safety for the AISC formulae. Fig. 2.13 shows the correlation of the predicted
nominal strengths with the experimental results.
2.7. RECOMMENDATION
2.7.1. BACKGROUND
By means of 12 full-scale experiments on isolated T-connections, conducted during phase one of the
research program, excessive safety was found to exist in the AISC 360 (2010) formula for the effective
elastic section modulus for in-plane bending for RHS moment T-connections (McFadden & Packer 2014).
The authors proposed a change to the current requirement that restricts the effective widths of welds to the
branch face from two times the chord wall thickness (2t) to a more reasonable limit of Bb/4. Their proposal
increases the effective length of the transverse weld elements in most RHS connections and was shown to
also be applicable to the formulae for the effective length of welds in axially-loaded RHS T- and X- (or
Cross) connections.
2.7.2. PROPOSAL
Since the same pattern is observed for RHS overlapped K-connections, it is proposed that the existing
formulae for the effective length of welds be modified in the same manner, by changing the requirement:

beov/2 shall not exceed 2tbj.”
to:
“When Bbi/B > 0.85 or θi > 50°, beoi/2 shall not exceed Bbi/4 and when Bbi/Bbj > 0.85 or (180° – θi – θj) >
50°, beov/2 shall not exceed Bbi/4.”
This change produces the correlation with the test data given by Fig. 2.14.
Fig. 2.13. Correlation with all test results for AISC Fig. 2.14. Correlation with all test results for modified
360-10 provisions AISC 360-10 provisions
Fig. 2.15. Correlation with all test results without using weld effective length rules

2.7.3. SAFETY LEVEL INHERENT IN RECOMMENDATION

The implied resistance factor, ϕ, is equal to 0.831 for the recommended modification to the AISC
(2010) specification provisions, which is still larger than the necessary resistance factors for fillet welds and
PJP groove welds. More importantly, using these modified AISC provisions for RHS overlapped K-
connections results in consistency for the aggregate recommended design rules for welds in RHS
connections, including axially-loaded RHS T- and X- (or Cross) connections, moment T-connections, and
overlapped K-connections.
2.7.4. COMMENTS
It is worth noting that if no effective length rules are applied, and the total weld length is used to
determine the strength of the welded joint to the overlapping branch, then the correlation with the test data
shown in Fig. 2.15 results. The implied resistance factor, ϕ, is equal to 0.564 which is less than the necessary
resistance factors for fillet and PJP welds, illustrating that such an approach provides an insufficient safety
margin. If historical tests (by Frater 1991) are omitted from the analysis, a marginal reduction to the inherent
safety factors is found; however, the previous discussion still applies, and the recommendation is found to be
safe. Correlations to this effect are given in Figs 2.16 – 2.18. All of the above still conservatively considers
test K-60-0.50 in the analysis.
Fig. 2.16. Correlation with current test results for Fig. 2.17. Correlation with current test results for
AISC 360-10 provisions modified AISC 360-10 provisions

Fig. 2.18. Correlation with current test results without using weld effective length rules
If the hidden toe of the overlapped branch was not welded, a smaller effective length would likely result
at the toe of the overlapping branch (beov) in connections with Ov = 30% since the restraint to transverse
deformation (stiffness) would be lost. This stands to reduce the mean test-to-predicted capacity ratio (ρP) in
Eq. (2.10), and hence reduce ρR in Eq. (2.9); however, by virtue of the test-to-predicted capacity ratio being
already higher for connections with Ov = 30% relative to the other connections, there would be some
counteracting decrease in VP, and hence VR, and thus a minimal effect on the reliability of the proposed
changes.

2.8. SUMMARY
Based on the results from nine full-scale tests on weld-critical RHS overlapped K-connections, and the
measured strength of two overlapped K-connection tests from a previous experimental program (Frater
1991), it has been found that:
1. The distribution of normal strain adjacent to the welded joint in RHS overlapped K-connections
is highly non-uniform.
2. As the overlap increases, stiffening the joint, the distribution of normal strain adjacent to the
welded joint becomes more uniform.
3. Transverse welds are only partially effective and generally become less effective as the -ratio
decreases, as Ov decreases, and as B/t (of the landing surface) increases.
The effective length rules defined by Equations K4-10 to K4-12 and given in Table K4.1 of AISC 360
(2010) for welds in RHS overlapped K-connections are quite conservative. It is recommended to modify the
requirement (AISC 360-10):
beov/2 shall not exceed 2tbj.”
to:
“When Bbi/B > 0.85 or θi > 50°, beoi/2 shall not exceed Bbi/4 and when Bbi/Bbj > 0.85 or (180° – θi – θj) >
50°, beov/2 shall not exceed Bbi/4.”
to increase the predicted strength of welded joints in RHS overlapped K-connections. This modification is
adopted from McFadden & Packer (2014) and has been shown to still be conservative yet generally provide a
more economical design approach for RHS T-, Y- and X- (or Cross-) connections subject to branch axial load
or branch bending. Using this recommendation would thus establish consistent rules across AISC 360 for the
design of welded truss connections between RHS.
2.8.1. FOOTNOTE
The above recommendation was duly implemented in Table K5.1 of AISC 360-16.

Chapter 3: ANALYSIS OF RHS TRUSSES WITH
3.1. INTRODUCTION
3.1.1. RECOMMENDATIONS FOR RHS TRUSS ANALYSIS
International (Packer et al. 2009) and Canadian (Packer & Henderson 1997) design guides recommend
that the static design of standard (e.g., Warren and Pratt) RHS trusses, having either gapped or overlapped
connections (Figs. 3.1a,b), be carried out on the basis of a force distribution obtained from an elastic analysis
of the truss assuming either: (a) all of the members are pin-connected, or (b) web members are pin-connected
to continuous chords, with the distance between the intersection of the web and chord centrelines (+e or –e in
Figs. 3.1a,b) modelled as an extremely stiff, or rigid, link.
Fig. 3.1. RHS (a) overlapped and (b) gapped K-connections, adapted from Packer & Henderson (1997)
Moments resulting from connection noding eccentricity (e) must be taken into account when designing
the chords (as beam-columns). Using method (a), these “eccentricity moments” can be easily overlooked,
since it is necessary to calculate and distribute the total panel point moment (i.e. the sum of horizontal web
member force components × e) to the chord (which is assumed to primarily resist these moments) in
28
Chapter 3: Analysis of RHS Trusses with Overlapped K-Connections 29
proportion to the chord stiffness on either side of the connection. This process is automated using method (b),
which was initially suggested in the 1980s with the popularity of personal computers (Packer & Davies 1989;
Wardenier & Packer 1992; Packer & Henderson 1992; Packer et al. 1992).
3.1.2. PREVIOUS RESEARCH

Under applied loads, significant chord face deformations are generally produced in RHS truss
connections. These contribute to the axial force and bending moment distributions, and the overall truss
deflections. Various methods, such as the “micro-bar” model (Czechowski et al. 1984) and finite element
(FE) methods (Coutie & Saidani 1989, 1991), have been used to show that by simulating chord face
deformations, the behaviour of experimental trusses can be replicated by theoretical models, which can
presumably be used for design of the truss. However, such models are impractical for design, since they
require experimental inputs (the micro-bar model) or an appreciable amount of modelling time (finite element
methods).
Philiastides (1988), whose test program involved RHS trusses with fully-overlapped web members (i.e.
Ov = 100%) (plus one gap-jointed truss) showed that there is little benefit in simulating chord face
deformations for an overlap-jointed truss. Practical models, like (a) and (b) above, produced similar results.
One exception was that, in his models with the webs rigidly connected to the chords, predictions of the web
member bending moments were not as good as the more rigorous methods of simulating chord face
deformations (i.e. micro-bar or finite element models). Additionally, neither (a) nor (b) were evaluated.
Frater & Packer (1992c) studied the behaviour of two large-scale gap-jointed RHS trusses and evaluated
both (a) and (b) on their ability to predict forces and deflections in the trusses, for the purpose of design. They
found that bending moments in the chords obtained using (b) showed poor numerical agreement with their
experimental results, but for the purpose of design, where it may only be necessary to predict the maximum
bending moment in the chord and to check its interaction with the maximum axial force, Frater & Packer
(1992c) concluded that (b) was generally successful. Frater & Packer (1992c) also found that a (concentric)
rigid-jointed model did not give good bending moment predictions for the web members. These moments
were found to be highly variable due to the secondary effects caused by chord face deformation.
The Canadian HSS Design Guide (Packer & Henderson 1997) suggests that secondary bending moments
in the webs (resulting from joint deformations) can be neglected in the design of the web members and
connections when the ratio of the web (or branch) member length between the node points (lb) to the member
height (Hbi, or Hbj) is six or more (i.e., lb/Hbi ≥ 6 and lb/Hbj ≥ 6), provided that connections are designed within
a certain range of validity; however, for cases where lb/Hbi < 6 or lb/Hbj < 6, these secondary moments need to
be considered.
Although it has been shown by Frater & Packer (1992c) that (b) produces an axial force and bending
moment distribution that is sufficient for the design of gap-jointed RHS trusses, there is a paucity of evidence

demonstrating it to be a suitable analysis method for overlap-jointed trusses. This Chapter evaluates several
truss analysis models that are available for obtaining axial force and bending moment distributions, including
(a) and (b), for the design of overlap-jointed RHS trusses.
A ten-metre span, simply-supported, overlap-jointed Warren truss (Fig. 3.2) (the same truss as in Chapter
2), with web members at 60° to the chord, was designed and fabricated using RHS with a minimum nominal
yield stress, Fy = 345 MPa. The truss chord members were comprised of two RHS sections, as noted in Table
3.1. Matched web members were overlapped at their connection to the chord, with 30% ≤ Ov ≤ 90% (see Fig.
3.1), and produced noding eccentricities (e) that ranged from 0 to -0.45 times the chord height (H). All of the
connections were designed within the parametric range of validity given by Packer & Henderson (1997).
The truss was tested under “service-load conditions” by applying a single quasi-static point load with a
universal test frame to one of seven top chord panel points (bold numbers 4 to 10 in Fig. 3.2). This was
facilitated by translating, rotating, or inverting the truss to invert the arrangement shown in Fig. 3.2. The
compression chord was braced against lateral instability by two frames that were post-tensioned to the
laboratory strong-floor, and supported out-of-plane by polytetrafluoroethylene (PTFE) plates. The PTFE
plates allowed vertical movement of the truss within the set-up, and two rollers provided the vertical support
to the truss at its ends. Centreline dimensions of the truss and e values at the connections are shown in Fig.
3.2. Note that the labelling convention in this figure bares no relation to that given in Fig. 2.1 for the weld
tests.
Fig. 3.2. Truss centreline dimensions and e values at connections
The actual (mechanical and geometrical) properties of the chord and web members were determined, for
use later in the analysis, as follows: the cross-sectional areas (A for the chord, and Ab for the webs) were

determined by cutting a prescribed length of the RHS, weighing it, and then using a density of 7850 kg/m3
(CISC 2016) to calculate the area; the Young’s modulus (E) was determined from tensile coupon tests in
accordance with ASTM A370 (2017); and the moment of inertia (I) was determined from formulae given in
Annex B of EN10219-2 (CEN 2006), using the measured widths, heights, thicknesses, and corner radii of the
RHS members. The results of the tensile coupon tests for the RHS truss members are provided in Appendix
A.3. The measured and calculated section properties for the members are shown in Table 3.1. Additional
measurements of the RHS, and the EN10219-2 (CEN 2006) section property formulae, are provided in
Appendices B.1 and B.2, respectively.
Table 3.1. Measured and calculated section properties for the RHS truss members
A or Ab E I
RHS Designation mm2 × 103 MPa × 106 mm4
HSS 127 × 127 × 8.0 (webs 1 – 10) 3626 193.0 8.25
HSS 178 × 178 × 13 (top chord 11 – 15; posts 22 and 23) 7776 193.0 33.9
HSS 254 × 254 × 9.5 (bottom chord 16 – 21) 8804 191.7 85.6
Instrumentation consisting of electrical resistance strain gauges (SGs) and linearly varying differential
transformers (LVDTs) was used to record the member axial forces and bending moments, and truss
deflection patterns. A total of 106 SGs (shown by symbols in Fig. 3.3) were installed on the top and bottom
of each member, centred in the plane of the truss, and oriented along the longitudinal axis of the members to
measure the in-plane strain profiles. Gauges were located (in pairs) at mid-length of each member and at a
distance from the connections (as shown in Fig. 3.3). LVDTs were placed along the underside of the bottom
chord at every panel point, and inside of the chord at corner joints (1 and 13, or 2 and 12, in Fig. 3.2). This
entailed a total of seven LVDTs when the truss was upright, and a total of six LVDTs when the truss was
inverted (see Fig. 3.3). Photographs of the LVDT orientations are provided in Appendix B.3. Two load cells
situated below the rollers were used to measure the reaction forces during the tests. These load cells are
apparent in the photographs of Appendix A.4. An overall view of the arrangement is shown in Fig. 3.4.
Fig. 3.3. SG and LVDT locations

Fig. 3.4. Overall view of the testing arrangement
3.3. RESULTS
The SG data and the actual member properties (Table 3.1) were used to determine the experimental axial
force (Pe) and bending moment (Me) at each pair of SGs in the truss, from strain distributions such as the one
shown in Fig. 3.5. According to the SG output, the truss remained linear elastic. Hence, Pe and Me for strain
distributions such as the one shown in Fig. 3.5 were calculated from Eqs. (3.1) and (3.2).
Fig. 3.5. Potential measured in-plane strain distribution for truss members
  top   bottom 
Pe  E  A    (3.1)
 2 

  top -  bottom 
Me  E  I    (3.2)
 H 
For a test designated as 4.1, the point load was applied to panel point 4 (Fig. 3.2), and the experimental
axial force and bending moment distributions at the peak load are shown in Fig. 3.6. A similar test, 4.2, gave
comparable Pe and Me distributions (see Appendices B.5 and B.6). [For test numbers, the first digit refers to
the loaded panel point, and the second digit refers to the first or second time that the panel point was loaded
in a series of tests (see Table 3.3 in Section 3.4.1.3)]. It may be noted that the large variations in axial force
along a member length are due to experimental error, and are not atypical of measurements obtained using
strain gauges (Frater & Packer 1992c). The axial force used for each segment is noted in Appendix B.5.

Note: bending moment diagrams are plotted on the side of the member subject to tensile bending stress
Fig. 3.6. Experimental (a) axial force and (b) bending moment distributions for test 4.1
In eight total tests, the panel point deflections and support settlements were determined from the LVDT
data. The maximum truss defection (Δmax) for each test is given in Table 3.3, and has been calculated by
taking into account the support settlements. Measured deflections at each panel point, and for each test, are
tabulated in Appendix B.4.

3.4. EVALUATION OF FRAME ANALYSIS MODELS

The experimental results have been compared with the theoretical behaviour/predictions given by four
practical frame analysis models using SAP2000 (CSI 2014), consisting of:
(a) a concentric pin-jointed model (P) (Fig. 3.7a);

(b) a combined pin and rigid connection model (PR), where web members are pin-connected to
continuous chord members (Fig. 3.7b);
(c) a concentric rigid-jointed model (R) (Fig. 3.7c); and
(d) an eccentric rigid-jointed model (RR) (Fig. 3.7d).
Methods (a) and (b) above are the same ones recommended by Packer & Henderson (1997), and have
been described in Section 3.1.1. Method (c) entails using fixed-end web members that are capable of
transmitting axial, in-plane bending and shear forces, thereby making it possible to plot a bending moment
distribution for both the web and chord members. In method (c), joints are modelled concentrically regardless
of the e value at the joint. Method (d) uses fixed-end web members that are connected to continuous chords,
with the distance between the intersection of the web and chord centrelines (+e or –e) modelled as a rigid
two-joint link.
Fig. 3.7. Assumptions in models (a) P, (b) PR, (c) R, and (d) RR with respect to noding at interior joints,
and at corner joints

The true effective depth (distance between chord centrelines) has been modelled for methods (a) and (c),
meaning that the true web-to-chord angle (θi = θj = 60°) has not been maintained. This assumption, for
“forced” concentric analyses, is claimed to give better results than the alternative (i.e. modelling the true web-
to-chord angles) (Philiastides 1988). Using methods (b) and (d), both the true effective depth and the true
web-to-chord angles have been maintained, since system lines for the models coincide with the member
centrelines. Assumptions with respect to noding at the corner joints are shown in Figs. 3.7e,f. One should
note that for concentrically-noded trusses (i.e. e = 0 at all connections), models R and RR are identical.
FE modelling of the truss (e.g. Coutie & Saidani 1989, 1991) (to explicitly take into account the semi-
rigid connection behaviour) was also considered; however, this was deemed an unlikely approach to the
design of planar HSS trusses. Such an approach is common as a research exercise only, or for the design of
high-value tubular frames (e.g. large-span 3D roof structures, offshore structures).
3.4.1. COMPARISON OF EXPERIMENTAL AND THEORETICAL TRUSS BEHAVIOUR
3.4.1.1. Axial Force Distributions

By comparing the mean percentage error for axial force predictions in each member (Table 3.2), it is
shown that models RR, PR, and R conservatively (over-)predict most member forces, and give similar axial
force distributions overall. Forces that were under-predicted by the models were always in web members, and
the maximum percentage errors, in absolute terms, were 4%, 1%, and 9% in test 4.1 (for models RR, PR and
R, respectively) and 7% (for model R) in test 4.2. It is found that model R most significantly under-predicts
member forces; thus, it is not recommended to use an axial force distribution obtained from this model for
design.
The larger mean percentage errors in test 4.2 are explained by weld fractures that, by design, occurred in
nine of the connections between tests 4.1 and 4.2 (i.e. the topic of Chapter 2). Fractured joints were ground,
re-welded, and sometimes plated between tests, thereby modifying the geometry and stiffness of the
connections, and hence the experimental axial force distribution. The sample standard deviation of the
percentage error for all predictions (19 total for each test/model), was less than 4% for both models RR and
PR, and greater than 6% for models R and P. Thus, the former two models can be deemed more precise with
respect to axial force predictions. The measured axial loads for each member are compared to the theoretical
(predicted) axial forces in Appendix B.5.

Table 3.2. Mean percentage errors for theoretical (predicted) axial forces
Test no. Model RR Model PR Model R Model P
4.1 Top chord (11 – 15) +5% +4% +5% +8%
Bottom chord (17 – 20) +7% +10% +8% +15%
Webs (1 – 10) +6% +6% +6% +7%
Mean percentage error a +6% +8% +7% +11%

No. of members under-predicted 1 1 2 0
4.2 Top chord (11 – 15) +8% +8% +9% +11%
Bottom chord (17 – 20) +9% +12% +11% +17%
Webs (1 – 10) +10% +10% +9% +11%
Mean percentage error +9% +11% +10% +14%

No. of members under-predicted 0 0 1 0
a
percentage error = [1 – (experimental/theoretical)] × 100% : positive indicates a conservative (over-)prediction
3.4.1.2. Bending Moment Distributions

Experimental and theoretical bending moment distributions (for models RR, PR and R) are compared in
Fig. 3.8 for test 4.1. With the exception of top chord segments 11 and 15, and bottom chord segments 16 and
21 (Fig. 3.2), models RR and PR give similar bending moment distributions in the chords. These are, in turn,
similar to the experimental results. Model R underestimates bending moments in the top chord away from the
applied load, and gives the incorrect sense of bending in the bottom chord (segment 17, where there is large e
at the adjacent connections). This is not surprising, since model R is capable of defining only the continuity
component of the bending moment distribution. Conversely, model P is only capable of defining the
eccentricity component of the bending moment distribution if it is calculated in accordance with
recommendations for method (a) by Packer & Henderson (1997), discussed in Section 3.1.1.
By the slopes of the experimental versus theoretical bending moment diagrams (Fig. 3.8), it is shown that
models PR and RR give reasonable shear forces in the chords; however, in this regard, model RR may be
deemed better, since model PR does not correctly give double curvature in chord segments 11, 12, 13 and 15
(top chord), and 19 (bottom chord). For segments 11 and 15, this is an effect caused by corner joint rigidity,
and since, by definition, model PR cannot employ web member end fixity, it cannot replicate the
experimental bending moments at these locations [which were also measured by Philiastides (1988), and
Frater & Packer (1992c)].
Bending moments in the webs are produced by models R and RR, and both distributions show the correct
sense of bending in all but two members (i.e. members 1 and 4). Both methods underestimate the magnitude
of the experimental results – a trend which has been reported by Coutie & Saidani (1989, 1991), Philiastides
(1988), and Frater & Packer (1992c). The maximum experimental web member bending moment in tests 4.1
and 4.2 (which occurred in member 3) was still low – less than 7% of the nominal yield moment of the
member (assuming a 345 MPa yield strength) – but this being (a) under service-load conditions, and (b)
2.5Bbi and 2.5Bbj away from the chord face (see Fig. 3.3). At ultimate loads, web member bending moments

will be larger, and may not redistribute as easily as in a gap-jointed truss since connections are more rigid.
Designers should therefore be mindful of cases where web moments could become significant. The
experimental bending moments for test 4.2, and a comparison to theoretical bending moments, are provided
in Appendix B.6.
Fig. 3.8. Comparison of experimental and theoretical bending moment distributions for test 4.1

3.4.1.3. Maximum Truss Deflections

Experimental and theoretical maximum truss deflections are compared in Table 3.3, and it is shown that
model RR (mean error = +3%) gives the most accurate prediction of the maximum truss deflection. Model
PR (mean error = +5%) is comparable. The similarity of these two models in predicting the maximum truss
deflection is unsurprising, since they also give similar chord bending moments, and axial forces overall. By
virtual work principles, the difference in given deflections is therefore made up in the bending of the webs,
which under experimental loading conditions were shown to be small. On average, the concentric models (P
and R) over-predict the maximum truss deflection by 13% and 9%, respectively. Nonetheless, all four models
(a - d) may be used to obtain a conservative (over-)prediction of the maximum truss deflection for design. A
summary of the theoretical deflections for each panel point compared to the actual deflections for all tests
done is provided in Appendix B.4.
Table 3.3. Comparison of experimental and theoretical (predicted) maximum truss deflections
Applied load Δmax Model RR Model PR Model R Model P
Test no. kN mm mm mm mm mm
Upright 4.2 490 4.50 4.42 (-2% a) 4.57 (+2%) 4.72 (+5%) 5.02 (+10%)
6.1 497 6.49 6.93 (+6%) 7.05 (+8%) 7.37 (+12%) 7.69 (+16%)
6.2 497 6.46 7.00 (+6%) 7.12 (+7%) 7.41 (+11%) 7.73 (+15%)
8 497 6.61 6.93 (+7%) 7.05 (+8%) 7.37 (+12%) 7.69 (+16%)
10 490 4.32 4.36 (+3%) 4.60 (+6%) 4.70 (+8%) 5.00 (+14%)
Inverted 7.1 483 6.21 6.57 (+5%) 6.67 (+7%) 6.95 (+11%) 7.22 (+14%)
7.2 480 6.35 6.54 (+3%) 6.63 (+4%) 6.91 (+8%) 7.19 (+12%)
9 496 6.00 5.84 (-3%) 5.95 (-1%) 6.18 (+3%) 4.44 (+3%)
Mean percentage error: (+3%) (+5%) (+9%) (+13%)

a
percentage error = [1 – (experimental/theoretical)] × 100% : positive indicates a conservative (over-)prediction

3.5. SUMMARY
The experimental behaviour of a large-scale overlap-jointed RHS truss has been compared with the
theoretical behaviour given by practical frame analysis models, consisting of:
(a) all joints pinned and concentric (P);

(b) all joints rigid and concentric (R);
(c) pin-ended web members connected eccentrically to continuous chords (PR); and
(d) rigid-ended web members connected eccentrically to continuous chords (RR).
The following conclusions have been reached for the analysis of RHS trusses with overlapped K-connections:
1. Model P provides a conservative (over-)prediction of the axial force distributions and truss
deflections; however, since it neglects continuity bending moments it will be a poor indicator of
the actual truss behaviour.
2. Model R is not suited for analysis since it has been shown to unsafely (under-)predict member
axial forces and, since it neglects eccentricity bending moments, it will be a poor indicator of
both the magnitude and sense of bending in chords.
3. Model PR gives generally good predictions for axial forces in chords and webs, bending
moments in the chords, and for the maximum truss deflection.
4. Model RR may be used for analysis since it provides even better predictions for the axial forces,
bending moments, and maximum truss deflection than model PR, but it does produce a bending
moment distribution for the web members which may cause difficulties using conventional truss
design recommendations, as these are based on axially-loaded web members.
The above conclusions are based on limited experimentation and pertain to RHS trusses of similar
geometry and, in particular, low span-to-depth ratios. It is hence recommended that the analysis of RHS
trusses with overlapped connections be done on the basis of a model with rigid-ended web members
connected eccentrically to continuous chords (Model RR). This is a logical extension of the recommendation
by Frater & Packer (1992c) to use Model PR for RHS trusses with gapped K-connections, wherein
connections are more flexible (due to chord face deformation) resulting in larger overall truss deformations
and different magnitudes of axial forces/bending moments in members.

Chapter 4: FILLET WELDS IN HSS-TO-RIGID PLATE
CONNECTIONS: EXPERIMENTATION
4.1. INTRODUCTION
4.1.1. WELD DESIGN PHILOSOPHIES FOR HSS CONNECTIONS
For steel hollow structural section (HSS) connections, current codes, standards, specifications and design
guides (Wardenier et al. 2008; Packer et al. 2009; Packer et al. 2010; ISO 2013) acknowledge two design
philosophies for proportioning welds:
(i) The weld can be proportioned so that it develops the yield strength of the connected branch
member. By setting the design strength of a fillet-welded joint equal to that of the yield strength
of the connected branch member, the required effective weld throat dimension (tw) can be
calculated in terms of the connected branch wall thickness (tb). This will produce an upper limit
on the weld size and hence be a conservative design procedure. Assuming an axially-loaded 90˚
T-connection between rectangular hollow sections (RHS) made to ASTM A500 Grade C (ASTM
2013) with matching electrodes, the results of Method (i) for the design of fillet welds in various
steel specifications/codes (CSA 2001; CEN 2005; AISC 2016; AWS 2015; CSA 2014) are listed
in Table 4.1 (updated from McFadden et al. 2013). Clearly, there is quite a disparity in fillet weld
design criteria in these steel specifications/codes.
(ii) As an alternative to Method (i), the weld can be proportioned so that it resists the actual forces in
the connected branch member. This approach involves a consideration of the effective length of
the weld group, since extensive research (Frater & Packer 1992a, 1992b; Packer & Cassidy 1995;
Packer & Sun 2011) has proven that the connected branch wall and the adjacent weld are
generally loaded in a highly non-uniform manner around the branch, because the base to which
the weld is attached is flexible in a typical HSS-to-HSS connection.
Method (i) is appropriate if there is low confidence in the design forces in the branch, or if there is
uncertainty regarding Method (ii), or if plastic stress redistribution is required in the connection. Method (i)
enables a prequalified weld size to be readily specified. However, Method (ii) generally allows for weld
“downsizing” and hence is popular. It is particularly appropriate if the branch forces are low relative to the
branch member capacity. AISC 360 (AISC 2016) has adopted Method (ii) in Chapter K for welded
41
Chapter 4: Fillet Welds in HSS-to-Rigid Plate Connections: Experimentation 42
connections to RHS by specifying various weld effective lengths for different connection types and loading
situations.
Table 4.1. Comparison of fillet weld effective throats required to develop the yield resistance of a connected
RHS branch member wall (updated from McFadden et al. 2013)
Specification or code tw
AISC 360-16 Table J2.5 (AISC 2016) 1.43 tb
AWS D1.1-15 Clause 9.6.1.3 and Fig. 9.10 (AWS 2015) 1.07 tb
CAN/CSA S16-01 Clause 13.13.2.2 (CSA 2001) 1.14 tb
CSA S16-14 Clause 13.13.2.2 (CSA 2014) 0.95 tb
EN1993-1-8: Directional method (CEN 2005) 1.28 tb
EN1993-1-8: Simplified method (CEN 2005) 1.57 tb
Note: This table assumes an axially-loaded 90˚ T-connection between RHS made to ASTM A500 Grade C with
matching electrodes.
The discrepancy between the tw values in Table 4.1 is due to simplified models for fillet weld design, as
well as use (or non-use) of the the (1.00+0.50sin1.5θ) directional strength-increase factor, discussed herein.
4.1.2. EFFECT OF LOADING ANGLE ON FILLET WELD BEHAVIOUR

Starting from the 1930s, experimental and theoretical investigations have been conducted on the
behaviour of fillet welds as a function of direction of loading with respect to the weld axis, mostly on fillet
welds in lap splice connections. The investigations that formed the basis of the modern fillet weld design
equations in North American and European specifications are discussed in this section.
4.1.2.1. Development of North American Fillet Weld Design Criteria

It is well-known that as the angle of loading increases (from θ = 0˚ for a longitudinally-loaded weld to θ
= 90˚ for a transversely-loaded weld), the strength of a fillet weld increases but its ductility decreases. Hence,
within a fillet weld group, the longitudinal weld tends to have the lower bound of strength but the upper
bound of ductility. Both American and Canadian steel design specifications, AISC 360-16 and CSA S16-14,
recognize the influence of the loading angle on the fillet weld strength and ductility.
Early tests performed by Butler & Kulak (1971) indicate that strength ratios of fillet welds with loading
angles of 30˚, 60˚, and 90˚ to longitudinal fillet welds (θ = 0˚) are 1.34, 1.41 and 1.44, respectively. Based on
experimental results, the analytical model developed by Kato & Morita (1974) predicts that a transverse fillet
weld is 46% stronger than a longitudinal fillet weld of the same size and length, corresponding to a
directional strength-increase factor of 1.46. The fillet weld design equations in current American and
Canadian steel design specifications originate from the research by Miazga & Kennedy (1989), where tests
were performed on 42 fillet-welded lap splice connection specimens with 5 or 9 mm fillet welds, with the
connection loaded in tension at angles to the weld axis from 0˚ to 90˚ in 15˚ increments. The strength of the

fillet weld gradually increased to 1.50 times as the loading angle increased from 0˚ to 90˚. Based on the
experimental results, Miazga & Kennedy (1989) proposed a method to predict the strength of fillet welds of
different orientations based on a maximum shear stress failure criterion. Later, Lesik & Kennedy (1990)
extended the work of Miazga & Kennedy (1989) and proposed a simplified equation which is a function only
of the loading angle [i.e. the (1.00+0.50sin1.5θ) directional strength-increase factor adopted in current
American (AISC 2016) and Canadian (CSA 2014) specifications]. It takes the form of a multiplier that is
applied to the longitudinal fillet weld strength.
The test program of Miazga & Kennedy (1989) included connection specimens lap-spliced by fillet welds
(Fig. 4.1a) using the shielded metal arc welding (SMAW) process, which is not commonly used in industry
for high-production welding. To re-evaluate the effectiveness of the (1.00+0.50sin1.5θ) factor on the more
prevalent flux-cored arc welding (FCAW) process, a series of investigations was conducted by Ng et al.
(2004a, 2004b) and Deng et al. (2006). Their reliability analyses showed that the design equations in the
American and Canadian standards provide an adequate level of safety for both welding processes.
Fig. 4.1. Fillet-welded (a) lap-splice, (b) cruciform, and (c) HSS-to-rigid plate test specimens
The tests performed by Ng et al. (2004a, 2004b) and Deng et al. (2006) consisted of concentrically loaded
fillet-welded connections with all welds having the same loading orientation. However, fillet-welded
connections commonly include welds at different orientations to the applied load, and the interaction between
fillet welds of different loading angles remained unknown. Hence, Callele et al. (2009) tested 19 lap splice
connections with multiple weld segments of different orientations. It was still found that the deformation

capacity of the fillet welds decreased as the angle of loading increased (i.e. maximum deformation capacity
for a weld element loaded longitudinally; minimum deformation capacity for a weld element loaded
transversely). As a result of such incompatibility, the transverse weld prevents the longitudinal weld from
reaching its full capacity before failure of the joint takes place. Hence, the tested weld groups possessed
capacities significantly lower than the sum of the individual weld segment strengths. Therefore, Callele et al.
(2009) proposed a simple method to account for this phenomenon conservatively by reducing the capacities
of the more ductile welds by 0 to 15%. For example, for a weld group containing longitudinal and transverse
welds, the longitudinal weld can only develop 85% of its full capacity before joint failure. This method has
been adopted by current American and Canadian steel design specifications. In order to investigate the
response of eccentrically loaded fillet welds, where the load is not in the plane of the weld group, Kanvinde et
al. (2009) performed 60 bend tests on cruciform connection specimens (Fig. 4.1b). It was found that a bearing
mechanism between the connected plates, which is not present for concentrically loaded joints, made an
obvious contribution to the strength of fillet-welded joints under out-of-plane eccentric loading. Hence, the
authors proposed a design approach which incorporated this beneficial effect.
Another important observation, based on the experimental evidence in the above research, is that the
actual weld fracture plane does not always coincide with the theoretical throat. Since the theoretical effective
throat thickness of a fillet weld is commonly defined, in various design specifications, as the height of the
largest triangle that can be inscribed within the fusion faces and the weld surface (i.e. the shortest distance
from the root to the face of the weld), the use of the theoretical effective throat thickness generally produces a
conservative strength prediction.
However, the application of this (1.00+0.50sin1.5θ) factor, also known as the “sinθ” factor, in the design
of fillet welds in HSS connections (Fig 4.1c) has been questioned since:
1. Unlike lap splice connections, fillet welds in many HSS connections have the welded attachment
loaded in tension or bending, rather than in shear.
2. Since welding can only be performed on one side of the hollow section wall, fillet welds to HSS
members will be subject to a local eccentricity. For example, tension loading in an attached wall
will produce additional tensile stress at the root of the weld (Fig. 4.1c). In fact, relevant codes and
standards recognize that eccentric loading on a fillet weld, causing tension at the weld root, may
reduce weld capacity. For example, CSA W59 (CSA 2013b) Clause 4.1.3.3.2 even states that …
“Single fillet and single partial joint penetration groove welds shall not be subjected to bending
about the longitudinal axis of the weld if it produces tension at the root of the weld”. EN 1993-1-
8 (CEN 2005) Clause 4.12 states that such local eccentricity, producing tension at the root of the
weld, should be taken into account, but it specifically notes that … “Local eccentricity need not
be taken into account if a weld is used as part of a weld group around the perimeter of a structural
hollow section”. The basis for this Eurocode waiver is unknown. AWS D1.1 (2015) Clause 2.6.2

states that, in the design of welded joints, the calculated stresses shall include those due to
eccentricity caused by alignment of the connected parts, size and type of welds, but this Section
pertains to connections which are “non-tubular”.
3. By the same token as 2., the HSS wall is not externally restrained from rotation.
4. It has been shown experimentally that the inclusion of the sinθ factor in the strength calculation
is generally unsafe, when applied to RHS-to-RHS fillet-welded connections and used in
conjunction with AISC 360-10 (and AISC 360-16) Chapter K weld effective lengths/properties,
because target structural reliability levels are not achieved (Packer & Sun 2011; McFadden &
Packer 2013; McFadden et al. 2013; McFadden & Packer 2014; Tousignant & Packer 2015a).
Thus, AISC does not permit the use of the sinθ factor when the “effective length method” is used
for proportioning fillet welds in hollow section connections (AISC 360-16 Commentary on K5).
The purpose of this Chapter is to determine if the sinθ factor is even applicable to fillet welds between an
HSS branch and a rigid base, where all of the weld length is effective (i.e. the AISC effective length method
is not applicable). Hence, in this investigation all connection specimens were made by welding HSS to rigid
steel plates to remove the effect of any flexibility of the surface on which the fillet weld lands.
4.1.2.2. Development of European Fillet Weld Design Criteria

It can be concluded, based on the prior literature review, that although a fillet weld is simple in concept,
the internal stress systems by which it transmits load are highly complex. The stresses over sections of the
fillet weld can be highly irregular due to stress-raising effects, depending on a number of factors such as
geometry of the weld, lack of or excessive penetration, geometry of the connection and residual stress.
However, for design the strength of a fillet weld is often described by simplifying the force system, assuming
a critical failure surface and distributing a mean stress over it. Same as the North American design criteria,
Eurocode 3 (CEN 2005) considers the effective throat as the critical failure surface over which the stress due
to the applied load is uniformly distributed. Different from the North American approach, Eurocode 3
requires the forces transferred by the fillet weld to be resolved into stress components in different directions
(𝜎⊥ , 𝜏⊥ and 𝜏∥ ) over the effective throat area (see Section 4.2.4).
The European fillet weld design criteria originate from the research conducted by Jensen (1934) and Kist
(1936) on fillet welds under consideration of constant deformation energy. Later, Vreedenburgh (1954)
extended the tests carried out by Jensen (1934) and Kist (1936), from which the early European fillet weld
design equation was developed. Later, IIW (1980) reported that the strength ratio of transverse to longitudinal
fillet welds was √ /√ = . . This ratio was recommended as a safe design value for the strength of
transverse welds, although much higher ratios had been observed in the North American investigations. IIW
suggested that such a difference was primarily due to friction and supporting effects between plates in the
tested lap splice connections. The ratio of 1.22 is implied in the modern fillet weld design equation in

EN1993-1-8 (CEN 2005), which was developed based on a von Mises hypothesis and verified experimentally
by assessing the strength of fillet welds loaded at different angles. Tests in the above research showed that the
strength of fillet welds under combined stresses, due to load applied at different angles, can be roughly
represented by an ellipsoid in the 𝜎⊥ , 𝜏⊥ , 𝜏∥ space. Recent European research by Björk et al. (2012, 2014) has
extended the fillet weld design rules to connections made of high and ultra-high-strength steel.
4.2. DESIGN SPECIFICATIONS

4.2.1. ANSI/AISC 360 (2016)
In Section J of AISC 360, unless overmatched weld metal is used, the design strength (Vrw = Pnw) of a
single fillet weld is based on the assumed single limit state of shear rupture along the plane of the weld
effective throat. It is taken as the product of the nominal stress of the weld metal (Fnw) and the weld effective
throat area (Aw) with a resistance factor ( = 0.75) applied. The nominal strength (Pnw) is determined as
follows:
Pnw  Fnw Aw (4.1a)
Fnw  0.60FEXX (4.1b)
where FEXX = ultimate strength of weld metal (= Xu in CSA S16).

Alternatively, for a linear weld group with a uniform leg size, loaded through the centre of gravity (i.e. all
weld elements are in a line or are parallel hence having the same deformation capacity under the applied
load), the provisions of Section J2.4(b) permit the use of the sinθ factor [Eq. (4.1c)] for the calculation of the
nominal stress of the weld metal (Fnw).
Fnw  0.60 FEXX (1.0  0.50sin1.5  ) (4.1c)
where θ = angle of loading measured from the weld longitudinal axis (in degrees).
As a special application of Section J2.4(b), Section J2.4(b)(2) gives provisions for concentrically loaded
connections with weld elements of multiple orientations (longitudinal and transverse to the direction of
applied load). It specifies that the nominal strength (Pnw) of concentrically loaded joints with both
longitudinal and transverse fillet welds be determined as the greater of Eqs. (4.2a) and (4.2b). This provision
is to account for the deformation incompatibility between longitudinal and transverse fillet welds.
Pnw  Pnwl  Pnwt (4.2a)

Pnw  0.85Pnwl  1.5Pnwt (4.2b)
where Pnwl = total nominal strength of longitudinally loaded fillet welds; and Pnwt = total nominal strength of
transversely loaded fillet welds with Fnw calculated using Eq. (4.1b).
4.2.2. CAN/CSA S16 (2001)
In the 2001 Canadian steel structures design standard, the design strength (Vrw = Pnw = 0.67Pnw) of a
fillet weld was taken as the lesser of two limit states: (i) shear rupture along the fusion face with the base
metal using Eq. (4.3a), and (ii) shear rupture along the plane of the weld effective throat using Eq. (4.3b),
which permits the use of a sinθ factor.
Pnw  0.67 Am Fu (4.3a)
Pnw  0.67 Aw FEXX (1.0  0.50sin1.5  ) (4.3b)
where Am = area of fusion face between weld and base metal; and Fu = ultimate tensile strength of base metal.
It should be noted that the design resistance (Vrw) calculated using Eqs. (4.1a) and (4.1c) per AISC 360-16 is
the same as that calculated using Eq. (4.3b) per CSA S16-01, because the terms ( = 0.75)(0.60FEXX) and (
= 0.67)(0.67FEXX) both equal 0.45FEXX.
4.2.3. CAN/CSA S16 (2014)

Similar to AISC 360 (2016), unless overmatched weld metal is used, the 2014 Canadian steel structures
design standard specifies that the design strength (Vrw = Pnw = 0.67Pnw) of a single fillet weld be evaluated
only based on the limit state of shear rupture along the plane of the weld effective throat. The nominal
strength of a fillet-welded joint is the sum of the nominal strengths of weld elements with different
orientations. The base metal strength check in CSA S16-01 was removed in CSA S16-09 and CSA S16-14
since, according to the research on fillet-welded lap splice connections by Ng et al. (2004a, 2004b), Deng et
al. (2006), and Callele et al. (2009), the base metal strength check might prevent the designer from taking
advantage of the full capacity of the weld. Another difference between CSA S16-01 (CSA 2001) and CSA
S16-14 (CSA 2014) is that the latter considers the deformation incompatibility between welds with different
orientations by introducing an “Mw factor”.
Pnw  0.67 Aw FEXX (1.0  0.50sin1.5  ) M w (4.4a)

0.85  1 / 600
Mw  (4.4b)
0.85   2 / 600
where θ and θ1 = angle of loading (in degrees) of the weld element under consideration; θ2 = angle of loading
(in degrees) of the weld element in the joint that is nearest to 90˚; Mw = strength reduction factor to allow for
the variation in deformation capacity of weld elements with different orientations, similar to the 0.85 factor in
Eq. (4.2b).
4.2.4. EN1993-1-8 (2005)

Eurocode 3 (CEN 2005) specifies that the design resistance of a fillet weld be determined using either the
Directional Method or the Simplified Method. For both methods, the assessment of the selected weld size is
based on the ultimate strength of the base material (Fu), which can be correlated to the ultimate strength of the
matching weld material using a “ w correlation factor”. Hence, it is generally safe if overmatched electrodes
are used.
The Directional Method requires resolution of the resultant design force transmitted by a unit length of
weld into components parallel and transverse to the longitudinal axis of the weld and normal and transverse
to the plane of its throat. Assuming a design throat area of Aw, the product of the effective throat thickness (tw)
and the unit weld length, the component forces can be used to calculate the component stresses (Fig. 4.2a) in
the same directions. The normal stress parallel to the weld axis (σ∥ ) is not considered when verifying the
design resistance of the weld. The design resistance of the fillet weld is deemed sufficient if Eqs. (4.5a) and
(4.5b) are satisfied along the entire length. Weld connecting elements with different material properties
should be designed using the properties of the lower strength material.
0.5
 2  3( 2   2 )   Fu / (  w M 2 ) (4.5a)
and    0.9Fu /  M 2 (4.5b)
where 𝜎⊥ = normal stress perpendicular to the throat; 𝜏⊥ = shear stress (in the plane of the throat)
perpendicular to the axis of the weld; 𝜏∥ = shear stress (in the plane of the throat) parallel to the axis of the
weld; 𝑀2 = partial safety factor for the resistance of weld equal to 1.25; and 𝑤 = correlation factor for fillet
welds.

Fig. 4.2. Stress components in the plane of the weld effective throat
Eq. (4.5a) can be simplified for 90° equal-legged welds to:
• For longitudinally-loaded welds (θ = 0˚)
Fu
Vrw  ( )twlw (4.6a)
3 w M 2
• For transversely-loaded welds (θ = 90˚)
Fu
Vrw  ( )twlw (4.6b)
2 w M 2
where Vrw = design resistance of the fillet weld. Intermediate steps in the simplification are shown in
Appendix C.1.
Thus, Eurocode 3 (CEN 2005) uses a relationship between the strength of a transverse weld to a
longitudinal weld of (√ /√ ) = . , which is significantly less than the 1.50 factor used in current North
American specifications [Eq. (4.1c) or Eq. (4.4a)].
The Simplified Method is an alternative to the Directional Method for fillet weld design. This method is
independent of the orientation of the weld throat plane with respect to the applied force. In fact, it is a

conservative alternative to Eq. (4.5a). The Simplified Method assumes that all welds are loaded in pure shear
parallel to the axis of the weld (Fig. 4.2b) and the welds can then be proportioned using Eq. (4.6a).
Since a primary objective of this study is to determine if the sinθ factor is applicable when the entire
length of a fillet weld in an HSS connection is effective (i.e. the effect of any flexibility of the surface on
which the fillet weld lands is removed), all connection specimens were made by welding either circular
hollow sections (CHS) or RHS to a rigid steel plate. For this work, 12 CHS-to-rigid plate connections with
different weld sizes, and angles between the HSS and plate of 60˚ and 90˚ (Fig. 4.3), were designed and
fabricated to be weld-critical under the applied tension loads. The fabrication drawings for these specimens
are provided in Appendix C.2. A further 21 tests on RHS-to-rigid plate connections conducted by Frater
(1986) and Oatway (2014) are also discussed. A total of 33 HSS-to-rigid plate connections have thus been
tested, and are analyzed herein.
Fig. 4.3. Connection specimens (with RHS or CHS members)
4.3.1. GEOMETRIC PROPERTIES

The RHS and CHS designation, angle between the branch and plate (θ), and plate thickness (tp) of all
connection specimens are given in Tables 4.2 to 4.4. Different weld sizes, with the intended leg size ranging
from 4 to 16 mm, were selected to investigate the validity of the sinθ factor comprehensively. Prior to testing,
all test welds were ground into a triangle shape using a hand-held grinder so that the weld leg sizes, as well as
the theoretical effective throat thickness, could be measured externally with accuracy using a standard or

skew-T fillet weld gauge. For each RHS connection specimen, the weld cross-sectional dimensions were
carefully measured at 20 locations around the footprint of the branch. For each CHS connection specimen,
the weld cross-sectional dimensions were measured at uniform increments of 25-30 mm around the footprint
of the branch. The averages of the theoretical effective throat thickness of the weld (tw), and the weld leg
length measured along the branch (lv) and along the plate (lh) are listed in Tables 4.2 - 4.4. Complete
measurements of the test welds in the CHS-to-rigid plate connections are provided in Appendix C.3.
Table 4.2. Measured geometric properties, failure loads, and failure modes for θ = 90° RHS-to-rigid plate
specimens
RHS Dimensions Exp.
Test no. θ Bb × Hb ×tb tp lw Average of tw Average of lv Average of lh Pa Fail.
° mm×mm×mm mm mm mm mm mm kN Mode
1† 3.59 5.45 4.78 831 W
† 127.0 × 127.0 × 7.78
2 5.91 8.30 8.43 1166 W
90˚ (r0 = 15.88 mm) 25.0 481
3† (ri = 8.98 mm) 5.29 7.73 7.25 1235 W
†
4 5.61 9.65 6.90 1311 W
†
5 6.38 9.28 8.78 2433 W
† 177.8 × 177.8 × 12.53
6 8.61 13.28 11.30 2574 W
90˚ (r0 = 35.0 mm) 25.0 651
7† (ri = 23.4 mm) 7.02 10.38 9.53 2525 W
†
8 7.35 12.23 9.20 2302 W
††
9 4.20 5.30 6.90 1020 W
††
10 3.57 4.40 6.10 960 W
††
11 3.24 3.90 5.80 840 W
12†† 127.6 × 127.6 × 9.54 4.74 6.30 7.20 1140 W
13 †† 90˚ (r0 = 19.08 mm) 19.0 478 5.69 8.10 8.00 1200 WP
†† (ri = 9.54 mm)
14 5.44 7.50 7.90 1207 WP
††
15 7.71 10.80 11.00 1494 P
16†† 9.33 13.20 13.20 1578 P
††
17 10.82 15.30 15.30 1788 P
Note 1: ro = average outside corner radius of RHS; ri = average inside corner radius of RHS.
Note 2: W = weld failure; WP = mixed failure mode of weld failure and partial plate failure; P = end-plate rupture along
at least one weldment.
†
Tests by Oatway (2014).
††
Tests by Frater (1986); ri and ro are taken as 1.0tb and 2.0tb, respectively, for all RHS used by Frater (1986).

Table 4.3. Measured geometric properties, failure loads, and failure modes for θ = 60° RHS-to-rigid plate
specimens
RHS Dimensions lw Average of tw Average of lv Average of lh Exp.
θ Bb × Hb ×tb tp mm mm mm mm Pa Fail.
Test no. ° mm×mm×mm mm a&b c d a&b c d a&b c d a&b c d kN Mode
18†† 137 119 119 6.10 4.20 8.30 8.00 8.40 10.8 9.40 8.60 8.80 1131 WP
†† 127.0 × 127.0 × 7.78
19 137 119 119 3.80 2.70 5.20 5.60 5.40 6.10 5.30 5.60 6.00 982 W
60˚ (r0 = 15.88 mm) 19.0
††
20 (ri = 8.98 mm) 137 119 119 4.90 3.80 8.10 8.30 8.60 11.2 6.00 6.80 8.50 1270 WP
††
21 137 119 119 9.20 6.80 12.4 12.5 13.2 14.5 13.5 13.9 14.2 1534 WP
Note 1: Locations a, b, c and d are indicated in Fig. 4.3.
Note 2: ro = average outside corner radius of RHS; ri = average inside corner radius of RHS.
Note 3: W = weld failure; WP = mixed failure mode of weld failure and partial plate failure; P = end-plate rupture along
at least one weldment.
††
Tests by Frater (1986); ri and ro are taken as 1.0tb and 2.0tb, respectively, for all RHS used by Frater (1986).
Table 4.4. Measured geometric properties, failure loads, and failure modes for θ = 90° and θ = 60° CHS-to-
rigid plate specimens
CHS Dimensions Exp.
θ Db ×tb tp lw Average of tw Average of lv Average of lh Pa Fail.
Test no. ° mm×mm×mm mm mm mm mm mm kN Mode
22 4.81 7.04 6.60 1261 W
90 167.9 × 6.70 25.0 528
23 6.63 9.64 9.13 1279 W
24 6.87 9.89 9.54 1459 W
90 127.4 × 11.55 25.0 401
25 7.98 11.23 11.34 1597 W
26 6.38 8.85 9.22 841 W
90 101.0 × 7.34 25.0 318
27 6.16 9.23 8.28 864 W
28 5.32 6.13 7.41 1450 W
60 167.9 × 6.70 25.0 569
29 5.73 6.88 7.71 1331 W
30 5.21 8.06 8.54 1109 W
60 127.4 × 11.55 25.0 432
31 6.78 10.50 11.13 1479 W
32 5.39 6.93 7.75 776 W
60 101.0 × 7.34 25.0 342
33 4.98 6.71 7.32 803 W
Note: W = weld failure; WP = mixed failure mode of weld failure and partial plate failure; P = end-plate rupture along at
least one weldment.
After testing and noting the actual load at failure, each connection was cut (where possible) normal to the
longitudinal axis of the weld at several locations around the footprint of the branch (two cuts per side for the
RHS, and at the locations of the weld cross-sectional dimension measurements for the CHS). After surface
polishing, all cross-sections were subjected to macroetch examinations in accordance with ASTM E340-15
(ASTM 2015b) using a 10% nital etchant solution to observe the weld profile. These cross-sections were then
scanned and input into AutoCAD so that the weld cross-sectional dimensions tw, lv and lh could be re-
measured with greater accuracy (Fig. 4.4). For these measurements, the throat thickness of the fillet weld was
taken as the height of the largest triangle that could be inscribed within the fusion faces and the weld surface.

Fig. 4.4. Examples of fillet weld throat measurements from the macroetch examinations
For the RHS, credence was given to the externally measured values of lv and lh using the fillet weld gauge
(presented in Tables 4.2 and 4.3), since all macroetch measurements showed good agreements with the
external measurements. The values of tw in Table 4.2 were then determined from the weld leg measurements
using Eq. (4.7), which takes into account the effect of unequal weld leg sizes and the local dihedral angle
(angle between the base metal fusion faces), ψ, on the orientation of the weld throat plane. A full derivation
of this equation is provided in Appendix C.4.
lv lh sin
tw  (4.7)
l  lh2  2lv lh cos
2
v
where lv = weld leg measured along the HSS branch; and lh = weld leg measured along the plate.
For the CHS, credence was given to the macroetch measurements of lv, lh and tw, since there was a
sizeable discrepancy with some of the external measurements. The values of tw, lv and lh in Table 4.4 and
Appendix C.3 hence correspond to these measurements.
The total weld lengths (lw in Tables 4.2 - 4.4) were determined based on the HSS perimeters and were
hence measured along the root of the weld considering the angle between the HSS and plate, and the corner
radii of the RHS.

4.3.2. MECHANICAL PROPERTIES

All RHS were cold-formed to CSA G40.20/G40.21 Grade 350W (CSA 2013a), and all CHS were cold-
formed to ASTM A500 Grade C (ASTM 2013). Tensile coupons were machined from the RHS (flat faces
away from the weld seam), CHS (at 90°, 180° and 270° from the weld seam), and from the intermediate rigid
plates, and tested in accordance with ASTM A370 (2017) to determine the material properties of the base
metals. The averages of the measured yield stresses (Fyb and Fyp, determined by the 0.2% strain offset
method) and ultimate strengths (Fu and Fup) are given in Table 4.5.
Table 4.5. Measured material properties for all 33 HSS-to-rigid plate connections
HSS End-plate Weld metal
Fyb Fu Fyb/Fu rup Fyp Fup Fyp/Fup rup Fyw FEXX Fyw/FEXX rup
Test No. MPa MPa MPa MPa MPa MPa
1-4 412 478 0.862 0.33 383 563 0.680 0.24 563 619 0.910 0.28
5-8 380 489 0.777 0.33 383 563 0.680 0.24 563 619 0.910 0.28
b a
9 - 11, 17 & 19 426 500 0.852 - 351 558 0.629 0.36 634 687 0.923 0.24
b a
12 - 16, 18, 20 & 21 426 500 0.852 - 351 558 0.629 0.36 641 739 0.867 0.24
22, 23, 28, 29 421 501 0.840 0.31 409 566 0.723 0.24 501 571 0.877 0.26
24, 25, 30, 31 431 488 0.883 0.35 409 566 0.723 0.24 501 571 0.877 0.26
26, 27, 32, 33 385 450 0.856 0.35 409 566 0.723 0.24 501 571 0.877 0.26
a
Coupons fractured at the gauge lines giving incorrect elongations.
b
Excludes results for corner coupons published by Frater (1986).
Matching electrodes with a nominal tensile strength of 490 MPa were used for all fillet welds. To
determine the material properties of the as-laid weld metals, all-weld-metal tensile coupons were created in
accordance with AWS D1.1 (2015). The averages of the measured yield stresses (Fyw, determined by the
0.2% strain offset method) and ultimate strengths (FEXX) of the as-laid weld metals are also given in Table
4.5. The results of the TC tests for the CHS, end plate, and weld metal used for the CHS-to-rigid plate tests
(nos. 22 - 33) are provided in Appendix C.5.
4.3.3. INSTRUMENTATION
For all RHS connection specimens, a group of four strain gauges (Group A) were mounted on the four
faces of the RHS well above the intermediate plate. The purpose of these strain gauges was to measure any
strain difference between opposite RHS faces during testing, thereby monitoring any unintended bending
moments. It was found that no bending moment was applied to any connection, hence all specimens were
loaded in pure tension during testing. In order to further verify that the weld elements were uniformly loaded
along their lengths, another group of eight strain gauges (Group B) were placed on two adjacent faces of the
RHS just above the intermediate plate for all connection specimens. Typical load-strain curves at four
different locations along one side of an RHS are shown in Fig. 4.5. Since the strains at four points along each

RHS wall remained almost identical, with increasing branch load, it was deduced that all welds were
uniformly loaded during each RHS connection test; hence, all of the weld length could be considered
effective. Fig 4.5 shows that, at high applied loads, strains along the RHS branch face start to deviate from
each other. This is due to progressive yielding of the branch (due to varying amounts of cold working along
the face), and the weld (due to the weld size being close to – but not exactly – uniform). Additionally, this is
caused by variations in the extent of the welding heat affected zone, but it has little effect on the findings.
For all CHS connection specimens, Group A and B consisted of four or eight strain gauges mounted with
uniform spacing around the CHS perimeter either well above the intermediate plate (Group A) or just above
the intermediate plate (Group B). Similarly, it was found that all welds were uniformly loaded during the
tests. Linearly varying differential transformers (LVDTs) were also used to measure the load-displacement
behaviour of the connection region (see Fig. 4.6a,b).
Fig. 4.5. Typical load-strain curves from four strain gauges on one side of RHS (Group B strain gauges,
shown for test no. 5)

(a) Instrumentation for RHS-to-rigid plate tests

(Oatway 2014)
(b) Instrumentation for CHS-to-rigid plate tests (c) overall test set-up (shown for a CHS-to-rigid plate test)
Fig. 4.6. Instrumentation and test set-up
4.3.4. CONNECTION TESTS

All connection specimens were tested using a 2700-kN capacity MTS universal testing machine, as
shown in Fig. 4.6c. The connection specimens were loaded to failure in a quasi-static manner to eliminate any
strain rate effect on the weld strength. Most RHS-to-rigid plate tests and all CHS-to-rigid plate tests failed by
weld rupture, as shown in Fig. 4.7. Photographs of the experimental tests and failure modes for CHS-to-rigid
plate tests are provided in Appendix C.6. The actual (experimental) failure loads (Pa), and failure modes, are
given in Tables 4.2 and 4.3 (RHS-to-rigid plate tests) and 4.4 (CHS-to-rigid plate tests).

(a) RHS-to-rigid plate (shown for test no. 1) (b) CHS-to-rigid plate (shown for test no. 22)
Fig. 4.7. RHS- and CHS-to-rigid plate specimens at failure
4.4. RESULTS AND ANALYSIS

For assessment of the various fillet weld design equations, analyses of test data have been performed
using the measured weld effective throat size (i.e. the minimum distance between the root of the fillet weld
and the face of the triangular weld profile), which is the weld theoretical or effective throat size that a
designer would use in calculations. This corresponds to the dashed line shown in the example of Fig. 4.4,
even though the typical fracture plane through the weld was generally closer to the HSS fusion face (as
indicated by the solid line in Fig. 4.4) and has a larger area. This measured weld throat size (tw) was
multiplied by lw to obtain the weld area, where the weld length was taken as the appropriate portion of the
HSS perimeter, considering the rounded corners of the RHS. Using this weld length provides a more
scientific evaluation of the sinθ effect, although it is recognized that most designers would simply calculate
the weld length for RHS from the overall height and width of the branch member, Hb and Bb, respectively
(especially if the branch was inclined). The Hb and Bb approach always produces a longer weld length, thus
exaggerating the real weld length and giving a higher predicted strength, which is unconservative for design.
The experimentally obtained weld strengths (Tables 4.2 - 4.4) can then be compared to the predictions in
accordance with each code/specification to determine if an appropriate safety index (or safety margin) is
achieved, as expected for brittle elements (welds) in structural design codes, both with and without
application of the sinθ factor. The strengths of the four weld elements in the 60˚ RHS connections (Specimen
Nos. 18 – 21) were calculated individually since they were loaded in a different manner. Similarly, the
strengths of the six 60˚ CHS specimens were calculated by summing up “component” weld strengths along
25 mm – 30 mm lengths of weld (tributary to each weld cross-sectional dimension measurement). Examples

of this procedure for several code methods applied to a 60° CHS-to-rigid plate connections are provided in
Appendix C.7.
4.4.1. ANSI/AISC 360 (2016)

The predicted nominal strengths (Pnw) of the test welds per AISC 360 without the sinθ factor are
compared to the actual strengths at failure in Fig. 4.8. For the 90˚ connections, the nominal strengths of the
test welds were determined using Eq. (4.1a,b). For the 60˚ connections, the nominal strengths of the test
welds were computed using Eq. (4.2a). In this case, Pnwl is applied to the RHS oblique welds at locations ‘a’
and ‘b’ (see Fig. 4.3) based on their real oblique weld lengths, and to the 60º CHS welds based on their real
elliptical weld length. Hence, all sinθ effects are omitted.
Fig. 4.8. Comparison of actual and nominal strengths Fig. 4.9. Comparison of actual and nominal strengths
per AISC 360-16 without the sinθ factor per AISC 360-16 with the sinθ factor
The predicted nominal strengths of the test welds per AISC 360 with the sinθ factor are compared to the
actual strengths at failure in Fig. 4.9. For the 90˚ connections, the nominal strengths of the test welds were
determined using Eq. (4.1a,c). For the 60˚ RHS connections, the nominal strengths of the test welds were
computed using Eq. (4.2b) with the 1.5 factor for Pnwt. Eq. (4.1a,b) are used to calculate Pnwt since the
directional strength-increase is considered by the 1.5 factor already. Eq. (4.1a,c) are used to calculate Pnwl to
account for the directional strength-increase factor for the 60˚ oblique welds. Also, Pnwl is multiplied by a
0.85 factor [similar to the Mw factor in Eq. (4.3a) per CSA S16-14] to account for the difference in
deformation capacity between the oblique and transverse weld elements. (In theory, the 0.85 factor should
really be higher since the 0.85 applies to longitudinal welds). For the 60˚ CHS connections, Eq. (4.1a-c) were
used to compute the strength of the weld component tributary to each measurement. These strengths were

then added together to calculate the nominal joint strength. The 0.85 factor in Eq. (4.2b), which could
technically apply - since the connection encompasses weld components with multiple orientations - was
deemed too general for this situation, and omitted from the analysis.
4.4.2. CAN/CSA S16 (2014)

Similarly, the predicted nominal strengths of the test welds per CSA S16-14 without and with the sinθ
factor are computed using Eq. (4.4a,b), and are compared to the actual strengths in Figs. 4.10 and 4.11. The
Mw factor [Eq. (4.4a)], akin to the 0.85 factor in Eq. (4.2b), is continuous for CHS joints with respect to θ,
and there was hence a rational basis to apply it to the 60° CHS connections. The value of θ used in both the
sinθ factor and Mw factor (as well as in the AISC 360-16 approach with the sinθ factor) to compute each
component strength of a CHS joint was determined by numerical integration of a θ(lw) curve, which was
derived using computer-programmed vector calculus equations in Matlab. The method used to derive this
function is illustrated in Appendix C.8, and the average values of θ along tributary weld lengths are provided
for 60° CHS connections in Appendix C.3.
per CSA S16-14 without the sinθ factor per CSA S16-14 with the sinθ factor
4.4.3. CAN/CSA S16 (2001)

The predicted nominal strength of each welded joint per CSA S16-01 without the sinθ factor was taken as
the least of the limit states of: (i) shear rupture along the fusion face along the HSS branch [using Fu, lv and
Eq. (4.3a)], (ii) shear rupture along the fusion face along the intermediate plate [using Fup, lh and Eq. (4.3a)],
and (iii) shear rupture along the plane of the weld effective throat [using FEXX, tw and Eq. (4.3b) without the

sinθ factor). The predicted nominal strength of each welded joint per CSA S16-01 with the sinθ factor was
determined by repeating the above procedures, with the sinθ factor in Eq. (4.3b) included. An example of this
procedure applied to a 60° CHS-to-rigid plate connection is provided is Appendix C.7.
All predictions per CSA S16-01 are compared to the actual strengths in Figs. 4.12 and 4.13. When the
sinθ factor is not used (Fig. 4.12), the predicted nominal strengths of nearly all of the 90° HSS connections
(16 of 17 RHS connections and 6 of 6 CHS connections) are governed by the limit state of shear rupture
along the plane of the weld effective throat, and all but one of the 60° HSS connections are governed by the
limit state of shear rupture along the fusion face with the base metal at some location along the weld length.
When the sinθ factor is included in the calculation (Fig. 4.13), the nominal strengths of all connections are
governed by fusion face rupture at some location along the weld length.
per CSA S16-01 without the sinθ factor per CSA S16-01 with the sinθ factor
4.4.4. EN1993-1-8 (2005)

Following the European fillet weld design criteria, the capacity of the tested fillet-welded joints was
calculated using the stress components on the theoretical throat plane, as illustrated previously in Fig. 4.2a.
The design strength of the weld joint in all RHS and CHS connections was determined using Eqs (4.8a) to
(4.8e), assuming a theoretical angle between the planes of the effective throat and the fusion face, ,
determined from the weld geometry and hence taking into account the effect of unequal weld leg sizes and
the local dihedral angle on the orientation of the weld throat plane. Although a more accurate comparison
between tested and calculated strengths may be conducted by measuring the actual angle between the planes
of the effective throat and the fusion face, the theoretical angle is used since it was not possible to perform

internal weld geometry measurement via sectioning on all connection specimens. Eq. (4.5b) was satisfied in
all cases. The correlation factor for fillet welds, w, was taken as 0.9 according to Table 4.1 in EN1993-1-8
(CEN 2005) for both cold-formed hollow sections (to EN10219) and hot-finished hollow sections (to
EN10210), for grade S355 (HSS with a nominal yield strength of 355 MPa). Since the target safety
(reliability) index for this Eurocode method is unknown, a comparison is performed against the limit states
design resistance, including the partial safety factor, 𝑀2 .
P cos 
  (4.8a)
t w lw
P sin  cos 
  (4.8b)
t w lw
P sin  sin 
  (4.8c)
t w lw
Substituting the above into Eq. (4.5a) gives:
P 0.5
sin 2  cos 2   3(sin 2  sin 2   cos 2  )   Fu / (  w M 2 )
t w lw (4.8d)
Fu t w lw Pnw
P   Vrw (4.8e)
(  w M 2 ) sin 2  cos 2   3(sin 2  sin 2   cos 2  )  0.5
M2
 
Design strengths (Vrw) calculated with Eq. (4.8e) are compared to the actual strengths in Fig. 4.14, which
shows that EN1993-1-8 produces safe predictions for all tested weld joints. The average actual-to-design
strength ratios for the RHS and CHS connections are 1.46 and 1.53, respectively (with an average of 1.48
overall). The predicted and/or design strengths discussed in this section (Section 4.4) are summarized for
each joint in Appendix C.10.

Fig. 4.14. Comparison of actual strengths and design strengths per EN1993-1-8 (2005): Directional Method

4.4.5. EVALUATION OF DIRECTIONAL STRENGTH-INCREASE FACTOR

To assess whether adequate safety margins are inherent in the correlations shown in Figs. 4.8 - 4.13, one
+
can check to ensure that the stipulated minimum target safety indices of = 4.0 (per Chapter B of the AISC
+
Specification Commentary) and = 4.5 (per Annex B of CSA S16) are achieved, using a reliability analysis.
The reliability analysis procedure is described in Section 2.6.2. The same values or equations used there for
ρM, ρG, ρP, VM, VG, VP, and ϕ + (see Section 2.6.2 for definitions) have also been used.
For the predicted nominal strengths per AISC 360-10 without the sinθ factor (Fig. 4.8), the application of
the reliability analysis produces  = 0.753 which exceeds 0.75 as specified by AISC 360-16. If  = 0.75 is
used to calculate the design strength, an implied safety index of 4.02 is achieved. Hence, on the basis of the
available experimental evidence, the predictions can be deemed adequately conservative. The predicted
nominal strengths per CSA S16-14 without the sinθ factor (Fig. 4.10) provide a calculated  = 0.604, which is
+
less than 0.67 as specified by CSA S16-14; however, although the implied safety index, = 4.03 < 4.5 as
specified by CSA, it exceeds that required by AISC.
However, the predicted nominal strengths per AISC 360-16 and CSA S16-14 with the sinθ factors (Figs.
4.9 and 4.11) are unsafe since the calculated  values (0.518 and 0.409, respectively) are much lower than the
corresponding specified resistance factor values (0.75 and 0.67, respectively for AISC and CSA). Viewed
+
another way, the implied safety indices (indicated by on Figs. 4.9 and 4.11) are well below the target
safety indices for AISC and CSA.
As shown in Figs. 4.12 and 4.13, the calculated -values for CSA S16-01 without and with the sinθ
+
factor, using = 4.5, are 0.639 and 0.587. CSA S16-01 (Fig. 4.13) is noticeably more conservative than
CSA S16-14 (Fig. 4.11) with the sinθ factor.
4.4.6. ISOLATION OF CHS-TO-RIGID PLATE TEST RESULTS

If only the results of the 12 CHS-to-rigid plate experimental tests are considered, the following results
+
with respect to the reliability analysis above. For AISC 360-16 and CSA S16-14 without the sinθ factor, =
5.16 (for both) ≥ 4.0 and 4.5. For AISC 360-16 and CSA S16-14 with the sinθ factor, +
= 3.12 (for both) <
4.0 and 4.5. For CSA S16-01 without and with the sinθ factor, +
= 5.28 and 4.41 (or and 0.659 for
+
= 4.5). Thus, the conclusions stated previously for all HSS are valid for the CHS considered separately
based on the experimental evidence. Still, an extension of this evidence to determine the effect of weld size
and other connection parameters when the weld is fully effective, is needed.

4.5. INFLUENCE OF WELD SIZE

One must bear in mind that the strength of fillet welds is also influenced by the amount of weld
penetration into the base metals at the root. Small fillet welds and large fillet welds tend to have the same
degree of root penetration; for large multi-pass welds the root penetration is generally determined just by the
root pass. In laboratory testing, then, the strength of small fillet welds will be aided proportionally more than
for large fillet welds, by the root penetration. The effect of this aid is generally linear for each branch type
(RHS or CHS) and angle (see Fig. 4.15), and the magnitude of the so-called “weld size effect” varies
depending on the model investigated. No trend is observed for the branch cross-sectional slenderness (Fig.
4.16). Since most laboratory tests on weld-critical joints involve relatively small welds (because the failure
mode of weld fracture must be achieved), the results obtained are likely more optimistic compared to results
from large-weld tests.
Fig. 4.15. Effect of weld size on actual-to-nominal Fig. 4.16. Effect of branch slenderness on actual-to-
strength ratio per AISC 360-16 without the sinθ factor nominal strength ratio per AISC 360-16 without the
sinθ factor

4.6. SUMMARY
In this chapter, 33 HSS-to-rigid plate weld-critical connections that were tested to failure under axial
tension loading have been compared. The design methods in CSA S16-01, EN1993-1-8:2005, AISC 360-16,
and CSA S16-14 for fillet welds to HSS members were evaluated by comparing the actual strengths of the
fillet welds to the predicted strengths. It can be concluded from this work that:
1. The Directional Method in Eurocode 3 produces safe strength predictions for fillet welds to HSS.
Hence, the Simplified Method is even more conservative.
2. When the (1.00+0.50sin1.5θ) directional strength increase factor is not included in the strength
calculation of fillet welds to HSS, the equations in both the current American and Canadian
specifications meet the target safety index according to AISC ( +
≥ 4.0).
3. Restrictions need to be placed in current North American steel design codes on the application of
such a directional strength increase factor for fillet welds in HSS connections. It should be noted
that the directional strength-increase factor was developed based on tests on fillet welds in lap
splice connections. According to this investigation of fillet welds in HSS connections, strength
calculation including a directional strength increase factor leads to predictions which do not have
a sufficient safety margin, even when it is not used in conjunction with the “effective length
method” of AISC 360 Chapter K.
4. The relative strength (per unit throat thickness) of small fillet welds is generally greater than for
large fillet welds.
5. CHS-to-rigid plate specimens generally exhibited higher average strengths than did RHS-to-rigid
plate specimens. This is believed to be due to the greater restraint from local rotation offered by
the constant curvature of the CHS wall. RHS, on the other hand, have less restraint from rotation
along the flats, especially for slender sections, which results in greater local rotation and higher
tension at the weld root. Additionally, non-uniform stiffness of the branch parallel to the branch
longitudinal axis (resulting in stress concentrations in the corners) may play a role. This is
investigated in Chapter 5.
4.6.1. FOOTNOTE
It is worth noting that the above tests were performed on connections between HSS and rigid end-plates.
In typical connections that occur between HSS and flexible end-plates, where the end plates are fastened to
adjacent structural components using fasteners or anchors, weld effective lengths would re-occur.

Chapter 5: FILLET WELDS IN HSS-TO-RIGID PLATE
CONNECTIONS: FINITE ELEMENT MODELLING
5.1. INTRODUCTION
When proportioning welds to resist the applied forces in axially-loaded branch members of hollow
structural section (HSS) connections, weld effective lengths are required (ISO 2013). Weld effective lengths
account for the non-uniform loading of the weld due to differences in the relative flexibility of the HSS chord
face, around the weld perimeter. Weld effective lengths for truss connections between rectangular hollow
sections (RHS) have been researched at the University of Toronto over the last several decades by Frater &
Packer (1992a, 1992b), Packer & Cassidy (1995), McFadden & Packer (2014), and Tousignant & Packer
(2015). AISC 360 Section K5 (AISC 2016), and Packer & Henderson (1997), present weld effective length
rules for designers derived from this research. McFadden & Packer (2014) and Chapter 2 have shown that the
fillet weld “directional strength-increase factor” (1.00+0.50sin1.5θ) in Section J2.4 of AISC 360 (AISC 2016),
and its equivalent of (1.00+0.50sin1.5θ)Mw in Clause 13.13.2.2 of CSA S16 (CSA 2014), does not provide
adequate structural reliability (or safety) when used in conjunction with the rules for weld effective lengths.
AISC (2011) and CISC (2016) hence disallow the use of the directional strength-increase, or “sinθ factor”,
when these effective length rules are used to design fillet welds in HSS-to-HSS connections.
Fig. 5.1. HSS-to-rigid plate connection specimens (with RHS or CHS members)
Chapter 4 presented a large number of laboratory tests on HSS-to-rigid plate connections (Fig. 5.1) to
investigate the applicability of the sinθ factor to single-sided fillet welds to HSS, joined to a rigid end-plate.
These experiments removed the influence of a flexible landing surface for the fillet weld, and hence removed
the weld effective length phenomenon. It was shown that HSS-to-rigid plate fillet welds still did not provide
adequate structural reliability if the sinθ factor was implemented.
66
Chapter 5: Fillet Welds in HSS-to-Rigid Plate Connections: Finite Element Modelling 67
The research herein is a finite element (FE) extension of the experimental work to determine (for HSS-to-
rigid end-plate connections):
(a) the extent to which fillet welds to RHS and circular hollow sections (CHS) are similar;
(b) the effect of relative weld size (tw/tb), branch wall slenderness (Bb/tb and Db/tb), and branch
inclination angle (θ) on the weld strength; and
(c) if alternate expressions are more appropriate for estimating the strength of fillet welds to HSS.
5.2. PREVIOUS EXPERIMENTATION

Chapter 4 presented the results of 33 total weld-critical HSS-to-rigid plate connections tested at the
University of Toronto between the mid-1980s and present. Thirteen of these tests were done by Frater (1986)
on RHS-to-rigid plate connections, with fillet weld throat dimensions (tw) ranging from 0.37 to 1.13 times the
branch wall thickness (tb). Frater (1986) found that, for connections with small welds (tw/tb ≤ 0.50) and a
branch inclination angle θ = 90°, rupture always occurred through the weld, around the entire branch
perimeter (failure mode W). When welds were only slightly larger (0.50 < tw/tb ≤ 0.60), rupture generally
occurred within the weld, but was accompanied by rupture of the end-plate near the middle of the RHS
branch member walls (failure mode WP). For connections with the largest welds tested (0.81 ≤ tw/tb ≤ 1.13),
end-plate rupture governed failure, with only some weld rupture occurring at sharp angles around the RHS
corners (failure mode P). When θ = 60°, Frater (1986) found that only the connection with the smallest weld
size (tw/tb = 0.42, where tw/tb ranged from 0.42 to 1.00) ruptured through the weld, around the entire branch
perimeter. The remaining three RHS connections with θ = 60° ruptured in the end-plate, instead of the weld,
along the heel of the connection.
Eight additional RHS-to-rigid plate connections, with θ = 90° and similar weld sizes to Frater (1986)
(0.46 ≤ tw/tb ≤ 0.76), were tested by Oatway (2014). In all of these tests, rupture occurred through the weld,
around the entire branch perimeter. In 12 further tests by the author on CHS-to-rigid plate connections
(Chapter 4), with 0.45 ≤ tw/tb ≤ 0.99, the same failure mode was observed. Three tests on CHS-to-rigid plate
connections were also performed at Tongji University, China (Wang et al. 2015); however, only one of them
failed entirely by weld rupture. A second test failed by weld rupture combined with branch rupture in the
heat-affected zone, and the third test did not reach the ultimate load.
5.3. FINITE ELEMENT MODELS

In order to validate the FE modelling procedure used herein, 33 RHS- and CHS-to-rigid plate connection
FE models, replicating experimental tests, were developed using ANSYS (Swanson Analysis Systems 2011).
These models covered the following geometric parameters: θ = 60° and 90°, 0.34 ≤ tw/tb ≤ 1.13, CHS
branches with 11.0 ≤ Db/tb ≤ 25.1, and RHS branches with 13.4 ≤ Bb/tb ≤ 16.3.

5.3.1. GENERAL
Each FE connection was comprised of a single tension-loaded branch welded to a rigid end-plate, with
fixed restraints applied to the nodes on the “underside” of the end-plate. When θ = 60°, one half of each FE
connection could be modelled using symmetry boundary conditions parallel to the axis of the toe and heel of
the connection (Figs. 5.2 and 5.3). When θ = 90°, an additional plane of symmetry, orthogonal to the first
plane and through the centre of the branch, allowed one quarter of each connection to be modelled (see
Appendix D.1). It was shown that these FE models provided identical load-displacement responses and the
same weld rupture loads as FE models that included two concentric, tension-loaded, branches (welded to
opposite sides of the end-plate). Since two branches were used for the experimental tests in the manner
described, it was necessary to demonstrate that the above models were equivalent, prior to evaluating them.
Fig. 5.2. FE CHS-to-rigid plate connection model with θ = 60°

Fig. 5.3. FE RHS-to-rigid plate connection model with θ = 60°
Fig. 5.4. Effective weld dimensions in FE models (shown for a model with θ = 90°)
To ensure that load transfer was only through the fillet weld, a gap was used to separate the branch and
the end-plate at their interface, as shown in Fig. 5.4. From preliminary analyses of the connections, the size of
the gap was determined to have a significant effect on the non-linear response of the fillet weld. To take into
account (and mitigate) this effect: (a) the size of the gap can be minimized (close to the Boolean tolerance of
the FE program), and (b) the calculation of the weld throat area (Aw) can be done based on the portion of the
fillet above the gap, as shown in Fig. 5.4. In applying (a) and (b), the load-displacement response and

ultimate fracture strain of the FE models were found to closely agree with the experimental tests. Generally,
however, when welds are much larger than the Boolean tolerance of the FE program, provision (b) is not
necessary, since the size of the gap resulting from provision (a) will be small compared to weld leg
dimensions (lv and lh).
Models were analyzed under static incremental displacements applied to the end of the branch, in the
theoretical constant stress region identified by Mehrotra & Govil (1972) – a distance of 3Bb or 3Db from the
weld toe. It was therefore only necessary to model the portion of the branch within this distance.
5.3.2. MATERIAL MODELLING

Multi-linear true stress-strain ( T- T) curves were developed from corresponding engineering stress-strain
( - ) curves for the HSS, rigid end-plate, and weld metal, and used to describe the behaviour of materials in
the FE models. Up to coupon necking, which corresponds to the ultimate engineering stress and strain ( u and
u, respectively), the - curve obtained from an average of three tensile coupon (TC) tests, in accordance
with ASTM A370 (ASTM 2015), was converted to a T- T curve using the following relationships (Boresi &
Schmidt 2003):
 T   (1   ) (5.1)
T  ln(1   ) (5.2)
It is well-known that different material properties are obtained in the HSS corners (higher yield stress,
higher ultimate stress, lower ductility) relative to the flats, where coupon measurements were taken; however,
the corner area as a proportion of the total HSS cross-sectional area is typically low, so these differences will
be diminished. Furthermore, the objective of the FE modelling exercise was primarily to capture the weld
behaviour and fracture, so the refinement in HSS modelling was considered to be of low influence on the
results.
Eqs. (5.1) and (5.2) are not valid past the coupon necking point, since the stress distribution in the
material is no longer uniaxial. Therefore, an iterative approach given by Ling (1996) was used to generate the
remainder of the T- T curve. Ling’s (1996) expression for the T- T curve in the necked region is given by Eq.
(5.3):
  T  
 T   T   w(1   T   T  )  (1  w)  T
 (5.3)
 T  
  T  

where 𝜎𝑇′ is the true stress at which necking starts, ε′𝑇 is the true strain at which necking starts, and w is a
weighting factor.
Ling’s (1996) approach is based upon weighting an approximate lower- and upper-bound to the T- T
curve past the coupon necking point. The lower bound (w = 0.0) represents a power law (Hollomon 1945),
and the upper bound (w = 1.0) is linear (Fig. 5.5).
Fig. 5.5. Post-necking generated T- T curve using the Ling (1996) procedure (shown for CHS 127.4 × 11.55
TC)
The weighting constant, w, is determined by selecting a trial value of w to generate points on the T- T
curve in the necked region, running an incremental load-step analysis of a TC up to fracture, and comparing
the - curve from the model, which is generated from applied loads and nodal displacements, to
experiments. The nodal displacements, and hence the FE strains, were calculated over a “virtual” 50-mm
gauge length from nodes on the exterior of the TC. This process is repeated until a value of w is found that
results in acceptable agreement between the experimental and FE - curves.
For this work, TCs were modelled using average measured dimensions of the coupon widths and
thicknesses (or diameters, for screw-type TCs), and nominal values of the grip dimensions, including the
radius of the machined fillet and the length of the reduced section outside of the TC gauge length (ASTM
2017). CHS TCs were modelled with curved geometries, as shown in Fig. 5.6, to replicate both the
experimental tests, and the in-situ condition of the material. ANSYS SOLID45 8-noded brick elements, with
non-linear material and geometric (large deformation) properties, were used.

Fig. 5.6. FE CHS TC (a) geometry and meshing, and (b) comparison of deformed shape and results at
fracture
The clip gauge used to measure strain during the experimental tests was removed at ≅ 0.20, to avoid
damage; however, the elongation of the coupon and the load at rupture were measured and recorded. In the
region of the - curve between first necking and the point where the clip gauge was removed, where Ling’s
(1996) approach was used, the experimental and numerical results always showed good agreement (see Fig.
5.7). At the point of coupon rupture, agreement varied. In two previous FE studies (Voth & Packer 2012a;
Martinez-Saucedo et al. 2006), it was determined that rupture in large-scale HSS connections typically occurs
at strains well below those present at the point of coupon rupture, due to boundary conditions and
confinement in connections that are different from those in TCs. The gap between the HSS and the end-plate
also creates a crack-like feature at the root of the weld, which contributes to earlier rupture in the HSS
connections relative to coupon specimens. Therefore, the variation in agreement between the numerical and
experimental - curves at the point of coupon rupture was deemed inconsequential and, for the same reason,
a fracture criterion was not calibrated for the TCs.
Fig. 5.7. Comparison of numerical and experimental - curves (shown for CHS 127.4 × 11.55 TC)

5.3.3. MODEL SENSITIVITY STUDY

To determine the element type and mesh arrangement best suited for modelling the HSS-to-rigid plate
connections, a sensitivity study was performed. The sensitivity study investigated the relative load-
displacement response of one RHS-to-rigid plate connection and one CHS-to-rigid plate connection, with
different element types, mesh densities, numbers of HSS through-thickness elements, and elements along the
weld face. The objective of the sensitivity study was twofold: (a) to determine the minimum mesh parameters
necessary to obtain numerical convergence, and (b) to select a set of parameters for modelling the remainder
of the RHS- and CHS-to-rigid end-plate connections, including those developed for the parametric study.
Two brick-type elements were examined for connection modeling: an 8-noded solid element with large
deformation and strain capabilities and three translational degrees of freedom per node (SOLID45), and a 20-
noded solid element capable of plasticity, creep, stress stiffening, large deflection, and large strain
(SOLID95). In addition to element type, three different mesh layouts were assessed (fine, medium, and
coarse, as shown in Fig. 5.8). Table 5.1 provides a summary of the different analyses performed for an RHS
and a CHS connection, showing what was varied. Regardless of the element type and mesh layout, all models
produced similar results in terms of initial stiffness (the initial slope of the load-displacement curve), non-
linear behaviour response, and ultimate load (Fig. 5.9).
Fig. 5.8. Mesh arrangements shown on a 1/4 θ = 90° FE CHS-to-rigid plate connection

Table 5.1. FE mesh sensitivity study results

HSS Plate
Test no. Element No. of No. of thickness thickness Weld face Mesh
(branch type) Type elements nodes elements elements elements type Pa PFE* Pa/PFE*
kN kN
SOLID45 15375 20554 3 7 Med. 1208 0.97
SOLID45 18225 23512 4 7 Med. 1235 0.94
2 (RHS) SOLID45 21075 26470 5 7 Med. 1240 0.94
3 1166
SOLID45 34340 43678 4 9 Fine 1232 0.95
SOLID45 9747 12832 4 5 Coarse 1240 0.94
SOLID95 9747 48140 4 5 Coarse 1268 0.92
SOLID45 11385 15470 3 7 Med. 828 1.02
SOLID45 13035 17206 4 7 Med. 832 1.01
26 (CHS) SOLID45 14685 18942 5 7 Med. 833 1.01
3 841
SOLID45 23160 30095 4 9 Fine. 832 1.01
SOLID45 5680 7708 4 5 Coarse 838 1.00
SOLID95 5680 28704 4 5 Coarse 842 1.00
Note 1: Test nos. correspond to Table 5.2.
Note 2: Pa = actual (experimental) weld rupture load; PFE* = FE load at experimental weld rupture displacement.
Fig. 5.9. Load-displacement curves for CHS test no. 26 with fine, medium and coarse mesh, and four CHS through-
thickness elements
A medium mesh and SOLID45 elements (using reduced integration and hourglass control) were selected
for the models based on the results of this study and previous HSS connection FE studies (Voth & Packer
2012a, 2012b) that also obtained good results with these parameters. Seven elements along the weld face, six
elements along the vertical weld leg, and six elements along the horizontal weld leg (see Fig. 5.4, shown
previously), were found to be suitable for achieving good resolution of the weld response under applied
loads. For the HSS, four through-thickness elements, biased towards the outside face of the branch in contact
with the weld, were used to capture local stresses and bending effects due to eccentric loading. Three

elements were used through the plate thickness; however, the number of plate through-thickness elements
was found not to affect the results.
5.3.4. MODEL FRACTURE CRITERION

In spite of the importance of fracture to structural engineering, there is no single widely accepted micro-
mechanics-based model to assess fracture in steel components. This is particularly the case in welded
connections, which contain geometric and material discontinuities, areas of high stress and strains, and
heterogeneous material properties (Kanvinde 2017). Previous researchers have had pragmatic success using a
simple FE fracture criterion based upon maximum equivalent strain ( ef) (Voth & Packer 2012a; Martinez-
Sauced et al. 2006). However, the use of this method requires robust calibration from large-scale tests where
local fracture occurs at the point of interest, because the value of ef to use in a model depends upon element
boundary conditions, mesh arrangement and loading. All of these factors affect the triaxiality at the point of
rupture – a key factor that, in turn, affects the rupture strain, and ef can thus vary greatly between connections
with different geometries and loading conditions. For example, Martinez-Saucedo et al. (2006) found that a
value of ef = 0.60 was appropriate to model the fracture behaviour of HSS in slotted end-plate connections,
while Voth & Packer (2012a) found that a value of ef = 0.20 was appropriate to model similar behaviour in
CHS branch-plate connections.
For the current work, ef for fracture in the weld ( ef,weld) and the end-plate ( ef,plate) was determined by
comparison of the finite element and experimental load-displacement results. For the former, 10 experimental
tests (four RHS-to-rigid plate tests and six CHS-to-rigid plate tests) were selected from the overall sample,
and corresponding numerical models of each connection were analyzed. A value of ef,weld was hence
determined, by trial and error, so that the FE displacement at failure matched the experimental displacement
at failure in the full-scale experimental connection tests. Using displacements instead of loads provided a
more robust fracture criterion, since fracture generally occurred on the plateau of the load-displacement
curve. The mean values of ef,weld for the RHS- and CHS-to-rigid plate connection tests obtained were nearly
equal (0.093 and 0.092, respectively), which hence gave credence to the use of an overall mean, 0.092, for
ef,weld. The five experimental tests on RHS-to-rigid plate connections that exhibited end-plate rupture were
selected from the overall sample to calibrate ef,plate, producing a mean of 0.011. The weld and plate fracture
modelled herein is a significantly less ductile connection than those studied by Martinez-Saucedo et al.
(2006) and Voth & Packer (2012a), both of which dealt with ductile connections that were not weld-critical.
Hence, a lower fracture strain is not surprising.
When the equivalent strain in an element reached ef, the stiffness and hence the stress of that element
were reduced to near zero (1 x 10-6). The inactive element(s) thereafter shed load to surrounding elements
(where the equivalent strain in the element, e < ef) and was permitted to freely deform. This element
behaviour is physically comparable to the initiation and propagation of a crack through the material.

The element “death feature” was thus programmed for elements in both the weld and the end-plate, as a
loop within each load step, to determine the FE ultimate load (PFE), the failure mode (W, P, or WP), and the
sequence of failure (the location of first-cracking and subsequent crack propagation).
5.4. FINITE ELEMENT MODELS EVALUATED AGAINST EXPERIMENTS

5.4.1. ULTIMATE LOADS AND FAILURE MODE
To validate the modelling techniques developed previously, the ultimate load and failure mode predicted
by the finite element models for each of the 33 HSS-to-rigid plate connections were compared to the
experimental results. Table 5.2 presents the key connection parameters (tw/tb, Bb/tb or Db/tb, and θ), actual
(experimental) ultimate load (Pa), experimental failure mode (W, P, or WP), FE-predicted ultimate load (PFE)
and FE-predicted failure mode for each test. The ratio of the actual-to-predicted ultimate load for each test
(Pa/PFE) is also given. Over all 33 tests, the mean of the ratio of actual-to-predicted ultimate load (A/P) is
1.00, with a coefficient of variation (COV) of 0.12. The correlations of A/P for each researcher, and branch
angle, are also shown in Table 5.2. The poorest correlation resulted for the RHS-to-rigid plate connections
tested by Frater (1986), with θ = 60°. All but one of these specimens failed experimentally by end-plate
rupture along the heel of the connection and weld rupture along the remaining three sides.
The numerical models provided good predictions for Pa, and generally gave the correct failure mode.
Only three out of the 33 models did not predict the correct failure mode (nos. 6, 8 and 14 in Table 5.2). From
separate analyses, in which failure was constrained to either the end-plate or the weld, it was found that
specimens nos. 6 and 8, which were predicted to fail at least partially in the plate, were within 3% of the load
at which weld failure would have governed. Similarly, the end-plate rupture load for specimen no. 14 was
only 5% higher than PFE (for weld failure) given in Table 5.2. The correlations presented in this table are
shown graphically in Appendix D.1.

Table 5.2. Comparison of experimental and FE results for HSS-to-rigid plate tests
Failure Mode Ultimate Load
Branch Test mean
Type No. θ Ab Bb/tb or Db/tb tw/tb Exp. FE Pa PFE Pa /PFE (A/P) COV
° mm2 kN kN
1 0.46 W W 831 841 0.99
2 0.76 W W 1166 1252 0.93
RHS 90 3563 16.3
3 0.68 W W 1235 1147 1.08
Oatway
(2014)
4 0.72 W W 1311 1358 0.97

1.04 0.10
5 0.51 W W 2433 2006 1.21
6 0.69 W P 2574 2525 1.02
RHS 90 7702 14.2
7 0.56 W W 2525 2222 1.14
8 0.59 W WP 2302 2460 0.94
9 0.44 W W 1020 953 1.07
10 0.37 W W 960 805 1.19
11 0.34 W W 840 733 1.15
12 0.50 W W 1140 1116 1.02
RHS 13 90 4271 13.4 0.60 WP WP 1200 1357 0.88 1.04 0.10
14 0.57 WP W 1207 1303 0.93
Frater a
(1986)
15 0.81 P P 1494 1444 1.03

16 0.98 P P 1578 1567 1.01
17 1.13 P P 1788 1662 1.08
18 0.66 WP W 1131 1531 0.74
19 0.42 W W 982 975 1.01
RHS 60 4271 13.4 0.83 0.14
20 0.57 WP WP 1270 1590 0.80
21 1.00 WP WP 1534 1905 0.81
22 0.72 W W 1261 1210 1.04
3393 25.1
23 0.99 W W 1279 1523 0.84
24 0.59 W W 1459 1337 1.09
CHS 90 4204 11.0 1.00 0.09
25 0.69 W W 1597 1530 1.04
26 0.87 W W 841 860 0.98
Chapter 4
2160 13.8
27 0.84 W W 864 877 0.99
28 0.79 W W 1450 1207 1.20
3393 25.1
29 0.86 W W 1331 1324 1.01
30 0.45 W W 1109 1278 0.87
CHS 60 4204 11.0 1.02 0.12
31 0.59 W W 1479 1601 0.92
32 0.73 W W 776 763 1.02
2160 13.8
33 0.68 W W 803 743 1.08
Note: W = weld failure; WP = mixed failure mode of weld failure and partial plate failure; P = end-plate rupture along at
least one weldment.
a
ri and ro were taken as 1.0tb and 2.0tb, respectively, for all RHS used by Frater (1986).
Frater’s (1986) tests showed that it is possible for cracks that begin in the weld to propagate into the
plate, or vice versa. Since the FE models were terminated once PFE was reached (to avoid problems that
would later occur with convergence), it is not possible to determine if this phenomenon would have occurred
in the numerical tests. However, to assess failure progression in the FE models from first cracking (the death
of the first element) to the ultimate load (PFE), a list containing the numbers of “killed” elements was output
from each load step. In the 90° RHS-to-rigid plate connections, the first weld elements to fail were generally
close to the RHS corners, whereas the first plate elements to fail were near the middle of the RHS wall. These
failure locations, and the progression of joint failure for test no. 13, which failed in the weld and the end-
plate, are illustrated in Fig. 5.10. The above behaviour suggests that non-uniform stress distributions in RHS

welded joints can still be caused by a difference in relative flexibility of the branch parallel to its surface,
around its perimeter, leading to weaker welds (per unit length) for RHS-to-rigid plate connections versus
CHS-to-rigid plate connections.
Fig. 5.10. Progression of fracture in FE RHS-to-rigid plate test no. 13
The first weld elements to fail in the θ = 60° RHS-to-rigid plate FE tests were located at the toe of the
connection; and when end-plate failure did occur numerically the first end-plate elements to fail were also
located at the toe [and not at the heel, as Frater (1986) observed]. Therefore, more experimental tests may be
necessary to understand the conditions that lead to end-plate rupture in skewed HSS-to-rigid plate
connections (θ < 90°). It is also worth noting that while element failure typically occurred over several load
steps in the RHS-to-rigid plate FE tests (i.e. a crack formed in the weld near the RHS corner, and
propagated), it usually occurred in a single load step in the CHS-to-rigid plate FE tests.
5.4.2. LOAD-DISPLACEMENT RESPONSE

The FE load-displacement curves are compared, for 10 tests, to the corresponding experimental load-
displacement curves in Figs. 5.11 and 5.12. For clarity, the experimental and FE load-displacement curves
have been divided over two plots within each figure. The FE results are shown by solid lines (red) and the
experimental results are shown by dashed lines (black). Differences between these data are believed to be
caused by variations in the experimental weld dimensions around the branch perimeters which are difficult to
mitigate (by grinding) – especially around RHS corners. For the connection FE modelling, average values of
the weld dimensions (tw, lv and lh) were used, and hence these variations were not included. Weld penetration
has also been ignored, but it is implicit in the calibration of ef,weld. Despite these features, the FE and
experimental ultimate loads, initial stiffnesses, and load-displacement responses generally showed good
agreement, except for some tests on RHS-to-rigid end-plate connections (Fig. 5.11). The discrepancies

therein with respect to the elastic limits, between the experimental and FE results, are believed to be due to
residual stresses in the RHS, which were not modelled.
Fig. 5.11. Comparison of RHS-to-rigid plate FE and experimental load-displacement responses
Fig. 5.12. Comparison of CHS-to-rigid plate FE and experimental load-displacement responses
5.5. FINITE ELEMENT PARAMETRIC STUDY

A finite element parametric study was conducted in which 65 90° fillet-welded RHS- and CHS-to-rigid
plate connections with six values of tw/tb (ranging from 0.35 - 1.41), and six values of the branch wall
slenderness (ranging from 9.1 - 50), were analyzed to determine the effect of these parameters on the weld
strength. The effect of θ on the weld strength was afterwards addressed by conducting eight more analyses:

• two on RHS-to-rigid plate specimens with Bb/tb = 12.5 and 50, and θ = 60° and tw/tb = 0.50;
• two on RHS-to-rigid plate specimens with Bb/tb = 12.5 and 50, and θ = 75° and tw/tb = 0.50;
• two on CHS-to-rigid plate specimens with Db/tb = 12.5 and 50, and θ = 60° and tw/tb = 0.50; and
• two on CHS-to-rigid plate specimens with Bb/tb = 12.5 and 50, and θ = 75° and tw/tb = 0.50.
These eight analyses were then compared to four of the previous FE analyses on specimens with θ = 90°
and tw/tb = 0.50 (two RHS-to-rigid plate specimens with Bb/tb = 12.5 and 50, and two CHS-to-rigid plate
specimens with Db/tb = 12.5 and 50). A total of 73 (65 + 8) analyses were hence performed.
The HSS width, or diameter, and the end-plate thickness were kept constant for all joints (Bb = 200 mm,
Db = 168 mm, and tp = 25 mm, respectively), and fillet welds were modelled with equal-sized legs. A 0.25-
mm gap between the branch member and the end-plate was used to restrict load transfer to the weld, as
discussed in Section 5.3.1. All models used SOLID45 8-noded hexahedral elements and were permitted to
fail either by weld or end-plate rupture, using the fracture criterion discussed in Section 5.3.4.
The – curves for each of the materials (cold-formed HSS, end-plate, and weld metal), developed from
experimental TC tests, are shown in Fig. 5.13. These material properties were used for all connections
irrespective of geometry and branch type. The material properties were: yield stress of HSS (Fyb) = 421 MPa,
ultimate stress of HSS (Fu) = 501 MPa, yield stress of end-plate (Fyp) = 409 MPa, ultimate stress of end-plate
(Fup) = 566 MPa, yield stress of weld metal (Fyw) = 501 MPa, and ultimate stress of weld metal (FEXX) = 571
MPa. These values were chosen to provide a similar level of base metal and weld metal over-strength (the
difference between the actual and minimum specified values used in design), to represent the standard
assumption in connection design that matched electrodes are used. Outside and inside corner radii of the RHS
(ro and ri) were taken as 2.0tb and 1.0tb, respectively.
Fig. 5.13. Engineering stress-strain curves for the cold-formed HSS, end plate, and weld metal used in the FE
parametric study

Early on in the parametric study, it was found that significant yielding of the branch occurred in CHS-to-
rigid plate connections with tw/tb > 1.06, preventing the analyses from completing. The range of tw/tb was
therefore reduced to 0.35 ≤ tw/tb ≤ 1.06 for the CHS specimens. Tables 5.3 and 5.4 show the non-dimensional
connection parameters (Bb/tb or Db/tb, and tw/tb) for the θ = 90° models in the parametric study.
Table 5.3. Non-dimensional parameters and FE parametric study results (PFE/AwFEXX and failure mode) for
RHS-to-rigid plate connections
(Bb = 200 mm) tw/tb
tb Bb/tb
mm 0.35 0.50 0.71 0.90 1.06 1.41
4.00 50 0.80 W 0.76 W 0.72 W 0.65 W 0.65 W 0.59 W a
5.00 40 0.80 W 0.77 W 0.73 W 0.69 W 0.67 W 0.59 P a
6.67 30 0.82 W 0.77 W 0.74 W 0.70 W 0.67 P 0.56 P a
10.00 20 0.84 W 0.80 W 0.76 W 0.71 P 0.64 P 0.54 P a
16.00 12.5 0.87 W 0.83 W 0.72 P 0.63 P 0.56 P 0.49 P a
22.00 9.1 0.88 W 0.84 W 0.64 P 0.57 P 0.53 P -a
a
Connections noted thus experienced nominal branch yielding.
Note: W = rupture through the weld, around the entire branch perimeter; P = rupture only in the end-plate.
Table 5.4. Non-dimensional parameters and FE parametric study results (PFE/AwFEXX and failure mode) for
CHS-to-rigid plate connections
(Db = 168 mm) tw/tb
tb Db/tb
mm 0.35 0.50 0.71 0.90 1.06 1.41
3.36 50 0.85 W 0.87 W 0.81 W 0.76 W 0.73 W a -a
4.20 40 0.89 W 0.86 W 0.82 W 0.76 W 0.73 W a -a
5.59 30 0.88 W 0.88 W 0.82 W 0.78 W 0.74 W a -a
8.40 20 0.90 W 0.88 W 0.85 W 0.80 W a 0.76 W a -a
13.44 12.5 0.93 W 0.89 W 0.87 W 0.79 P a 0.71 P a -a
18.48 9.1 0.92 W 0.90 W 0.83 P 0.71 P 0.65 P a -a
a
Connections noted thus experienced nominal branch yielding.
Note: W = rupture through the weld, around the entire branch perimeter; P = rupture only in the end-plate.
5.5.1. RESULTS AND ANALYSIS

Weld rupture occurred in 22/35 FE tests on 90° RHS-to-rigid plate connections, and 25/30 FE tests on
90° CHS-to-rigid plate connections. In 21 and 20 of these tests, respectively, weld rupture occurred before
branch yielding. End-plate rupture occurred before weld rupture and branch yielding in nine of the RHS-to-
rigid plate FE tests, and two of the CHS-to-rigid plate FE tests. The non-dimensional average stress on the
weld throat at failure (PFE/AwFEXX), and the failure mode, are given in Tables 5.3 and 5.4. For the tests ending
in weld rupture, the factor PFE/AwFEXX relates the shear strength of the weld metal (Fnw) to the electrode
ultimate strength. In general, PFE/AwFEXX decreased as the RHS or CHS branch became more slender, and as
the weld became larger with respect to the branch thickness.

5.5.1.1. Geometric Limits for Weld Rupture

The FE RHS-to-rigid plate tests that failed by end-plate rupture, both before and after yielding of the
branch, had low values of Bb/tb and high values of tw/tb. It was determined that the ratio of tw/tb to Bb/tb was
always greater than 0.035. For the CHS-to-rigid plate tests that failed in the same manner, the ratio of tw/tb to
Db/tb was always greater than about twice that value, 0.072. It was thus deduced that:
For weld rupture to govern over plate rupture in RHS-to-rigid plate connections:
tw
tb t
 w  0.035
Bb Bb (5.4)
tb
For weld rupture to govern over plate rupture in CHS-to-rigid plate connections:
tw
tb t
 w  0.072 (5.5)
Db Db
tb
When tw/Bb > 0.035 or tw/Db > 0.072, plate rupture can be expected to govern. The use of non-
dimensional parameters to study the behaviour of HSS connections is a common practice which allows
results to be generalized for all sizes of HSS; thus, Eqs. (5.4) and (5.5), and the remainder of the results, are
believed to be valid for smaller or larger sizes of Bb or Db than those selected in this study.
All of the end-plate ruptures in the FE RHS-to-rigid plate tests were observed at mid-wall of the RHS, a
critical location validated by experimental tests (Frater 1986). Furthermore, the values of tw/Bb for the
experimental RHS-to-rigid plate tests that failed by end-plate rupture satisfied Eq. (5.4). Since end-plate
rupture did not occur in any of the experimental CHS-to-rigid plate tests, more work is necessary to validate
Eq. (5.5).
5.5.1.2. Effect of Weld Size

Fig. 5.14 illustrates the effect of tw/tb on the weld strength, given by the ratios PFE/AwFEXX and Pa/AwFEXX.
For both RHS- and CHS-to-rigid plate tests, as tw/tb increases (for constant values of Bb/tb and Db/tb),
PFE/AwFEXX decreases. The magnitude of the effect of tw/tb on the weld strength was found to be similar for
both the RHS- and CHS-to-rigid plate tests; however, the welds in the CHS-to-rigid plate tests generally
exhibited a higher average strength. In general, as tw/tb increases, the eccentricity (equal to the distance from
the middle of the branch wall to the centroid of the fillet weld) increases, and the moment transmitted about

the longitudinal axis of the weld increases. This moment causes bending of the branch wall inward and
tension near the weld root, which is well-known to be a situation that can lead to premature weld failure. This
detrimental feature of “single-sided fillet welds” is inherent to all HSS fillet-welded joints. Fig. 5.15, from FE
analysis, illustrates this rotation of the branch wall due to eccentricity in two fillet-welded RHS joints. In Fig.
5.15, the von Mises equivalent stress is overlaid atop the deformed joint at failure. This effect is greater for
RHS than CHS, which have less restraint from rotation along the flats, especially for slender sections.
Fig. 5.14. Effect of the ratio tw/tb on fillet weld strength in θ = 90° HSS-to-rigid plate connections: (a) for
RHS branches; (b) for CHS branches

Fig. 5.15. Examples of branch rotation in fillet-welded joints to RHS
5.5.1.3. Effect of Branch Slenderness

Fig. 5.16 illustrates the effect of Bb/tb and Db/tb on PFE/AwFEXX. As Bb/tb and Db/tb increases (i.e. the
branch became more slender), the strength of joints that failed by weld rupture slightly decreases. This effect
is more pronounced for the RHS-to-rigid plate tests than the CHS-to-rigid plate tests. At low values of Bb/tb,
the fillet-weld strength in the RHS-to-rigid plate tests approaches a similar strength to that of a fillet weld in
CHS-to-rigid plate tests with low to moderate values of Db/tb, corroborating the presence of local stiffness
effects in RHS branches. The non-uniform load distribution patterns for RHS can be inferred from Fig. D.4 in
Appendix D.1.
Fig. 5.16. Effect of the ratios Bb/tb and Db/tb on fillet weld strength in θ = 90° HSS-to-rigid plate connections:
(a) for RHS branches; (b) for CHS branches

Connections with slender branches and high values of tw/tb, especially RHS connections, exhibited
significant rotations of the branch wall adjacent to the weld inwards at failure, as shown previously in Fig.
5.15. When this rotation was small (i.e. branches were stocky, and tw/tb was small), PFE/AwFEXX was closer to
the value predicted using the sinθ factor, thus illustrating that fillet-welded RHS- and CHS-to-rigid plate
connections, and other single-sided fillet welded connections, merit special attention. This is effect is also
believed to explain the higher strength of CHS connections compared to RHS connections.
5.5.1.4. Effect of Branch Inclination Angle

Fig. 5.17 illustrates PFE/AwFEXX as a function of θ for θ = 60°, 75° and 90°, and two different values of the
branch wall slenderness. Although the absolute strength of the weld increases due to the longer weld length
as θ varies from 90° to 60°, the effect on PFE/AwFEXX is negligible. The curves predicted using the sinθ factor
in AISC 360 (2016) for the average loading angle of welds in RHS- and CHS-to-rigid plate connections with
various branch inclination angles are also plotted in Fig. 5.17. Note that these equations neglect any
provisions for deformation compatibly. For 60° ≤ θ ≤ 90° in an HSS joint, this reduction is less than 4% using
the Mw factor in CSA S16 (2014) with the average loading angle of the weld, and is hence small. While the
sinθ factor predicts a reduction in PFE/AwFEXX as θ decreases from 90° to 60°, Fig. 5.17 illustrates that this is
not always the case. Therefore, the complex effect of θ on the fillet weld strength may reasonably be ignored
for weld joints to HSS. Moreover, the weld strength (per unit length) may be considered independent of the
branch angle.
Fig. 5.17. Effect of θ on fillet weld strength in HSS-to-rigid plate connections

5.6. PROPOSED ALTERNATE MODEL FOR FILLET WELD STRENGTH

It can be seen from the previous comparisons that PFE/AwFEXX in HSS-to-rigid plate connections is non-
constant, and that it is a function of both tw/tb and Bb/tb (or Db/tb). Linear regression analyses were hence
performed on the results of the FE RHS- and CHS-to-rigid plate tests to develop alternate, yet still simple,
equations for predicting the strength of fillet welds therein. Only the data from FE tests that failed by weld
rupture before branch yielding was used. The following linear equations for the nominal strength of fillet
welds (Pnw) to HSS were determined:
 B   t 
For RHS: Pnw  0.954  0.00193  b   0.210  w   Aw FEXX (5.6)
  tb   tb  
 D   t 
For CHS: Pnw  1.009  0.00137  b   0.197  w   Aw FEXX (5.7)
  tb   tb  
which provide R2 values of 0.97 and 0.93, respectively, and mean actual-to-predicted (A/P) ratios and COVs
of 1.00 and 0.02 (both are the same) when compared to the finite element results.
It is also possible to neglect the effect of branch wall slenderness and still produce a sufficiently accurate
equation for Pnw, using linear regression, as follows:
  t 
For RHS: Pnw  0.924  0.262  w   Aw FEXX (5.8)
  tb  
  t 
For CHS: Pnw  0.984  0.226  w   Aw FEXX (5.9)
  tb  
Eqs. (5.8) and (5.9) provide R2 values of 0.83 and 0.77, mean A/P ratios of 1.02 and 1.00, and COVs of
0.04 and 0.03, respectively, when compared to the finite element results. The correlations of Eqs. (5.8) and
(5.9) against both finite element and experimental results are shown as dashed lines in Fig. 5.14 (presented
earlier). While these are simple equations for predicting the strength of fillet welds to HSS, they are not well-
suited for design because they are quadratic with respect to tw (since Aw = twlw). By simplifying the constants
in Eqs. (5.8) and (5.9), and substituting Pr/Py for tw/tb (where Pr = required weld strength, in units of force, and
Py = yield force of the branch), Eqs. (5.8) and (5.9) can be rewritten as:

 P 
For RHS: Pnw  0.90  0.25  r   Aw FEXX (5.10)

  Py  
 P 
For CHS: Pnw  1.00  0.25  r   Aw FEXX (5.11)
  Py
  
In a worst-case scenario, the required weld strength Pr will equal the weld strength (Pa or PFE), which
becomes disproportionately larger as tw increases. The branch yield load Py is linearly related to tb. By
comparing the relationship between the plotted points and the diagonal lines on Figs. 5.14 and 5.18, it can be
seen that substituting tw/tb with Pr/Py still provides a good prediction of the weld rupture load. Importantly, it
also allows a solution to be found for tw using a linear equation. Graphical comparisons of Pa/Py and PFE/Py to
tw/tb are provided in Appendix D.1.
Comparing Eqs. (5.10) and (5.11) to the FE results yields mean A/P ratios of 1.03 (both are the same),
and COVs of 0.04 and 0.05, respectively. Comparing the same equations to the weld-critical experimental
results yields mean A/P values of 1.08 and 1.04, and COVs of 0.17 and 0.10, respectively. The correlation of
Eq. (5.10) against the FE and experimental RHS results, and the correlation of Eq. (5.11) against the FE and
experimental CHS results, is shown in Fig. 5.18.

Fig. 5.18. Evaluation of design equations against all 90° (FE and experimental) HSS-to-rigid plate test
results: (a) for RHS branches; (b) for CHS branches
5.7. RELIABILITY ANALYSIS OF FE RESULTS FOR DESIGN CODES

Reliability analyses, taking into account the mean values and variations in the material, geometric and
professional factors in the manner discussed in Section 2.6.2, were performed for the following design
procedures:
(a) AISC 360 (2016) and CSA S16 (2014) (with and without the sinθ factor);
(b) EN 1993-1-8 (CEN 2005) Directional Method and Simplified Method; and

(c) the alternate method given by Eqs. (5.10) and (5.11).
The predicted nominal weld strengths (Pnw) for each of the θ = 90° connections that failed by weld
rupture were calculated as described in Sections 4.4.1 (for AISC) 4.4.2 (for CSA) and 4.4.4 (for EN1993-1-
8), and are given in Appendix D.2.
Because the product 0.67(ϕ=0.67) = 0.60(ϕ=0.75), the predicted weld resistances – and, therefore, the
resulting reliability index – are identical for AISC 360 (AISC 2016) and CSA S16 (CSA 2014), both with and
+
without the sinθ factor. The following calculations for are therefore presented only for AISC 360 (AISC
2016) with respect to (a) above.
Tables 5.5 and 5.6 summarize the key parameters of the reliability analysis. In both tables, Eqs. (5.10) or
(5.11) have been assessed using ϕ = 0.75 and 0.67, to determine their reliability when used with the resistance
factors for weld metal in AISC 360 (AISC 2016) and CSA S16 (CSA 2014), respectively.
Table 5.5. Reliability analysis parameters with respect to 21 weld-critical RHS-to-rigid plate FE test results
with θ = 90°
AISC 360-16 EN1993-1-8:2005 Eq. (5.10)
0.60 0.60(1.0+0.50sin1.5θ) Directional Method Simplified Method ϕ = 0.75 ϕ = 0.67
ϕ 0.75 0.75 1/ M2 = 0.80 1/ M2 = 0.80 0.75 0.67
ρM 1.12 1.12 1.12 1.12 1.12 1.12
VM 0.12 0.12 0.12 0.12 0.12 0.12
ρG 1.03 1.03 1.03 1.03 1.03 1.03
VG 0.10 0.10 0.10 0.10 0.10 0.10
ρP 1.27 0.85 1.11 1.36 1.03 1.03
VP 0.09 0.09 0.09 0.09 0.05 0.05
ρR 1.47 0.98 1.28 1.57 1.19 1.19
VR 0.18 0.18 0.18 0.18 0.16 0.16
ϕ+ 0.84 1.02 0.92 0.84 0.91 0.86
+
5.00 a 2.85 b 3.90 4.99 4.05 4.67
a +
identical to CSA S16-14 method without the sinθ factor.
b +
identical to CSA S16-14 method with the sinθ factor.
Table 5.6. Reliability analysis parameters with respect to 20 weld-critical CHS-to-rigid plate FE test results θ
= 90°
AISC 360-16 EN1993-1-8:2005 Eq. (5.11)
0.60 0.60(1.0+0.50sin1.5θ) Directional Method Simplified Method ϕ = 0.75 ϕ = 0.67
ϕ 0.75 0.75 0.80 0.80 0.75 0.67
ρM 1.12 1.12 1.12 1.12 1.12 1.12
VM 0.12 0.12 0.12 0.12 0.12 0.12
ρG 1.03 1.03 1.03 1.03 1.03 1.03
VG 0.10 0.10 0.10 0.10 0.10 0.10
ρP 1.43 0.95 1.24 1.52 1.03 1.03
VP 0.06 0.06 0.06 0.06 0.04 0.04
ρR 1.65 1.10 1.43 1.75 1.19 1.19
VR 0.17 0.17 0.17 0.17 0.16 0.16
ϕ+ 0.78 0.95 0.86 0.78 0.91 0.86
+
5.87 a 3.57 b 4.69 5.87 4.09 4.72
a +
identical to CSA S16-14 method without the sinθ factor.
b +
identical to CSA S16-14 method with the sinθ factor.

By examining the -values in Tables 5.5 and 5.6, it can be seen that the sinθ factor is still unsafe
(provides less than the target value) when applied to fillet welds in HSS-to-rigid plate connections,
irrespective of whether the branch is an RHS or CHS member, and even when a wider range of connection
parameters (weld size and branch slenderness) is considered. By examining ρP and +
in both tables, it can be
seen that Eqs. (5.10) and (5.11) provide accurate (ρP close 1) and safe predictions ( +
≥ 4.0 and 4.5) of the
fillet weld ultimate strength in RHS- and CHS-to-rigid plate connections.
5.8. FILLET WELD SIZE TO DEVELOP HSS BRANCH YIELD STRENGTH

Eqs. (5.10) and (5.11) were used to determine the weld size required to develop the yield strength of a
connected RHS or CHS branch member – a highly-debated topic internationally, and among code
committees. In order to solve for tw as a function of the branch wall thickness (tb) using Eqs. (5.10) and
(5.11), it is necessary to assume – conservatively – that the branch cross-sectional area (Ab) is equal to the
product of the branch perimeter (i.e. the weld length, lw) and tb. By setting the nominal weld strength Pnw and
the required weld strength Pr equal to Py, it can be shown that for an RHS branch:
1  Fyb 
tw    tb (5.12)
0.65  FEXX 
will develop the branch yield strength. Therefore, for a 350W RHS branch member (Fyb = 350 MPa) welded
all around to a plate with a matched electrode (FEXX = 490 MPa):
tw  1.10tb (5.13)
Similarly, for CHS members:
1  Fyb  (5.14)
tw    tb
0.75  FEXX 
(5.15)
tw  0.95tb
These values are shown plotted on Fig. D.5 in Appendix D.

To achieve the necessary level of reliability in North America, the values of 1.10tb and 0.95tb should be
multiplied by the ratio of the resistance factor for branch yielding to the resistance factor for the weld metal
[i.e. 0.90/0.75 = 1.20 in AISC 360 (AISC 2016), and 0.90/0.67 =1.34 in CSA S16 (CSA 2014)].

A more reasonable estimate for inclined branches (θ < 90°) can be obtained by taking into account the
longer weld length due to the branch angle, for example, by dividing Eq. (5.15) for CHS branches by the
weld length factor (Ka) given by the American Welding Society (AWS 2015):
1  1 / sin 
Ka  (5.16)
2

5.9. SUMMARY
Finite element models were developed for fillet-welded RHS-to-rigid plate and CHS-to-rigid plate
connections and these were validated against 33 laboratory tests on such welded joints. A parametric study
was then performed in which a total of 73 FE RHS- and CHS-to-rigid plate connections were tested to failure
under axial tension loading. The parametric study investigated the effect of the ratio of the weld throat
dimension to the branch wall thickness (tw/tb), the branch wall slenderness (Db/tb and Bb/tb) and the branch
inclination angle (θ) on the weld strength. It was found that:
1. As tw/tb increases, the average stress on the weld throat area at failure significantly decreases.
2. As Db/tb and Bb/tb increase, the average stress on the weld throat area at failure slightly decreases.
3. The branch inclination angle (θ) has a negligible effect on the weld strength per unit throat area;
however, the longer weld length that results from a reduction in branch angle increases the
absolute strength of the weld.
Equations were then developed to estimate the strength of fillet welds in RHS- and CHS-to-rigid plate
connections. When subjected to a reliability analysis with respect to the weld-critical FE results, these
equations were found to provide an adequate level of safety for use in North America. It is shown, using these
equations, that a weld throat dimension equal to 1.10tb (for an RHS branch) and 0.95tb (for a CHS branch)
will develop the yield strength of the connected branch member at 90° to a rigid plate, for 350 MPa yield-
strength HSS with matching electrode. It was hence found that welds in RHS-to-rigid plate connections were
generally not as strong as welds in CHS-to-rigid plate connections (per unit length) due to non-uniform
membrane stresses parallel to the branch, which peak at the RHS corners, and less restraint from local
rotations in RHS flats versus curved CHS walls.
The design methods for fillet welds to HSS members given in AISC 360 (AISC 2016), CSA S16 (CSA
2014), and EN1993-1-8 (CEN 2005) were evaluated with respect to North American safety index
requirements, looking at fillet welds in RHS- and CHS-to-rigid plate connections independently, and using a
reliability analysis that included the mean values and variations in material, geometric, and professional
factors. The sinθ factor in AISC 360 (AISC 2016) and CSA S16 (CSA 2014) was found to provide
inadequate levels of the safety index when used to design fillet welds to HSS.
For all HSS connections, including HSS-to-HSS connections where the effective length concept is used,
and even HSS connections in which the welds are fully effective, it is therefore recommended that the
provisions of AISC 360 (AISC 2016) and CSA S16 (CSA 2014) for the design of fillet welds be used without
the sinθ factor [i.e. taking θ = 0° in the term (1.00 + 0.50sin1.5θ)], or that the alternate method, developed
herein, be used with North American resistance factors.

Chapter 6: FILLET WELD EFFECTIVE LENGTHS IN CHS X-
6.1. INTRODUCTION
When welding to hollow structural sections (HSS), welds can be proportioned: (a) to achieve the capacity
of the connected branch member wall; or (b) to be “fit for purpose” (ISO 2013). By designing welds as fit for
purpose – to resist the actual forces present in the branch member – smaller, more appropriate weld sizes
typically result.
In order to account for the non-uniform loading of the weld perimeter due to differences in the relative
flexibilities of the chord loaded normal to its surface, and the branch(es) carrying membrane stresses parallel
to its surface, weld effective properties – including weld effective lengths and weld effective section moduli –
are used. These properties are determined by discounting segments of the weld which do not contribute to its
overall resistance.
Over the last 30 years, much research has been conducted at the University of Toronto to determine weld
effective lengths for rectangular hollow section (RHS) connections, including gapped K-connections, T-, Y-
and X- (or Cross-) connections, moment-loaded T-connections, and overlapped K-connections (Frater &
Packer 1992a, 1992b; Packer & Cassidy 1995; McFadden & Packer 2014; and the work of Chapter 2).
Recommendations based on this research have been adopted as code in North America, by the American
Institute of Steel Construction (AISC) in Section K5: Welds of Plates and Branches to RHS of their latest
specification, AISC 360-16.
Since the addition of Section K5 (formerly Section K4, in the 2010 specification), weld effective
properties for circular hollow section (CHS) connections have been an issue faced by code writers, including
AISC and the American Welding Society (AWS), since load transfer around a welded CHS joint can be
highly non-uniform (Marshall 1992) (e.g. Fig. 6.1).
93
Chapter 6: Fillet Weld Effective Lengths in CHS X-Connections: Experimentation 94
Fig. 6.1. Variation of X-connection stress distribution
While AISC 360-16 is noticeably silent regarding weld effective lengths for CHS connections, AWS
D1.1 Structural Welding Code – Steel (AWS 2015) implies, in Clause 9.6.1.3(4), that the weld effective
length in axially loaded CHS connections is equal to 1/1.5 of the total weld length under factored loads,
regardless of the joint geometry. While believed to be conservative, this rule is not supported by experimental
evidence.
A laboratory-based test program was hence conducted to assess the performance of welds in CHS
connections. For the first time, weld-critical tests (where failure occurs by weld fracture) have been
completed on fillet welds in full-scale CHS connections, and the structural reliability (safety index) of the
existing AWS, AISC and CSA specification provisions for the design of such welds is determined. The effect
of key connection parameters on the weld strength is also investigated, and an empirical method to quantify
the weld effective length is proposed. The results of this chapter are vital for determining a strategy for the
fit-for-purpose design of welds in CHS connections is both accurate (reasonably predicts the correct failure
load) and safe (meets or marginally exceeds target reliability indices provided by design codes).
6.2. TEST SPECIMEN PREPARATION AND MATERIAL TESTING

Six CHS X-connections were designed and fabricated from ASTM A500 (ASTM 2013) dual-certified
Grade B/C cold-formed CHS, and fillet welded using a semi-automatic flux-cored arc welding (FCAW)
process with full CO2 shielding gas. As it was speculated that the strength of welds in CHS X-connections

depends on branch-to-chord diameter ratio ( = Db/D), chord wall slenderness (D/t), and branch inclination
angle (θ), the chord and branch members were selected to cover a wide range of these values, as shown in
Table 6.1. These values were selected to be within limits for fillet welds to develop the full throat thickness,
as given by AWS D1.1-15 (see Section 6.2.1). Four connections had branches at θ = 90° to the chord, and
two connections had branches at θ = 60° to the chord. The connection layout is shown in Fig. 6.2, which
includes definitions for the nomenclature used herein. Appendix E.1 provides specimen fabrication drawings.
Table 6.1. Measured properties of 12 CHS X- (test) connections

Test a CHS branch member CHS chord member θ D/t Pa d Paꞌ e
Db × tb Ab b Fyb c
D×t Ab Fy c
mm × mm mm2 MPa mm × mm mm2 MPa ° kN kN

102-273-90a 102.0 × 7.34 2161 373 273.5 × 11.69 9614 460 0.37 23.4 0.63 672 672
102-273-90b 102.0 × 7.34 2161 373 273.5 × 11.69 9614 460 0.37 23.4 0.63 678 678
102-406-90a 102.0 × 7.34 2161 373 406.5 × 12.34 15283 355 0.25 32.9 0.59 608 608
102-406-90b 102.0 × 7.34 2161 373 406.5 × 12.34 15283 355 0.25 32.9 0.59 540 540
90
127-273-90a 127.4 × 11.55 4207 431 273.5 × 11.69 9614 460 0.47 23.4 0.99 653 653
127-273-90b 127.4 × 11.55 4207 431 273.5 × 11.69 9614 460 0.47 23.4 0.99 609 653
127-406-90a 127.4 × 11.55 4207 431 406.5 × 12.34 15283 355 0.31 32.9 0.94 557 557
127-406-90b 127.4 × 11.55 4207 431 406.5 × 12.34 15283 355 0.31 32.9 0.94 556 557
102-406-60a 102.0 × 7.34 2161 373 410.0 × 12.21 15260 373 0.25 33.6 0.60 721 721
102-406-60b 102.0 × 7.34 2161 373 410.0 × 12.21 15260 373 0.25 33.6 0.60 538 721
60
127-406-60a 127.4 × 11.55 4207 431 410.0 × 12.21 15260 373 0.31 33.6 0.95 761 761
127-406-60b 127.4 × 11.55 4207 431 410.0 × 12.21 15260 373 0.31 33.6 0.95 798 850
a
In the test designation: the first number represents the nominal branch diameter; the second number represents the
nominal chord diameter; the third number represents the branch inclination angle (θ); and a/b represent the side of the
connection, since each connection had two fillet welds (a = top; b = bottom).
b
Cross-sectional areas determined by cutting a prescribed length of CHS, weighing it, and then using a density of 7850
kg/m3 (CISC 2016) to calculate its cross-sectional area.
c
Yield strength of all CHS determined from tensile coupon tests performed according to ASTM A370 (2017) while
maintaining the curved shape.
d
Actual (experimental) weld fracture load.
e
Greatest load sustained by the weld.

Fig. 6.2. Connection layout and nomenclature
6.2.1. CONNECTION GEOMETRIC CONSIDERATIONS

CHS members and connection geometry had to be carefully designed, to maintain the local dihedral angle
(Ѱ) of the joints between 60° and 120°, to develop the full fillet weld throat thickness (tw). According to Note
4 in Fig. 9.10 of AWS D1.1-15, when Ѱ < 60° the Z loss values in Table 9.5 of AWS D1.1-15, for PJP welds,
apply. To keep Ѱ within this range, the complex effect of and θ on Ѱ, which changes continuously around
the joint, was studied using a vector-calculus approach given by Luyties & Post (1988). Using a subtended
angle increment (Δρ) equal to 1°, as suggested by Luyties & Post (1988), it was determined that must not
exceed 0.50 for 90° CHS connections and 0.28 for 60° CHS X-connections to maintain 60° ≤ Ѱ ≤ 120°.
While having θ less than 60°, and thus Ѱ less than 60°, would adversely affect the weld strength by
contributing to the Z loss (loss of the weld throat) at the root of the weld, according to Table 9.5 of AWS
D1.1-15, having slightly larger -values, and thus Ѱ slightly greater than 120° does not. For the design of test
connections it was therefore deemed necessary to keep θ between 60° and 90°, while a minor deviation from
the stated limits was considered acceptable. Local dihedral angle curves for each test connection showing
the variation of Ѱ along the weld length are shown in Fig. 6.3. These curves have been determined by
applying the method given by Luyties & Post (1988) in Matlab, with Δρ equal to 1° and measured values of
Db and D. The letters “a” and “b” at the end of the test designation in Table 6.1 have been omitted since the
curves apply to both branch connections on either side of the chord (i.e. “a” and “b”). Local dihedral angle

curves for connections with different values, calculated using the same method, have also been included in
Fig. 6.3. These curves can be compared to those given in Informative Annex O of AWS (2015) to verify the
approach used.
(a) θ = 90° connections (b) θ = 60° connections
Fig. 6.3. Local dihedral angle curves for test joints, with subtended angle measured from the crown heel
The branches were cut to a minimum branch length (lb) of 6Db, to avoid shear lag effects at mid-length,
from both ends (Mehrota & Govil 1972), and profiled to saddle perfectly onto the chords, without edge
bevelling (Fig. 6.4). The branches were capped by a tee connection through which load was applied. The tee
connection was designed, using Section 7.6 of Packer & Henderson (1997), to develop the member capacity.
The chords were cut to an overall chord length (l) to avoid end effects at the connection (van der Vegte &
Makino 2010). To economize on material, they were left unrestrained (uncapped) at both ends. The average
measured material properties for the CHS branch and chord members were determined from three tensile
coupon tests per section, performed according to ASTM A370 (ASTM 2017) while maintaining the curved
shape. One tensile coupon was taken from each CHS directly across from the weld seam, and the other two
were taken from the CHS faces orthogonal to the weld seam.

(a) Typical 90° connection (shown for 102-273-90a) (b) Typical 60° connection (shown for 102-406-60b)
Fig. 6.4. Fit-up of branch to chord after profiling and tack welding
6.2.2. GEOMETRIC PROPERTIES OF THE AS-LAID WELDS

Correct measurement of the geometric properties of the welds, which comprise a complex saddle shape in
CHS connections, is critical to the subsequent scientific analysis of the weld strength; hence, great care was
taken to very accurately obtain these measurements.
The total weld length (lw), and the weld length tributary to each throat size measurement (which is
necessary to determine the average throat size for the joint), were calculated by modifying the vector-calculus
approach used previously to determine Ѱ by Luyties & Post (1988) to give a near-perfect solution for the
distance between points along the weld root, and then summing up these distances. This can be done using
computer-programmed equations (e.g. in Matlab) as follows:
Step 1: Starting at a subtended angle (ρ) equal to 0° (i.e. the heel of the connection, Fig. 6.5a) (or the
beginning of the interval of interest), compute the coordinates of the branch-chord intersection at ρ and ρ +
Δρ using Eq. (6.1a). Eq. (6.1a) gives the position vector 𝑃⃗𝑖 at point i along the branch-chord intersection,
where i equals ρ or ρ + Δρ. The notation [( ), ( ), ( )] represents the three vector components in the branch
coordinate system (Fig. 6.5b).
 D  D 
Pi   lt  ,  b sin i  ,  b cos i   (6.1a)
  2   2 
where:

D (1  cos i) D  D   Db sin i 
2 2
(6.1b)
lt  b 
2 tan  2sin 
Eq. (6.1b) is derived in Appendix E.2.

⃗ ) connecting these points: 𝑉
Step 2: Calculate the vector (𝑉 ⃗ = 𝑃⃗𝜌+𝛥𝜌 – 𝑃⃗𝜌 .
⃗ by taking the square root of the sum of squares of its three vector
Step 3: Compute the magnitude of 𝑉
components. This is the approximation to the weld length between 𝑃⃗𝜌 and 𝑃⃗𝜌+𝛥𝜌 . The smaller Δρ is, the closer
the approximation will be to the actual weld length.
⃗ to the
Step 4: Increment ρ by Δρ, and repeat Steps 1 to 3, adding the new result for the magnitude 𝑉
previous results. For the total weld length, continue repeating Steps 1 to 3 until ρ is equal to 360° - Δρ. For an
interval length along the weld, continue repeating Steps 1 to 3 until ρ is exactly Δρ less than ρ at the end of
the interval.
The weld lengths herein were calculated as described above with Δρ equal to 1°. This generally gave the
same answer for total weld length as an exact solution based on calculus, but could be more easily applied to
a range of different joint geometries. Lie et al. (2001) also provided equations to describe the geometry of
butt welds in HSS Y-connections. If used to calculate Ѱ or lw for the current tests, the results would match the
Authors’ because both approaches are based strictly on the connection geometry.

(a) scalar and vector parameters
(b) branch coordinate system
Fig. 6.5. Vector calculus method used to determine weld lengths
After being laid, welds were ground into a triangular shape, with a near-uniform throat size (tw) around
each joint, and flat weld faces. Flat weld faces allowed tw to be obtained from a 3D model of the weld’s exact
geometry, as shown in Fig. 6.6. Using this approach, the orientation of tw and the weld legs (lv and lh) must be
established correctly: in the plane of Ѱ, perpendicular to the weld root, between tangents to the outside
surfaces of the branch and the chord. The computer-aided design (CAD) program Solidworks was employed
to exact this requirement.
First, local components of lv and lh in a plane containing the branch axis and the normal to the branch face
were measured at uniform increments of ρ along the weld length. The component of lv parallel to the branch
was first measured by wrapping a mat board collar with reinforced edges around the branch at a fixed

distance (or “offset”) from the root of the weld at the heel of the connection. The offset distance (Fig. 6.5a)
was then measured using a Mitutoyo Digimatic calliper (with a specified resolution of 0.01 mm). The
distance between the collar and the weld toe along the branch (x) was then measured, using the same calliper,
at uniform increments of ρ along the weld length. The component of lv parallel to the branch at ρ could then
be calculated by subtracting this measured distance (x) from the theoretical distance between the collar and
the weld root (= lt + offset). Historically, lt refers to the template length, which is the length (parallel to the
branch) of a steel-cutting template that was wrapped around a CHS branch and used as a guide for profiling
with a torch. The component of lh perpendicular to the face of the branch at ρ was measured by laying a
standard fillet weld gauge along the axis of the branch and measuring the distance to the weld toe. The weld
profile around the entire joint was then modelled in Solidworks using these measurements and the measured
values of Db and D. Finally, sections were taken through the weld in the plane of Ѱ using Solidworks
geometry tools, and lv, lh and tw were precisely measured, as shown in Fig. 6.6.
The weld area (Aw) was determined by summing up: tw × tributary weld length (Fig. 6.6) around the entire
joint. The measured fillet-weld geometric properties are shown in Table 6.2, and the individual measurements
along each weld, made externally and with Solidworks, are provided in Appendix E.3. Appendix E.4
provides the 3D Solidworks models of the weld profiles generated for all 12 test joints.
Fig. 6.6. 3D Solidworks models of weld profile and weld dimensions

Table 6.2. Weld dimensions and predicted fracture loads for test joints according to existing AWS D1.1-15
provisions for weld effective lengths in CHS X-connections
Test Average measured weld dimensions Pnwa
lv lh tw lw Aw
mm mm mm mm mm kN
102-273-90a 6.86 6.17 4.08 322 1312 303
102-273-90b 7.23 6.65 4.37 322 1405 324
102-406-90a 5.16 5.78 3.56 320 1139 263
102-406-90b 4.54 5.08 3.14 320 1004 232
127-273-90a 5.94 5.93 3.63 406 1475 340
127-273-90b 7.05 6.06 4.00 406 1625 375
127-406-90a 4.83 5.03 3.16 403 1273 294
127-406-90b 5.60 5.19 3.47 403 1410 323
102-406-60a 5.83 5.59 3.58 345 1235 285
102-406-60b 6.29 5.83 3.79 345 1307 302
127-406-60a 5.68 8.01 3.95 434 1716 396
127-406-60b 5.39 6.00 3.38 434 1468 339
a
Nominal predicted fracture load according to existing AWS (2015) specification provisions, calculated using Eqs.
(6.2a,b) and (6.3), using Aw and FEXX determined from tensile coupon tests (= 577 MPa).
6.2.2.1. Post-Rupture Macro-Etch Examinations

To verify the values of lv and lh obtained using Solidworks, post-rupture macro-etch examinations of the
fillet welds were performed after several tests. Four cross sections of the fillet weld profile (at ρ = 0°, 90°,
180°, and 270°) were cut in the plane of Ѱ using a drop saw. The cross sections were then hand polished,
macro-etched using a 5% nital etchant solution, and digitized at a scale of 1:1. Using the program AutoCAD,
lv, lh, and tw were re-measured from the cross sections. The weld throat (tw) was taken as the shortest distance
from the root to the face of the diagrammatic weld. The average value of tw for the macro-etched connections
was then determined, using a tributary width for each measurement equal to 0.25lw, and compared to the
previous measurements. The mean ratio of the measurements (Solidworks/macro-etch) was found to be 0.96.
The individual macro-etch measurements and their comparison to the experimental results are provided in
Appendix E.5.
It may be noted in Table E.15 (Appendix E.5) that the ratio Solidworks/macro-etch for test 127-406-60b
(0.78) is much lower than the average reported above (0.96). This is due to the difference in tw values
measured at the heel of the connection (ρ = 0°), where external measurement is made difficult by the small
angle between the adjoining base metals (Ѱ = 60°). The macro-etch values at ρ = 90°, 180°, and 270° (all
other locations where macro-etching was performed) showed good agreement, which gave credence to the
Solidworks-based dimensions derived from external calliper measurements, and the use of them for all tests
herein.

6.2.3. MECHANICAL PROPERTIES OF THE AS-LAID WELDS

The mechanical properties of the as-laid welds were determined by tensile coupon tests in accordance
with ASTM A370 (ASTM 2017). A summary of the all-weld-metal tensile coupon test results is given in
Table 6.3.
The average yield stress from three coupon tests (by the 0.2% strain offset method) was 517 MPa and the
average ultimate stress (FEXX) was 577 MPa with 28.1% elongation at rupture. The measured ultimate
strength was 17.8% greater than the specified nominal strength (490 MPa) of the E71T-1C electrode used.
The welding process specifications used for the joints were: voltage = 25 V, amperage = 260 A, and travel
speed = 230 mm/min. Welding was performed principally in the flat position, by a welder qualified with the
Canadian Welding Bureau for the position and the FCAW process used. The tensile coupon test results for
the weld metal, as well as the results for the CHS material used, are provided in Appendix E.6.
Table 6.3. All-weld-metal tensile coupon test results for CHS X-connection tests
All-weld-metal coupon Yield stress Young’s modulus, FEXX Rupture strain,
designation E εrupa
MPa MPa MPa %
[i] 510 189000 573 30.6
[ii] 520 201000 576 27.4
[iii] 521 235000 581 26.4
Average 517 208000 577 28.1
a
Rupture strain determined by re-joining the fractured coupon and measuring: change in gauge length / initial gauge
length
6.3. CONNECTION TESTS AND INSTRUMENTATION

Quasi-static axial tension load was applied to the end of each branch on either side of the connection, and
hence to the weld, by a 2700-kN capacity universal testing machine (UTM). The typical testing arrangement
is shown in Fig. 6.7.

Fig. 6.7. Typical testing arrangement (shown for test 127-273-90a)
Four linear strain gauges (SGs), equally spaced around the perimeter of the branch at mid length (≥ 3Db
from the welded test joint and the end), and oriented along its longitudinal axis, were used to measure the
uniformity of load being applied to the branch. Equal strains were typically measured at all four locations
over all tests, indicating that predominantly axial load was applied.
Seven additional SGs, with the same orientation, were used around half the weld perimeter (i.e. on one
side of the branch only, due to symmetry) to measure non-uniform loading of the weld perimeter (Fig. 6.8).
For this purpose, SGs were centred 20 mm away from the weld toe, to avoid stress concentrations that occur
there due to the notch effect (Packer & Cassidy 1995), and located at ρ = 0°, 30°, 60°, 90°, 120°, 150°, and
180°. A single SG in the saddle position on the opposite side (at ρ = 270°) was used to verify symmetry of the
strain distribution about the plane of the connection. In all, 12 welds were tested to rupture (two per
connection). All welds failed in a brittle manner, by fracture along a plane through the weld. Four typical
failures are shown in Fig. 6.9, for both 90° connections (Figs. 6.9a,b) and 60° connections (Figs. 6.9c,d).
Additional photographs of the test set-up and failure mode for each test are provided in Appendix E.7.

Fig. 6.8. Strain gauges near weld toe (and weld fracture) in test 127-273-90a
After the first test weld (e.g. side a) ruptured in each connection, the branch was re-positioned within the
UTM and tack welded back in place. The entire connection was then removed from the UTM, and fully re-
welded (nominally in the flat position) to ensure that separation of the same branch did not occur again. The
connection was re-placed in the UTM, and tested until rupture of the second test weld (e.g. side b) occurred.
Chord deformation ( ) was continuously monitored throughout both tests with three LED targets: one on each
branch, 50-mm above the crown; and one at the connection work point on the chord face parallel to the plane
of the connection (Figs. 6.7, 6.10). The value of , which is defined as the outward displacement, normal to
the chord, of a single branch from the chord centreline (Packer et al. 2012), was taken as the normal
component of half of the displacement between the LEDs on each branch (Fig. 6.10). It therefore represents
the average deformation on both sides of the connection. The actual weld fracture loads (and hence Pa and
Paꞌ, given in Table 6.1) were obtained from load cells in-line with the UTM actuator, and verified by
comparison with forces computed from average SG readings of strain at mid-length of the branch and the
measured branch cross-sectional area (Ab) and Young’s modulus (E).

(a) Test 102-273-90a (θ = 90°) (b) Test 127-406-90a (θ = 90°)
(c) Test 102-406-60a (θ = 60°) (d) Test 102-406-60a (θ = 60°)
Fig. 6.9. Typical weld fractures
6.4. DISCUSSION OF RESULTS

6.4.1. APPLIED LOAD VERSUS DEFORMATION RESPONSE
Fig. 6.10 shows the relationship between , expressed as a fraction of the chord diameter ( /D), and the
applied load (P) for several representative tests. The six curves shown on Fig. 6.10 correspond to the first
weld tested on each connection (e.g. side a). To determine the relationship between /D and the applied load
for the second tests, residual displacements from the first test needed to be taken into account. The residual
displacements were estimated using an unloading curve with the same slope as the initial connection
stiffness. This curve was projected back, from the point of rupture of the first test, onto the x-axis, as shown
in Fig. 6.11. This marked the origin for the measured curve from the second test. The overall /D versus

applied load response for the second test was then obtained by extending the corresponding first curve past its
point of rupture until it met the second curve. This part of the curve is illustrated by the dashed black line in
Fig. 6.11. The chord deformation at rupture ( a) for the six second welds tested (e.g. side b) could then be
obtained. Despite having only small fillet welds, chord plastification in excess of the 3%D deformation limit
(Lu et al. 1994) occurred before rupture in seven out of the 12 tests (Table 6.4).
Fig. 6.10. Typical load versus chord deformation Fig. 6.11. Procedure for calculation of load versus
relationships deformation for second welds tested
Table 6.4. Residual chord deformation (at start of test) and chord deformation at rupture for all 12 tests
Test Residual chord deformation (as percent of D) a/D
% %
102-273-90a 0 3.23
102-273-90b 1.732 3.68
102-406-90a 1.577 4.70
102-406-90b 0 3.52
127-273-90a 0 2.06
127-273-90b 0.914 2.07
127-406-90a 0 2.61
127-406-90b 0.962 2.78
102-406-60a 0 3.34
102-406-60b 1.759 3.63
127-406-60a 0 2.34
127-406-60b 1.712 3.60
Note: residual chord deformations are equal to zero for first weld tested on each connection

6.4.2. NON-UNIFORM STRAIN DISTRIBUTIONS ADJACENT TO THE WELD

Representative graphs of the strain distribution around the branch adjacent to the test weld, at fractions of
the actual (experimental) weld fracture load (Pa), are given in Figs. 6.12 and 6.13. Moreover, Figs. 6.12 and
6.13 present the strains measured during the test in which the corresponding weld actually fractured.
(a) Test 127-273-90a ( = 0.47, θ = 90o) (b) Test 127-406-90a ( = 0.31, θ = 90o)
Fig. 6.12. Typical strain distributions adjacent to test weld (θ = 90° connections)
(a) Test 102-406-60a ( = 0.25, θ = 60o) (b) Test 127-406-60a ( = 0.31, θ = 60o)
Fig. 6.13. Typical strain distributions adjacent to test weld (θ = 60° connections)
It is shown that, for θ = 90° connections (Figs. 6.12a,b), the tensile strain (and hence tensile load)
decreases as a function of distance away from the saddle (ρ = 90° point). The tensile strain is therefore

smallest at the crown (ρ = 0° and 180° points), with much of the weld even remaining in compression for the
entire tension load range. This phenomenon equates to a non-uniform loading of the weld perimeter – which
is expectedly more pronounced for connections with higher -values, where stiff membrane action dominates
load transfer at the saddle. It can thus be concluded that weld effective lengths are present in CHS-to-CHS
connections. An illustration of the effect that causes compression at the crown is shown in Fig. 6.14.
Fig. 6.14. Effect causing compressive strains at the crown (ρ = 0° and 180° points) (θ = 90° connections)
The largest tensile strains for θ = 60° connections were initially measured at the saddle (Figs. 6.13a,b).
As the load increased, the strain adjacent to the saddle, on the heel side of the connection, began to increase at
a faster rate than the strain adjacent to the saddle on the toe side of the connection. This is due to secondary
bending effects from connection flexibility and joint rotation, which may not exist in real structures where the
chord ends are prevented from rotating.
6.5. EVALUATION OF AWS D1.1 (2015)

6.5.1. EXISTING PROVISIONS FOR WELD EFFECTIVE LENGTHS IN CHS X-CONNECTIONS
According to Clause 9.5.3 of AWS D1.1 (2015), the nominal strength of fillet welds in CHS X-
connections designed as fit-for-purpose (Pnw) is based on the limit state of shear rupture along the plane of the
weld effective throat in accordance with Eqs. (6.2a,b):

Pnw  Qwle (6.2a)
Qw  0.60tw FEXX (6.2b)
where le = weld effective length (AWS 2015). An LRFD resistance factor for fillet welds, ϕ, equal to 0.80, is
then applied to determine the design strength.
In Clause 9.5.4, simplified equations are given to compute weld lengths for CHS connections under axial
load, which can be traced back to Appendix C of British Standard 449 (BSI 1959). These factors can be
shown to calculate the total weld length, rather than the effective weld length (see Section 6.5.4). A branch
stress/load factor of 1.50 is specified by AWS in Clause 9.6.1.3(4) “Uneven Distribution of Load (Weld
Sizing)”, for design using the LRFD method. This factor, established in the 1980s, is used to prevent
progressive weld failure due to non-uniform load transfer across the weld when welds are designed as fit-for-
purpose. In modern day LRFD, the approach is to apply a reduction to the resistance of the weld, by
calculating a weld effective length, rather than to increase the design load. Hence, it is deduced that the weld
effective length implied by AWS Clause 9.6.1.3(4) is the inverse of the stress/load factor:
1 2
le  lw  lw (6.3)
1.5 3
6.5.2. SAFETY LEVEL INHERENT IN AWS D1.1 (2015)

To assess whether adequate or excessive safety margins are inherent, the structural reliability (or safety
index) ( +) can be calculated and compared to the minimum target value in North America (4.0, as currently
adopted by AISC (2016) per Section B3.1 of the AISC 360-16 Commentary), using the reliability analysis
discussed in Chapter 2 (Section 2.6.2).
The mean test-to-predicted capacity ratio (ρP) was taken as the average over all tests of Paꞌ (Table 6.1)
divided by Pnw (Table 6.2), with Pnw calculated using Eqs. (6.2a,b) and (6.3), and the measured values of Aw
and FEXX. The same values or equations used in previous Chapters for the other reliability analysis parameters
(ρM, ρG, ρP, VM, VG, VP, and ϕ +) have also been used herein. These parameters, and the results of the
reliability analysis, are shown in Table 6.5.

Table 6.5. Reliability analysis parameters for 12 CHS X-Connections

AWS (2015) AISC (2016) CSA (2014)
le/lw 2/3 unity unity unity
ϕ 0.80 0.80 0.75 0.67
ρM 1.12 1.12 1.12 1.12
VM 0.12 0.12 0.12 0.12
ρG 1.03 1.03 1.03 1.03
VG 0.10 0.10 0.10 0.10
ρP 2.13 1.42 1.42 1.27
VP 0.13 0.13 0.13 0.13
ρR 2.48 1.65 1.65 1.47
VR 0.21 0.21 0.21 0.21
ϕ+ 0.72 0.85 0.82 0.82
+
7.0 4.9 5.2 5.2
+
The implied safety index, , is equal to 7.0 for the existing AWS D1.1-15 specification provisions,
which is much larger than the minimum target safety index of 4.0 in North America. However, it should be
noted that ϕ = 0.80 in AWS (2015) is believed to be an error, as it should match AISC 360 (2016) where ϕ =
0.75. This indicates that a high level of conservatism is present in the AWS formulae. Fig. 6.15 shows the
correlation of the existing AWS predicted nominal strengths with the experimental results. The actual
strength in Fig. 6.15 is taken as the greatest load sustained by the weld (Paꞌ). On average, the actual strength
is 2.13 times larger than that predicted by AWS.
If, instead, no effective length rules are applied, and the total weld length is used to determine the
strength of the welded joint, then the correlation in Fig. 6.16 results. The implied safety index is then 4.9.
+
The mean experimental-to-predicted strength is 1.42. As > 4.0, it can be concluded that, for the range of
parameters studied, weld effective lengths are not required in conjunction with the AWS (2015) code design
method evaluated. The predicted strengths for each connection according to AWS, and the code methods that
follow, are provided in Appendix E.8.

Fig. 6.15. Correlation of existing AWS D1.1-15 provisions with the test results, with weld effective lengths
Fig. 6.16. Correlation of existing AWS D1.1-15 provisions (excluding weld effective lengths) and AISC 360-
16 provisions with test results
6.5.3. COMPARISON TO AISC 360 (2016) AND CSA S16 (2014)

AISC 360-16 gives the same equations (Eqs. 6.2a,b) for the nominal strength of fillet welds via Clause
J.2.4a with le = lw; however, to calculate the design strength, a resistance factor, ϕ = 0.75 (instead of 0.80), is
+
used. The implied safety index, , is equal to 5.2 for AISC Clause J.2.4a (Table 6.6 and Fig. 6.16), which is
expectedly larger than the minimum target safety index of 4.0, and the implied safety index of 4.9 when AWS
D1.1 is used without weld effective lengths. The foregoing evaluations of both the AWS and AISC fillet weld
design provisions assume that the (1.00+0.50sin1.5θ) directional strength-enhancement factor is not used

(AISC Clause J2.4b and AWS Clause 2.6.4.2), because it has been shown in Chapters 4 and 5 to be generally
unsafe for the design of fillet welds in HSS connections.
The ultimate strength of fillet welds in the Canadian steel code, CSA S16-14, is also based on the limit
state of shear rupture along the weld effective throat; however, CSA gives a different equation than AWS and
AISC for the nominal strength (Pnw) of fillet welds (Clause 13.13.2.2):
Pnw  0.67 Aw FEXX (6.4)
where Aw = twlw.
An LRFD resistance factor for fillet welds, ϕ, equal to 0.67, is then applied to determine the design
strength. As discussed for AISC Clause J2.4b (AISC 2016), the above equation also excludes the
1.00+0.50sin1.5θ directional strength-enhancement factor.
CSA gives a higher nominal strength than AISC for fillet welds (0.67 versus 0.60 for the shear strength
+
factor) and a proportionally lower resistance factor (ϕ = 0.67 versus 0.75). The reliability index, , implied
+ +
by CSA Clause 13.13.2.2 is therefore the same as implied by AISC Clause J.2.4a ( = 5.2, in Table 6.5,
which is greater than 4.5, the target safety index per Annex B of CSA S16-14). Fig. 6.17 shows the
correlation of the CSA predicted nominal strengths with the experimental results. On average, the
experimental strength is only 1.27 times larger than that predicted with Eq. (6.4), using the measured values
of Aw and FEXX. This value is the closest to unity amongst all methods investigated (AWS with/without weld
effective lengths, AISC, and CSA). Still, more work is needed to determine the effect of connection
parameters outside of the range studied on the weld strength.
Fig. 6.17. Correlation of CSA S16-14 provisions with test results

6.5.4. EVALUATION OF AWS D1.1 (2015) TOTAL WELD LENGTH APPROXIMATIONS

In AWS D1.1-15, the total weld length is determined from the following equation (Clause 9.5.4):
lw   Db Ka (6.5)
where Ka = weld length factor, given as:
2
1  3 2    1  3   
2

2
1 1
Ka      3      2 
(6.6)
2 sin  3  2   2   2 sin    3  2    
Eq. (6.6) gives the projection of the weld root along an inclined branch onto a cylindrical surface. It takes
into account both the branch-angle and beta-ratio distortion of the weld length. If one considers a CHS branch
welded to a flat plate at θ = 90°, with lw then equal to πDb, branch-angle distortion is the transformation of the
circular weld into an ellipse caused by a change in θ. Beta-ratio distortion occurs when the flat plate is
replaced by a cylindrical surface, causing the plane of the weld to distort into a saddle shape. Despite its
complex appearance, Eq. (6.6) is only an approximation to the weld length. Note that when equals zero and
θ equals 90°, such as the case for welding a CHS branch at 90° to a flat plate in the example above, Ka does
not equal exactly 1.00 using Eq. (6.6). Instead, it equals 0.99. Hence, as part of a comprehensive evaluation
of the AWS code, it is necessary to evaluate the error associated with this method to calculate lw.
Fig. 6.18. Comparison of lw/πDb using Eq. (6.6) (AWS 2015) and the vector-calculus method
Fig. 6.18 shows the relationship between lw/πDb (= Ka) determined using Eq. (6.6) (shown by the solid
red lines) and lw/πDb determined from the vector-calculus method (shown by the data points), as used herein,

for a range of values. It is shown that Eq. (6.6) is conservative as a design tool, i.e. it under-predicts the
weld length. The maximum error is only 0.6% over the range of parameters studied (for = 0.10 and θ just
less than 90°). Thus, despite its complexity, Eq. (6.6) gives a predicted weld length very close to the actual
weld length.
AWS notes that the following formula for Ka may be conservatively used instead of Eq. (6.6) to calculate
lw for CHS connections:
1  1 / sin 
Ka  (6.7)
2
Fig. 6.19 (analogous to Fig. 6.18) shows the relationship between lw/πDb determined using Eq. (6.7) and
lw/πDb determined from the vector-calculus method. It is shown that Eq. (6.7) is even more conservative than
Eq. (6.6) as a design tool. The maximum error is still only 1.9% over the range of parameters studied (for =
0.50 and θ = 90°), which is expectedly larger than the error associated with Eq. (6.6). However, Eq. (6.7) is
always conservative.
Fig. 6.19. Comparison of lw/πDb using Eq. (6.7) (AWS 2015) and the vector-calculus method
One could argue that, since these lower-bound approximations give actual-to-predicted weld lengths
greater than 1.00, mean values and variations in the actual-to-predicted weld length should be included in the
factors ρG and VG in the reliability analysis. Including these variations would marginally increase the
reliability index ( +). However, with CAD now widely used for design and analysis, it seems increasingly
likely that designers will opt to find the total weld length using software, rather than using Eqs. (6.6) or (6.7).
It was therefore deemed prudent to omit variations in the actual-to-predicted weld length in the factors ρG and

VG, because including them would be un-conservative. Eqs. (6.6) and (6.7) are still, however, useful (and
conservative) design tools.
6.6. EVALUATION OF PROCEDURES TO DETERMINE WELD EFFECTIVE

LENGTHS FROM TESTS
For calculating weld effective lengths from tests, which have been shown to exist for CHS-to-CHS X-
connections in Section 6.4, three methods are evaluated:
• Method 1: The ratio of le to lw is taken as the ratio of nominal-to-peak elastic strain. The nominal
elastic strain was calculated by multiplying strain measured in the branch adjacent to the weld (as shown in
Figs. 6.12a – d and 6.13a,b when P = 0.25Pa), by the weld length tributary to the measurement, then dividing
the sum of the results by the sum of the tributary weld lengths. Since the strain is elastic, the result of using
Paꞌ instead of Pa in this process is not significant. Either the entire weld length, or a length of weld between a
plane or planes of symmetry, should be instrumented with SGs when using this method. Compressive strains
in the branch adjacent to the weld were taken as zero strain (rather than negative strain), because they do not
increase the weld effective length. Wang et al. (2015) used this method to study weld effective lengths for
CHS branch-to-RHS chord connections. Caulkins (1968) used a similar method to study welds in CHS T-
connections. His method used forces from finite element models instead of elastic strains. Due to
proportionality, however, it yields the same results. The appeal of this method is that weld effective lengths
can be determined from elastic tests, and it logically takes into account stress concentrations along the weld
length. Moreover, weld-critical tests, which are difficult to achieve, are not needed. This method is based
explicitly on elastic load/stress distribution, and it does not take into account load/stress re-distribution in the
weld before rupture. It is therefore likely to be a lower-bound.
• Method 2: the weld effective length is empirically determined by comparison of actual-to-predicted
strengths, with Pnw calculated using an accurate predictor such as Eq. (6.8), with actual values of tw, tb, Db, Aw,
and FEXX (Chapter 5):
 D t 
Pnw  1.009  0.00137 b  0.197 w  Aw FEXX
 tb tb  (6.8)
Eq. (6.8) was developed from regression of a large database of weld-critical CHS-to-rigid plate
connection finite element results (see Section 5.6). As such, it is tailored to the unique loading on single-sided
fillet welds to CHS branches (which produces tension at the weld root under branch axial tension), and
moreover takes into account the principal influential geometric parameters of the CHS member and weld
joint. It is used to predict the strength of a fillet weld to a CHS branch when the weld is fully effective (i.e. le
= lw). The ratio of le to lw for each test is hence the ratio of Paꞌ (Table 6.1) to the predicted value using Eq.

(6.8). Method 2 takes into account load/stress re-distribution in the weld before rupture, unlike Method 1. It
also utilizes an accurate formula for predicting the nominal weld strength [Eq. (6.8)]. Since Eq. (6.8) was
developed from extensive testing and finite element analysis on CHS connections, Method 2 is believed to be
the most accurate way to determine the true weld effective lengths.
• Method 3: the weld effective length is empirically determined by comparison of actual-to-predicted

strengths, with Pnw calculated using Eqs. (6.2a,b), with le = lw and actual values of tw, lw, and FEXX. The ratio of
le to lw is hence the ratio of Paꞌ (Table 6.1) to the predicted nominal weld strength using simple code
equations. Method 3 provides the values of le for each fillet weld that would result in an actual-to-predicted
strength ratio of 1.0 when used in conjunction with the AWS (2015) fillet weld design provisions [Eqs.
(6.2a,b)], and also the AISC 360 (2016) fillet weld design provisions in Clause J2.4a.
Table 6.6 gives the values of le/lw computed for each of the 12 fillet welds in the CHS X-connections
tested herein, using Methods 1 to 3.
Table 6.6. Weld effective length ratios for CHS X-connections determined using three possible methods
Weld effective length ratio, le/lw
Test Method 1 Method 2 Method 3
102-273-90a 0.29 1.01 1.48
102-273-90b 0.29 0.96 1.39
102-406-90a 0.28 1.03 1.54
102-406-90b 0.28 1.03 1.55
127-273-90a 0.28 0.82 1.28
127-273-90b 0.28 0.75 1.16
127-406-90a 0.27 0.81 1.26
127-406-90b 0.27 0.74 1.15
102-406-60a 0.33 1.13 1.69
102-406-60b 0.33 1.08 1.59
127-406-60a 0.32 0.83 1.28
127-406-60b 0.32 1.07 1.67
With Method 1, le/lw is always less than 1.0, and moreover less than 2/3 [Clause 9.6.1.3(4) of AWS
(2015)]. Since le = (2/3)lw has already been shown to be very conservative (Section 6.5.2), it can be concluded
that Method 1 is even more conservative and inaccurate. The aim in assessing the values of le/lw in Table 6.6
is to achieve a weld effective length ratio between 2/3 and 1.0.
With Method 2, le/lw is between 0.75 and 1.13 (Table 6.6). Generally, fillet welds in connections with
similar parameters ( , D/t, ) have similar ratios of le/lw, in accordance with expectations based on previous
research (Caulkins 1968).
For Method 3, using Eqs. (6.2a,b), the weld effective length is always greater than the real length, which
cannot be true. This outcome is not unexpected, since Eq. (6.2) (and AISC 360-16 Clause J2.4a) has already
been shown to be conservative for CHS connections with fully effective welds in Chapter 5.

Method 2 is therefore a logical procedure to determine weld effective lengths from tests. It is thus
deduced that weld effective lengths for CHS vary as a function of connection parameters, and rupture tests
(experimental and/or numerical) over a broad range of geometric parameters are necessary to reasonably
determine weld effective lengths.

6.7. SUMMARY
Based on 12 careful laboratory tests on CHS X-connections under branch axial tension, which all failed
by rupture along a plane through the connecting fillet weld, it is shown that weld effective lengths exist in
CHS-to-CHS connections, and that the existing AWS (2015) code provisions for weld effective lengths in
such connections, given by Clause 9.6.1.3(4), are very conservative. Furthermore, it is shown that the current
AWS (2015), AISC (2016), and CSA (2014) specification provisions provide adequate structural reliability
( +
≥ 4.0 or 4.5) without weld effective lengths (i.e. using the total weld length to determine the weld
strength), assuming the fillet weld directional strength enhancement factor is not used. This is due to the
simplicity of the fillet weld nominal strength formula, because weld lengths were shown (by strain
distributions) to be less than 100% effective. These conclusions are currently limited to 0.25 ≤ ≤ 0.47, 23 ≤
D/t ≤ 34, 0.6 ≤ ≤ 1.0 and θ = 60° or 90°.
A systematic approach to calculating the total weld length in CHS-to-CHS connections has also been
presented. This approach is based on vector calculus and can be easily programmed to allow designers,
fabricators, and researchers of tubular structures to accurately calculate weld lengths. The approximations
given in AWS (2015) Clause 9.5.4 for the total weld length (lw) in CHS connections were compared to the
vector-calculus approach and found to be useful lower-bound design tools for connections with 0.1 ≤ ≤ 0.5
and 60° ≤ θ ≤ 90°.
It is shown that the ratio of the weld effective length to the total weld length (le/lw) is not constant for all
CHS connections, as the current AWS specification currently implies via Clause 9.6.1.3(4), and varies as a
function of connection parameters. It is also concluded that rupture tests on weld-critical connections are
necessary to reasonably determine weld effective lengths.
A comprehensive parametric modelling study, using finite element methods, and a secondary reliability
analysis are henceforth conducted, to determine: (a) if these findings are applicable to a wider range of fillet-
welded CHS X-connections, and (b) the effect of connection parameters , θ, D/t, and on the weld strength.

Chapter 7: FILLET WELD EFFECTIVE LENGTHS IN CHS X-
CONNECTIONS: FINITE ELEMENT MODELLING STUDY
7.1. INTRODUCTION
To date, only one experimental study (discussed in Chapter 6) has been conducted to evaluate the
performance of welds in hollow structural section (HSS) connections to a circular hollow section (CHS)
chord. Without experimental evidence, specifications have been reluctant to provide a “fit-for-purpose”
approach to weld design that takes into account the non-uniform contribution of the weld perimeter by using
weld effective lengths. Currently, only AWS D1.1 Structural Welding Code – Steel (AWS 2015) gives such a
method, in Clause 9.6.1.3(4), but it is badly defined. Instead, welds in connections to a CHS chord are
routinely sized to develop the yield capacity of the connected branch member. This is done by meeting
prescriptive requirements for the weld throat dimension (tw) as a function of the branch thickness (tb). In rare
cases, it is suggested that welds be designed for the connection capacity, instead of the yield capacity of the
branch (AWS 2015). According to ISO (2013), this approach is not common.
Fig. 7.1. CHS X-connection general configuration and non-dimensional parameters
In the lone experimental study (Chapter 6) that evaluated the performance of welds in CHS-to-CHS X-
connections (Fig. 7.1), linear strain gauge (SG) measurements adjacent to the weld showed that load transfer
was highly non-uniform, and peaked in the saddle position. This indicates that a weld effective length
120
Chapter 7: Fillet Weld Effective Lengths in CHS X-Connections: Finite Element Modelling 121
phenomenon exists in CHS connections. A reliability analysis of the experimental results determined that the
weld effective length provisions given by Clause 9.6.1.3(4) of AWS (2015) are highly conservative. For the
range of parameters studied, the AWS (2015), AISC (2016), and CSA (2014) provisions for fillet weld design
+
exceeded the minimum safety index in North America ( > 4.0) even without weld effective lengths.
This chapter presents a parametric modelling study that was performed, using finite element (FE)
methods, to determine: (a) the effect of key connection parameters on weld strength in CHS X-connections;
(b) if these findings are applicable to a wider range of fillet-welded CHS X-connections; and (c) a better
method for estimating fillet weld strength, using weld effective lengths.
7.2. FILLET WELD DESIGN FOR CHS X-CONNECTIONS

7.2.1. AWS D1.1 (2015)
The definitive guidance on weld design for CHS-to-CHS connections in North America is given by AWS
D1.1 (2015). AWS uses the term “T-, Y-, and K-connections” generically to describe welded connections
between CHS, and also refers to X-connections as double-tee connections (Clause 9.6.1.5). For such
connections, AWS D1.1 Clause 9.6.1.3 points out that “…due to differences in the relative flexibilities of the
main member loaded normal to its surface, and the branch member carrying membrane stresses parallel to its
surface, transfer of load across the weld is highly non-uniform, and local yielding can be expected before the
connection reaches its design load”. AWS gives two possible methods to proportion fillet welds in CHS
connections: a fit-for-purpose approach, using a weld effective length, and a prequalified weld size approach.
7.2.1.1. Weld Effective Length Design Approach

AWS D1.1-15 Clause 9.5.3 permits the nominal strength (Pnw) of fillet welds to be calculated according
to Eqs. (7.1) and (7.2):
Pnw  Qwle (7.1)
Qw  0.60tw FEXX (7.2)
where le = weld effective length, and FEXX = ultimate strength of weld metal (a minimum specified tensile
strength is used). An LRFD resistance factor for fillet welds, ϕ, equal to 0.80, is then applied to determine the
design strength of the fillet weld (ϕPnw).
Equations for the total weld length (lw) are given in AWS D1.1-15 Clause 9.5.4, in the following form:
lw   Db Ka (7.3)

where Db = diameter of the CHS branch, and Ka = weld length factor, giving the ratio of the total weld length
(taken along the weld root) to the branch circumference. The simplest Ka factor used by AWS is:
1  1 / sin 
Ka  (7.4)
2
where θ = branch inclination angle.

The effective weld length (le) in AWS D1.1-15 is implied by a branch stress/load factor of 1.5 specified
in Clause 9.6.1.3(4), for LRFD, which is stated to “… account for non-uniform distribution of load”. The
inverse of this factor gives le as a fraction of the total weld length (lw):
2
le  lw (7.5)
3
7.2.1.2. Prequalified Fillet Weld Design Approach

AWS D1.1-15 Clause 9.6.1.3(3) states that if prequalified fillet weld details are used per their Figure
9.10, and CHS have a yield strength (Fy) > 280 MPa, then tw may be specified as the lesser of Eqs. (7.6) and
(7.7):
To develop the capacity of the connected member walls:
tw  1.07tb (7.6)
To develop the local “punching shear” connection capacity of the chord:
tw  t (7.7)
where t = CHS chord thickness.

AWS (2015) limits these prequalified fillet weld sizes to connections with ≤ 0.33 per Clause 9.5.1.2.
7.2.2. CIDECT (WARDENIER ET AL. 2008)

The CIDECT design guide for statically loaded CHS connections (Wardenier et al. 2008) gives a
prequalified fillet weld design approach based on the Directional Method of EN 1993-1-8 (CEN 2005). By
setting the weld resistance equal to the design capacity of the branch, tw = 1.10tb is given to develop the
capacity of the connected member walls for steel with Fy = 355 MPa. This is determined by assuming a 90°
branch welded to a flat surface, and a weld effective length equal to πDb (i.e. le = lw).

Fleischer & Herion (2015) aimed to reduce the prequalified weld throat from tw = 1.10tb by taking into
account the actual weld length in CHS connections. The actual weld length is a function of the branch
inclination angle (θ) and the branch-to-chord diameter ratio ( = Db/D) (Fig. 7.1). To take the actual weld
length into account, the authors borrowed a more accurate Ka value from AWS D1.1-15 Clause 9.5.4 (Eq.
6.10, in Chapter 6). They then computed the weld strength using the Simplified Method of EN1993-1-8 (CEN
2005), which is a conservative (lower-bound) design alternative to the Directional Method. It is easier to
apply to CHS connections because it does not require stress components on the throat plane to be determined.
Using the CIDECT methodology (i.e. setting weld strength equal to branch capacity), it was shown that the
prequalified fillet weld size could theoretically be reduced for some joints. The required fillet weld throat was
a complex function of , the branch slenderness ratio (Db/tb), and θ (Fig. 7.1). Design charts were hence given
for simplicity. This work was still based on the design philosophy of developing the capacity of the
connected member walls. Moreover, weld effective lengths were not determined.
7.2.3. CAULKINS (1968) AND MARSHALL (1992)

Caulkins (1968) did work relevant to a fit-for-purpose approach to weld design for CHS connections. He
showed, using a shell theory solution, that load transfer through the weld peaks at the saddle points in CHS-
to-CHS T-connections. Recent experimental research (presented in Chapter 6) has confirmed this for CHS-to-
CHS X-connections. Marshall (1992) used the same data to demonstrate that load transfer efficiency is
greater for T-connections with low D/t (i.e. a stocky chord member), low , and low branch-to-chord
thickness ( ) (Fig. 7.2). In his attempt to generalize the results, he found that evidence on the effect of high
was mixed.
Fig. 7.2. Load transfer efficiency across the weld of a θ = 90° CHS-to-CHS T-connection, adapted from
Marshall (1992)

In the work by Caulkins (1968) and Marshall (1992), load transfer efficiency was defined as the ratio of
nominal-to-peak elastic load at the connection. This ratio was believed to be related to the weld effective
length. In another study by Wang et al. (2015), load transfer efficiency was defined as the ratio of nominal-to-
peak elastic strain measured adjacent to the weld. This study investigated weld effective lengths for
connections between CHS branches and RHS chords. The definition of load transfer efficiency is marginally
different, but the methods provide the same results because load and strain are linearly related in the elastic
range. Chapter 6 showed that these methods provide a very conservative lower-bound when used to explicitly
determine le.
7.3. FINITE ELEMENT MODELLING

To validate the FE modelling approach used herein, 12 CHS-to-CHS X-connection models were
developed to replicate previous experimental tests, with the same geometric and material properties as the
experimental specimens in Chapter 6. The geometric properties and experimental results are summarized in
Table 7.1.
Table 7.1. Geometric properties of CHS X- (test) connections and comparison of experimental and FE results
Test θ
CHS branch member CHS chord member tw Pa a P aꞌ b a
c
PFE d FE
d
Paꞌ/PFE a/ FE
Db × tb D×t
° mm × mm mm × mm mm kN kN % kN %
102-273-90a 102.0 × 7.34 273.5 × 11.69 4.08 672 672 3.23 655 2.72 1.03 1.19
102-273-90b 102.0 × 7.34 273.5 × 11.69 4.37 678 678 3.68 690 3.03 0.98 1.21
102-406-90a 102.0 × 7.34 406.5 × 12.34 3.56 608 608 4.70 543 3.31 1.12 1.42
102-406-90b 102.0 × 7.34 406.5 × 12.34 3.14 540 540 3.52 495 2.87 1.09 1.23
90
127-273-90a 127.4 × 11.55 273.5 × 11.69 3.63 653 653 2.06 762 2.67 0.86 0.77
127-273-90b 127.4 × 11.55 273.5 × 11.69 4.00 609 653 2.07 811 2.98 0.80 0.69
127-406-90a 127.4 × 11.55 406.5 × 12.34 3.16 557 557 2.61 631 3.21 0.88 0.81
127-406-90b 127.4 × 11.55 406.5 × 12.34 3.47 556 557 2.78 617 3.01 0.90 0.92
102-406-60a 102.0 × 7.34 410.0 × 12.21 3.58 721 721 3.34 640 2.62 1.13 1.28
102-406-60b 102.0 × 7.34 410.0 × 12.21 3.79 538 721 3.63 672 2.84 1.07 1.28
60
127-406-60a 127.4 × 11.55 410.0 × 12.21 3.95 761 761 2.34 903 3.58 0.84 0.65
127-406-60b 127.4 × 11.55 410.0 × 12.21 3.38 798 850 3.60 798 2.87 1.06 1.25
a
Actual (experimental) force in branch member at weld fracture.
b
Greatest load sustained by the weld.
c
Actual (experimental) chord deformation at weld fracture (as % of measured CHS chord dimeter D) averaged over both
branches.
d
PFE and FE are analogous to Paꞌ and a, but refer to the FE values at weld fracture.
The 12 connections were modelled using the commercially available software package ANSYS 14.0
(Swanson Analysis Systems 2011). Although it was possible to model just one eighth of the non-inclined (θ =
90°) connections due to symmetry about three principal planes passing through the connection work point,
one half of each connection was modelled instead, for all connections, to accommodate the inclined branch
cases. This simplified the parametric programming of the models, which was later done using ANSYS batch
files. Symmetry boundary conditions were applied in the plane of the connection (i.e. along the cut face). To

replicate the experimental connection of the branches to the chord, a 0.25-mm gap was modelled between the
two members. This gap ensured that the applied load was only transferred through the fillet weld (i.e. it
prevented direct load transfer between the branch and chord members). The general geometry (showing the
gap), mesh layout, and boundary conditions are shown in Fig. 7.3 (for a non-inclined connection) and Fig. 7.4
(for an inclined connection).
All FE analyses were conducted by applying static incremental displacements (non-linear time-step
analysis) to the ends of the specimen (e.g. the ends of each branch, or for coupon tests, the ends of each
coupon). Large deformation allowance (non-linear geometry) and non-linear material properties (for each
different material) were included.
Fig. 7.3. FE CHS-to-CHS X-connection geometry, mesh layout, and boundary conditions with θ = 90°

Fig. 7.4. FE CHS-to-CHS X-connection geometry, mesh layout, and boundary conditions with θ < 90° (θ =
60° shown)
7.3.1. MATERIAL PROPERTIES

Multi-linear true stress-strain curves for each different material (i.e. the weld metal, and each different
branch and chord member) were derived from tensile coupon tests conducted in accordance with ASTM
A370 (2017). The procedure used is as follows:
Prior to necking: the average engineering stress ( ) and engineering strain ( ) ordinates from the tests
were converted to true stress ( T) and true strain ( T). The following well-known relationships were used
(Boresi & Schmidt 2003):
 T   (1   ) (7.8)
T  ln(1   ) (7.9)
After necking: an iterative approach based on weighting an approximate lower- and upper-bound to the
true stress versus true strain response was used to determine the ordinates on the curve:

  T  
 T   T   w(1   T   T  )  (1  w)  T
 (7.10)
 T  
  T  
where 𝜎𝑇′ = true stress at the start of necking, ε′𝑇 = true strain at the start of necking, and w = weighting factor
(Ling 1996).
The weighting factor in Eq. (7.10) was derived for each different material by matching the engineering
stress-strain curve of a tensile coupon modelled in ANSYS to the average engineering stress-strain curve
from the experimental tests. The FE coupon was modelled using the average measured geometry from the
tests, with the original curved shape for CHS coupons. The engineering strains for the FE coupons were
calculated over a 50-mm gauge length from nodes on the exterior of the coupon to closely mimic the
experimental method using a clip gauge. Typical engineering stress-strain curves and the corresponding true
stress-strain curves derived using the above process are shown in Figs. 7.5a,b. In Figs. 7.5a,b the solid line is
the average experimental curve and the dashed line is the FE-generated curve. The true stress-strain curves
continue indefinitely along the x-axis since a fracture criterion was not calibrated for the coupons. Material
properties for the 12 X-connections are given in Chapter 6.
(a) Engineering stress-strain (b) True stress-strain
Fig. 7.5. Comparison of typical experimental (solid line) and FE (dashed line) stress-strain curves
During the experimental tensile coupon tests, it was necessary to remove the clip gauge shortly after
necking to prevent damage to it. The comparison of the FE and experimental engineering stress-strain curves
therefore involved the pre-necked portion and a small variable amount of the post-necked portion of the
curve, and the ordinates at rupture (shown with ‘x’ and ‘o’ markers in Fig. 7.5a). The engineering rupture

stress was determined by dividing the load just before rupture by the initial cross-sectional area of the
coupon, and the rupture strain was determined by rejoining the fractured coupon, measuring it to determine
the final gauge length, and dividing the final gauge length by the initial gauge length. At rupture, the necked
shape of the FE coupons closely resembled that of the experimental coupons, with necking at the mid-point.
The same elements ultimately used for the CHS X-connection models (eight-noded solid brick elements)
were used.
7.3.2. MODEL SENSITIVITY STUDY

To determine the best-suited mesh layout and element type for the CHS X-connection models, a model
sensitivity study was performed. Figs. 7.6a - c show the typical mesh layouts studied, which were adapted to
60° and 90° connections. The mesh layout in Fig. 7.6a has the same number of elements circumferentially
around the branch as the mesh layout in Fig. 7.6b, but fewer elements on the chord adjacent to the weld. The
mesh layout in Fig. 7.6c has more elements circumferentially around the branch than the mesh layouts in Fig.
7.6a,b, as well as more elements on the chord adjacent to the weld. Table 7.2 gives additional parameters for
the meshes studied.
Fig. 7.6. Mesh layouts used in the mesh sensitivity study
The performance of each mesh layout and element type was evaluated by comparing the load-
deformation response of the FE connection to the experimental test (see Chapter 6) (Fig. 7.7a). Table 7.2
compares the values of the initial stiffness (ka = actual (experimental) initial stiffness, kFE = FE initial
stiffness) and the load at a chord deformation ( ) equal to 3%D (the connection plastification limit) (Lu et al.
1994) (P3%D,a = actual load at connection plastification limit, P3%D,FE = FE load at connection plastification
limit). The value of was taken as the chord face deformation (normal to the chord, relative to the chord
centre line) but averaged for both sides of the connection. Displacement was measured between points 50 mm
away from the crown, as show in Fig. 7.7a. In general, as the number of nodes increased by using a finer
mesh, a higher-order element, or more CHS through-thickness elements, the stiffness of the connection

increased and so did the solution time. Fig. 7.7b demonstrates that the solution typically converged with four
elements through the CHS thickness.
Table 7.2. Mesh sensitivity study results for test 102-273-90a

CHS Weld
Mesh thickness face P3%D,a/
layout Element type No. of elements No. of nodes elements elements kFE ka/kFE P3%D,FE P3%D,FE
kN/mm kN
B SOLID45 22520 33885 2 7 140.0 1.21 636 1.03
C SOLID45 33184 49863 2 139.8 1.21 634 1.03
B SOLID45 32724 44520 3 157.2 1.08 663 0.99
B SOLID95 32724 165913 3 7 169.4 1.00 670 0.98
C SOLID45 48456 65644 3 157.0 1.08 662 0.99
B SOLID45 42928 55155 4 163.2 1.04 669 0.98
C SOLID45 63728 81564 4 7 162.8 1.04 667 0.98
A SOLID45 36288 46685 4 166.6 1.01 677 0.97
Note: ka = 169.0 kN/mm and P3%D,a = 654 kN.
(a) Load-deformation response (b) FE convergence iterations
Fig. 7.7. Comparison of FE and experimental mesh sensitivity parameters for test 102-273-90a
The best overall agreement with the experimental results was obtained using a medium-density mesh
(Mesh layout A or B), with four elements through the CHS branch and chord member thickness and eight-
noded solid brick elements (SOLID45 in ANSYS). Elements used had reduced integration and hourglass
control. To decide between Mesh layouts A and B, analyses of connections with different branch angles,
values, weld sizes, and CHS member thicknesses were undertaken to foresee any problems associated with
these mesh types. Mesh layout A was found to be capable of mapping to a wider range of connection
geometries (i.e. some models could not be generated using mesh layout B), and was therefore used.

7.3.2.1. Weld Fracture Criterion

An important part of the current study is modelling fillet weld fracture to determine the ultimate load for
weld-critical connections. A method used in previous studies to model fracture in steel (Martinez-Saucedo et
al. 2006; Voth & Packer 2012a; Chapter 5) was adopted herein. Fracture was modelled using the ANSYS
element death feature, which was initiated by an equivalent strain fracture criterion ( ef). The ANSYS element
death feature reduces the stress and stiffness of “killed” elements to near-zero, allowing the element to freely
deform. The equivalent strain fracture criterion for the weld in the FE models was calibrated from six
experimental tests (referred to as the “training set”), to match the load and chord deformation at weld
fracture. Figs. 7.8a - c show the ratios of the experimental weld strength (taken as Paꞌ, the maximum load
experienced by the weld) to the FE weld strength (PFE), and the ratio of the actual (experimental) deformation
at rupture ( a) to the FE deformation at rupture ( FE), for the training set for different values of ef,weld. The
curves are spread over three plots for clarity.
These figures illustrate that as ef,weld was increased, the ratios of experimental-to-FE load and
deformation predictably decreased, because PFE and FE become greater. They also illustrate that connections
with similar values of required similar values of εef,weld to obtain Paꞌ/PFE and a/ FE equal to 1 (i.e. to match
the FE and experimental rupture loads and rupture deformations). To minimize error associated with this
scatter, the average best-fit value over the six tests ( ef,weld = 0.32) was selected as the fracture criterion. All 12
connections were then analysed using ef,weld = 0.32 for the fillet weld.
The difference in ef,weld between the current Chapter ( ef,weld = 0.32) and Chapter 5 ( ef,weld = 0.092) is due
to different triaxiality at the location of rupture in welds in HSS-to-rigid plate connections and CHS-to-CHS
connections. Such differences are caused by element boundary conditions, mesh arrangement and loading in
the model (Kanvinde & Deierlein 2006; Voth & Packer 2012a), which affect triaxiality, as discussed in
Section 5.3.4.

(a) Tests 102-273-90a and 127-273-90a
(b) Tests 102-406-90b and 127-406-90a (c) Tests 102-406-60a and 127-406-60a
Fig. 7.8. Ratios of Paꞌ/PFE and a/ FE for the training set for different values of ef.weld
7.3.2.2. Effect of Model Size (Scalability)

The effect of absolute model size (e.g. using the same non-dimension parameters but different absolute
values of the geometry) was investigated. The results of typical FE analyses are shown in Table 7.3. The
same modelling techniques as above were used. The results in Table 7.3 indicate that the model and the weld
strength are not sensitive to the absolute size of the model. Moreover, the normalized rupture load
(PFE/AwFEXX, where Aw = twlw) is the same for models with the same non-dimensional parameters.

Table 7.3. Effect of X-connection model size (scalability)

Branch diameter Non-dimensional parameters Weld dimensions
θ D/t tw lw PFE PFE/AwFEXX
mm ° mm mm kN
100 90 0.50 10 1.0 0.50tb 320 1866 1.012
200 90 0.50 10 1.0 0.50tb 639 7467 1.012
300 90 0.50 10 1.0 0.50tb 959 16799 1.012
100 90 0.10 30 1.0 0.50tb 314 3103 1.026
200 90 0.10 30 1.0 0.50tb 629 12281 1.017
Note: all other geometry is calculated from the branch diameter, using the non-dimensional parameters shown.
7.4. FINITE ELEMENT MODELS EVALUATED AGAINST EXPERIMENTAL

RESULTS
Figs. 7.9a,b compare the experimental and FE load-deformation curves for the six specimens in the
training set. The deformations have been normalized, by dividing by D, so that curves for connections with
different chord diameters can be presented on the same plot. For clarity, the six curves have been divided
between two graphs. Chord deformations were measured as described in Section 7.3.2. Fig. 7.9 illustrates that
the FE models are capable of predicting the actual response of the experiments. Figs. 7.10a,b show the
correlation of the experimental and FE ultimate strengths and deformations (at rupture) for all 12 tests (see
values in Table 7.1). The mean actual-to-FE predicted (A/P) rupture load was 0.98, with a coefficient of
variation (COV) of 0.12. These values indicate that the model made acceptable predictions of Paꞌ across all
tests (including tests not in the training set). The mean actual-to-FE predicted deformation was 1.06, with a
COV of 0.25. A better correlation is obtained for the rupture load (Fig. 7.10a) than the deformation (Fig.
7.10b) because displacements were small, and fracture typically occurred on the non-linear part of the load-
deformation curve, as shown previously in Figs. 7.9a,b.

(a) Tests 102-273-90a, 102-406-90b, and 127-273-90a (b) Tests 127-406-90a, 102-406-60a, and 127-406-60a
Fig. 7.9. Comparison of experimental and FE load-deformation curves
(a) Rupture load (b) Chord deformation at rupture
Fig. 7.10. Correlation of experimental and FE results
Experimental spot strain measurements from SGs located 20 mm from the weld (around one half of the
weld, due to symmetry about the plane of the connection), oriented along the axis of the branch, indicated
that elastic load transfer peaked at the saddle point of the connection (subtended angle, ρ = 90°) (see Chapter
6). The sign convention for ρ is shown on Fig. 7.6b (clockwise from the heel). Typical measurements of
elastic strain parallel to the branch along a line intersecting these gauges in the FE models (indicated by a
dashed line in Fig. 7.6b) are shown in Figs. 7.11a,b.

(a) Test 127-273-90a ( = 0.47, θ = 90o, = 0.99) (b) Test 102-406-60a ( = 0.25, θ = 60o, = 0.60)
Fig. 7.11. Comparison of typical experimental and FE longitudinal strain distributions adjacent to the weld
In Figs. 7.11a,b, elastic branch strains from experiments are represented by solid black lines at an applied
load (P) equal to 0.25Pa (i.e. 25% of the load at which the weld actually ruptured in the experiments, during
the same monotonic loading phase), with the FE strains (solid red lines) predicting the same trends and values
as shown by experimental strains. It is shown that agreement between the FE and experimental strain
distributions under elastic load (P = 0.25Pa) was generally good. At ultimate, the FE strain distributions
showed the same trend; however, they poorly predicted the exact spot strain values measured in the
experiments. This is believed to be due to progressive and non-uniform yielding of the weld along its length,
due to variations in the experimental weld geometry that were not captured in the FE models. Nonetheless,
the FE models provided further evidence that load peaked in the saddle position, and revealed that weld
rupture initiated in the saddle position (i.e. the first killed elements were at this location). As load increased
further, failure propagated in both directions towards the crown points.

7.5. FINITE ELEMENT PARAMETRIC STUDY

A parametric code was developed to consistently apply the modelling procedure (e.g. same mesh layout,
ef,weld = 0.32, boundary conditions, and solver) to FE analysis in ANSYS. The range of non-dimensional
connection parameters was chosen to cover all permissible fillet-welded connections subject to the following
restrictions:
(a) the local dihedral angle (Ѱ) limits imposed by AWS (2015) Fig. 9.10 and Table 9.5 (60° ≤ Ѱ ≤
120°);
(b) the limits of applicability of connection design formulae in AISC (2016) Table K3.1, which are
given in AISC (2016) Table K3.1A; and
(c) the range of standard CHS sections available for designers in Table 1-13 of the Steel
Construction Manual (AISC 2011).
The parameters varied were: the branch inclination angle (θ = 60°, 70°, 80°, and 90°); the chord
slenderness (D/t = 10, 20, 30, 40, and 50); the branch-to-chord dimeter ratio ( = 0.10, 0.20, 0.30, 0.40, and
0.50); and the branch-to-chord thickness ratio ( = 0.20, 0.40, 0.60, 0.80, and 1.00). Although a total of 500
permutations exist for the values given, there are several practical limitations that must be considered. First,
available CHS sections limit branch slenderness ratios (Db/tb) to between about 10 and 50. Secondly, not all
combinations of and θ produce Ѱ between 60° and 120° (to qualify as a fillet weld) along the entire weld
length. Fig. 7.12 shows the results of a study conducted to determine compatible values of and θ that meet
this requirement. It is shown that permissible values of range from 0.50 (for 90° connections) to 0.28 (for
60° connections).
A comprehensive parametric study was performed by modelling up to 0.30 for 60° connections and
up to 0.50 for all other branch angles. A total of 256 CHS X-connection models was analyzed.

Fig. 7.12. Compatible values of and θ to keep Ѱ between 60° and 120° along the entire weld length,
determined using methodology by Luyties & Post (1988)
7.5.1. DETAILS OF PARAMETRIC MODELS

The parametric FE models had a constant branch diameter (Db) of 200 mm. Although this is larger than
Db used in the experiments (Table 7.1), Section 7.3.2.2 demonstrated that the the normalized rupture load
(PFE/AwFEXX) remains largely unaffected. All models also contained an average weld throat dimension (tw)
equal to 0.50tb to ensure that the branch yield capacity was not reached before weld fracture. As in the
experiments, the ends of the chords were uncapped and unrestrained. The length of the chord (l) was 10D
(when D/t > 25) or 6D (when D/t ≤ 25) to prevent chord end effects at the connection (van der Vegte &
Makino 2010). The length of the branches (lb) was 3Db, and load was applied to their ends. Load application
was therefore in the theoretical constant stress region (Mehrota & Govil 1972).
All models used the same set of material properties for the weld, the branches, and the chord. These
material properties were based on materials tested experimentally that gave the most nominally matched weld
metal and base metals. Fig. 7.13 shows the engineering stress-strain curves for these materials, which were
converted to true-stress strain curves (see Section 7.3.1) for use in ANSYS. Table 7.4 summarizes the yield
stress (Fy) and ultimate stress (Fu, or FEXX for weld metal) of these materials.

Fig. 7.13. Engineering stress-strain curves for materials used in the parametric models
Table 7.4. Yield stress and ultimate stress of materials used in the parametric models
Yield stress Ultimate stress, Fu (or FEXX for weld metal)
MPa MPa
Weld 517 577
Chord 460 540
Branches 431 488
Both welds were modelled as the same size, but the fracture criterion was only assigned to the upper
(test) weld (e.g. as indicated previously in Figs. 7.3 and 7.4). The other weld, the branches, and the chord
were not permitted to fracture. The model was loaded by applying uniform incremental displacements to the
ends of each branch. The applied load (P) was obtained by summing up nodal forces parallel to the branch at
the end of one branch, and multiplying by two to account for the half model. Chord deformation ( ) was
obtained as described in Section 7.3.2, with the vertical displacement taken between nodes on two branches at
50 mm from the crown (Fig. 7.7a, shown previously). No axial load was applied to the chord.
7.6. RESULTS AND EVALUATION OF THE PARAMETRIC STUDY

All 256 FE tests failed by weld fracture and the branches of the connections always remained elastic.
Fracture initiated in the weld (evidenced by killed elements) at the saddle point (ρ = 90° in Fig. 7.6b) and
propagated away from the saddle towards the crown.
The FE chord deformation at weld rupture ( FE) was compared to the 3%D chord plastification limit,
which is internationally accepted as a good estimator of the deformation-based capacity of ductile CHS
connections (Lu et al. 1994). The ratio of FE/0.03D ranged from 0.10 to 2.58, and “chord plastification”
( FE/0.03D ≥ 1) was reached or surpassed in 97 out of the 256 FE tests. The weld sizes studied thus spanned a
broad range of welded joint situations, from well below to well beyond “connection strength”. Fig. 7.14
shows the relationship between , expressed as a fraction of the chord diameter ( /D), and the applied load,

normalized by the weld area and the electrode ultimate strength (P/AwFEXX), for typical tests. It is shown that
deformation ( /D) and strength (P/AwFEXX) varied widely across the FE tests at rupture. A complete list of
results from the parametric study are provided in Appendix F.1.
Fig. 7.14. Load-deformation curves for typical FE tests
To determine the weld effective length (le), first the strength of the fillet weld in each connection was
predicted using Eq. (7.11), which was developed from regression of a large database of weld-critical tests on
CHS connections that had fully effective fillet welds (Chapter 5):
 D   t 
Pnw  1.009  0.00137  b   0.197  w   Aw FEXX (7.11)
  tb   tb  
The fraction of the weld length that is effective (le/lw) was then found by dividing the weld FE strength by
the predicted weld strength from Eq. (7.11), corresponding to Method 2 in Chapter 6 (Section 6.6).
Determined in this manner, 138 (out of 256) FE connections had weld effective lengths less than 1.0. The
smallest weld effective length was 0.58 times the total weld length (for D/t = 50, = 0.50, = 1.0 and θ =
80°), and the largest weld effective length was theoretically 1.22 times the total weld length (for D/t = 50, =
0.10, = 0.2 and θ = 90°). The fact that weld effective lengths were sometimes greater than 1.0 is believed to
be due, in part, to the scatter associated with the best-fit prediction equation (Eq. 7.11), and also due to the
effect of secondary forces on the weld due to local bending of the chord.

7.6.1. EFFECT OF CHORD SLENDERNESS AND BRANCH-TO-CHORD DIAMETER RATIO

As D/t increases, CHS X-connections become more flexible, due to a decrease in bending stiffness of the
chord wall. Chord deformations at rupture become larger, and calculated weld effective lengths at rupture are
generally smaller. Fig. 7.15a shows the relationship between the calculated weld effective length, expressed
as a fraction of the total weld length (le/lw) and D/t. Fig. 7.15b shows the relationship between le/lw and . Fig.
7.16 shows the relationship between le/lw and the product (D/t) . For clarity, the results have been shown as
mean values with ± one standard deviation bars. Smaller scatter bars indicate a more statistically significant
correlation between le/lw and the independent variable. It can be seen that the quantity (D/t) has a strong
correlation to le/lw. For values of (D/t) greater than approximately eight, this relationship resembles a power
law. For values of (D/t) less than or equal to eight, le/lw plateaus (at a value greater than 1).
(a) Chord slenderness (b) Branch-to-chord diameter ratio
Fig. 7.15. Effect of chord slenderness and branch-to-chord diameter ratio on effective length

Fig. 7.16. Effect of the product (D/t) on effective length
7.6.2. EFFECT OF BRANCH INCLINATION ANGLE

Fig. 7.17 shows the relationship between θ and le/lw for connections with = 0.10, 0.20 and 0.30, since
connections with θ = 60° were only tested for up to 0.30. Fig. 7.17 indicates a marginal decrease in le/lw as θ
increases from 60° to 90°. The change in mean value over this range is only 3.5%. Overlapping of the ± one
standard deviation bars for all points indicates that the effect of θ is not statistically significant. This is also
indicated by the similar load-deformation curves in Fig. 7.14 for connections with θ = 60° and 90° and all
other parameters constant. The branch inclination angle is thus found to have no effect.
Fig. 7.17. Effect of branch inclination angle on effective length

7.6.3. EFFECT OF BRANCH-TO-CHORD THICKNESS RATIO

As increased in the experimental tests (in Chapter 6), a tendency for less stress re-distribution and a
lower average weld strength was observed. Fig. 7.18 shows the relationship between and le/lw for the FE
tests. As increases from 0.2 to 1.0, the average value of le/lw decreases. The relationship is approximately
linear.
Fig. 7.18. Effect of branch-to-chord thickness ratio on effective length
7.7. REGRESSION ANALYSIS

A non-linear regression analysis was performed with Eq. (7.12) as a basis:
le C
 b (7.12)
lw  (  ) a
where a, b, and C = regression constants, and = half-diameter-to-thickness ratio of the CHS chord (= D/2t).
The half-diameter-to thickness-ratio of the CHS chord is a standard parameter for HSS connection design.
The arrangement of variables in Eq. (7.12) was determined empirically, with efforts made to take into
account the relationships discussed in Section 7.6. The values of a, b, and C were determined by least-squares
regression of the 138 FE results with le/lw < 1, since above this value the relationship changes and it is
inapplicable for design. The “best-fit” equation is given by Eq. (7.13):
le 1.786
 (7.13)
lw  0.1007 (  )0.440

Eq. (7.13) gives a mean value of actual-to-predicted rupture strength of 1.001 over the 138 tests with a
COV of 0.030. Eq. (7.13) with = 0.2 and Eq. (7.13) with = 1.0 are plotted as dashed red lines on Fig. 7.16.
These two curves represent the extremes of the data, and also show that Eq. (7.12) as a basis for the
regression is appropriate because it describes the predominant data trends. For a lower-bound to Eq. (7.13),
the value of can be taken as 1.0 (the maximum in this study and a recommended maximum in practice)
which is synonymous with taking b equal to zero in Eq. (7.12). Eq. (7.14) is proposed as an even more
simple equation for determining the weld effective length (le) as a ratio of the total weld length in CHS X-
connections:
le 2
 1 (7.14)
lw 
Eq. (7.14) is plotted as a solid line in Fig. 7.16. It implies that weld effective lengths do not exist in CHS
X-connections when ≤ 4. Eq. (7.14), plotted in Fig. 7.19, embodies the trend established by Caulkins
(1968) and Marshall (1992) for the range of and D/t in this study (0.10 ≤ ≤ 0.50 and 10 ≤ D/t ≤ 50). As
and D/t increase, the load transfer efficiency (or for this study, le/lw) decreases. The rate of decrease is greater
for lower values of and D/t, when > 4.
Fig. 7.19. Weld effective lengths in CHS-to-CHS X-connections with 0.10 ≤ ≤ 0.50 and 10 ≤ D/t ≤ 50
according to Eq. (7.14)

7.8. EVALUATION OF DESIGN METHODS

7.8.1. FIT-FOR-PURPOSE DESIGN METHODS FOR FILLET WELDS
The nominal strength (Pnw) of fillet welds in CHS X-connections designed as fit-for-purpose is generally
given by:
Pnw  Fnwtwle (7.15)
where Fnw = nominal weld strength. An LRFD or LSD resistance factor for fillet welds, ϕ, is then applied to
determine ϕPnw. Note that Eqs. (7.1) and (7.2) can be written in this form.
7.8.1.1. Proposed Procedure

Step 1) Calculate Fnw using (1 – 0.25Pr/Py), where Pr = required strength of the weld and Py = branch
yield load (= AbFy). This equation is a simplification of Eq. (7.11), and is based on fillet-welded CHS-to-rigid
end-plate experimental and FE research (Chapters 4 and 5) in which the total weld length is effective.
Step 2) Determine the total weld length (lw) by an acceptable means (e.g. AWS D1.1-15 Clause 9.5.4, or
CAD).
Step 3) Determine the ratio le/lw using Eq. (7.14).
Step 4) Calculate le by multiplying the results of Steps 2) and 3).
Step 5) Calculate Pnw using Eq. (7.15), for a given value of tw.
Step 6) Apply the appropriate ϕ value to determine ϕPnw for design.
7.8.1.2. Safety Level Inherent in Proposed Procedure

To assess whether adequate or excessive safety margins are inherent, the structural reliability (or safety
index) ( +) can be calculated and compared to the minimum target value in North America (4.0, as currently
adopted by AISC (2016) per Section B3.1 of the AISC 360-16 Commentary), using the reliability analysis
discussed in Chapter 2 (Section 2.6.2).
The mean test-to-predicted capacity ratio (ρP) was taken as the average over all experimental and FE tests
of Paꞌ (or PFE) calculated using Steps 1) - 5), using measured (or actual) values of tw and FEXX. In Step 1), Pr
was taken as Paꞌ or PFE for a worst-case scenario. The total weld length (lw) was determined using a highly
accurate vector-calculus approach described in Chapter 6, and the ratio of le/lw was calculated using Eq. 7.14
with le/lw capped at a maximum of unity. The same values or equations used previously for the other
reliability analysis parameters (ρM, ρG, ρP, VM, VG, VP, and ϕ +, as discussed in Section 2.6.2) have also been
used herein (Table 7.5). A total of 268 tests (12 experimental plus 256 FE) were included in the reliability
analysis.

Table 7.5 gives the results of the reliability analysis using three different values of ϕ. The safety indices
calculated for the proposed design method for each value of ϕ can be compared to target values given in the
+ +
United States ( = 4.0, per Section B3.1 of the AISC 260-16 Commentary) and Canada ( = 4.5, per Annex
B of CSA S16-14).
The implied safety index ( +) ≥ 4.0 when ϕ = 0.80 (as used by AWS D1.1-15), +
≥ 4.0 when ϕ = 0.75 (as
used by AISC 360-16), and +
≥ 4.5 when ϕ = 0.67 (as used by CSA S16-14). This indicates that the method
meets U.S. and Canadian safety indices. Fig. 7.20 shows the correlation of the predicted nominal strengths
using Steps 1) – 5) for the proposed procedure with the experimental and FE results. On average, the test
capacity is only 1.12 times larger than the predicted rupture load, with a COV of 0.08.
Table 7.5. Reliability analysis parameters for proposed procedure with ϕ = 0.80, 0.75 and 0.67
ϕa 0.80 0.75 0.67
ρM 1.12 1.12 1.12
VM 0.12 0.12 0.12
ρG 1.03 1.03 1.03
VG 0.10 0.10 0.10
ρP 1.12 1.12 1.12
VP 0.08 0.08 0.08
ρR 1.29 1.29 1.29
VR 0.18 0.18 0.18
ϕ+ 0.91 0.89 0.84
+
4.0 4.3 5.0
a
resistance factors for fillet welds according to AWS (2015), AISC (2016), and CSA (2014) (see Chapter 6).
Fig. 7.20. Correlation of proposed design method with all test results
7.8.1.3. Safety Level Inherent in AWS (2015), AISC (2016) and CSA (2014)
The reliability analysis was repeated, using all 268 numerical and FE tests, to determine the implied
+
safety index, , for the current AWS D1.1-15 provisions, with and without the (2/3)lw weld effective length

factor, and the AISC 360-16 and CSA S16-14 provisions, as presented in Chapter 6. The analysis determined
that +
= 6.2 ≥ 4.0 for AWS (2015) with le = (2/3)lw, +
= 4.3 ≥ 4.0 for AWS (2015) with le = lw, and +
= 4.6
for both AISC (2016) and CSA (2014) > 4.0 and 4.5. A table summarizing the reliability analysis parameters,
and correlation plots for each method, is provided in Appendix F.2. It can therefore still be concluded that
weld effective lengths are never required for fillet-welded CHS X-connections in conjunction with these code
design methods. The mean test-to-predicted capacity ratio (ρP) and Vp for these methods (ρP = 2.07 for AWS
D1.1-15 with weld effective lengths, ρP = 1.38 for AWS D1.1-15 without weld effective lengths and AISC
360-16, ρP = 1.24 for CSA S16-14, and Vp = 0.19 for all methods) indicate that the proposed procedure
(Section 7.8.1.1) is much more accurate (ρP in Table 7.5 closer to 1) and precise (VP in Table 7.5 closer to 0)
than the code methods for predicting fillet weld strength in CHS X-connections.

7.9. SUMMARY
Non-linear FE models with weld fracture were developed for fillet-welded CHS X-connections. These
models were validated by comparison with 12 weld-critical tests on CHS X-connections. A parametric study
was then performed in which 256 FE models with 60° ≤ θ ≤ 90°, 10 ≤ D/t ≤ 50, 0.10 ≤ ≤ 0.50, and 0.20 ≤
≤ 1.00 were analysed. All models were shown to fail by weld fracture. Based on the parametric study, the
following can be deduced:
1. The weld effective length in CHS X-connections decreases as the branch slenderness (D/t),
branch-to-chord dimeter ratio ( ), and branch-to-chord thickness ratio ( ) increase.
2. The branch inclination angle θ has an insignificant effect on the weld effective length.
3. Theoretically, the weld effective length can be as low as 0.58 times the total weld length.
4. The weld length is 100% effective for ≤ 4.
5. For CHS X-connections load transfer peaks at the saddle position.
6. Stress re-distribution occurs prior to weld rupture, even in connections with small welds.
Based on a reliability analysis to determine the safety index ( +) for 268 experimental and FE tests
covering the same range of parameters, the following can also be deduced:
7. The existing AWS (2015) specification provisions for fillet welds in CHS X-connections in
+
Clause 9.6.1.3(4), with le = (2/3)lw, are very conservative (safety index, = 6.2 > 4.0).
8. The existing AWS (2015) specification provisions for fillet welds in CHS X-connections in
+
Clause 9.6.1.3(4), with le = lw, are also conservative (safety index, = 4.3 > 4.0).
9. The existing AISC (2016) provisions in Clause J2.4a and the existing CSA (2014) provisions in
+
Clause 13.13.2.2, with le = lw, are conservative (safety index, = 4.6 for both > 4.0 and 4.5).
The evaluations of AWS, AISC and CSA fillet weld design provisions assume that the (1+0.50sin 1.5θ)
directional strength-enhancement factor is not used (AWS D1.1-15 Clause 2.6.4.2, AISC 360-16 Clause
J2.4b, and CSA S16-14 Clause 13.13.2.2), because it has been shown, in Chapters 4 and 5, to be generally
unsafe for the design of fillet welds in HSS connections.
An alternative method for the design of fillet welds, based on rational weld effective lengths as a function
of non-dimensional connection parameters, was proposed. This method was shown be more accurate and
precise for predicting fillet weld strength in CHS X-connections than AWS (2015), AISC (2016) and CSA
(2014). A reliability analysis with respect to all experimental and FE tests showed that this proposed method
provides an adequate level of safety for use with AWS, AISC and CSA codes.

Chapter 8: CONCLUSIONS
8.1. CONCLUSIONS AND IMPACTS OF DESIGN RECOMMENDATIONS

Based on results from full-scale tests on weld-critical rectangular hollow section (RHS) overlapped K-
connections, and the measured strength of overlapped K-connection tests from a previous experimental
program (Frater 1991) (Chapter 2), it has been found that the effective length rules defined by Equations K4-
10 to K4-12 and given in Table K4.1 of the American Institute of Steel Construction (AISC) Specification for
Structural Steel Buildings (AISC 2010) for welds in RHS overlapped K-connections are quite conservative. It
has been recommended that a modification to these formulae be made. The modification, which has already
been adopted by AISC, in AISC 360 (2016), has established a more economical and yet still safe weld design
method for welds to rectangular HSS overlapped K-connections. This work has been published in Tousignant
& Packer (2014, 2015a, 2015b), and is cited, via Tousignant & Packer (2015b), in the most-recent (11th)
edition of the Canadian Institute of Steel Construction (CISC) Handbook of Steel Construction (CISC 2016),
in Part II: CISC Commentary on S16-14 and Part III: Connections and Tension Members. Recommendations
for overlap-jointed RHS truss analysis have also been made (Chapter 3). These are published in Tousignant &
Packer (2016a).
Based on the results from 33 large-scale experimental tests and subsequent FE analysis on HSS-to-rigid
plate connections, where the weld is fully effective, it has been proven that the (1.00+0.50sin1.5θ) directional
strength-increase factor for fillet welds is unsafe for hollow structural sections (HSS) (Chapters 4 and 5). A
recommendation to prohibit the use of the directional strength-increase factor for HSS was hence made. The
experimental work has been published in Packer et al. (2016) for all tests, and in Tousignant & Packer
(2016b) for tests on circular hollow sections (CHS). The numerical finite element (FE) work has been
published in Tousignant & Packer (2017a). This work is believed to extend to all single-sided fillet welds
(e.g. to plates as well as HSS). In light of this work, the AISC 360 Connections technical committee (TC6)
will be voting on a proposal to prohibit the directional strength increase factor in connections where local
bending of the weld is not restrained, including hollow structural member connections, for incorporation in
AISC 360-22. This motion has also been put forward to the CSA S16 committee, and is under consideration
for adoption in CSA S16-19. More accurate design formulae tailored for fillet welds to RHS and CHS have
also been proposed, and the fillet weld size required to develop the yield capacity of RHS and CHS branch
member walls – a highly controversial topic – has been determined.
147
Chapter 8: Conclusions 148
Fracture-critical welds in CHS-to-CHS connections have been studied for the first time, and it has been
determined that effective lengths therein are a function of branch-to-chord width ratio ( ), chord wall
slenderness (2 ), and branch-to-chord thickness ratio (Chapters 6 and 7). Welds are proven to be fully
effective for ≤ 4. A weld effective length design approach has been established to provide a more
economical and yet still safe weld design method for welds in CHS-to-CHS X-connections; however it is
unlikely that this approach will replace those given by the American Welding Society (AWS), AISC, and
CSA at this time. It has hence been shown that these current code methods provide sufficient reliability, even
without weld effective lengths, as long as the (1.00+0.50sin1.5θ) factor is not used. The experimental and FE
portions of the CHS-to-CHS X-connection test program have been submitted as papers for review
(Tousignant & Packer 2017b, 2017c, 2017d).
8.2. RECOMMENDATIONS FOR FUTURE RESEARCH

Although the scope of this work is expansive, and has covered welds for the connection of RHS and CHS
members to plates, and other HSS members, additional research in the following areas is suggested to
enhance the design recommendations made:
• The range of RHS overlapped K-connections in this study was limited, and covered only
connections with θbi = θbj = 60°. Although this is deemed a critical case by AISC, it is advisable
to evaluate and improve upon the weld effective length formulae for such connections over a
wider range of branch angles.
• Similarly, the previous work done on weld effective lengths in RHS-to-RHS moment T-
connections (McFadden & Packer 2014) could be expanded using FE analysis.
• As stated in Section 5.3.4, in spite of the importance of fracture to structural engineering, there is
no single widely accepted micro-mechanics-based model to predict ductile fracture in steel
components. Previous researchers have proposed several fracture models for monotonic and
cyclic loading [e.g. stress-modified critical strain model, void growth model (Kanvinde &
Deierlein 2006); cyclic void growth model (Kanvinde & Deierlein 2007); Xue model (Xue
2007); modified Xue model (Xue 2009); and simplified damage plasticity model (Ma et al.
2015)], but most of these models require complex calibration. They are therefore difficult to
implement in the analysis of large-scale tests. Confidence in modelling ductile fracture in such
tests could be increased by:
o Compiling/developing a database of large-scale fracture-critical experimental tests,
complete with ancillary coupon tests of the materials; and
o Evaluating the many models on the same set of tests, over a wide range of parameters, to
assess their relative practical merits.

Chapter 8: Conclusions 149
• Research on analytical methods for welds that explicitly take into account secondary bending
moments, including not only local eccentricity, but also the secondary moments produced by
rotation of the attached base metal would be highly informative. Limited data for simple tests on
fillet welds between plates at 90°, given by Chen et al (2001), could be used as a starting point;
however, more such tests on a wider range of plate and fillet weld parameters would be desirable.
Although this research is simple in theory, the results of it may help to further explain the trends
observed in the current work concerning relative weld strength with respect to weld size, branch
slenderness, branch-to-chord thickness ratio, and branch-to-chord width/diameter ratio.

REFERENCES
American Institute of Steel Construction (AISC) 2010. ANSI/AISC 360-10. Specification for structural steel
buildings. Chicago, IL, USA.
American Institute of Steel Construction (AISC) 2011. Steel construction manual, 14th ed., Chicago, IL, USA.
American Institute of Steel Construction (AISC) 2016. ANSI/AISC 360-16. Specification for structural steel
buildings. Chicago, IL, USA.
ASTM International (ASTM) 2013. ASTM A500-13. Standard specification for cold-formed welded and
seamless carbon steel structural tubing in rounds and shapes. West Conshohocken, PA, USA.
ASTM International (ASTM) 2015a. ASTM A1085-15. Standard specification for cold-formed welded steel
hollow structural sections (HSS). West Conshohocken, PA, USA.
ASTM International (ASTM) 2015b. ASTM E340-15. Standard test method for macroetching metals and
alloys. West Conshohocken, PA, USA.
ASTM International (ASTM) 2017. ASTM A370-17. Standard test methods and definitions for mechanical
testing of steel products. West Conshohocken, PA, USA.
American Welding Society (AWS). 2015. AWS D1.1/D1.1M:2015. Structural welding code – Steel, 23rd ed.,
Miami, FL, USA.
Björk, T., Toivonen, J. & Nykänen, T. 2012. Capacity of fillet welded joints made of ultra high-strength steel.
Welding in the World 56(3-4): 71-84.
Björk, T., Penttilä, T. & Nykänen, T. 2014. Rotation capacity of fillet weld joints made of high-strength steel.
Welding in the World 58(6): 853-863.
Boresi, A. P. & Schmidt, R.J. 2003. Advanced mechanics of materials, 6 th ed. NJ, USA: John Wiley and
Sons, Inc.
150
References 151
British Standards Institution (BSI) 1959. BS 449:1959. Specification for the use of structural steel in
building. London, England.
Butler, L. J. & Kulak, G. L. 1971. Strength of fillet welds as a function of direction of load. Welding Journal,
American Welding Society 50(5): 231-234.
Callele, L. J., Driver, R. G. & Grondin, G. Y. 2009. Design and behavior of multi-orientation fillet weld
connections. Engineering Journal, American Institute of Steel Construction 46(4): 257-272.
Canadian Institute of Steel Construction (CISC). 2016. Handbook of steel construction, 11 th ed., Toronto,
Canada: Canadian Institute of Steel Construction.
Canadian Standards Association (CSA) 2001. CAN/CSA S16-01. Limit states design of steel structures.
Toronto, Canada: Canadian Standards Association.
Canadian Standards Association (CSA) 2013a. CSA G40.20-13/G40.21-13. General requirements for rolled
or welded structural quality steel/structural quality steel. Toronto, Canada: Canadian Standards
Association.
Canadian Standards Association (CSA) 2013b. CSA W59-13. Welded steel construction (metal arc welding).
Toronto, Canada: Canadian Standards Association.
Canadian Standards Association (CSA) 2014. CSA S16-14. Design of steel structures. Toronto, Canada:
Canadian Standards Association.
Caulkins, D. W. 1968. CDG Report 15. Parameter study for FRAMETI elastic stress in tubular joints.
Houston, TX, USA: Shell Oil Company.
Chen, Y., Shen, Z., Zheng, Q. & Chen, C. 2001. Experimental study on the performance of single weld joints
in H-shaped steel members. International Journal of Steel Structures 1(3): 201-211.
Computers & Structures Inc. 2014. SAP2000 ver. 16.1. Walnut Creek, CA, USA.
Coutie, M. G. & Saidani, M. 1989. The use of finite element techniques for the analysis of RHS structures
with flexible joints. In Niemi, E. & Mäkeläinen, P. (Eds.) Tubular Structures III; Proc. Intern. Symp.,
Lappeenranta 1-2 September 1989. London, UK: Elsevier, 224-231.

References 152
Coutie, M. G. & Saidani, M. 1991. Comparison of the theoretical behaviour of two rectangular hollow
section trusses. In Wardenier, J. & Shahi, E. P. (Eds.) Tubular Structure IV; Proc. Intern. Symp., Delft 26-
28 June 1991. Delft, The Netherlands: Delft University Press, 334-343.
Czechowski, A., Gasparski, T., Zycinski, J., & Brodka, J. 1984. IIW Doc. XV-562-84. Investigation into the
static behaviour and strength of lattice girders made of RHS. Paris, France: International Institute of
Welding.
Davies, G. & Packer, J. A. 1982. Predicting the strength of branch plate - RHS connections for punching
shear. Canadian Journal of Civil Engineering 9(3): 458-467.
Deng, K., Grondin, G. Y. & Driver, R. G. 2006. Effect of loading angle on the behavior of fillet welds.
Engineering Journal, American Institute of Steel Construction 43(1): 9-23.
European Committee for Standardization (CEN) 2005. EN 1993-1-8:2005. Eurocode 3: Design of steel
structures – Part 1-8: Design of joints. Brussels, Belgium.
European Committee for Standardization (CEN). 2006. EN 10219-2:2006. Cold formed welded structural
hollow sections of non-alloy and fine grain steels: Tolerances, dimensions and sectional properties.
Brussels, Belgium.
Fisher, J. W., Galambos, T. V., Kulak, G. L. & Ravindra, M. K. (1978). Load and resistance factor design
criteria for connectors. Journal of the Structural Division, American Society of Civil Engineers 104(9):
1427-1441.
Fleischer, O. & Herion, S. 2015. Reduction of fillet weld sizes. In Batista, E., Vellasco, P. & Lima, L. (eds.),
Tubular Structures XV; Proc. Intern. Symp., Rio de Janeiro 27-29 May 2015. Rotterdam: Balkema, 473-
480.
Franchuk, C. R., Driver, R. G. & Grondin, G.Y. 2002. Block shear failure of coped steel beams. Proc. Annual
Conf. of the Canadian Society for Civil Engineering, Montreal, 5-8 June 2002. 1000-1009.
Frater, G. S. 1986. Weldment design for hollow structural section joints. M.A.Sc. thesis. Toronto, Canada:
University of Toronto.
Frater, G. S. 1991. Performance of welded rectangular hollow structural section trusses. PhD thesis. Toronto,
Canada: University of Toronto.

References 153
Frater, G. S. & Packer, J. A. 1992a. Weldment design for RHS truss connections. I: Applications. Journal of
Structural Engineering, American Society of Civil Engineers 118(10): 2784-2803.
Frater, G. S. & Packer, J. A. 1992b. Weldment design for RHS truss connections. II: Experimentation.
Journal of Structural Engineering, American Society of Civil Engineers 118(10): 2804-2820.
Frater, G. S., & Packer, J. A. 1992c. Modelling of hollow structural section trusses. Canadian Journal of
Civil Engineering 19(6): 947-959.
Hollomon, J. H. (1945). Tensile deformation. Transactions of the Metallurgical Society of American Institute
of Mining Metallurgical, and Petroleum Engineers 162: 268-290.
International Institute of Welding (IIW) 1980. IIW Doc. XV-467-80. Deformation curves of fillet welds.
London, England.
International Institute of Welding (IIW) 1989. IIW Doc. XV-701-89. Design recommendations for hollow
section joints – Predominatly statically loaded, 2nd ed., Paris, France.
International Institute of Welding (IIW) 2012. IIW Doc. XV-1402-12. Design recommendations for hollow
section joints – Predominatly statically loaded, 3rd ed., Paris, France.
International Organization for Standardization (ISO). 2013. ISO 14346:2013 (E). Static design procedure for
welded hollow section joints – Recommendations. Geneva, Switzerland.
Jensen, C. D. 1934. Combined stresses in fillet welds. Welding Journal, American Welding Society 13(2): 17-
21.
Kanvinde, A. M. 2017. Predicting fracture in civil engineering steel structures: State of the art. Journal of
Structural Engineering, American Society of Civil Engineers 143(3): 03116001-1 – 03116001-15.
Kanvinde, A. M. & Deierlein, G. G. 2006. Void growth model and stress modified critical strain model to
predict ductile fracture in structural steels. Journal of Structural Engineering, American Society of Civil
Engineers 132(12): 1907-1918.
Kanvinde, A. M. & Deierlein, G. G. 2007. Cyclic void growth model to assess ductile fracture initiation in
structural steels due to ultra low cycle fatigue. Journal of Structural Engineering, American Society of
Civil Engineers 133(6): 701-712.

References 154
Kanvinde, A. M., Grondin, G. Y., Gomez, I. R. & Kwan, Y. 2009. Experimental investigation of fillet-
welded joints subjected to out-of-plane eccentric loads. Engineering Journal, American Institute of Steel
Construction 46(3): 197-212.
Kato, B. & Morita, K. 1974. Strength of transverse fillet welded joints. Welding Journal, American Welding
Society 53(2): 59-64.
Kist, N. C. 1936. Berechnung der schweißnähte unter berücksichtigung konstanter gestaltänderungsenergie /
Calculation of welds under consideration of constant deformation energy. Vorbericht 2, Kongress Int. Ver.
für Brückenbau und Hochbau, Berlin, Germany. Proceedings, 511-522.
Lesik, D. F. & Kennedy, D. J. 1990. Ultimate strength of fillet welded connections loaded in plane. Canadian
Journal of Civil Engineering 17(1): 55-67.
Lie, S. T., Lee, C. K. & Wong S. M. 2001. Modelling and mesh generation of weld profile in tubular Y-joint.
Journal of Constructional Steel Research 57(5): 467-580.
Ling, Y. 1996. Uniaxial true stress-strain after necking. AMP Journal of Technology 5(1): 37-48.
Lu, L. H., de Winkel, G. D., Yu, Y. & Wardenier, J. 1994. Deformation limit for the ultimate strength of
hollow section joints. In Grundy, P., Holgate, A. & Wong, B. (Eds.), Tubular Structures VI; Proc. Intern.
Symp., Melbourne 14-16 December 1994. Rotterdam: Balkema, 341-348.
Luyties, W. H. & Post, J. W. 1988. Local dihedral angle equations for tubular joints and related applications.
Welding Journal, American Welding Society 77(4): 51-60.
Ma, X., Wang, W., Chen, Y. & Qian, X. 2015. Simulation of ductile fracture in welded tubular connections
using a simplified damage plasticity model considering the effects of stress triaxialty and Lode angle.
Journal of Constructional Steel Research 114: 217-236.
Marshall, P.W. 1992. Design of welded tubular connections – Basis and use of AWS code provisions.
Amsterdam, The Netherlands: Elsevier.
Martinez-Saucedo, G., Packer, J. A. & Willibald, S. 2006. Parametric finite element study of slotted end
connections to circular hollow sections. Engineering Structures 28(14): 1956-1971.
McFadden, M. R. & Packer, J. A. 2013. Effective weld properties for RHS-to-RHS moment T-connections.
Phase 1 Report to the American Institute of Steel Construction. Toronto, Canada: University of Toronto.

References 155
McFadden, M. R. & Packer, J. A. 2014. Effective weld properties for hollow structural section T-connections
under branch in-plane bending. Engineering Journal, American Institute of Steel Construction 51(4): 247-
266.
McFadden, M. R., Sun, M. & Packer, J. A. 2013. Weld design and fabrication for RHS connections. Steel
Construction – Design and Research 6(1): 5-10.
Mehrotra, B. L. & Govil, A. K. 1972. Shear lag analaysis of rectangular full-width tube connections. Journal
of the Structural Division, American Society of Civil Engineers 98(ST1): 287-305.
Miazga, G. S. & Kennedy, D. J. 1989. Behaviour of fillet welds as a function of the angle of loading.
Canadian Journal of Civil Engineering 16(4): 583-599.
Miller, D. K. 2002. Designing fillet welds for skewed T-joints – Part 1. Welding Innovation 19(1): 7-11.
Moore, A. M., Rassati G. A. & Swanson J. A. 2010. An experimental analysis of strength and ductility of
high strength fasteners. Engineering Journal, American Institute of Steel Construction 47(3): 161-174.
Ng, A. K. F., Deng, K., Grondin, G. Y. & Driver, R. G. 2004a. Behavior of transverse fillet welds:
experimental program. Engineering Journal, American Institute of Steel Construction 41(2): 39-54.
Ng, A. K. F., Driver, R. G. & Grondin, G. Y. 2004b. Behavior of transverse fillet welds: parametric and
reliability analyses. Engineering Journal, American Institute of Steel Construction 41(2): 55-67.
Oatway, P. 2014. Fillet-welded end-plate connections to square and rectangular HSS. M.Eng. thesis. Toronto,
Canada: University of Toronto.
Packer, J. A. & Cassidy, C. E. 1995. Effective weld length for HSS T, Y, and X connections. Journal of
Structural Engineering, American Society of Civil Engineers 121(10): 1402-1408.
Packer, J. A., Choo, Y. S., Shen, W., Wardenier, J., van der Vegte, G. J., & Mustard, T. 2012. CIDECT
Report 5BW-2/12. Axially loaded T and X joints of elliptical hollow sections. Geneva, Switzerland:
CIDECT.
Packer, J. A. & Davies, G. 1989. On the use and calibration of design standards for SHS joints. The
Structural Engineer 67 (21): 377-386.
Packer, J. A. & Frater, G. S. 2005. Recommended effective throat sizes for flare groove welds to HSS.
Engineering Journal, American Institute of Steel Construction 42(1): 31-44.

References 156
Packer, J. A. & Henderson, J. E. 1992. Design guide for hollow structural section connections, 1st ed.,
Toronto, Canada: Canadian Institute of Steel Construction.
Packer, J. A. & Henderson, J. E. 1997. Hollow structural section connections and trusses – A design guide,
2nd ed., Toronto, Canada: Canadian Institute of Steel Construction.
Packer, J. A., Sherman, D. R. & Leece, M. 2010. Hollow structural section connections, AISC Steel Design
Guide No. 24. Chicago, IL, USA: American Institute of Steel Construction.
Packer, J. A. & Sun, M. 2011. Weld design for rectangular HSS connections. Engineering Journal, American
Institute of Steel Construction 48(1): 31-48.
Packer, J. A., Sun, M., & Tousignant, K. 2016. Experimental evaluation of design procedures for fillet welds
to hollow structural sections. Journal of Structural Engineering, American Society of Civil Engineers
142(5): 0416007-1 – 04016007-12.
Packer, J. A., Wardenier, J., Kurobane, Y., Dutta, D. & Yeomans, N. 1992. Design guide for rectangular
hollow section (RHS) joints under predominantly static loading, CIDECT Design Guide No. 3, 1st ed.,
Köln, Germany: CIDECT & Verlag TÜV Rheinland GmbH.
Packer, J. A., Wardenier, J., Zhao, X. L., van der Vegte, G. J. & Kurobane, Y. 2009. Design guide for
rectangular hollow section (RHS) joints under predominantly static loading, CIDECT Design Guide No.
3, 2nd ed., Geneva, Switzerland: CIDECT.
Philiastides, A. 1988. Fully overlapped rolled hollow section welded joints in trusses. PhD Thesis.
Nottingham, UK: University of Nottingham.
Ravindra, M. K. & Galambos, T. V. 1978. Load and resistance factor design for steel. Journal of the
Structural Division. American Society of Civil Engineers 104(9): 1337-1353.
Swanson Analysis Systems. 2011. ANSYS ver. 14.0. Houston, TX, USA.
Tousignant, K. & Packer, J. A. 2014. Design of welds in rectangular HSS overlapped K-connections. Report
to the American Institute of Steel Construction. Toronto, Canada: University of Toronto, 283 p.
Tousignant, K. & Packer, J. A. 2015a. Weld effective lengths for rectangular HSS overlapped K-connections.
Engineering Journal, American Institute of Steel Construction 52 (4): 259-282.

References 157
Tousignant, K. & Packer, J. A. 2015b. Investigation of weld effective length rules for RHS overlapped K-
connections. In Batista, E., Vellasco, P. & Lima, L. (Eds.), Tubular Structures XV; Proc. Intern. Symp.,
Rio de Janeiro 27-29 May 2015. Rotterdam: Balkema, 357-364.
Tousignant, K. & Packer, J.A. 2016a. Analysis of rectangular hollow section trusses with overlapped K-
connections. 5th International Structural Specialty Conference, Canadian Society of Civil Engineering,
London, ON, 1-4 June 2016. Proceedings, 987-1 – 987-10.
Tousignant, K. & Packer, J. A. 2016b. Experimental evaluation of directional strength-enhancement factor
for fillet welds to CHS. Connections VIII; Proc. Intern. Workshop on Connections in Steel Structures,
Boston, 24-25 May 2016. Proceedings (AISC), 295-304.
Tousignant, K. & Packer, J. A. 2017a. Numerical investigation of fillet welds in HSS-to-rigid end-plate
connections. Journal of Structural Engineering, American Society of Civil Engineers 143(12): 04017165-1
– 04017165-16.
Tousignant, K. & Packer, J. A. 2017b. Fillet weld effective lengths in CHS X-connections. I: Experiments.
Journal of Constructional Steel Research 138: 420-431.
Tousignant, K. & Packer, J. A. 2017c. Fillet weld effective lengths in CHS X-connections. II: Finite element
modelling, parametric study and design. Journal of Constructional Steel Research, submitted for review.
Tousignant, K. & Packer, J. A. 2017d. Investigation of weld effective length rules for CHS X-connections.
Tubular Structures XVI; Proc. Intern. Symp., Melbourne 4-6 December 2017, accepted for publication.
van der Vegte, G.J. & Makino, Y. 2010. Further research on chord length and boundary conditions of CHS T-
and X-joints. Advanced Steel Construction 6(3): 879-890.
Voth, A. P. & Packer, J. A. 2012a. Branch plate-to-circular hollow structural section connections. I:
Experimental investigation and finite-element modeling. Journal of Structural Engineering, American
Society of Civil Engineers 138(8): 995-1006.
Voth, A. P. & Packer, J. A. 2012b. Branch plate-to-circular hollow structural section connections. II: X-type
parametric numerical study and design. Journal of Structural Engineering, American Society of Civil
Engineers 138(8): 1007-1018.

References 158
Vreedenburgh, G. G. J. 1954. New principles for the calculation of welded joints. Welding Journal, American
Welding Society 33(8): 743-751.
Wang, W., Gu, Q., Ma, X. & Wang, J. 2015. Axial tensile behavior and strength of welds for CHS branches
to SHS chord joints. Journal of Constructional Steel Research 115: 303-315.
Wardenier J., Kurobane, Y., Packer, J. A., van der Vegte, G. J. & Zhao, X. L. 2008. Design guide for circular
hollow section (CHS) joints under predominantly static loading, CIDECT Design Guide No. 1, 2nd ed.
Geneva, Switzerland: CIDECT.
Wardenier, J., and Packer, J. A. 1992. Connections between hollow sections. In Constructional Steel Design
– An International Guide. London, UK: Elsevier.
Xue, L. 2007. Damage accumulation and fracture initiation in uncracked ductile solids subject to triaxial
loading. International Journal of Solids and Structures 44(16): 5163-5181.
Xue, L. 2009. Stress based fracture envelope for damage plastic solids. Engineering Fracture Mechanics
76(3): 419-438.

WELD EFFECTIVE LENGTHS IN RHS
Appendix A provides the supplementary data for Chapter 2: Weld Effective Lengths in RHS Overlapped
K-Connections.
A.1. TRUSS FABRICATION DRAWINGS

A list of fabrication drawings for the RHS Warren truss, and the drawings themselves (in U.S. Customary
units), are provided herein.
Table A.1. List of truss fabrication drawings

Sheet No. Drawing No. Description
1 T-1 Truss elevation, joint layout and typical connection detail
2 F-1 Web members 11-1, 1-4, 4-5 and 5-8 cut details (according to drawing T-1)
3 F-2 Web members 8-9, 9-7, 7-6 and 6-3 cut details (according to drawing T-1)
4 F-3 Web member 3-2 and 2-12 cut details (according to drawing T-1)
5 F-4 Upper reinforcement details
6 F-5 Lower reinforcement details
7 F-6 Weld-critical test joints 1, 2, 5, 6 and 9 (according to drawing T-1)
8 F-7 Weld-critical test joints 3, 4, 7 and 8 (according to drawing T-1)
159
Appendix A: Weld Effective Lengths in RHS Overlapped K-Connections 160
Fig. A.1. Truss elevation, joint layout and typical connection detail (drawing T-1)

Fig. A.2. Web members 11-1, 1-4, 4-5 and 5-8 cut details (according to drawing T-1)

Fig. A.3. Web members 8-9, 9-7, 7-6 and 6-3 cut details (according to drawing T-1)

Fig. A.4. Web member 3-2 and 2-12 cut details (according to drawing T-1)

Fig. A.5. Upper reinforcement details

Fig. A.6. Lower reinforcement details

Fig. A.7. Weld-critical test joints 1, 2, 5, 6 and 9 (according to drawing T-1)

Fig. A.8. Weld-critical test joints 3, 4, 7 and 8 (according to drawing T-1)

A.2. MEASURED WELD DIMENSIONS FOR FILLET AND PJP WELDS IN RHS
The measured weld dimensions for each of the fillet and partial joint penetration (PJP) welds in the nine
RHS overlapped K-connection test joints are provided below. The average values along each weld element
are also given. Note that the apparent fillet weld leg along the overlapping branch (Fig. A.9a) is provided for
the fillet welds in Tables A.2 – A.10. The actual leg dimension (lv) is equal to the values shown multiplied by
1/sinѰ. The local dihedral angle (Ѱ) = 90° for weld elements a and b, and Ѱ = 60° for weld elements c and d.
Fig. A.9. Weld dimensions for (a) fillet weld elements and (b) PJP weld elements
Table A.2. Weld measurements for test K-90-0.50a

Weld element Apparent lv (mm) Apparent lh (mm) tw (mm)
a 4.78 6.20 3.45
Average 4.78 6.20 3.45
a’ PJP PJP 3.33
a’ PJP PJP 3.18
a’ PJP PJP 3.18
a’ PJP PJP 3.02
Average PJP PJP 3.18
b 5.08 8.23 3.12
Average 5.08 8.23 3.12
b’ PJP PJP 3.66
b’ PJP PJP 3.18
b’ PJP PJP 3.18
b’ PJP PJP 4.29
c 4.37 4.50 3.68
c 6.05 4.34 3.89
c 4.85 4.42 3.76
c 4.88 5.33 3.91
c 4.88 4.42 3.58
Average 5.00 4.60 3.76
d 6.02 5.41 4.19
d 4.85 4.52 3.91
d 5.13 4.34 3.81
d 5.99 7.39 5.13
d 4.67 5.97 4.29
Average 5.33 5.54 4.27

Table A.3. Weld measurements for test K-60-0.50

a 4.24 4.78 3.25
a 2.51 4.72 2.08
Average 3.38 4.75 2.67
a’ PJP PJP 3.33
a’ PJP PJP 3.51
a’ PJP PJP 4.60
b 2.77 6.35 2.51
b 2.21 4.62 2.26
Average 2.51 5.49 2.39
b’ PJP PJP 3.18
b’ PJP PJP 3.18
b’ PJP PJP 4.29
c 3.73 3.99 3.53
c 4.29 3.38 3.40
c 4.39 3.86 3.68
c 6.20 5.89 4.93
c 4.52 4.39 3.76
Average 4.65 4.29 3.86
d 4.52 4.19 3.48
d 5.77 5.13 4.47
d 5.49 6.60 4.29
d 5.66 5.36 4.24
d 6.48 5.94 4.57
Average 5.56 5.46 4.22

a 3.89 3.99 3.18
Average 3.89 3.99 3.18
a’ PJP PJP 3.81
a’ PJP PJP 3.51
a’ PJP PJP 3.81
a’ PJP PJP 4.45
b 3.71 4.17 3.18
Average 3.71 4.17 3.18
b’ PJP PJP 3.66
b’ PJP PJP 3.51
b’ PJP PJP 3.66
b’ PJP PJP 4.45
c 3.58 3.71 3.02
c 6.86 4.67 3.96
c 6.27 4.75 4.09
c 4.55 3.43 3.38
c 3.96 5.46 3.73
Average 5.03 4.39 3.63
d 4.17 5.69 4.06
d 3.38 4.52 3.71
d 3.68 4.62 3.61
d 3.78 3.53 3.25
d 6.40 5.84 4.57
Average 4.29 4.85 3.84


a 6.55 5.11 4.06
a 5.66 4.01 3.94
Average 6.10 4.57 3.99
a’ PJP PJP 3.18
a’ PJP PJP 3.02
a’ PJP PJP 4.29
b 5.21 4.88 3.91
b 5.46 5.18 3.81
Average 5.33 5.00 3.86
b’ PJP PJP 3.02
b’ PJP PJP 3.02
b’ PJP PJP 3.33
c 4.65 6.45 4.04
c 5.54 5.94 4.04
c 4.98 4.65 3.76
c 4.78 4.75 3.66
c 3.58 5.82 3.43
Average 4.70 5.51 3.78
d 4.17 5.31 3.78
d 4.78 3.86 3.35
d 5.64 5.77 4.42
d 4.34 6.15 3.86
d 5.59 5.05 3.91
Average 4.88 5.23 3.86
Table A.6. Weld measurements for test K-90-0.50b

a 6.93 4.85 4.60
Average 6.93 4.85 4.60
a’ PJP PJP 3.33
a’ PJP PJP 3.33
a’ PJP PJP 3.18
a’ PJP PJP 3.96
b 6.35 5.79 3.81
Average 6.35 5.79 3.81
b’ PJP PJP 3.33
b’ PJP PJP 3.33
b’ PJP PJP 3.66
b’ PJP PJP 4.29
c 4.17 4.39 3.51
c 4.17 6.55 3.99
c 5.69 6.86 3.84
c 5.26 5.13 3.99
c 4.17 6.50 3.84
Average 4.67 5.89 3.84
d 4.93 4.78 3.86
d 4.88 5.13 4.27
d 6.48 5.46 4.37
d 5.11 7.42 4.39
d 6.99 6.99 4.47
Average 5.66 5.97 4.27


a 4.88 6.15 2.92
a 3.86 4.04 3.20
a 5.33 5.33 3.91
a 3.91 4.39 3.43
Average 4.52 4.98 3.35
a’ PJP PJP 4.60
b 4.37 4.50 2.57
b 4.55 5.00 2.69
b 4.88 5.92 3.15
b 5.69 6.05 3.43
Average 4.88 5.36 2.95
b’ PJP PJP 3.96
c 6.63 5.08 4.22
c 5.77 4.83 4.34
c 5.18 5.33 4.52
c 7.44 6.32 4.37
c 7.59 4.90 4.24
Average 6.50 5.28 4.34
d 6.17 5.18 3.61
d 5.82 4.98 4.45
d 4.65 3.86 3.38
d 4.60 4.19 3.61
d 4.85 3.86 3.10
Average 5.21 4.42 3.63

a 2.72 4.37 2.26
a 5.11 6.48 4.60
Average 3.89 5.41 3.43
a’ PJP PJP 3.02
a’ PJP PJP 3.02
a’ PJP PJP 4.60
b 3.07 3.99 2.49
b 4.32 7.52 3.94
Average 3.68 5.74 3.23
b’ PJP PJP 3.96
b’ PJP PJP 3.02
b’ PJP PJP 4.29
c 4.27 3.89 3.38
c 3.73 5.18 4.88
c 3.53 5.97 3.43
c 3.02 3.56 2.84
c 5.82 6.43 4.62
Average 4.06 5.00 3.84
d 4.22 4.78 3.73
d 5.21 4.32 3.71
d 6.83 6.10 4.75
d 4.50 4.06 3.63
d 6.45 4.47 3.99
Average 5.44 4.75 3.96


a 5.51 6.48 4.42
a 5.18 6.35 4.24
a 6.05 7.65 4.98
a 6.65 6.63 4.65
Average 5.84 6.78 4.57
a’ PJP PJP 4.93
b 2.69 7.77 2.31
b 5.31 5.89 3.96
b 5.31 6.71 4.27
b 2.92 4.45 3.05
Average 4.06 6.20 3.40
b’ PJP PJP 4.78
c 6.10 5.18 4.11
c 6.22 5.87 4.37
c 6.48 5.36 3.58
c 3.43 5.46 4.95
c 6.38 6.48 4.24
Average 5.72 5.66 4.27
d 6.91 5.31 4.80
d 3.86 4.04 3.05
d 5.56 5.16 4.09
d 4.55 4.85 3.89
d 3.38 4.83 3.10
Average 4.85 4.85 3.78

a 5.77 7.26 4.60
a 4.55 4.57 3.25
a 3.58 4.39 2.67
a 2.87 4.34 2.54
Average 4.19 5.16 3.28
a’ PJP PJP 4.29
b 4.60 5.64 3.71
b 4.47 4.75 2.87
b 3.68 4.75 3.10
b 3.71 4.27 2.54
Average 4.11 4.85 3.05
b’ PJP PJP 4.29
c 5.23 5.33 3.96
c 5.28 4.52 3.89
c 5.00 4.09 3.91
c 5.46 6.02 4.19
c 5.92 4.80 4.17
Average 5.36 4.95 4.01
d 4.45 4.85 3.48
d 4.95 3.73 3.30
d 5.99 4.67 4.09
d 5.26 4.34 3.73
d 3.51 4.01 3.10
Average 4.83 4.34 3.53

A.3. TENSILE COUPON TEST RESULTS FOR RHS AND WELD METAL
Results of tensile coupon tests conducted for the RHS and weld metal in Chapter 2: Weld Effective
Lengths in RHS Overlapped K-Connections are given below.
(a) HSS 254.4 × 254.4 × 9.24 (heat no. 813767) (b) HSS 178.7 × 178.7 × 12.53 (heat no. L62502)
(c) HSS 127.0 × 127.0 × 7.78 (heat no. L62680)
Fig. A.10. TC test results for RHS branch and chord materials
Fig. A.11. All-weld-metal TC test results (E71T-1C electrode)

A.4. PHOTOGRAPHS OF TRUSS SUPPORTS AND POINT LOAD DEVICE

The truss in-plane and out-of-plane supports, and the 2700-kN capacity MTS ram head used to apply the
point load to panel points are depicted in the photographs below.
(a) Point load device
(b) Roller end support (c) Out-of-plane compression chord support
Fig. A.12. Truss load point, reaction, and out-of-plane support

A.5. PHOTOGRAPHS OF TEST CONNECTIONS AND FAILURE MODES

This sub-appendix contains photographs of all nine tests on welds to RHS overlapped K-connections with
instrumentation immediately prior to testing, and after testing, illustrating the weld rupture failure mode
achieved for all nine connections.
K-90-0.50 (a) (b) K-60-0.50 (c) K-90-0.71
(d) K-60-0.71a (e) K-90-0.50b (f) K-30-0.50a
(g) K-60-0.71b (h) K-30-0.71 (i) K-30-0.50b
Fig. A.13. Weld-critical overlapped K-connections with instrumentation immediately prior to testing

K-90-0.50 (a) (b) K-60-0.50 (c) K-90-0.71
(d) K-60-0.71a (e) K-90-0.50b (f) K-30-0.50a
(g) K-60-0.71b (h) K-30-0.71 (i) K-30-0.50b
Fig. A.14. Weld-critical overlapped K-connections with instrumentation immediately after testing

A.6. VARIATIONS IN STRAIN AROUND THE BRANCH PERIMETER FOR

ADDITIONAL RHS OVERLAPPED K-CONNECTIONS
The figures below show the variation in strain measured by 13 strain gauges (SGs) adjacent to the test
welds at the initial unloaded stage and at 50%, 80% and 100% of the weld rupture load. The results for six of
the nine connections are shown below; the remaining three have been included in Section 2.5 as typical
examples.
Fig. A.15. Distribution of normal strain around branch perimeter for test K-90-0.50a
Fig. A.16. Distribution of normal strain around branch perimeter for test K-30-0.50b

Fig. A.17. Distribution of normal strain around branch perimeter for test K-90-0.71
Fig. A.18. Distribution of normal strain around branch perimeter for test K-60-0.71a

Fig. A.19. Distribution of normal strain around branch perimeter for test K-60-0.71b
Fig. A.20. Distribution of normal strain around branch perimeter for test K-30-0.71

ANALYSIS OF RHS TRUSSES WITH
Appendix B provides the supplementary data for Chapter 3: Analysis of RHS Trusses with Overlapped
K-Connections.
B.1. MEASUREMENTS AND SECTION PROPERTIES FOR RHS TRUSS

MEMBERS
Appendix B.1 provides additional measurements and calculated section properties for the rectangular
hollow section (RHS) truss members.
Table B.1. Measured RHS section properties

RHS Designation Average of measured values
B and H t (or tb) ro ri slice weight slice thickness
(or Bb and Hb)
mm mm mm mm g mm
HSS 254 × 254 × 9.5 254.4 9.24 24.56 14.65 1758.9 25.46
HSS 178 × 178 × 13 178.7 12.53 34.97 23.41 1550.7 25.41
HSS 127 × 127 × 8.0 127.0 7.78 15.84 8.98 727.3 25.56
Table B.2. Calculated RHS section properties

RHS Designation Calculated values per EN10219-2:2006
A (or Ab) a I S r Z
mm2 × 106 mm4 × 103 mm3 mm × 103 mm3
HSS 254 × 254 × 9.5 8804 85.7 673 98.6 792
HSS 178 × 178 × 13 7776 33.9 380 66.0 469
HSS 127 × 127 × 8.0 3626 8.25 129.9 47.4 157.0
Note: the values above have been calculated according to the section property formulae of EN10219-2 (2006).
a
Cross-sectional areas of the branch and chord members were obtained by weighing a slice of the section, and
calculating the areas using the slice weight and slice thickness (Table A.1), with a density of steel = 7850 kg/m3 (CISC
2016).
180
Appendix B: Analysis of RHS Trusses with Overlapped K-Connections 181
B.2. SECTION PROPERTY FORMULAE ACCORDING TO EN10219-2 (2006)

The following formulae were used to calculate the RHS section properties from the measured dimensions in
the previous sub-appendix. Typical equations are shown for the chord subject to major axis bending. Branch
properties can be calculated by substituting the variables Bb, Hb and tb for B, H and t. These variables are
defined in the Symbols and Abbreviations section of the thesis. For minor axis bending, interchange section
widths (e.g. B) with section heights (e.g. H).
BH 3  B  2t  H  2t 
3
I   4  I g  Ag hg2   4  I   A h2  (B.1)

12 12
2I
S (B.2)
H
I
r (B.3)
A
BH 2  B  2t  H  2t 
2
Z   4  Ag hg   4  A h  (B.4)
4 4
where:
1  1  4
I g      ro (B.5)
 3 16 3 12  3  
 
Ag  1   ro2 (B.6)
 4
H  10  3 
hg    ro (B.7)
2  12  3 
1  1  4
I      ri (B.8)
 3 16 3 12  3  

 
A  1   ri 2 (B.9)
 4
H  2t  10  3 
h    ri (B.10)
2  12  3 

B.3. TRUSS ELEMENT DESIGNATIONS AND LVDT PHOTOGRAPHS

In this sub-appendix, Fig. B.1 gives the truss member and panel point designations (for reference in
Sections B.4 and B.5), and Fig. B.2 shows the orientation of the LVDTs used to measure truss deflections.
e = eccentricity v v v v
Fig. B.1. Truss member and panel point designations used in Appendix B
(a) (b)
Fig. B.2. LVDTs transverse to the chord longitudinal axis (a) along the chord and (b) at the support

B.4. PANEL POINT DEFLECTION DATA

In the tables that follow (Appendix B.4), panel point deflections for the truss are measured from a straight
line drawn between supports to account for rigid-body motion; support deflections are hence 0 mm. A
negative value indicates a deflection of the panel point in the direction of gravity. Percentage error is taken as
[1 – (experimental/theoretical)] × 100%, with positive indicating that a conservative (over-)prediction is
made by the theoretical model. Percentage errors are shown only for the panel point with the maximum
deflection, as indicated by underlines in the tables. In the results shown below, tests 4.1, 4.3, and 5 were
deemed spurious and are hence not included in the analysis in Chapter 3. Refer to Fig. B.1 (Appendix B.3)
for panel point designations.
Table B.3. Panel point deflections in test 4.1 (applied load = 483.3 kN to panel point 4)
Deflection (mm)
Panel point number: 3 5 7 9 11 % error a
Experimental -2.789 -5.207 -4.887 -3.609 -2.106 -
Model RR -2.398 -4.366 -3.876 -2.612 -1.006 -19%
Model PR -2.53 -4.51 -3.991 -2.697 -1.055 -15%
Model R -2.559 -4.655 -4.107 -2.769 -1.059 -12%
Model P -2.857 -4.955 -4.299 -2.908 -1.137 -5%
a
Spurious results omitted from analysis in Chapter 3.
Deflection (mm)
Panel point number: 3 5 7 9 11 % error
Experimental -1.849 -4.496 -4.115 -2.715 -1.031 -
Model RR -2.429 -4.423 -3.927 -2.646 -1.019 -2%
Model PR -2.563 -4.569 -4.043 -2.732 -1.069 2%
Model R -2.592 -4.716 -4.161 -2.805 -1.073 5%
Model P -2.894 -5.020 -4.355 -2.946 -1.152 10%
Deflection (mm)
Panel point number: 3 5 7 9 11 % error a
Experimental -2.710 -4.943 -4.630 -3.324 -1.882 -
Model RR -2.408 -4.384 -3.892 -2.623 -1.010 -13%
Model PR -2.541 -4.528 -4.007 -2.708 -1.059 -9%
Model R -2.570 -4.674 -4.124 -2.780 -1.064 -6%
Model P -2.869 -4.975 -4.316 -2.920 -1.142 1%
a

Table B.6. Panel point deflections in test 10 (applied load = 490.4 kN applied to panel point 10)
Deflection (mm)
Experimental -0.927 -2.549 -3.907 -4.322 -2.316 -
Model RR -1.034 -2.634 -3.922 -4.435 -2.435 3%
Model PR -1.081 -2.723 -4.047 -4.599 -2.572 6%
Model R -1.081 -2.788 -4.141 -4.699 -2.567 8%
Model P -1.159 -2.929 -4.336 -5.003 -2.871 14%
Table B.7. Panel point deflections in test 6.1 (applied load = 496.5 kN applied to panel point 6)
Deflection (mm)
Experimental -2.352 -5.648 -6.489 -4.611 -1.719 -
Model RR -2.467 -6.062 -6.929 -4.803 -1.856 6%
Model PR -2.575 -6.188 -7.050 -4.923 -1.932 8%
Model R -2.598 -6.456 -7.367 -5.085 -1.955 12%
Model P -2.766 -6.769 -7.690 -5.279 -2.088 16%
Deflection (mm)
Experimental -1.787 -4.750 -6.612 -6.077 -2.441 -
Model RR -1.896 -4.815 -6.998 -6.113 -2.443 6%
Model PR -1.973 -4.926 -7.124 -6.258 -2.553 7%
Model R -1.992 -5.111 -7.408 -6.472 -2.569 11%
Model P -2.125 -5.305 -7.734 -6.784 -2.739 15%
Deflection (mm)
Experimental -2.326 -5.614 -6.456 -4.612 -1.711 -
Model RR -2.467 -6.026 -6.929 -4.803 -1.856 7%
Model PR -2.575 -6.188 -7.050 -4.923 -1.923 8%
Model R -2.598 -6.456 -7.367 -5.085 -1.955 12%
Model P -2.766 -6.769 -7.690 -5.279 -2.088 16%
Deflection (mm)
Panel point number: 4 6 8 10 % error
Experimental -2.989 -4.942 -5.999 -4.202 -
Model RR -2.513 -4.587 -5.844 -4.208 -3%
Model PR -2.578 -4.679 -5.954 -4.335 -1%
Model R -2.652 -4.845 -6.175 -4.444 3%
Model P -2.760 -5.003 -6.442 -4.695 7%

Deflection (mm)
Panel point number: 4 6 8 10 % error a
Experimental 0.164 -8.242 -6.273 -3.462 -
Model RR -4.202 -5.794 -4.579 -2.489 -42%
Model PR -4.314 -5.887 -4.662 -2.555 -40%
Model R -4.456 -6.154 -4.848 -2.624 -34%
Model P -4.707 -6.425 -5.006 -2.732 -28%
a
Deflection (mm)
Experimental -3.460 -6.178 -6.210 -3.634 -
Model RR -3.666 -6.520 -6.570 -3.650 5%
Model PR -3.752 -6.609 -6.667 -3.742 7%
Model R -3.866 -6.920 -6.950 -3.840 11%
Model P -4.009 -7.193 -7.222 -3.893 14%
Deflection (mm)
Experimental -3.585 -6.326 -6.353 -3.597 -
Model RR -3.647 -6.486 -6.540 -3.630 3%
Model PR -3.732 -6.575 -6.632 -3.723 4%
Model R -3.846 -6.884 -6.911 -3.820 8%
Model P -3.988 -7.156 -7.185 -3.963 12%

B.5. AXIAL LOAD DATA

Tables B.14 and B.16 summarize the experimental axial forces in each member obtained from strain
gauge measurements, and the theoretical (predicted) axial forces from each of the four models investigated in
tests 4.1 and 4.2, respectively. Note that in the test designation, the first number refers to the panel point
subjected to point loading (see Fig. B.1). The second number differentiates between the first and second test,
in chronological order. Tables B.15 and B.17 provide the mean percentage errors in axial force predictions by
locations of interest in each of the tests. Member designations are indicated in Fig. B.1 (Appendix B.3).
Table B.14. Axial forces by truss model in test 4.1

Theoretical axial force (kN)
Experimental
Member axial force (kN) Model RR Model PR Model R Model P
a
11 -195.2 -208.9 7% -205.9 5% -211.6 8% -224.3 13%
12 -354.9 -364.7 3% -364.6 3% -366.2 3% -376.3 6%
13 -253.5 -267.9 5% -268.5 6% -266.9 5% -268.4 6%
14 -149.7 -158.9 6% -159 6% -158.9 6% -160.4 7%
15 -52.1 -54.22 4% -53.67 3% -55 5% -56.09 7%
DD
1 351 354.5 1% 390.1 10% 377.1 7% 446.1 21%
2 -434.7 -419.6 -4% -431.4 -1% -420.7 -3% -436 0%
3 -107.2 -110.9 3% -114.2 6% -97.98 -9% -110.5 3%
4 80.4 91.90 13% 93.74 14% 95.11 15% 111 28%
5 -87.8 -98.98 11% -98.46 11% -105.8 17% -112.2 22%
6 100.3 110.1 9% 112.3 11% 110.7 9% 111.6 10%
7 -95.4 -105.6 10% -106.8 11% -109.1 13% -111.6 15%
8 95.1 103.8 8% 103.7 8% 109 13% 111 14%
9 -95.5 b -103.9 8% -106.9 11% -106.9 11% -110.5 14%
10 88.6 94.83 7% 104.4 15% 99.91 11% 112.7 21%
DD
17 390.5 418.5 7% 421.7 7% 410.5 5% 427.9 9%
18 301 318.8 6% 317.8 5% 320.3 6% 324.6 7%
19 200 212.2 6% 212.4 6% 212.5 6% 214.4 7%
20 100 106.7 6% 107.2 7% 106.1 6% 107.7 7%
Note: positive forces are tensile; negative forces are compressive. (Applied load = 483.3 kN)
a
Percentage error = [1 – (experimental/theoretical)] × 100% : positive indicates a conservative (over-)prediction
b
Indicated forces were derived from a strain gauge pair away from the centre of the member.
Table B.15. Mean percentage errors in axial force predictions in test 4.1
Mean percentage error
Location Model: RR PR R P
Top chord 5% 4% 5% 8%
Webs 7% 10% 8% 15%
Bottom chord 6% 6% 6% 7%
Overall 6% 8% 7% 11%
Overall standard deviation 3.7% 4.0% 6.2% 7.3%

Table B.16. Axial forces by truss model in test 4.2

Theoretical axial force (kN)
Experimental
Member axial force (kN) Model RR Model PR Model R Model P
11 -194.2 -211.6 8% a -208.584 7% -214.4 9% -227.2 15%
12 -341.8 -369.5 7% -369.353 7% -371.0 8% -381.2 10%
13 -250.4 -271.4 8% -272 8% -270.4 7% -271.9 8%
14 -145.7 -161.0 9% -161.073 10% -161.0 9% -162.5 10%
15 -50.03 -54.9 9% -54.3696 8% -55.7 10% -56.8 12%
DD
1 358.9 359.1 0% 395.1851 9% 382.0 6% 451.9 21%
2 -410.6 -425.1 3% -437.023 6% -426.2 4% -441.7 7%
3 -106.3 -112.3 5% -115.689 8% -99.3 -7% -111.9 5%
4 75.99 b 93.1 18% 94.96194 20% 96.3 21% 112.4 32%
5 -85.25 -100.3 15% -99.7435 15% -107.2 20% -113.7 25%
6 101.2 111.5 9% 113.7639 11% 112.1 10% 113.1 10%
7 -94.28 -107.0 12% -108.192 13% -110.5 15% -113.1 17%
8 93.4 105.2 11% 105.0518 11% 110.4 15% 112.4 17%
9 -95.6 b -105.3 9% -108.293 12% -108.3 12% -111.9 15%
10 88.8 b 96.1 8% 105.7609 16% 101.2 12% 114.2 22%
DD
17 380.2 424.0 10% 427.197 11% 415.9 9% 433.5 12%
18 296.8 323.0 8% 321.9426 8% 324.5 9% 328.8 10%
19 194.3 215.0 10% 215.1687 10% 215.3 10% 217.2 11%
20 95.9 108.1 11% 108.5974 12% 107.5 11% 109.1 12%
Note: Positive forces are tensile; negative forces are compressive. (Applied load = 489.6 kN)
a
Percentage error = [1 – (experimental/theoretical)] × 100% : positive indicates a conservative (over-)prediction
b
Indicated forces were derived from a strain gauge pair away from the centre of the member.
Table B.17. Mean percentage errors in axial force predictions in test 4.2
Mean percentage error
Location Model: RR PR R P
Top chord 8% 8% 9% 11%
Webs 9% 12% 11% 17%
Bottom chord 10% 10% 9% 11%
Overall 9% 11% 10% 14%
Overall standard deviation 3.9% 3.5% 6.0% 6.8%

B.6. BENDING MOMENT DATA FOR TEST 4.2

Fig. B.3 shows the experimental bending moment distribution obtained using strain gauges for test 4.2.
The experimental distribution of bending moments is compared to theoretical predictions provided by three
of the truss analysis models in Fig. B.4.
Units are in kNm ggggggggg
Fig. B.3. Experimental bending moment distributions for test 4.2

Fig. B.4. Comparison of experimental and theoretical bending moment distributions for test 4.2

FILLET WELDS IN HSS-TO-RIGID PLATE
Appendix C provides the supplementary data for Chapter 4: Fillet Welds in HSS-to-Rigid Plate
Connections: Experimentation.
C.1. FILLET WELD RESISTANCE ACCORDING TO EN1993-1-8 (2005) FOR

A SKEWED-T FILLET WELD
The derivation of fillet weld resistance (Vrw), and a subsequent simplification of the resistance for Ѱ = 90°
equal-legged fillet welds, loaded either longitudinally (θ = 0°) or transversely (θ = 90°), are shown below.
Fig. C.1. Stress components in the plane of the weld effective throat for Ѱ = θ = 90°
EN1993-1-8 gives the following as one possible design formula for the resistance of a fillet weld:
0.5
 2  3( 2   2 )   Fu / (  w M 2 ) (C.1)
The stress components on the weld can be written as follows:
191
Appendix C: Fillet Welds in HSS-to-Rigid Plate Connections: Experimentation 192
P cos 
  (C.2a)
t w lw
P sin  cos 
  (C.2b)
t w lw
P sin  sin 
  (C.2c)
t w lw
Substituting the above three equations into Eq. (C.1) gives:
P 0.5
sin 2  cos 2   3(sin 2  sin 2   cos 2  )   Fu / (  w M 2 )
t w lw (C.3a)
Fu t w lw Pnw
P   Vrw (C.3b)
(  w M 2 ) sin  cos   3(sin  sin   cos  ) 
2 2 2 2 2 0.5
M2
 
For the case of an equal-legged longitudinally-loaded (θ = 0°) fillet weld in a Ѱ = 90° connection ( = 45°),
Eq. (C.3b) simplifies to:
Fu
Vrw  ( )twlw (C.4a)
3 w M 2
For the same case (Ѱ = 90° and = 45°), but transversely-loaded (θ = 90˚), Eq. (C.3b) simplifies to:
Fu
Vrw  ( )twlw (C.4b)
2 w M 2

C.2. CHS-TO-RIGID PLATE FABRICATION DRAWINGS

A list of fabrication drawings for the CHS-to-rigid plate test specimens, and the drawings themselves, are
provided herein. Additional fabricated specimens, such as trial weld specimens and all-weld-metal tensile
coupon test specimens, are included on Sheet No. 3.
Table C.1. List of CHS-to-rigid plate test specimen fabrication drawings

1 1 Ninety degree CHS-to-rigid plate test specimens
2 2 Sixty degree CHS-to-rigid plate test specimens
3 3 All-weld-metal TC test specimen and trial weld specimens for CHS-to-rigid plate
tests

Fig. C.2. Ninety degree CHS-to-rigid plate test specimens

Fig. C.3. Sixty degree CHS-to-rigid plate test specimens

Fig. C.4. All-weld-metal TC test specimen and trial weld specimens for CHS-to-rigid plate tests

C.3. WELD MEASUREMENTS FOR CHS-TO-RIGID PLATE CONNECTIONS

Appendix C.5 contains the weld measurements for all 12 CHS-to-rigid plate connections. Each table
presents the location of the measurement (given by ρ, measured clockwise from either the heel of the
connection for θ = 60° connections, or the same location on a predetermined side for θ = 90° connections),
the weld leg dimensions (lv and lh), the local dihedral angle (Ѱ), the weld throat dimension (tw), the theoretical
inclination of the failure plane ( ), and the tributary weld length. Average values of lv, lh, and tw, and the total
weld length (lw), are also given. In the tables for θ = 60° CHS-to-rigid plate connections, the theoretical
loading angle of the fillet weld (θ) has also been included. For each measurement, this angle has been taken
as the integral average of the constantly varying loading angle over the tributary length (see Appendix C.9).
The average loading angle for each joint has been obtained by taking the sum of θ × tributary weld length (for
each measurement), and diving the total quantity by lw; average values of lv, lh and tw were obtained in the
same manner.
Table C.2. Weld measurements for CHS-to-rigid plate test no. 22

ρ Tributary weld Ѱ lv lh tw
length
° mm ° mm mm mm °
0 26.4 90 6.10 6.30 4.38 45.9
18 26.4 90 7.28 6.98 5.10 43.1
36 26.4 90 6.74 6.02 4.49 41.8
54 26.4 90 7.86 7.10 5.27 42.1
72 26.4 90 8.29 6.62 5.20 38.2
90 26.4 90 5.38 7.22 4.33 53.1
108 26.4 90 6.61 6.97 4.83 46.1
126 26.4 90 7.33 7.29 5.17 44.8
144 26.4 90 6.77 7.38 4.99 47.5
162 26.4 90 7.07 6.49 4.78 42.6
180 26.4 90 5.86 5.14 3.92 40.3
198 26.4 90 7.12 5.77 4.48 39.0
216 26.4 90 6.18 6.10 4.34 44.6
234 26.4 90 7.52 6.40 4.87 40.4
252 26.4 90 6.71 6.88 4.85 45.2
270 26.4 90 7.34 6.10 4.69 39.7
288 26.4 90 7.66 6.53 4.97 40.4
306 26.4 90 7.59 6.77 5.10 41.1
324 26.4 90 7.51 6.95 5.12 42.5
342 26.4 90 7.85 7.03 5.24 41.8
Total (lw): 528 Average: 7.04 6.60 4.81 -


length
0 26.4 90 8.34 7.98 5.76 43.7
18 26.4 90 8.40 7.72 5.69 42.6
36 26.4 90 8.60 7.85 5.80 42.4
54 26.4 90 9.41 8.86 6.45 43.3
72 26.4 90 9.20 8.86 6.38 43.9
90 26.4 90 9.32 8.65 6.34 42.9
108 26.4 90 10.72 8.90 6.85 39.7
126 26.4 90 9.58 8.90 6.52 42.9
144 26.4 90 9.58 8.86 6.50 42.8
162 26.4 90 9.65 8.86 6.53 42.6
180 26.4 90 9.65 10.13 6.99 46.4
198 26.4 90 9.88 9.45 6.83 43.7
216 26.4 90 10.84 10.25 7.45 43.4
234 26.4 90 10.72 10.20 7.42 43.3
252 26.4 90 10.72 10.00 7.36 42.6
270 26.4 90 10.52 10.13 7.30 43.9
288 26.4 90 9.34 9.87 6.79 46.6
306 26.4 90 9.46 9.87 6.83 46.2
324 26.4 90 9.41 8.74 6.40 42.9
342 26.4 90 9.41 8.61 6.35 42.5
Total (lw): 528 Average: 9.64 9.13 6.63 -

length
0 26.7 90 9.31 9.68 6.75 45.8
24 26.7 90 9.08 8.75 6.34 43.6
48 26.7 90 9.42 9.81 6.86 45.6
72 26.7 90 10.91 10.00 7.45 41.8
96 26.7 90 9.88 10.00 7.10 44.8
120 26.7 90 10.00 8.70 6.62 40.4
144 26.7 90 9.88 7.70 6.14 37.1
168 26.7 90 9.46 10.65 7.09 48.3
192 26.7 90 10.57 8.59 6.69 38.8
216 26.7 90 10.57 9.08 6.91 40.4
240 26.7 90 10.26 10.65 7.42 45.9
264 26.7 90 9.08 11.85 7.23 52.4
288 26.7 90 9.19 8.80 6.40 43.3
312 26.7 90 10.91 9.70 7.30 41.2
336 26.7 90 9.77 9.08 6.70 42.4
Total (lw): 401 Average: 9.89 9.54 6.87 -


length
0 26.7 90 11.14 9.27 7.19 39.1
24 26.7 90 11.65 11.41 8.19 44.1
48 26.7 90 11.63 11.23 8.14 43.5
72 26.7 90 10.66 11.10 7.69 46.2
96 26.7 90 15.24 11.59 9.28 36.8
120 26.7 90 10.98 11.05 7.79 45.2
144 26.7 90 10.95 10.82 7.77 44.1
168 26.7 90 10.30 11.82 7.77 48.9
192 26.7 90 10.85 12.08 8.07 48.1
216 26.7 90 10.78 11.31 7.85 46.0
240 26.7 90 12.09 11.26 8.29 42.6
264 26.7 90 10.00 12.24 7.80 50.4
288 26.7 90 11.16 11.69 8.14 45.9
312 26.7 90 10.58 12.18 8.07 48.5
336 26.7 90 10.42 11.08 7.64 46.4
Total (lw): 401 Average: 11.23 11.34 7.98 -

length
0 26.4 90 9.53 10.77 7.13 48.5
30 26.4 90 9.06 9.68 6.61 46.9
60 26.4 90 8.90 8.58 6.18 44.0
90 26.4 90 10.31 9.83 7.13 43.5
120 26.4 90 10.15 10.77 7.42 46.4
150 26.4 90 10.15 8.58 6.55 40.2
180 26.4 90 9.21 8.58 6.34 42.4
210 26.4 90 7.81 9.21 6.00 49.3
240 26.4 90 8.74 8.12 6.00 42.3
270 26.4 90 7.81 9.83 6.15 51.3
300 26.4 90 7.81 9.21 6.00 49.3
330 26.4 90 6.71 7.49 5.00 48.1
Total (lw): 318 Average: 8.85 9.22 6.38 -


length
0 26.4 90 10.12 7.56 6.06 36.8
30 26.4 90 11.61 8.93 7.08 37.6
60 26.4 90 8.65 8.40 6.03 44.2
90 26.4 90 10.87 9.48 7.14 41.1
120 26.4 90 10.93 9.10 6.99 39.8
150 26.4 90 10.30 7.61 6.12 36.5
180 26.4 90 7.93 8.24 5.71 46.1
210 26.4 90 7.96 8.45 5.79 46.7
240 26.4 90 6.75 7.55 5.03 48.2
270 26.4 90 8.44 8.60 6.02 45.5
300 26.4 90 7.44 8.25 5.53 48.0
330 26.4 90 9.81 7.24 5.83 36.4
Total (lw): 318 Average: 9.23 8.28 6.11 -

ρ Tributary weld Ѱ θ lv lh tw
length
° mm ° ° mm mm mm °
0 26.4 60.0 87.4 6.26 9.81 7.24 42.5
18 26.8 61.6 79.9 4.87 7.64 5.60 42.9
36 27.9 66.1 71.3 7.49 6.58 6.57 2.9
54 29.1 72.9 65.1 6.87 7.40 6.70 25.2
72 30.1 81.1 61.3 6.36 6.09 5.53 24.8
90 30.4 90.0 60.1 5.99 6.72 5.24 38.9
108 30.1 98.9 61.3 5.49 6.81 4.62 47.2
126 29.1 107.1 65.1 5.97 7.40 4.62 51.4
144 27.9 113.9 71.3 6.24 7.74 4.40 55.3
162 26.8 118.4 79.9 6.49 8.05 4.30 57.7
180 26.4 120.0 87.4 6.92 6.95 4.06 54.3
198 26.8 118.4 79.9 7.79 8.05 4.74 53.9
216 27.9 113.9 71.3 6.24 9.29 4.78 59.0
234 29.1 107.1 65.1 6.27 7.03 4.66 48.5
252 30.1 98.9 61.3 5.49 6.81 4.69 46.4
270 30.4 90.0 60.1 6.28 6.02 5.45 25.1
288 30.1 81.1 61.3 5.20 7.16 5.27 42.7
306 29.1 72.9 65.1 5.38 7.77 5.81 41.7
324 27.9 66.1 71.3 6.24 7.35 6.57 26.7
342 26.8 61.6 79.9 4.87 8.45 5.66 48.0
Total (lw): 569 Average: 70.3 6.13 7.41 5.32 -


length
0 26.4 60.0 87.4 7.64 6.38 6.37 3.6
18 26.8 61.6 79.9 7.52 6.65 5.82 28.9
36 27.9 66.1 71.3 7.14 7.66 6.56 31.1
54 29.1 72.9 65.1 6.72 7.6 4.93 49.6
72 30.1 81.1 61.3 6.83 7.23 5.26 43.4
90 30.4 90.0 60.1 6.02 7.51 4.62 52.0
108 30.1 98.9 61.3 5.38 6.65 4.51 47.3
126 29.1 107.1 65.1 6.12 8.1 5.72 45.1
144 27.9 113.9 71.3 6.60 8.24 6.30 40.1
162 26.8 118.4 79.9 7.03 9.51 6.23 49.1
180 26.4 120.0 87.4 5.71 8.62 5.56 49.8
198 26.8 118.4 79.9 7.23 9.28 6.56 45.0
216 27.9 113.9 71.3 6.33 9.92 5.69 55.0
234 29.1 107.1 65.1 7.06 8.44 5.18 52.2
252 30.1 98.9 61.3 6.41 6.11 4.44 43.4
270 30.4 90.0 60.1 6.55 6.28 4.69 41.7
288 30.1 81.1 61.3 6.35 6.83 5.03 42.6
306 29.1 72.9 65.1 7.70 8.4 6.58 38.4
324 27.9 66.1 71.3 9.29 8.45 8.40 6.3
342 26.8 61.6 79.9 8.26 6.84 6.83 2.7
Total (lw): 569 Average: 70.3 6.88 7.71 5.73 -

length
0 26.7 60.0 86.6 8.51 8.43 6.67 37.7
24 27.4 62.8 76.9 8.00 9.85 6.74 46.9
48 29.0 70.5 66.9 7.82 8.99 6.18 46.5
72 30.4 81.1 61.4 6.60 8.87 5.15 54.5
96 30.7 93.0 60.3 7.38 7.90 4.77 52.8
120 29.8 104.5 63.6 7.98 7.84 4.38 56.0
144 28.2 113.9 71.4 8.99 8.30 4.28 59.0
168 26.9 119.3 83.2 8.45 9.38 4.08 64.2
192 26.9 119.3 83.2 8.16 8.37 3.79 63.1
216 28.2 113.9 71.4 8.06 7.35 3.81 58.7
240 29.8 104.5 63.6 8.20 9.05 4.79 58.1
264 30.7 93.0 60.3 7.66 8.78 5.11 54.4
288 30.4 81.1 61.4 7.40 7.98 5.21 49.3
312 29.0 70.5 66.9 9.02 7.71 6.16 37.0
336 27.4 62.8 76.9 8.92 9.52 7.14 41.4
Total (lw): 432 Average: 70.3 8.06 8.54 5.21 -


length
0 26.7 60.0 86.6 13.49 11.07 2.15 78.8
24 27.4 62.8 76.9 13.37 9.74 3.22 70.7
48 29.0 70.5 66.9 10.12 10.75 2.21 78.1
72 30.4 81.1 61.4 9.40 11.14 8.32 41.7
96 30.7 93.0 60.3 9.66 10.58 7.58 44.2
120 29.8 104.5 63.6 10.12 12.38 7.24 54.2
144 28.2 113.9 71.4 10.44 11.94 6.16 58.9
168 26.9 119.3 83.2 8.59 12.71 6.34 60.1
192 26.9 119.3 83.2 8.67 11.91 6.18 58.7
216 28.2 113.9 71.4 9.85 13.37 7.24 57.2
240 29.8 104.5 63.6 9.20 10.36 7.51 43.5
264 30.7 93.0 60.3 9.60 11.66 7.80 48.0
288 30.4 81.1 61.4 10.31 9.89 9.10 23.1
312 29.0 70.5 66.9 12.27 10.18 9.80 15.7
336 27.4 62.8 76.9 12.82 10.00 10.00 0.0
Total (lw): 431.8 Average: 70.3 10.50 11.13 6.78 -

length
0 26.5 60.0 85.7 9.06 9.99 8.74 28.9
30 27.6 64.3 74.1 7.29 10.00 7.36 42.6
60 29.5 75.5 63.7 5.09 7.07 4.87 46.5
90 30.4 90.0 60.3 6.57 7.21 5.16 44.3
120 29.5 104.5 63.7 6.76 7.82 4.72 52.8
150 27.6 115.7 74.1 6.86 6.40 3.76 54.0
180 26.5 120.0 85.7 8.70 8.32 4.53 57.0
210 27.6 115.7 74.1 7.60 7.20 4.20 54.3
240 29.5 104.5 63.7 6.40 7.07 4.39 51.7
270 30.4 90.0 60.3 5.91 6.85 4.76 46.0
300 29.5 75.5 63.7 5.09 5.96 4.59 39.6
330 27.6 64.3 74.1 8.38 9.60 8.01 33.5
Total (lw): 342 Average: 70.3 6.93 7.75 5.39 -


length
0 26.5 60.0 85.7 7.98 9.58 7.49 38.6
30 27.6 64.3 74.1 6.93 9.31 6.23 48.0
60 29.5 75.5 63.7 6.43 7.03 4.73 47.7
90 30.4 90.0 60.3 6.28 6.64 4.78 44.0
120 29.5 104.5 63.7 6.16 7.60 4.90 49.9
150 27.6 115.7 74.1 5.87 7.45 3.99 57.6
180 26.5 120.0 85.7 5.52 6.55 3.79 54.6
210 27.6 115.7 74.1 6.61 6.63 3.79 55.1
240 29.5 104.5 63.7 6.18 6.55 4.35 48.4
270 30.4 90.0 60.3 5.52 6.01 4.36 43.5
300 29.5 75.5 63.7 6.52 5.60 4.20 41.4
330 27.6 64.3 74.1 10.82 9.27 7.40 37.0
Total (lw): 342 Average: 70.3 6.71 7.32 4.98 -

C.4. THEORETICAL THROAT DIMENSION OF A SKEWED-T FILLET WELD
Fig. C.5. Skewed-T fillet weld profile dimensions
The following derivation was performed in order to calculate the theoretical throat dimension (tw) of a
skewed-T fillet weld from the measured weld leg dimensions (lv and lh) and the local dihedral angle. The
result is a general equation applicable to all welded joints (not just tubular joints), which can be simplified to
produce more specific and practical formulae for design of skewed-T joints (i.e. joints with lv = lh), as given
by Miller (2002). A general equation such as the one provided below is not given therein. The dimensions
referenced below are shown in Fig. C.5 (Note: Ѱ = + ).
tw
 cos  (C.5)
lv
tw
 cos   cos(   ) (C.6a)
lh
tw
 cos cos   sin sin  (C.6b)
lh
Substituting Eq. (C.5) into Eq. (C.6b), expanding, and then simplifying:
2
tw t t 
 cos w  sin 1   w  (C.7a)
lh lv  lv 

tw t
2  cos w
 tw  lh lv (C.7b)
1   
 lv  sin
2 2
 tw  tw2 t 
2    2cos  cos 2   w  (C.7c)
t 
1  w    h   lv 
l lv lh
 lv  sin 2
2 2 2
t  t  t2 t 
sin   sin   w    w   2cos w  cos 2   w 
2 2 (C.7d)
 lv   lh  lv lh  lv 
 1 2cos cos 2   sin 2  

sin 2   tw2  2    (C.7e)
 lh lv lh lv2 
 1 2cos 1 
sin 2   tw2  2   2 (C.7f)
 lh lv lh lv 
sin 2 
tw2 
 1 2cos 1  (C.7g)
 2  2
 lh lv lh lv 
lv lh sin
 tw  (C.8)
l  lh2  2lv lh cos
2
v

C.5. MATERIAL PROPERTY TEST RESULTS

This sub-appendix provides the results of tensile coupon tests conducted for the circular hollow sections
(CHS), the weld metal, and the end plate used for tests 22 - 33 in Chapter 4.
(a) HSS 167.9 × 6.70 (heat no. 703172) (b) HSS 127.4 × 11.55 (heat no. 1421012)
(c) HSS 101.0 × 7.34 (heat no. 819720-1)
Fig. C.6. TC test results for CHS branch materials
Fig. C.7. All-weld-metal TC test results (E71T-1C- Fig. C.8. TC test results for end-plate material (heat
H8 electrode) no. 4506365)

C.6. PHOTOGRAPHS OF EXPERIMENTAL TESTS AND FAILURE MODES FOR

CHS-TO-RIGID PLATE TESTS
Appendix C.6 contains photographs of the experimental test set-ups and weld rupture failure modes for
all 12 experimental CHS-to-rigid plate tests (test nos. 22-33 in Chapter 4). For clarity in the photographs
depicting weld rupture failure, instrumentation has been moved out of the way, or removed.
(a) Experimental test set-up (b) Weld rupture failure
Fig. C.9. Photographs of test 22

(a) Experimental test set-up (post weld-rupture) (b) Weld rupture failure

(a) Weld rupture failure in test 26 (b) Weld rupture failure in test 27
Fig. C.13. Photographs of tests 26 and 27

(a) Experimental test set-up (showing re-welded connection (b) Weld rupture failure
on opposite side)


C.7. EXAMPLES OF WELD STRENGTH CALCULATION METHOD FOR

FILLET-WELDED CHS-TO-RIGID PLATE CONNECTIONS
The following tables illustrate the procedure used to calculate the nominal weld strength as the sum of the
strengths of “weld components” tributary to each of the 12 – 20 weld dimension measurements for CHS-to-
rigid plate test connections. The measurements referenced in the footnotes to the tables below are tabulated in
Table C.13.
Table C.14. Application of component approach to CSA S16-01 with and without the sinθ factor for test no.
33
ρ Weld rupture strength Base metal rupture strength CSA S16-01 component strength e
with sinθ a without sinθ b with lh c with lv d without sinθ with sinθ
° kN kN kN kN kN kN
0 113.9 76.0 96.4 63.9 63.9 63.9
30 96.7 65.7 97.3 57.6 57.6 57.6
60 76.1 53.4 78.7 57.2 53.4 57.2
90 78.2 55.7 76.7 57.6 55.7 57.6
120 78.8 55.3 85.1 54.8 54.8 54.8
150 61.9 42.1 77.9 48.8 42.1 48.8
180 57.6 38.5 65.9 44.2 38.5 44.2
210 58.8 40.0 69.3 54.9 40.0 54.9
240 70.0 49.1 73.3 55.0 49.1 55.0
270 71.3 50.8 69.4 50.7 50.7 50.7
300 67.5 47.4 62.7 58.0 47.4 58.0
330 114.8 78.0 96.9 89.9 78.0 89.9
f
Pnw = 631 693
a
Eq. (4.3b) with Aw = tw × tributary weld length for each component.
b
Eq. (4.3b) without the sinθ factor, with Aw = tw × tributary weld length.
c
Eq. (4.3a) with Am = lh × tributary weld length.
d
Eq. (4.3a) with Am = lv × tributary weld length.
e
Minimum of columns 3-5 for CSA S16-01 without the sinθ factor, or the minimum of columns 2,4 and 5 for CSA S16-
01 with the sinθ factor.
f
Sum of component strengths.

Table C.15. Application of component approach to AISC 360 (2016), CSA S16 (2014), and EN1993-1-8
(2005) for test no. 33
ρ AISC 360-16 CSA S16-14 EN1993-1-8 e
without sinθ a with sinθ b
without sinθ c with sinθ d
Directional method
° kN kN kN kN kN
0 68.1 102.0 76.0 113.1 74.4
30 58.8 86.6 65.7 94.1 58.3
60 47.8 68.1 53.4 72.7 46.3
90 49.9 70.0 55.7 74.3 48.9
120 49.5 70.6 55.3 75.4 47.4
150 37.7 55.4 42.1 60.3 35.0
180 34.5 51.6 38.5 57.2 32.9
210 35.8 52.7 40.0 57.2 33.7
240 44.0 62.7 49.1 66.9 42.4
270 45.5 63.9 50.8 67.8 44.7
300 42.5 60.5 47.4 64.6 42.8
330 69.9 102.8 78.0 111.8 75.6
f
Pnw = 584 847 652 915 582
a
Eq. (4.1b) with Aw = tw × tributary weld length.
b
Eq. (4.1c) with Aw = tw × tributary weld length, and θ for each component.
c
Eq. (4.4a) with Aw = tw × tributary weld length for each component, but omitting the sinθ and Mw factors.
d
Eq. (4.4a) with Aw = tw × tributary weld length, and θ for each component, and Eq. (4.4b) with θ1 = θ and θ2 = 90°.
e
Eq. (4.8e) with tw, lw = tributary weld length, θ and for each component.
f
Sum of component strengths.

C.8. APPROACH TO CALCULATING LOADING ANGLE FOR CHS-TO-PLATE

CONNECTIONS
The local loading angle of the weld (θ) at any point (ρ) along the length of intersection of the CHS and
the plate can be calculated using computer-programmed equations (e.g. in Matlab) as follows:
Step 1: Starting at a subtended angle (ρ) equal to the point of interest, compute the coordinates of the
branch-plate intersection at ρ – Δρ and ρ + Δρ using Eq. (C.9a). Eq. (C.9a) gives the position vector 𝑃⃗𝑖 at
point i along the branch-plate intersection, where i equals ρ – Δρ or ρ + Δρ. The notation [( ), ( ), ( )]
represents the three vector components in the branch coordinate system (shown in Fig. 6.5b, in Chapter 6).
 D  D 
Pi   lt  ,  b sin i  ,  b cos i   (C.9a)
  2   2 
where:
Db
lt  1  cos i  (C.9b)
2 tan 
The derivation of lt for CHS-to-plate connections is provided in Appendix C.9.

⃗ ) connecting these points: 𝑉
Step 2: Calculate the vector (𝑉 ⃗ = 𝑃⃗𝜌+𝛥𝜌 – 𝑃⃗𝜌−𝛥𝜌 . This is the approximation to
the weld axis at point 𝑃⃗𝜌 (the point of interest), which lies between 𝑃⃗𝜌+𝛥𝜌 and 𝑃⃗𝜌−𝛥𝜌 . The smaller Δρ is, the
closer the approximation will be to the actual weld axis.
Step 4: For axial load, the line of action of the applied force (𝐹 ) will fall along the branch axis. Thus:
F  1 ,  0 ,  0 (C.10)
Step 5: Using the dot product, calculate the angle between the line of action of the applied force and the
weld axis (θ):
 
V F 
  cos 1  (C.11)
V F 
 
Step 6: Increment ρ by Δρ, and repeat Steps 1 to 5 to determine the variation in θ with ρ. This variation
can then be transformed to determine the variation in θ with respect to the weld length by correlating the
point ρ to the subtended weld length. This approach is explained in Chapter 6 for CHS-to-CHS connections,
but for CHS-to-plate connections, it is analogous. The average loading angle (between any two points, a and
b) can be calculated by numerically integrating the resulting θ(lw) curve between a and b, and dividing by (b –

a), according to Eq. (C.12). This approach, as written, can also be applied to CHS-to-CHS connections, by
substituting the appropriate formula for lt.
1 b
b  a a
 avg   d (lw ) (C.12)

C.9. TEMPLATE LENGTH FORMULA FOR CHS-TO-PLATE CONNECTIONS
Fig. C.3. CHS-to-plate template geometry
The template length (lt) can be determined for an arbitrary point “A” along the intersection surface of the
tube and the plate as follows:
Step 1: Construct ΔABC (Fig. C.3), where AB is along the plate surface and ∠BAC is equal to the branch
inclination angle (θ). The length AC is equal to lt, and can be determined thus (as a function of the subtended
angle, ρ):
BC
lt  AC  (C.13)
tan 
D D  D
BC   b  b cos    b 1  cos   (C.14)
 2 2  2
Substitute Eq. (C.14) into Eq. (C.13):
Db
 lt  1  cos   (C.15)
2 tan 

C.10. SUMMARY OF PREDICTED WELD STRENGTHS FOR HSS-TO-RIGID
PLATE CONNECTIONS
The tables below gives the predicted weld strengths according to codes for all RHS- and CHS-to-rigid
plate connections discussed.
Table C.16. Summary of actual and predicted strengths for all experimental RHS- and CHS-to-rigid plate
connections according to AISC 360-16, CSA S16-14, and CSA S16-01
Pnw
Test no. AISC 360-16 CSA S16-14 CSA S16-01
Pa without sinθ with sinθ without sinθ with sinθ without sinθ with sinθ
kN kN (A/P) kN (A/P) kN (A/P) kN (A/P) kN (A/P) kN (A/P)
1 831 641 1.30 962 0.86 716 1.16 1074 0.77 716 1.16 840 0.99
2 1166 1056 1.10 1584 0.74 1179 0.99 1768 0.66 1179 0.99 1279 0.91
3 1235 945 1.31 1418 0.87 1055 1.17 1583 0.78 1055 1.17 1191 1.04
4 1311 1002 1.31 1503 0.87 1119 1.17 1679 0.78 1119 1.17 1252 1.05
5 2433 1583 1.54 2374 1.02 1768 1.38 2651 0.92 1768 1.38 2031 1.20
6 2574 2136 1.20 3204 0.80 2385 1.08 3578 0.72 2385 1.08 2847 0.90
7 2525 1742 1.45 2612 0.97 1945 1.30 2917 0.87 1945 1.30 2272 1.11
8 2302 1823 1.26 2735 0.84 2036 1.13 3054 0.75 2036 1.13 2318 0.99
9 1020 822 1.24 1234 0.83 918 1.11 1377 0.74 843 1.21 843 1.21
10 960 699 1.37 1048 0.92 781 1.23 1171 0.82 700 1.37 700 1.37
11 840 634 1.32 952 0.88 708 1.19 1063 0.79 621 1.35 621 1.35
12 1140 998 1.14 1497 0.76 1115 1.02 1672 0.68 1002 1.14 1002 1.14
13 1200 1198 1.00 1798 0.67 1338 0.90 2007 0.60 1289 0.93 1289 0.93
14 1207 1146 1.05 1719 0.70 1279 0.94 1919 0.63 1193 1.01 1193 1.01
15 1494 1624 0.92 2436 0.61 1813 0.82 2720 0.55 1719 0.87 1719 0.87
16 1578 1965 0.80 2948 0.54 2194 0.72 3291 0.48 2100 0.75 2100 0.75
17 1788 2119 0.84 3178 0.56 2366 0.76 3548 0.50 2366 0.76 2435 0.73
18 1131 1400 0.81 1872 0.60 1563 0.72 2207 0.51 1375 0.82 1460 0.77
19 982 826 1.19 1105 0.89 922 1.06 1303 0.75 879 1.12 972 1.01
20 1270 1220 1.04 1648 0.77 1362 0.93 1932 0.66 1215 1.05 1294 0.98
21 1534 2126 0.72 2846 0.54 2374 0.65 3353 0.46 2123 0.72 2250 0.68
22 1261 869 1.45 1303 0.97 970 1.30 1455 0.87 970 1.30 1224 1.03
23 1279 1197 1.07 1796 0.71 1337 0.96 2006 0.64 1337 0.96 1700 0.75
24 1459 942 1.55 1412 1.03 1051 1.39 1577 0.93 1052 1.39 1280 1.14
25 1597 1094 1.46 1641 0.97 1222 1.31 1833 0.87 1222 1.31 1450 1.10
26 841 693 1.21 1040 0.81 774 1.09 1161 0.72 774 1.09 847 0.99
27 864 664 1.30 997 0.87 742 1.16 1113 0.78 742 1.16 864 1.00
28 1450 1037 1.40 1501 0.97 1158 1.25 1621 0.89 1073 1.35 1169 1.24
29 1331 1117 1.19 1621 0.82 1248 1.07 1752 0.76 1222 1.09 1305 1.02
30 1109 771 1.44 1116 0.99 861 1.29 1205 0.92 861 1.29 1073 1.03
31 1479 1002 1.48 1445 1.02 1119 1.32 1556 0.95 1112 1.33 1251 1.18
32 776 632 1.23 918 0.85 706 1.10 993 0.78 640 1.21 714 1.09
33 803 584 1.38 847 0.95 652 1.23 915 0.88 631 1.27 693 1.16
ρP = 1.21 0.82 1.09 0.74 1.13 1.02
VP = 0.18 0.18 0.18 0.19 0.17 0.16
217
Appendix D: Fillet Welds in HSS-to-Rigid Plate Connections: Finite Element Modelling 218
Table C.17. Summary of actual and predicted strengths for RHS- and CHS-to-rigid plate connections
according to EN1993-1-8:2005 Directional Method
Test no. EN1993-1-8:2005 Directional Method
Pa Pnw (nominal strength) Vrw (design strength)
kN kN (A/P) kN (A/P)
1 831 670 1.24 536 1.55
2 1166 1063 1.10 850 1.37
3 1235 1059 1.17 847 1.46
4 1311 1132 1.16 906 1.45
5 2433 1661 1.46 1329 1.83
6 2574 2304 1.12 1844 1.40
7 2525 2175 1.16 1740 1.45
8 2302 2104 1.09 1683 1.37
9 1020 737 1.38 590 1.73
10 960 619 1.55 495 1.94
11 840 555 1.51 444 1.89
12 1140 856 1.33 685 1.66
13 1200 1065 1.13 852 1.41
14 1207 1002 1.20 802 1.51
15 1494 1432 1.04 1146 1.30
16 1578 1741 0.91 1392 1.13
17 1788 2019 0.89 1615 1.11
18 1131 1271 0.89 1017 1.11
19 982 793 1.24 635 1.55
20 1270 1182 1.07 946 1.34
21 1534 1877 0.82 1502 1.02
22 1261 1017 1.24 814 1.55
23 1279 1395 0.92 1116 1.15
24 1459 1069 1.36 855 1.71
25 1597 1226 1.30 981 1.63
26 841 710 1.18 568 1.48
27 864 706 1.22 565 1.53
28 1450 1198 1.21 958 1.51
29 1331 1339 0.99 1071 1.24
30 1109 811 1.37 649 1.71
31 1479 1120 1.32 896 1.65
32 776 644 1.20 515 1.51
33 803 582 1.38 466 1.72
ρP = 1.19 Mean = 1.48
VP = 0.155

FILLET WELDS IN HSS-TO-RIGID PLATE

Appendix D provides the supplementary data for Chapter 5: Fillet Welds in HSS-to-Rigid Plate
Connections: Finite Element Modelling.
D.1. SUPPLEMENTARY FIGURES

Appendix D.1 provides general supplementary figures.
Fig. D.1. FE CHS-to-rigid plate connection model with θ = 90°

Fig. D.2. FE RHS-to-rigid plate connection model with θ = 90°
Fig. D.3. Graphical comparison of FE versus predicted ultimate load (showing FE failure modes)

(a) Test with Bb/tb =12.5 and tw/tb = 0.90 (b) Test with Bb/tb =30.0 and tw/tb = 0.90
(a) Test with Bb/tb =50 and tw/tb = 0.90
Fig. D.4. Non-uniform FE strain distributions adjacent to the weld in RHS-to-rigid plate tests

(a) RHS-to-rigid plate tests (b) CHS-to-rigid plate tests
Fig. D.5. Comparison of Pa/Py and PFE/Py to tw/tb for FE and experimental HSS-to-rigid plate tests

D.2. SUMMARY OF ACTUAL AND PREDICTED STRENGTHS ACCORDING TO

CODE DESIGN METHODS FOR FE CHS- AND RHS-TO-RIGID PLATE TESTS
A summary of the actual and predicted strengths for weld-critical FE CHS- and RHS-to-rigid plate tests
is provided in the following tables.
Table D.1. Summary of actual and predicted nominal strengths for 21 FE RHS-to-rigid plate connections
according to AISC 360-16, CSA S16-14, and EN1993-1-8:2005 Directional and Simplified Methods (θ = 90°
connections)
Pnw
tw/tb Bb/tb AISC 360-16 CSA S16-14 EN 1993-1-8:2005
Pa without sinθ a with sinθ b without sinθ c with sinθ d Directional e Simplified f
kN kN (A/P) kN (A/P) kN (A/P) kN (A/P) kN (A/P) kN (A/P)
0.35 50.0 508 381 1.33 571 0.89 425 1.19 638 0.80 438 1.16 357 1.42
0.50 50.0 679 539 1.26 808 0.84 602 1.13 902 0.75 619 1.10 505 1.34
0.71 50.0 911 762 1.20 1143 0.80 851 1.07 1276 0.71 875 1.04 715 1.27
0.90 50.0 1049 970 1.08 1455 0.72 1083 0.97 1624 0.65 1114 0.94 910 1.15
1.06 50.0 1247 1143 1.09 1714 0.73 1276 0.98 1914 0.65 1313 0.95 1072 1.16
0.35 40.0 631 474 1.33 711 0.89 529 1.19 794 0.79 545 1.16 445 1.42
0.50 40.0 855 670 1.28 1006 0.85 749 1.14 1123 0.76 770 1.11 629 1.36
0.71 40.0 1147 948 1.21 1422 0.81 1059 1.08 1588 0.72 1089 1.05 890 1.29
0.90 40.0 1380 1207 1.14 1810 0.76 1348 1.02 2022 0.68 1387 1.00 1132 1.22
1.06 40.0 1578 1422 1.11 2134 0.74 1588 0.99 2382 0.66 1634 0.97 1334 1.18
0.35 30.0 855 628 1.36 942 0.91 701 1.22 1052 0.81 721 1.19 589 1.45
0.50 30.0 1142 888 1.29 1332 0.86 991 1.15 1487 0.77 1020 1.12 833 1.37
0.71 30.0 1546 1256 1.23 1883 0.82 1402 1.10 2103 0.74 1443 1.07 1178 1.31
0.90 30.0 1872 1598 1.17 2397 0.78 1785 1.05 2677 0.70 1836 1.02 1499 1.25
0.35 20.0 1291 927 1.39 1391 0.93 1036 1.25 1553 0.83 1066 1.21 870 1.48
0.50 20.0 1748 1312 1.33 1967 0.89 1465 1.19 2197 0.80 1507 1.16 1230 1.42
0.71 20.0 2338 1855 1.26 2782 0.84 2071 1.13 3107 0.75 2131 1.10 1740 1.34
0.35 12.5 2088 1444 1.45 2166 0.96 1612 1.29 2419 0.86 1659 1.26 1355 1.54
0.50 12.5 2836 2042 1.39 3063 0.93 2280 1.24 3420 0.83 2346 1.21 1916 1.48
0.35 9.1 2831 1931 1.47 2896 0.98 2156 1.31 3234 0.88 2218 1.28 1811 1.56
0.50 9.1 3804 2730 1.39 4095 0.93 3049 1.25 4573 0.83 3137 1.21 2561 1.49
ρP = 1.27 0.85 1.14 0.76 1.11 1.36
VP = 0.09 0.09 0.09 0.09 0.09 0.09
a
Eq. (4.1b) with Aw = tw[4(Bb-4tb)+4πtb].
b
Eq. (4.1c) with Aw = tw[4(Bb-4tb)+4πtb] and θ = 90°.
c
Eq. (4.4a) with Aw = tw[4(Bb-4tb)+4πtb], but omitting the sinθ and Mw factors.
d
Eq. (4.4a) with Aw = tw[4(Bb-4tb)+4πtb], and θ = θ1 = θ2 = 90°.
e
Eq. (4.6b).
f
Eq. (4.6a).

Table D.2. Summary of actual and predicted nominal strengths for 20 FE CHS-to-rigid plate connections
according to AISC 360-16, CSA S16-14, EN1993-1-8:2005 Directional and Simplified Methods (θ = 90°
connections)
Pnw
tw/tb Db/tb AISC 360-16 CSA S16-14 EN 1993-1-8:2005
Pa without sinθ a with sinθ b without sinθ c without sinθ a with sinθ b without sinθ c
kN kN (A/P) kN kN (A/P) kN kN (A/P) kN kN (A/P) kN
0.35 50.0 304 215 1.42 322 0.94 240 1.27 360 0.84 247 1.23 202 1.51
0.50 50.0 441 304 1.45 456 0.97 339 1.30 509 0.87 349 1.26 285 1.55
0.71 50.0 578 430 1.35 644 0.90 480 1.20 720 0.80 494 1.17 403 1.43
0.90 50.0 693 547 1.27 820 0.84 611 1.13 916 0.76 628 1.10 513 1.35
0.35 40.0 400 269 1.49 403 0.99 300 1.33 450 0.89 308 1.30 252 1.59
0.50 40.0 544 380 1.43 570 0.96 424 1.28 636 0.86 436 1.25 356 1.53
0.71 40.0 732 537 1.36 806 0.91 600 1.22 899 0.81 617 1.19 504 1.45
0.90 40.0 868 683 1.27 1025 0.85 763 1.14 1145 0.76 785 1.11 641 1.35
0.35 30.0 524 358 1.47 536 0.98 399 1.31 599 0.87 411 1.28 336 1.56
0.50 30.0 741 506 1.47 759 0.98 565 1.31 847 0.87 581 1.28 474 1.56
0.71 30.0 982 715 1.37 1073 0.92 799 1.23 1198 0.82 822 1.19 671 1.46
0.90 30.0 1176 910 1.29 1366 0.86 1017 1.16 1525 0.77 1046 1.12 854 1.38
0.35 20.0 804 537 1.50 806 1.00 600 1.34 899 0.89 617 1.30 504 1.60
0.50 20.0 1117 759 1.47 1139 0.98 848 1.32 1272 0.88 873 1.28 712 1.57
0.71 20.0 1520 1074 1.42 1611 0.94 1199 1.27 1799 0.84 1234 1.23 1008 1.51
0.35 12.5 1329 859 1.55 1289 1.03 959 1.39 1439 0.92 987 1.35 806 1.65
0.50 12.5 1805 1215 1.49 1823 0.99 1357 1.33 2035 0.89 1396 1.29 1140 1.58
0.71 12.5 2499 1718 1.45 2578 0.97 1919 1.30 2878 0.87 1974 1.27 1612 1.55
0.35 9.1 1817 1181 1.54 1772 1.03 1319 1.38 1979 0.92 1357 1.34 1108 1.64
0.50 9.1 2495 1671 1.49 2506 1.00 1866 1.34 2799 0.89 1920 1.30 1567 1.59
ρP = 1.43 0.95 1.28 0.85 1.24 1.52
VP = 0.06 0.06 0.06 0.06 0.06 0.06
a
Eq. (4.1b) with Aw = twπDb.
b
Eq. (4.1c) with Aw = twπDb and θ = 90°.
c
Eq. (4.4a) with Aw = twπDb, but omitting the sinθ and Mw factors.
d
Eq. (4.4a) with Aw = twπDb, and θ = θ1 = θ2 = 90°.
e
Eq. (4.6b).
f
Eq. (4.6a).

FILLET WELD EFFECTIVE LENGTHS IN CHS X-
Appendix E provides the supplementary data for Chapter 6: Fillet Weld Effective Lengths in CHS X-
Connections: Experimentation.
E.1. CHS-TO-CHS X-CONNECTION FABRICATION DRAWINGS

A list of fabrication drawings for the CHS-to-CHS X-connections, and the drawings themselves, are
provided herein. Additional fabricated specimens, such as trial weld specimens and all-weld-metal tensile
coupon test specimens, are included on Sheet No. S7.
Table E.1. List of CHS-to-CHS X-connection fabrication drawings

S1 1 Test 127-406-90a/b specimen
S7 7, 8, 9 All-weld-metal TC test specimen, trial weld specimens, and parts list for CHS X-
connection tests
225
Appendix E: Fillet Weld Effective Lengths in CHS X-Connections: Experimentation 226
Fig. E.1. Test 127-406-90a/b specimen






Fig. E.7. All-weld-metal TC test specimen, trial weld specimens, and parts list for CHS X-connection tests

E.2. TEMPLATE LENGTH FORMULA FOR CHS-TO-CHS CONNECTIONS
Fig. E.8. CHS-to-CHS template geometry view no. 1
Fig. E.9. CHS-to-CHS template Fig. E.10. CHS-to-CHS template

geometry view no. 2 geometry view no. 3
The template length (lt) can be determined, for an arbitrary point “A” along the intersection surface of the
tubes, as follows:
Step 1: Construct ΔABC (Fig. E.8), where AB is perpendicular to the centreline of the branch and ∠BCA
is equal to the branch inclination angle (θ).
Step 2: Construct ΔCDE where CE is perpendicular to the centreline of the cord and ∠CDE is equal to θ.
Step 3: Define M at the crown of the chord along the line ECB; define Q such that ΔCDE and ΔMEQ are
similar triangles; define N at the heel of the connection; and define K as the point along the branch centreline
such that ∠MKN is equal to 90°.

Step 4: Compute the template length (lt) as follows:
lt  KM  EM  BC  CE (E.1)
Db
Db (E.2)
KM  2 
tan  2 tan 
D
D (E.3)
EM  2 
sin  2sin 
With reference to Fig. E.9:
Db
AB  cos  (E.4a)
2
AB Db cos 
BC   (E.4b)
tan  2 tan 
D
CD  cos P (E.5a)
2
CD D cos P
CE   (E.5b)
sin  2sin 
With reference to Fig. E.10, transform P to ρ:
D D
sin P  b sin  (E.6c)
2 2
Db
sin P  sin  (E.6b)
D

2
D 
cos P  1   sin    1   b sin   (E.6c)
2
 D 
Hence:
D 2   Db sin  
2
(E.7)
CE 
2sin 
Substitute Eqs. (E.2), (E.3), (E.4b) and (E.7) into Eq. (E.1):
D 2   Db sin  
2
Db D Db cos  (E.8)
lt    
2 tan  2sin  2 tan  2sin 
Db (1  cos  ) D  D   Db sin  
2 2
(E.9)
 lt  
2 tan  2sin 

E.3. MEASURED FILLET WELD DIMENSIONS FOR CHS-TO-CHS X-

CONNECTIONS
The external measurements of the weld leg components, and the weld legs and weld throat dimensions
obtained via Solidworks, are shown for each of the 12 CHS-to-CHS X-connections in the following tables.
The explanation of component measurements is given in Section 6.2.2. The offset dimension, given as
footnotes to the tables, is shown in Fig. 6.5a. Where the value of x is not given (e.g. at the toe and heel, ρ = 0°
and 180°), the component of lv was obtained by direct measurement on the specimen.
Table E.2. Weld measurements for 102-273-90a

External measurements Solidworks measurements
Component Component
ρ Tributary weld length lt x of lh of lv lh lv Ѱ tw
° mm mm mm mm mm mm mm ° mm
0 26.6 0 - 5.70 6.90 5.70 6.90 90.0 4.39
30 26.9 2.37 42.98 5.28 7.90 5.21 7.93 95.3 4.16
60 26.9 7.24 47.61 5.52 7.30 5.48 7.65 106.1 3.81
90 26.6 9.75 48.95 6.69 6.40 6.69 6.69 111.8 3.75
120 26.9 7.24 46.49 6.64 6.70 6.57 7.02 106.1 4.08
150 26.9 2.37 42.36 5.90 7.60 5.85 7.64 95.3 4.43
180 26.6 0 - 5.60 7.30 5.60 7.30 90.0 4.44
210 26.9 2.37 40.98 7.28 6.20 7.20 6.23 95.3 4.49
240 26.9 7.24 46.65 6.48 6.60 6.39 6.91 106.1 3.99
270 26.6 9.75 49.2 6.44 5.20 6.44 5.65 111.8 3.37
300 26.9 7.24 46.64 6.49 5.00 6.39 5.24 106.1 3.45
330 26.9 2.37 41.62 6.64 7.10 6.57 7.14 95.3 4.61
Total (lw): 322 Average: 6.17 6.86 - 4.08
Note: offset = 45.89 mm.

Table E.3. Weld measurements for 102-273-90b

Component Component
0 26.6 0 30.58 7.86 6.80 7.86 6.80 90.0 5.14
30 26.9 2.37 32.66 8.15 8.00 8.04 8.04 95.3 5.42
60 26.9 7.24 38.51 7.17 7.20 7.04 7.55 106.1 4.38
90 26.6 9.75 41.29 6.90 6.00 6.90 6.52 111.8 3.76
120 26.9 7.24 39.03 6.65 5.20 6.55 5.45 106.1 3.57
150 26.9 2.37 33.80 7.01 7.60 6.93 7.64 95.3 4.89
180 26.6 0 32.50 5.94 7.60 5.94 7.60 90.0 4.68
210 26.9 2.37 34.89 5.92 7.10 5.88 7.14 95.3 4.33
240 26.9 7.24 38.55 7.13 6.80 7.04 7.12 106.1 4.25
270 26.6 9.75 42.02 6.17 6.30 6.17 6.85 111.8 3.64
300 26.9 7.24 40.12 5.56 7.80 5.49 8.18 106.1 3.91
330 26.9 2.37 34.77 6.04 7.80 5.91 7.84 95.3 4.50
Total (lw): 322 Average: 6.65 7.23 - 4.37

Component Component
0 26.6 0 - 5.00 5.20 5.00 5.20 90.0 3.60
30 26.7 1.59 65.84 5.75 4.80 5.72 4.81 93.6 3.57
60 26.7 4.80 68.43 6.37 5.10 6.34 5.20 100.8 3.63
90 26.6 6.43 70.39 6.04 5.80 6.04 6.01 104.4 3.69
120 26.7 4.80 68.27 6.53 5.20 6.47 5.30 100.8 3.70
150 26.7 1.59 64.89 6.70 4.90 6.68 4.91 93.6 3.84
180 26.6 0 - 5.70 4.50 5.70 4.50 90.0 3.53
210 26.7 1.59 65.33 6.26 4.80 6.25 4.81 93.6 3.69
240 26.7 6.43 69.66 6.77 5.10 6.70 5.20 100.8 3.71
270 26.6 6.43 71.81 4.61 5.10 4.62 5.28 104.4 3.01
300 26.7 4.80 70.14 4.66 5.30 4.62 5.41 100.8 3.17
330 26.7 1.59 66.40 5.19 5.30 5.16 5.31 93.6 3.58
Total (lw): 320 Average: 5.78 5.16 - 3.56


Component Component
0 26.6 0 - 4.60 4.10 4.60 4.10 90.0 3.06
30 26.7 1.59 48.02 4.86 4.50 4.83 4.51 93.6 3.19
60 26.7 4.80 50.96 5.13 4.90 5.10 5.00 100.8 3.22
90 26.6 6.43 52.79 4.93 4.20 4.93 4.35 104.4 2.83
120 26.7 4.80 51.56 4.53 4.80 4.50 4.90 100.8 2.99
150 26.7 1.59 47.81 5.07 4.80 5.04 4.81 93.6 3.37
180 26.6 0.00 - 4.40 4.50 4.40 4.50 90.0 3.15
210 26.7 1.59 47.88 5.00 3.90 5.01 3.91 93.6 2.99
240 26.7 6.43 50.94 6.78 4.40 6.75 4.48 100.8 3.39
270 26.6 6.43 51.98 5.74 4.50 5.74 4.66 104.4 3.14
300 26.7 4.80 50.62 5.47 4.60 5.45 4.59 100.8 3.17
330 26.7 1.59 48.28 4.60 4.70 4.58 4.71 93.6 3.18
Total (lw): 320 Average: 5.08 4.54 - 3.14

Component Component
0 26.7 0 - 6.30 5.70 6.30 5.70 90.0 4.23
24 27.1 2.48 47.40 5.13 5.20 5.03 5.22 94.4 3.48
48 27.4 8.45 52.85 5.65 6.10 5.46 6.32 104.9 3.56
72 27.0 14.15 - 4.61 5.20 4.51 5.81 114.9 2.72
96 26.8 15.56 - 4.86 4.00 4.85 4.55 117.4 2.44
120 27.3 11.62 56.83 4.84 5.90 4.73 6.35 110.4 3.08
144 27.3 5.23 50.05 5.23 5.10 4.64 5.23 99.3 3.18
168 26.8 0.64 43.82 6.87 5.10 6.81 5.10 91.2 4.04
192 26.8 0.64 44.08 6.61 6.10 6.58 6.10 91.2 4.43
216 27.3 5.23 48.67 6.61 7.20 5.91 7.39 99.3 4.24
240 27.3 11.62 - 6.90 6.60 6.68 7.12 110.4 3.93
264 26.8 15.56 - 5.81 5.10 5.80 5.82 117.4 3.02
288 27.0 14.15 56.58 7.62 5.30 7.16 6.05 114.9 3.52
312 27.4 8.45 50.85 7.65 5.90 7.46 6.13 104.9 4.09
336 27.1 2.48 45.34 7.19 6.20 7.10 6.22 94.4 4.50
Total (lw): 406 Average: 5.93 5.94 - 3.63


Component Component
0 26.7 0 - - - 6.90 7.00 90.0 4.91
24 27.1 2.48 - - - 5.68 7.22 94.4 4.29
48 27.4 8.45 - - - 4.93 8.00 104.9 3.66
72 27.0 14.15 - - - 5.82 7.16 114.9 3.45
96 26.8 15.56 - - - 4.98 6.97 117.4 3.00
120 27.3 11.62 - - - 5.92 6.78 110.4 3.60
144 27.3 5.23 - - - 6.39 7.00 99.3 4.32
168 26.8 0.64 - - - 6.98 7.10 91.2 4.93
192 26.8 0.64 - - - 5.87 6.90 91.2 4.43
216 27.3 5.23 - - - 6.05 6.39 99.3 4.02
240 27.3 11.62 - - - 5.71 6.77 110.4 3.53
264 26.8 15.56 - - - 6.15 6.97 117.4 3.39
288 27.0 14.15 - - - 7.09 7.26 114.9 3.86
312 27.4 8.45 - - - 6.21 7.17 104.9 4.05
336 27.1 2.48 - - - 6.29 7.12 94.4 4.53
Total (lw): 406 Average: 6.06 7.05 - 4.00

Component Component
0 26.7 0 - 4.20 4.20 4.20 4.20 90.0 2.97
24 26.9 1.66 71.38 6.13 4.80 6.13 4.81 93.0 3.69
48 27.0 5.59 75.35 6.09 4.60 5.98 4.67 100.0 3.35
72 26.8 9.24 79.82 5.27 5.00 5.22 5.23 106.5 3.13
96 26.7 10.13 81.43 4.54 4.80 4.54 5.07 108.1 2.81
120 26.9 7.63 78.92 4.55 4.30 4.51 4.43 103.6 2.76
144 27.0 3.48 74.92 4.41 4.30 4.36 4.33 96.2 2.90
168 26.7 0.43 71.79 4.49 4.40 4.48 4.40 90.8 3.12
192 26.7 0.43 71.62 4.66 4.20 4.67 4.20 90.8 3.10
216 27.0 3.48 73.08 6.25 4.70 6.19 4.73 96.2 3.56
240 26.9 7.63 78.35 5.13 5.60 5.06 5.77 103.6 3.33
264 26.7 10.13 80.76 5.21 5.40 5.22 5.71 108.1 3.20
288 26.8 9.24 79.68 5.40 5.10 5.39 5.34 106.5 3.21
312 27.0 5.59 76.74 4.70 5.20 4.65 5.28 100.0 3.18
336 26.9 1.66 72.56 4.94 4.20 4.91 4.21 93.0 3.11
Total (lw): 403 Average: 5.03 4.83 - 3.16


Component Component
° mm mm mm mm mm mm mm °
0 26.7 0 - 4.70 5.50 4.70 5.50 90.0 3.57
24 26.9 1.66 51.48 5.07 5.70 5.06 5.71 93.0 3.69
48 27.0 5.59 54.36 6.12 5.30 6.05 5.39 100.0 3.66
72 26.8 9.24 58.80 5.33 5.80 5.29 6.07 106.5 3.38
96 26.7 10.13 60.25 4.77 5.80 4.77 6.13 108.1 3.14
120 26.9 7.63 57.30 5.22 6.00 5.15 6.19 103.6 3.47
144 27.0 3.48 52.58 5.79 5.20 5.72 5.23 96.2 3.65
168 26.7 0.43 49.63 5.69 5.20 5.68 5.20 90.8 3.81
192 26.7 0.43 49.92 5.40 5.30 5.39 5.30 90.8 3.75
216 27.0 3.48 53.22 5.15 5.40 5.09 5.43 96.2 3.51
240 26.9 7.63 57.38 5.14 5.30 5.10 5.46 103.6 3.26
264 26.7 10.13 59.53 5.49 5.40 5.43 5.71 108.1 3.27
288 26.8 9.24 59.02 5.11 5.40 5.08 5.65 106.5 3.20
312 27.0 5.59 54.92 5.56 5.80 5.50 5.89 100.0 3.66
336 26.9 1.66 51.70 4.85 5.20 4.82 5.21 93.0 3.45
Total (lw): 403 Average: 5.19 5.60 - 3.47

Component Component
0 26.8 0.00 36.80 8.06 4.10 8.06 4.73 60.0 4.71
30 28.7 5.74 44.10 6.50 4.80 5.96 5.15 68.4 4.54
60 31.2 20.13 59.30 5.69 5.50 4.82 5.51 86.6 3.73
90 30.6 36.63 75.00 6.49 5.00 5.65 5.17 104.3 3.31
120 28.4 49.40 88.30 5.96 6.20 5.58 6.88 115.4 3.28
150 26.9 56.44 95.10 6.20 5.90 6.14 6.79 119.4 3.25
180 26.6 58.54 97.00 6.40 6.10 6.40 7.04 120.0 3.35
210 26.9 56.44 95.90 5.40 5.70 5.10 6.71 119.4 2.91
240 28.4 49.40 88.90 5.36 6.20 5.23 6.86 115.4 3.16
270 30.6 36.63 75.10 6.39 5.80 5.55 5.98 104.3 3.53
300 31.2 20.13 60.10 4.89 4.70 4.26 4.71 86.6 3.25
330 28.7 5.74 45.30 5.30 4.40 4.71 4.72 68.4 3.90
Total (lw): 345 Average: 5.59 5.83 - 3.58


Component Component
0 26.8 0.00 65.20 8.76 4.80 8.76 5.54 60.0 5.48
30 28.7 5.74 74.30 5.40 5.25 4.97 5.64 68.4 4.35
60 31.2 20.13 88.20 5.89 5.70 5.05 5.71 86.6 3.89
90 30.6 36.63 104.40 6.19 6.80 5.30 7.02 104.3 3.68
120 28.4 49.40 118.20 5.16 7.00 4.85 7.81 115.4 3.16
150 26.9 56.44 124.10 6.30 6.30 6.28 7.27 119.4 3.39
180 26.6 58.54 125.40 7.10 5.70 7.10 6.58 120.0 3.41
210 26.9 56.44 123.80 6.60 6.30 6.57 7.35 119.4 3.50
240 28.4 49.40 116.30 7.06 6.20 6.22 6.93 115.4 3.50
270 30.6 36.63 104.90 5.69 6.20 5.02 6.39 104.3 3.43
300 31.2 20.13 88.20 5.89 4.50 4.99 4.51 86.6 3.44
330 28.7 5.74 73.60 6.10 4.65 5.44 4.98 68.4 4.29
Total (lw): 345 Average: 5.83 6.29 - 3.79

Component Component
0 26.8 0.00 70.00 8.46 4.30 8.46 4.30 60.0 4.30
24 28.5 5.08 77.60 5.94 4.40 5.57 4.80 66.3 4.29
48 31.2 18.57 88.80 8.23 4.50 7.34 4.55 81.1 4.12
72 31.9 35.99 104.50 9.95 5.10 8.29 5.14 97.7 4.09
96 30.2 52.21 121.50 9.17 6.10 8.09 6.57 110.9 4.10
120 28.2 63.90 132.70 9.66 6.00 9.33 6.76 118.3 4.00
144 27.1 70.51 138.30 10.67 6.60 10.51 7.67 120.3 4.39
168 26.7 73.25 142.20 9.51 6.20 9.49 7.17 120.1 4.07
192 26.7 73.25 142.20 9.51 6.00 9.50 6.94 120.1 3.99
216 27.1 70.51 140.00 8.97 6.10 8.95 7.09 120.3 3.93
240 28.2 63.90 135.60 6.76 5.50 6.54 6.27 118.3 3.28
264 30.2 52.21 123.80 6.87 5.30 5.99 5.69 110.9 3.31
288 31.9 35.99 106.50 7.95 4.70 6.53 4.74 97.7 3.58
312 31.2 18.57 88.20 8.83 3.80 7.37 3.84 81.1 3.60
336 28.5 5.08 73.90 9.64 4.05 8.96 4.41 66.3 4.39
Total (lw): 434 Average: 8.01 5.68 - 3.95


Component Component
0 26.8 0.00 40.90 7.66 4.00 7.66 4.00 60.0 4.00
24 28.5 5.08 48.20 5.44 4.05 5.07 4.42 66.3 3.93
48 31.2 18.57 63.50 3.63 4.10 3.04 4.14 81.1 2.62
72 31.9 35.99 79.80 4.75 4.10 4.13 4.13 97.7 2.72
96 30.2 52.21 94.60 6.17 5.70 5.42 6.10 110.9 3.25
120 28.2 63.90 107.00 5.46 5.50 5.22 6.28 118.3 2.92
144 27.1 70.51 112.00 7.07 6.30 7.03 7.31 120.3 3.57
168 26.7 73.25 114.70 7.11 6.20 7.10 7.18 120.1 3.56
192 26.7 73.25 114.70 7.11 5.00 7.10 5.78 120.1 3.18
216 27.1 70.51 112.10 6.97 5.40 6.90 6.30 120.3 3.28
240 28.2 63.90 106.10 6.36 6.20 6.08 7.11 118.3 3.36
264 30.2 52.21 94.60 6.17 5.80 5.44 6.19 110.9 3.28
288 31.9 35.99 77.30 7.25 4.10 5.89 4.13 97.7 3.16
312 31.2 18.57 58.50 8.63 4.00 7.28 4.04 81.1 3.74
336 28.5 5.08 45.70 7.94 4.00 7.44 4.37 66.3 4.28
Total (lw): 434 Average: 6.00 5.39 - 3.38
Note: offset = 56.16 mm

E.4. 3D SOLIDWORKS WELD PROFILES FOR ALL TEST WELDS

Appendix E.4 provides the 3D Solidworks weld profiles (built from actual weld measurements) for all
test welds. The triangular cross sections along each weld length are shown in the plane of the local dihedral
angle (Ѱ).
(a) 102-273-90a (b) 102-273-90b
(c) 102-406-90a (d) 102-406-90b
Fig. E.11. 3D Solidworks models of weld profile for 102-series tests with θ = 90°

(a) 127-273-90a (b) 127-273-90b
(c) 127-406-90a (d) 127-406-90b
Fig. E.12. 3D Solidworks models of weld profile for 127-series tests with θ = 90°

(a) 102-406-60a (b) 102-406-60b
(c) 127-406-60a (d) 127-406-60b
Fig. E.13. 3D Solidworks models of weld profile for tests with θ = 90°

E.5. POST-RUPTURE MACROETCH MEASUREMENTS

The table below compares the average weld throat dimensions obtained from Solidworks measurements prior
to testing to those obtained from post-rupture macro-etch examinations after testing for six of the tested weld
specimens.
Table E.14. Fillet weld macroetch measurements for CHS X-connections

Test ρ lv lh Ѱ tw average (tw)
° mm mm ° mm mm
0 6.88 6.34 90 4.66
90 6.36 6.59 117 3.38
102-273-90b 4.13
180 7.57 6.79 90 5.05
270 6.04 7.15 117 3.42
0 5.96 4.38 91 3.50
90 6.54 6.40 108 3.80
102-406-90a 3.54
180 4.71 6.86 89 3.91
270 4.65 6.06 112 2.93
0 5.81 7.57 90 4.61
90 5.82 8.96 116 3.71
127-273-90b 4.01
180 6.36 6.36 90 4.50
270 5.67 7.54 120 3.23
0 6.46 5.76 89 4.34
90 4.90 6.42 113 3.06
127-406-90b 3.82
180 6.00 6.90 91 4.49
270 5.00 8.09 112 3.41
0 13.55 5.96 60 5.95
90 5.42 5.64 110 3.17
102-406-60b 3.86
180 6.26 6.91 121 3.23
270 4.96 6.10 111 3.09
0 12.66 7.30 59 7.28
90 6.63 5.23 109 3.38
127-406-60b 4.31
180 7.00 7.04 120 3.51
270 5.18 7.26 120 3.05
Table E.15. Comparison of average weld throat dimensions obtained from post-rupture macro-etch
examinations and external measurements
Test Average weld throat dimension (tw) A/B
Solidworks measurements Macro-etch measurements
prior to testing (A) after testing (B)
mm mm
102-273-90b 4.37 4.13 1.06
102-406-90a 3.56 3.54 1.01
127-273-90b 4.00 4.01 1.00
127-406-90b 3.47 3.82 0.91
102-406-60b 3.79 3.86 0.98
127-406-60b 3.38 4.31 0.78

E.6. MATERIAL PROPERTY TEST RESULTS

This sub-appendix provides the results of tensile coupon tests conducted for the circular hollow sections
(CHS) and the weld metal for tests in Chapter 6. Note that several CHS branch members are the same as
those used in Chapter 4, and their property tests have not been repeated here.
(a) HSS 273.5 × 11.69 (heat no. 778039) (b) HSS 406.5 × 12.34 (heat no. 823306)
(c) HSS 410.0 × 12.21 (heat no. 777481) (d) HSS 102.0 × 7.34 (heat no. 819720-2)
Fig. E.14. TC test results for CHS chord and branch materials
Fig. E.15. All-weld-metal TC test results (E71T-1C electrode)

E.7. PHOTOGRAPHS OF EXPERIMENTAL TESTS AND FAILURE MODES

Appendix E.7 contains photographs of the experimental test set-ups and weld rupture failure modes for
all 12 tests on CHS-to-CHS X-connections. For clarity in the photographs depicting weld rupture failure,
instrumentation has been moved out of the way, or removed.
Fig. E.16. Photographs of test 102-273-90a
(a) Weld rupture failure in test 102-273-90b (b) Weld rupture failure in test 102-406-90a
Fig. E.17. Photographs of test 102-273-90b and 102-406-90a

(a) Weld rupture failure (b) Overwelding of test joint (after test)
Fig. E.18. Photographs of test 102-406-90b
(a) Weld rupture failure in test 127-273-90a (b) Weld rupture failure in test 127-273-90b
Fig. E.19. Photographs of tests 127-273-90a and 127-273-90b

Fig. E.21. Photograph of test 127-406-90b

Fig. E.23. Photograph of test 102-406-60b

(a) Weld rupture failure in test 127-406-60a (b) Weld rupture failure in test 127-406-60b
Fig. E.24. Photographs of tests 127-406-60a and 127-406-60b

E.8. SUMMARY OF ACTUAL AND PREDICTED STRENGTHS ACCORDING TO

CODE DESIGN METHODS FOR FE CHS X-CONNECTION TESTS
A summary of the actual and predicted strengths for 12 weld-critical CHS X-connection tests is provided
in the following table.
Table E.16. Summary of actual and predicted strengths for all experimental CHS X-connections according to
AWS D1.1-15, AISC 360-16, and CSA S16-14
Pnw
Test AWS D1.1-15 AISC 360-16 CSA S16-14
Paꞌ le = (2/3)lw le = lw le = lw le = lw
kN kN (A/P) kN (A/P) kN (A/P) kN (A/P)
102-273-90a 672 303 2.22 454 1.48 454 1.48 507 1.33
102-273-90b 678 324 2.09 486 1.39 486 1.39 543 1.25
102-406-90a 608 263 2.31 394 1.54 394 1.54 440 1.38
102-406-90b 540 232 2.33 348 1.55 348 1.55 388 1.39
127-273-90a 653 340 1.92 511 1.28 511 1.28 570 1.15
127-273-90b 653 375 1.74 563 1.16 563 1.16 628 1.04
127-406-90a 557 294 1.90 441 1.26 441 1.26 492 1.13
127-406-90b 557 323 1.73 484 1.15 484 1.15 540 1.03
102-406-60a 721 285 2.53 428 1.69 428 1.69 477 1.51
102-406-60b 721 302 2.39 453 1.59 453 1.59 505 1.43
127-406-60a 761 396 1.92 594 1.28 594 1.28 663 1.15
127-406-60b 850 339 2.51 508 1.67 508 1.67 568 1.50
ρP = 2.13 1.42 1.42 1.27
VP = 0.13 0.13 0.13 0.13

FILLET WELD EFFECTIVE LENGTHS IN CHS X-
Appendix F provides the supplementary data for Chapter 7: Fillet Weld Effective Lengths in CHS X-
Connections: Finite Element Modelling.
F.1. COMPLETE PARAMETRIC STUDY RESULTS

Table F.1 lists the complete results of the parametric study.
Table F.1. Parametric study results

Test ID θ D/t Db/tb tw lw PFE FE le/lw FE/0.03D
° mm mm kN mm
1 90 0.5 50 1.0 25.0 4.00 639 781 39.8 0.60 1.66
2 90 0.5 50 0.8 31.3 3.20 639 629 29.0 0.61 1.21
3 90 0.5 50 0.6 41.7 2.40 639 478 20.0 0.63 0.83
4 90 0.5 40 1.0 20.0 5.00 639 1069 40.9 0.66 1.70
5 90 0.5 40 0.8 25.0 4.00 639 865 28.9 0.67 1.20
6 90 0.5 40 0.6 33.3 3.00 639 662 19.3 0.69 0.80
7 90 0.5 40 0.4 50.0 2.00 639 439 11.2 0.71 0.47
8 90 0.5 30 1.0 15.0 6.67 639 1650 42.7 0.75 1.78
9 90 0.5 30 0.8 18.8 5.33 639 1351 28.8 0.78 1.20
10 90 0.5 30 0.6 25.0 4.00 639 1034 17.7 0.80 0.74
11 90 0.5 30 0.4 37.5 2.67 639 719 10.7 0.85 0.45
12 90 0.5 20 1.0 10.0 10.0 639 2979 39.8 0.90 1.66
13 90 0.5 20 0.8 12.5 8.00 639 2485 26.2 0.94 1.09
14 90 0.5 20 0.6 16.7 6.00 639 1963 16.0 1.00 0.67
15 90 0.5 20 0.4 25.0 4.00 639 1402 9.7 1.08 0.40
16 90 0.5 20 0.2 50.0 2.00 639 707 4.6 1.14 0.19
17 90 0.5 10 0.4 12.5 8.00 639 3048 7.0 1.16 0.29
18 90 0.5 10 0.2 25.0 4.00 639 1559 4.3 1.21 0.18
19 90 0.4 50 1.0 20.0 5.00 635 1067 59.6 0.66 1.99
20 90 0.4 50 0.8 25.0 4.00 635 868 47.0 0.68 1.57
21 90 0.4 50 0.6 33.3 3.00 635 652 29.5 0.69 0.98
22 90 0.4 50 0.4 50.0 2.00 635 424 17.5 0.69 0.58
23 90 0.4 40 1.0 16.0 6.25 635 1470 61.1 0.72 2.04
24 90 0.4 40 0.8 20.0 5.00 635 1201 43.1 0.74 1.44
25 90 0.4 40 0.6 26.7 3.75 635 913 27.9 0.76 0.93
26 90 0.4 40 0.4 40.0 2.50 635 605 16.3 0.77 0.54
27 90 0.4 30 1.0 12.0 8.33 635 2265 61.0 0.83 2.03
28 90 0.4 30 0.8 15.0 6.67 635 1861 40.8 0.86 1.36
29 90 0.4 30 0.6 20.0 5.00 635 1440 26.1 0.89 0.87
30 90 0.4 30 0.4 30.0 3.33 635 994 15.2 0.94 0.51
31 90 0.4 20 0.8 10.0 10.0 635 3429 37.6 1.04 1.25
32 90 0.4 20 0.6 13.3 7.50 635 2697 22.4 1.10 0.75
33 90 0.4 20 0.4 20.0 5.00 635 1858 12.8 1.15 0.43
254
Appendix F: Fillet Weld Effective Lengths in CHS X-Connections: Finite Element Modelling 255

° mm mm kN mm
34 90 0.4 20 0.2 40.0 2.50 635 918 6.0 1.17 0.20
35 90 0.4 10 0.4 10.0 10.0 635 3795 8.1 1.15 0.27
36 90 0.4 10 0.2 20.0 5.00 635 1924 4.7 1.19 0.16
37 90 0.3 50 1.0 15.0 6.67 632 1613 87.4 0.74 2.19
38 90 0.3 50 0.8 18.8 5.33 632 1295 64.3 0.75 1.61
39 90 0.3 50 0.6 25.0 4.00 632 971 42.3 0.76 1.06
40 90 0.3 50 0.4 37.5 2.67 632 652 26.1 0.78 0.65
41 90 0.3 40 1.0 12.0 8.33 632 2242 88.2 0.83 2.21
42 90 0.3 40 0.8 15.0 6.67 632 1800 61.4 0.83 1.54
43 90 0.3 40 0.6 20.0 5.00 632 1374 40.7 0.85 1.02
44 90 0.3 40 0.4 30.0 3.33 632 953 25.1 0.90 0.63
45 90 0.3 30 0.8 11.3 8.89 632 2795 55.3 0.96 1.38
46 90 0.3 30 0.6 15.0 6.67 632 2156 35.6 1.00 0.89
47 90 0.3 30 0.4 22.5 4.44 632 1507 22.1 1.06 0.55
48 90 0.3 30 0.2 45.0 2.22 632 809 11.2 1.18 0.28
49 90 0.3 20 0.6 10.0 10.0 632 3718 26.8 1.14 0.67
50 90 0.3 20 0.4 15.0 6.67 632 2512 15.9 1.16 0.40
51 90 0.3 20 0.2 30.0 3.33 632 1275 7.8 1.21 0.20
52 90 0.3 10 0.2 15.0 6.67 632 2573 5.8 1.19 0.15
53 90 0.2 50 1.0 10.0 10.0 630 3017 136.8 0.93 2.28
54 90 0.2 50 0.8 12.5 8.00 630 2405 100.1 0.93 1.67
55 90 0.2 50 0.6 16.7 6.00 630 1795 67.6 0.93 1.13
56 90 0.2 50 0.4 25.0 4.00 630 1219 42.0 0.96 0.70
57 90 0.2 50 0.2 50.0 2.00 630 669 22.0 1.09 0.37
58 90 0.2 40 0.8 10.0 10.0 630 3344 92.9 1.03 1.55
59 90 0.2 40 0.6 13.3 7.50 630 2503 60.0 1.03 1.00
60 90 0.2 40 0.4 20.0 5.00 630 1722 37.2 1.07 0.62
61 90 0.2 40 0.2 40.0 2.50 630 925 19.3 1.19 0.32
62 90 0.2 30 0.6 10.0 10.0 630 3667 47.0 1.13 0.78
63 90 0.2 30 0.4 15.0 6.67 630 2480 28.4 1.15 0.47
64 90 0.2 30 0.2 30.0 3.33 630 1269 14.1 1.21 0.24
65 90 0.2 20 0.4 10.0 10.0 630 3834 18.9 1.18 0.32
66 90 0.2 20 0.2 20.0 5.00 630 1953 10.0 1.22 0.17
67 90 0.2 10 0.2 10.0 10.0 630 3860 7.5 1.18 0.13
68 90 0.1 50 0.4 12.5 8.00 629 3009 63.7 1.16 0.53
69 90 0.1 50 0.2 25.0 4.00 629 1553 31.6 1.22 0.26
70 90 0.1 40 0.4 10.0 10.0 629 3649 56.6 1.12 0.47
71 90 0.1 40 0.2 20.0 5.00 629 1882 27.0 1.17 0.23
72 90 0.1 30 0.2 15.0 6.67 629 2580 17.6 1.20 0.15
73 90 0.1 20 0.2 10.0 10.0 629 3749 14.4 1.15 0.12
74 80 0.5 50 1.0 25.0 4.00 644 759 45.2 0.58 1.85
75 80 0.5 50 0.8 31.3 3.20 644 612 33.5 0.59 1.37
76 80 0.5 50 0.6 41.7 2.40 644 469 23.6 0.62 0.97
77 80 0.5 40 1.0 20.0 5.00 644 1044 47.8 0.64 1.96
78 80 0.5 40 0.8 25.0 4.00 644 848 35.3 0.65 1.45
79 80 0.5 40 0.6 33.3 3.00 644 649 23.3 0.67 0.96
80 80 0.5 40 0.4 50.0 2.00 644 451 14 0.72 0.57
81 80 0.5 30 1.0 15.0 6.67 644 1608 51.4 0.73 2.11
82 80 0.5 30 0.8 18.8 5.33 644 1305 34.8 0.74 1.43
83 80 0.5 30 0.6 25.0 4.00 644 994 21.2 0.76 0.87
84 80 0.5 30 0.4 37.5 2.67 644 704 12.8 0.83 0.53
85 80 0.5 20 1.0 10.0 10.0 644 2879 48.8 0.86 2.00
86 80 0.5 20 0.8 12.5 8.00 644 2378 32.7 0.89 1.34
87 80 0.5 20 0.6 16.7 6.00 644 1881 20 0.95 0.82
88 80 0.5 20 0.4 25.0 4.00 644 1349 11.4 1.03 0.47
89 80 0.5 20 0.2 50.0 2.00 644 702 5.3 1.12 0.22


° mm mm kN mm
90 80 0.5 10 0.4 12.5 8.00 644 2776 6.1 1.04 0.25
91 80 0.5 10 0.2 25.0 4.00 644 1439 3.7 1.10 0.15
92 80 0.4 50 1.0 20.0 5.00 640 1009 66.7 0.62 2.19
93 80 0.4 50 0.8 25.0 4.00 640 826 50.6 0.64 1.66
94 80 0.4 50 0.6 33.3 3.00 640 633 34.5 0.66 1.13
95 80 0.4 50 0.4 50.0 2.00 640 423 20.5 0.68 0.67
96 80 0.4 40 1.0 16.0 6.25 640 1407 69.7 0.69 2.29
97 80 0.4 40 0.8 20.0 5.00 640 1151 52.2 0.71 1.71
98 80 0.4 40 0.6 26.7 3.75 640 881 33.3 0.73 1.09
99 80 0.4 40 0.4 40.0 2.50 640 607 19.8 0.77 0.65
100 80 0.4 30 1.0 12.0 8.33 640 2137 70.4 0.78 2.31
101 80 0.4 30 0.8 15.0 6.67 640 1770 48.3 0.81 1.59
102 80 0.4 30 0.6 20.0 5.00 640 1368 30.8 0.84 1.01
103 80 0.4 30 0.4 30.0 3.33 640 962 17.9 0.90 0.59
104 80 0.4 20 0.8 10.0 10.0 640 3215 45.2 0.97 1.48
105 80 0.4 20 0.6 13.3 7.50 640 2512 26.2 1.02 0.86
106 80 0.4 20 0.4 20.0 5.00 640 1768 14.8 1.08 0.49
107 80 0.4 20 0.2 40.0 2.50 640 911 7.1 1.15 0.23
108 80 0.4 10 0.4 10.0 10.0 640 3497 7.6 1.06 0.25
109 80 0.4 10 0.2 20.0 5.00 640 1790 4.4 1.10 0.14
110 80 0.3 50 1.0 15.0 6.67 637 1524 99.1 0.70 2.44
111 80 0.3 50 0.8 18.8 5.33 637 1255 74.8 0.72 1.84
112 80 0.3 50 0.6 25.0 4.00 637 938 50.5 0.73 1.24
113 80 0.3 50 0.4 37.5 2.67 637 631 29.6 0.75 0.73
114 80 0.3 40 1.0 12.0 8.33 637 2113 102.2 0.77 2.52
115 80 0.3 40 0.8 15.0 6.67 637 1698 70.7 0.78 1.74
116 80 0.3 40 0.6 20.0 5.00 637 1325 47.9 0.82 1.18
117 80 0.3 40 0.4 30.0 3.33 637 892 27.2 0.84 0.67
118 80 0.3 30 0.8 11.3 8.89 637 2634 66.1 0.90 1.63
119 80 0.3 30 0.6 15.0 6.67 637 2031 41.9 0.93 1.03
120 80 0.3 30 0.4 22.5 4.44 637 1441 24.8 1.00 0.61
121 80 0.3 30 0.2 45.0 2.22 637 792 12.4 1.14 0.31
122 80 0.3 20 0.6 10.0 10.0 637 3491 29.8 1.06 0.73
123 80 0.3 20 0.4 15.0 6.67 637 2367 27 1.08 0.66
124 80 0.3 20 0.2 30.0 3.33 637 1201 7.9 1.13 0.19
125 80 0.3 10 0.2 15.0 6.67 637 2349 4.9 1.08 0.12
126 80 0.2 50 1.0 10.0 10.0 635 2871 157.4 0.87 2.58
127 80 0.2 50 0.8 12.5 8.00 635 2305 117.1 0.88 1.92
128 80 0.2 50 0.6 16.7 6.00 635 1695 76.1 0.87 1.25
129 80 0.2 50 0.4 25.0 4.00 635 1175 47 0.91 0.77
130 80 0.2 50 0.2 50.0 2.00 635 642 23.4 1.04 0.38
131 80 0.2 40 0.8 10.0 10.0 635 3195 111.8 0.97 1.84
132 80 0.2 40 0.6 13.3 7.50 635 2437 73.9 0.99 1.21
133 80 0.2 40 0.4 20.0 5.00 635 1638 41.6 1.01 0.68
134 80 0.2 40 0.2 40.0 2.50 635 890 20.4 1.14 0.33
135 80 0.2 30 0.6 10.0 10.0 635 3538 56.8 1.08 0.93
136 80 0.2 30 0.4 15.0 6.67 635 2378 32.7 1.09 0.54
137 80 0.2 30 0.2 30.0 3.33 635 1225 15.1 1.16 0.25
138 80 0.2 20 0.4 10.0 10.0 635 3562 19 1.08 0.31
139 80 0.2 20 0.2 20.0 5.00 635 1816 9.7 1.12 0.16
140 80 0.2 10 0.2 10.0 10.0 635 3561 6.4 1.08 0.11
141 80 0.1 50 0.4 12.5 8.00 634 2875 69.7 1.10 0.57
142 80 0.1 50 0.2 25.0 4.00 634 1496 33.5 1.17 0.27
143 80 0.1 40 0.4 10.0 10.0 634 3646 53.8 1.11 0.44
144 80 0.1 40 0.2 20.0 5.00 634 1878 26.1 1.16 0.21
145 80 0.1 30 0.2 15.0 6.67 634 2508 19.1 1.16 0.16


° mm mm kN mm
146 80 0.1 20 0.2 10.0 10.0 634 3649 12.6 1.11 0.10
147 70 0.5 50 1.0 25.0 4.00 660 808 44.1 0.61 1.73
148 70 0.5 50 0.8 31.3 3.20 660 655 33.4 0.62 1.31
149 70 0.5 50 0.6 41.7 2.40 660 492 22.6 0.63 0.88
150 70 0.5 40 1.0 20.0 5.00 660 1106 47.2 0.66 1.85
151 70 0.5 40 0.8 25.0 4.00 660 899 34.4 0.67 1.35
152 70 0.5 40 0.6 33.3 3.00 660 682 22.4 0.69 0.88
153 70 0.5 40 0.4 50.0 2.00 660 470 13.5 0.73 0.53
154 70 0.5 30 1.0 15.0 6.67 660 1696 49.1 0.75 1.92
155 70 0.5 30 0.8 18.8 5.33 660 1377 33.5 0.77 1.31
156 70 0.5 30 0.6 25.0 4.00 660 1062 21 0.80 0.82
157 70 0.5 30 0.4 37.5 2.67 660 747 12.6 0.86 0.49
158 70 0.5 20 1.0 10.0 10.0 660 3024 46.1 0.89 1.80
159 70 0.5 20 0.8 12.5 8.00 660 2508 31.7 0.92 1.24
160 70 0.5 20 0.6 16.7 6.00 660 1972 18.9 0.97 0.74
161 70 0.5 20 0.4 25.0 4.00 660 1391 11.2 1.04 0.44
162 70 0.5 20 0.2 50.0 2.00 660 720 5.2 1.12 0.20
163 70 0.5 10 0.4 12.5 8.00 660 2818 6.6 1.04 0.26
164 70 0.5 10 0.2 25.0 4.00 660 1430 3.6 1.07 0.14
165 70 0.4 50 1.0 20.0 5.00 656 1089 67.5 0.65 2.11
166 70 0.4 50 0.8 25.0 4.00 656 892 51 0.67 1.60
167 70 0.4 50 0.6 33.3 3.00 656 676 34.4 0.69 1.08
168 70 0.4 50 0.4 50.0 2.00 656 461 20.9 0.72 0.65
169 70 0.4 40 1.0 16.0 6.25 656 1474 69.5 0.70 2.18
170 70 0.4 40 0.8 20.0 5.00 656 1212 50.7 0.73 1.59
171 70 0.4 40 0.6 26.7 3.75 656 939 33.2 0.76 1.04
172 70 0.4 40 0.4 40.0 2.50 656 649 19.7 0.80 0.62
173 70 0.4 30 1.0 12.0 8.33 656 2252 69 0.80 2.16
174 70 0.4 30 0.8 15.0 6.67 656 1845 48.1 0.82 1.51
175 70 0.4 30 0.6 20.0 5.00 656 1448 30.6 0.87 0.96
176 70 0.4 30 0.4 30.0 3.33 656 994 16.6 0.91 0.52
177 70 0.4 20 0.8 10.0 10.0 656 3363 43.1 0.99 1.35
178 70 0.4 20 0.6 13.3 7.50 656 2607 25.1 1.03 0.79
179 70 0.4 20 0.4 20.0 5.00 656 1790 13.7 1.07 0.43
180 70 0.4 20 0.2 40.0 2.50 656 897 6.2 1.11 0.19
181 70 0.4 10 0.4 10.0 10.0 656 3522 8.2 1.04 0.26
182 70 0.4 10 0.2 20.0 5.00 656 1783 4.3 1.07 0.13
183 70 0.3 50 1.0 15.0 6.67 652 1622 98.9 0.73 2.32
184 70 0.3 50 0.8 18.8 5.33 652 1306 73.5 0.74 1.73
185 70 0.3 50 0.6 25.0 4.00 652 1005 50.3 0.76 1.18
186 70 0.3 50 0.4 37.5 2.67 652 676 29.5 0.78 0.69
187 70 0.3 40 1.0 12.0 8.33 652 2199 100.6 0.78 2.36
188 70 0.3 40 0.8 15.0 6.67 652 1810 70.3 0.81 1.65
189 70 0.3 40 0.6 20.0 5.00 652 1385 46.8 0.83 1.10
190 70 0.3 40 0.4 30.0 3.33 652 955 27 0.88 0.63
191 70 0.3 30 0.8 11.3 8.89 652 2757 66.9 0.92 1.57
192 70 0.3 30 0.6 15.0 6.67 652 2118 39.3 0.95 0.92
193 70 0.3 30 0.4 22.5 4.44 652 1504 23.7 1.02 0.56
194 70 0.3 30 0.2 45.0 2.22 652 789 11 1.11 0.26
195 70 0.3 20 0.6 10.0 10.0 652 3561 28.9 1.05 0.68
196 70 0.3 20 0.4 15.0 6.67 652 2388 15.7 1.07 0.37
197 70 0.3 20 0.2 30.0 3.33 652 1219 8.4 1.12 0.20
198 70 0.3 10 0.2 15.0 6.67 652 2319 4.5 1.04 0.11
199 70 0.2 50 1.0 10.0 10.0 650 2999 154.0 0.89 2.41
200 70 0.2 50 0.8 12.5 8.00 650 2409 113.2 0.90 1.77
201 70 0.2 50 0.6 16.7 6.00 650 1825 77.3 0.91 1.21


° mm mm kN mm
202 70 0.2 50 0.4 25.0 4.00 650 1242 46.1 0.94 0.72
203 70 0.2 50 0.2 50.0 2.00 650 683 23.2 1.08 0.36
204 70 0.2 40 0.8 10.0 10.0 650 3350 108.4 1.00 1.70
205 70 0.2 40 0.6 13.3 7.50 650 2518 69.9 1.00 1.09
206 70 0.2 40 0.4 20.0 5.00 650 1726 40.7 1.04 0.64
207 70 0.2 40 0.2 40.0 2.50 650 903 19.1 1.12 0.30
208 70 0.2 30 0.6 10.0 10.0 650 3589 52.8 1.07 0.83
209 70 0.2 30 0.4 15.0 6.67 650 2396 29.3 1.08 0.46
210 70 0.2 30 0.2 30.0 3.33 650 1232 13.9 1.13 0.22
211 70 0.2 20 0.4 10.0 10.0 650 3573 19.6 1.07 0.31
212 70 0.2 20 0.2 20.0 5.00 650 1825 9.2 1.10 0.14
213 70 0.2 10 0.2 10.0 10.0 650 3573 7.6 1.07 0.12
214 70 0.1 50 0.4 12.5 8.00 649 3023 69.3 1.13 0.54
215 70 0.1 50 0.2 25.0 4.00 649 1520 31.4 1.16 0.25
216 70 0.1 40 0.4 10.0 10.0 649 3665 49.2 1.09 0.39
217 70 0.1 40 0.2 20.0 5.00 649 1888 24.2 1.14 0.19
218 70 0.1 30 0.2 15.0 6.67 649 2424 16.1 1.09 0.13
219 70 0.1 20 0.2 10.0 10.0 649 3711 12.4 1.10 0.10
220 60 0.3 50 1.0 15.0 6.67 682 1775 97.5 0.76 2.11
221 60 0.3 50 0.8 18.8 5.33 682 1453 73.2 0.78 1.58
222 60 0.3 50 0.6 25.0 4.00 682 1112 48.9 0.81 1.06
223 60 0.3 50 0.4 37.5 2.67 682 764 29.2 0.85 0.63
224 60 0.3 40 1.0 12.0 8.33 682 2446 99.1 0.83 2.15
225 60 0.3 40 0.8 15.0 6.67 682 2007 70.2 0.86 1.52
226 60 0.3 40 0.6 20.0 5.00 682 1542 45.4 0.89 0.98
227 60 0.3 40 0.4 30.0 3.33 682 1071 26.7 0.94 0.58
228 60 0.3 30 0.8 11.3 8.89 682 3053 63.6 0.98 1.38
229 60 0.3 30 0.6 15.0 6.67 682 2336 38.6 1.00 0.84
230 60 0.3 30 0.4 22.5 4.44 682 1605 22.5 1.04 0.49
231 60 0.3 30 0.2 45.0 2.22 682 817 10 1.10 0.22
232 60 0.3 20 0.6 10.0 10.0 682 3752 26.3 1.06 0.57
233 60 0.3 20 0.4 15.0 6.67 682 2507 15.6 1.07 0.34
234 60 0.3 20 0.2 30.0 3.33 682 1233 7 1.08 0.15
235 60 0.3 10 0.2 15.0 6.67 682 2471 5.6 1.06 0.12
236 60 0.2 50 1.0 10.0 10.0 679 3235 150.3 0.92 2.17
237 60 0.2 50 0.8 12.5 8.00 679 2623 110.8 0.94 1.60
238 60 0.2 50 0.6 16.7 6.00 679 1988 74.1 0.95 1.07
239 60 0.2 50 0.4 25.0 4.00 679 1366 44.3 0.99 0.64
240 60 0.2 50 0.2 50.0 2.00 679 730 21.5 1.11 0.31
241 60 0.2 40 0.8 10.0 10.0 679 3537 102.3 1.01 1.48
242 60 0.2 40 0.6 13.3 7.50 679 2656 64.6 1.01 0.93
243 60 0.2 40 0.4 20.0 5.00 679 1812 38 1.05 0.55
244 60 0.2 40 0.2 40.0 2.50 679 932 17.7 1.11 0.26
245 60 0.2 30 0.6 10.0 10.0 679 3756 47.4 1.07 0.68
246 60 0.2 30 0.4 15.0 6.67 679 2494 28 1.07 0.40
247 60 0.2 30 0.2 30.0 3.33 679 1248 12.8 1.10 0.18
248 60 0.2 20 0.4 10.0 10.0 679 3814 18.5 1.08 0.27
249 60 0.2 20 0.2 20.0 5.00 679 1889 9 1.09 0.13
250 60 0.2 10 0.2 10.0 10.0 679 3748 8.5 1.07 0.12
251 60 0.1 50 0.4 12.5 8.00 677 3154 64.7 1.13 0.47
252 60 0.1 50 0.2 25.0 4.00 677 1597 29 1.17 0.21
253 60 0.1 40 0.4 10.0 10.0 677 3948 50.1 1.13 0.36
254 60 0.1 40 0.2 20.0 5.00 676 1961 22.1 1.14 0.16
255 60 0.1 30 0.2 15.0 6.67 676 2675 18.5 1.16 0.13
256 60 0.1 20 0.2 10.0 10.0 676 6758 3959 1.13 0.10

F.2. CORRELATION PLOTS AND RELIABILITY ANALYSIS PARAMETERS

FOR CODE DESIGN METHODS
This sub-appendix presents the reliability analysis parameters and correlation plots for the code design
methods for fillet welds in CHS-to-CHS X-connections evaluated in Chapter 7.
Table F.2. Reliability analysis parameters for code design methods

AWS D1.1-15 AISC 360-16 CSA S16-14
le / lw 2/3 unity unity unity
ϕ 0.80 0.80 0.75 0.67
ρM 1.12 1.12 1.12 1.12
VM 0.12 0.12 0.12 0.12
ρG 1.03 1.03 1.03 1.03
VG 0.10 0.10 0.10 0.10
ρP 2.07 1.38 1.38 1.24
VP 0.19 0.19 0.19 0.19
ρR 2.39 1.59 1.59 1.43
VR 0.24 0.24 0.24 0.24
ϕ+ 0.77 0.89 0.87 0.87
+
6.2 4.3 4.6 4.6
Fig. F.1. Correlation of existing AWS D1.1-15 provisions with all test results, with weld effective lengths

Fig. F.2. Correlation of existing AWS D1.1-15 provisions (excluding weld effective lengths) and AISC 360-
16 provisions with all test results
Fig. F.3. Correlation of CSA S16-14 provisions with all test results

Weld Effective Lengths For Hollow Structural Section Connections PDF

Загружено:

Сведения о документе

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Weld Effective Lengths For Hollow Structural Section Connections PDF

Загружено:

Авторское право:

Доступные форматы

WELD EFFECTIVE LENGTHS FOR

HOLLOW STRUCTURAL SECTION CONNECTIONS

A thesis submitted in conformity with the

2.8.1. Footnote ..................................................................................................................... 27

Weld Effective Lengths for Hollow Structural Section Connections

5.2. Previous Experimentation ....................................................................................................... 67

Weld Effective Lengths for Hollow Structural Section Connections

7.2.2. CIDECT (Wardenier et al. 2008) ............................................................................. 122

Weld Effective Lengths for Hollow Structural Section Connections

Weld Effective Lengths for Hollow Structural Section Connections

Table A.1. List of truss fabrication drawings ............................................................................................... 159

Weld Effective Lengths for Hollow Structural Section Connections

Weld Effective Lengths for Hollow Structural Section Connections

Table E.12. Weld measurements for 127-406-60a......................................................................................... 241

Weld Effective Lengths for Hollow Structural Section Connections

Fig. 1.1. Applications of HSS ........................................................................................................................... 1

Weld Effective Lengths for Hollow Structural Section Connections

Weld Effective Lengths for Hollow Structural Section Connections

Weld Effective Lengths for Hollow Structural Section Connections

Weld Effective Lengths for Hollow Structural Section Connections

Fig. E.9. CHS-to-CHS template geometry view no. 2 ................................................................................ 233

Weld Effective Lengths for Hollow Structural Section Connections

AISC = American Institute of Steel Construction

CHS = circular hollow section

A = cross-sectional area of the HSS chord

Db = overall diameter of the CHS branch

Weld Effective Lengths for Hollow Structural Section Connections

Weld Effective Lengths for Hollow Structural Section Connections

Weld Effective Lengths for Hollow Structural Section Connections

Weld Effective Lengths for Hollow Structural Section Connections

1.1. HOLLOW STRUCTURAL SECTIONS

Fig. 1.1. Applications of HSS

1.2. WELD EFFECTIVE LENGTHS

Weld Effective Lengths for Hollow Structural Section Connections

1.3. RESEARCH PROGRAM GOAL AND OVERVIEW

Weld Effective Lengths for Hollow Structural Section Connections

Weld Effective Lengths for Hollow Structural Section Connections

Weld Effective Lengths for Hollow Structural Section Connections

2.2.2. TRUSS DESIGN

Weld Effective Lengths for Hollow Structural Section Connections

Table 2.1. Measured properties of nine RHS overlapped K- (test) connections

Weld Effective Lengths for Hollow Structural Section Connections

Fig. 2.1. Elevation of the truss, dimensions and joint designations

2.2.2.2. Weld Joint Details

Weld Effective Lengths for Hollow Structural Section Connections

Weld Effective Lengths for Hollow Structural Section Connections

2.3. GEOMETRIC AND MECHANICAL PROPERTIES

Weld Effective Lengths for Hollow Structural Section Connections

Fig. 2.4. Typical measurement procedure and mould profile

Weld Effective Lengths for Hollow Structural Section Connections

2.4. INSTRUMENTATION AND LOADING STRATEGY

Weld Effective Lengths for Hollow Structural Section Connections

2.4.2. LOADING STRATEGY

Weld Effective Lengths for Hollow Structural Section Connections

Weld Effective Lengths for Hollow Structural Section Connections

(a) Test K-60-0.71a (b) Test K-30-0.50a

(c) Test K-30-0.50b (d) Test K-90-0.71

Fig. 2.7. Typical weld fractures (and instrumentation)

Weld Effective Lengths for Hollow Structural Section Connections

Weld Effective Lengths for Hollow Structural Section Connections

Weld Effective Lengths for Hollow Structural Section Connections

Weld Effective Lengths for Hollow Structural Section Connections

2.6. EVALUATION OF AISC 360 (2010)