Thesis Example

Rectangular Hollow Sections with circular
web openings
Fundamental behaviour in torsion, bending and shear
by D.J.Ridley-Ellis, B.Eng. M.Res.
Thesis submitted to the University of Nottingham
for the degree of Doctor of Philosophy, October 2000

ii
Since men were of an imitative and teachable nature, they

boasted of their inventions as they daily showed their various
achievements in building, and thus, exercising their talents in
rivalry, were rendered of better judgement daily.
Vitruvius, circa 27 B.C.E.

De Architectura, (Liber Secundus, I)
(Granger Translation, University College Nottingham, 1931)
iii
Table of contents
Table of contents iii

List of figures vii
List of tables xiii
Abstract (long version) xv
Abstract (short version) xviii
Acknowledgements xx
Author’s declaration xxi
Notation and symbols xxii
List of abbreviations and trademarks xxviii
1 Introduction and background 1-1
1.1 Introduction 1-1
1.2 Background 1-1
1.2.1 An outline of the slim floor method of construction 1-1
1.2.2 Slimdek 1-4
1.2.3 Slim floor edge beams 1-5
1.3 Methodology 1-9
1.4 Scope of study and dimensions of cross-section 1-10
1.5 The contents of this Thesis 1-13
2 Relevant knowledge and theory 2-1
2.2 The stress-strain behaviour of hot finished steel 2-2
2.2.1 Introduction 2-2
2.2.2 The stress-strain behaviour of hot finished steel and BS EN 10002-1 2-3
2.2.3 A mathematical model of stress-strain behaviour 2-5
2.2.4 Applying the mathematical approximation to experimental data 2-8
2.2.4.1 Applying the mathematical approximation to a single coupon result 2-8
2.2.4.2 Measuring the suitability of the fit 2-11
2.2.4.3 Applying the mathematical model to a group of coupon results 2-15
2.2.5 Multi-linear approximations 2-16
2.3 Finite Element analysis 2-17
2.3.2 The Finite Element approach 2-17
2.3.3 Good practice 2-19
2.3.4 Stress accuracy 2-21
2.3.5 Non-linear problems 2-22
2.3.6 Shell elements 2-23
2.3.7 The ABAQUS Finite Element analysis program 2-24
2.4 Experimental strain and stress measurement 2-25
2.4.2 The components of stress 2-26
2.4.3 The components of strain 2-29
2.4.4 Strain measurement 2-32
2.4.4.1 Introduction 2-32
2.4.4.2 Electrical resistance strain gauges 2-34
iv
2.4.4.3 Brittle coatings 2-38

2.5 Torsion theory 2-41
2.5.2 Basic torsion theory 2-41
2.5.3 Warping 2-48
2.5.4 Calculation of torsional capacity 2-53
2.5.4.1 Expressions for elastic and plastic capacity 2-53
2.5.4.2 The shear yield stress and tensile yield stress 2-54
2.5.4.3 Experimental measurement of the shear yield stress for hot finished steel 2-56
2.5.5 Experimental studies of the torsional properties of tubes 2-59
2.6 Summary 2-60
3 Bending (four-point loading) 3-1
3.2 A description of the apparatus and testing procedure 3-2
3.2.1 The four-point bending test apparatus 3-2
3.2.2 The four-point bending test specimens 3-5
3.2.3 Instrumentation 3-6
3.2.3.1 Measurement of bending moment 3-6
3.2.3.2 Measurement of curvature 3-6
3.2.3.3 Measurement of strain 3-7
3.2.3.4 Other measurements 3-9
3.2.4 Testing procedure 3-9
3.3 Specimen types, dimensions and properties 3-10
3.3.1 Specimens tested 3-10
3.3.2 Material properties 3-11
3.3.3 Dimensions and geometrical properties 3-12
3.4 The experimental results 3-14
3.5 A comparison of experimental results and analytical predictions 3-20
3.6 Concluding remarks 3-32
4 Shear (three-point loading) 4-1
4.2.1 The three-point bending test apparatus 4-2
4.2.2 The three-point bending test specimens 4-4
4.2.3.1 Measurement of shear force and bending moment 4-5
4.2.3.2 Measurement of deflection 4-5
5 Full-scale torsion 5-1
5.2.1 The torsion test apparatus 5-1
5.2.2 The torsion test specimens 5-4
5.2.3 Connection of the specimen and rig 5-5
5.2.4.1 Measurement of torque 5-6
5.2.4.2 Measurement of rotation 5-9
v

5.6 The validity of the experimental findings 5-32
5.6.2 Measurement of rotation 5-32
5.6.3 Measurement of torque 5-34
5.6.4 Measurement of strain 5-35
5.6.5 Repeatability of experimental results 5-36
5.6.6 Accuracy of the data measurement and logging system 5-36
5.6.7 Additional loads on the specimen 5-37
5.6.8 The validity of the theoretical and analytical predictions 5-38
5.6.8.1 Introduction 5-38
5.6.8.2 Boundary conditions 5-38
5.6.8.3 Cross-section dimensions 5-40
5.6.8.4 Material properties 5-40
5.6.9 Marshall’s results 5-48
5.6.10 Summary 5-51
5.7 Suitable schemes for further investigation 5-52
6 Small-scale torsion 6-1
6.2.1 The torsion test apparatus and specimens 6-2
6.2.2 Types of specimen and scope of study 6-4
6.3.1 Measured torque-rotation relationships 6-10
6.3.2 Mechanisms of failure 6-14
6.4.2 Deformed shapes 6-18
6.4.3 Torque-rotation relationships 6-20
6.5 Stress distributions 6-25
7 Parametric study and design recommendations 7-1
7.2 Scope and limitations 7-2
7.3 General design procedure 7-5
7.4 The parametric Finite Element study 7-8
7.5 Bending 7-13
7.5.2 Concentration factors for elastic stress 7-17
7.5.3 Resistance 7-24
7.5.4 Deflection 7-37
7.5.5 Zone of influence 7-40
7.5.6 Design recommendation for calculation of resistance 7-41
7.6 Shear 7-43
7.7 Torsion 7-60
vi

7.8 Combinations 7-77
8 Conclusions 8-1
8.2 Laboratory testing and calibration of Finite Element models 8-1
8.2.1 Four-point bending (bending) 8-1
8.2.2 Three-point bending (shear) 8-2
8.2.3 Full-scale torsion 8-3
8.2.4 Small-scale torsion 8-4
8.3 The parametric Finite Element study 8-5
8.3.1 Bending 8-5
8.3.2 Shear 8-6
8.3.3 Torsion 8-7
8.5 Lists of conclusions 8-9
8.5.1 Bending of perforated RHS (without elastic instability) 8-9
8.5.2 Shearing of perforated RHS (without elastic instability) 8-10
8.5.3 Torsion of perforated RHS (without elastic instability) 8-11
Appendices I
Table of contents II
List of figures III
List of tables IV
A. Calibration of the Avery torsion testing machine VI
B. Sensitivity of sectional properties XVII
C. Formulae for the calculation of the properties of cross-section XXV
D. Material properties XXIX
E. Stress concentration factors (Finite Element) LII
F. Calibration of testing equipment LVIII
References LX
Standards and technical design documents LXVI
Papers and project reports LXVIII
Errata LXIX
vii
List of figures
Figure 1-1: Slim floor and conventional construction 1-3

Figure 1-2: Nineteenth century ‘slim floor’ construction 1-3
Figure 1-3: Examples of internal slim floor beams 1-4
Figure 1-4: The Slimdek internal beam (ASB) 1-5
Figure 1-5: Composite and non-composite edge beam construction 1-7
Figure 1-6: Slim floor edge beam construction 1-8
Figure 1-7: Slim floor components 1-8
Figure 1-8: Perforations 1-11
Figure 1-9: Dimensions (cross-section and corner profile) 1-12
Figure 1-10: Dimensions (web opening) 1-12
Figure 2-1: The stress-strain behaviour of steel from hot finished RHS 2-4
Figure 2-2: The yield plateau 2-5
Figure 2-3: The mathematical model 2-7
Figure 2-4: Evaluating the difference between two functions using area 2-13
Figure 2-5: The difference between the approximation and the experimental data 2-14
Figure 2-6: The mathematical approximation and the yield plateau 2-14
Figure 2-7: The mathematical approximation and the experimental data 2-15
Figure 2-8: A multi-linear model 2-17
Figure 2-9: Components of stress acting on a small cubic element 2-28
Figure 2-10: A rotation of the coordinate system 2-29
Figure 2-11: Mohr’s circle for stress (two-dimensions) 2-29
Figure 2-12: Mohr’s circle for strain (two-dimensions) 2-32
Figure 2-13: An electrical resistance strain gauge 2-37
Figure 2-14: A three-element strain gauge rosette 2-38
Figure 2-15: Measurements of strain on a three-element strain gauge rosette 2-38
Figure 2-16: Three types of yield lines in brittle coatings 2-40
Figure 2-17: Yield lines in a brittle coating due to bending 2-40
Figure 2-18: Stress distributions around web openings (yield line patterns) 2-41
Figure 2-19: Torsional inertia constant (Marshall and BS EN 10210 calculations) 2-46
Figure 2-20: Torsional modulus constant (Marshall and BS EN 10210 calculations) 2-46
Figure 2-21: Torsional inertia constant (thin and thick wall calculations) 2-47
Figure 2-22: Torsional modulus constant (thin and thick wall calculations) 2-47
Figure 2-23: Warping of rectangular cross-sections 2-51
viii
Figure 2-24: Warping of RHS in torsion (theory and FE) 2-52

Figure 2-25: The effect of end restraint on torsional behaviour of RHS 2-52
Figure 2-26: Warping at an unrestrained end adjacent to a torsional buckle 2-53
Figure 2-27: The von Mises and Tresca failure envelopes 2-56
Figure 2-28: The shear testing coupon 2-58
Figure 2-29: Shear testing coupon results 2-58
Figure 3-1: Schematic of the four-point bending test 3-3
Figure 3-2: The four-point loading test apparatus 3-3
Figure 3-3: The supports (rocker and fixed) 3-4
Figure 3-4: A dual axis rocker bearing 3-4
Figure 3-5: Four-point bending test specimen 3-5
Figure 3-6: Measurement of curvature with linear potentiometers 3-6
Figure 3-7: Strain gauge pattern (not REPEAT1, REPEAT4, DREB1 and DREB2) 3-7
Figure 3-8: Strain gauge pattern (tests DREB1 and DREB2) 3-8
Figure 3-9: Strain gauge pattern (test REPEAT4) 3-8
Figure 3-10: Experimental results RHS 200x100x8 no holes 3-17
Figure 3-12: Experimental results RHS 200x100x8 one hole 165 mm dia 3-18
Figure 3-14: Experimental results RHS 200x100x8 two holes 165 mm dia 3-19
Figure 3-16: FE and experimental (tests REPEAT1 and REPEAT5) 3-25
Figure 3-17: FE and experimental (tests REPEAT3 and REPEAT4) 3-25
Figure 3-18: FE and experimental (test TEST6) 3-26
Figure 3-19: FE and experimental (test DREB1) 3-26
Figure 3-20: FE and experimental (tests TEST5A and TEST5B) 3-27
Figure 3-21: FE and experimental (test DREB2) 3-27
Figure 3-22: Displaced shape RHS 200x100x8 no holes 3-28
Figure 3-23: Displaced shape RHS 200x100x8 one hole 3-28
Figure 3-24: Displaced shape RHS 200x100x8 two holes 3-28
Figure 3-25: Displaced shape RHS 250x250x10 no holes 3-29
Figure 3-26: Displaced shape RHS 250x250x10 one hole 3-29
Figure 3-27: Displaced shape RHS 250x250x10 two holes 3-29
Figure 3-28: Displaced shape RHS 200x100x8 two holes (superposition) 3-30
Figure 3-29: Longitudinal strain for RHS 250x250x10 unperforated (REPEAT4) 3-30
Figure 3-30: Principal strains for RHS 250x250x10 one hole (unperforated web) 3-31
Figure 3-31: Principal strains for RHS 250x250x10 one hole (perforated web) 3-31
Figure 4-1: Schematic of the three-point bending test 4-3
Figure 4-2: The three-point bending test apparatus 4-3
Figure 4-3: Three-point bending test specimen 4-4
ix
Figure 4-4: Measurement of vertical displacement with linear potentiometers 4-5

Figure 4-5: Strain gauge pattern (all tests) 4-6
Figure 4-6: Inclinometers for measurement of twist 4-7
Figure 4-7: Measurement of twist at the rocker support 4-8
Figure 4-8: Experimental results RHS 200x100x8 4-14
Figure 4-9: Experimental results RHS 250x250x10 4-14
Figure 4-10: Shear force-displacement relationship RHS 200x100x8 no holes 4-15
Figure 4-11: Displacement profiles RHS 200x100x8 no holes 4-15
Figure 4-12: Shear force-displacement relationship RHS 200x100x8 one hole 4-16
Figure 4-13: Displacement profiles RHS 200x100x8 one hole (perforated web) 4-17
Figure 4-14: Displacement profiles RHS 200x100x8 one hole (unperforated web) 4-17
Figure 4-15: Shear force-displacement relationship RHS 200x100x8 two holes 4-18
Figure 4-16: Displacement profiles RHS 200x100x8 two holes 4-18
Figure 4-17: FE and experimental (test TESTS1) 4-22
Figure 4-23: FE and experimental (twist of TESTS3) 4-25
Figure 4-24: FE and experimental (twist of TESTS6) 4-25
Figure 4-25: Displaced shapes (Finite Element, perforated) 4-26
Figure 4-26: Displaced shapes (experimental, perforated) 4-26
Figure 4-27: Displaced shapes (Finite Element, unperforated) 4-27
Figure 4-28: Displaced shapes (Experimental, unperforated) 4-27
Figure 4-29: Displaced shape RHS 200x100x8 two holes (superposition) 4-28
Figure 5-1: The torsion testing apparatus (schematic) 5-3
Figure 5-2: The torsion testing apparatus 5-3
Figure 5-3: Torsion test specimen 5-4
Figure 5-4: The connecting plates and bolting arrangement 5-6
Figure 5-5: The calibration shaft 5-7
Figure 5-6: Calibration data 5-8
Figure 5-7: Calibration data (averaged measurements) 5-8
Figure 5-8: Calibration of the load-indicating shaft 5-9
Figure 5-9: Inclinometers for measurement of rotation 5-10
Figure 5-10: Inclinometer locations 5-10
Figure 5-11: Strain gauge pattern (tests TT1-TT6) 5-11
Figure 5-12: Strain gauge pattern (test TT7) 5-11
Figure 5-14: Experimental results RHS 150x150x6.3 no holes 5-18
x
Figure 5-16: Experimental results RHS 150x150x6.3 one hole 99 mm dia 5-19
Figure 5-18: Experimental results RHS 150x150x6.3 two holes 99 mm dia 5-20
Figure 5-19: Experimental result RHS 150x150x6.3 two holes 38.1 mm dia 5-21
Figure 5-20: FE and experimental (test TT4) 5-25
Figure 5-26: FE and experimental (tests TT7 and TT14) 5-28
Figure 5-33: Displaced shape RHS 150x150x6.3 one hole 99 mm dia 5-31
Figure 5-34: Displaced shape RHS 200x100x8 two holes 165 mm dia 5-32
Figure 5-35: Rotation measurements 5-33
Figure 5-36: Load measurements 5-35
Figure 5-37: Boundary conditions RHS 200x100x8 grade S275J2H 1 hole 5-39
Figure 5-38: Boundary conditions RHS 150x150x6.3 grade S275J2H 1 hole 5-39
Figure 5-39: Reduced yield RHS 200x100x8 grade S275J2H no holes 5-44
Figure 5-40: Reduced yield RHS 150x150x6.3 grade S275J2H no holes 5-44
Figure 5-41: Reduced yield RHS 200x100x8 grade S275J2H one hole 5-45
Figure 5-42: Reduced yield RHS 150x150x6.3 grade S275J2H one hole 5-45
Figure 5-43: Reduced yield RHS 200x100x8 grade S275J2H two holes 5-46
Figure 5-44: Reduced yield RHS 150x150x6.3 grade S275J2H two holes 5-46
Figure 5-45: Strain measurements (test TT4 south web) 5-47
Figure 5-46: Strain measurements (test TT4 north web) 5-47
Figure 5-47: Measurements and predictions of torque at yield (unperforated) 5-51
Figure 6-1: The small-scale torsion rig 6-3
Figure 6-2: Small-scale torsion test specimen 6-3
Figure 6-3: The inserts and clamping plates 6-4
Figure 6-4: Types of specimen (developed views) 6-7
Figure 6-5: Stress-strain relationship from tensile tests 6-7
Figure 6-6: Stress-strain relationship for cold-formed and hot finished steel (1%) 6-9
Figure 6-7: Stress-strain relationship for cold-formed and hot finished steel 6-9
xi
Figure 6-8: Torque rotation curves – One hole (Type II) 6-12
Figure 6-9: Torque rotation curves – Two holes facing (Type III) 6-12
Figure 6-10: Peak torque and hole diameter – Types I, II & III 6-13
Figure 6-11: Torque-rotation curves for type IV specimens 6-13
Figure 6-12: Peak torque and hole separation (Type IV) 6-14
Figure 6-13: Post-yield failure by buckling 6-16
Figure 6-14: Failure of type II specimens 6-16
Figure 6-15: Failure of type III specimens 6-16
Figure 6-16: Failure of type IV specimens 6-17
Figure 6-17: Regions of high tensile stress around the hole 6-17
Figure 6-18: Displaced shape (type I) 6-18
Figure 6-19: Displaced shape (type II 10 mm dia) 6-18
Figure 6-20: Displaced shape (type II 30 mm dia) 6-19
Figure 6-21: Displaced shape (type III 20 mm dia) 6-19
Figure 6-22: Displaced shape (type III 30 mm dia) 6-19
Figure 6-23: FE and experimental for type I specimens (no holes) 6-23
Figure 6-24: FE and Experimental for type II specimens (one hole) 6-23
Figure 6-25: FE and experimental for type III specimens (two holes facing) 6-24
Figure 6-26: The influence of boundary conditions 6-25
Figure 6-27: Concentration factors for Von Mises stress around a small hole 6-26
Figure 6-28: Concentration factors for Von Mises stress around a large hole 6-27
Figure 7-1: Perforated and unperforated sub-members 7-7
Figure 7-2: Locations for stress concentration 7-7
Figure 7-3: Model coordinate axes 7-12
Figure 7-4: Finite Element meshes for parametric investigation 7-12
Figure 7-5: Cross-section classifications and moment-rotation relationship 7-17
Figure 7-6: Distribution of elastic bending stress 7-21
Figure 7-7: Distribution of von Mises stress for RHS 250x250x10 in bending 7-22
Figure 7-8: Parametric sampling for bending 7-23
Figure 7-9: Stress concentration factors for bending 7-23
Figure 7-10: Critical hole size ratio for bending 7-24
Figure 7-11: Elastic moment capacity 7-29
Figure 7-12: The significance of strain hardening for beam ductility 7-30
Figure 7-13: Distortions under the action of bending (two holes) 7-31
Figure 7-14: Distortions under the action of bending (one hole) 7-31
Figure 7-15: Moment-curvature relationship for RHS 250x250x10 one hole 7-32
Figure 7-16: Moment-curvature relationship for RHS 250x250x10 two holes 7-32
xii
Figure 7-21: Yield in bending for sections with one hole (FE and design) 7-35
Figure 7-22: Yield in bending for sections with two holes (FE and design) 7-35
Figure 7-23: Maximum moment for sections with one hole (FE and design) 7-36
Figure 7-24: Maximum moment for sections with two holes (FE and design) 7-36
Figure 7-25: Maximum moment without strain hardening 7-37
Figure 7-26: Bending displacement of RHS 200x100x8 two holes 140.8 mm dia 7-40
Figure 7-27: Critical cantilever length for shear 7-45
Figure 7-28: Distribution of elastic shear stress 7-48
Figure 7-29: Parametric sampling for shear 7-48
Figure 7-30: Distribution of von Mises stress for RHS 200x100x8 in shear 7-49
Figure 7-31: Stress concentration factors for shear 7-50
Figure 7-32: Shear force-shearing angle relationship for RHS 250x250x10 one hole 7-53
Figure 7-33: Shear force-shearing angle relationship for RHS 250x250x10 two holes 7-53
Figure 7-38: The vierendeel mechanism of shear failure 7-56
Figure 7-39: Distortions under the action of shear 7-57
Figure 7-40: Shear displacement of RHS 200x100x8 two holes 140.8 mm dia 7-57
Figure 7-41: Parametric sampling for torsion 7-64
Figure 7-42: Stress concentration factors for torsion 7-64
Figure 7-43: Distribution of von Mises stress for RHS 250x250x10 in torsion 7-65
Figure 7-44: Torque-rotation relationship for RHS 250x250x10 one hole 7-68
Figure 7-45: Torque-rotation relationship for RHS 250x250x10 two holes 7-68
Figure 7-50: Maximum torque for RHS 250x250x10 7-71
Figure 7-51: Maximum torque for RHS 200x100x8 7-71
Figure 7-52: Distortions under the action of torsion 7-73
Figure 7-53: Torsional twist of RHS 250x250x10 two holes 7-74
Figure 7-54: Torsional twist of RHS 200x100x8 two holes 7-74
Figure 7-55: Stiffness reduction factors for torsion 7-75
Figure 7-56: Contours of bending capacity with coexistent shear and torsion 7-80
Figure 7-57: Bending capacity with coexistent shear or torsion 7-80
Figure 7-58: Possible stiffening or reinforcement of web openings 7-83
xiii
List of tables
Table 3-1: Specimens tested 3-11

Table 3-2: Material models 3-12
Table 3-3: Dimensions of cross-section (measured) 3-13
Table 3-4: Properties of cross-section (from measured dimensions) 3-13
Table 3-5: Sectional properties (measured dims and material props, theoretical) 3-13
Table 3-6: Sectional properties (nominal, theoretical) 3-14
Table 3-7: Experimental results – stiffness and maximum bending moment 3-16
Table 3-8: Experimental results and theory– stiffness and max bending moment 3-16
Table 3-9: Finite Element results – stiffness and maximum bending moment 3-23
Table 3-10: Experimental and FE results – stiffness and maximum bending moment 3-24
Table 4-7: Experimental results – stiffness and maximum shear force 4-13
Table 4-8: Finite Element results – stiffness and maximum shear force 4-21
Table 4-9: Experimental and FE results – stiffness and maximum shear force 4-21
Table 5-7: Experimental results – stiffness and maximum torque 5-17
Table 5-8: Measurements of torsional shear strain in unperforated sections 5-17
Table 5-9: Finite Element results – stiffness and maximum torque 5-23
Table 5-10: Experimental and FE results – stiffness and maximum torque 5-24
Table 5-11: Dimensions and material properties of Marshall’s torsion test specimens 5-49
Table 5-12: Properties of cross-section for Marshall’s torsion test specimens 5-49
Table 5-13: Theoretical plastic torque and torque at yield for Marshall’s torsion tests 5-50
Table 5-14: Plastic torque and torque at yield measured by Marshall 5-50
xiv

Table 6-2: ABAQUS material models for cold-formed tube 6-8
Table 6-6: Maximum torque (average experimental and FE) 6-22
Table 6-7: Elastic stiffness from Finite Element (average over 240 mm length) 6-22
Table 6-8: Concentration factors for Von Mises stress 6-26
Table 7-1: Equivalence of material grades for design 7-2
Table 7-2: RHS in the European Standard range 7-5
Table 7-3: Material model for parametric study 7-11
Table 7-4: ABAQUS material model for parametric study 7-11
Table 7-5: Boundary conditions for parametric FE models 7-11
Table 7-6: Deflection in bending 7-39
Table 7-7: Stress concentration factors for bending, ψbending 7-43
Table 7-8: Critical hole size ratio for bending, Φcr 7-43
Table 7-9: Stress concentration factors for shear, ψshear 7-59
Table 7-10: Stress concentration factors for torsion, ψtorsion 7-77
xv
Abstract (long version)
The use of Rectangular Hollow Sections (RHS) as edge beams in steel frame
building construction is becoming increasingly common. Edge beams are subject to
out of balance loads and RHS members are usually specified to take advantage of
their comparatively high torsional rigidity. Although the tubular nature of the RHS
offers an ideal opportunity for improved service integration, the influence of an
opening in the web upon the structural performance of the section has been found to
be significant.
The primary aim of the project was to develop the basis for a set of rules to
be used in the design of RHS slim floor edge beams with web openings. However,
since the research is fundamental in nature, the findings are also applicable to other
situations where it may be advantageous to cut holes in load bearing tubular
members of rectangular cross-section.
The study considered the influence of the number and size of holes upon
resistance and stiffness in bending, shear, and torsion and combined analytical Finite
Element modelling with large and small scale pseudo-static (short-time static
loading) laboratory testing.
Large web openings were found to cause a significant reduction in both the
torsional capacity (up to 60%) and stiffness (up to 40%). The reduction in stiffness
was due to the perforated zone being much more flexible than the unperforated beam
and severe deformations in this zone were observed. The reduction in capacity was
xvi
found to be comparable to the reduction in the cross-sectional area of the perforated
web.
The reduction in the shear capacity due to large web openings was found to
be as much as 85%, and the shear stiffness of the perforated zone was also reduced
significantly. The reduction in the shear capacity was found to be comparable to the
reduction in the cross-sectional area of the webs at the location of the hole.
The reduction in the bending capacity due to large web openings was found
to be as much as 30%, but the reduction in elastic stiffness was slight (less than 5%).
The reduction in the bending capacity was found to be comparable to the reduction
in the elastic and plastic section moduli at the location of the hole. In some cases,
web openings were observed to decrease the plastic stability of the cross-section
causing a reduction in the rotation capacity.
In all but one category of tests (see below), good agreement was achieved
between experimentally measured quantities (such as capacities, deflections and
strains) and the corresponding Finite Element predictions, allowing parametric
investigations to be conducted with calibrated analytical models. Generally, the
predictions of elastic and plastic capacities were within 15% of the measured
quantities and predictions of elastic stiffness within 20%.
Preliminary design recommendations are presented based on the results of
the parametric study and laboratory tests. The design advice was developed with
regard to existing recommendations for the design of perforated I-beams, and was
produced in a form that allows integration with modern limit state design codes.
Aspects of behaviour requiring further investigation have been identified and
categorised.
xvii
Torsion tests on full-scale RHS without web openings yielded some
unexpected results. Elastic and plastic capacities measured in the laboratory were
significantly lower (12–20%) than those predicted by the Finite Element models and
the thick walled torsion theory used as the basis of the British and European design
procedures. Attempts were made to determine the cause of this behaviour and a
number of possibilities were eliminated. Although the anomalous results have not
been fully explained, evidence of similar behaviour in previous full-scale testing was
discovered.
xviii
Abstract (short version)
The primary aim of the project was to develop the basis for a set of rules to
be used in the design of RHS slim floor edge beams with web openings. However,
since the research is fundamental in nature, the findings are also applicable to other
situations where it may be advantageous to cut holes in load bearing tubular
members of rectangular cross-section.
The study considered the influence of the number and size of holes upon
resistance and stiffness in bending, shear, and torsion and combined analytical Finite
Element modelling with large and small scale pseudo-static (short-time static
loading) laboratory testing.
In all but one category of tests (see below), good agreement was achieved
between experimentally measured quantities (such as capacities, deflections and
strains) and the corresponding Finite Element predictions, allowing parametric
investigations to be conducted with calibrated analytical models.
Preliminary design recommendations are presented based on the results of
the parametric study and laboratory tests. The design advice was developed with
regard to existing recommendations for the design of perforated I-beams, and was
produced in a form that allows integration with modern limit state design codes.
Aspects of behaviour requiring further investigation have been identified and
categorised.
xix
Torsion tests on full scale RHS without web openings yielded some
unexpected results. Elastic and plastic capacities measured in the laboratory were
significantly lower (12–20%) than those predicted by the Finite Element models and
the thick walled torsion theory used as the basis of the British and European design
procedures. Attempts were made to determine the cause of this behaviour and a
number of possibilities were eliminated. Although the anomalous results have not
been fully explained, evidence of similar behaviour in previous full-scale testing was
discovered.
xx
Acknowledgements
I would like to offer my thanks for the invaluable support, guidance and
motivation provided by my supervisors, Dr. Gwynne Davies and Dr. John Owen, of
the University of Nottingham and by my project advisors, Eddie Hole and Noel
Yeomans of British Steel Tubes and Pipes plc (now known as Corus Tubes).
Thanks are also due to Gordon Hardy, Gary Hayes, Geoff Mitchell, Bal Loyla, Brian
Whitehouse, Nigel Rook and Bob Collins for their expedient and proficient technical
assistance, and to the University for the use of its facilities. I would also like to
thank everyone else who has helped me on the way, most notably my parents for
their continuing patience.
Funding for this research was provided jointly and graciously by the
University of Nottingham and by British Steel Tubes and Pipes plc.

xxi
Author’s declaration
I declare that, unless otherwise stated in the text, the contents of this thesis
are the results of my own work conducted under the guidance of my supervisors.
Daniel James Ridley-Ellis

Digitally signed by
Daniel Ridley-Ellis
Daniel DN: cn=Daniel

Ridley-Ellis, o=University
of Nottingham,
ou=School of Civil
Signature
Not Verified
Ridley-Ellis Engineering, c=GB
Date: 2001.09.17
15:45:43 Z
xxii
Notation and symbols
Latin
a Horizontal semi axis of the strain hardening ellipse
A Area (of cross-section)
Ah Area enclosed by mean perimeter
Av Shear area
Aζ Intermediate quantity (BS EN 10210-2:1997)
Aξ Intermediate quantity (BS EN 10210-2:1997)
b Length of the shorter side of a Rectangular Hollow Section (width)
b Vertical semi axis of the strain hardening ellipse
Ct Torsional modulus constant
D The diameter of a circular web opening
E Young’s modulus
Eplateau Stiffness of the yield plateau
fu Ultimate (material) strength
fy Yield strength
fy1 Yield strength (mathematical model)
fy2 Stress at onset of strain hardening
fy,reduced Reduced yield strength allowing for shear and/or torsion
G Shear modulus
h Length of the longer side of a Rectangular Hollow Section (height)
hc Mean perimeter
hζ Intermediate quantity (BS EN 10210-2:1997)
hξ Intermediate quantity (BS EN 10210-2:1997)
iy Radius of gyration (axis parallel to flanges)
It Torsional inertia constant
It,unperf Torsional inertia constant for an unperforated section
Iy Second moment of area (axis parallel to flanges)
Iy,unperf Second moment of area for an unperforated section (axis parallel to flanges)
lV,cr Critical cantilever length for shear
xxiii
Iζζ Intermediate quantity (BS EN 10210-2:1997)

Iξξ Intermediate quantity (BS EN 10210-2:1997)
K The suitability of fit (total energy error proportion)
K Intermediate quantity (BS EN 10210-2:1997)
Kends The stiffness of the small-scale torsion testing machine
l Separation of strain gauges
lcr Critical cantilever length for shear
L Length of a member
M Bending moment
Mc,y,Rd Design moment of resistance (bending about axis parallel to flanges)
Me Elastic moment
Mel,y Moment at yield (bending about axis parallel to flanges)
Mel,y,Rd,1hole Moment at yield for a section with one opening (design value)
Mel,y,Rd,2holes Moment at yield for a section with two openings (design value)
Mel,y,Rd,unperf Moment at yield for an unperforated section (design value)
Mp Full plastic moment
Mpl,y Full plastic moment (bending about axis parallel to flanges)
Mpl,y,Rd,1hole Moment of resistance for a section with one opening (design value)
Mpl,y,Rd,2holes Moment of resistance for a section with two openings (design value)
Mpl,y,Rd,unperf Moment of resistance for an unperforated section (design value)
My,Sd Applied moment (about axis parallel to flanges)
n Number of openings
r Radius of web opening (not ro as in EC3 Annex N)
r Radius of a circular bar
r Radius to a strain gauge from the centre of an opening
ro External corner radius
ri Internal corner radius
R Rotation capacity
Rc Mean corner radius
ReH Upper yield strength as defined by BS EN 10002-1:1990
ReL Lower yield strength as defined by BS EN 10002-1:1990
Rm Tensile strength as defined by BS EN 10002-1:1990
s The distance between two openings in the same web
s x-coordinate of the centre of the strain hardening ellipse
t Wall thickness
t y-coordinate of the centre of the strain hardening ellipse
xxiv
T Torque (torsional moment)

Tel Torque at yield
Tel,Rd,1hole Torque at yield for a section with one opening (design value)
Tel,Rd,2holes Torque at yield for a section with two openings (design value)
Tel,Rd,unperf Torque at yield for an unperforated section (design value)
Tpl Torque of resistance
Tpl,Rd,1hole Torque of resistance for a section with one opening (design value)
Tpl,Rd,2holes Torque of resistance for a section with two openings (design value)
Tpl,Rd,unperf Torque of resistance for an unperforated section (design value)
TSd Applied torque
Vel,y,Rd,1hole Design shear resistance for a section with one opening (yield limited)
Vel,y,Rd,2holes Design shear resistance for a section with two openings (yield limited)
Vpl,y,Rd,unperf Design shear resistance for an unperforated section (plastic)
Vy Shear resistance (the product of the shear area and the shear yield stress)
Vy,Sd Applied shear force (parallel to webs)
w Longitudinal warping distortion at the corners of an RHS
Wel,y Elastic section modulus (axis parallel to flanges)
Wel,y,1hole Elastic section modulus for an RHS with one hole (axis parallel to flanges)
Wel,y,2holes Elastic section modulus for an RHS with two holes (axis parallel to flanges)
Wel,y,unperf Elastic section modulus for an unperforated section (axis parallel to flanges)
Wpl,y Plastic section modulus (axis parallel to flanges)
Wpl,y,1hole Plastic section modulus for an RHS with one hole (axis parallel to flanges)
Wpl,y,2holes Plastic section modulus for an RHS with two holes (axis parallel to flanges)
Wpl,y,unperf Plastic section modulus for an unperforated section (axis parallel to flanges)
x Rectangular coordinate
x’ Rectangular coordinate (rotated)
xx Axis along the member
y Rectangular coordinate
y’ Rectangular coordinate (rotated)
yy Axis of the cross-section (parallel to flanges, normally the major axis)
z Rectangular coordinate
zz Axis of the cross-section (parallel to webs)
xxv
Standard Greek
α Aspect ratio
γ Partial safety factor
γ Shear strain
γM0 Material partial factor
γMBen1 Material partial factor for bending of perforated RHS
γMBen2 Material partial factor for bending of perforated RHS
γMShe1 Material partial factor for shear of perforated RHS
γMShe2 Material partial factor for shear of perforated RHS
γMTor1 Material partial factor for torsion of perforated RHS
γMTor2 Material partial factor for torsion of perforated RHS
γxy Shear strain in the y-direction on a plane perpendicular to the x-axis
γyz Shear strain in the z-direction on a plane perpendicular to the y-axis
γzx Shear strain in the x-direction on a plane perpendicular to the z-axis
ε Engineering (normal) strain
ε1 Maximum principal strain
ε2 Minimum principal strain
εA Normal strain in the A direction
εB Normal strain in the B direction
εC Normal strain in the C direction
εT True (normal) strain
εu Total elongation at maximum stress
εxx Normal strain in the x-direction on a plane perpendicular to the x-axis
εy1 Total normal strain at yield
εy2 Total normal strain at onset of strain hardening
εyy Normal strain in the y-direction on a plane perpendicular to the y-axis
εzz Normal strain in the z-direction on a plane perpendicular to the z-axis
θ Rotation of a Finite Element node or twist of a beam in torsion
θA Inclination of the A direction
θB Inclination of the B direction
θC Inclination of the C direction
θel Rotation at yield
κ Curvature
xxvi
κe Curvature at elastic moment

κp Curvature at plastic moment (assuming elastic flexural rigidity)
λESHF Elliptical strain hardening factor
λflange Flange slenderness
λplateau Plateau ratio
λstrength Strength ratio
λweb Web slenderness
ν Poisson’s ratio
π Pi (3.14159265359)
ρtorsion Stiffness reduction factor for torsion
σ Direct stress
σ Engineering stress
σ1 Maximum principal stress
σ2 Minimum principal stress
σxx Direct stress in the x-direction on a plane perpendicular to the x-axis
σx’x’ Direct stress in the x’-direction on a plane perpendicular to the x’-axis
σyy Direct stress in the y-direction on a plane perpendicular to the y-axis
σy’y’ Direct stress in the y’-direction on a plane perpendicular to the y’-axis
σzz Direct stress in the z-direction on a plane perpendicular to the z-axis
σMises,Max The maximum von Mises stress
σT True stress
τ Shear stress
τi Shear stress at the internal surface
τMax Maximum shear stress
τo Shear stress at the external surface
τxy Shear stress in the y-direction on a plane perpendicular to the x-axis
τx’y’ Shear stress in the y’-direction on a plane perpendicular to the x’-axis
τxz Shear stress in the z-direction on a plane perpendicular to the x-axis
τy Shear stress at yield
τyx Shear stress in the x-direction on a plane perpendicular to the y-axis
τyz Shear stress in the z-direction on a plane perpendicular to the y-axis
τzx Shear stress in the x-direction on a plane perpendicular to the z-axis
τzy Shear stress in the y-direction on a plane perpendicular to the z-axis
φ Angle of principal strain
xxvii
Φ Hole size ratio

Φcr Critical hole size ratio for bending
ψbending Stress concentration factor for bending
ψshear Stress concentration factor for shear
ψtorsion Stress concentration factor for torsion
xxviii
List of abbreviations and trademarks
ASB Asymmetric Slimflor Beam

CF210 ComFlor 210 mm deep profiled steel deck
ESHF Elliptical Strain Hardening Factor
FE Finite Element
FEA Finite Element Analysis
RHS Rectangular Hollow Section
RHSFB Rectangular Hollow Slimflor beam
SD225 225 mm deep profiled steel deck for Slimdek
UK United Kingdom
ABAQUS is a registered trademark of Hibbit, Karlsson and Sorrensen Inc.

ANSYS is a registered trademark of Swanson Analysis Systems Inc.
NASTRAN is a registered trademark of the National Aeronautics and Space
Administration Federal Agency.
ComFlor is a registered trademark of Precision Metal Forming Ltd.
Slimdek and Slimflor are registered trademarks of British Steel plc (now known as
Corus).
UNIX is a registered trademark of the American Telephone & Telegraph Company.
VAX and VMS are registered trademarks of Digital Equipment Corporation.
Rectangular Hollow Sections with circular web openings 1-1
1 Introduction and background
1.1 Introduction
This Thesis describes an investigation into the fundamental behaviour of
Rectangular Hollow Sections (RHS), with circular web cut-outs, under the actions of
bending, shear and torsion. Full and small-scale laboratory tests were conducted,
and the results compared with Finite Element (FE) models and other theoretical
predictions, in order to develop the basis for design recommendations. The
application in mind initially was that of RHS edge beams with web openings for
simple services in steel framed buildings of slim floor construction, but, since the
research is fundamental in nature, the results have a more general applicability.
1.2 Background
1.2.1 An outline of the slim floor method of construction
Slim floor is a method of construction for multi-storey steel-framed buildings
in which the structural depth of each floor is minimised by incorporating the steel
floor beams within the depth of the concrete floor slab. Conventional steel-frame
construction places the floor slab above the (downstand) beam such that the
structural floor depth is the sum of the slab depth and the depth of the beam. The
slim floor approach permits a more efficient use of vertical space, which can result
D.J.Ridley-Ellis / School of Civil Engineering / University of Nottingham / Ph.D. thesis / October 2000
in a significant reduction in building height and a corresponding reduction in
cladding costs. Since the slim floor beams are integrated into the floor structure
there are no downstand beams and the floor can be prepared with either a flat soffit,
or a false ceiling with no obstructions for service runs.
Floor slabs may comprise of pre-cast concrete units or in-situ concrete
placed on profiled metal decking. The use of a composite deck can reduce the
weight of the floor and allows the accommodation of minor services in the voids
between the ribs of the deck.
Like many innovations, the technique was not entirely new and forms similar
to slim floor (Figure 1-2) can be seen in a number of nineteenth century buildings.
One such example is Salt’s Mill (a textile mill built by Lockwood and Mawson in
the early 1850s) at Saltaire near Bradford.
The modern slim floor technique was originally developed in Scandinavia
during the 1980s where it has markedly increased the market share for steel framed
multi-storey buildings (Lawson and Mullett (1993)). The original Scandinavian
beams were termed ‘hat’ or ‘top hat’ beams because of their shape, but slim floor
beams have since developed into a number of alternative configurations. Figure 1-3
is a schematic illustration of several typical internal beams.
Method (a) is the classic hat shape and consists of four plates welded to form
a box. This form has been used previously in the UK. Method (b) uses tee-sections.
This and similar forms are common in Norway and Finland. Method (c) is marketed
by ConstrucThor plc and is a patented system used in the UK and Scandinavia.
Method (d) makes use of a Universal Column (UC) section. The floor units are pre-
cast planks and are supported by the bottom flange. Method (e) is British Steel’s
Slimflor and uses a UC section welded to a bottom plate. This system involves
more fabrication than method (d), but allows easier construction and the use of
composite deep deck slabs (principally CF210). Method (f) is British Steel’s
Slimdek system, incorporating a special asymmetric rolled I-beam, and is described
in detail in the following Section.
(a) Conventional construction (b) Slim floor construction
Floor slab
Floor beam
Figure 1-1: Slim floor and conventional construction
Iron beam Brick arch
Iron column
Figure 1-2: Nineteenth century ‘slim floor’ construction
(a) (b) (c)
(d) (e) (f)
Figure 1-3: Examples of internal slim floor beams
1.2.2 Slimdek
British Steel markets a proprietary slim floor system under the registered
trademark Slimdek. Slimdek is a development of the Slimflor beam with new
elements specifically designed for slim floor construction. The complete Slimdek
system consists of the ‘Asymmetric Slimflor Beam’ (ASB) used in conjunction with
a composite slab of in-situ concrete placed on a profiled metal deck (Figure 1-4).
The asymmetric beam can be up to 25% lighter than conventional Slimflor
beams (Lawson et al 1997), and will permit larger openings (up to 160 mm
diameter) for services in the web. An embossed diamond pattern rolled on to the
beam during manufacture is said to provide composite action with the floor slab
without the need for welded shear connectors (Lawson et al 1997). ASB sections
are designed for use only with deep deck composite slabs, and an improved deck
profile (SD225) was developed to be part of the Slimdek system. The SD225 deck
has a greater load carrying capacity and a longer span range than the old CF210
system. It also has an improved ceiling fixing and service integration capability.
Slab reinforcement In-situ concrete
Asymmetric Slimflor Beam

(ASB)
Rib reinforcement
Profiled metal deck

Service run
Figure 1-4: The Slimdek internal beam (ASB)
1.2.3 Slim floor edge beams
Beams on the perimeter of a building are loaded, chiefly, on one side only.
The ASB and similar open section beams are generally inefficient when subjected to
out of balance (combined flexure and torsion) loads. If used inappropriately, they
may develop high degrees of rotational twist, causing damage to the building
finishes and cladding. For this reason, the Steel Construction Institute recommends
the use of an RHS based edge beam (Mullett 1997). In addition to the advantages
afforded by the superior torsional rigidity, preliminary studies (Mullett et al 1995)
have shown that there is sufficient weight saving to compensate for the higher steel
basis cost. RHS edge beams also offer architectural advantages over downstand
beam alternatives (Mullett 1997).
The Rectangular Hollow Slimflor beam (RHSFB) consists of a hot finished
RHS welded to a 15 mm thick steel flange plate (Figure 1-5). The plate projects on
one side to support the floor slab, and usually projects a small amount on the other
to permit welding of the section without the need for turning. RHS used in edge
beam applications are normally between 200 and 300 mm deep, with wall
thicknesses in the range: 6.3 mm to 16 mm (Mullett 1997). Beams are usually of
grade S355J2H steel (BS EN 10210-1:1994), but grade S275J2H can be more
economical in situations when design is governed by serviceability
The RHSFB may be designed to act compositely or non-compositely with
the floor slab depending on the level of structural performance required (Figure 1-5).
Composite action requires the use of shear studs tied to the reinforcement mesh in
the in-situ concrete. Composite edge beam construction requires an RHS of
sufficient wall thickness (8 mm) to prevent burn-through when attaching the shear
studs, but torsional loads on the beam are reduced as a result of the composite
action. Design advice is available from the Steel Construction Institute (Mullett
1997). Figure 1-6 shows the main elements of slim floor construction using deep
decking and an RHS edge beam.
One of the main features of the Slimdek system is the improved service
integration capability made possible by the new internal beam and deck profile. The
ability to incorporate basic building services in the structural zone capitalises on the
principal advantage of the slim floor method by introducing a further economy of
vertical space. At the perimeter of the building, such services may be required to
pass through the edge beam to the exterior of the building or to pass within the void
of the edge beam itself (Figure 1-7). In both instances, openings are required in the
webs of the RHS forming the edge beam.
Although the flexural behaviour of I-beams with web openings has been well
researched and procedures exist for design (e.g. Annex N of Eurocode 3), there has
been little study of the effects of holes in the webs of RHS. The primary aim of this
project was to study, experimentally and analytically, the behaviour of RHS with
web openings, and to develop the basis for a set of rules for the design of perforated
RHS edge beams (presented in Chapter 7). However, since the research is
fundamental in nature, the findings are also applicable to other situations where it
may be advantageous to cut holes in the webs of load bearing tubular members of
rectangular cross-section.
(a) Composite (b) Non-composite
RHS slim floor edge beam End diaphragm

Concrete slab Shear stud
Profiled metal deck Transverse reinforcement
Slab reinforcement mesh Cold formed angle section
Figure 1-5: Composite and non-composite edge beam construction
Figure based on illustrations in reference Mullett 1997
Column
In-situ concrete Slab reinforcement
RHS edge beam
Profiled metal deck
Service run
Tie beam
Rib reinforcement
Figure 1-6: Slim floor edge beam construction
Web opening
Profiled metal deck (SD225) RHS edge beam Service run
Figure 1-7: Slim floor components
1.3 Methodology
As stated above, the primary aim of this project was to study the behaviour
of RHS with web openings, and to develop the basis for a set of design rules for
perforated RHS edge beams. To this end, the fundamental structural behaviour of
RHS with web openings was examined by an investigation that combined analytical
study with both small and full-scale testing. Representative laboratory tests were
conducted to provide sufficient data for FE models to be calibrated. The FE models
were then used to study the behaviour in greater detail, over a wider range of
parameters.
Laboratory testing compared the performance of perforated RHS with that of
unperforated control specimens. Three types of full-scale test were used to
investigate each of three actions: bending (Chapter 3), shear (Chapter 4) and torsion
(Chapter 5). The full-scale tests were conducted on sizes of RHS similar to those
used in slim floor construction (Section 1.2.3). A preliminary study, consisting of
small-scale torsion testing (Chapter 6), was used to aid the design of the full-scale
tests, and the results were also used to augment the calibration of the FE models.
Once the FE models were shown to predict the experimental response to a
sufficiently accurate and reliable degree, they were used in a parametric
investigation to study the importance of variables such as hole diameter and web
slenderness. The FE models were used to obtain information that would have been
impractical to obtain from laboratory testing alone, and the results, combined with
the experimental observations, were used to develop the basis for the design rules
(Chapter 7).
1.4 Scope of study and dimensions of cross-section
The scope of the research is limited to tubes of constant wall thickness, with
circular holes (also referred to as ‘perforations’, ‘web openings’ and ‘cut-outs’)
placed at mid-depth in the webs. Two cases of perforated section are considered,
depending on whether or not both webs contain holes (Figure 1-8). In the case of
the doubly perforated RHS, the two holes are of equal diameter, and are situated
symmetrically with one in each web.
Standard section sizes for hot-finished RHS are specified by BS EN 10210-
2:1997. The three primary dimensions: depth (h), width (b) and wall thickness (t),
are defined by BS EN 10210-2:1997 as shown in Figure 1-9. Web and flange
slenderness (λweb and λflange) are defined by Equation 1-1 and Equation 1-2. An
additional dimensionless parameter, the hole size ratio (Φ), is used here to indicate
the size of the perforation (Equation 1-3 and Figure 1-10).
λ web = (h − 3t ) t
Equation 1-1
λ flange = (b − 3t ) t
Equation 1-2
Φ = 2r (h − 3t )
Equation 1-3
BS EN 10210-2:1997 prescribes a standard nominal corner profile, which
relates the internal and external radii to the section thickness. Although most
producers roll their own corner profiles (within the permitted tolerances), the section
properties quoted in technical documents are calculated from the nominal corner
profile. Section properties quoted in this Thesis are based on measured corner
profiles, as are FE models of the laboratory tests.
The formulae in Annex A of BS EN 10210-2:1997 for the calculation of the
geometrical properties of RHS are also given in Appendix C of this Thesis.
(a) Unperforated (b) One hole (c) Two holes
Figure 1-8: Perforations
Width, b
z
Internal radius, ri Thickness, t
height, h
y y
External radius, ro
Figure 1-9: Dimensions (cross-section and corner profile)
Width, b
z
h/2
Hole radius, r
Thickness, t
Height, h
x x y y
Hole
h/2
diameter, D
Figure 1-10: Dimensions (web opening)
1.5 The contents of this Thesis
The Thesis began with an introduction to the project and the methodology
employed. The slim floor method of construction was briefly described with
particular emphasis towards edge beams, design and the British Steel Slimdek
system (of which, edge beams based on RHS are an integrated element).
The following Chapter contains a review of the existing knowledge and
theory relevant to this research. Discussion is limited to factors that are of greatest
relevance to the main body of the thesis and those which are most likely to be
unfamiliar to the reader. Consequently, the theories of torsion are discussed in
particular detail, while a sound understanding of the theories of flexure is presumed.
The experimental results and complementary FE work are described in four
chapters, that are grouped according to the test series to which they belong. In each
case, the Chapter begins with a description of the experimental procedure, before
discussing the results and the theoretical and FE predictions. The four laboratory
test series are: full-scale four-point bending (bending), full-scale three-point bending
(shear), full-scale torsion and small-scale torsion.
The penultimate and largest Chapter contains the principal conclusions of the
research in the form of design recommendations. The recommendations were
developed following the calibrated FE parametric study, which is also described.
The justifications for the recommendations are outlined with reference to the
analytical and experimental data.
The final Chapter summarises the conclusions that were drawn at each stage
of the project and ends with a brief critical analysis of the project as a whole.
2 Relevant knowledge and theory
2.1 Introduction
The aim of this project, as stated in the previous chapter, was to study the
behaviour of Rectangular Hollow Sections (RHS) with web openings, and to
develop the basis for a set of design rules for perforated RHS beams in flexure and
torsion. To this end, the fundamental structural behaviour of RHS with web
openings was examined by an investigation that combined analytical study with
laboratory testing. This Chapter is a review of the existing knowledge and theory
pertinent to this research. Discussion is limited to factors that are of greatest
relevance to the main body of the Thesis and those which are most likely to be
unfamiliar to the reader.
Section 2.2 discusses the stress-strain behaviour of hot finished steel as
measured by tensile testing. The features of the stress-strain curve are described
with reference to the representation of material non-linearity in analytical models.
The material model used in the course of this research is explained and the various
parameters that are used to describe the properties of the material are defined.
In this study, the analytical modelling was performed using a commercial
Finite Element (FE) program and the underlying principles of the FE method are
outlined in Section 2.3. The FE method has certain inherent imperfections, and
predictions that are both reliable and accurate can only be obtained if the method is
used correctly. The Section contains a discussion of what is normally regarded as
good practice, and concludes by summarising the rationale behind the various
decisions that were made regarding the analytical component of the research.
The topic of strain measurement is discussed in Section 2.4 with particular
emphasis on electrical resistance strain gauges of the type used in the experimental
investigation. The strengths and limitations of the technique are outlined with
reference to the validation of theoretical and analytical predictions.
The theories of torsion are introduced in Section 2.5. Discussion focuses on
torsion of closed sections and describes the development of torsion theory from
classical thin wall theory to the sophisticated thick wall theory used in design today.
The differences between the various approaches are quantified with reference to
experimental torsion studies conducted by other investigators.
Issues relating specifically to the design of beams in flexure and torsion are
discussed in Chapter 7, which presents the findings of the parametric FE study in the
form of recommendations for the design of RHS with web openings.
2.2 The stress-strain behaviour of hot finished steel
2.2.1 Introduction
This Section describes a mathematical approximation, suitable for modelling
the stress-strain behaviour of hot finished steel, for use in FE analysis. It describes
how a material can be modelled using data obtained from tensile tests, and proposes
a quantitative measure of the similarity between the experimental data and the
approximation. The approximation is compared to standard measures of material
properties as defined by BS EN 10002-1:1990.
2.2.2 The stress-strain behaviour of hot finished steel and BS EN 10002-1
Figure 2-1 shows a typical stress-strain curve for hot finished steel, when
tested at ambient temperature in accordance with BS EN 10002-1:1990. The figure
shows four distinct regions of behaviour:
1) The elastic region: Stress is proportional to strain and all deformation is elastic
and recoverable.
2) The yield plateau (also known as ‘Lüder’s extension’): At yield the stress
reaches a local peak (the upper yield strength), before decreasing rapidly to
fluctuate above a lower value (the lower yield strength). This region has a lower
average stiffness (which may be negative) than the elastic region (Figure 2-2).
Occasionally an apparent reduction in strain will follow the upper yield peak (as
in Figure 2-2). This will occur if the location of first yield lies outside the gauge
length of the extensometer and is a consequence of the upper and lower yield
behaviour of the steel.
3) Strain hardening: Engineering stress increases with increasing strain to reach a
maximum value (the ultimate tensile strength). The stiffness at any value of
strain within the strain hardening region decreases with increasing strain and is
zero at the maximum value of stress.
4) Necking and fracture: Engineering stress falls with increasing strain as the
material thins and eventually fractures.
BS EN 10002-1:1990 provides definitions of upper yield strength (ReH)
lower yield strength (ReL) and tensile strength (Rm) (Figure 2-1):
1) Upper yield strength: ‘The value of [engineering] stress at the moment when the
first decrease in force is observed.’
2) Lower yield strength: ‘The lowest value of [engineering] stress during plastic
yielding, ignoring any transient effects.’ This value may correspond to different
locations in the yield plateau depending on the shape of the plateau.
3) Tensile strength: ‘The [engineering] stress corresponding to the maximum
force.’ The maximum force is defined as ‘the greater force which the test piece
withstands during the test once the yield point has been passed.’
Engineering stress σ
Rm
ReH
ReL
Initial transient effect
εu
Elastic region Strain hardening Engineering strain ε
Yield plateau Necking and fracture
Figure 2-1: The stress-strain behaviour of steel from hot finished RHS
500
Upper yield
450
Engineering stress (N/mm )
2
Onset of strain
hardening
400
350
Yield plateau
300
RHS 200x100x8 S275J2H bar 1 coupon A4

250
0 1 2 3 4 5 6 7 8
Engineering strain (%)
Figure 2-2: The yield plateau
2.2.3 A mathematical model of stress-strain behaviour
The author proposes that a stress-strain relationship of the type discussed in
Section 2.2.2 can be represented by the mathematical approximation shown in
Figure 2-3. The approximation uses simple geometric shapes to model the
experimental curve, and is therefore a way of expressing, in a simple form, the
behaviour as measured by a tensile test (it is not model of material behaviour per se).
The elastic region can be represented by a straight line up to a value of yield
stress (fy1). This is consistent with the assumption that hot finished steel is a
perfectly elastic material in this range. The modulus of elasticity or Young’s
modulus (E) is defined as the ratio of yield stress (fy1) and yield strain (εy1) as given
in Equation 2-1.
f y1
E=
ε y1
Equation 2-1
The initial upper yield peak and the yield plateau are the most variable of all
the aspects of the stress-strain behaviour. These parameters are sensitive to the
nature and quantity of impurities in the metal, and the shape, temperature, loading
rate and stress history of the coupon (Nádai 1931). The upper and lower yield
strengths and the oscillation of the plateau can vary significantly between different
tests of the same batch of material.
The average stress-strain relationship for the whole yield plateau can be
approximated by a second straight line to a second value of yield stress (fy2) and
yield strain (εy2). The values of the first and second yield stress (fy1 and fy2) do not
necessarily correspond to the values of upper and lower yield (ReH and ReL) as
defined by BS EN 10002-1:1990. The definition of lower yield can vary depending
on the shape of the yield plateau and upper yield is highly variable, and is not
appropriate for use in material models for many applications.
The first yield stress (fy1) is the stress at which the elastic line meets the
plateau line and is a simplification of true yield behaviour that removes the
variability of transient effects (the upper yield spike and the subsequent downward
spike). The second yield point (fy2 , εy2) marks the end of the yield plateau and the
onset of strain hardening. This point is often well defined in experimental stress-
strain curves.
The strain hardening behaviour can be approximated by a segment of an
ellipse up to the tensile strength (fu) and total strain at maximum stress (εu). The
strain hardening segment of the ellipse will always terminate at the apex, but may
begin at different locations depending on the material. The location of the ellipse is
defined by its centre (s , t) and the size of its semi axes (a and b). The strain at
maximum stress (εu) defines the x-coordinate of the strain hardening ellipse (s) as
stated in Equation 2-2.
s = εu
Equation 2-2
The final part of the stress-strain curve is difficult to model since the coupon
necks and fractures over a small portion of the gauge length. Engineering stress
may be assumed to remain at a constant value (that of the tensile strength) for
increasing strain provided this assumption is valid in respect to the end use of the
material model.
fu
fy2 b
fy1
fy2 - t (s,t)
εy1 εy2 εu Engineering strain ε
Figure 2-3: The mathematical model
2.2.4 Applying the mathematical approximation to experimental data
2.2.4.1 Applying the mathematical approximation to a single coupon result
The mathematical approximation described in Section 2.2.3 can be applied to
a single coupon result by inspection if sufficient fixed points (such as yield stress,
Young’s modulus and tensile strength) are obtained by established methods. If a
more scientific approach is required, the approximation can be applied as detailed
below.
The function defining the approximated stress-strain relationship can be
defined by four equations. The elastic region is defined by Equation 2-3. The yield
plateau is defined by Equation 2-4. The strain hardening function is defined by
Equation 2-5 and the final portion of the curve can be defined by Equation 2-6 if
appropriate.
f y1
σ= ε 0 ≤ ε < ε y1
ε y1
Equation 2-3
 f y 2 − f y1 
σ = f y1 +  (ε − ε y1 ) ε y1 ≤ ε < ε y 2
ε − ε 
 y2 y1 
Equation 2-4
(ε − ε u ) 2
σ = ( fu − t) 1 − +t ε y2 ≤ ε < ε u
a2
Equation 2-5
σ = fu εu ≤ ε
Equation 2-6
The semi axes of the ellipse (a and b) are determined by the onset of strain
hardening, as defined by the point (fy2 , εy2). Equation 2-7 and Equation 2-8 express
a and b explicitly.
(ε y2 − ε u ) ( fu − t)
2 2
a=
( f u − t )2 − ( f y 2 − t )2
Equation 2-7
b = fu − t
Equation 2-8
The tensile strength (fu) can be determined directly from the coupon results
and will correspond to the tensile strength as defined by BS EN 10002-1:1990. The
six remaining variables can be determined by minimising the difference between the
coupon data and the approximation. It may be possible to determine the total strain
at maximum stress (εu) in advance, if the precise value is apparent from the
experimental data.
A possible measure of the suitability of the fit is introduced in the following
section. The calculation is computationally intensive, but can be performed easily
on a computer. The approximation should be visually compared to the original data
to confirm the appropriateness of the fit.
The variables are (as defined earlier):
fy1 Yield stress
εy1 Total strain at yield
fy2 Stress at onset of strain hardening
εy2 Total strain at onset of strain hardening
t y-coordinate of the centre of the strain hardening ellipse
These six variables together with the tensile strength (fu) are sufficient to
define the complete stress-strain relationship (with the exception of necking and
fracture). They may be used to determine the following quantities, which provide a
more immediate understanding of the stress-strain behaviour:
fy1 Yield stress
E Young’s modulus (Equation 2-1)
Eplateau Stiffness of the yield plateau (Equation 2-9)
λplateau Plateau ratio (Equation 2-10)
λESHF Elliptical strain hardening factor (Equation 2-11)
and either
fu Tensile strength
or
λstrength Strength ratio (Equation 2-12)
The plateau ratio (λplateau) is a measure of the length of the yield plateau
compared to the length of the elastic region. The elliptical strain hardening factor
(λESHF) is a measure of the initial gradient of the strain hardening effect. A value
close to zero indicates a marked onset of strain hardening while a larger value
indicates a less well defined boundary.
f y 2 − f y1
E plateau =
ε y 2 − ε y1
Equation 2-9
ε y1 − ε y 2
λ plateau =
ε y1
Equation 2-10
f y2 − t
λ ESHF =
fu − t
Equation 2-11
fu
λ strength =
f y1
Equation 2-12
2.2.4.2 Measuring the suitability of the fit
In order to determine the parameters that define the best approximation to an
experimental coupon result, it is necessary to define what is meant by ‘best’. If the
experimental data consists of a series of stress-strain coordinates then it is possible,
using the equations in Section 2.2.4.1, to sum up the differences between the
experimental stresses, and the stresses calculated at the corresponding strains using
the approximation. The sum of these differences can be minimised to find a best fit.
This method can be adversely biased by the distribution of data points and by
transient effects such as a sharp upper yield spike. A more stable solution, which is
not influenced by such aspects to the same degree, is explained below.
The difference between two functions, f(x) and g(x), can be quantified in
terms of the sum of the areas between them in xy space as indicated by the shaded
region in Figure 2-4. In the case of a stress-strain curve, the area below the curve
represents the energy required to strain a unit volume of material. The area between
two curves is the difference between them in terms of strain energy. This area, or
strain energy, measurement of fit is not biased by the distribution of data points (e.g.
between A and B in Figure 2-4) or, if the functions are expressed by a sufficient
number of points, by transient effects (e.g. point C), since they do not represent
significant energy features.
If the two functions can be evaluated at the same values of x, then the error
area can be calculated by applying the trapezium rule to the modulus of the
difference between the functions. The total area difference has the dimensions
[Mass/LengthTime2] and can be non-dimensionalised by expressing it as a percentage of
the area below the experimental curve.
In this Thesis, the quantitative measure of the suitability of fit of the
mathematical approximation to the experimental curve is given the symbol K. The
suitability of fit is the sum of the areas between the mathematical approximation and
the experimental curve (calculated to the strain at maximum stress) expressed as a
percentage of the area below the experimental curve (to the same strain).
When the best fit solution is found by minimising the energy error (K) the
difference between the stress predicted by the mathematical approximation and that
observed in the laboratory is greatest in the yield plateau region as shown by
example in Figure 2-5, Figure 2-6 and Figure 2-7.
y y = f(x)
y = g(x)
A B
Figure 2-4: Evaluating the difference between two functions using area
Yield plateau Strain hardening Maximum stress

20
15
10
Error (N/mm )
5
2
-5
RHS 250x250x10 bar 2 coupon LOE2
-10
(A negative error is a prediction of stress which is
-15
lower than the experimental value)
-20
0 2 4 6 8 10 12 14 16
Strain (%)
Figure 2-5: The difference between the approximation and the experimental data
500
Upper yield
450
2
Onset of strain
hardening
400
350
Yield plateau
300
RHS 200x100x8 S275J2H bar 1 coupon A4
Best fit approximation
250
0 1 2 3 4 5 6 7 8
Figure 2-6: The mathematical approximation and the yield plateau
600
Mathematical
model
500
400
Stress (N/mm )
Tensile
2
test result
300
200
100 Tensile test result

RHS 250x250x10 bar 2 coupon LOE2 Mathematical model
0
0 5 10 15 20 25 30
Strain (%)
Figure 2-7: The mathematical approximation and the experimental data
2.2.4.3 Applying the mathematical model to a group of coupon results
A best fit for a series of coupon results can be obtained by averaging the seven
parameters of the best fit for each individual stress-strain curve (as defined earlier):
fy1 Yield stress
E Young’s modulus
Eplateau Stiffness of the yield plateau
λplateau Plateau ratio
λESHF Elliptical strain hardening factor
and either one of:
fu Tensile strength
λstrength Strength ratio
A quantitative measure of the suitability of the fit can be obtained in the
same way as for the individual results by averaging the sums of the areas between
all the experimental curves and the averaged approximation. There may be other
material parameters that give a lower energy error (K) but they may not be realistic
values.
2.2.5 Multi-linear approximations
For some applications, the stress-strain behaviour of a material must be
expressed as a series of straight lines. It is possible to divide the strain hardening
ellipse into linear regions. One such scheme results in total of six lines by splitting
the ellipse into three straight lines at 5%, 20% and 50% of the distance between the
strain at the onset of strain hardening and the strain at maximum force (Figure 2-8).
A multi-linear approximation can be converted from engineering stress and
strain (ε and σ) to true stress and strain (εT and σT) using Equation 2-13 and
Equation 2-14.
ε T = ln(1 + ε )
Equation 2-13
σ T = σ (1 + ε )
Equation 2-14
fu
fy2
fy1
5%
20%
50%
100%
εy1 εy2 εu Engineering strain ε
Figure 2-8: A multi-linear model
2.3 Finite Element analysis
2.3.1 Introduction
This Section is a brief introduction to the FE method that summarises the
various decisions that were made regarding the analytical component of the
research. Comprehensive discussions of topics relating to the FE technique can be
found in such works as Cook et al (1988), Zienkiewicz et al (1994) and Rockey et al
(1986).
2.3.2 The Finite Element approach
Most numerical techniques in continuum mechanics are based on the
principle that it is possible to describe the behaviour of a complex body, with a
reasonable degree of accuracy, by dividing the body into a large number of small,
but not infinitesimal, parts. Predictions about the body as a whole can be made by
describing how each part acts individually, and by deriving the relationships that
link the parts together. The techniques can be classified broadly as belonging to one
of three categories (discussion here is limited to the Finite Element method):
1) Finite Difference
2) Boundary Element
3) Finite Element (FE)
In the FE approach, the body is described by a number of small units known
as ‘elements’. Each element is described separately by a set of mathematical
equations, and is linked to neighbouring elements by relationships considering
equilibrium of forces and continuity of the body. The FE method is well suited to
many engineering problems since it offers versatility in the choice of the mesh and is
able to cope with complex geometry.
The FE procedure requires manipulation of large sparsely populated
matrices. Consequently, the major part of the development of the theory coincided
with the introduction of digital computers. Much of the initial work was conducted
within the aerospace industry, but the technique gained widespread recognition after
it was shown to have a sound mathematical foundation in the mid 1960s.
FE models were initially constructed on an ad hoc basis, but large general-
purpose FE computer programs were beginning to emerge by the late 1960s.
Programs such as ANSYS and NASTRAN included several different types of
elements for performing different types of problem. The development of pre-
processors (for data input) and post processors (for evaluation of results) coincided
with the boom in computer graphics technology in the early 1980s.
Modern FE packages offer a wide variety of different element types for use
in different types of problem. Each type of element can vary in complexity
according to parameters such as the number of nodes, the number of integration
points, and the active degrees of freedom. Much of the skill involved in FE analysis,
using a standard package, lies in the choice of element and the formation of the
mesh. The following Section discusses good practice in the use of FE analysis.
2.3.3 Good practice
FE solutions are not exact and the predictions, at best, are only an
approximation to the true situation. Generally, there are three main sources of error
in an FE solution:
1) Modelling errors: Errors due to geometry not being accurately modelled or
boundary conditions not being properly applied
2) Mesh errors: Errors due to ‘bad mesh’ (including poor element formulation)
3) Numerical errors: Rounding errors in computations, errors in numerical
integration techniques, and errors due to an ill-conditioned solution matrix
These errors can be minimised by good practice, but the solutions should,
whenever possible, be compared with data obtained by other methods (such as
laboratory experimentation). It is generally agreed among experienced users of FE
analysis that the following general guidelines constitute, at least in part, good
practice:
1) Choose the most appropriate element for the task
2) Use a ‘good mesh’
3) Use a fine mesh in regions where stress is of particular interest
4) Start with a relatively course mesh and refine in stages until there is little change
in the solution
5) Check there are no significant discontinuities in the stress values between
elements
6) Check that the externally applied forces are consistent with the reactions at the
restrained nodes
7) Prevent rigid body motion unless it is required as part of the problem
8) Before attempting a new analysis, use a ‘benchmark’ to test the accuracy of the
FE package
Precisely what constitutes a ‘good mesh’ will depend on the program being used.
General pointers are:
1) Use the most appropriate element type
2) Ensure a good fit to the boundary
3) Use higher mesh densities in regions where the loading, internal stresses, or
geometry change rapidly
4) Ensure a gradual change in size between neighbouring elements
5) Ensure the correct connectivity between neighbouring elements
6) Ensure the elements have suitable aspect ratios
FE programs use a displacement-based formulation that gives an upper
bound solution to the true stiffness matrix. Displacements and stresses are therefore
under-estimated, and the error in a correctly formulated FE problem will always be
negative. As the density of the mesh increases, the FE solution will approach the
exact solution from below. In this research, the sensitivity of the solutions to mesh
types and densities was carefully studied to make sure that the FE predictions were
close to the exact analytical solution. Ultimately, the complexity and computational
overhead of the problem must be balanced against the level of precision required.
2.3.4 Stress accuracy
In FE solutions, the internal stresses are calculated from the nodal
displacements by considering each element as an individual entity. These stresses
are not continuous across element boundaries and may not necessarily satisfy
equilibrium between elements. Most commercial FE programs calculate the stress at
a node in one of three ways:
1) Stress at a node is an average taking into consideration all elements connected to
the node
2) Stresses at the nodes are calculated by a combination of extrapolation and
interpolation between the Gaussian integration points
3) Stresses at the nodes are calculated by a combination of extrapolation and
interpolation between points identified as being appropriate for the element type
Increasing the number of integration points in an element increases the
accuracy of the stress calculation, but adds to the complexity of the problem being
solved. For some problems (particularly non-linear problems), it is advisable to use
fewer integration points (‘reduced integration elements’) rather than more (‘full
integration elements’) because a better displacement solution can be obtained. Since
the FE method always underestimates the stresses, the loss in accuracy associated
with reduced integration is compensated by an improvement in the modelling of
overall behaviour. However, a numerical instability problem known as
‘hourglassing’ may occur if reduced integration elements are used unwisely.
2.3.5 Non-linear problems
There are four categories of non-linear problem:
1) Material non-linearity: The stress-strain relationship is non-linear, but the strain-
displacement relationship remains linear (eg elasto-plastic material behaviour in
a metal)
2) Geometric non-linearity: The strain-displacement relationship is non-linear, but
the stress-strain relationship remains linear (eg elastic buckling)
3) Combined geometric and material non-linearity: The stress-strain relationship
and the strain-displacement relationship are both non-linear (eg plastic collapse)
4) Boundary non-linearity: Deformations and stresses are not linearly dependent
on the applied loads (eg contact problems)
Most modern commercial FE packages arrive at a final solution in a non-
linear problem by a combination of incremental and iterative procedures.
Incremental procedures treat non-linearity in a ‘piecewise linear’ manner by
dividing the total load applied into a number of smaller steps (or ‘increments’). An
iterative procedure applies the full load or deformation in the first step and calculates
the correction to be applied in the next (the ‘unbalanced’ portion). The iteration
proceeds until the unbalanced forces and deformations are within acceptable limits.
By using a combination of increments and iterations, a more reliable and
accurate solution can be obtained. This usually occurs at the expense of higher
computation time and greater computational effort. Incremental and iterative
procedures both require numerical search techniques. The exact method by which
FE solvers arrive at the solution to a non-linear problem is one of the biggest
differences between modern commercial FE packages. Certain solution techniques
are more suited to certain types of problems and particular FE programs will be
better at some types of problem than others, even when the same types of elements
are used.
2.3.6 Shell elements
Shell elements are used to model structures where the thickness is
significantly smaller than the other dimensions, and the stresses in the direction of
the thickness are negligible. Shells normally fall into one of two categories
depending on whether or not they model transverse shear deformation. Shells that
model shear flexibility are known as ‘thick shells’ and those that do not are known
as ‘thin shells’.
In addition to the different shear characteristics of shells, there are also
variations in the calculation of membrane strains. In geometrically non-linear
analysis (Section 2.3.5), certain types of shell are able to model the changes in
thickness that result from extreme deformation. Shells in which thickness changes
are neglected are known as ‘small strain shells’ and are not recommended for use in
situations where strains exceed 5%.
A further distinction between different shell types lies in the number of
nodes and integration points, both of which lie on the shell’s reference surface.
Shells with mid-side nodes have quadratic shape functions are able to model curved
surfaces more accurately than shells without mid-side nodes, which have linear
shape functions. Shells are able to model variations in stress through the thickness
by including additional calculation points known as ‘section points’. Section points
exist at each integration point and are positioned at various levels in the shell
thickness.
2.3.7 The ABAQUS Finite Element analysis program
The FE models used in this research were created and solved using the
general purpose ABAQUS/Standard program marketed by Hibbitt, Karlsson &
Sorensen Inc. Early models were solved using version 5.5.1 running on a
VAX/VMS system, while later models were solved using version 5.8.1 running on a
newer and better-equipped UNIX system. In the context of this research, the two
versions of the code were functionally identical and gave identical results when
presented with the same problem.
The FE models included full material and geometric non-linearity, which
were both necessary for accurate modelling of the large displacements and high
strains present in the laboratory tests. Material non-linearity was provided by a
classical metal plasticity model of the type recommended for time-independent
stress analysis at relatively low temperatures. In ABAQUS, the isotropic metal
plasticity model uses the standard von Mises yield surface and is defined by giving
the uniaxial yield stress as a function of the uniaxial equivalent plastic strain.
The elastic region is defined by the Young’s modulus (N/mm2) and
Poisson’s ratio in the card following the keyword, *ELASTIC. The post-yield
behaviour is approximated by a series of straight lines defined by a set of cards
following the keyword, *PLASTIC. The first figure in each card is the true plastic
strain and the second figure is the true stress (N/mm2). The program assumes a
constant response outside the range defined by the input data.
ABAQUS provides a number of different shell elements, but only one shell
type combines the ability to model finite membrane strain, and transverse shear
deformation. The shell is of the reduced integration type with four nodes known by
the designation ‘S4R’. Since four noded shells are formulated by a linear shape
function, a large number were required to accurately model the curved surfaces of
circular openings and corner radii.
2.4 Experimental strain and stress measurement
2.4.1 Introduction
This Section summarises the equilibrium based stress relationships, and
displacement based strain relationships, that are used commonly to interpret
experimental measurements of surface strain using electrical resistance gauges.
Detailed discussion of various strain measurement techniques, and the theories
behind them, can be found in reference Dally and Riley (1991). Derivations of the
fundamental elastic theories of stress and strain can be found in reference
Timoshenko and Goodier (1970).
The following analyses are based on the assumption that the magnitudes of
the strains are small enough to assume that, over a small distance, planes remain
plane during deformation (the products and squares of the strain gradients may be
neglected). This assumption is valid for the working range of the basic types of
electrical resistance strain gauges. For such strains, the magnitudes of true and
engineering strain (Equation 2-13) are sufficiently similar for them to be treated as
being equal, as are the values of true and engineering stress (Equation 2-14).
The following equations for stress (Section 2.4.2) are based on consideration
of equilibrium and are valid for strains outside the elastic range of the material.
Similarly, the equations for strain (Section 2.4.3) are based on consideration of the
geometrical limitations on displacement, and are also valid for strains outside the
elastic range. However, the expressions that relate the components of stress to the
components of strain are only valid in the elastic range and for isotropic and
homogeneous materials.
2.4.2 The components of stress
The stresses acting on the surfaces of a small cubic element of a body, the
faces of which are normal to an arbitrarily chosen system of rectangular coordinates
(x, y and z), can be resolved into Cartesian stress components as shown in Figure
2-9. Vector components of direct, or normal, stress (σ) act normal to the six faces of
the cube and components of shear stress (τ) act in the planes of the six faces, in the
directions of the axes.
The sign convention is such that, if the outer normal defining the cube face is
in the direction of increasing x, y or z, then the associated direct and shear stresses
are also in the direction of positive x, y, z. If the direction of the outer normal is
negative, then the stresses are in the direction of decreasing x, y, z. In Figure 2-9
and the following discussion, the first subscript refers to the outer normal that
defines the face upon which the stress component acts, and the second subscript
indicates the direction. For normal stresses, positive magnitudes indicate tension
and negative magnitudes indicate compression.
If the cube is small enough to neglect any spatial variation in the stress
system, consideration of equilibrium indicates that only six of the components are
necessary to describe the three-dimensional system of stress at a point. For
translational equilibrium to be maintained, the three components acting on each face
must be equal in magnitude and opposite in direction to the corresponding
components on the opposite face. Similarly, the shear stresses on adjacent faces
must be consistent with rotational equilibrium (Equation 2-15).
τ xy = τ yx , τ xz = τ zx , τ yz = τ zy
Equation 2-15
For the most part, strain gauges are limited to measuring strains on a free
surface, where the two-dimensional state of stress that exists, can be expressed in
terms of the three non-zero Cartesian vector components of stress (σxx, σyy and τxy).
The magnitudes and directions of these components depend on the orientation of the
coordinate system. The expressions for the components, for a rotation (θ) of the
coordinate system (Figure 2-10) are given in Equation 2-16, Equation 2-17 and
Equation 2-18. These relationships, that govern the transformation of one
coordinate system to another, can be represented graphically by a Mohr’s circle of
stress (Figure 2-11).
σ xx + σ yy σ xx − σ yy
σ x'x' = + cos (2θ ) + τ xy sin (θ )
2 2
Equation 2-16
σ yy + σ xx σ yy − σ xx
σ y'y' = + cos (2θ ) − τ xy sin (θ )
2 2
Equation 2-17
σ yy − σ xx
τ x' y' = sin (2θ ) + τ xy cos(2θ )
2
Equation 2-18
For a two-dimensional state of stress, there exist two mutually perpendicular
principal stresses (σ1 and σ2), given by Equation 2-19 and Equation 2-20, and a
single value of maximum shear stress (τMax), given by Equation 2-21. An additional
coordinate independent value of stress, the von Mises stress, is extremely useful in
relating a three-dimensional stress system to material properties measured in a one-
dimensional tensile test (Section 2.5.4.2). The von Mises stress, for a two
dimensional system of stress, is given by Equation 2-22.
2
σ xx + σ yy  σ xx − σ yy 
σ1 = +   + τ xy2
2  2 
Equation 2-19
2
σ xx + σ yy  σ xx − σ yy 
σ2 = −   + τ xy2
2  2 
Equation 2-20
σ1 −σ 2
τ Max =
2
Equation 2-21
σ Mises = σ 12 − σ 1 σ 2 + σ 22 = σ xx2 − σ xx σ yy + σ yy2 + 3τ xy2

Equation 2-22
z z
σzz
τzy
τzx
τyz
τxz
σyy
τxy τyx
y σxx y
x x
Figure 2-9: Components of stress acting on a small cubic element
y y’ y
σyy τx’y’
τxy σy’y’
x’
σx’x’
σxx x θ x
Figure 2-10: A rotation of the coordinate system
τ
Shear stress τMax in xy plane
component (σyy , τxy)
σ1
(σy’y’ , τx’y’) 2θ σ
Normal stress
component
(σx’x’ , τx’y’)
σ2
-τMax (σxx , τxy)

Compressive Tensile θ=0
Figure 2-11: Mohr’s circle for stress (two-dimensions)
2.4.3 The components of strain
Each of the six vector components of stress, that describe a three-
dimensional state of stress, is associated with a co-directional vector component of
strain. In the elastic range, the three components of normal strain (εxx, εyy and εzz)
are related to the three direct stresses (σxx, σyy and σzz) as described by generalised
Hooke’s law (Equation 2-23). The components of shear stress (τxy, τxz and τyz) are
proportional to the corresponding components of shear strain (γxy, γxz and γyz) and
shear modulus (G) of the material (Equation 2-24). The Young’s modulus (E) and
the shear modulus (also known as ‘modulus of rigidity’ and ‘modulus of elasticity in
shear’) are two of the five elastic constants used in engineering to describe the
elastic properties of isotropic, homogenous materials. The other three elastic
constants are Lamé’s constant, bulk modulus and Poisson’s ratio (ν). However,
although there are five elastic constants, there are only two independent values. The
shear modulus can be related to the Young’s modulus and Poisson’s ratio
(approximately 0.3 for steel) by the expression given in Equation 2-25.
(σ − ν (σ yy + σ zz ))
ε xx =
xx
E
(σ − ν (σ xx + σ zz ))
ε yy =
yy
E
(σ zz − ν (σ xx + σ zz ))
ε zz =
E
Equation 2-23
1 1 1
γ xy = τ xy , γ xz = τ xz , γ yz = τ yz
G G G
Equation 2-24
E
G=
2(1 + υ )
Equation 2-25
As is the case for stress, the magnitudes of the components of normal and
shear strain are dependent on the choice of coordinate system. For a two-
dimensional state of stress, the expressions for the three components of strain, for a
rotation (θ) of the coordinate system (Figure 2-10) are given in Equation 2-26,
Equation 2-27 and Equation 2-28. The Mohr’s circle for strain (Figure 2-12) is
similar to that for stress, but requires half the shear strain component to be plotted on
the ordinate axis.
ε xx + ε yy ε xx − ε yy γ xy sin (θ )
ε x 'x ' = + cos (2θ ) +
2 2 2
Equation 2-26
ε yy + ε xx ε yy − ε xx γ xy sin (θ )
ε y'y' = + cos (2θ ) −
2 2 2
Equation 2-27
γ x ' y ' = (ε yy − ε xx ) sin (2θ ) + γ xy cos(2θ )

Equation 2-28
For a two-dimensional state of stress, there exist two mutually perpendicular
principal strains (ε1 and ε2), given by Equation 2-29 and Equation 2-30, and a single
value of maximum shear strain (γMax), given by Equation 2-31.
2
ε xx + ε yy  ε xx − ε yy  γ xy
2
ε1 = +   +
2  2  2
Equation 2-29
2
ε xx + ε yy  ε xx − ε yy  γ xy
2
ε2 = −   +
2  2  2
Equation 2-30
γ Max = ε 1 − ε 2
Equation 2-31
γ/2
Half γMax / 2 in xy plane
shear strain (εxx , γxy / 2)
component
ε1
(εy’y’ , γx’y’ / 2)
2θ ε
Normal strain
component
(εy’y’ , γx’y’ / 2)
ε2
(εxx , γxy / 2)
Compressive Tensile -γMax / 2 θ=0
Figure 2-12: Mohr’s circle for strain (two-dimensions)
2.4.4 Strain measurement
2.4.4.1 Introduction
For the most part, strain gauges are limited to measurements on a free
surface, where the two dimensional system of strains can be represented by the three
Cartesian strain components (εxx, εyy and γxy). Most types of strain gauge measure
one component of normal strain over a small region of the body, known as the
‘gauge length’. Since the measurement is made over a finite length, rather than at a
point, the measurement is not exact for non-uniform strain fields. The error depends
on the gauge length and the variation of the strain gradient over this distance. If the
material is perfectly homogenous and isotropic, the measurement error can be
reduced by using a strain gauge with as small a gauge length as possible. However,
metallurgical study of steel reveals that, at a microscopic scale, it is composed of a
number of different types of crystal, and is far from homogeneous and isotropic over
small distances.
The error due to variations in the strain gradient over the gauge length must
be balanced against the need to take measurements over lengths large enough to
average out the variations in strain due to small scale features of the steel. Once the
steel begins to yield, it becomes highly variable at the small end of the macroscopic
scale (Section 2.4.4.3) and it becomes extremely difficult to take accurate
measurements of strain with small gauge length strain gauges. When strain
measurements are to be used in the calibration of FE models, it is wisest to make
measurements of strain where the stress distribution is governed by the property
under examination, but is expected to be nearly uniform over the gauge length.
Measurements of strain after yielding are unlikely to match those predicted by FE
models due to small-scale variations in the steel, but qualitative comparisons and
quantitative comparisons based on the average of several strain measurements are
still possible.
There are many different types of strain gauge, but no single type of gauge is
without shortcomings and the choice of gauge depends largely on the requirements
of the experiment. The electrical resistance strain gauge (Section 2.4.4.2) currently
dominates the strain gauge market, but there are a number of alternative gauges
based on a variety of different principals:
1) Mechanical
2) Optical
3) Piezoresistive
4) Acoustic
5) Pneumatic
6) Electrical (resistance, inductance and capacitance)
2.4.4.2 Electrical resistance strain gauges
Although, there are many different types of strain gauge, there is no single
measurement device that possesses all the characteristics required for optimum
performance. However, the electrical resistance strain gauge (Figure 2-13) has
proved to be extremely successful as it enables precise results of surface strain to be
obtained quickly, using relatively simple and inexpensive equipment. Electrical
resistance strain gauges are also widely used as the principal measurement device in
sensors such as load cells and extensometers.
The electrical resistance strain gauge is based on the principle that the
electrical resistance of metal wire changes with strain. They have been in use for
some 60 years, and have developed into a versatile, robust and easy to use device.
Gauges come in a large number of different configurations, but normally consist of a
length of conductive foil or wire of suitable resistance (frequently 120 Ω or 350 Ω)
wound back and forth over a short gauge length. The most frequently employed
gauges in current use consist of an etched foil grid, bonded to a thin insulating
substrate that can be glued to the surface of the test specimen. The sensitivity of the
gauge, normally specified in the accompanying documentation, is dependent on the
choice of alloy for the gauge wire and the configuration of the gauge.
In order to obtain accurate strain measurements, it is necessary to handle the
gauges carefully and follow the mounting procedures specified in the gauge
documentation. After the adhesive has cured, it is good practice to inspect the
quality of the bond, and replace the gauge if it is found to be damaged, or
improperly attached to the specimen. If possible, it is also advisable to monitor the
performance of the gauge while applying a small number of elastic load cycles to the
specimen, prior to the actual test.
The maximum strain that can be measured with an electrical resistance strain
gauge is known as the ‘elongation limit’ or ‘strain limit’. After this limit, the
linearity of the gauge readout is likely to be reduced by damage to the foil grid, or
substrate material. The types of gauges employed in this project were of the ‘low
strain’ type, which had elongation limits between about 3% and 5%. However, it is
possible for a gauge, or adhesive bond, to become damaged before the elongation
limit quoted by the manufacturer is achieved, and careful inspection of the gauges
throughout a test is necessary to ensure the results are not contaminated by spurious
data. During the experimental component of this research, it was observed that
gauges were particularly prone to early de-bonding when used in situations with
high biaxial shear (such as torsion).
Gauges can be configured to measure strain in single direction (known as
‘linear gauges’), but are also available with stacked or in-plane grids for measuring
strain in more than one direction at a single location (known as a ‘rosette’). The
three-element rectangular rosette (Figure 2-14) is commonly used configuration, as
the three strain measurements (εA, εB and εC) at 0o, 45o and 90o (Figure 2-15) can be
used to calculate the two-dimensional system of strain that exists at the surface with
a simple set of equations based on those quoted in Section 2.4.3. The three
Cartesian strain components (εxx, εyy and γxy) can be calculated using Equation 2-32,
Equation 2-33 and Equation 2-34.
ε xx = ε A
Equation 2-32
ε yy = ε C
Equation 2-33
γ xy = 2ε B − ε A − ε C
Equation 2-34
The principal strains (ε1 and ε2) can be calculated from Equation 2-35 and
Equation 2-36. The angle between the x-axis and the principal strain directions,
known at the principal angle (φ) can be calculated using Equation 2-37.
ε A + εC 1
ε1 = + (ε A − ε C )2 + (2 ε B − ε A − ε C )2
2 2
Equation 2-35
ε A + εC 1
ε2 = − (ε A − ε C )2 + (2 ε B − ε A − ε C )2
2 2
Equation 2-36
1  2ε B − ε A − ε C 
φ = atan 
2  ε A − εC 
Equation 2-37
In the elastic range, the two principal stresses can be calculated using
generalised Hooke’s law (Equation 2-23) using Equation 2-38 and Equation 2-39.
 ε A + εC 1 
σ 1 = E  + (ε A − ε C )2 + (2 ε B − ε A − ε C )2 
 2 (1 + υ ) 2 (1 + υ ) 
Equation 2-38
 ε A + εC 1 
σ 2 = E  − (ε A − ε C )2 + (2 ε B − ε A − ε C )2 
 2 (1 + υ ) 2 (1 + υ ) 
Equation 2-39
In the full-scale experimental component of this research, strain gauges
manufactured by Tokyo Sokki Kenkyujo and by Kyowa Electronic Instruments were
used to measure surface strains and provide data for the calibration of FE models.
Linear strain gauges, with gauge lengths ranging from 5 mm to 20 mm, were used to
measure strains on the flanges, while three-element rosettes, with gauge lengths of 5
mm and 6 mm, were used to measure the more complex strains in the webs. Torque
gauges (two grids used to measure shear strain) with a gauge length of 2 mm were
used to measure the applied torque in the full-scale torsion tests (Chapter 5), and a
small number of 30 mm sandwich gauges (two parallel linear grids between a thick
substrate) were used measure bending strain in the four-point loading tests (Chapter
3).
Grid
Lead wires
Backing film Position

markings
Figure 2-13: An electrical resistance strain gauge
Lead wires
Position
markings
Backing film
Grid
Figure 2-14: A three-element strain gauge rosette
y θA = 0° C
θB = 45° B
θC = 90°
C
B
θB
θC
A
x
Figure 2-15: Measurements of strain on a three-element strain gauge rosette
2.4.4.3 Brittle coatings
A brittle coating is a thin layer of material applied to the surface of the body
being tested. The coating is bonded to the surface and the strains in the body are
transmitted to the coating. The patterns of cracks that form in the coating are
observed in order to make deductions about the distribution of stress at the surface
of the body.
Although brittle coating analysis can be highly advanced, an extremely
effective, if primitive, coating can be used at very little cost and without the need for
special materials and surface treatments. It is common practice, when testing hot
finished steel, to make use of the brittle oxide coating (mill-scale) that forms on the
surface of the steel during manufacture. The oxide coating fails by flaking and
cracking when the base material yields. To improve the visibility of the yield lines,
the steel is often coated with a lime wash (also a primitive form of brittle coating) so
that the dark coloured lines show up against the white background. In this project,
all of the full-scale test specimens were painted with a lime wash consisting of,
approximately, four parts of hydrated lime (by volume) to three parts water.
In this investigation, three, individually distinct, types of yield line were
observed in the lime wash/mill-scale coating. The most common type was the shear
slip line, collectively known as Lüder’s lines, that forms in the direction of
maximum shear stress (Figure 2-16a). Lines due to high compressive strain in the
direction of the maximum principal stress were also observed at 45 degrees to the
Lüder’s lines (Figure 2-16b). A third, but much less common category of lines was
observed that arose as a result of high tensile strain (Figure 2-16c). Lüder’s lines
and compression yield lines can be seen in Figure 2-17 (four-point bending test:
RHS 250x250x10 no holes).
The yield line pattern provides information about the distribution of plastic
stress within the steel at all stages of the test, and can be compared with FE
predictions of the directions of the principal stresses. Example yield line patterns
are presented in Figure 2-18a (torsion test: RHS 150x150x6.3 with two holes of 38.1
mm diameter) and Figure 2-18b (four-point bending test: RHS 250x250x10 with
two holes of 165 mm diameter).
Lüder’s lines represent the intersection of the surface of the body with planes
on which shear stress has produced plastic slip or gliding. In hot finished steel, it is
possible for the material between the Lüder’s lines to remain elastic, because of the
yield plateau behaviour of the steel (Section 2.2.2). It is also possible for material
on either side of a Lüder’s line to have very different systems of strain. Over small
distances, the steel ceases to be homogenous and information gained from the brittle
coating may help to interpret the information gained from strain gauges.
(a) Shear (b) Compression (c) Tension
Figure 2-16: Three types of yield lines in brittle coatings
Figure 2-17: Yield lines in a brittle coating due to bending
(a) (b)
Figure 2-18: Stress distributions around web openings (yield line patterns)
2.5 Torsion theory
2.5.1 Introduction
This Section is a summary of the torsion theory relevant to this research. It
covers both thin and thick-wall treatments of RHS and describes how stiffness and
resistance can be calculated by both methods. The effect of warping restraint is
discussed, with reference to theoretical and FE predictions. Finally, the findings of
previous experimental investigations are summarised.
2.5.2 Basic torsion theory
The first advances in torsion theory were made during the Industrial
Revolution, mainly as a consequence of the introduction of steam power and the
development of iron and steel as structural materials. Coulomb (1787) established
the relationship between the torque and angle of twist of a solid bar of circular
section, and the theory was developed to cover bars of non-circular section by
Navier (1826 and 1864) and Saint-Venant (1855).
For a circular bar, the angle of twist (θ) is dependent on the applied torque
(T), the length of the member (L), the second polar moment of area (J) and the shear
modulus of the material (G), as described by Equation 2-43. The shear stress at the
surface of the member, due to the torque, is proportional to the applied torque and
the radius of the bar (r), and is inversely proportional to the second polar moment of
area, as described by Equation 2-50. For a circular bar, the second polar moment of
area (sometimes called the polar moment of inertia) is given by Equation 2-42.
TL
θ=
GJ
Equation 2-40
τ o = Tr J
Equation 2-41
1
J = πr4
2
Equation 2-42
The torsional properties of a prismatic bar with a non-circular cross-section
are governed by a shape dependent parameter known as the ‘torsional inertia
constant’ (It), which is never greater than the second polar moment of area. If the
member is free to warp at both ends without restraint (Section 2.5.3), then the angle
of twist can be described by Equation 2-43 and the shear stress at the external
surface by Equation 2-44. In Equation 2-44, the relationship between the torque and
shear stress is described by a shorthand parameter known as the ‘torsional modulus
constant’ (Ct). The torsional modulus constant is related to the torsional inertia
constant and analogous to the elastic section modulus in bending theory.
TL
θ=
GI t
Equation 2-43
τ o = T Ct
Equation 2-44
The values of the torsional constants are usually estimated by approximate
mathematical methods based on the ‘stress function’ or ‘membrane analogy’
proposed by Prandtl (1903), and developed by Brendt (1896) and Griffith and
Taylor (1917). In summary, the function describing the variation in the shear stress,
in a member of constant cross-section, is represented by an imaginary homogeneous
membrane supported at the edges, and capped by rigid weightless plates over any
internal voids. The shear stress is represented by the slope of the membrane and the
torque is represented by twice the volume enclosed. The ratio of the pressure under
the membrane to the membrane tension represents twice the product of the angle of
twist and the shear modulus. Early investigators studied the torsion of bars with
non-circular cross-sections by measuring soap films (e.g. Trayer and March 1930),
but the development of computers allowed later investigators to use more accurate
methods, such as Finite Difference (e.g. Marshall 1971) and Finite Element (e.g.
Schultz and Filippou 1998), to evaluate the stress function directly.
For a tubular cross-section, such as an RHS, the shear stress due to torsion is
greatest at the outside surface and decreases through the thickness. Estimates of the
torsional constants fall into two categories depending on how the variation of shear
stress through the thickness is treated, viz ‘thin wall’ and ‘thick wall’.
Thin wall theories are simple to apply, but are only accurate when the
thickness is small because they assume that the shear stress is uniform through the
thickness. The classic thin wall theory is that of Brent, restated here in Equation
2-45 (see also Timoshenko and Goodier 1970). The expressions given in Table 20
of Roark’s Formulas (Young 1989) for an RHS with sharp re-entrant corners are
based on Brent’s formula and are therefore thin wall estimates. Roark’s expressions
are given in Equation 2-46 and Equation 2-47, where the width (b), depth (h), and
thickness (t), are as defined in Figure 1-9.
4 Ah2 t
I t , Brent =
hc
Equation 2-45
Where:
Ah = Area enclosed by mean perimeter
hc = Mean perimeter
2t (b − t ) (h − t )
2 2
I t ,no radii , Roark =
b + h − 2t
Equation 2-46
C t ,no radii = 2t (h − t )(b − t )

Equation 2-47
Thick wall theories are those which attempt to account for the variation in
shear stress through the thickness and, perhaps, at the corners. Abramyan (1951),
developed accurate numerical series solutions for boxes with walls of finite
thickness that were significant improvements upon the foregoing semi-empirical
formulae, but were complicated to implement. A decade later, Byrne and Carré
(1962) were able to apply computer technology to solve the more accurate, but
mathematically cumbersome problem, of box sections with (very tight) internal
corner radii. Byrne and Carré found that localised high stresses are present at the
internal radii, which are potentially higher than those on the external surface. This
observation is also made by Marshall (1971 and 1972) who confidently states that
the measured behaviour of RHS in the laboratory has shown that thin-walled theory
is inappropriate for design. Marshall’s values of torsional inertia constant and
torsional modulus constant, calculated by the application of the Finite Difference
method, are practically identical to those calculated from the formulae1 in Annex A
of BS EN 10210-2:1997 (Figure 2-19 and Figure 2-20). Close inspection of the
formulae in Annex A reveals that they are based on the approximate formulae
proposed by Marshall (1970), with an allowance for the corner radii. Design,
however, does not take into account the raised stresses at internal radii that,
according to Marshall, can be significantly higher (up to 30% for typical radii) than
elsewhere. This may not be important as Marshall also states that the area of
concentrated stress does not exceed 0.6% of the total area of cross-section.
Values of torsional inertia constant calculated by thin wall theory are, for
typical sizes of RHS, between 5% and 1% lower than values calculated by thick wall
theory (Figure 2-21). Values of torsional modulus constant calculated by thin wall
theory are, for typical sizes of RHS between 5% and 25% higher those calculated by
thick wall theory (Figure 2-22).
1
The formulae in BS EN 10210-2:1997 are also given in Appendix C of this Thesis.
Torsional inertia constant, I t
100.5%
BS EN 10210 value / Marshall value
Aspect ratio 1.0

100.0%
Aspect ratio 1.5
99.5% Aspect ratio 2.0
99.0%
98.5%
98.0%
97.5% Points are for selected RHS in the

European standard range (BS EN 10210) (Ref: Marshall 1971)
97.0%
0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035
Thickness / Perimeter, t /(2h +2b )
Figure 2-19: Torsional inertia constant (Marshall and BS EN 10210 calculations)
Torsional modulus constant, C t
102.0%
Aspect ratio 1.0
BS EN 10210 value / Marshall value
101.5% Aspect ratio 1.5

Aspect ratio 2.0
101.0%
100.5%
100.0%
99.5%
Points are for selected RHS in the
European standard range (BS EN 10210) (Ref: Marshall 1971)
99.0%
0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035
Figure 2-20: Torsional modulus constant (Marshall and BS EN 10210 calculations)
Torsional intertia constant, I t
100%
"Thin walled" RHS 150x150x6.3
Roark value / BS EN 10210 value
99%
RHS 200x100x8
98%
97%
RHS 250x250x10
96%
95% Points are for RHS in the European

"Thick walled"
standard range (BS EN 10210)
94%
0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035
Figure 2-21: Torsional inertia constant (thin and thick wall calculations)
Torsional modulus constant, C t
130%
Roark value / BS EN 10210 value
125%
RHS 200x100x8
"Thick walled"
120%
RHS 150x150x6.3
115%
RHS 250x250x10
110%
105%
Points are for RHS in the European
"Thin walled" standard range (BS EN 10210)
100%
0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035
Figure 2-22: Torsional modulus constant (thin and thick wall calculations)
2.5.3 Warping
The numerous symmetries of a circle imply certain constraints on the
torsional deformation of members with solid circular, and concentrically hollow,
cross-sections:
1) All sections must remain circular and rotate around their centres: By symmetry,
all radial lines are equivalent and the deformation along those lines must be
identical for all radii
2) Cross-sections remain plane: Longitudinal displacement would be in violation of
the symmetry of all cross-sections
3) Radial lines must remain straight during deformation: Curvature of radial lines
would be in violation of the symmetry about the sectional plane
4) The angle between any two radial lines remains constant with twist: Any change
in this angle would be in violation of the rotational symmetry about the
longitudinal axis
These assumptions formed the basis for Coulomb’s theory for the torsion of
circular shafts (Coulomb (1789)), and were later applied by Navier (1826), when he
expanded the theory to cover prismatic bars of non-circular cross-section. However,
the above statements cannot be made for a general shape, and Navier’s analysis was
erroneous, implying a solution that is in violation of equilibrium at the boundary of
the cross-section. Saint-Venant modified Navier’s approach to include the
possibility of longitudinal, or warping, distortion of cross-sections (Saint-Venant
(1855) and Navier (1864)).
The warping of cross-sections in torsion becomes important when it is
suppressed, as warping restraint induces longitudinal direct stresses that act in
addition to the ‘Saint-Venant’ shear stresses. The portion of the externally applied
torque carried by these longitudinal stresses is known as the ‘warping torsion’ and
the portion of the externally applied torque resisted by the shear stresses is known as
the ‘Saint-Venant torsion’. In open sections, the warping torsion is a significant
proportion of the externally applied torque, but for tubular sections the warping
torsion is extremely small and usually represents less than 1% of the applied torque.
For a prismatic bar, warping restraint can take one or more of the following forms:
1) Forces are present that resist the longitudinal displacement of the cross-section at
particular locations e.g. the end of a beam, welded to a stiff foundation or joint
2) The applied torque varies along the beam: Adjacent cross-sections subject to
different torsional moments will attempt to warp by different amounts and
provide a degree of restraint to one another e.g. a beam loaded by an eccentric
distributed load
3) The rate twist changes along the beam: Adjacent cross-sections subject to
different rates of twist will attempt to warp by different amounts and provide a
degree of restraint to one another e.g. formation of a torsional buckle in a thin
walled tube
The effect of warping restraint, in members such as RHS, is slight (see
below), and it is usually neglected in practical applications. For this reason, the
majority of the textbooks that discuss torsion of structural sections choose to ignore
warping in tubes, and concentrate instead on open-sections where consideration of
warping is more important. However, a commentary on the theory of warping in
hollow box sections can be found in references: Young (1982), Megson (1990), and
Trahair and Bradford (1998). The theory was originally proposed by von Kármán
and Chien (1946), who studied torsion of aircraft boxes (thin webs between
longitudinal stringers) and was later developed to cover heavy box beams (similar to
RHS) used in the construction of earth moving machinery (Thomas (1969), Smith et
al (1970), Nitzsche and Miller (1976), and Just and Walley (1978)).
In contrast to the warping of members with solid rectangular cross-sections
(Figure 2-23a), the warping distortion in an RHS is linear across the four faces, and
takes the form shown in Figure 2-23b. Consideration of the symmetry of an RHS
implies the following:
1) If warping is not restrained, the distribution is identical at all cross-sections
2) Warping is zero at all points where the cross-section meets a transverse axis of
symmetry
3) The warping distortion at diametrically opposite corners is equal and opposite
4) The warping distortion at the corners of square cross-sections is zero
The longitudinal warping distortion (w) at the corner of an RHS is
proportional to the applied torque and is given by Equation 2-48. When the section
depth (h) equals the width (b), the longitudinal corner warp predicted by Equation
2-48 is zero. Closed section beams that do not warp in torsion are known as
‘Neuber beams’ and a square tube with a constant wall thickness is one example.
The theoretical prediction of the warping distortion is very close to that observed in
FE models (Figure 2-24).
T × (h − b )
w=
8hbtG
Equation 2-48
When warping is restrained in an RHS, the warping shear and normal
stresses reduce to zero within a short distance of the point of restraint
(approximately twice the section depth), where the warping distortion returns to the
unrestrained distribution (Thomas (1969) and Smith et al (1970)). Warping restraint
increases the von Mises stress at the ends (by 8% for RHS 200x100x8), where it also
increases the torsional stiffness slightly, but it has little effect on the performance of
the whole member (unless it is very short). Figure 2-24 shows the torque-rotation
behaviour for three FE models: one for which warping is restrained at both ends, one
for which warping and longitudinal movement (shortening) are restrained at both
ends, and one for which warping is restrained at one end and free at the other. The
Figure shows that elastic stiffness and torsional capacity are not sensitive to the
effects of warping and longitudinal restraint, and that differences are only noticeable
at high values of twist. The decrease in torque at high plastic twist was caused by
the formation of a torsional buckle, which resulted in shortening and additional
warping of the cross-section (Figure 2-26).
+w
-w
(a) Solid rectangular section
-w
(b) Hollow rectangular section

+w
Figure 2-23: Warping of rectangular cross-sections
0.07
RHS 250x250x10 Theory
0.06 RHS 250x250x10 FE
Longitudinal corner warp, w (mm)

0.05
RHS 300x200x12 FE
0.04
RHS 200x100x8 FE
0.03
0.02
0.01
0.00
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Torque / Torque at yield (theory, unperforated)
Figure 2-24: Warping of RHS in torsion (theory and FE)
1.2
Torque / Torque at yield (theory, free to warp)
1.0
0.8
0.6
RHS 200x100x8 unperforated

0.4
Warping prevented & ends restrained longitudinally
Warping prevented & ends unrestrained longitudinally
0.2
Free to warp at one end & both ends unrestrained longitudinally
Shape factor for torsion (theoretical plastic torque)
0.0
0 1 2 3 4 5 6
Rotation / Rotation at yield (theory, free to warp)
Figure 2-25: The effect of end restraint on torsional behaviour of RHS
RHS 200x100x8 unperforated

Free to warp at this end and unrestrained longitudinally
Figure 2-26: Warping at an unrestrained end adjacent to a torsional buckle
2.5.4 Calculation of torsional capacity
2.5.4.1 Expressions for elastic and plastic capacity
As implied by Equation 2-44, the elastic torsional capacity (Tel) is
proportional to the torsional modulus constant and can be calculated from the shear
yield stress (τy) as shown in Equation 2-49. The elastic torsional capacity is the
torque at which the shear stress at the external surface begins to yield. The
calculation does not account for the stress concentrations at the inside surface of
sharp corners.
The plastic torsional capacity (Tpl) can be calculated by considering the flow
of plastic shear around the cross-section (Equation 2-50). The plastic torsional
capacity does not account for strain hardened shear stresses, which may permit a
higher ultimate torque than the plastic torsional capacity.
Tel = τ y C t
Equation 2-49
Aτ y
Tpl = (h(b − t ) + b(h − t ))
2(b + h )
Equation 2-50
2.5.4.2 The shear yield stress and tensile yield stress
Calculations of both the elastic and torsional capacities require knowledge of
the shear yield stress. When direct measurements are required, the shear yield stress
of materials is usually calculated from torsion tests. However, the American Society
for Metals (ASM 1985) lists two methods of direct measurement, in addition to the
torsion test and those reserved for sheet blanking and testing of fasteners:
1) The Miyauchi shear test (Miyauchi 1984)
2) The Marciniak in-plane sheet torsion test (Marciniak 1973)
These shear tests require special testing equipment and are normally reserved
for specialist applications. In practice, the shear properties of structural grade steels
are usually inferred from the results of uniaxial tensile testing and the various
methods by which this can be achieved are discussed by Timoshenko and Goodier
(1970). The two principal methods of calculating the shear yield stress from the
uniaxial tensile yield strength, commonly referred to as ‘von Mises’ and ‘Tresca’
(Figure 2-27), are summarised below.
The ratio between the uniaxial yield stress (σy) and the shear stress at yield
(τy) can be estimated in a number of different ways, by the various material failure
models that describe yield for arbitrary systems of stresses. Current practice favours
the von Mises failure criterion (Mises 1913 and Huber 1904), which is based on
strain energy. In brief, the strain energy is split into two portions: strain energy due
to volume change and strain energy due to distortion. Failure is determined by the
strain energy due to distortion and, for pure shear, gives the familiar ratio expressed
in Equation 2-51.
τy =σy 3 = 0.577 × σ y
Equation 2-51
Experimental work in combined axial load and torsion by investigators such
as Lode (1926 and 1928), Taylor and Quinney (1928), and Hohenemser and Prager
(1932) have shown that the von Mises failure criterion is particularly successful at
modelling the behaviour of structural grade steels. The same ratio was also deduced
(apparently independently) from shear strain measurements by experimental torsion
investigators Stang et al (1937). Timoshenko (1955) summarises the results of
various investigators in the statement that, for steel, the experimentally determined
ratio usually lies between 0.55 and 0.60 (See also Willems et al (1981).
For metals, the main alternative to the von Mises yield criterion is the
Tresca, or ‘principal shear stress theory’, which assumes that the effect of all the
shear stresses within the material is proportional to the maximum shear stress. The
criterion is very easy to implement, but, as Shanley (1957) points out, is based on
assumptions that are least valid for the situation of pure shear. The Tresca criterion
can be applied in a number of different ways, but usually relates the shear yield
stress to the uniaxial yield stress by a factor of one half.
Knowledge of the relationship between the shear yield stress and the uniaxial
yield stress was critical for the full-scale torsion testing component of this research
(Chapter 5). For this reason, an attempt was made to experimentally determine the
ratio for the hot finished steel of the RHS, without recourse to applications of torsion
theory. The method by which this was achieved and the results are described in the
following Section.
1.5
Tresca
Pure shear σ 1 = -σ 2
Von Mises
1.0 Shear coupon results
Principal stress, σ 2
0.5
0.0
-0.5
-1.0
-1.5
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5
Principal stress, σ 1
Figure 2-27: The von Mises and Tresca failure envelopes
2.5.4.3 Experimental measurement of the shear yield stress for hot finished steel
A specially designed tensile test coupon was used to study the failure of the
steel in pure shear (Figure 2-28). Shear failure occurred in the middle of the coupon
where the transverse and longitudinal legs of the cross meet. Strains in this region,
on both sides of the coupon, were measured using strain gauges (Section 2.4). The
thickness of the coupon was sufficient to prevent buckling of the transverse
(compression) legs.
Two coupons were made from the webs of untested RHS 150x150x6.3 grade
S275J2H material from the same bar used in the full-scale torsion tests (Chapter 5).
When tested, the coupons indicated yielding in shear at a stress slightly higher
(approximately 16%) than that predicted by the von Mises yield criterion (Figure
2-27 and Figure 2-29). This measurement is consistent with the findings of the
experimental studies mentioned in the previous Section.
The two coupons measured different values of shear yield stress, which
suggests variability in the material. A torsion test subjects the whole tube to pure
shear and failure would occur first where the steel is weakest. However, it is not
possible to make any assessment about the range of the variability in the tube with
just two measurements.
180
60
10 Dia
10
Shear zone
30
5
5 Dia
120
10
40
Tension leg 5 Milled void

Compression leg
60
Figure 2-28: The shear testing coupon
1.4 Coupon A side 1

Maximum shear strain / Theoretical maximum
1.3 Coupon A side 2

Yield of A
1.2 Coupon B side 1
Yield as predicted by
1.1 the von Mises criterion Coupon B side 2 Yield of B
1.0
shear strain at yield
0.9
0.8 Yield as predicted by
0.7 the Tresca criterion
0.6 The maximum shear strains were calculated
0.5 from the strains measured by strain gauge
0.4 rosettes on each side of the coupons. The
0.3 theoretical maximum shear strain at yield was
calculated from the von Mises criterion,
0.2
universal Hooke's law and the uniaxial stress-
0.1
strain relationship measured by tensile testing.
0.0
0 10 20 30 40 50 60 70
Load (kN)
Figure 2-29: Shear testing coupon results
2.5.5 Experimental studies of the torsional properties of tubes
Many of the experimental studies of the torsional properties of tubes were
conducted in the years leading to and during the Second World War. At the time,
thin tubes of various cross-sections were used to transmit torques to the control
surfaces of aircraft. One such series of tests was commissioned by the National
Advisory Committee for Aeronautics (now known as the National Aeronautics and
Space Administration Federal Agency). Stang et al (1937) tested steel and
aluminium tubes of circular cross-section and concluded that relationships between
torque, twist and shear stress could be adequately described by the basic torsion
theory expressed by Equation 2-40 through Equation 2-42. Later, Moore (Moore
and Paul 1943, and Moore 1947) tested tubes of various materials and concluded
that torsional stiffness, maximum shear stress and torsional capacity could be
adequately described by the formulae quoted by well-known sources such as Roark
(1938) and Timoshenko (1941b). Both Stang et al and Moore observed the ratio of
uniaxial stress to shear stress expressed by Equation 2-51.
Extensive torsion tests on I and H-sections were conducted by Darwish and
Johnson (1965) and the results were compared favourably with soap film predictions
made by Lyse and Johnson (1936). In the discussion following the paper, Roop and
Letherbury described torsion tests of thick walled square tube, but their
understanding of the fundamental theory was flawed (in assuming that the tube
could be analysed by an inappropriate adaptation of the theory for solid sections)
and their conclusions were of limited value.
Tests on hot finished RHS, of sizes similar to those used in this study
(Chapter 5), were conducted by Marshall (1972), who compared the results to the
predictions made by his thick wall torsion theory (Section 2.5.2). Marshall
measured shear stresses on the internal and external surfaces of the RHS in addition
to torque and twist, and concluded that torsional stiffness and shear stress were
adequately predicted by his theoretical expressions. Marshall also measured the
elastic and plastic torsional capacities, but chose to use the Tresca yield criterion
rather than the von Mises criterion (Section 2.5.4.2) in his theoretical calculations.
Marshall’s decision is curious2 as, in the same document, he references the second
edition of Timoshenko’s ‘Strength of materials’ (Timoshenko 1941a and 1941b).
Marshall briefly discusses alternative yield criteria, but does not mention von Mises,
and justifies the use of the Tresca criterion with an obscure reference to an old
technical paper (Marin 1937). The implications of Marshall’s theoretical treatment
of torsional capacity are discussed in Section 5.6.9 in relation to the results of the
full-scale torsion testing component of this research.
2.6 Summary
This Chapter began with a discussion of the stress-strain behaviour of hot-
finished steel (Section 2.2). The various features of the stress-strain curve were
described, and since they have a direct influence on the behaviour of perforated and
unperforated RHS, will be referred to later in the Thesis.
The topics of FE analysis, and experimental strain measurement, were
introduced in Sections 2.3 and 2.4. Data from strain gauge measurements and the
2
Timoshenko (1941b) also references Marin (1937) and states that Tresca is often used in machine
design (as it is both easy to use and conservative), but goes on to discuss von Mises and the good
agreement with experimental observations (Lode (1926) and (1928)).
observation of brittle coatings were used in the calibration of the FE models, which
were later used in the parametric study (Chapter 7). However, as discussed above,
there are limitations in both the strain measurement technique and the FE method,
which meant that comparisons of FE predictions and measurements of strain were
expected to diverge with increasing plastic strain. Nevertheless, quantitative
comparisons of measured and predicted strains in the elastic range, and at low
plastic strain were expected to be reliable and therefore suitable for use in the
calibration of FE models.
The theories of torsion were discussed in some detail in Section 2.5. During
the project, it was observed that some of the experimental results contradicted the
accepted theoretical model. As a consequence, the various aspects of torsion theory
are referred to later, in the Chapter concerned with the full-scale torsion tests
(Chapter 5).
The experimental results and complementary FE work are described in the
following four Chapters, grouped according to the test series to which they belong.
In each case, the Chapter begins with a description of the experimental procedure,
before discussing the results and the theoretical and FE predictions. The discussion
of the experimental investigation begins with the four-point bending tests.
3 Bending (four-point loading)
3.1 Introduction
This Chapter contains a description of the full-scale four-point loading
flexural testing component of the project. The testing apparatus and procedure are
described, and the details of the test specimens are listed. The tests were conducted
to investigate the behaviour of perforated Rectangular Hollow Sections (RHS) in
uniform bending (no externally applied shear). The full-scale three-point loading
flexural testing programme, which was used to investigate the behaviour of
perforated RHS in combined shear and bending, is described in Chapter 4. The full-
scale and small-scale torsion testing programmes are described in Chapters 5 and 6
respectively.
In this Chapter, the experimental results of the four-point bending tests are
summarised, as are the comparisons with conventional beam bending theory and
Finite Element (FE) predictions. The FE models that were calibrated, using the data
from the four-point loading tests, were later used in a parametric study to investigate
the influence of hole diameter, and section size, on the flexural performance of
perforated RHS (Section 7.5).
3.2 A description of the apparatus and testing procedure
3.2.1 The four-point bending test apparatus
The set-up for the four-point bending tests, which were performed in a 2MN
Instron universal testing machine, is shown in Figure 3-1 and Figure 3-2. The
beams were supported on semi-cylindrical bearings (Figure 3-3) with a span of 1750
mm. Nylon pads were used to reduce friction between the specimen and supports.
The load was applied vertically by the hydraulic actuator through a heavily
reinforced spreader beam, and a pair of rocker bearings placed 1000 mm apart on the
compression flange. Each rocker bearing (Figure 3-4) consisted of a roller bearing
that was free to rotate about (and slide along) its longitudinal axis. The cradle,
housing the roller bearing, was free to rotate, within its base, about an axis
perpendicular to (and in-plane with) the axis of the roller.
The rocker bearings, together with a rocker support (Figure 3-3a), allowed
twisting and lozenging of the specimens to occur, without restraint, during the test.
This precaution was necessary because specimens may have been caused to twist or
lozenge under load as a result of any asymmetry of the web openings (intentional or
otherwise), or any distortions present in the specimens as manufactured.
North Load
Spreader beam
Rocker bearings
Web opening RHS 200x100x8 or

RHS 250x250x10
1000 mm
Rocker support Fixed support
1750 mm
375 mm 500 mm 500 mm 375 mm
Figure 3-1: Schematic of the four-point bending test
Figure 3-2: The four-point loading test apparatus
(a) Rocker support (b) Fixed support
Figure 3-3: The supports (rocker and fixed)
Roller bearing
Cradle
Base
Figure 3-4: A dual axis rocker bearing
3.2.2 The four-point bending test specimens
Each four-point loading test specimen comprised of a 1.9 m length of RHS
as shown in Figure 3-1 and Figure 3-5. Steel plates were welded to the flanges, in
the shear span, to prevent failure in combined flexure and shear. Sealing plates and
bearing stiffeners were added to prevent bearing failure at the supports and load
points. The web openings were cold cut to a tolerance of ±0.1 mm in shape and
±0.1 mm in position. Further details of the specimens are given in Section 3.3.
The flange stiffening plate

A mild steel plate welded to the flange to stiffen
and strengthen the shear length of the RHS
Bearing stiffener
A mild steel plate welded to
the web at the end of the tube
to prevent bearing failure
Web opening
A circular hole cut in the web of the tube
165 mm diameter
Rectangular Hollow Section

RHS 200x100x8 or
RHS 250x250x10 Sealing plate
A mild steel plate welded
over the end of the tube to
prevent bearing failure
Figure 3-5: Four-point bending test specimen
3.2.3 Instrumentation
3.2.3.1 Measurement of bending moment
The vertical load was measured using the load cell of the Instron machine
which had been rated as a grade 0.5 device in compression to BS 1610-1:1992
(Appendix F). The bending moment was calculated from the applied vertical load
and the magnitude of the shear span lever arm (375 mm). Preliminary calculations
(including FE analysis) had shown that the change in the shear span lever arm with
increasing curvature was small enough to be neglected.
3.2.3.2 Measurement of curvature
The average curvature of the central span was calculated from the measured
vertical displacement, of the tension flange, at the loading points and in the middle
of the central span. Vertical displacements were measured by linear potentiometers
and the layout of the instrumentation is shown in Figure 3-6.
North

RHS 250x250x10
500 mm 500 mm
200 mm for RHS 250x250x10 & 50 mm for RHS 200x100x8
Bottom flange = Linear potentiometer
Figure 3-6: Measurement of curvature with linear potentiometers
3.2.3.3 Measurement of strain
Several of the bending test specimens were instrumented with electrical
resistance strain gauges of the type shown in Figure 2-13 and Figure 2-14. The
layouts are shown in Figure 3-7, Figure 3-8 and Figure 3-9.
All specimens were painted with a brittle lime wash to increase the visibility
of slip lines in the steel and mill scale (Section 2.4.4.3). Examination of the
formation of the slip line patterns provided information about the onset of yield and
the distribution of plastic shear.
Linear gauge Top flange l = 200 mm for RHS 250x250x10

& 50 mm for RHS 200x100x8
l
Linear gauge
Rosette Webs Locations of rosettes on

webs varied between tests
Rosette
Linear gauge Bottom flange
Linear gauge Linear gauges on flange

only for test REPEAT3
Figure 3-7: Strain gauge pattern (not REPEAT1, REPEAT4, DREB1 and DREB2)
10 mm 10 mm
Rosette Rosette
250 mm
m
m
5
16
Unperforated web Perforated webs
Figure 3-8: Strain gauge pattern (tests DREB1 and DREB2)
Figure 3-9: Strain gauge pattern (test REPEAT4)
3.2.3.4 Other measurements
In the tests of the specimens with single web openings (i.e. asymmetric),
gravity inclinometers (measuring rotation) and additional linear potentiometers were
used in an attempt measure secondary structural responses, such as twisting about
the longitudinal axis and bending about the secondary transverse axis. No
measurable twisting or secondary bending was observed, although the RHS
200x100x8 one hole specimen was observed to lozenge at high curvature.
In every test, linear potentiometers (which can be seen in Figure 3-3) were
used to measure the horizontal movement of the specimen at the supports as the
bottom flange lengthened with increasing curvature. The measurements confirmed
the efficacy of the precautionary procedures aimed at ensuring that significant
additional loads were not placed on the specimens by excessive friction.
3.2.4 Testing procedure
Testing proceeded as outlined below. In all cases, elastic tests were
performed by repeated loading and unloading within the elastic range. One test
(REPEAT1) was conducted in load control rather than displacement control.
1) The specimen was inserted into the loading apparatus and carefully positioned
2) The strain gauges were initialised (zeroed) using the data logger
3) A reading was taken at zero load
4) The spreader beam was lowered onto the specimen, and its weight (7.137 kN)
included in the calculation of the applied load
5) A reading was taken
6) An increment of displacement was applied (either manually, or automatically
using a predefined ramp)
8) Steps 6 and 7 were repeated until the desired curvature was achieved
9) The specimen was unloaded in increments and readings were taken at each
10) The spreader beam was removed and a reading was taken at zero load
11) The specimen was removed from the loading apparatus
3.3 Specimen types, dimensions and properties
3.3.1 Specimens tested
The specimens tested are listed in Table 3-1. Five different hot finished bars
were used in the manufacture of the specimens (two section sizes of grade
S275J2H). The material was provided and manufactured by British Steel (now
Corus).
Unfortunately, an intermittent valve servo fault in the hydraulic control
system of the testing machine caused large errors in the measurement of load for a
number of the tests; particularly those involving the highest loads. The results of
tests TEST1, TEST2, TEST3 and TEST4 were affected by the fault, which became
known when it caused the accidental uncontrolled premature destruction of test
REPEAT2. Additional tests were conducted to replace those affected, and the
results are presented here in lieu of the originals. However, measurements of elastic
stiffness and elastic strain were unaffected by the fault, and those measurements
were made use of in the calibration of the FE models.
Table 3-1: Specimens tested

ID Date tested Section size Material grade Number Hole diameter Hole size
Code [Bar number] of holes (mm) ratio, Φ
TEST4 06/02/1998 RHS 200x100x8 S275J2H [#1] 0 - -
REPEAT1 11/09/1998 RHS 200x100x8 S275J2H [#1] 0 - -
REPEAT2 17/09/1998 RHS 200x100x8 S275J2H [#1] 0 - -
REPEAT5 21/01/1999 RHS 200x100x8 S275J2H [#2] 0 - -
TEST6 11/03/1998 RHS 200x100x8 S275J2H [#2] 1 165.0 0.94
TEST5A 12/02/1998 RHS 200x100x8 S275J2H [#1] 2 165.0 0.94
TEST5B 23/02/1998 RHS 200x100x8 S275J2H [#1] 2 165.0 0.94
TEST1 24/10/1997 RHS 250x250x10 S275J2H [#1] 0 - -
REPEAT3 07/10/1998 RHS 250x250x10 S275J2H [#2] 0 - -
REPEAT4 20/10/1999 RHS 250x250x10 S275J2H [#2] 0 - -
TEST3 25/11/1997 RHS 250x250x10 S275J2H [#1] 1 165.0 0.75
DREB1 24/01/2000 RHS 250x250x10 S275J2H [#3] 1 165.0 0.75
TEST2 20/11/1997 RHS 250x250x10 S275J2H [#1] 2 165.0 0.75
DREB2 19/01/2000 RHS 250x250x10 S275J2H [#3] 2 165.0 0.75
Note 1: Results of tests TEST1, TEST2, TEST3 and TEST4 affected by testing machine fault
Note 2: Test REPEAT2 destroyed by failure of the hydraulic control system of the testing machine
Note 3: The hole size ratio (Φ) is defined by Equation 1-3
Note 4: Test REPEAT1 tested in load control (all others tested in displacement control)
3.3.2 Material properties
The properties of the five bars are summarised in Table 3-2. The results of
the individual tensile tests are given in Appendix D. The definitions of the
parameters used can be found in Section 2.2 and BS EN 10002-1:1990.
Table 3-2: Material models

Section size K fy1 E λplateau Eplateau λESHF fu εu
2 2 2 2
Grade [Bar no] (%) (N/mm ) (kN/mm ) (kN/mm ) (N/mm ) (%)
RHS 200x100x8 1.21 355.4 199.5 13.4 -0.07 0.17 477.4 16.2
S275J2H [#1]
RHS 200x100x8 2.53 335.1 182.0 11.7 0.50 0.21 475.5 16.0
S275J2H [#2]
RHS 250x250x10 1.48 347.7 210.0 14.8 0.52 0.24 463.8 15.6
S275J2H [#1]
RHS 250x250x10 3.16 343.2 197.6 12.9 0.21 0.22 468.0 15.9
S275J2H [#2]
RHS 250x250x10 1.23 364.1 204.0 11.0 0.31 0.27 490.0 15.0
S275J2H [#3]
Note: Refer to Section 2.2
3.3.3 Dimensions and geometrical properties
The measured dimensions of the five bars, used in the manufacture of the test
specimens, are given in Table 3-3. Dimensions are as defined by BS EN 10210-
2:1997 (Figure 1-9). The geometrical properties, obtained using the formulae given
in BS EN 10210-2:1997 (Appendix C), are listed in Table 3-4.
Sectional properties, based on measured material properties and dimensions,
are given in Table 3-5. The theoretical shear capacities of the bars are also listed
(for completeness) and the method by which they were calculated is explained in
Section 4.3.3. For comparison, the nominal section properties are given in Table
3-6.
Table 3-3: Dimensions of cross-section (measured)

Section size Material grade Height Width Thickness Corner radius (mm)
[Bar number] h (mm) b (mm) t (mm) Internal External
RHS 200x100x8 S275J2H [#1] 197.5 100.0 7.79 5.0 12.5
RHS 200x100x8 S275J2H [#2] 198.0 100.0 7.73 5.0 12.5
RHS 250x250x10 S275J2H [#1] 250.0 250.0 9.70 8.0 15.0
RHS 250x250x10 S275J2H [#2] 249.5 249.5 9.74 8.0 15.0
RHS 250x250x10 S275J2H [#3] 250.5 251.0 10.22 8.0 15.0
Table 3-4: Properties of cross-section (from measured dimensions)

Section size Material grade Shear area Second moment Elastic section Plastic section
2
[Bar number] Av (cm ) of area Modulus Modulus
4 4
Iy (cm ) Wel,y (cm ) Wpl,y (cm3)
RHS 200x100x8 S275J2H [#1] 28.38 2074 210 265
RHS 200x100x8 S275J2H [#2] 28.25 2074 209 265
RHS 250x250x10 S275J2H [#1] 45.66 8699 696 817
RHS 250x250x10 S275J2H [#2] 45.74 8675 695 817
RHS 250x250x10 S275J2H [#3] 48.11 9179 733 862
Note: Shear area calculated as in Table 5.16 of Eurocode 3 and Clause 4.2.3d of BS 5950
Table 3-5: Sectional properties (measured dims and material props, theoretical)
Section size Material Curvature Elastic Plastic Shape Shear
grade at yield moment moment factor resistance
[Bar number] κe (km )
-1
Mel,y (kNm) Mpl,y (kNm) Mpl,y / Mel,y Vy (kN)
RHS 200x100x8 S275J2H [#1] 18.02 74.78 94.47 1.26 583
RHS 200x100x8 S275J2H [#2] 18.58 70.32 88.81 1.26 548
RHS 250x250x10 S275J2H [#1] 13.24 242.4 284.6 1.17 918
RHS 250x250x10 S275J2H [#2] 13.91 239.1 280.8 1.17 908
RHS 250x250x10 S275J2H [#3] 14.24 267.3 314.4 1.18 1013
Note: Shear resistance calculated as in Eurocode 3 (Clause 5.5.1) and BS 5950 (Clause 4.2.3)
Table 3-6: Sectional properties (nominal, theoretical)

-1
[Bar number] (km ) (kNm) (kNm) (kN)
RHS 200x100x8 S275J2H [1] 13.10 (73%) 61.42 (82%) 77.54 (82%) 1.26 (100%) 438 (75%)
RHS 200x100x8 S275J2H [2] 13.10 (71%) 61.42 (87%) 77.54 (87%) 1.26 (100%) 438 (80%)
RHS 250x250x10 S275J2H [1] 10.48 (79%) 199.2 (82%) 233.9 (82%) 1.17 (100%) 754 (82%)
RHS 250x250x10 S275J2H [2] 10.48 (75%) 199.2 (83%) 233.9 (83%) 1.17 (100%) 754 (83%)
RHS 250x250x10 S275J2H [3] 10.48 (74%) 199.2 (75%) 233.9 (74%) 1.17 (99%) 754 (74%)
Note 1: The percentages in parentheses denote the nominal value as a percentage of the measured
Note 2: Shear resistance calculated as in Eurocode 3 (Clause 5.5.1) and BS 5950 (Clause 4.2.3)
3.4 The experimental results
Measured elastic stiffnesses and maximum bending moments are listed in
Table 3-7. The full moment-curvature relationships are presented in Figure 3-10
through Figure 3-15. The bending moment and curvature have been non-
dimensionalised against the theoretical moment and curvature at yield for the
corresponding unperforated section. The curvature is the average value over the
uniform bending length (1000 mm).
The theoretical moments and curvatures were calculated using conventional
(Euler-Bernoulli-Parent) beam bending theory and were based on measured
dimensions and material properties. The shape factor is also plotted on the Figures,
and is shown as a horizontal dashed line. The shape factor is the ratio of the elastic
and plastic section moduli and, in this instance, represents the theoretical plastic
capacity of the corresponding unperforated section.
The unperforated sections behaved as predicted by conventional bending
theory, although the RHS 250x250x10 specimens showed limited plastic rotation
capacity due to compression flange buckling. Calculation of the slenderness ratio of
the flange indicated that the RHS 250x250x10 section was at the very limit of the
Eurocode 3 (Clause 5.3) and BS 5950 (Clause 3.5) class 1 definition, and although
the measured rotation capacities were sufficient for plastic design (Section 7.5.1),
plastic buckling at high curvature was not unexpected.
The presence of the web openings was observed to decrease the stability of
the compression elements of the perforated zone, and inelastic buckling in this
region was observed in all the tests of perforated sections. The destabilising effect
of the web openings is discussed further in Sections 3.5 and 7.5.
Measurements of elastic stiffness and maximum bending moment are
compared to the theoretical values of the corresponding unperforated sections in
Table 3-9. The measurements show that the unperforated specimens were able to
achieve capacities slightly higher (up to 10%) than the plastic moment of resistance.
This is likely to be the result of strain hardening at the location of the plastic hinge.
In the case of the RHS 200x100x8 with two holes, the capacity was reduced
by around 30%, but in the case of the RHS 250x250x10, the reduction was around
5%. A qualitative difference of this type was expected, as, in the case of the RHS
250x250x10 beam, the holes were smaller in relation to the section depth, and the
webs contributed a smaller proportion of the section modulus.
No conclusions could be drawn, from the experimental results alone, about
the effect of the web openings on stiffness, as the measurements showed a relatively
large amount of variability, even for unperforated specimens. This is because the
effect of web openings on bending stiffness, discussed further in Sections 3.5 and
7.5.4, is slight.
In the cases where duplicate tests were performed, comparisons of the
complete moment-curvature relationships indicated good repeatability, implying
good reliability of the testing method.
Table 3-7: Experimental results – stiffness and maximum bending moment

ID Section size Material grade Number Hole Stiffness Moment @
2
Code [Bar number] of holes diameter (MNm ) (kNm)
(mm)
TEST4 RHS 200x100x8 S275J2H [#1] 0 - 4.59 - -
REPEAT1 RHS 200x100x8 S275J2H [#1] 0 - 4.18 94.3 Max
TEST6 RHS 200x100x8 S275J2H [#2] 1 165.0 4.22 81.4 Max
TEST5A RHS 200x100x8 S275J2H [#1] 2 165.0 4.33 71.5 Max
TEST5B RHS 200x100x8 S275J2H [#1] 2 165.0 4.57 70.1 Max
TEST1 RHS 250x250x10 S275J2H [#1] 0 - 19.1 - -
REPEAT3 RHS 250x250x10 S275J2H [#2] 0 - 18.2 303 Max
TEST3 RHS 250x250x10 S275J2H [#1] 1 165.0 18.0 - -
DREB1 RHS 250x250x10 S275J2H [#3] 1 165.0 17.8 309 Max
TEST2 RHS 250x250x10 S275J2H [#1] 2 165.0 18.4 - -
Table 3-8: Experimental results and theory– stiffness and max bending moment
ID Section size Material grade Number Hole Stiffness Moment
Code [Bar number] of holes diameter (% of (% of
(mm) theory) theory)
TEST4 RHS 200x100x8 S275J2H [#1] 0 - 110.6% -

REPEAT1 RHS 200x100x8 S275J2H [#1] 0 - 100.7% 99.8%
REPEAT5 RHS 200x100x8 S275J2H [#2] 0 - 114.4% 102.7%
TEST6 RHS 200x100x8 S275J2H [#2] 1 165.0 111.5% 91.7%
TEST5A RHS 200x100x8 S275J2H [#1] 2 165.0 104.3% 75.7%
TEST5B RHS 200x100x8 S275J2H [#1] 2 165.0 110.1% 74.2%
TEST1 RHS 250x250x10 S275J2H [#1] 0 - 104.3% -
REPEAT3 RHS 250x250x10 S275J2H [#2] 0 - 105.9% 107.9%
REPEAT4 RHS 250x250x10 S275J2H [#2] 0 - 101.2% 103.6%
TEST3 RHS 250x250x10 S275J2H [#1] 1 165.0 98.3% -
DREB1 RHS 250x250x10 S275J2H [#3] 1 165.0 94.8% 98.3%
TEST2 RHS 250x250x10 S275J2H [#1] 2 165.0 100.5% -
DREB2 RHS 250x250x10 S275J2H [#3] 2 165.0 95.4% 94.8%
Note: Theoretical quantities are for the corresponding unperforated section
RHS 200x100x8 grade S275J2H no holes
1.4 Plastic moment

Moment / Moment at yield (theory, unperforated)
1.2
1.0
0.8
Repeat1
0.6
Repeat5
0.4
0.2
0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Curvature / Curvature at yield (theory, unperforated)
Figure 3-10: Experimental results RHS 200x100x8 no holes
1.4
Plastic moment
1.2
1.0
0.8
Repeat3
0.6
Repeat4
0.4
0.2
0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
RHS 200x100x8 grade S275J2H one hole
1.4
Plastic moment
1.2
1.0
0.8
0.6 BT6
0.4
0.2
0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Figure 3-12: Experimental results RHS 200x100x8 one hole 165 mm dia
1.4
Plastic moment
1.2
1.0
0.8
0.6 B1
0.4
0.2
0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
RHS 200x100x8 grade S275J2H two holes
1.4
Plastic moment
1.2
1.0
0.8
BT5A
0.6
BT5B
0.4
0.2
0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Figure 3-14: Experimental results RHS 200x100x8 two holes 165 mm dia
1.4
Plastic moment
1.2
1.0
0.8
0.6 B2
0.4
0.2
0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
3.5 A comparison of experimental results and analytical predictions
The laboratory tests were modelled using the ABAQUS FEA program
(versions 5.5.1 and 5.8.1) and the experimental results were compared with the FE
predictions. Models included both material and geometric non-linearity and
contained between 4200 and 5800 four-noded shell elements (S4R). These elements
were chosen to create the models because they allowed changes in thickness at high
strains to be modelled (Section 2.3.6). The FE models were based on the actual
measured dimensions of the test specimens and included the corner radii of the tube.
The stiffening and strengthening plates were also included, and were modelled by
locally increasing the thickness of the elements. The support conditions and
bearings at the loading points were modelled by defining constraint equations to
describe the permitted displacements in the nodal degrees of freedom.
The initial stiffnesses and maximum bending moments predicted by the FE
models are listed in Table 3-9. The differences between the experimental values
(Table 3-7) and the FE predictions are listed in Table 3-10.
The complete moment-curvature relationships are presented in Figure 3-16
through Figure 3-21. The bending moment and curvature have been non-
dimensionalised against the theoretical moment and curvature at yield of the
corresponding unperforated section (Table 3-5). The moment-curvature
relationships predicted by the FE models matched those measured in the laboratory,
and the FE was able to correctly model the negative stiffness in the collapse stage.
The FE models tended to under-predict the maximum bending resistance by
around 10%. Predictions of elastic stiffness were within 14% of those measured
experimentally. However, this difference in elastic stiffness is accounted for, in
part, by the difficulty in obtaining accurate measurements of Young’s modulus from
tensile tests.
The FE models of the unperforated sections failed to achieve the full
theoretical plastic capacity due to the long yield plateau in the material model and
the large curvature required to achieve yield in the entire cross-section. In the
laboratory tests, the plastic deformation was concentrated over a short length (a
plastic hinge) where strain hardened stresses were able to develop. However, in the
FE models, the homogeneity of the material meant that plastic strain occurred
uniformly over the whole bending length of the beam, and strains were too small to
result in strain hardened stresses.
The FE predictions of the elastic stiffness for the unperforated sections were
close to those predicted by bending theory, although, in the case of the RHS
250x250x10 sections, the FE predictions were slightly lower (around 3%). The FE
models of the perforated sections indicated that the reduction in stiffness, due to the
presence of the hole, was small (less than 5% in the case of the RHS 200x100x8 and
less than 1% in the case of the RHS 250x250x10). The effect of the web openings
on the elastic stiffness is discussed further in Section 7.5.4.
As was the case with the laboratory tests, the FE model of the
RHS250x250x10 unperforated section showed limited ductility due to compression
flange buckling. The natural tendency of the compression flange in the FE model
was to buckle outwards, while inward buckles were observed in the laboratory. A
slight imperfection (0.05% of the section width) was introduced into the flange of
the FE model used in the calibration, in order to induce an inward buckle. A
parametric investigation showed that the ultimate resistance of the beam was
insensitive to the size of the imperfection, although the rotation capacity could be
increased (up to 20%) by balancing the imperfection with the natural tendency of the
flange to buckle outwards.
The FE models matched closely the deformed shapes observed in the
laboratory (Figure 3-22 through Figure 3-27). The deformation in the perforated
zone was modelled extremely well, as can be seen in Figure 3-28, which shows a
deformed FE mesh superimposed on a photograph of the deformed test specimen
(RHS 200x100x8 with two holes).
In the perforated sections, the presence of the web openings reduced the
plastic stability of the compression elements in the perforated zone. In the case of
the RHS 200x100x8 sections, plastic collapse of the section occurred through
arching of the cee-shaped compression zone above the hole (Figure 3-23, Figure
3-24 and Figure 3-28). In the case of the RHS 250x250x10 sections, plastic collapse
of the section occurred through buckling of the compression flange because the
perforation(s) reduced the stabilising influence of the webs (Figure 3-26 and Figure
3-27). For the RHS 250x250x10 sections, the buckle in the compression flange was
slightly off-centre, in both the FE models and the laboratory tests (Figure 3-26 and
Figure 3-27).
In the elastic range, the theoretical and FE predictions of surface strain for
unperforated sections were close to those measured in the laboratory. However, the
measurements and predictions diverged with increasing strain (this is explained in
Section 2.6). Figure 3-29 shows a comparison of the longitudinal strain-curvature
gradient for various points around the circumference of an unperforated RHS
250x250x10 specimen (test REPEAT4, Figure 3-9). In this case, the measurements
of the bending strain in the webs were close to those predicted by conventional beam
bending theory, but measurements of bending strain in the flanges were slightly
lower (15%). It is possible that this difference is a result of shear lag, although no
shear lag was observed in the FE models.
In the elastic range, the FE predictions of surface strain for perforated
sections were also close to those measured in the laboratory. As was the case with
the perforated sections, the measurements and predictions diverged with increasing
strain (this is explained in Section 2.6). In this respect, the comparisons of measured
strains and FE predictions, for an RHS 250x250x10 with one hole (test DREB1),
shown in Figure 3-30 and Figure 3-31 are typical. The Figures show the maximum
and minimum principal strains for a point 10 mm above the top of the hole, for the
perforated (Figure 3-31) and unperforated (Figure 3-30) webs (see also Figure 3-8).
The effect of the perforation on the stress distribution in the beam, evident in these
results, is discussed further in Section 7.5.2.
Table 3-9: Finite Element results – stiffness and maximum bending moment
ID Section size Material grade Number Hole Stiffness Moment @
2
Code [Bar number] of holes diameter (MNm ) (kNm)
(mm)
TEST4 RHS 200x100x8 S275J2H [#1] 0 - 4.16 - -
TEST6 RHS 200x100x8 S275J2H [#2] 1 165.0 3.72 74.0 Max
TEST5A RHS 200x100x8 S275J2H [#1] 2 165.0 3.96 63.0 Max
TEST5B RHS 200x100x8 S275J2H [#1] 2 165.0 3.96 63.0 Max
TEST1 RHS 250x250x10 S275J2H [#1] 0 - 17.8 - -
TEST3 RHS 250x250x10 S275J2H [#1] 1 165.0 17.8 - -
TEST2 RHS 250x250x10 S275J2H [#1] 2 165.0 17.8 - -
Table 3-10: Experimental and FE results – stiffness and maximum bending moment
ID Section size Material grade Number Hole Difference Difference
Code [Bar number] of holes diameter stiffness moment
(mm) (%) (%)
TEST4 RHS 200x100x8 S275J2H [#1] 0 - 10.4 -
REPEAT1 RHS 200x100x8 S275J2H [#1] 0 - 0.35 3.00
REPEAT5 RHS 200x100x8 S275J2H [#2] 0 - 14.1 6.02
TEST6 RHS 200x100x8 S275J2H [#2] 1 165.0 13.5 9.87
TEST5A RHS 200x100x8 S275J2H [#1] 2 165.0 9.20 13.5
TEST5B RHS 200x100x8 S275J2H [#1] 2 165.0 15.2 11.3
TEST1 RHS 250x250x10 S275J2H [#1] 0 - 7.42 -
REPEAT3 RHS 250x250x10 S275J2H [#2] 0 - 8.41 10.0
REPEAT4 RHS 250x250x10 S275J2H [#2] 0 - 4.11 5.66
TEST3 RHS 250x250x10 S275J2H [#1] 1 165.0 0.71 -
DREB1 RHS 250x250x10 S275J2H [#3] 1 165.0 -1.34 7.47
TEST2 RHS 250x250x10 S275J2H [#1] 2 165.0 3.62 -
DREB2 RHS 250x250x10 S275J2H [#3] 2 165.0 -0.79 10.2
Note: A negative percentage indicates experimental values lower than Finite Element values
1.4
1.2
1.0
0.8
0.6
0.4
FE (s0calbt)
0.2 Experimental (Repeat1 and Repeat5)
Shape factor
0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Figure 3-16: FE and experimental (tests REPEAT1 and REPEAT5)
1.4
1.2
1.0
0.8
0.6
0.4
FE (l0calct)
0.2 Experimental (Repeat3 and Repeat4)
Shape factor
0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Figure 3-17: FE and experimental (tests REPEAT3 and REPEAT4)
1.4
1.2
1.0
0.8
0.6
0.4
FE (s1calat)
0.2
Experimental (BT6)
Shape factor
0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Figure 3-18: FE and experimental (test TEST6)
1.4
1.2
1.0
0.8
0.6
0.4
FE (fecalb1s)
0.2
Experimental (B1)
Shape factor
0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Figure 3-19: FE and experimental (test DREB1)
1.4
1.2
1.0
0.8
0.6
0.4
FE (s2calbt)
0.2 Experimental (BT5A and BT5B)
Shape factor
0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Figure 3-20: FE and experimental (tests TEST5A and TEST5B)
1.4
1.2
1.0
0.8
0.6
0.4
FE (fecalb2s)
0.2
Experimental (B2)
Shape factor
0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Figure 3-21: FE and experimental (test DREB2)
Figure 3-22: Displaced shape RHS 200x100x8 no holes
Figure 3-23: Displaced shape RHS 200x100x8 one hole
Figure 3-24: Displaced shape RHS 200x100x8 two holes
Figure 3-25: Displaced shape RHS 250x250x10 no holes
Figure 3-26: Displaced shape RHS 250x250x10 one hole
Figure 3-27: Displaced shape RHS 250x250x10 two holes
Figure 3-28: Displaced shape RHS 200x100x8 two holes (superposition)
150
Experimental (Repeat4)
Theory
Longitudinal strain / curvature (µε.km)
100
50
0
0 250 500 750 1000
-50
-100
Top flange East web Bottom flange West web

-150
Distance around perimeter
Figure 3-29: Longitudinal strain for RHS 250x250x10 unperforated (REPEAT4)
(Distance around perimeter in millimetres)
3
250 mm 10 mm Maximum principal strain
Strains are higher in the perforated web

2 than the unperforated web as a direct
165 mm consequence of the presence of the hole.
This is discussed further in Section 7.5.2.
1
Strain (%)
0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
-1
-2 Experimental (DREB1)
FE (fecalb1s) Minimum principal strain

-3
Curvature / curvature at yield (theory, unperforated)
Figure 3-30: Principal strains for RHS 250x250x10 one hole (unperforated web)
3
250 mm 10 mm Maximum principal strain
2
165 mm
1
Strain (%)
0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
-1
-2 Experimental (DREB1)
FE (fecalb1s) Minimum principal strain

-3
Curvature / curvature at yield (theory, unperforated)
Figure 3-31: Principal strains for RHS 250x250x10 one hole (perforated web)
3.6 Concluding remarks
A series of 13 full-scale four-point bending tests were conducted on two
sizes of RHS to assess the effect of web openings upon structural performance in
uniform bending. During the tests, the moment and average curvature were
measured. In 12 of the tests, strains on the external surface of the flanges and webs
were also measured.
The results of the tests showed that the reduction in the bending capacity due
to the web openings was as much as 30%, and that the openings caused no
measurable reduction in elastic stiffness. It was also observed that the presence of
the web openings reduced the plastic stability of the compression elements in the
perforated zone, causing a decrease in the rotation capacity.
The experimental results were compared with FE predictions and a close
agreement was observed in terms of displaced shapes, elastic stiffnesses and strains
in the elastic range. The experimental moment-curvature relationships matched
those predicted by the FE models, although the FE tended to under-predict the
bending resistance by around 10%.
The FE models of the perforated sections indicated that the reduction in
elastic stiffness, due to the presence of the hole, was small. The reduction was less
than 5% in the case of the RHS 200x100x8 and less than 1% in the case of the RHS
250x250x10.
The full-scale three-point loading flexural testing programme, which was
used to investigate the behaviour of perforated RHS in combined shear and bending,
is described in the following Chapter.
4 Shear (three-point loading)
4.1 Introduction
This Chapter contains a description of the full-scale three-point loading
flexural testing component of the project. The testing apparatus and procedure are
described, and the details of the test specimens are listed. The tests were conducted
to investigate the behaviour of perforated Rectangular Hollow Sections (RHS) in
combined bending and shear. The high shear resistance of the RHS prevented the
beams being tested under the action of shear in the absence of significant bending
forces. The full-scale four-point loading flexural testing programme (bending
without shear) is described in Chapter 3, and the full-scale and small-scale torsion
testing programmes are described in Chapters 5 and 6 respectively.
In this Chapter, the experimental results of the three-point loading tests are
summarised, as are the comparisons with conventional beam bending theory and
Finite Element (FE) predictions. The FE models that were calibrated, using the data
from the three-point loading tests, were later used in a parametric study to
investigate the influence of hole diameter, and section size, on the shear resistance of
perforated RHS (Section 7.6).
4.2.1 The three-point bending test apparatus
The set-up for the three-point bending tests, which were performed in the
same 2MN Instron universal testing machine as the four-point loading tests (Chapter
3), is shown in Figure 4-1 and Figure 4-2. The beams were supported on semi-
cylindrical bearings (Figure 3-3) with a span of 1750 mm. Nylon pads were used to
reduce friction between the specimen and supports.
The load was applied vertically by the hydraulic actuator through a single
rocker bearing (Figure 3-4) placed centrally on the compression flange. The web
openings were placed at a distance of 500 mm from the loading point and were
subject to combined shear and bending moment.
The rocker bearings, together with a rocker support (Figure 3-3a), allowed
twisting and lozenging of the specimens to occur, without restraint, during the test.
This precaution was necessary because specimens may have been caused to twist or
lozenge under load as a result of any asymmetry of the web openings (intentional or
otherwise), or any distortions present in the specimens as manufactured.
North Load
Rocker bearing

RHS 250x250x10
500 mm
875 mm 875 mm
1750 mm
Figure 4-1: Schematic of the three-point bending test
Figure 4-2: The three-point bending test apparatus
4.2.2 The three-point bending test specimens
Each three-point bending test specimen comprised of a 1.9 m length of RHS,
as shown in Figure 4-1 and Figure 4-3. Steel plates were welded to the flanges in
the supporting span to ensure failure occurred in the 875 mm span containing the
web opening. In addition, sealing plates and bearing stiffeners were added to
prevent bearing failure at the supports and load points. The web openings were cold
cut to a tolerance of ±0.1 mm in shape and ±0.5 mm in position. Further details of
the specimens are given in Section 4.3. Unperforated beams, which failed in
bending close to the load point, were included in the testing programme to aid in the
calibration of the FE models.
The flange stiffening plate

A mild steel plate welded to the flange to stiffen
and strengthen the shear length of the RHS
RHS 200x100x8 or
RHS 250x250x10
Bearing stiffener
A mild steel plate welded to
the web at the end of the tube
to prevent bearing failure
Web opening
165 mm diameter
Flange bearing stiffener

A mild steel plate welded to the Sealing plate
bottom flange at the end of the A mild steel plate welded
tube to prevent bending failure over the end of the tube to
prevent bearing failure
Figure 4-3: Three-point bending test specimen
4.2.3.1 Measurement of shear force and bending moment
The shear force and bending moment were calculated from the vertical load,
which was measured using the load cell of the Instron machine (rated as a grade 0.5
device in compression to BS 1610-1:1992, see Appendix F).
4.2.3.2 Measurement of deflection
The total central deflection of the beam was calculated from the measured
vertical displacement of the tension flange at the load point and at the supports
(settlement). Vertical deflections of the tension flange were also measured at each
side of the web opening to measure shear deformation. The layout of the linear
potentiometers, used to measure these vertical deflections, is shown in Figure 4-4.
North 500 mm

RHS 250x250x10
90 mm 165 mm 90 mm
200 mm for RHS 250x250x10 & 50 mm for RHS 200x100x8
Bottom flange = Linear potentiometer
Figure 4-4: Measurement of vertical displacement with linear potentiometers
The test specimens were instrumented with electrical resistance strain gauges
of the type shown in Figure 2-13 and Figure 2-14. The layouts are shown in Figure
4-5. The specimens were also painted with a brittle lime wash to increase the
visibility of slip lines in the steel and mill scale (Section 2.4.4.3). Examination of
the formation of the slip line patterns provided information about the onset of yield
and the distribution of plastic shear.
Top flange Linear gauge l = 200 mm for RHS 250x250x10

& 50 mm for RHS 200x100x8
l
m
Linear gauge
m
95
Rosette Webs
45o 200 mm
Rosette
Support end Load end
Linear gauge
Linear gauge
Bottom flange
Figure 4-5: Strain gauge pattern (all tests)
In all the tests, gravity inclinometers (measuring rotation) were used to
measure twisting about the longitudinal axis. The inclinometers (Figure 4-6 and
Figure 4-7) were placed at the ends of the specimens in order to measure the total
twist along the whole length (although most of this twist occurred in the perforated
zone).
In addition, linear potentiometers were used to measure the horizontal
movement of the specimen at the supports as the bottom flange lengthened with
increasing curvature. The measurements confirmed the efficacy of the precautionary
procedures aimed at ensuring that significant additional loads were not placed on the
specimens by excessive friction.
Inclinometer for
measurement of twist Reference inclinometer

RHS 250x250x10
= Inclinometer
Figure 4-6: Inclinometers for measurement of twist
Test specimen
Inclinometer
Rocker support
Figure 4-7: Measurement of twist at the rocker support
Testing proceeded as outlined below. In some cases, elastic tests were
performed by repeated loading and unloading within the elastic range.
1) The specimen was inserted into the loading apparatus and carefully positioned
2) The strain gauges were initialised (zeroed) using the data logger
4) An increment of displacement was applied (either manually, or automatically
using a predefined ramp)
6) Steps 4 and 5 were repeated until the desired central displacement was achieved
7) The specimen was unloaded in increments and readings were taken at each
9) The specimen was removed from the loading apparatus
The specimens tested are listed in Table 4-1. Three different hot finished
bars were used in the manufacture of the specimens (two section sizes of grade
S275J2H). The material was provided and manufactured by British Steel (now
Corus).

Code [Bar number] of holes (mm) ratio, Φ
TESTS1 07/12/1998 RHS 200x100x8 S275J2H [#2] 0 - -
TESTS3 14/12/1998 RHS 200x100x8 S275J2H [#2] 1 165.0 0.94
TESTS2 09/12/1998 RHS 200x100x8 S275J2H [#2] 2 165.0 0.94
TESTS4 16/12/1998 RHS 250x250x10 S275J2H [#2] 0 - -
TESTS6 13/01/1999 RHS 250x250x10 S275J2H [#2] 1 165.0 0.75
TESTS5 19/01/1999 RHS 250x250x10 S275J2H [#1] 2 165.0 0.75
Note: The hole size ratio (Φ) is defined by Equation 1-3
The properties of the three bars are summarised in Table 4-2. The results of
the individual tensile tests are given in Appendix D. The definitions of the
parameters used can be found in Section 2.2 and BS EN 10002-1:1990.

2 2 2 2
RHS 200x100x8 2.53 335.1 182.0 11.7 0.50 0.21 475.5 16.0
S275J2H [#2]
RHS 250x250x10 1.48 347.7 210.0 14.8 0.52 0.24 463.8 15.6
S275J2H [#1]
RHS 250x250x10 3.16 343.2 197.6 12.9 0.21 0.22 468.0 15.9
S275J2H [#2]
The measured dimensions of the three bars, used in the manufacture of the
test specimens, are given in Table 4-3. Dimensions are as defined by BS EN 10210-
in BS EN 10210-2:1997 (Appendix C), are listed in Table 4-4. The shear area was
calculated as specified in Table 5.16 of Eurocode 3 and Clause 4.2.3d of BS 5950.
Sectional properties, based on measured material properties and dimensions,
are given in Table 4-5. The theoretical shear capacity was calculated from the
product of the shear yield stress and the shear area. The shear yield stress was
calculated from the yield stress, as measured from direct tensile testing, and the von
Mises failure criterion (Section 2.5.4.2). This method of calculating the shear
capacity is the same as that used by the Eurocode 3 (Clause 5.5.1) and BS 5950
(Clause 4.2.3) design procedures. For comparison, the nominal section properties
are given in Table 4-6.

RHS 200x100x8 S275J2H [#2] 198.0 100.0 7.73 5.0 12.5
RHS 250x250x10 S275J2H [#1] 250.0 250.0 9.70 8.0 15.0
RHS 250x250x10 S275J2H [#2] 249.5 249.5 9.74 8.0 15.0

Section size Material grade Shear area Second moment Elastic section Plastic section
[Bar number] Av (cm2) of area Modulus Modulus
4 4
Iy (cm ) Wel,y (cm ) Wpl,y (cm3)
RHS 200x100x8 S275J2H [#2] 28.25 2074 209 265
RHS 250x250x10 S275J2H [#1] 45.66 8699 696 817
RHS 250x250x10 S275J2H [#2] 45.74 8675 695 817
Note: Shear area calculated as in Table 5.16 of Eurocode 3 and Clause 4.2.3d of BS 5950
[Bar number] κe (km ) -1
Mel,y (kNm) Mpl,y (kNm) Mpl,y / Mel,y Vy (kN)
RHS 200x100x8 S275J2H [#2] 18.58 70.32 88.81 1.26 548
RHS 250x250x10 S275J2H [#1] 13.24 242.4 284.6 1.17 918
RHS 250x250x10 S275J2H [#2] 13.91 239.1 280.8 1.17 908
Note: Shear resistance calculated as in Eurocode 3 (Clause 5.5.1) and BS 5950 (Clause 4.2.3)

-1
[Bar number] (km ) (kNm) (kNm) (kN)
RHS 200x100x8 S275J2H [#2] 13.10 (71%) 61.42 (87%) 77.54 (87%) 1.26 (100%) 438 (80%)
RHS 250x250x10 S275J2H [#1] 10.48 (79%) 199.2 (82%) 233.9 (82%) 1.17 (100%) 754 (82%)
RHS 250x250x10 S275J2H [#2] 10.48 (75%) 199.2 (83%) 233.9 (83%) 1.17 (100%) 754 (83%)
Note 1: The percentages in parentheses denote the nominal value as a percentage of the measured
Note 2: Shear resistance calculated as in Eurocode 3 (Clause 5.5.1) and BS 5950 (Clause 4.2.3)
Measured elastic stiffnesses and maximum shear forces are listed in Table
4-7. The full shear force-central displacement relationships are presented in Figure
4-8 and Figure 4-9. The shear force has been non-dimensionalised against the
theoretical shear capacity of the corresponding unperforated section (Section 4.3.3).
The central displacement has been non-dimensionalised against the theoretical
central displacement at which the corresponding unperforated section begins to
yield, in bending, at the load point (calculated using conventional beam bending
theory).
The unperforated sections failed in bending rather than shear on the
unstrengthened side of the loading point where the bending moment was largest.
The shear forces corresponding to the theoretical moments required to cause bending
failure at the load point are plotted, in Figure 4-8 and Figure 4-9, as horizontal
dotted lines, which are labelled ‘Bending failure elastic’ and ‘Bending failure
plastic’ corresponding to the theoretical elastic and plastic moment capacities. The
unperforated sections behaved as predicted by beam bending theory, and were able
to resist loads higher than the theoretical plastic moment capacity due to high (strain
hardening) strains at the hinge location. The unperforated RHS250x250x10
specimen reached a maximum load when buckling of the compression flange
occurred.
The sections with web openings failed in shear in the perforated zone,
although significant bending forces were also present at this location. The results of
the tests showed that the reduction in the shear capacity due to the web openings
was as much as 85%, and that the openings caused a significant reduction in the
shear stiffness of the perforated zone. The reductions in the shear resistance of the
RHS 250x250x10 specimens (around 70%) were less than those for the RHS
200x100x8 specimens (around 80%) because the holes were smaller in relation to
the section depth. Asymmetry of the web openings caused twisting and lozenging of
the sections due to unbalanced shear deformation of the perforated and unperforated
webs. Symmetrical sections (unperforated and two holes) did not twist or lozenge.
Displacement profiles, for the RHS 200x100x8 specimens, are shown in
Figure 4-10 through Figure 4-16. The measurements were made using the linear
potentiometers shown in Figure 4-4. In the symmetrical cases (no holes and two
holes), the profile shown is the average of the measured profiles of the two webs. In
the unsymmetrical case (one hole), the profiles of the perforated and unperforated
webs are shown separately. The numbered symbols on the profiles relate to the
numbered symbols on the corresponding shear force-displacement plots. The
Figures illustrate the modes of failure discussed above.
Table 4-7: Experimental results – stiffness and maximum shear force

ID Section size Material grade Number Hole Stiffness Shear force @
Code [Bar number] of holes diameter (kN/mm) (kN)
(mm)
TESTS1 RHS 200x100x8 S275J2H [#2] 0 - 29.7 291 25 mm
TESTS3 RHS 200x100x8 S275J2H [#2] 1 165.0 26.0 224 Max
TESTS4 RHS 250x250x10 S275J2H [#2] 0 - 112.2 803 Max
Note: Unperforated sections failed in bending rather than shear
RHS 200x100x8 grade S275J2H
0.30
Shear force / Shear capacity (unperforated)
0.25
No holes
0.20
Bending failure (plastic)
0.15 Bending failure (yield)
0.10 Two holes
0.05
One hole
0.00
0 1 2 3 4 5 6 7 8 9 10 11 12
Central displacement / Central displacement at yield (theory, unperforated)
Figure 4-8: Experimental results RHS 200x100x8
RHS 250x250x10 grade S275J2H
0.50
No holes
0.45
0.40 One hole
0.35
Two holes
0.25
0.20
0.15
0.10
0.05
0.00
0 1 2 3 4 5 6 7 8 9 10 11 12
Figure 4-9: Experimental results RHS 250x250x10
0.30
7 8
6 10
9
0.25
5
2 4
3
0.20

1
0.10
0.05
Experimental (ST1)
0.00
0 1 2 3 4 5 6 7 8 9 10 11 12
Figure 4-10: Shear force-displacement relationship RHS 200x100x8 no holes
0 1
2 RHS 200x100x8
3 Grade S275J2H
Vertical displacement (mm)
-10
4 No holes
5
-20
6
7
-30
8
9
-40
10
-50
-60
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Distance North - South (mm)
Figure 4-11: Displacement profiles RHS 200x100x8 no holes
RHS 200x100x8 grade S275J2H one hole 165 mm diameter
0.25

0.20
3 4 5 6
2
0.10
1
0.05
Experimental (ST3)
0.00
0 1 2 3 4 5
Figure 4-12: Shear force-displacement relationship RHS 200x100x8 one hole
0
1 RHS 200x100x8
Grade S275J2H
-5
2 One hole
-10 165 mm diameter
3 Perforated web
-15
4
-20 5
-25 6
-30
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Figure 4-13: Displacement profiles RHS 200x100x8 one hole (perforated web)
0
1 RHS 200x100x8
Grade S275J2H
-5
2 One hole
-10 165 mm diameter
3 Unperforated web
-15
4
-20 5
6
-25
-30
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Figure 4-14: Displacement profiles RHS 200x100x8 one hole (unperforated web)
RHS 200x100x8 grade S275J2H two holes 165 mm diameter
0.16
0.14
2 7
3 4 5 6 8
0.12
1 9 10
0.10
0.08
0.06
0.04
0.02
Experimental (ST2)
0.00
0 1 2 3 4 5 6 7 8 9 10 11 12
Figure 4-15: Shear force-displacement relationship RHS 200x100x8 two holes
0 1
2 RHS 200x100x8
Grade S275J2H
-10 3
4 Two holes
-20 5 165 mm diameter
6
7
-30 8
9
-40
10
-50
-60
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Figure 4-16: Displacement profiles RHS 200x100x8 two holes
The experimental tests were modelled using the ABAQUS v5.5.1 FEA
program and the experimental results were compared with the FE predictions.
Models included both material and geometric non-linearity and contained between
6700 and 7600 four-noded shell elements (S4R). These elements were chosen to
create the models because they allowed changes in thickness at high strains to be
modelled (Section 2.3.6). The FE models were based on the actual measured
dimensions of the test specimens and included the corner radii of the tube. The
stiffening and strengthening plates were also included, and were modelled by locally
increasing the thickness of the elements. The support conditions and bearings at the
loading points were modelled by defining constraint equations to describe the
permitted displacements in the nodal degrees of freedom.
The initial stiffnesses and maximum shear forces predicted by the FE models
are listed in Table 4-8. The differences between the experimental values (Table 4-7)
and the FE predictions are listed in Table 4-9.
The full shear force-central displacement relationships are presented in
Figure 4-17 through Figure 4-22. The shear force has been non-dimensionalised
against the theoretical shear capacity of the corresponding unperforated section
(Section 4.3.3). The central displacement has been non-dimensionalised against the
theoretical central displacement at which the corresponding unperforated section
begins to yield, in bending, at the load point (calculated using conventional beam
bending theory). The shear force-displacement relationships predicted by the FE
models closely matched those measured in the laboratory.
In contrast to the four-point bending (Section 3.5), the FE models tended to
over-predict the maximum resistance, but by no more than 5%. Predictions of
elastic stiffness were less accurate, but within 17% of those measured
experimentally. However, this difference in elastic stiffness is accounted for, in
part, by the difficulty in obtaining accurate measurements of Young’s modulus from
tensile tests.
As was the case for the laboratory tests, the FE models of the unperforated
sections behaved as predicted by beam bending theory, and were able to resist loads
higher than the theoretical plastic moment capacity due to high (strain hardening)
strains at the hinge location. The unperforated RHS250x250x10 specimen reached a
maximum load when buckling of the compression flange occurred. However, the
buckle predicted by the FE model was a slightly different shape to that observed
experimentally (see below).
In the case of the RHS 200x100x8 specimen with one hole, the FE model
was able to correctly predict the twisting of the beam with increasing shear
displacement (Figure 4-23). The prediction of twisting for the RHS 250x250x10
specimen with one hole was less successful, although the twist was correctly
predicted for shear deformations in the elastic range (Figure 4-24).
The FE models matched closely the deformed shapes observed in the
laboratory (Figure 4-25 through Figure 4-28). The deformation in the perforated
zone was modelled extremely well, as can be seen in Figure 4-29, which shows a
deformed FE mesh superimposed on a photograph of the deformed test specimen
(RHS 200x100x8 with two holes). However, in the case of the unperforated RHS
250x250x10 (Figure 4-27 and Figure 4-28), the buckle predicted by the FE model
(outwards) was a slightly different shape to that observed experimentally (inwards).
In the elastic range, the theoretical and FE predictions of surface strain were
close to those measured in the laboratory. However, the measurements and
predictions diverged with increasing strain (this is explained in Section 2.6).
Table 4-8: Finite Element results – stiffness and maximum shear force
ID Section size Material grade Number Hole Stiffness Shear force @
Code [Bar number] of holes diameter (kN/mm) (kN)
(mm)
TESTS1 RHS 200x100x8 S275J2H [#2] 0 - 25.5 294 25 mm
TESTS4 RHS 250x250x10 S275J2H [#2] 0 - 99.4 808 Max
Note: Unperforated sections failed in bending rather than shear
Table 4-9: Experimental and FE results – stiffness and maximum shear force
ID Section size Material grade Number Hole Difference Difference
Code [Bar number] of holes diameter stiffness Shear force
(mm) (%) (%)
TESTS1 RHS 200x100x8 S275J2H [#2] 0 - 16.5 -0.95
TESTS3 RHS 200x100x8 S275J2H [#2] 1 165.0 12.9 -4.86
TESTS2 RHS 200x100x8 S275J2H [#2] 2 165.0 16.5 -2.90
TESTS4 RHS 250x250x10 S275J2H [#2] 0 - 12.9 -0.56
TESTS6 RHS 250x250x10 S275J2H [#2] 1 165.0 3.99 -0.10
TESTS5 RHS 250x250x10 S275J2H [#1] 2 165.0 0.04 -3.00
Note 1: A negative percentage indicates experimental values lower than Finite Element values
Note 2: Unperforated sections failed in bending rather than shear
0.30
0.25
0.20
0.10
0.05 Finite Element (s0calsat and s0calsbt)

Experimental (ST1)
0.00
0 1 2 3 4 5 6 7 8 9 10 11 12
Figure 4-17: FE and experimental (test TESTS1)
0.50
0.45
0.40
0.35
0.25
0.20
0.15
0.10
Finite Element (l0calsct)
0.05
Experimental (ST4)
0.00
0 1 2 3 4 5 6 7 8 9 10 11 12
0.30
0.25
0.20
0.10
0.05 Finite Element (s1calsat)

Experimental (ST3)
0.00
0 1 2 3 4 5 6 7 8 9 10 11 12
0.50
0.45
0.40
0.35
0.25
0.20
0.15
0.10
Finite Element (l1calsat)
0.05
Experimental (ST6)
0.00
0 1 2 3 4 5 6 7 8 9 10 11 12
0.30
0.25
0.20
0.10
0.05 Finite Element (s2calsat)

Experimental (ST2)
0.00
0 1 2 3 4 5 6 7 8 9 10 11 12
0.50
0.45
0.40
0.35
0.25
0.20
0.15
0.10
Finite Element (l2calsdt)
0.05
Experimental (ST5)
0.00
0 1 2 3 4 5 6 7 8 9 10 11 12
RHS 200x100x8 grade S275J2H one hole 165 mm diameter (twist)
6
Finite Element (s1calsat)
5
Experimental (ST3)
Twist (degrees)
0
0 1 2 3 4 5 6 7 8
Figure 4-23: FE and experimental (twist of TESTS3)
RHS 250x250x10 grade S275J2H one hole 165 mm diameter (twist)
6
Finite Element (l1calsat)
5
Experimental (ST6)
Twist (degrees)
0
0 1 2 3 4 5 6 7 8
Figure 4-24: FE and experimental (twist of TESTS6)
(a) RHS 200x100x8 one hole (b) RHS 200x100x8 two holes
(c) RHS 250x250x10 one hole (d) RHS 250x250x10 two holes
Figure 4-25: Displaced shapes (Finite Element, perforated)
(a) RHS 200x100x8 one hole (b) RHS 200x100x8 two holes
(c) RHS 250x250x10 one hole (d) RHS 250x250x10 two holes
Figure 4-26: Displaced shapes (experimental, perforated)
RHS 200x100x8 no holes RHS 250x250x10 no holes
Figure 4-27: Displaced shapes (Finite Element, unperforated)
(a) RHS 200x100x8 no holes (b) RHS 250x250x10 no holes (buckle)
(c) RHS 250x250x10 no holes
Figure 4-28: Displaced shapes (Experimental, unperforated)
Figure 4-29: Displaced shape RHS 200x100x8 two holes (superposition)
A series of six full-scale three-point bending tests were conducted on two
combined shear and bending. During the tests, the shear force, bending moment and
displacement were measured. In all six tests, strains on the external surface of the
flanges and webs were also measured.
The results of the tests showed that the reduction in the shear capacity due to
the web openings was as much as 85%, and that the openings caused a significant
reduction in the shear stiffness of the perforated zone.
in the elastic range. The experimental shear force-displacement relationships closely
matched those predicted by the FE models with a maximum difference of only 5%.
The full-scale torsion testing programme, which was used to investigate the
behaviour of perforated RHS in torsion, is described in the following Chapter.
5 Full-scale torsion
5.1 Introduction
This Chapter contains a description of the full-scale torsion testing
component of the project. The testing apparatus and procedure are described, and
the details of the test specimens are listed. The small-scale torsion testing
programme is described in Chapter 6, and the full-scale bending and shear testing
programmes are described in Chapters 3 and 4.
The experimental results are summarised and compared with conventional
torsion theory (Section 2.5) and Finite Element (FE) predictions. Differences
between the experimental observations and theoretical and analytical models were
found, and arguments supporting the validity of the experimental data are discussed.
A possible explanation of the differences is proposed and suitable schemes for
further investigation are suggested.
5.2.1 The torsion test apparatus
The torsion testing rig comprised of a fixed reaction beam and a pivoted
rotating beam, between which a length of Rectangular Hollow Section (RHS) could
be caused to twist about its longitudinal axis (Figure 5-1 and Figure 5-2). The
apparatus was designed to function correctly and safely up to a torque of 100 kNm.
The tee-shaped fixed reaction beam was held in place by two sets of columns
bolted to a structural strong floor. A crosspiece of grade S275J2H RHS 250x250x10
resisted the torque by bearing vertically on the columns and a short shaft consisting
of doubled-up grade S275J2H RHS 200x100x8 transmitted the torque from the
specimen to the crosspiece.
The torsional action was applied through the rotation of a further grade
S275J2H RHS 250x250x10 crosspiece. This beam was supported on the axis of the
rig by a 50 mm diameter mild steel bar and a bearing block. The bearing was held
in place by a pair of columns bolted to the strong floor. Excessive movement of the
rotating beam out of the vertical plane was prevented by two sets of guide columns,
also bolted to the strong floor.
The rotating end of the specimen was connected to the rotating beam through
a 500 mm length of grade S355J2H Circular Hollow Section. This length of circular
shaft was used to measure the torque and was machined internally and externally to
an external diameter of 242.5 mm and a thickness of 10.5 mm, with a tolerance of
±0.05 mm.
The torque was created by a matched pair of double acting hydraulic
cylinders mounted on the guide columns. Each cylinder was operated from a
separate oil source allowing the two loads to be balanced manually during the test.
Control over the unloading of the specimens was achieved by the regulation of the
out-flowing oil through a pair of needle valves.
Twisting end Jack point (down) South Fixed end
Guide columns
Bearing house
700 mm
Calibration shaft Reaction beam

700 mm
Specimen
Connecting plates and bolts
Jack point (up)

250 500 mm 2000 mm 350 250
mm mm mm
Figure 5-1: The torsion testing apparatus (schematic)
Figure 5-2: The torsion testing apparatus
5.2.2 The torsion test specimens
Each torsion test specimen comprised of a two-metre length of RHS as
shown in Figure 5-3. The details of the specimens, which were tested with the plane
of the major axis horizontal, are given in Section 5.3. The web openings were cold
cut to a tolerance of ±0.1 mm in shape and ±0.1 mm in position, and the overall
length of the specimens was controlled to ±0.5 mm. The torsional action from the
loading apparatus was transmitted to the specimen through bolted connections at
both ends.
Connecting plate
A mild steel plate welded to the end of the tube
to allow a bolted connection with the testing rig

RHS 200x100x8 or
RHS 150x150x6.3
Web opening
165 mm diameter for RHS 200x100x8
99 and 38.1 mm diameter for RHS 150x150x6.3
Figure 5-3: Torsion test specimen
5.2.3 Connection of the specimen and rig
Connectivity of the specimens with the testing apparatus was provided by
410 mm square mild steel plates, which were joined to the rig shafts and specimens
by single bevel full penetration butt welds. The plates were machined from a
thickness of 30 mm to 25 mm to insure the mating surfaces were both flat and
perpendicular to the axis of the testing rig. Each plate contained 12 equally spaced
holes of 26 mm diameter, at a radius of 170 mm from the axis of the rig. The
connecting plates on the specimen and the rig were bolted together with M24x150
mm high strength friction grip bolts. A chain-like series of 12 mild steel links
provided additional support to the specimen-side connecting plates (Figure 5-4).
The bolts were manually tightened to a torque of 550 Nm after the free movement
due to the oversized holes was taken up. This torque was not sufficient to generate
the full design bolt pre-load required for ‘friction grip’, but was sufficient to prevent
slippage of the connection. The bolts were tightened in an alternating sequence to
avoid uneven fixity.
Figure 5-4: The connecting plates and bolting arrangement
5.2.4.1 Measurement of torque
Measurement of the torque was achieved by monitoring the shear strain on
the outside surface of the circular shaft (Figure 5-5). Six evenly spaced shear strain
(or torsion) gauges (Section 2.4.4.2) provided an unbiased measure of average shear
strain (Figure 5-6 and Figure 5-7). Measurement of the torque in this way ensured
that the measurement was unaffected by friction and the change in the lever arm
with rotation. The slight difference between the measured shear strains and the
theoretical shear strains in Figure 5-6 and Figure 5-7 is due to the assumptions made
about the Young’s modulus and Poisson’s ratio of the shaft (no direct measurements
were made).
The circular shaft was calibrated, prior to the torsion tests, by comparing the
output from the shear strain gauges with applied torques of known magnitude. The
shaft was calibrated in a vertical position (Figure 5-8), by suspending the rotating
beam from the fixed reaction beam. The torque was applied through the action of
two hydraulic cylinders placed at equal distances from the vertical axis. The
magnitude of the torque was calculated from the forces measured by two load cells,
mounted on spherical seatings between the hydraulic cylinders and the rotating
beam. The fixed reaction beam resisted the torque applied to the calibration shaft by
bearing on two sets of columns that were bolted to the strong floor. Under this
arrangement, the rotating beam was freely suspended and was in contact solely with
the jacks and reaction beam. The applied torque was not diminished by friction, and
the high stiffness of the apparatus minimised the effect of rotation on the torque
calculation. The shaft was calibrated to a torque of 90 kNm and linear behaviour
was confirmed over this range (Figure 5-7).
Figure 5-5: The calibration shaft
1600
1400
1200
Shear strain (microstrain)
Measured shear strains

1000
Theoretical shear strain
800
600
400
Assumed properties of calibration shaft
2
200 Young's Modulus = 210 kN/mm
Poisson's ratio = 0.3
0
0 10 20 30 40 50 60 70 80 90 100
Torque (kNm)
Figure 5-6: Calibration data
1600
Gradient of average measured = 13.76 microstrain / kNm
1400 Gradient of theoretical = 14.55 microstrain / kNm
Difference = 5.7%
1200
Shear strain (microstrain)
Average of measured shear strains

1000
Theoretical shear strain
800
600
400
Assumed properties of calibration shaft
2
200 Young's Modulus = 210 kN/mm
Poisson's ratio = 0.3
0
0 10 20 30 40 50 60 70 80 90 100
Torque (kNm)
Figure 5-7: Calibration data (averaged measurements)
Reaction beam Frame
Calibration shaft
Reaction column
Rotating lever arm Hydraulic jack
Figure 5-8: Calibration of the load-indicating shaft
5.2.4.2 Measurement of rotation
Rotation was measured at each end of the specimen using gravity
inclinometers (Figure 5-9) mounted on the connecting plates. Two additional
inclinometers measured the rotation at the quarterpoints of the specimen. The twist
was calculated by referencing the rotations measured at locations 1, 2 and 3 (Figure
5-10) to the fiduciary rotation at the fixed end, location 4. Measurement of the twist
in this way was independent of the stiffness of the testing apparatus and bolted
connections.
Figure 5-9: Inclinometers for measurement of rotation
= Inclinometer Specimen Connecting plate
No. 3 No. 2
No. 4 No. 1
500 mm 1000 mm 500 mm
Figure 5-10: Inclinometer locations
Several of the torsion specimens were instrumented with electrical resistance
strain gauges of the type shown in Figure 2-14. The layouts are shown in Figure
5-11 and Figure 5-12. The strain gauges provided data for use in the calibration of
the FE models, which were later used in the parametric study (Chapter 7). The
arrangement, shown in Figure 5-12, was used to study the variation in torsional
shear stress along the length of the specimen.
All specimens were painted with a brittle lime wash to increase the visibility
of slip lines in the steel and mill scale (Section 2.4.4.3). Examination of the
formation of the slip line patterns provided information about the onset of yield and
the distribution of plastic shear.
r
Rosette 4 s, Rosette 1
iu
R ad For
45o RHS 200x100x8 r = 95 mm
RHS 150x150x6.3 r = 57 mm
Up
West Rosette 3 Rosette 2 East
South web
r
Rosette 8 s, Rosette 5
iu
R ad
45o
Up
East Rosette 7 Rosette 6
North web
Figure 5-11: Strain gauge pattern (tests TT1-TT6)
Specimen Connecting plate
250 mm 250 mm 250 mm 250 mm 250 mm 250 mm 250 mm

125 mm 125 mm
Rosette Rosette Rosette Rosette Rosette Rosette Rosette Rosette
1 2 3 4 5 6 7 8
Figure 5-12: Strain gauge pattern (test TT7)
The loads applied by the hydraulic jacks were measured by load cells to aid
in the balancing of the vertical forces. The movement of the end of the rotating
beam was measured by a linear potentiometer to assist in the determination of the
loading increments. These devices also provided secondary, but less reliable,
measurements of torque and rotation.
Testing proceeded as outlined below. In some cases, elastic tests were
performed by repeated loading and unloading within the elastic range.
1) The specimen was inserted into the loading apparatus. The bolts were inserted,
but not tightened.
2) The strain gauges were initialised (zeroed) using the data logger.
3) The specimen was rotated by the hydraulic jacks until a small torque was
measured by the load cells. This procedure removed the rotation due to the
oversized holes in the connecting plates. The load was then released.
4) The bolts were tightened in alternating sequence.
5) The strain gauge output was reviewed to verify that there was no load on the
specimen.
6) A reading was taken at zero load.
7) An increment of displacement was applied.
8) A reading was taken.
9) Steps 7 and 8 were repeated until the desired total rotation was achieved.
10) The specimen was unloaded in increments and readings were taken at each.
11) A reading was taken at zero load.
12) The specimen was de-mounted.
The specimens tested are listed in Table 5-1. Four different hot finished bars
were used in the manufacture of the specimens (two section sizes each of two
different grades). The grade S275J2H material was provided and manufactured by
British Steel (now Corus). The grade S355J2H material was provided by British
Steel and manufactured by Vallourec and Mannesmann Tubes, Germany.

code [Bar number] of holes (mm) ratio, Φ
TT4 12/08/1999 RHS 200x100x8 S275J2H [#3] 0 - -
TT6 19/08/1999 RHS 200x100x8 S275J2H [#3] 1 165.0 0.94
TT2 03/08/1999 RHS 200x100x8 S275J2H [#3] 2 165.0 0.94
TT3 09/08/1999 RHS 150x150x6.3 S275J2H [#1] 0 - -
TT5 17/08/1999 RHS 150x150x6.3 S275J2H [#1] 1 99.0 0.76
TT1 27/07/1999 RHS 150x150x6.3 S275J2H [#1] 2 99.0 0.76
TT7 03/03/2000 RHS 200x100x8 S355J2H [#1] 0 - -
TT14 16/03/2000 RHS 200x100x8 S355J2H [#1] 0 - -
TT12 14/03/2000 RHS 200x100x8 S355J2H [#1] 1 165.0 0.94
TT9 08/03/2000 RHS 200x100x8 S355J2H [#1] 2 165.0 0.94
TT8 07/03/2000 RHS 150x150x6.3 S355J2H [#1] 0 - -
TT13 15/03/2000 RHS 150x150x6.3 S355J2H [#1] 1 99.0 0.76
TT11 13/03/2000 RHS 150x150x6.3 S355J2H [#1] 2 38.1 0.29
TT10 10/03/2000 RHS 150x150x6.3 S355J2H [#1] 2 99.0 0.75
Note 1: Only one hydraulic cylinder used for TT1 (acting upwards)
Note 2: The hole size ratio (Φ) is defined by Equation 1-3
The properties of the four bars are summarised in Table 5-2. The results of
individual tensile tests are given in Appendix D. The definitions of the parameters
used can be found in Section 2.2 and BS EN 10002-1:1990.

2 2 2 2
RHS 200x100x8 0.96 344.4 195.2 13.7 0.71 0.25 476.2 16.2
S275J2H [#3]
RHS 150x150x6.3 1.02 329.3 195.2 16.7 -0.16 0.14 457.0 17.9
S275J2H [#1]
RHS 200x100x8 0.71 380.9 203.1 7.0 -0.09 0.24 523.7 12.9
S355J2H [#1]
RHS 150x150x6.3 0.98 404.7 199.9 7.1 0.13 0.34 544.7 13.1
S355J2H [#1]
The measured dimensions of the four bars, used in the manufacture of the
test specimens, are given in Table 5-3. Dimensions are as defined by BS EN 10210-
in BS EN 10210-2:1997 (Appendix C), are listed in Table 5-4. Sectional properties,
based on measured material properties and dimensions, are given in Table 5-5.
Plastic torque was calculated by considering the flow of plastic shear strain around a
simplified cross-section (Equation 2-50), and the torque and rotation at yield were
calculated as in BS EN 10210-2:1997. For comparison, the nominal section
properties are given in Table 5-6.

RHS 200x100x8 S275J2H [#3] 197.5 99.5 7.79 5.0 12.5
RHS 150x150x6.3 S275J2H [#1] 149.0 149.0 6.09 5.0 11.0
RHS 200x100x8 S355J2H [#1] 199.0 100.0 7.93 5.0 12.5
RHS 150x150x6.3 S355J2H [#1] 149.5 149.5 5.79 5.0 11.0

Section size Material grade Cross-sectional area Torsional inertia Torsional modulus
[Bar number] A (cm ) 2 constant, It (cm4) constant, Ct (cm3)
RHS 200x100x8 S275J2H [#3] 42.7 1717 242

RHS 150x150x6.3 S275J2H [#1] 34.0 1816 230
RHS 200x100x8 S355J2H [#1] 43.7 1777 248
RHS 150x150x6.3 S355J2H [#1] 32.5 1755 222
Section size Material grade Rotation at Torque at Plastic torque Shape factor
[Bar number] yield, θel yield, Tel Tpl for torsion
(o/m) (kNm) (kNm) Tpl / Tel
RHS 200x100x8 S275J2H [#3] 2.13 48.1 52.9 1.10

RHS 150x150x6.3 S275J2H [#1] 1.83 43.7 46.2 1.06
RHS 200x100x8 S355J2H [#1] 2.25 54.7 60.3 1.10
RHS 150x150x6.3 S355J2H [#1] 2.20 51.9 54.7 1.05

Section size Material grade Rotation at Torque at Plastic torque Shape factor
[Bar number] yield yield (kNm) for torsion
(o/m) (kNm)
RHS 200x100x8 S275J2H [#3] 1.61 (76%) 39.9 (83%) 44.5 (84%) 1.12 (102%)
RHS 150x150x6.3 S275J2H [#1] 1.45 (79%) 38.0 (87%) 40.8 (88%) 1.07 (101%)
RHS 200x100x8 S355J2H [#1] 2.07 (92%) 51.5 (94%) 57.5 (95%) 1.12 (102%)
RHS 150x150x6.3 S355J2H [#1] 1.87 (85%) 49.1 (95%) 52.7 (96%) 1.07 (102%)
Note: The percentages in parentheses denote the nominal value as a percentage of the measured
Measured initial stiffnesses and maximum torques are listed in Table 5-7. In
cases where no maximum was reached during the test, the torque is quoted for a
particular value of twist. The complete torque-rotation relationships are presented in
Figure 5-13 through Figure 5-19. Torque and rotation have been non-
dimensionalised against the theoretical torque and rotation at yield of the
corresponding unperforated section (Table 5-5).
The results indicated that large openings in the webs caused a significant
reduction in both the torsional stiffness and capacity. The reduction in stiffness (an
average measurement over the whole length of the specimen) was due to the
perforated zone being much more flexible than the unperforated beam, and severe
rotational deformations were observed around the openings. A slight reverse twist
was observed in the beam at either side of the opening due warping and lozenging of
the cross-section. The reduction in stiffness was less severe for the specimens with
square cross-sections than for the specimens with rectangular cross-sections, as, in
the former, the webs contributed a smaller proportion toward the total resistance of
the cross-section to the applied torque.
Observation of the Lüders lines in the steel indicated that yielding occurred
at low magnitudes of torque, due to stress concentrations around the web openings.
However, perforated sections were able to resist torques much higher than those at
the onset of yield, but at the expense of plastic deformation around the hole (hence
the hysteresis in the unloading curves).
The results also indicated that the unperforated sections were unable to
achieve the theoretical torsional capacity (Section 2.5.4), although the experimental
measurements of torsional stiffness and shear strain (Table 5-8) in the elastic range
were close the theoretical predictions. The contradictions between the experimental
observations and the theoretical predictions are discussed later in this Chapter.
Table 5-7: Experimental results – stiffness and maximum torque

ID Section size Material grade Number Hole diameter Stiffness Torque @
o
Code [Bar number] of holes (mm) (kNm/ ) (kNm)
TT4 RHS 200x100x8 S275J2H [#3] 0 - 12.6 43.1 Max
TT6 RHS 200x100x8 S275J2H [#3] 1 165 8.32 20.8 10o
TT2 RHS 200x100x8 S275J2H [#3] 2 165 7.33 16.2 10o
TT3 RHS 150x150x6.3 S275J2H [#1] 0 - 13.4 38.8 Max
TT5 RHS 150x150x6.3 S275J2H [#1] 1 99 12.0 24.3 7o
TT1 RHS 150x150x6.3 S275J2H [#1] 2 99 10.2 23.3 Max
TT7 RHS 200x100x8 S355J2H [#1] 0 - 12.3 53.2 Max
TT14 RHS 200x100x8 S355J2H [#1] 0 - 13.1 53.6 Max
TT12 RHS 200x100x8 S355J2H [#1] 1 165 9.44 26.4 10o
TT9 RHS 200x100x8 S355J2H [#1] 2 165 6.87 23.2 Max
TT8 RHS 150x150x6.3 S355J2H [#1] 0 - 11.9 49.6 Max
TT13 RHS 150x150x6.3 S355J2H [#1] 1 99 11.0 32.1 10o
TT11 RHS 150x150x6.3 S355J2H [#1] 2 38.1 12.6 49.2 7o
TT10 RHS 150x150x6.3 S355J2H [#1] 2 99 10.5 28.1 Max
Table 5-8: Measurements of torsional shear strain in unperforated sections

ID Section size Torsional modulus constant, Ct Shear stress / twist
code Material grade [Bar] Measured % of theory % of theory
3
(cm )
TT4 RHS 200x100x8 270.8 112.1 99.9
S275J2H [#3]
TT3 RHS 150x150x6.3 259.3 113.0 99.5
S275J2H [#1]
TT7 RHS 200x100x8 269.15 108.3 92.4
S355J2H [#1]
Note: Measured values obtained from strain gauges on the specimens
RHS 200x100x8 no holes
1.2
1.0
0.8
0.6
0.4
Grade S275J2H (TT4)
Grade S355J2H (TT7 & TT14)
0.2
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Rotation / Rotation at yield (theory, unperforated)
RHS 150x150x6.3 no holes
1.2
Torque / Torque yield (theory, unperforated)
1.0
0.8
0.6
0.4
Grade S275J2H (TT3)
Grade S355J2H (TT8)
0.2
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Rotation / Rotation at yield (theory,unperforated)
Figure 5-14: Experimental results RHS 150x150x6.3 no holes
RHS 200x100x8 one hole 165 mm diameter
1.2
1.0 Grade S275J2H (TT6)

Grade S355J2H (TT12)
0.8
0.6
0.4
0.2
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
RHS 150x150x6.3 one hole 99 mm diameter
1.2

0.8
0.6
0.4
0.2
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Figure 5-16: Experimental results RHS 150x150x6.3 one hole 99 mm dia
RHS 200x100x8 two holes 165 mm diameter
1.2

Grade S355J2H (TT9)
0.8
0.6
0.4
0.2
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
RHS 150x150x6.3 two holes 99 mm diameter
1.2

0.8
0.6
0.4
0.2
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Figure 5-18: Experimental results RHS 150x150x6.3 two holes 99 mm dia
RHS 150x150x6.3 two holes 38.1 mm diameter
1.2
1.0
0.8
0.6
0.4

0.2
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Figure 5-19: Experimental result RHS 150x150x6.3 two holes 38.1 mm dia
The laboratory tests were modelled using the ABAQUS v5.8.1 FEA program
(Section 2.3). Models included both material and geometric non-linearity and
contained between 6200 and 8300 four-noded shell elements (S4R). These elements
were chosen to create the models because they allowed changes in thickness at high
strains to be modelled (Section 2.3.6). The FE models were based on the actual
measured dimensions of the test specimens and included the corner radii of the tube.
The connecting plates were also included, and the support conditions were modelled
by defining constraint equations to describe the permitted displacements in the nodal
degrees of freedom.
The initial stiffnesses and maximum torques predicted by the FE models are
listed in Table 5-9. In cases where no maximum was reached during the test, the
torque is quoted for the same value of twist as in Table 5-7. The differences
between the experimental values and the FE predictions are listed in Table 5-10.
The complete torque-rotation relationships are presented in Figure 5-20
through Figure 5-32. Torque and rotation have been non-dimensionalised against
the theoretical torque and rotation at yield of the corresponding unperforated section
(Table 5-5).
Although the FE models matched the displaced shapes observed
experimentally (e.g. Figure 5-33 and Figure 5-34), in some instances the torque-
rotation relationships were not good matches. For an unperforated section in
particular, the simplicity of the testing arrangement should lead to good agreement
with standard theory and FE predictions. A difference in excess of 10% is indicative
of an important aspect of behaviour that is missing from the analytical models. FE
models of unperforated sections agreed almost exactly with standard theory, but not
with the behaviour observed in the laboratory.
In most cases, the elastic stiffnesses predicted by FE were close to those
measured experimentally. Stiffness is proportional to the Young’s modulus of the
steel, which is difficult to measure accurately with tensile tests. For perforated
sections, the elastic range was very small and is likely to have been affected by the
loads imposed on the specimens during manufacturing and handling.
Yielding occurred at small rotations for sections with large openings. The
post-yield stiffness is governed by the yield stress and, to a lesser extent, strain
hardening. In the case of the grade S355J2H material with large openings, FE
predictions and experimental results agreed closely over the full range of rotation.
In the case of the grade S275J2H RHS 200x100x8, the specimens were less stiff and
weaker than predicted. For the grade S275J2H RHS 150x150x6.3 with one hole,
post-yield stiffness was similar to the FE prediction in the early stages of the test,
but the maximum torque was lower. For two holes the experimental result and FE
prediction are similar.
Despite the poor agreement with accepted torsion theory, tests TT7 and
TT14 gave near identical results indicating good experimental repeatability.
Table 5-9: Finite Element results – stiffness and maximum torque

ID Section size Material grade Number Hole diameter Stiffness Torque @
o
code [Bar number] of holes (mm) (kNm/ ) (kNm)
TT4 RHS 200x100x8 S275J2H [#3] 0 - 11.3 54.0 Max
TT6 RHS 200x100x8 S275J2H [#3] 1 165 7.59 23.2 10o
TT2 RHS 200x100x8 S275J2H [#3] 2 165 6.02 20.0 10o
TT3 RHS 150x150x6.3 S275J2H [#1] 0 - 11.9 47.4 Max
TT5 RHS 150x150x6.3 S275J2H [#1] 1 99 11.2 27.1 7o
TT1 RHS 150x150x6.3 S275J2H [#1] 2 99 10.4 23.7 Max
TT7 RHS 200x100x8 S355J2H [#1] 0 - 12.1 61.2 Max
TT14 RHS 200x100x8 S355J2H [#1] 0 - 12.1 61.2 Max
TT12 RHS 200x100x8 S355J2H [#1] 1 165 8.29 26.7 10o
TT9 RHS 200x100x8 S355J2H [#1] 2 165 6.67 23.5 Max
TT8 RHS 150x150x6.3 S355J2H [#1] 0 - 11.9 56.3 Max
TT13 RHS 150x150x6.3 S355J2H [#1] 1 99 11.2 32.4 10o
TT11 RHS 150x150x6.3 S355J2H [#1] 2 38.1 11.8 53.6 7o
TT10 RHS 150x150x6.3 S355J2H [#1] 2 99 10.2 27.7 Max
Table 5-10: Experimental and FE results – stiffness and maximum torque

ID Section size Material grade Number Hole diameter Difference Difference
code [Bar number] of holes (mm) stiffness (%) torque (%)
TT4 RHS 200x100x8 S275J2H [#3] 0 - 12.2 -20.2
TT6 RHS 200x100x8 S275J2H [#3] 1 165 9.66 -10.3
TT2 RHS 200x100x8 S275J2H [#3] 2 165 21.7 -19.2
TT3 RHS 150x150x6.3 S275J2H [#1] 0 - 12.3 -18.2
TT5 RHS 150x150x6.3 S275J2H [#1] 1 99 7.76 -10.2
TT1 RHS 150x150x6.3 S275J2H [#1] 2 99 -2.49 -1.84
TT7 RHS 200x100x8 S355J2H [#1] 0 - 1.96 -13.1
TT14 RHS 200x100x8 S355J2H [#1] 0 - 8.70 -12.4
TT12 RHS 200x100x8 S355J2H [#1] 1 165 13.9 -1.09
TT9 RHS 200x100x8 S355J2H [#1] 2 165 3.04 -1.03
TT8 RHS 150x150x6.3 S355J2H [#1] 0 - -0.65 -11.9
TT13 RHS 150x150x6.3 S355J2H [#1] 1 99 -1.39 -0.76
TT11 RHS 150x150x6.3 S355J2H [#1] 2 38.1 6.31 -8.24
TT10 RHS 150x150x6.3 S355J2H [#1] 2 99 2.48 1.49
Note: A negative percentage indicates experimental values lower than Finite Element values
1.2
1.0
0.8
0.6
0.4 FE (s0clbtat)
Experimental (TT4)
0.2 Shape factor
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Figure 5-20: FE and experimental (test TT4)
RHS 150x150x6.3 grade S275J2H no holes
1.2
1.0
0.8
0.6
0.4 FE (l0clbtat)
Experimental (TT3)
0.2 Shape factor
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
1.2
1.0 FE (s1clbtat)
Experimental (TT6)
0.8 Shape factor
0.6
0.4
0.2
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
RHS 150x150x6.3 grade S275J2H one hole 99 mm diameter
1.2
1.0 FE (l1clbtdt)
Experimental (TT5)
0.8 Shape factor
0.6
0.4
0.2
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
1.2
1.0 FE (s2clbtat)
Experimental (TT2)
0.8 Shape factor
0.6
0.4
0.2
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
RHS 150x150x6.3 grade S275J2H two holes 99 mm diameter
1.2
1.0 FE (l2clbtat)
Experimental (TT1)
0.8 Shape factor
0.6
0.4
0.2
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
1.2
1.0
0.8
0.6
0.4 FE (fecalt7t)
Experimental (TT7 and TT14)
0.2 Shape factor
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Figure 5-26: FE and experimental (tests TT7 and TT14)
1.2
1.0
0.8
0.6
0.4 FE (fecalt8t)
Experimental (TT8)
0.2 Shape factor
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
1.2
1.0 FE (feclt12t)
Experimental (TT12)
0.8 Shape factor
0.6
0.4
0.2
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
RHS 150x150x6.3 grade S355J2H one hole 99 mm diameter
1.2
1.0 FE (feclt13t)
Experimental (TT13)
0.8 Shape factor
0.6
0.4
0.2
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
1.2
1.0 FE (fecalt9t)
Experimental (TT9)
0.8 Shape factor
0.6
0.4
0.2
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
RHS 150x150x6.3 grade S355J2H two holes 99 mm diameter
1.2
1.0 FE (feclt10t)
Experimental (TT10)
0.8 Shape factor
0.6
0.4
0.2
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
RHS 150x150x6.3 grade S355J2H two holes 38.1 mm diameter
1.2

1.0
0.8
0.6
0.4 FE (feclt11t)
Experimental (TT11)
Drop of torque in FE at large twist due to Shape factor
0.2
buckling (no buckling in experimental).
Buckling sensitive to tube thickness.
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Figure 5-33: Displaced shape RHS 150x150x6.3 one hole 99 mm dia
Figure 5-34: Displaced shape RHS 200x100x8 two holes 165 mm dia
5.6 The validity of the experimental findings
5.6.1 Introduction
The experimental method was thoroughly examined in an attempt to provide
an explanation for the differences between the measured torsional behaviour and that
predicted by FE and standard torsion theory. The validity of the experimental
results is discussed in detail in Project Report SC2000-010, but the main points are
summarised in the following Sections.
5.6.2 Measurement of rotation
Measurement of rotation was achieved using gravity inclinometers. These
devices were particularly susceptible to external influences such as the stability of
the input voltage, vibration and draughts. These influences had a random effect on
the instrument output, which appeared as noise on the channel. The accuracy of the
inclinometers quoted by the manufacturer was compromised by these, relatively
large, random external influences. Observation of their performance in the
laboratory suggested the following statistics:
Reliability (inspection of experimental results) better than ±0.05o
Precision (observation of data logger performance) around 0.001o
Accuracy (calibration against known quantities) better than ±0.1o
Although the inclinometers were not being used to their full potential, their
performance was adequate for the purposes of the torsion tests. Furthermore, in
every test, the rotation calculated from the vertical movement of the lever arm was
consistent with the rotation measured by the inclinometers (Figure 5-35 is a typical
example).
RHS 200x100x8 grade S275J2H no holes (test TT4)
20
18
Rotation from pot (degrees)
16
14
12
10
8
6
4
2
0
0 2 4 6 8 10 12 14 16 18 20
Rotation from inclinometer (degrees)
Figure 5-35: Rotation measurements
5.6.3 Measurement of torque
In the tests, the torque was measured by monitoring the shear strain on the
external surface of the calibration shaft (Figure 5-5). Six electrical resistance shear
strain gauges (Section 2.4.4.2) were used for this purpose, giving six separate
measurements of torque. In all the tests, the difference between the lowest and
highest indicated torque (the six measurements before averaging) was never more
than 2.5 kNm. Furthermore, a three-element strain gauge rosette, also placed on the
shaft, confirmed that the orientation of the principal strain axes was consistent with
pure torsion in the shaft. Observation of the performance of the calibration shaft in
the laboratory suggested the following statistics for the averaged indicated torque:
Reliability (inspection of experimental results) better than ±0.5 kNm
Precision (observation of data logger performance) around 0.1 kNm
Accuracy (calibration against known quantities) better than ±0.5 kNm
The performance of the calibration shaft was adequate for the purposes of the
torsion tests. Inaccurate measurement of torque is unlikely to be the cause of the
disparity between the laboratory measurements and the predictions of theory and FE
for the following reasons:
1) In the cases of the grade S355J2H material with large openings, the experimental
torque agreed very closely with the FE predictions. In the other cases, the elastic
stiffness was close to that predicted by FE and theory
2) A corresponding error in the measurement of rotation would be necessary to
maintain the agreement with the measured stiffnesses, and those predicted by FE
and theory (Section 5.6.2)
3) The shear strains measured at the surface of the calibration shaft were similar to
those predicted by standard torsion theory (Figure 5-7)
4) The jack forces measured during the tests corresponded to the torque measured
by the calibration shaft (Figure 5-36 is a typical example)
5) In the elastic range, the strains measured on the surface of the specimen were
consistent with those expected for the measured value of torque
50
Torque from strain of calibration shaft
45
40 Unloading
35 after test
Difference at high
30
torque and unloading
(kN)
25 due to large rotation

20 (lever arm no longer
Loading perpendicular to jacks)
15
10
5
0
0 10 20 30 40 50
Torque calculated from load cells on jacks (kN)
Figure 5-36: Load measurements
5.6.4 Measurement of strain
Inspection of the experimental results suggested a reliability and precision in
the order of ±0.1 microstrain. The manufacturer’s documentation quotes an
accuracy of ±1 % for strains less than 0.04. This level of accuracy is only possible if
the gauges are used correctly.
5.6.5 Repeatability of experimental results
Repeatability of results is necessary to prove the reliability of the testing
method. Two unperforated specimens cut from the same length of RHS 200x100x8
S355J2H (tests TT7 and TT14) gave very similar results (Figure 5-13) indicating
good repeatability and hence good reliability of the testing method. Good reliability
can also be inferred from the good repeatability within individual tests (elastic tests
and unloading-reloading after plastic deformation).
5.6.6 Accuracy of the data measurement and logging system
An error introduced by the data measurement and logging equipment could
affect all the experimental measurements, but is unlikely to be the cause of the
disparity between the experimental measurements and the predictions of FE and
theory for the following reasons:
1) The measurements would have to be changed in different directions to result in
an error of this type, while still maintaining consistency between strain, rotation
and torque in the elastic range
2) An error in the measurement and logging equipment would be likely to affect all
tests equally, but the differences between the experimental measurements and FE
predictions were not equal for all tests
3) The equipment was in frequent use for other work and no problems were
reported
4) The accuracy of the measurement and logging equipment was checked with
known voltages, loads and displacements, and was found to be well within
acceptable limits
5.6.7 Additional loads on the specimen
Additional loads on the specimens could have caused them to fail at lower
torques than those predicted by the FE models, if those additional loads were not
included in the FE models. Potentially, these loads may have resulted from friction,
out-of-balance hydraulic jacks, warping restraint, bending due to asymmetry, and/or
axial restraint. However, this is an unlikely explanation for the disparity between
the experimental measurements and the predictions of FE and theory for the
following reasons:
1) The elastic strains measured on the specimens do not indicate anything other
than pure torsional shear
2) According to accepted theory, the reduction in the torsional capacity due to
warping stresses in box sections is so small it may be neglected (Section 2.5.3)
3) Warping restraint is included in the FE models
4) Square cross-sections do not warp, but produced similar results to the
rectangular cross-sections
5) Friction between the rotating beam and the guide columns would not have
affected the torque measured on the calibration shaft
6) The rig structure was too flexible to exert the enormous bending and axial
forces required to reduce the torsional capacity by 10%
7) External forces would also have affected the linearity of the elastic response of
the unperforated specimens
8) FE models, of unperforated beams with axial restraint, predicted identical
capacities to the FE models without axial restraint (since changes in length due
to deformation were very small e.g. Figure 2-25)
When the unperforated specimens were tested, Lüders lines were first
observed at the corners where the RHS was welded to the connecting plates.
Localised effects due to captured thermal stresses, warping, and heat induced
metallurgical changes are likely to be responsible for this effect, but they do not
explain the differences between the measured torque-rotation response and the FE
predictions. At maximum torque, all the unperforated specimens exhibited plastic
deformation along the whole length. Perforated specimens formed plastic hinges
around the holes without plastic deformation at the ends.
5.6.8 The validity of the theoretical and analytical predictions
5.6.8.1 Introduction
It is possible that the experimental measurements were correct, but that
theoretical and FE models may have been incorrectly applied.
5.6.8.2 Boundary conditions
Correct identification of the boundary conditions in particular has been found
to be an important aspect of the FE modelling. In an FE model, the ends of the
specimen may, or may not be, restrained from rotating about the major and minor
axes of the cross-section. For an asymmetric case (one hole), a pinned boundary
condition results in a lower capacity than a fixed boundary condition (Figure 5-37
and Figure 5-38). However, boundary conditions have no effect on the response of
the symmetric cases.
RHS 200x100x8 grade S275J2H 1 hole 165 mm diameter
0.8
0.7
0.6
0.5
0.4
0.3
0.2 FE, fixed boundary conditions (s1caltbt)

FE, pinned boundary conditions (s1caltet)
0.1 FE, fixed about the vertical axis pinned about the horizontal axis (s1clbtat)
Experimental (TT6)
0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Figure 5-37: Boundary conditions RHS 200x100x8 grade S275J2H 1 hole
RHS 150x150x6.3 grade S275J2H 1 hole 99 mm diameter
0.8
0.7
0.6
0.5
0.4
0.3
0.2 FE, fixed boundary conditions (l1caltbt)

FE, pinned boundary conditions (l1caltet)
0.1 FE, fixed about the vertical axis pinned about the horizontal axis (l1clbtat)
Experimental (TT5)
0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Figure 5-38: Boundary conditions RHS 150x150x6.3 grade S275J2H 1 hole
5.6.8.3 Cross-section dimensions
If incorrect dimensions had been used in the calculations, then the FE and
theoretical predictions would be in error. Consideration of the theoretical equations
governing behaviour of an unperforated RHS in torsion allowed the sensitivity of the
theoretical torsional properties to changes in the dimensions to be quantified
(Appendix B). However, the use of erroneous dimensions is an unlikely explanation
for the disparity between the experimental measurements and the predictions of FE
and theory for the following reasons:
1) No single parameter affects behaviour in the correct way, and unfeasibly large
changes are required to reduce torsional capacity by 10%
2) Grade S355J2H specimens with large holes agreed with FE, yet they have the
same cross-section dimensions as the unperforated specimens
3) Specimens were re-measured after the tests and the dimensions were found to be
correct
5.6.8.4 Material properties
If incorrect material properties had been used in the calculations, then the FE
and theoretical predictions would be in error. Consideration of the theoretical
equations governing behaviour of an unperforated RHS in torsion allowed the
sensitivity of the theoretical torsional properties to changes in the material properties
to be quantified (Appendix B).
Yield stress influences torsional capacity, without changing stiffness, but is
easy to measure accurately with tensile tests. Young’s modulus is difficult to
measure accurately with tensile tests, but does not influence capacity, unless there is
instability. The value of poisson’s ratio used in the analysis (0.3) is an assumed
rather than a measured quantity, but only influences stiffness. Residual stresses in
hot formed steel tubes are small and do not reduce plastic capacity when they are
present.
It is possible that the use of erroneous values of yield stress, in the theoretical
and FE calculations, is the cause of the disparity between the predictions of torsional
capacity and the experimental measurements. The following evidence supports this
conjecture:
1) Some of the theoretical and FE predictions can be made to match the
experimental results by using a lower yield stress (see below)
2) Experimental observations of strain in the specimen indicate plastic strain
occurring when measured material properties predict elastic behaviour (see
below)
However, there is also evidence that suggests that the correct value of yield stress
was used in the theoretical and FE calculations:
1) The material properties were measured by a large number of tensile tests in
accordance with BS EN 10002 (Appendix D)
2) The variation in yield stress as measured by tensile testing of coupons from
individual bars was small, indicating homogeneity and good experimental
repeatability (Appendix D)
3) FE predictions of grade S355J2H specimens with large holes matched the
experimental results when the measured material properties were used
4) Similar RHS were tested in three and four point bending, and measured material
properties were consistent with experimental results (Chapters 3 and 4)
5) Two coupons, inclined to the axis of the tube were tested in an attempt to
measure anisotropy, but no anisotropy was observed (Appendix D)
6) One coupon was tested at a very low strain rate, but no difference was observed
when compared with the BS EN 10002 coupons (Appendix D)
7) Special coupons, designed to investigate the shear yield behaviour of the steel,
indicated that the correct values of shear yield stress were used in the
calculations (Section 2.5.4.3)
Despite the numerous arguments against the ‘reduced yield’ explanation, it
remains a feasible mechanism that explains how torsional capacity could potentially
have been reduced, without affecting the elastic response.
When a yield stress of 275 N/mm2 was used in an FE model of the RHS
200x100x8 grade S275J2H no holes specimen, the torque-rotation relationship
matched the experimental result (Figure 5-39). When a yield stress of 267 N/mm2
was used in an FE model of the RHS 150x150x6.3 grade S275J2H no holes
specimen, a similar match was achieved (Figure 5-40). When these reduced yield
stresses were applied to the specimens with holes, the torque-rotation relationships
matched for the RHS 200x100x8 case, but not for the RHS 150x150x6.3 (Figure
5-41 through Figure 5-44).
Similar reductions in yield stress for the unperforated grade S355J2H
specimens were used to match the FE predictions with the experimental results, but
no reductions were required to match the behaviour of the specimens with large
holes.
When the strain gauge data was examined for the unperforated specimens, it
showed behaviour similar to the examples shown in Figure 5-45 and Figure 5-46.
At low twist, strains were uniform and linear and matched those predicted by FE and
theory. However, a departure from linearity occurred at strains that should have
been elastic. In the examples here, the strain was, at first, maintained at a constant
level indicating that failure was occurring at some other location in the specimen.
With more rotation, the strains began to separate and increase, indicating that plastic
deformation was occurring at the location of the gauges.
The difference between the experimentally measured torque and the torque
calculated, using torsion theory, from the experimentally measured strains in the
specimen can be used to back-calculate a yield stress from the torsion test. The
yield stresses back-calculated from strain data were the same as the reduced yield
stresses necessary to match experimental and FE torque-rotation relationships.
It appears from the observations that the steel does not obey the von Mises
failure criterion, but this is contrary to the results of previous investigators (Section
2.5.4.2) and the results of attempts to measure the shear yield stress directly (Section
2.5.4.3). It is unclear why some specimens behaved as predicted using measured
yield stresses, and why some specimens behaved as predicted using reduced yield
stresses. It is possible that this could be a result of variability within the steel, but no
evidence is available to support this suggestion.
1.2
Torque / Torque at yield (theory, unperf, red)
1.0
0.8
0.6
0.4
FE material model FE with reduced yield (s0caltdt)

0.2
yield stress reduced to Experimental (TT4)
2
275 kN/mm Shape factor
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Rotation / Rotation at yield (theory, unperforated, reduced)
Figure 5-39: Reduced yield RHS 200x100x8 grade S275J2H no holes
1.2
1.0
0.8
0.6
0.4
FE material model FE with reduced yield (l0caltdt)

0.2
yield stress reduced to Experimental (TT3)
2
267 kN/mm Shape factor
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Figure 5-40: Reduced yield RHS 150x150x6.3 grade S275J2H no holes
RHS 200x100x8 grade S275J2H 1 hole 165 mm diameter
0.7
0.6
0.5
0.4
0.3
0.2
FE material model
FE with reduced yield (s1caltdt)
0.1 yield stress reduced to
2
275 kN/mm Experimental (TT6)
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Figure 5-41: Reduced yield RHS 200x100x8 grade S275J2H one hole
RHS 150x150x6.3 grade S275J2H 1 hole 99 mm diameter
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2 FE material model

yield stress reduced to FE with reduced yield (l1caltdt)
0.1 267 kN/mm
2
Experimental (TT5)
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Figure 5-42: Reduced yield RHS 150x150x6.3 grade S275J2H one hole
RHS 200x100x8 grade S275J2H 2 holes 165 mm diameter
0.45
0.40
0.35
0.30
0.25
0.20
0.15
0.10 FE material model

FE with reduced yield (s2caltdt)
yield stress reduced to
0.05 2
0.00
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
Figure 5-43: Reduced yield RHS 200x100x8 grade S275J2H two holes
RHS 150x150x6.3 grade S275J2H 2 holes 99 mm diameter
0.7
0.6
0.5
0.4
0.3
0.2
FE material model
yield stress reduced to FE with reduced yield (l2caltdt)
0.1
2
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
Figure 5-44: Reduced yield RHS 150x150x6.3 grade S275J2H two holes
3.0
Maximum shear strain
Max shear strain / Shear strain at yield
South web
2.5
Rose 1
2.0 Rose 2
Rose 3
1.5 Rose 4
FE (s0clbtat)
1.0
0.5
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Total rotation / Rotation at yield (theory, unperforated)
Figure 5-45: Strain measurements (test TT4 south web)
RHS 200x100x8 grade S275J2H no holes (TT4)
3.0
Maximum shear strain
Max shear strain / Shear strain at yield
North web
2.5
Rose 5
2.0 Rose 6
Rose 7
1.5 Rose 8
FE (s0clbtat)
1.0
0.5
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Total rotation / Rotation at yield (theory, unperforated)
Figure 5-46: Strain measurements (test TT4 north web)
5.6.9 Marshall’s results
Torsion tests on unperforated hot finished RHS, of sizes similar to those used
in this study, were conducted by Marshall (1972), who compared the results to the
predictions made by his thick wall torsion theory (Section 2.5.2). As is the case with
the unperforated sections tested in this project, Marshall achieved good agreement
between the theoretical predictions (of shear strain and torsional stiffness) in the
elastic range and the experimentally measured values.
However, when Marshall compared his measurements of elastic and plastic
torsional capacities, he chose to use the Tresca yield criterion rather than the von
Mises criterion (Sections 2.5.4.2 and 2.5.5) in his theoretical calculations. As a
result, his theoretical predictions of the torsional capacities were 15% lower than
they would have been if he had used the von Mises criterion, which is widely
regarded as the more appropriate criterion for such calculations. When Marshall’s
test data was re-examined, using the theoretical treatments employed to analyse
torsional behaviour in this project, the results showed evidence of behaviour similar
to that observed in the full-scale torsion testing component of this research.
The dimensions and material properties of Marshall’s test specimens, as
quoted in reference Marshall (1972), are listed in Table 5-11. The torsional
constants, calculated from these dimensions, are listed in Table 5-12, and the
theoretical torsional capacities (using the von Mises criterion) are listed in Table
5-13. Marshall’s measurements of the torsional capacities are listed in Table 5-14,
and compared with the theoretical predictions. The calculations indicate that the
experimental measurements were up to 18% lower than the theoretical predictions,
and typically 10% lower for the larger sections (Figure 5-47).
Table 5-11: Dimensions and material properties of Marshall’s torsion test specimens
Test Height Width Thickness Yield stress Young’s Poisson’s
h (mm) b (mm) t (mm) fy1 (N/mm2) modulus ratio
E (kN/mm2) ν
A 50.8 50.8 4.9 307.3 211.6 0.29
B 63.5 63.5 4.9 307.3 211.6 0.29
C 76.2 76.2 4.9 432.4 211.6 0.29
D 101.6 101.6 6.4 307.3 211.6 0.29
E 101.6 101.6 4.1 307.3 211.6 0.29
F 127.0 50.8 6.4 307.3 211.6 0.29
G 76.2 38.1 4.1 307.3 211.6 0.29
H 76.2 50.8 3.3 307.3 211.6 0.29
I 101.6 50.8 3.3 307.3 211.6 0.29
Note: Data obtained from reference Marshall (1972)
Table 5-12: Properties of cross-section for Marshall’s torsion test specimens

Test Cross sectional area Torsional inertia constant Torsional modulus constant
2 4
A (cm ) It (cm ) Ct (cm3)
A 8.70 49.25 17.1
B 11.2 102.0 28.9
C 13.7 182.4 43.8
D 23.8 565.5 102
E 15.7 384.8 71.5
F 20.5 228.8 57.4
G 8.44 47.66 17.1
H 7.72 66.58 20.3
I 9.37 99.57 27.7
Note 1: Calculations based on data obtained from reference Marshall (1972)
Note 2: Calculations made using equations in BS EN 10210-2:1997 (Also in Appendix C)
Table 5-13: Theoretical plastic torque and torque at yield for Marshall’s torsion tests
Test Rotation at yield Torque at yield Plastic torque Shape Factor for torsion
θel ( /m)
o
Tel (kNm) Tpl (kNm) Tpl / Tel
A 4.30 3.03 3.55 1.17
B 3.51 5.13 5.83 1.14
C 4.19 10.9 12.2 1.11
D 2.24 18.1 20.1 1.11
E 2.30 12.7 13.6 1.07
F 3.11 10.2 12.1 1.19
G 4.43 3.02 3.50 1.16
H 3.78 3.60 3.95 1.10
I 3.44 4.91 5.36 1.09
Note 1: Calculations based on data obtained from reference Marshall (1972)
Note 2: Calculations made using equations in BS EN 10210-2:1997 (Also in Appendix C)
Note 3: Plastic torque calculated using Equation 2-50
Table 5-14: Plastic torque and torque at yield measured by Marshall

Test Marshall’s measurements Difference measured - theory
Torque at yield Plastic torque Torque at yield Plastic torque
Ty (kNm) Tpl (kNm) (% of theory) (% of theory)
A 2.78 3.67 -8.10 3.50
B 4.56 5.31 -11.2 -8.80
C 9.36 10.2 -14.3 -15.7
D 15.9 17.7 -11.9 -11.8
E 11.6 12.0 -8.20 -11.4
F 8.35 10.4 -18.0 -14.0
G 2.78 3.29 -8.00 -6.00
H 3.54 3.75 -1.70 -5.30
I 4.56 4.81 -7.20 -10.3
Note 1: Data obtained from reference Marshall (1972)
Note 2: A negative percentage indicates experimental values lower than theoretical values
Theoretical torque at yield (kNm)

0%
0 10 20 30 40 50 60
measured - theoretical (% of theoretical)
-5% This project (grade S275)
Difference in torque at yield
This project (grade S355)

Marshall's tests (Marshall 1972)
-10%
-15%
-20%
-25%
Figure 5-47: Measurements and predictions of torque at yield (unperforated)
5.6.10 Summary
Some torsional capacities measured in the laboratory were significantly
lower than the theoretical and FE predictions. The validity of the experimental
findings has been verified leading to the conclusion that there is some unknown
aspect of behaviour, which is not included in the theoretical and FE models. One
explanation (that appears to fit most of the available evidence) is that the steel does
not always comply with the von Mises failure criterion. Evidence of similar
behaviour is present in the results of full-scale testing of hot-finished RHS by
Marshall (1972).
5.7 Suitable schemes for further investigation
Further experimental work is necessary if the differences between the
experimental results, and the theoretical and FE predictions, are to be explained.
The following schemes are suitable:
1) Additional torsion tests of unperforated tubes of different sizes and grades
2) Torsion tests of Circular Hollow Sections – to investigate the response of the
steel to pure shear without the complications of shape
3) Torsion tests of cold formed or aluminium tubes – to investigate the torsional
behaviour of box sections made from different materials
4) Additional shear testing of the remaining material – to investigate the variability
of the shear yield stress (Section 2.5.4.3)
A series of 14 torsion tests were performed on two sizes and grades of RHS
with and without circular openings. For each test, the torque and twist were
measured. In half the tests, strains on the external surface of the RHS were also
measured.
When the experimental results were compared with FE predictions, good
matches were observed in terms of displaced shapes, elastic stiffness and strains in
the elastic range. Torque-rotation relationships and torsional resistances matched
FE predictions for some cases, but were significantly different in others. The largest
difference occurred for the unperforated sections where the best agreement was
anticipated.
The experimental measurements were re-examined and found to be valid.
One explanation for the differences between measured and predicted behaviour, that
fits most of the available evidence, is that the steel did not always comply with the
von Mises failure criterion. Attempts were made to measure this effect, but without
success (Section 2.5.4.3).
Although the anomalous results have not been fully explained, evidence of
similar behaviour in previous full-scale tests, conducted by Marshall (1972), was
discovered. Paradoxically, the torsional capacities measured by the small-scale
torsion tests, discussed in the following Chapter, agree extremely well with the
predictions of FE and theory.
6 Small-scale torsion
6.1 Introduction
The full-scale torsion tests described in the previous Chapter required the
preparation of large and expensive testing equipment, with consequent time
penalties. The availability of a small-scale torsion testing machine (Figure 6-1),
used usually for undergraduate teaching, enabled an extensive series of tests to be
carried out during the preliminary stages of the project.
The small-scale study indicated some notable similarities and differences
when compared to the behaviour observed in the full-scale torsion study (Chapter 5).
In addition to providing valuable insights into the nature of torsional behaviour at
both scales, it acted as an additional source of experimental data for calibration of
the Finite Element (FE) models. The small-scale torsion study was presented in a
conference paper (Ridley-Ellis et al (1998)) shortly after it was conducted.
However, this Chapter contains substantial additional discussion and analysis, which
extends the published research.
In the small-scale torsion testing series, a large number of tests on cold-
formed steel tube in torsion revealed the effects of various arrangements of circular
cut-outs on torsional capacity and modes of failure. The tests provided an
understanding of how the diameter of the circular openings can influence both
torsional resistance and stiffness. Comparison of the experimental results with FE
predictions indicated good agreement over the whole range of hole diameters.
6.2.1 The torsion test apparatus and specimens
The small-scale torsion tests were conducted on lengths of square section
cold-formed steel tube, the dimensions of which are shown in Figure 6-2. The
cross-section of the tube was close to being geometrically similar to the RHS
250x250x10 used in the three and four-point testing programmes (Chapters 3 and 4)
and the RHS 150x150x6.3 used in the full-scale torsion study (Chapter 5). The
small tube had a web slenderness of 23.6 compared with 22.0 for the RHS
250x250x10 and 20.8 for the RHS 150x150x6.3. The specimens were 300 mm in
length, but were restrained at the ends by solid metal inserts and stiff clamping
plates (Figure 6-2 and Figure 6-3). The inserts provided connectivity with the
testing rig (Figure 6-1).
In all the tests described in this Chapter, the length of the portion of the
specimen that was free to twist under load was 240 mm. A number of sections of
different lengths, but without holes, were also tested as part of an exercise to
determine the stiffness of the testing machine (Appendix A). In all cases, the torque
(T) was applied at a constant rate of twist (31/3 degrees per minute).
Torque dial
Fixed Free travelling

specimen Test specimen straining unit
holder
Electric motor
Load
Gearbox
indicating
cabinet
Rotating specimen holder
Figure 6-1: The small-scale torsion rig
Test specimen
Clamping plate
Centreline corner radius
1 mm (approx.)
Solid metal insert (b) 38.0 mm

Clamping stud
(b) 38.0 mm
(t) 1.43 mm
Driving flat
Figure 6-2: Small-scale torsion test specimen
Note: Photograph is of a 400 mm specimen

used in the calibration of the testing machine
Figure 6-3: The inserts and clamping plates
6.2.2 Types of specimen and scope of study
The test specimens were classified according to the number and position of
holes. Figure 6-4 shows the types of test pieces. The figure shows all four faces of
the tube as if it was opened out; the dotted lines represent the corners and the tube is
such that the lines labelled AB coincide. Table 6-1 is a summary of the dimensions
of the specimens tested.
Specimens of type I formed the control group. They had no holes and their
torsional resistance and stiffness could therefore be predicted by established
formulae (Section 2.5). The behaviour of the specimens with web openings was
compared directly with the behaviour of the control group to quantify the influence
of the hole(s).
Specimens of type II had a single circular hole. The diameter (D) varied, but
the hole was always placed centrally in one face. Specimens of type III had two
circular holes. The diameter varied, but the holes were always placed centrally in
opposing faces. Specimens of type IV had two circular holes. The holes are placed
symmetrically in the same face at a series of different separations (s).
The material properties of the test pieces were determined from direct tensile
testing of coupons cut from the walls of the tube. Tensile test piece dimensions
were specified in accordance with BS EN 10002–1:1990. The results of the
materials tests are shown in Figure 6-5. The torsion specimens were cut from one of
three bars, with similar material properties. Material models for FE calibration were
based on the coupon results for the corresponding bar (Table 6-2).
The dimensions of the tube are listed in Table 6-3 and the properties of the
cross-section are listed in Table 6-4. Sectional properties based on the average
material properties of the three bars are listed in Table 6-5. Sectional properties
were calculated using the formulae in BS EN 10210 (Appendix C) with the
exception of plastic torque, which was calculated from Equation 2-50.
The nature of the stress-strain relationship for the cold-formed steel meant
that unperforated specimens were able to generate a maximum torque significantly
higher (23%) than the theoretical plastic torque. The cold-formed steel exhibited
two regions of strain hardening. The first region consisted of a rapid gain in stress
immediately following the yield point, in contrast to the very long flat yield plateau
in the hot finished RHS (Figure 6-6). The second, more gradual, strain hardening
region was much shorter than the strain hardening region for the hot finished RHS.
The tensile strength (engineering stress) was reached at 4% strain in contrast to 16%
strain for the hot finished RHS. The strains generated in the small and full-scale
torsion tests were of similar magnitude, but the relatively rapid and early onset of
strain hardening in the cold-formed tube meant that strain hardening played a much
more significant role in the small-scale tests. Table 6-5 contains a theoretical
estimate of the maximum torque based on a plastic shear stress flow equivalent to
the 0.5% proof stress. The theoretical maximum torque (Tmax) was calculated using
Equation 2-50 substituting the 0.5% proof stress for the yield stress.

Type Number of Hole diameter Hole separation
specimens Diameter Hole size D/h (%) Separation s/D (%)
tested D (mm) ratio, Φ s (mm)
I 6 - - - - -
II 2 5 0.15 13.1 - -
2 10 0.30 26.3 - -
2 15 0.44 39.5 - -
2 20 0.59 52.6 - -
2 25 0.74 65.8 - -
2 30 0.89 78.9 - -
III 2 5 0.15 13.1 - -
2 10 0.30 26.3 - -
2 15 0.44 39.5 - -
2 20 0.59 52.6 - -
2 25 0.74 65.8 - -
2 30 0.89 78.9 - -
IV 1 20 0.59 52.6 5 25
1 20 0.59 52.6 10 50
1 20 0.59 52.6 15 75
1 20 0.59 52.6 20 100
1 20 0.59 52.6 25 125
Note: The hole size ratio (Φ) is the ratio of the hole diameter to h-3t (refer to Figure 1-9)
A Type I: control (no holes) B A Type II: single hole B
Hole diameter (D)

A B A B
A Type III: two holes facing B A Type IV: two holes same side B
D = Hole diameter
Hole diameter (D)
Hole separation (s)
Hole diameter (D) D D

A B A B
Figure 6-4: Types of specimen (developed views)

500 500
2
2
400 400
2
300 300 Tensile strength 401 N/mm
2
0.2% proof stress 364 N/mm
200 200 2
0.5% proof stress 370 N/mm
100 100
25% strain at fracture
0 0
0 1 2 3 4 5 0 5 10 15 20 25 30
Engineering strain (%) Engineering strain (%)
Figure 6-5: Stress-strain relationship from tensile tests
Table 6-2: ABAQUS material models for cold-formed tube

Bar 1 Bar 2 Bar 3
*ELASTIC *ELASTIC *ELASTIC
169902.2 , 0.3 197231.9 , 0.3 181362.6 , 0.3
*PLASTIC *PLASTIC *PLASTIC
305.55 , 0 305.47 , 0 291.76 , 0
350.82 , 0.000282 340.61 , 0.000071 332.15 , 0.000256
371.11 , 0.000811 350.70 , 0.000220 353.16 , 0.001008
375.38 , 0.001484 357.07 , 0.001185 387.02 , 0.012055
393.82 , 0.012571 387.73 , 0.012923 407.00 , 0.027315
404.88 , 0.022310 404.52 , 0.025564 422.20 , 0.050426
411.41 , 0.031980 420.53 , 0.046658 440.55 , 0.092881
451.95 , 0.137102 457.70 , 0.137441 457.70 , 0.137238
Note: Refer to Sections 2.2.5 & 2.3.7

Section Height Width Thickness Corner radius (mm)
h (mm) b (mm) t (mm) Internal External
Cold formed tube 38.0 38.0 1.43 0.3 1.7

Section Cross-sectional area, Torsional inertia Torsional modulus
A (cm2) constant, It (cm4) constant, Ct (cm3)
Cold formed tube 2.07 7.082 3.56
Section Rotation at Torque at Plastic Shape factor Maximum
yield, θel yield, Tel torque, Tpl for torsion torque, Tmax
(o/m) (kNm) (kNm) Tpl / Tel (kNm)
Cold formed tube 7.13 0.618 0.657 1.06 0.807
Note: Maximum torque is plastic torque based on 0.5% proof stress rather than yield
500
0.2% proof stress line 0.5% proof stress line
450 Yield for hot finished
400
2
350
300 First strain hardening zone

250
Yield for cold-formed
200
150
100 Cold-formed (bar 3)

50 Hot finished (RHS 250x250x10 bar 1)
0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Figure 6-6: Stress-strain relationship for cold-formed and hot finished steel (1%)
500
Second strain
450 hardening zone
400
2
350
300
250
200
150
100
Cold-formed (bar 3)
50
Hot finished (RHS 250x250x10 bar 1)
0
0 2 4 6 8 10 12 14 16
Figure 6-7: Stress-strain relationship for cold-formed and hot finished steel
6.3.1 Measured torque-rotation relationships
The measured torque-rotation histories for the one and two hole specimens
(type II and type III) are shown in Figure 6-8 and Figure 6-9. In the figures, the
torque is non-dimensionalised against the theoretical torque at yield of the
unperforated tube (Tel), and rotation is non-dimensionalised against the theoretical
rotation at yield (θel). The theoretical maximum torque (Tmax) is also plotted.
These two figures also include the torque-rotation histories for the control
specimens (type I). The initial elastic behaviour is followed by extensive plastic
deformation (10 to 15 times the elastic rotation). The plastic deformation shows
marginally positive stiffness that is associated with strain hardening and the effects
of large deflection. This is followed by a significant drop in torque, associated with
large out-of-plane distortion of the walls near the ends of the tube adjacent to the
ends of the metal inserts. Although the full cross-section was observed to reach the
yield stress, the sections failed by buckling before the full strain hardened tensile
strength observed in the coupon tests could develop.
Figure 6-8 also shows the behaviour of specimens with single holes (type II).
The effect of the diameter of the hole was seen to be important: small holes had little
effect on the overall resistance but decreased the extent of plastic behaviour.
However, for larger holes (Φ greater than 0.3) the reduction in resistance was seen
to be significant. For large holes, the subsequent negative stiffness, associated with
collapse at the location of the opening, was seen to be reduced, and hardly existed
for Φ greater than 0.7.
The onset of non-linearity was seen to be more pronounced in the specimens
with larger holes. As in the full-scale torsion tests (Chapter 5), this is due to stress
concentrations and rotational displacements that were concentrated over a short
length around the hole. It was not possible to identify the point at which permanent
damage first occurred from the experimental results alone, but the FE study (Section
6.4) was used examine the distribution of stress indirectly.
The web openings reduced the initial stiffness of the section, although the
reduction was less pronounced for a hole of small diameter than for one of large
diameter. The stiffness described here is an average value, since the rotation is
measured over the full 240 mm length of the specimen.
Figure 6-9 shows the torque-rotation histories for the control specimens (type
I) and the specimens with two holes facing (type III). As in the full-scale torsion
testing study (Chapter 5), the results showed the double holes influencing torsional
performance in a similar way to the single holes. A double hole was found to have a
similar effect to that of a single hole of equal diameter (Figure 6-10). However, the
negative stiffness effect as shown in Figure 6-9 remained strong for all the
specimens tested compared with the single sided holes shown in Figure 6-8.
The complete torque-rotation relationship for type IV specimens (two holes
in the same side) is shown in Figure 6-11. Figure 6-12 shows how the distance
between the holes affected the peak torque. For large separations the capacity was
similar to that of a specimen with a single hole of equal diameter. For smaller
separations, interaction between the holes resulted in a reduction in capacity. Type
IV specimens also show a greater rotation before the peak torque was reached and
reduced subsequent negative stiffness, as the distortion was shared between the two
openings.
Maximum torque
1.4
(based on 0.5% proof stress)
1.2
Φ=0.30 Φ=0.15 Φ=0.00
Φ=0.44
1.0
Φ=0.59
0.8
Φ=0.74
0.6
0.4 Φ=0.89
0.2
0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Figure 6-8: Torque rotation curves – One hole (Type II)
Maximum torque
1.4 (based on 0.5% proof stress)
1.2
Φ=0.30 Φ=0.15 Φ=0.00
1.0
Φ=0.44
Φ=0.59
0.8
Φ=0.74
0.6
0.4
Φ=0.89
0.2
0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Rotation / Rotation at yield (theory, unperforared)
Figure 6-9: Torque rotation curves – Two holes facing (Type III)
120
% of unperforated (average, experimental)
100
80
Maximum torque
60
Type II
(one hole)
40 No holes (Type I)
Type III
One hole (Type II)
(two holes)
20
Two holes (Type III)
0
0.00 0.20 0.40 0.60 0.80 1.00
Hole diameter (D/b)
Figure 6-10: Peak torque and hole diameter – Types I, II & III
1.0 20 mm diameter hole Φ = 0.59
0.9
0.8
0.7
0.6
Single hole (Type II)
0.5 Two holes; s = 5 mm (Type IV)
0.0
0 1 2 3 4 5 6 7 8 9 10 11 12
Figure 6-11: Torque-rotation curves for type IV specimens
20 mm diameter hole Φ = 0.59
68
66
% of unperforated (average, experimental)
64
62
Maximum torque
60
58
56
54 Double: One side (Type IV)

52
Single (Type II)
50
0 5 10 15 20 25
Separation s (mm)
Figure 6-12: Peak torque and hole separation (Type IV)
6.3.2 Mechanisms of failure
The specimens all failed in one of two ways: either a local end failure after
global plastic strain, or a local failure in the region of the hole due to the weakening
influence of the perforation.
The failure of the stronger sections (those without holes or with Φ less than
0.2) was accompanied by the formation of a buckle on each of the four faces at 45o
to the longitudinal axis of the section as shown in Figure 6-13. Sections with larger
holes failed in the region of the hole in the manner shown in Figure 6-14 (Type II) &
Figure 6-15 (Type III). The failure mode of the type IV specimens is shown in
Figure 6-16.
Figure 6-13 shows local failure initiated adjacent to the metal insert and
clamp. Plastic strain occurred along the whole length of the specimens that failed in
this manner indicating that such sections had developed the full plastic (non-strain-
hardened) torsional capacity prior to the formation of the buckle.
In type II specimens (unsymmetrical) most of the rotation took place in the
region of the hole and was about an axis parallel to, but offset from the longitudinal
axis of the section. The face opposite the perforation remained essentially plane,
which resulted in bending of the section. It was observed that the unsymmetrical
mode of failure caused the cross-section to lozenge (Figure 6-14).
In type III specimens (symmetrical, two holes facing. Figure 6-15) the
rotation took place about the longitudinal axis of the section, and there was no
bending or lozenging effect.
The mode of failure for type IV specimens seen in Figure 6-16 strongly
resembles that of the type II specimens, but with a shearing of the material between
the holes. Initial plastic deformation was observed to occur at equal rates at both
holes before the diagonal inward buckles formed around the (marginally) weaker of
the two cut-outs.
A feature common to all the modes of failure in the local region around the
hole was the sudden change in curvature at the distorted corners of the section
marked with an asterisk in the Figures. This zone was associated with the high
tensile stresses that developed around the hole (Figure 6-17). Necking and even
tearing could be seen in this region in the specimens with larger holes.
The test showed that web openings could have a profound effect on the
torsional performance of the specimens. They also showed failure by several
different mechanisms depending on the number, diameter and position of the holes.
View along axis of specimen
Failure by formation
of inward buckles
on each face
Plastic twist along whole length
Figure 6-13: Post-yield failure by buckling
Lozenging of cross section Inward buckle
Tension
*
Outward buckle
Tension
Tension
Outward buckle
* * = Sudden change of curvature
Figure 6-14: Failure of type II specimens
Inward buckle
*
Tension
*
Outward buckle
Tension
Tension
Outward buckle
Figure 6-15: Failure of type III specimens
Inward buckle
* = Sudden change of curvature
Tension
*
Outward buckle Tension
Shear
Tension
Outward buckle
Figure 6-16: Failure of type IV specimens
* = Sudden change of curvature

*
High
High tensile
tensile stress
stress
Figure 6-17: Regions of high tensile stress around the hole
6.4.1 Introduction
The laboratory tests were modelled using the ABAQUS v5.8.1 FEA program
(Section 2.3). Models included both material and geometric non-linearity and
contained between 3100 and 5900 four-noded shell elements (S4R). The FE models
were based on the actual measured dimensions of the test specimens and included
the corner radii of the tube. The support conditions and were modelled by defining
constraint equations to describe the permitted displacements in the nodal degrees of
freedom. Type IV specimens were not included in the analytical modelling.
The Finite Element models were calibrated against the experimental
observations by comparing torque-rotation histories (torsional stiffness and capacity)
and deformed shapes. In some instances, additional features of the failure
mechanism could be identified such as localised high tensile stress (evident as
visible local tensile failure of the material).
6.4.2 Deformed shapes
The deformed shapes obtained from the FE models match closely those
observed during the laboratory testing (Figure 6-18 through Figure 6-22 are typical).
0 holes (type I)
4634 elements (S4R)
Figure 6-18: Displaced shape (type I)
1 hole (type II)

10 mm diameter
3394 elements (S4R)
Figure 6-19: Displaced shape (type II 10 mm dia)
1 hole (type II)

30 mm diameter
3290 elements (S4R)
Figure 6-20: Displaced shape (type II 30 mm dia)
2 holes (type III)

20 mm diameter
3442 elements (S4R)
Figure 6-21: Displaced shape (type III 20 mm dia)
2 holes (type III)

30 mm diameter
3442 elements (S4R)
Figure 6-22: Displaced shape (type III 30 mm dia)
6.4.3 Torque-rotation relationships
The maximum torques predicted by the FE models and those measured in the
tests are listed in Table 6-6. The agreement between the FE predictions and the
measured values is extremely good with a maximum difference of only 3%, with FE
tending to under-predict the maximum torque in most cases. The good agreement is
true for the entire torque-rotation relationship (Figure 6-23, Figure 6-24 and Figure
6-25) with FE correctly modelling the onset of non-linearity, the maximum torque
and the plastic collapse. The largest difference occurred for the plastic collapse of
the unperforated specimens, which was particularly sensitive to the mesh density.
The correct prediction of the torque at yield and maximum torque for the
unperforated specimens and the specimens with small holes is in direct contrast to
the poor agreement obtained at full-scale (Section 5.5)
The initial elastic stiffnesses were difficult to measure accurately in the
laboratory due to limitations of the testing method, but the FE models were used to
obtain indirect measures (Table 6-7). The reduction in stiffness is slight for small
holes, but becomes important for holes larger than Φ = 0.44. Sections with two
holes are less stiff than sections with a single hole of corresponding diameter.
The FE predictions of elastic stiffness are significantly lower (up to 50%)
than the experimentally measured values. This is partly due to the low values of
Young’s modulus used in the FE material models. Young’s modulus is difficult to
measure accurately using a tensile test, particularly when the coupons are very thin.
Although the tensile tests indicated values of Young’s modulus in the order of 180
kN/mm2, when the deflection of a 5.5 m span of the cold-formed tube was measured
in bending tests, the Young’s modulus was calculated to be 214 kN/mm2. In the
absence of the opportunity to obtain more reliable material property data, the tensile
test results (with low E values) were used in the theoretical and FE models.
The boundary conditions in the FE models were found to be important for
unsymmetrical cases. Figure 6-26 shows how, for a single hole diameter, three FE
models can give very different results depending on the boundary conditions at the
ends of the specimen. When the boundary conditions are fixed in both transverse
directions at both ends (labelled ‘fixed’), the specimen is able to reach a higher value
of peak torque than if the boundary conditions are pinned (labelled ‘pinned’). Fixed
boundary conditions resist the tendency of the unsymmetrical specimens to bend at
high twist. Their influence is more significant for the small-scale specimens than
the full-scale torsion specimens (Section 5.6.8) because the values of twist are
higher and the specimens are shorter (in relation to section depth). The true
condition is that each end is pinned in one transverse direction and fixed in the other
and that the pinned axes at each end are perpendicular (labelled ‘cross-pinned’).
This condition represents the experimental situation, where the driving flats on each
specimen were initially mutually perpendicular (to prevent accidental transmission
of bending to the specimen). The cross-pinned condition reaches a maximum torque
between that of the pinned and fixed cases, but is able to maintain the torque with
increasing twist.
Table 6-6: Maximum torque (average experimental and FE)

Type Number of Hole diameter Maximum torque
specimens D (mm) Φ Av exp FE Difference
tested
(% of Tmax) (% of Tmax) (%)
I (control) 6 - - 102.7 104.3 1.52
II 2 5 0.15 100.6 101.5 0.84
(one hole) 2 10 0.30 94.60 94.90 0.30
2 15 0.44 81.87 83.61 1.75
2 20 0.59 67.87 67.76 -0.12
2 25 0.74 52.83 52.58 -0.25
2 30 0.89 46.88 45.04 -1.84
III 2 5 0.15 100.76 99.48 -1.28
(two holes) 2 10 0.30 93.06 90.13 -2.94
2 15 0.44 79.77 79.23 -0.53
2 20 0.59 64.72 63.05 -1.67
2 25 0.74 49.33 47.45 -1.88
2 30 0.89 35.55 33.75 -1.80
Note: Tmax is the theoretical capacity of the unperforated section (Table 6-5)
Table 6-7: Elastic stiffness from Finite Element (average over 240 mm length)
Finite Element elastic stiffness (% of theory unperforated)
Hole diameter (mm) 0 1 2 5 10 15 20 25 30
Hole size ratio, Φ 0 0.03 0.06 0.15 0.30 0.44 0.59 0.74 0.89
No holes (type I) 98.18 - - - - - - - -
Ono hole (type II) - 98.17 98.15 98.02 97.41 96.02 93.10 86.68 65.15
Two holes (type III) - 98.17 98.13 97.84 96.63 93.92 88.50 77.49 56.12
Torque / Torque at yield (Ty)

1.4 1.4
1.2 1.2
1.0 1.0
0.8 0.8
0.6 Type I 0.6 Type I
0.4 (No holes) 0.4 (No holes)
FE FE
0.2 Experimental 0.2 Experimental
0.0 0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 0 5 10 15 20 25
Rotation / Rotation at yield (θy) Rotation / Rotation at yield (θy)
Figure 6-23: FE and experimental for type I specimens (no holes)

1.4 Torque / Torque at yield (Ty) 1.4

1.2 1.2
1.0 1.0
0.8 0.8
0.6 Type II 0.6 Type II
0.4 (One hole) 0.4 (One hole)
FE FE
5 mm dia. 10 mm dia.
0.0 0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 0 1 2 3 4 5 6 7 8 9 10 11 12
1.2 1.0
1.0 0.8
0.8
0.6
0.6
Type II 0.4 Type II
0.4 (One hole) (One hole)
FE FE
0.2 15 mm dia. 0.2 20 mm dia.
Experimental Experimental
0.0 0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 0 1 2 3 4 5 6 7 8 9 10 11 12
0.8 0.8
0.7 0.7
0.6 0.6
0.5 0.5
0.4 0.4
Type II Type II
0.3 0.3
(One hole) (One hole)
0.2 FE 0.2 FE
0.0 0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 0 1 2 3 4 5 6 7 8 9 10 11 12
Figure 6-24: FE and Experimental for type II specimens (one hole)

1.4 1.4
1.2 1.2
1.0 1.0
0.8 0.8
0.6 Type III 0.6 Type III
0.4 (Two holes) 0.4 (Two holes)
FE FE
0.0 0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 0 1 2 3 4 5 6 7 8 9 10 11 12

1.2 1.0
1.0 0.8
0.8
0.6
0.6
Type III 0.4 Type III
0.4 (Two holes) (Two holes)
FE FE
0.2 15 mm dia. 0.2 20 mm dia.
Experimental Experimental
0.0 0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 0 1 2 3 4 5 6 7 8 9 10 11 12
0.8 0.8
0.7 0.7
0.6 0.6
0.5 0.5
0.4 0.4
Type III Type III
0.3 0.3
(Two holes) (Two holes)
0.2 FE 0.2 FE
0.0 0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 0 1 2 3 4 5 6 7 8 9 10 11 12
Figure 6-25: FE and experimental for type III specimens (two holes facing)
Small-scale torsion 1 hole 30 mm diameter
0.9
0.8
0.7
0.6
0.5
0.4
0.3 FE, fixed boundary conditions
0.2 FE, pinned boundary conditions

FE, crosspinned boundary conditions
0.1
Experimental
0
0 2 4 6 8 10 12
Figure 6-26: The influence of boundary conditions
6.5 Stress distributions
A number of FE models with an increased mesh density (up to 10500
elements) were used to study stress distributions. Figure 6-27 and Figure 6-28 show
typical surface distribution of von Mises stress around a small hole, and large hole
respectively. In the figures, the von Mises stress is plotted in terms of the stress
concentration factor for torsion (ψtorsion). The stress concentration factor for torsion
is defined as the ratio of the maximum von Mises stress in the unperforated zone, to
the maximum von Mises stress which would be present in an equivalent
unperforated beam at the same torque. The torque is such that the maximum von
Mises stress in the perforated zone is in the elastic range. The values obtained are
listed in Table 6-8.
The FE study showed that the stress distribution in the web of an RHS with
two holes is nearly identical to that in the web of an RHS with a single hole of equal
diameter. Even small holes caused stress concentrations, but the highest stresses
were generated by the largest holes. In all cases, in the elastic range, the disturbance
of the otherwise uniform stress flow caused by the hole was limited to a short
distance on either side of the hole. The locations of the highest stress in the FE
models coincided with the locations in the experimental specimens where the
metallurgical effects of high stress were most evident.
Table 6-8: Concentration factors for Von Mises stress

Stress concentration factor (ψtorsion)
Hole diameter (mm) 0 1 2 5 10 15 20 25 30
Hole size ratio, Φ 0 0.03 0.06 0.15 0.30 0.44 0.59 0.74 0.89
No holes (type I) 1.03 - - - - - - - -
Ono hole (type II) - 1.72 1.93 2.22 2.55 3.14 4.15 6.28 12.29
Two holes (type III) - 1.72 1.93 2.22 2.55 3.15 4.16 6.30 12.29
Figure 6-27: Concentration factors for Von Mises stress around a small hole
Figure 6-28: Concentration factors for Von Mises stress around a large hole
The small-scale experimental investigation showed the extent to which
circular openings in the walls of square-section steel tube can have a significant
influence on the torsional capacity and stiffness of the section.
A single-sided circular opening reduced the ultimate torsional capacity by as
much as 70%. A small opening (Φ less than 0.2) did not reduce ultimate capacity or
stiffness, but did reduce the extent of plastic deformation at peak torque. A larger
opening (Φ greater than 0.3) reduced significantly both ultimate capacity and
stiffness, and resulted in non-linear behaviour with asymmetry causing secondary
bending effects at high twist due to differential shear in the webs.
A double-sided circular opening was found to have a similar effect to that of
a single sided opening of equal diameter having a capacity slightly lower (0 to 10%).
The symmetry of the condition meant there were no bending effects at high twist
although there was a more rapid loss of post failure stiffness.
The sections with web openings behaved in a largely non-linear manner.
The presence of the hole(s) and associated stress concentrations constituted a
flexible zone about which the majority of twist occurred, and this was found to
govern the nature of the failure mechanism.
FE models were calibrated against the experimental observations by
comparing torque-rotation histories and deformed shapes. The deformed shapes
obtained from the FE models match closely those observed during the laboratory
testing, as do the torque-rotation relationships. The agreement between the FE
predictions and the measured values of maximum torque is extremely good with a
maximum difference of only 3%. This is particularly encouraging as it extends
confidence in the use of small-scale testing for studying a wider range of parameters
than can be investigated in expensive full-scale torsion tests.
The following Chapter, describes how the FE models, calibrated against the
experimental data obtained in the small and full-scale experimental studies, were
used in a parametric study in order to develop the preliminary design
recommendations for perforated RHS.
7 Parametric study and design recommendations
7.1 Introduction
This Chapter contains recommendations for the design of hot-finished steel
Rectangular Hollow Sections (RHS) with circular web openings under the actions of
torsion, bending and shear. Equations are given permitting the calculation of static
resistance, subject to specified limitations of web slenderness. The
recommendations are presented in a form that allows integration with modern design
codes and reference is made to parallel procedures in BS 5950-1:1990 and Eurocode
3 (EC3). The recommendations are justified by the findings of an analytical
parametric study (also presented in this Chapter) using Finite Element (FE) models
calibrated by laboratory testing (Chapters 3, 4, 5 and 6).
In the case of torsion, laboratory tests indicated lower resistances than those
predicted by the theory underlying the prevailing design procedures (Section 5.4).
Design recommendations based on current practice are provided here, as the
experimental results are not yet fully understood.
The actions of torsion, bending and shear are isolated before proposals are
made about the treatment of combinations of these actions. At each step, the
justification for each recommendation is outlined with reference to analytical and
experimental data. Finally, the design procedure is summarised, and the potential
for further development of design guidance is discussed.
The terminology and symbols have been selected to maintain consistency
with EC3, although the use of the decimal point for numerical values has been
retained. Some additional subscripts and symbols not found in EC3 are introduced
for clarity. Part 2 of Annex N in EC3 deals with the design of I-beams with
individual web openings and provided a useful model for a design procedure
concerned with web openings in RHS.
The 1993 version of EC3 refers to steel grades Fe 430 and Fe 510 in
accordance with BS EN 10025:1990. Here, use is made of the revised terminology
specified by BS EN 10025:1993 and more particularly BS EN 10210-1:1994, which
applies specifically to hot-finished hollow sections. Fortunately, for the purposes of
design, the equivalence between codes is elementary (Table 7-1).
Table 7-1: Equivalence of material grades for design

BS 4360:1990 BS EN 10025:1990 BS EN 10210-1:1994
Design grade 43 Fe 430 S275
Design grade 50 Fe 510 S355
7.2 Scope and limitations
The scope of the design advice is limited to tubes of constant wall thickness,
with circular holes placed at mid-depth in the webs. Two cases of perforated section
are considered, depending on whether just one web, or both webs contain holes
(Figure 1-8). These cases are termed ‘one hole’ and ‘two holes’ and, for the
purposes of design, are compared with unperforated sections (‘unperforated’ or ‘no
holes’). In the case of the doubly perforated RHS, the two holes are of equal
diameter, and are situated symmetrically with one in each web.
The topic of elastic instability is briefly discussed, but no attempt has been
made to provide rules for determining the limits of stability for use in design. It is
clear that such a system of stability classification is necessary, but there is
insufficient experimental evidence to form generalised conclusions from this
research. A study of stability limits would be, by necessity, more detailed than this
broader, and more fundamental, investigation would allow.
Perforations have the effect of concentrating deformations in one location
and therefore higher strains develop than in equivalent unperforated sections.
Because of this, strain hardening plays a much more important role in the
development of the ultimate resistance of a perforated section than an unperforated
section. In the absence of an opportunity to study the importance of strain hardening
in sufficient detail, the design recommendations here must remain conservative, and
neglect any strengthening effect that results from strains exceeding the yield stress
of the material.
These design recommendations apply principally to hot-finished steel RHS
of grades S275J2H and S355J2H since they are based largely upon experimental
work performed on those materials. Much of what follows is also applicable to
other materials, such as cold-formed steel and aluminium. However, no advice is
given regarding the particular properties of these materials, or how the behaviour of
RHS produced from them differs, in general, to hot-finished steel RHS.
Standard section sizes for hot-finished RHS are specified by BS EN 10210-
2:1997. The principal European producers roll the whole range, and commonly
include additional sizes based on other national and international standards.
Reference here is restricted to sections covered by the European Standard, but the
design recommendations proposed are applicable to any section size, provided the
limitations accompanying the rules are met.
BS EN 10210-2:1997 prescribes a standard nominal corner profile, which
relates the internal and external radii to the section thickness. Although most
producers roll their own corner profiles (within the permitted tolerances) the section
properties quoted in technical documentation are calculated from the nominal corner
profile (and the equations3 given in BS EN 10210-2:1997). For unperforated
sections, the difference is negligible. The corner profile has a small influence upon
the behaviour of RHS with large holes (Φ greater than 0.9) but the affect is too slight
to merit inclusion here. The nominal corner profile (or a close approximation) was
used in the parametric study.
BS EN 10210-2:1997 specifies a standard range of 267 different RHS, within
which are 24 different depths and eight different aspect ratios (Table 7-2). Sections
range from 20 mm to 500 mm in depth and from 2 mm to 20 mm in thickness. Web
slenderness ranges from 4.5 to 55.3. Of the standard sections, 180 (67%) have a
web slenderness less than or equal to 22.0, but the majority of these are at the small
end of the scale (150 have depths of 200 mm or less).
While it is possible to use the standard sections with the depth and width
transposed, thereby giving seven additional aspect ratios less than 1.0, it is
uncommon in practice. The scope of this document is limited to aspect ratios within
in the standard range (between 1.0 and 2.0).
3
The equations in BS EN 10210-2:1997 are also given in Appendix C.
Table 7-2: RHS in the European Standard range

Aspect ratio Number of Number of Number with web slenderness
different sections different depths less than or equal to 22.0
1.00 131 22 98
1.40 7 1 1
1.44 7 1 3
1.50 27 4 20
1.67 28 5 18
1.75 6 1 4
1.80 19 3 9
2.00 42 7 27
7.3 General design procedure
The size of an RHS can be described by three parameters. The common
method is to identify the depth, width and wall thickness (Figure 1-10). An
alternative method is also used here, which uses the three parameters that control the
structural behaviour: aspect ratio (α), section depth (h) and web slenderness (λweb).
As in the foregoing Chapters, a fourth parameter, the hole size ratio (Φ), is used to
indicate the size of the perforation. Aspect ratio is defined by Equation 7-1,
slenderness by Equation 1-1 and Equation 1-2, and hole size ratio by Equation 1-3.
α =h b
Equation 7-1
The design of an RHS with web perforations can be facilitated by
considering an idealised member consisting of a number of perforated and
unperforated sub-members. The sub-members can be designed individually using
the existing rules for unperforated sub-members, and the new recommendations for
perforated sub-members. Multiple perforations must be placed at sufficient
distances apart to ensure there is no detrimental interaction. The permissible spacing
of holes is based on the notion of zones of influence, which are discussed later in
this Chapter. This is the same approach as the one used in Annex N of EC3, which
deals with the design of I-section beams with web openings.
The design recommendation for bending is based partly on the strength limit
state approach used in Annex N of EC3. The EC3 approach was modified to include
a maximum stress component for when a yield based serviceability criterion is also
required (such as in BS 5950-1:1990).
A yield based serviceability approach is recommended for design in shear
and torsion. Web openings in RHS under the action of shear or torsion induce large
stresses and a serviceability approach based on a maximum stress criterion results in
a large reduction in design capacity. Unperforated RHS are naturally strong in both
shear and torsion and large reductions in design shear and torsion capacity may not
necessarily result in a requirement for a larger section. However, it is likely that a
strength limit state approach coupled with a displacement based serviceability limit
will result in a more economical method of design. Unfortunately, there was
insufficient time available to develop fully this approach within this project.
The stress based design recommendations presented in this Chapter are based
on the observation of stress distributions obtained from a parametric FE study. In
the following sections, the locations of raised stress will be referenced to the regions
labelled , and in Figure 7-2. The regions are:
The top surface of the upper flange above the opening and the bottom surface
of the lower flange below the opening
The edge of the opening at the top and bottom
The four locations where the edge of the opening meets the diagonals
In design, a factor of safety is normally included in the expression for design
resistance. In EC3 this safety factor is referred to as a ‘partial material factor’ and is
denoted by the symbol γM0. In the following discussions, where the design resistance
of an unperforated RHS or a perforated I-beam is mentioned, the reference is to the
value produced by the code expressions when the factor of safety is unity.
Zone of influence Zone of influence
Unperforated Unperforated Unperforated
Perforated Perforated
Figure 7-1: Perforated and unperforated sub-members
45o
x x y y
Figure 7-2: Locations for stress concentration
7.4 The parametric Finite Element study
The design recommendations presented in this Chapter were formulated with
regard to knowledge obtained from a parametric FE study. Perforated and
unperforated RHS of varying dimensions were modelled using the ABAQUS v5.8.1
FEA program (Section 2.3). Models included material and geometric non-linearity
and each contained between 3200 to 5700 four-noded shell elements (S4R). The
sensitivity of the solutions to meshes and element types was studied to ensure that
the findings of the parametric study were reliable. The FE models were calibrated
against the results of laboratory tests of perforated and unperforated RHS in torsion,
bending and shear (Chapters 3 to 6).
Three section sizes where chosen to investigate the full load-displacement
response (resistance and ductility) for each action (116 models in total). The three
sizes had the same web slenderness (22.0) as the sections used in the experimental
study and had aspect ratios (1.0, 1.5 and 2.0) representing the range of European
standard sizes (BS EN 10210-2:1997).
A standardised material model was used in the study (Table 7-3 and Table
7-4). The standardised material has typical yield plateau and strain hardening
behaviour (Section 2.2). The material fits the criteria for both S275 and S355
grades, being at the lower boundary of S355 and the upper boundary of S275. The
models included strain hardening because the parametric study was designed to
investigate the behaviour of actual perforated RHS in service. For purely design
purposes, it would have been more appropriate to use a simple elastic, perfectly
plastic material definition.
A larger number of models were displaced in the elastic range to study stress
distributions and concentrations over a wider range of section size (66 models for
bending, 84 models for shear and 73 models for torsion). The models contained the
full definition of the non-linear material used in the full plastic parametric study, but
the displacement steps were reduced to keep stresses within the elastic range.
Geometrical non-linearity was also accounted for, but its influence was insignificant
for such small deflections. Elastic parametric models contained more elements than
plastic parametric models to ensure reliable stress concentration data.
All the parametric models were based on the same generic mesh definition.
The boundary conditions (Table 7-5 and Figure 7-3) were selected to re-create the
typical conditions of RHS beams in service (RHS beams bolted between columns).
Actions were induced by controlling the displacements at the ends of the RHS.
Fixed and active boundary conditions were achieved by linking (via mathematical
relationships) the displacement of the cross-section at each end, to the displacement
of a single node.
Models used to study the action of bending (Figure 7-4a) were displaced at
the ends by rotations about the horizontal transverse axis. The rotations were equal
in magnitude, but opposite in direction, to produce a uniform bending moment over
whole length of the beam. Since the bending moment was uniform, no external
shear forces were produced. Models were long (2.5 m) to ensure that the end effects
did not interfere with the behaviour of the perforated zone.
Models used to study the action of shear (Figure 7-4b) were also displaced at
the ends by rotations about the horizontal transverse axis. The rotations were equal
in magnitude, and in direction, to produce a uniform shear force over whole length
of the beam, with zero bending moment in the middle (the perforated zone). Models
in the full plastic study were short (0.7 m) to ensure that bending failure did not
occur at the ends. However, the high shear resistance meant that only RHS with
large openings could be caused to fail in shear without any bending failure. In
addition, the short length meant that some interaction between the end effects and
the perforated zone was unavoidable. Elastic parametric models were longer in
length (2.0 m) and were used to study holes of all sizes without end effect
interaction or bending failure.
Models used to study the action of torsion (Figure 7-4c) were displaced at
the ends by rotations about the longitudinal axis. The rotations were equal in
magnitude, but opposite in direction, to produce a uniform torsional moment over
whole length of the beam. No external shear forces or bending moments were
applied. Models were long (2.0 m) to ensure that the end effects did not interfere
with the behaviour of the perforated zone.
Table 7-3: Material model for parametric study

fy1 E λplateau Eplateau λESHF fu εu ν
2 2 2 2
(N/mm ) (kN/mm ) (kN/mm ) (N/mm ) (%)
Material model for 360.0 205.0 13.0 0.0 0.20 510.0 16.0 0.3
parametric FE
Note 1: Typical material based on observation of tensile tests of steel from RHS (Appendix D)
Note 2: Refer to Section 2.2 for definitions of quantities
Table 7-4: ABAQUS material model for parametric study

*ELASTIC
205540.3 , 0.3
*PLASTIC
360.63 , 0
368.85 , 0.022493
403.30 , 0.028912
461.61 , 0.048132
530.81 , 0.085696
591.60 , 0.145542
624.24 , 0.199087
Table 7-5: Boundary conditions for parametric FE models

Model Axis 1 Axis 2 Axis 3
type (longitudinal) (horizontal transverse) (vertical transverse)
Translation Rotation Translation Rotation Translation Rotation
Bending End 1 Free Fixed Fixed Active, +θ Fixed Fixed
End 2 Fixed Fixed Fixed Active, -θ Fixed Fixed
Shear End 1 Free Fixed Fixed Active, +θ Fixed Fixed
End 2 Fixed Fixed Fixed Active, +θ Fixed Fixed
Torsion End 1 Free Active, +θ Fixed Fixed Fixed Fixed
End 2 Fixed Active, -θ Fixed Fixed Fixed Fixed
End 1 End 2
Axis 3
Axis 1
Axis 2
Figure 7-3: Model coordinate axes
(a) Bending
(b) Shear
(c) Torsion
Figure 7-4: Finite Element meshes for parametric investigation
7.5 Bending
7.5.1 Introduction
When designing a beam in bending using EC3 or BS 5950 (which both use
the limit state approach) the first step is the classification of the cross-section. The
section classification gives an indication of the expected performance (Figure 7-5) of
the beam with respect to the elements of the cross-section that are in compression
and hence determines the method of design.
Class 1 (plastic) cross-sections as those that can develop the plastic moment
of resistance (Mp) and can form a plastic hinge with a rotation capacity sufficient for
plastic design (allowing for the redistribution of moment within a structure). Class 1
sections have a long plateau of rotation, where the plastic hinge can form without
the occurrence of local buckling.
Class 2 (compact) cross-sections are those that can develop the plastic
moment of resistance (Mp), but have limited rotation capacity once this is reached
(due to local buckling), and are not suitable for plastic design.
Class 3 (semi-compact) cross-sections are those in which the calculated
stress at the extreme fibre of the member can reach its yield strength, but in which
local buckling prevents the development of the plastic moment of resistance (Mp).
Class 3 sections may only be designed in the elastic range up to the design yield
stress (elastic moment Me). The behaviour of class 3 sections does not necessarily
have to be linear up to the yield point.
Class 4 (slender) cross sections are those for which it is necessary to make
allowances for the effect of local buckling when determining the moment of
resistance. Class 4 sections can buckle locally before reaching the elastic moment
(Me).
The treatment of the various classifications of section in EC3 is similar to
that in BS 5950 with one significant exception. The design of class 4 sections in
EC3 proceeds by an effective area method rather than one of reduced stresses as in
BS 5950. The effective area method results in a more realistic and economical
design (Joannides et al 1993).
For sections of class 1 and 2 there is no requirement for further checking for
local buckling, although only class 1 sections may be used at plastic hinges.
Restrictions on use of class 3 and class 4 sections ensure that local buckling is
considered in design.
In practice, the classification of a section is determined by limits of web and
flange slenderness, with a modification for the grade of material (Clauses 5.3 of EC3
and 3.5 of BS 5950). A section can be described by the individual classifications of
the web and flange, or by the classification of the cross-section as a whole. For
bending, with the neutral axis at mid-depth, all the RHS in the European standard
range have class 1 webs for both S275 and S355 steel grades (EC3 or BS 5950
limits). The flange classification varies from class 1 to class 4.
Although the determination of class 1 cross-sections in design is determined
by geometric ratios, the actual definition of class 1 behaviour is based on rotation
capacity (R). Rotation capacity is defined by Equation 7-2 where κ is the curvature
at which the moment falls below Mp and κp is the curvature at which the plastic
moment is reached assuming an elastic flexural rigidity (Equation 7-3). Eurocode 3
requires R ≥ 3 for plastic design, but some standards require R ≥ 4, particularly those
covering seismic regions.
R = κ κ p −1
Equation 7-2
κ p = M p EI
Equation 7-3
The current web slenderness limits for RHS classification are based on tests
of I-sections where more work has been done on plastic design, but flanges in RHS
have different support conditions than those in I-sections. Flange slenderness limits
for RHS have been studied by a number of investigators (e.g. Dwyer and Galambos
(1965), Korol and Hudoba (1972), Hasan and Hancock (1988), and Zhao and
Hancock (1991)).
Recent research has resulted in cause to doubt the design code treatments of
RHS classification, and the way in which flanges and webs are treated in isolation
without consideration of interaction (Zhao and Hancock (1990) and (1992),
Wilkinson and Hancock (1997)). There has also been some cause to reflect on the
plastic performance of hot-finished sections, which might be considered to be
compromised by the yield plateau (Sedlacek et al (1998)). To add to the confusion
the Australian standard AS 4100 actually requires a sizeable yield plateau (λplateau ≥
6) as part of the material ductility requirement for plastic classification.
Once a section has been classified, its design moment of resistance (elastic or
plastic) is calculated using equations based on beam bending theory. To comply
with the ultimate limit state requirement, the factored load (EC3 refers to this as the
design load) must be less than the design moment of resistance (with a reduction
factor for safety). Both codes have two approaches: one for bending coexistent with
high shear, and one for bending coexistent with low shear load. Design for bending
about one axis with low shear load is dealt with in Clauses 4.2.5 of BS 5950 and
5.4.5.2 of EC3. The design of RHS with web openings under the action of shear and
bending is dealt with later in Section 7.8.
In EC3, the moment of resistance is calculated using Equation 7-4 for
sections of classes 1 and 2 and Equation 7-5 for sections of class 3. Elastic and
plastic section moduli are denoted by Wel,y and Wpl,y respectively, fy is the yield
strength (Table 3.1 of EC3) and γM0 is the material partial factor (a safety factor).
Class 4 sections are dealt with differently using area reduction factors. Section 7.5.3
contains analogous expressions that can be used to calculate the design moment of
resistance for RHS with web openings.
M c , y , Rd = M pl , y , Rd = W pl , y f y γ M 0
Equation 7-4
M c , y , Rd = M el , y , Rd = Wel , y f y γ M 0
Equation 7-5
For the serviceability limit state, the deflection of the beam must be limited
so as not to impair the strength or efficiency of the structure. Normally this is done
by checking that the calculated deflection is within the limiting values recommended
in the codes (Clauses 4.2 of EC3 and 2.5.1 of BS 5950). Deflection limits in BS
5950 are for imposed loads only, while EC3 considers deflections arising from both
dead and imposed loads. In BS 5950, the strength clause (Clause 4.2.5) prohibits
plasticity at working load so deflections can be calculated by elastic methods. There
is no such clause in EC3 and calculations of deflection must allow for plasticity.
Section 7.5.4 contains expressions that allow the calculation of the elastic deflection
under bending of RHS with web openings.
Moment
Class 1
Mp
Class 2
Me
Class 3 Class 1
definition
rotation
Class 4
capacity
Rotation
Figure 7-5: Cross-section classifications and moment-rotation relationship
7.5.2 Concentration factors for elastic stress
The distribution of elastic bending stresses around an opening is shown
schematically in Figure 7-6. Examples of FE contour plots of von Mises stress are
shown in Figure 7-7. The stress distribution in the web of an RHS with one hole is
nearly identical to that in the web of an RHS with two holes. Small differences
occur due to asymmetry, but are too slight to merit discussion here.
An RHS without an opening (Figure 7-6a) will have a linear distribution of
stress as predicted by Euler-Bernoulli-Parent beam bending theory. Under a
uniform bending moment, the cross-sections at all points along the beam remain
plane. The bending stress in a longitudinal fibre is proportional to the perpendicular
distance from the neural axis. Yield occurs when the stress in the extreme fibres
reaches the yield stress (fy).
A small circular opening on the neutral axis (Figure 7-6b) will disturb the
uniform system of stress, causing longitudinal warping of the cross-section (planes
no longer remain plane). Longitudinal stress close to the neutral axis is interrupted
by the opening and must flow around it, raising the stress above and below the
opening (location in Figure 7-2). Longitudinal stress closer to the flanges is less
disturbed by the flow of stress around the hole and bending stress distant from the
neutral axis returns to the same linear distribution of stress present in the
unperforated beam, remote from the opening. Since the highest stresses are present
at the extreme fibres, yield occurs simultaneously with yield in the unperforated
beam. The higher stresses around the opening compensate for the reduced second
moment of area and the moment at yield is not reduced by the presence of the hole.
An opening of a particular size (Figure 7-6c) will raise the stress above and
below the opening (location in Figure 7-2), to the same magnitude as that at the
extreme fibres (location in Figure 7-2). This size of hole is, here, termed the
‘critical hole size’, and marks the boundary between small holes that do not reduce
the moment at yield and large holes that do.
Holes larger than the critical hole size (Figure 7-6d) raise the stress above
and below the opening to values larger than the bending stress at the extreme fibre of
the unperforated beam distant from the opening. Stress in the flanges immediately
above and below the hole is also raised by the proximity to the hole (the reduction in
second moment of area), but yield occurs in the webs first and the moment at yield
is reduced.
A beam with a large opening that does not produce a stress concentration
effect (such as a very long elongated hole with a gradual change in cross-section)
has a linear stress distribution (Figure 7-6e). The stress distribution at the opening is
analogous to that in the unperforated section, but higher stresses (in all longitudinal
fibres) will result for a particular bending moment because of the reduced second
moment of area. A beam with a circular hole large enough to remove the entire web
at the central cross-section behaves in a similar way, because the flow of
concentrated stress is confined to the webs (although stress in the corner radii
becomes very high).
A number of FE models (66) have been used to study the stress
concentrations around openings in the webs of RHS in pure bending. The models
used in the parametric study are shown diagrammatically in Figure 7-8. Stress
concentrations were found to be dependent on hole size and web slenderness, but not
aspect ratio. The web may be considered as a rectangular beam acting as an element
of the whole beam. The dependence of stress concentration on the web slenderness
is analogous to the dependence upon thickness, of the stress concentration around a
circular hole in an infinite strip in bending (as considered by investigators such as
Isida (1952) and Peterson (1974)).
The FE models were used to obtain a representative sample of stress
concentration factors for bending. The stress concentration factor for bending,
Ψbending, is defined as the ratio of the maximum von Mises stress in the perforated
beam (at any location) against the maximum von Mises stress in the unperforated
beam. The two maximum stresses are compared at the same bending moment and
the moment is such that both maximum stresses are in the elastic range.
The stress concentration factors obtained from the FE study were
approximated by an empirical mathematical function, which was fitted to the data
using a least squares method (Equation 7-6). For the 66 models included in the
study, the maximum differences (in absolute terms) between the empirical function
and the FE values were 0.14 (empirical value less than FE value) and 0.07
(empirical value greater than FE value). The square root of the average square error
is 0.04. The actual stress concentration data from the FE study is presented in
Appendix E.
 λ2web λ web  λ2 λ 
ψ bending = Maximum − + + 1.5054  Φ + web − web − 0.3288 , 1
 439.2 5.916  1200 14.16 
Equation 7-6
The empirical function for stress concentration factor is shown graphically in
Figure 7-9. Stress concentration factors increase with increasing hole size and are
larger for more slender webs. The function is valid only for hole size ratios between
0 and 0.9 and for web slendernesses between 9.0 and 32.0. The function actually
reduces slightly at the extreme limit of slenderness (λweb > 29.0): this is a
consequence of its empirical nature and is not indicative of true behaviour.
Holes smaller than the critical hole size ratio (see above) have stress
concentration factors of unity. The critical hole size is represented by the
intersection of the stress concentration function with the plane Ψbending = 1 (Figure
7-10).
(a) +fy (b) +fy (c) +fy (d) (e) +fy

+fy
+fy
-fy
-fy
-fy -fy -fy -fy
z z z z z
(a) (b) (c) (d) (e)
y y y y y y y y y y
z z z z z
Figure 7-6: Distribution of elastic bending stress
RHS 250x250x10 RHS 250x250x10

Two holes Φ = 0.0 Two holes Φ = 0.1
Bending Bending
Moment = 179 kNm Moment = 179 kNm
RHS 250x250x10 RHS 250x250x10

Bending Bending
RHS 250x250x10 RHS 250x250x10

Bending Bending
Figure 7-7: Distribution of von Mises stress for RHS 250x250x10 in bending
Von Mises stress plotted on external surface

(Original in colour)
50
45
40
Web slenderness, λ web
35
Aspect ratio 1.00
30
Aspect ratio 1.50
25 Aspect ratio 1.67
Aspect ratio 1.75
20
Aspect ratio 2.00
15
10
5
critical hole size
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Hole size ratio, Φ
Figure 7-8: Parametric sampling for bending
2.4
2.2
2.0
1.8 Stress 2.2-2.4

concentration factor 2-2.2
1.6 ψ bending
1.8-2
1.4 1.6-1.8
1.2 1.4-1.6
29 1.2-1.4
1.0
22 1-1.2
0.90
0.80
Web slenderness 15
0.70
λ web
Hole size ratio
0.60
8
Φ
0.50
Figure 7-9: Stress concentration factors for bending
0.69
Computed
0.68
Proposed design values
0.67
Critical hole size ratio, Φ cr
0.66
0.65
0.64
0.63
0.62
0.61
0.60
0.59
0.58
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32
Web slenderness , λ web
Figure 7-10: Critical hole size ratio for bending
7.5.3 Resistance
While it is clear that a system of stability classification is necessary for
design, there is insufficient experimental evidence to form generalised conclusions
from this research. However, a number of preliminary observations can be made
based on this study and on parallel design procedures for I-sections with web
openings. The proposed expressions for calculation of resistance which follow, are
based on the premise that elastic instability will not occur.
Class 1 sections differ from class 2 sections by having sufficient rotation
capacity for plastic design. However, plastic design is not practicable for perforated
sections (no flame cut edges or punched holes: Clause 7.3 of EC3). The critical
requirements for classification therefore lie at the boundary between class 2 and
class 3 behaviour, which governs the design resistance (elastic or plastic), and the
boundary between class 3 and class 4 behaviour, which marks the point where
elastic stability must be considered.
Annex N of EC3 deals with the design of I-section beams with web openings
and specifies that the flange of the unperforated section must be of class 1 or class 2
(Clause N.1.7.1). It is recommended that this specification be adopted for RHS, as
class 3 flanges will be susceptible to flange buckling. Clause N.1.7.1 also specifies
that the web should be of class 1, 2, or 3, but this requirement is not critical for RHS
where webs are of class 1 for the entire standard European range.
Clause N.2.1.4 of Annex N deals with the calculation of design resistance,
and gives two approximate methods. The first method is for sections of class 1 and
class 2, and is based on the plastic capacity. Equation N.10b in the code is based on
the plastic section modulus of the reduced section (for RHS: Equation 7-7 and
Equation 7-8). The analogous expressions for the plastic moment of resistance of
perforated RHS are Equation 7-9 (for one hole) and Equation 7-10 (for two holes).
W pl , y ,1hole = W pl , y ,unperf − tr 2
Equation 7-7
W pl , y , 2holes = W pl , y ,unperf − 2tr 2

Equation 7-8
 tr 2 
M pl , y , Rd ,1hole = M pl , y , Rd ,unperf 1 − 
 W 
 pl , y ,unperf 
Equation 7-9
 2tr 2 
M pl , y , Rd , 2holes 
= M pl , y , Rd ,unperf 1 − 
 W 
Equation 7-10
The second method in clause N.2.1.4 of Annex N is for class 3 sections and
is based on the elastic capacity. Equation N.10d in the code is based on the elastic
section modulus of the reduced section. However, the examination of stress
distributions (Section 7.5.2) has shown that holes smaller than the critical hole size
ratio (Φcr) do not reduce the elastic bending capacity (hence Equation 7-11).
For RHS with holes larger than the critical hole size, the reduced section
expression produces a more economical design than the maximum stress approach.
The capacity can be reduced by an appropriate proportion towards the fully
perforated capacity (when Φ = 1, Figure 7-6e) as shown diagrammatically in Figure
7-11. The mathematical expressions are Equation 7-12 (for one hole) and Equation
7-13 (for two holes).
M el , y , Rd ,1hole or M el , y , Rd , 2holes = M el , y , Rd ,unperf = f yWel , y ,unperf

For Φ ≤ Φ cr
Equation 7-11
  t (h − 3t )    Φ − Φ cr 
3
M el , y , Rd ,1hole = M el , y , Rd ,unperf −  M el , y , Rd ,unperf − f y Wel , y ,unperf −  
 1− Φ

 6 h
   cr 
For Φ > Φ cr
Equation 7-12
  t (h − 3t )    Φ − Φ cr 
3
M el , y , Rd , 2holes = M el , y , Rd ,unperf  
− M el , y , Rd ,unperf − f y Wel , y ,unperf −   
 3h  1− Φ
   cr 
For Φ > Φ cr
Equation 7-13
The expressions for elastic moment above do not predict the moment at
which first yield occurs. Actual material yield occurs at a lower moment due to
stress concentrations at the opening. However, the area of yielded material is small,
and does not influence the behaviour of the beam as a whole. In this instance, the
elastic moment defines the extent of the linear region of the behaviour of the whole
beam.
A section without strain hardening is incapable of achieving a moment
greater than its plastic moment of resistance. For an unperforated hot-finished
section, the strains required for strain hardening are large and correspond to
enormous curvatures (hence Sedlacek et al (1998)). However, the presence of the
opening in a perforated RHS induces high strains by focusing displacements in the
flexible zone. Consequently, the strain hardening behaviour has a significant role in
the plastic performance.
Although strain hardening is not considered in design (here or in Annex N),
it has, in practice, a significant effect on the plastic capacity of a perforated section.
Strain hardening increases the ductility of the section allowing it to reach a higher
maximum moment, as is shown by example in Figure 7-12. In the figure, the
moment has been reduced to a dimensionless ratio against the theoretical elastic
moment of the unperforated section. Curvature has been reduced to a dimensionless
ratio against the theoretical curvature of the unperforated section at yield. The
curvature is an average measure made along the whole beam and is therefore
dependent on the length.
For sections of classes 1 and 2, the maximum moment is limited by the
inelastic stability of the elements of the cross-section in compression. The principal
buckling mechanisms are arching of the compression cee (Figure 7-13a) and
buckling of the compression flange (Figure 7-13b). Flange buckling occurs for
squarer RHS where the flange slenderness is greater and is similar to flange
buckling in unperforated sections. Larger holes reduce the stabilising influence of
the web and induce flange buckling at lower curvatures. Arching of the
compression cee occurs for more oblong RHS with larger holes. Both mechanisms
of plastic collapse were observed in the experimental study (Chapter 3). Sections
with single holes also experience arching (Figure 7-14a), and this can induce
sympathetic buckling of the unperforated web (Figure 7-14b).
The moment-curvature relationships for the three sizes of RHS used in the
plastic parametric study are shown in Figure 7-15 through Figure 7-20. Moment and
curvature are reduced to non-dimensional ratios in the same way as in Figure 7-12.
The reductions in the elastic capacity (global rather than first yield) due to
perforations are small (less than 25%) as the webs contribute a minor part of the
elastic section modulus. Small holes remove small amounts of material close to the
neutral axis and the reduction in elastic capacity is correspondingly small. The
elastic capacities of the three sections as determined by FE are plotted in Figure 7-21
and Figure 7-22. The elastic capacities as determined by the design expressions
(Equation 7-12 and Equation 7-13) are also plotted. The FE elastic capacities are, in
all cases, greater than the design expressions predict.
The maximum moments achieved by the three sections as determined by FE
are plotted in Figure 7-23 and Figure 7-24. The plastic moments of resistance as
determined by the design expressions (Equation 7-9 and Equation 7-10) are also
plotted.
All three RHS chosen for the full plastic study were capable of generating
the plastic moment of resistance and, therefore, can be described as class 2. The
three sizes have the same web slenderness (22.0), and since they have aspect ratios
at both ends of the standard European range, indicate that all standard sections with
web slendernesses less than 22.0 will also be class 2.
Strain hardening has the effect of increasing the maximum capacity moment
beyond the design values and FE models without strain hardening in the material
model have maximum moments equal to the design values (Figure 7-25).
Elastic moment capacity
Capacity of unperforated RHS

Proposed design expression
Reduced elastic section

modulus approach
Capacity of fully perforated RHS, Φ = 1
Hole size ratio, Φ

0 Φcr 1
Figure 7-11: Elastic moment capacity
RHS 200x100x8 two holes parametric bending (FE)
1.4
Plastic moment (theory, unperforated)

1.2
1.0 Φ=0.4
0.8 Φ=0.8 Φ=0.7
0.6
0.4
Φ = 0.4, 0.7, 0.8 With strain hardening
0.2
Without strain hardening
0.0
0 1 2 3 4 5 6 7 8
Figure 7-12: The significance of strain hardening for beam ductility
(a) (b)
(c) (d)
Figure 7-13: Distortions under the action of bending (two holes)
(a) (b)
Figure 7-14: Distortions under the action of bending (one hole)
RHS 250x250x10 one hole parametric bending (FE)
1.4
1.2
1.0
Φ=0.9 Φ=0.7 Φ=0.0 & 0.1
0.8
Φ=0.5
0.6 Φ=0.3
0.4
Φ = 0.0, 0.1, 0.3, 0.5, 0.7, 0.9
0.2
0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Figure 7-15: Moment-curvature relationship for RHS 250x250x10 one hole
1.4
1.2
Φ=0.2
1.0 Φ=0.4 Φ=0.3
Φ=0.5
Φ=0.6 Φ=0.7
0.8 Φ=0.8 Φ=0.0 & 0.1
Φ=0.9
0.6
0.4
Φ = 0.0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9
0.2
0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Figure 7-16: Moment-curvature relationship for RHS 250x250x10 two holes
1.4

1.2
Φ=0.0, 0.1, 0.3, 0.5, 0.7
1.0
Φ=0.9
0.8
0.6
0.4
Φ = 0.0, 0.1, 0.3, 0.5, 0.7, 0.9
0.2
0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
1.4

1.2 Φ=0.0, 0.1, 0.3, 0.5
Φ=0.7
1.0
0.8
Φ=0.9
0.6
0.4
Φ = 0.0, 0.1, 0.3, 0.5, 0.7, 0.9
0.2
0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
1.4

1.2 Φ=0.0, 0.1, 0.3, 0.5, 0.7
Φ=0.9
1.0
0.8
0.6
0.4
Φ = 0.0, 0.1, 0.3, 0.5, 0.7, 0.9
0.2
0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
1.4

1.2 Φ=0.0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7
Φ=0.8
1.0
0.8
0.6 Φ=0.9
0.4
Φ = 0.0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9
0.2
0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Yield is the moment at which the structural response is no longer linear -

1.02
Moment at yield / Moment at yield (theory, unperforated) not the moment when material yield first occurs.
1.00
0.98 RHS 250x250x10 (FE)
0.96 RHS 300x200x12 (FE)

RHS 200x100x8 (FE)
0.94 RHS 250x250x10 (design expression)
RHS 200x100x8 (design expression)
0.90
0.88
0.86
0.84
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Hole size ratio, Φ
Figure 7-21: Yield in bending for sections with one hole (FE and design)
1.05 Yield is the moment at which the structural response is no longer linear -
Moment at yield / Moment at yield (theory, unperforated)
not the moment when material yield first occurs.
1.00
RHS 250x250x10 (FE)

0.95
RHS 300x200x12 (FE)
RHS 200x100x8 (FE)
0.90
0.85
0.80
0.75
0.70
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Hole size ratio, Φ
Figure 7-22: Yield in bending for sections with two holes (FE and design)
Max moment / Plastic moment (theory, unperforated) 1.05

1.00
0.95
0.90
0.85
0.80
0.75
RHS 250x250x10 (FE)
0.70
RHS 300x200x12 (FE)
0.65 RHS 200x100x8 (FE)
0.55
0.50
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Hole size ratio, Φ
Figure 7-23: Maximum moment for sections with one hole (FE and design)
1.05
Max moment / Plastic moment (theory, unperforated)
1.00
0.95
0.90
0.85
0.80
0.75
RHS 250x250x10 (FE)
0.70
RHS 300x200x12 (FE)
0.65 RHS 200x100x8 (FE)
0.55
0.50
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Hole size ratio, Φ
Figure 7-24: Maximum moment for sections with two holes (FE and design)
Max moment / Plastic moment (theory, unperforated) 1.05

1.00
0.95
0.90
0.85
0.80 RHS 200x100x8 (FE, with strain hardening)
0.75 RHS 200x100x8 (FE, no strain hardening)
0.65
0.60
0.55
0.50
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Hole size ratio, Φ
Figure 7-25: Maximum moment without strain hardening
7.5.4 Deflection
Clause N.2.3 of Annex N in EC3 deals with the deflection of I-beams with
single perforations. The clause states that the deflection should be calculated from
the overall bending deformation of the unperforated beam and the additional
deformation of the perforated part, but does not provide a method by which this
might be achieved. The requirements of clause N.2.3 when transposed to the
bending deformation of a perforated RHS are:
(1) The additional deformation of the perforated beam should be determined taking
into account:
- The effect of global bending on the overall deformation of the perforated
cross-section
- The effect of localised bending deformation of the top and bottom cees
(2) The deformations due to the presence of the web opening should be analysed
taking slope compatibility at the ends of the opening into account
In the elastic range, the displaced shape of a perforated RHS in bending can be
calculated as follows:
(1) By treating the perforated zone as a sub-member of one hole diameter in length
with a reduction in the stiffness proportional to the reduction in the second
moment of area at the central cross-section
(2) By taking slope compatibility at the ends of the perforated zone into account
The displaced shape calculated by this method is a good match for the
displaced shape produced by FE (e.g. Figure 7-26) but the resulting deflections are
only slightly larger than for an unperforated beam (Table 7-6). Short beams and
beams with multiple perforated zones are most affected, but the additional deflection
for a 2.5 m beam with a single perforated zone is less than 5%.
Table 7-6: Deflection in bending

Section Hole size One hole Two holes
ratio, Φ Deflection (% of unperforated) Deflection (% of unperforated)
FE Calculation FE Calculation
RHS 250x250x10 0.1 99.98 100.00 99.97 100.00
RHS 250x250x10 0.2 - - 99.98 100.01
RHS 250x250x10 0.3 99.99 100.01 99.99 100.03
RHS 250x250x10 0.4 - - 100.02 100.09
RHS 250x250x10 0.5 100.05 100.11 100.11 100.22
RHS 250x250x10 0.6 - - 100.27 100.46
RHS 250x250x10 0.7 100.27 100.42 100.65 100.88
RHS 250x250x10 0.8 - - 101.35 101.55
RHS 250x250x10 0.9 101.21 101.20 102.64 102.60
RHS 200x100x8 0.1 99.98 100.00 99.96 100.00
RHS 200x100x8 0.2 - - 99.97 100.01
RHS 200x100x8 0.3 99.99 100.02 99.99 100.04
RHS 200x100x8 0.4 - - 100.06 100.12
RHS 200x100x8 0.5 100.52 100.15 100.21 100.30
RHS 200x100x8 0.6 - - 100.52 100.65
RHS 200x100x8 0.7 102.06 100.59 101.12 101.26
RHS 200x100x8 0.8 - - 102.22 102.29
RHS 200x100x8 0.9 102.24 101.72 104.33 104.01
RHS 300x200x12 0.1 99.98 100.00 99.96 100.00
RHS 300x200x12 0.3 99.99 100.02 99.98 100.05
RHS 300x200x12 0.5 100.10 100.18 100.19 100.37
RHS 300x200x12 0.7 100.48 100.70 101.10 101.48
RHS 300x200x12 0.9 102.06 102.01 102.88 104.53
Note: Elastic defection of a 2.5 m beam in pure bending
120%
(percent of central displacement of unperforated)
100%
Vertical displacement
80%
60%
40%
Finite Element RHS 200x100x8

20% Two holes Φ = 0.8
Design recommendation Bending
0%
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Longitudinal position
(percent of span)
Figure 7-26: Bending displacement of RHS 200x100x8 two holes 140.8 mm dia
7.5.5 Zone of influence
The zone of influence is the length of the RHS over which the web opening
influences the stress distribution, and hence marks the boundary between the
perforated and unperforated zones (Figure 7-1). In the elastic range, the zone of
influence for bending has been observed to extend a distance in the order of one hole
diameter on either side of the edge of the hole. Consequently, the zone is
approximately three diameters in length (e.g. Figure 7-7).
Openings placed in the same web at sufficient separations, such that the
zones of influence do not overlap, will not cause the stress distributions to interact,
and hence will not cause a further reduction in resistance (provided there is no
instability e.g. web post buckling).
In Clause N.2.1.1 of Annex N (EC3) the allowable clear width between
circular holes in I-beams is dependent on the shear force but the minimum value is
one and a half diameters.
7.5.6 Design recommendation for calculation of resistance
It has been observed that RHS with web slenderness less than or equal to
22.0 can be treated as class 2 sections. For the present, it is recommended that a
conservative approach be adopted for perforated RHS with more slender webs:
treating them as class 3 sections. Very slender sections are likely to exhibit class 4
behaviour. Since perforated sections are not practicable for plastic design, there is
no requirement for a class 1 definition.
If all plasticity is to be avoided, the elastic bending capacity can be
calculated using Equation 7-14, where γMBen1 and γMBen2 are possible partial safety
factors. The stress concentration factor for bending (ψbending) can be found from
Table 7-7. The expression is valid for hole size ratios between 0 and 0.9 and for
web slendernesses between 10.0 and 34.0, but does not allow for class 4 behaviour.
M el , y , Rd ,unperf
M el,y , Rd,1hole or M el,y , Rd,2holes =
(ψ bending + γ MBen1 ) × γ MBen2
Equation 7-14
Alternatively, the design elastic moment for sections with holes larger than
the critical size (Table 7-8) can be calculated using Equation 7-15 for sections with
one hole, and Equation 7-16 for sections with two holes, where γM0 is a possible
partial safety factor. The design elastic moment for sections with holes smaller than
the critical hole size is equal to the design elastic moment of the unperforated
section.
  t (h − 3t )    Φ − Φ cr 
3
M el , y , Rd ,unperf  
 6 h  1 − Φ
   cr 
M el , y , Rd ,1hole =
γ M0
Equation 7-15
  t (h − 3t )    Φ − Φ cr 
3
M el , y , Rd ,unperf  
 3h  1 − Φ
   cr 
M el , y , Rd , 2holes =
γ M0
Equation 7-16
The design plastic moment of can be calculated using Equation 7-17 for
sections with one hole, and Equation 7-18 for sections with two holes, where γM0 is a
possible partial safety factor.
M pl , y , Rd ,unperf  tr 2 
M pl , y , Rd ,1hole =  1− 
γ M0  W 
Equation 7-17
M pl , y , Rd ,unperf  2 
M pl , y , Rd , 2holes = 1 − 2tr 
γ M0  W 
Equation 7-18
Table 7-7: Stress concentration factors for bending, ψbending

Hole size Web slenderness, λweb
ratio, Φ 10 14 18 22 26 30 34
<=0.575 1.00 1.00 1.00 1.00 1.00 1.00 1.00
0.600 1.00 1.00 1.00 1.00 1.01 1.02 1.01
0.625 1.00 1.00 1.05 1.10 1.12 1.13 1.12
0.650 1.00 1.07 1.15 1.20 1.23 1.24 1.24
0.675 1.05 1.16 1.24 1.30 1.34 1.36 1.35
0.700 1.13 1.24 1.34 1.41 1.45 1.47 1.47
0.725 1.20 1.33 1.43 1.51 1.56 1.58 1.58
0.750 1.27 1.41 1.53 1.61 1.67 1.70 1.70
0.775 1.35 1.50 1.62 1.72 1.78 1.81 1.81
0.800 1.42 1.59 1.72 1.82 1.89 1.92 1.93
0.825 1.50 1.67 1.81 1.92 2.00 2.04 2.04
0.850 1.57 1.76 1.91 2.02 2.10 2.15 2.16
0.875 1.65 1.84 2.00 2.13 2.21 2.26 2.28
0.900 1.72 1.93 2.10 2.23 2.32 2.38 2.39
Note: Values obtained from Equation 7-6
Table 7-8: Critical hole size ratio for bending, Φcr

Web slenderness, λweb 10-11 11-12 12-14 14-16 16-18 18-23 23-32
Critical hole size ratio, Φcr 0.65 0.64 0.63 0.62 0.61 0.60 0.59
Note: Values obtained from Figure 7-10
7.6 Shear
7.6.1 Introduction
Clauses 5.5.1 of EC3 and 4.2.3 of BS 5950 deal with the shear resistance of
unperforated beams. The design value of shear force must be less than the plastic
shear resistance (Vpl,Rd) defined by Equation 7-19.
Av f y
V pl , Rd =
3 ×γ M 0
Equation 7-19
The plastic shear resistance is the product of the shear stress at yield (0.6fy in
BS 5950) and the shear area (Av) modified by a material partial safety factor (γM0).
The shear area is defined by Equation 7-20 (Table 5.16 in EC3 and Clause 4.2.3d of
BS 5950), where A is the total area of the cross-section. The shear area is therefore
the area of the flat part of the webs (that used to calculate web slenderness and hole
size ratio) and a portion of area of the radii.
Av = Ah (b + h )
Equation 7-20
RHS are much stronger in shear than other types of section as they have two
relatively thick webs. In practice, the shear resistance is so high that it is rarely
critical for design, as can be illustrated by the following generalised example.
A cantilever loaded at the unsupported end by a point load is subject to
uniform shear force and has a maximum bending moment at the supported end, that
is equal to the product of the load and span. A critical cantilever length may be
defined as that at which the shear resistance is equal to the shear force and the elastic
bending resistance is equal to the maximum bending moment. For an RHS, this
critical length (lcr) is given by Equation 7-21. This expression is based on a simple
calculation that assumes the cross-section remains plane regardless of the high shear
forces and short span.
l cr
=
(
3 bh 3 − (b − 2t )(h − 2t )
3
)
h 12t (h − t ) h 2
Equation 7-21
The critical cantilever length for the sizes of RHS in the standard European
range (BS EN 10210) is shown in Figure 7-27. The critical length increases with
increasing slenderness, but reaches a limit (Equation 7-22) which is dependent only
on aspect ratio. Any beam under any system of loads with a span longer than twice
the critical cantilever length can never be made to fail in shear as it will fail in
bending first. Since the maximum critical length (for α = 1) is only 1.16 times the
section depth, the shear resistance is rarely critical in practice.
 l cr  0.866
  = + 0.289
 h  max α
Equation 7-22
1.2
1.1 α = 1.00
Critical length / Section depth
1.0
0.9 α = 1.40
α = 1.44
0.8 α = 1.50
α = 1.67
α = 1.75
0.7
α = 1.80
α = 2.00
0.6
0.5
0 10 20 30 40 50 60 70
Web slenderness
Figure 7-27: Critical cantilever length for shear
An RHS without an opening in the web in pure shear will have ‘top hat’
distribution of elastic shear stress (Figure 7-28) similar in shape to the distribution of
shear stress in an I-section. A small circular opening on the neutral axis will disturb
the system of stress where it is most uniform, and produce a stress pattern similar to
that around a hole in an infinite plate in pure shear. Stress is raised at four points
around the edge of the hole (location in Figure 7-2). Larger holes raise the stress
further, as less web is available to transmit the shear across the width of the hole.
concentrations around openings in the webs of RHS in shear with zero bending
moment at the location of the hole. The models used in the parametric study are
shown diagrammatically in Figure 7-29. Stress concentrations were found to be
dependent on hole size and web slenderness, but not aspect ratio. The dependence
of stress concentration on the web slenderness is analogous to the dependence upon
thickness of the stress concentration around a circular hole in an infinite plate in
pure shear. The FE contour plots of von Mises stress shown in Figure 7-30 are
typical of the stress patterns observed.
concentration factors for shear. The stress concentration factor for shear, Ψshear, is
defined by Equation 7-23 and is the ratio of the plastic shear resistance of the
unperforated beam to the shear force at first yield for the perforated beam. The
shear force (V) is such that the maximum von Mises stress in the perforated RHS
(σMises,Max) is in the elastic range.
σ Mises , Max
ψ shear = Av
3V
Equation 7-23
The stress distribution in the web of an RHS with one hole is similar to that
in the web of an RHS with two holes, although maximum stresses are slightly lower,
particularly for large holes. To simplify the parametric study, a conservative
approach was used and stress concentration factors were obtained for sections with
two holes, to be applied also to sections with one hole.
using a least squares method (Equation 7-24). For the models (with two holes)
included in the study, the maximum differences (in absolute terms) between the
empirical function and the FE values were 0.76 (empirical value less than FE value)
and 2.01 (empirical value greater than FE value). The square root of the average
square error was 0.44. The actual stress concentration data from the FE study is
presented in Appendix E.
ψ shear = 
 1 
(
− 1 0.02705λ web − 1.149Φ 2 + 0.152Φ + 1.573 + 2 )
1− Φ 
Equation 7-24
larger for more slender webs. The function is valid for hole size ratios between 0
and 0.9 and for web slendernesses between 9.5 and 47.0.
140%
'Average shear stress' = Shear force (V ) / Shear area (A v ) RHS 250x250x10
RHS 200x100x8
120%
RHS 300x200x12
Shear stress (% of average)
100%
Shear stress z
80%
60%
y y
40%
20%
z
0%
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Position (% of section depth)
Figure 7-28: Distribution of elastic shear stress
50
45
40
35
30
Aspect ratio 1.00
Aspect ratio 2.00
20
15
10
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Hole size ratio, Φ
Figure 7-29: Parametric sampling for shear
(a)
(b)
RHS 200x100x8
One hole Φ = 0.7
Shear
Shear force = 64.9 kN
Figure 7-30: Distribution of von Mises stress for RHS 200x100x8 in shear

22.0
20.0
18.0
20-22
16.0
18-20
14.0 Stress
16-18
12.0 concentration factor
10.0 ψ shear 14-16
8.0 12-14
6.0 10-12
4.0 8-10
42
2.0 6-8
31
0.90
4-6
0.70
Web slenderness 20
2-4
0.50
λ web
Hole size ratio
0.30
9
Φ
0.10
Figure 7-31: Stress concentration factors for shear
7.6.3 Resistance
In BS 5950, clause 4.2.4 states that the shear stress in sections with web
openings significantly larger than those required for fasteners should be calculated
from first principles assuming elastic behaviour. The clause imposes a maximum
limit on shear stress that is higher (17%) than the design shear yield stress.
In EC3, the shear resistance of I-beams with individual web openings is dealt
with in clause N.2.1.3 of Annex N. The clause provides an equation (N.7c) for
calculation of the plastic shear resistance of I-beams with circular web openings that
is based on the reduced shear area. The clause also provides an equation for the
calculation of shear buckling resistance (N.8b).
The stress concentration factors determined in the previous section can be
used to determine the shear resistance based on a maximum stress criterion (as in BS
5950). If stress is limited to the design yield stress, the yield based shear resistance
of the perforated section can be determined by dividing the plastic shear resistance
of the unperforated section by the stress concentration factor (Equation 7-25).
Vel,y , Rd,1hole or Vel,y , Rd,2holes = V pl , y , Rd ,unperf ψ shear

Equation 7-25
An alternative approach based on the reduced section (similar to than in
Annex N of EC3 but less sophisticated) is to base the shear resistance of the
perforated RHS on a simple ratio of the portion of remaining web at the centre of the
hole (Equation 7-26). This approach results in a higher resistance than the
maximum stress approach for all hole sizes, but means that there is some plasticity
present at design loads.
Vel , y , Rd,1hole or Vel , y , Rd , 2hole = (1 − Φ ) × V pl , y , Rd ,unperf

Equation 7-26
Only sections with large holes (Φ ≥ 0.6) were used for the full plastic
parametric FE study as sections with small holes were too strong to be failed in
shear. The shear force-displacement plots are shown in Figure 7-32 through Figure
7-37. The non-dimensional ratio of shear force against the plastic shear capacity of
the unperforated section is plotted on the ordinate axis and shear angle is plotted on
the abscissa. The shear angle is the angle between the original longitudinal axis of
the RHS and the displaced longitudinal axis at each end. The shear angle is
dimensional and is dependent on the length of the beam. The shear resistance
calculated from Equation 7-26 is also plotted and shown as a circle symbol on the
corresponding data series.
For the range of sections chosen for the full plastic parametric study, the
shear resistance based on the reduced section is less than the shear force at the onset
of non-linearity. However, sections with more slender webs may be susceptible to
the Vierendeel shear failure mechanism and shear resistance may have to be reduced
accordingly. The relative thickness of the webs means that RHS are more resistant
to Vierendeel shear failure than I-sections and a consideration of the mechanism
may not be necessary in practice.
The Vierendeel shear force can be estimated using an equivalent rectangular
opening (Figure 7-38a) and plastic stress block for the hinges (Figure 7-38b) as
previously used in allowable stress design (e.g. Constrado document Cl/SfB 1968
‘Holes in beam webs’). Alternatively a more sophisticated method for calculating
internal forces and moments may be employed such as the one used in Clause
N.3.3.2.3 of Annex N (for beams with multiple circular openings).
The shear resistance of a section with one hole is larger than the shear
resistance of the same section with two holes, particularly for squarer sections. In
the unperforated zone, the webs each carry the same amount of shear force.
However, in the perforated zone, the torsional and lozenging rigidity of the box
shape allows the unperforated web to carry a greater portion of the shear force than
the perforated web. This effect is exaggerated by the boundary conditions and short
span used in the parametric FE models. Asymmetry induces internal torsional and
lozenging forces that, for high plastic shear loads in longer spans, can cause
distortion of the cross-section at some considerable distance from the hole. In
design, the distortion of the cross-section is of much less importance than the (much
larger) vertical deformations across the hole.
RHS 250x250x10 one hole parametric shear (FE)
1.0
0.9
Φ=0.6
0.8 Φ=0.7
Φ=0.8
0.7
Φ=0.9
0.6
0.5
0.4
0.3 0.7 m
0.2
Shearing angle
0.1 Shearing angle
Φ = 0.6, 0.7, 0.8, 0.9
0.0
0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035 0.040
Shearing angle (radians)
Figure 7-32: Shear force-shearing angle relationship for RHS 250x250x10 one hole
RHS 250x250x10 two holes parametric shear (FE)
1.0
0.9 0.7 m
0.8 Shearing angle

0.7 Shearing angle Φ=0.6
0.6
Φ=0.7
0.5
Φ=0.8
0.4
0.3 Φ=0.9
0.2
0.1
Φ = 0.6, 0.7, 0.8, 0.9
0.0
0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035 0.040
Figure 7-33: Shear force-shearing angle relationship for RHS 250x250x10 two holes
1.0
0.9 Φ=0.6
0.8 Φ=0.7
0.7 Φ=0.8
0.6 Φ=0.9
0.5
0.4
0.3 0.7 m
0.2
Shearing angle
0.1 Shearing angle
Φ = 0.6, 0.7, 0.8, 0.9
0.0
0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035 0.040
1.0
0.9 0.7 m
0.8 Shearing angle

0.7 Shearing angle Φ=0.6
0.6
Φ=0.7
0.5
0.4 Φ=0.8
0.3
Φ=0.9
0.2
0.1
Φ = 0.6, 0.7, 0.8, 0.9
0.0
0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035 0.040
1.0
0.9
0.8
0.7
Φ=0.6
0.6
0.5 Φ=0.9
0.4
0.3 0.7 m
0.2
Shearing angle
0.1 Shearing angle
Φ = 0.6, 0.7, 0.8, 0.9
0.0
0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035 0.040
1.0
0.9 0.7 m
0.8 Shearing angle

0.7 Shearing angle
0.6
Φ=0.6 Φ=0.7
0.5
Φ=0.8
0.4
0.3
Φ=0.9
0.2
0.1
Φ = 0.6, 0.7, 0.8, 0.9
0.0
0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035 0.040
Hinge Hinge
x
x
Plastic hinge
Hinge Hinge stress block
(a) (b)
Figure 7-38: The vierendeel mechanism of shear failure
7.6.4 Deflection
Clause N.2.3 of Annex N in EC3 deals with the deflection of I-beams with
single perforations. The clause states that the calculation should include a
component of the shear deformation across the opening. Shear deformation at
extreme displacements, for RHS with large openings (Φ ≥ 0.6) takes the form shown
in Figure 7-39 and is similar for sections with similar hole sizes, regardless of the
aspect ratio.
If the shear resistance is calculated according to a maximum elastic stress
criterion, the strains at design loads will be limited to the elastic range and shear
deformation will be small enough to be neglected.
The deformation of the perforated zone is crucial in design, as displacements
are concentrated over a short length. The displaced shape shown in Figure 7-40 is
typical of the shear deformation of RHS with large holes. If a reduced shear area
approach is adopted for design, a reliable method of calculating the vertical
displacement across the width of the hole must be developed as an accompanying
serviceability limit.
(a) (b)
Figure 7-39: Distortions under the action of shear
0.60
0.50
0.40
Vertical displacement
0.30
(mm)
0.20
Finite Element
0.10
0.00 RHS 200x100x8

Two holes Φ = 0.8
-0.10 Shear
Shear force = 81.7 kN
-0.20
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
(percent of span)
Figure 7-40: Shear displacement of RHS 200x100x8 two holes 140.8 mm dia
The zone of influence describes the length of the RHS over which the web
opening influences the stress distribution, and hence marks the boundary between
the perforated and unperforated zones (Figure 7-1). In the elastic range, the zone of
influence for shear has been observed to extend a distance in the order of one hole
diameter on either side of the edge of the hole. Consequently, the zone is
approximately three diameters in length.
instability).
The shear capacity based on a yield criterion can be calculated using
Equation 7-27, where γMShe1 and γMShe2 are possible partial safety factors. The stress
concentration factor for shear (ψshear) can be found from Table 7-9. The expression
is valid for hole size ratios between 0 and 0.9 and for web slendernesses between
10.0 and 46.0, but does not allow for instability of the cross-section (such as shear
buckling of the webs).
V pl , y , Rd ,unperf
Vel,y , Rd,1hole or Vel,y , Rd,2holes =
(ψ shear + γ MShe1 ) × γ MShe2
Equation 7-27
An alternative shear capacity based on a simplified reduced shear area
approach can be calculated from Equation 7-28. This does not allow for instability
of the cross-section or shear failure by the Vierendeel mechanism, and requires a
displacement based serviceability criterion to be useful in design.
V pl , y , Rd ,unperf (1 − Φ )
Vel,y , Rd,1hole or Vel,y , Rd,2holes =
γ M0
Equation 7-28
For both approaches, the shear capacity of RHS with one hole is taken to be
the same as the shear capacity of an RHS with two holes.
Table 7-9: Stress concentration factors for shear, ψshear

ratio, Φ 10 14 18 22 26 30 34 38 42 46
0.1 2.21 2.22 2.23 2.24 2.25 2.27 2.28 2.29 2.30 2.31
0.2 2.46 2.48 2.51 2.54 2.57 2.59 2.62 2.65 2.67 2.70
0.3 2.77 2.81 2.86 2.90 2.95 3.00 3.04 3.09 3.14 3.18
0.4 3.15 3.22 3.29 3.36 3.44 3.51 3.58 3.65 3.72 3.80
0.5 3.63 3.74 3.85 3.96 4.07 4.17 4.28 4.39 4.50 4.61
0.6 4.28 4.44 4.61 4.77 4.93 5.09 5.26 5.42 5.58 5.74
0.7 5.24 5.49 5.74 5.99 6.25 6.50 6.75 7.00 7.26 7.51
0.8 6.92 7.35 7.78 8.22 8.65 9.08 9.52 9.95 10.38 10.81
0.9 11.45 12.42 13.39 14.37 15.34 16.32 17.29 18.26 19.24 20.21
7.7 Torsion
7.7.1 Introduction
Rules for design of members in torsion can be found in Annex G of EC3.
BS 5950 does not contain any specific rules for torsion although British Steel
publication TD 365, ‘SHS Design to BS 5950 Part 1’ contains a short section for
torsional design of RHS. The Steel Construction Institute also provides advice for
the design of members in torsion (Nethercot et al 1989).
Clause G.2 of Annex G (EC3) states that designing to resist loads by torsion
is not usually an efficient method of load transfer and should be avoided wherever
practicable. The clause recommends the use of hollow sections or box girders where
torsion is unavoidable because box sections are more efficient at transmitting
torsional loads than open sections. The design rules in Annex G are applicable only
to members with class 1, 2 or 3 cross-sections, and where the webs are not
susceptible to shear buckling.
The total internal torque in a thin walled member under the action of torsion
is the sum of Saint-Venant torsion and the warping torsion, but the torsional rigidity
of RHS is very large compared to the warping rigidity and design may proceed by
neglecting warping torsion, without any loss of accuracy (Section 2.5.3).
The plastic torsional capacity (Tpl) of an unperforated RHS can be calculated
by considering the flow of plastic shear around the cross-section (Equation 7-29).
The full plastic torque is only reached at large angles of twist and design is normally
based on the elastic torsional capacity (Tel). The elastic torsional capacity is
proportional to the torsional modulus constant (Ct) and can be calculated from
Equation 7-30. The shear stress at the external surface (τo) is proportional to the
applied torque (T) and inversely proportional to the torsional modulus constant
(Equation 7-31).
Af y
T pl = (h(b − t ) + b(h − t ))
2 3 (b + h )
Equation 7-29
Tel = f y C t 3
Equation 7-30
τ o = T Ct
Equation 7-31
The total angle of torsional twist (θ) is proportional to the length of the
member (L) and the applied torque (T) and is inversely proportional to the shear
modulus of the material (G) and the torsional inertia constant (It). The total angle of
twist (Equation 7-32) and the twist per unit length are both important for the
serviceability limit state, but permissible values will be depend on the structure
being designed.
TL
θ=
GI t
Equation 7-32
Annex G of EC3 gives formulae for the calculation of torsional constants,
but not for box sections. However, formulae for calculation of the torsional inertia
constant and torsional modulus constant for RHS can be found in Annex A of BS
EN 10210 part 2 (see also Section 2.5 and Appendix C).
FE models of unperforated RHS closely match theoretical predictions using
the BS EN 10210 values for the torsional constants. However, full-scale tests
conducted as part of this research project indicate elastic and plastic capacities
below (12 to 20%) those predicted by Equation 7-29 and Equation 7-30 (Section
5.4). FE models of perforated sections were considerably more successful at
predicting behaviour in the laboratory (Section 5.4).
An unperforated RHS in pure torsion will have a (reasonably) uniform
distribution of elastic shear stress that varies through the thickness and is greater
(typically 20 to 30%) on the external surface (Section 2.5.2). A small circular
opening on the neutral axis will disturb the system of stress and produce a stress
pattern similar to that around a hole in an infinite plate in pure shear (See also
Section 6.5). Stress is raised at four points around the edge of the hole (location
in Figure 7-2). Larger holes raise the stress further as less web is available to
transmit the shear across the width of the hole.
concentrations around openings in the webs of RHS under the action of uniform
external torsion. The models used in the parametric study are shown
diagrammatically in Figure 7-41. Stress concentrations were found to be dependent
on hole size and web slenderness, but not aspect ratio. The dependence of stress
concentration on the web slenderness is analogous to the dependence upon
thickness, of the stress concentration around a circular hole in an infinite plate in
pure shear. The FE contour plots of von Mises stress shown in Figure 7-43 are
typical of the stress patterns observed.
concentration factors for torsion. The stress concentration factor for torsion, Ψtorsion,
is defined as the ratio of the maximum von Mises stress in the unperforated beam, to
the maximum von Mises stress in the perforated beam. The torque is such that the
maximum von Mises stress in the perforated RHS is in the elastic range. The stress
distribution in the web of an RHS with one hole is nearly identical to that in the web
of an RHS with two holes, although the maximum stress is slightly higher for
sections with two holes for large diameters.
using a least squares method (Equation 7-33). For the models included in the study,
the maximum differences between the empirical function and the FE values were
1.03 (empirical value less than FE value) and 1.25 (empirical value greater than FE
value). The square root of the average square error was 0.23. The actual stress
concentration data from the FE study is presented in Appendix E.
 1
ψ torsion = 

(
− 1 0.038λ web − 4.847Φ 2 + 6.562Φ − 1.734 + 2 )
1− Φ 
Equation 7-33
larger for slender webs. The function is valid for hole size ratios between 0 and 0.9
and for web slendernesses between 13.0 and 38.7.
50
45
40
35

Aspect ratio 1.50
25
Aspect ratio 1.67
15
10
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Hole size ratio, Φ
Figure 7-41: Parametric sampling for torsion
18.0
16.0
14.0
12.0 16-18
Stress
10.0 concentration factor 14-16
ψ torsion 12-14
8.0
10-12
6.0
8-10
4.0 6-8
2.0 4-6
30 0.90 2-4
0.70
Web 21
0.50
slenderness Hole size ratio
λ web 12 0.30 Φ
0.10
Figure 7-42: Stress concentration factors for torsion
RHS 250x250x10 RHS 250x250x10

Torsion Torsion
Torque = 22.3 kNm Torque = 22.3 kNm
RHS 250x250x10 RHS 250x250x10

Torsion Torsion
RHS 250x250x10 RHS 250x250x10

Torsion Torsion
Figure 7-43: Distribution of von Mises stress for RHS 250x250x10 in torsion

7.7.3 Resistance
The stress concentration factors determined in the previous section can be
used to determine the torsional resistance based on a maximum stress criterion. If
stress is limited to the design yield stress, the yield based torsional resistance of the
perforated section can be determined by dividing the elastic torsional capacity of the
unperforated section by the stress concentration factor (Equation 7-34).
Tel , Rd,1hole or Tel,Rd,2holes = Tel , Rd ,unperf ψ torsion

Equation 7-34
An alternative approach based on the reduced section (similar to that for
shear) is to base the torsional resistance of the perforated RHS on a simple ratio of
the portion of remaining web at the centre of the hole (Equation 7-35). This
approach results in a higher resistance than the maximum stress approach for all hole
sizes, but means that there is some plasticity present at design loads.
Tel , Rd,1hole or Tel , Rd , 2hole = (1 − Φ ) × Tel , Rd ,unperf

Equation 7-35
The torque-rotation plots obtained from the full plastic parametric study are
shown in Figure 7-44 through Figure 7-49. The non-dimensional ratio of torque
against the elastic torsional capacity of the unperforated section is plotted on the
ordinate axis. The non-dimensional ratio of the total rotation against the rotation at
yield of the unperforated section is plotted on the abscissa. The total rotation is the
sum of the rotations across the perforated and unperforated parts and is dependent on
the member length. The torsional resistance calculated from Equation 7-35 is also
plotted and shown as a circle symbol on the corresponding data series.
For the range of sections chosen for the full plastic parametric study, the
torsional resistance based on the reduced section is less than the torque at the onset
of non-linearity. However, sections with more slender webs may be susceptible to a
mechanism of failure similar to Vierendeel shear failure and torsional resistance may
have to be reduced accordingly. This type of failure can be evaluated with yield line
analysis. Maximum plastic torque may be evaluated by the same method, but is
unlikely to be useful in design, as, for most applications, high twist is likely to cause
a loss of structural integrity before plastic collapse of the member occurs.
Maximum torques are shown in Figure 7-50 and Figure 7-51. For small
holes and unperforated sections, the maximum torque is equal to the theoretical
plastic torque (Equation 7-29). Strain hardening is unable to generate higher
torques, as plastic strains are relatively small. Large holes reduce the elastic
capacity by a significant amount, but the reduction in maximum torque is
comparatively low. This is partly due to strain hardening, which is able to develop
due to the enormous strains that result from the gross torsional twist across the hole.
The ultimate torsional resistance of a section with one hole is larger than the
ultimate torsional resistance of the same section with two holes. In the perforated
zone, the torsional and lozenging rigidity of the box shape allows the unperforated
web to carry a greater portion of the shear flow than the perforated web.
Asymmetry induces internal torsional and lozenging forces that, for high plastic
torsional loads, can cause distortion of the cross-section at some considerable
distance from the hole. In design, the distortion of the cross-section is of much less
importance than the angular deformation across the hole.
RHS 250x250x10 one hole parametric torsion (FE)
1.2
Φ=0.0 Plastic torque (theory, unperforated)

1.0 Φ=0.1, 0.3
Φ=0.5
0.8
Φ=0.7
Φ=0.9
0.6
0.4
0.2
Φ = 0.0, 0.1, 0.3, 0.5, 0.7, 0.9
0.0
0 1 2 3 4 5 6
Figure 7-44: Torque-rotation relationship for RHS 250x250x10 one hole
RHS 250x250x10 two holes parametric torsion (FE)
1.2
Φ=0.0 Plastic torque (theory, unperforated)

1.0 Φ=0.1
Φ=0.3
0.8
Φ=0.5
0.6
Φ=0.7
0.4
Φ=0.9
0.2
Φ = 0.0, 0.1, 0.3, 0.5, 0.7, 0.9
0.0
0 1 2 3 4 5 6
Figure 7-45: Torque-rotation relationship for RHS 250x250x10 two holes
1.2
Plastic torque (theory, unperforated)
Φ=0.0, 0.1
Φ=0.3
1.0 Φ=0.5
Φ=0.7
0.8
Φ=0.9
0.6
0.4
0.2
Φ = 0.0, 0.1, 0.3, 0.5, 0.7, 0.9
0.0
0 1 2 3 4 5 6
1.2
Plastic torque (theory, unperforated)
Φ=0.0, 0.1
1.0
Φ=0.3 Φ=0.5
0.8
Φ=0.7
0.6
0.4 Φ=0.9
0.2
Φ = 0.0, 0.1, 0.3, 0.5, 0.7, 0.9
0.0
0 1 2 3 4 5 6
1.2 Plastic torque (theory, unperforated)

Φ=0.3
1.0
Φ=0.1, 0.3 Φ=0.0
Φ=0.5
Φ=0.7
0.8
Φ=0.9
0.6
0.4
0.2
Φ = 0.0, 0.1, 0.3, 0.5, 0.7, 0.9
0.0
0 1 2 3 4 5 6
1.2 Φ=0.0 Plastic torque (theory, unperforated)

1.0
Φ=0.3
Φ=0.5
Φ=0.1, 0.3
0.8
Φ=0.7
0.6
0.4 Φ=0.9
0.2
Φ = 0.0, 0.1, 0.3, 0.5, 0.7, 0.9
0.0
0 1 2 3 4 5 6
RHS 250x250x10 parametric torsion (FE)
1.2
Max torque / Plastic torque (theory, unperforated)
1.0
0.8
0.6
0.4
One hole
0.2
Two holes
0.0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Hole size ratio, Φ
Figure 7-50: Maximum torque for RHS 250x250x10
RHS 200x100x8 parametric torsion (FE)
1.2
Max torque / Plastic torque (theory, unperforated)
1.0
0.8
0.6
0.4
One hole
0.2
Two holes
0.0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Hole size ratio, Φ
Figure 7-51: Maximum torque for RHS 200x100x8
7.7.4 Deflection
Torsional deformation at extreme displacements, for perforated RHS, takes
the form shown in Figure 7-52 and, for small diameters is similar for sections with
similar hole sizes, regardless of the aspect ratio (Figure 7-52a and b). For large
holes, oblong sections distort through shearing of the flanges (Figure 7-52c), while
square sections distort through shearing of the webs (Figure 7-52d).
If the torsional resistance is calculated according to a maximum elastic stress
criterion, the strains at design loads will be limited to the elastic range and the
additional angular deformation across the width of the hole will be small enough to
be neglected.
As for members in shear, the deformation of the perforated zone is crucial in
design, as displacements are concentrated over a short length. The displaced shapes
shown in Figure 7-53 and Figure 7-54 are typical of the torsional deformation of
RHS with web openings. If a reduced area approach is adopted for design, a reliable
method of calculating the angular displacement across the width of the hole must be
developed as an accompanying serviceability limit.
In the elastic range, the total twist (θ) can be calculated from the sum of the
twist in the unperforated and perforated parts. The twist across the perforated zone
can be estimated by applying a reduced stiffness over a length equal to the depth of
section. For a member with one perforated zone the total twist can be estimated
from Equation 7-36. The stiffness reduction factor (ρtorsion) given in Equation 7-37,
and as described by Equation 7-36, is conservative for all the sections chosen for the
fully plastic parametric study (Figure 7-55).
T × (L − h ) T × h × ρ torsion
θ= +
It ×G It ×G
Equation 7-36
ρ torsion = 1 / (1 − Φ )
Equation 7-37
(a) (b)
(c) (d)
Figure 7-52: Distortions under the action of torsion
200%
Φ = 0.0, 0.1, 0.3, 0.5, 0.7, 0.9
150%
(percent of total twist of unperforated)
100%
50%
Twist
0%
-50% Φ=0.0, 0.1, 0.3, 0.5
-100%
Φ=0.7 Φ=0.5
-150% Φ=0.7
Φ=0.9 Hole diameters Φ=0.9
-200%
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
(percent of span)
Figure 7-53: Torsional twist of RHS 250x250x10 two holes
200%
Φ = 0.0, 0.1, 0.3, 0.5, 0.7, 0.9
150%
(percent of total twist of unperforated)
100%
50%
Twist
0%
-50% Φ=0.0, 0.1, 0.3, 0.5
-100%
Φ=0.7 Φ=0.5
-150% Φ=0.7
Φ=0.9 Hole diameters Φ=0.9
-200%
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
(percent of span)
Figure 7-54: Torsional twist of RHS 200x100x8 two holes
Stiffness reduction factor for torsion, ρtorsion 10

RHS 250x250x10 one hole
9
RHS 250x250x10 two hole
8
7 RHS 300x200x12 two holes
6
RHS 200x100x8 two holes
5 Design expression
4
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Hole size ratio, Φ
Figure 7-55: Stiffness reduction factors for torsion
The zone of influence describes the length of the RHS over which the web
opening influences the stress distribution, and hence marks the boundary between
the perforated and unperforated zones (Figure 7-1). In the elastic range, the zone of
influence for torsion has been observed to extend a distance in the order of two hole
diameters on either side of the edge of the hole. Consequently, the zone is
approximately five diameters in length (e.g. Figure 7-43).
instability).
The elastic torsion capacity can be calculated using Equation 7-38, where
γMTor1 and γMTor2 are possible partial safety factors. The stress concentration factor
for shear (ψtorsion) can be found from Table 7-10. The expression is valid for hole
size ratios between 0 and 0.9 and for web slendernesses between 14.0 and 38.0, but
does not allow for instability of the cross-section.
Tel , Rd ,unperf
Tel,Rd,1hole or Tel,Rd,2holes =
(ψ torsion + γ MTor1 ) × γ MTor2
Equation 7-38
An alternative torsional capacity based on a simplified reduced area
approach can be calculated from Equation 7-39. This does not allow for instability
of the cross-section or failure by a torsional Vierendeel mechanism, and requires a
displacement based serviceability criterion to be useful in design.
T pl , Rd ,unperf (1 − Φ )
Tel , Rd,1hole or Tel,Rd,2holes =
γ M0
Equation 7-39
For both approaches, the torsional capacity of RHS with one hole is taken to
be the same as the torsional capacity of an RHS with two holes.
Table 7-10: Stress concentration factors for torsion, ψtorsion

ratio, Φ 14 18 22 26 30 34 38
0.1 1.93 1.95 1.97 1.98 2.00 2.02 2.04
0.2 1.98 2.02 2.06 2.09 2.13 2.17 2.21
0.3 2.14 2.21 2.27 2.34 2.40 2.47 2.53
0.4 2.43 2.53 2.63 2.74 2.84 2.94 3.04
0.5 2.87 3.02 3.17 3.32 3.48 3.63 3.78
0.6 3.49 3.71 3.94 4.17 4.40 4.63 4.85
0.7 4.37 4.73 5.08 5.44 5.79 6.14 6.50
0.8 5.78 6.39 7.00 7.61 8.21 8.82 9.43
0.9 9.00 10.37 11.74 13.10 14.47 15.84 17.21
7.8 Combinations
7.8.1 Introduction
Torsion and shear force both produce shear stress in a member while bending
produces longitudinal direct stress. If deflections are small enough (no geometric
non-linearity), the resulting system of stress is the combination (superposition) of
the three systems of stress that would be present for each individual action. The
shear stress resulting from the torsion combines with the shear stress due to the shear
force and the direct stress due to the bending to produce an increased von Mises
stress in the member. Therefore, a perforated RHS subject to a combination of
actions can be designed by the superposition of the stresses calculated by the
reduced area method introduced above.
Clause G.5 in Annex G of EC3 deals with design of unperforated RHS in
combined flexure and bending. The shear stress due to shear force and the shear
stress due to torsion are combined by additive superposition (Equation G.46d), but
the bending resistance is not reduced unless the combined shear stress (as a von
Mises equivalent) represents more than half of the design yield strength (Equation
G.47d). This approach is similar to the approach used when designing for a member
in shear and bending.
When torsion is coexistent with bending, design should allow for the effect
of the torsional rotation of the cross-section relative to its unloaded position. A
beam under the action of bending about its major axis will, when twisted by the
action of torsion be obliged to bend about its minor axis also. Similarly, the bending
deformations resulting from the bending about the major and minor axes may have
an influence on the torsional loading of the beam. Clause G.5 of EC3 provides
procedures for design that allow for the second order effects due to the rotational
and translational deformations. The procedure for perforated RHS outlined below is
a simplified treatment with no explicit allowance for second order effects.
In the case of an eccentrically loaded beam, such as a slim floor edge beam,
the principal structural action is that of bending. Hence, a convenient approach for
design is one that reduces the bending capacity to allow for coexistent shear force
and torsion. One way to achieve this is to use a reduced yield stress (fy,reduced) in the
bending resistance calculation (plastic and elastic moment capacity, perforated and
unperforated) to allow for the coexistent shear stress. Equation 7-40 is based on the
von Mises model for yield and a simplified reduced area calculation for shear stress
due to shear force (Vy,Sd) and torque (TSd). Since bending capacity is proportional to
yield stress, the ratio of the bending capacity in pure bending to the bending capacity
with coexistent shear and torsion is the same as the ratio fy / fy,reduced (leading to
Figure 7-56, or Figure 7-57 for only two actions). Equation 7-41 is for the special
case of an RHS with one hole where the shear force opposes the torsional shear in
the perforated web. Additional calculations are required to ensure that failure will
not occur by a Vierendeel shear mode (torsional or shear) or by instability of the
cross-section. The parametric FE study indicates that instability and the Vierendeel
shear mode will not occur if the web slenderness is less than 22.0.
2
 V y , Sd TSd 
f y ,reduced = f − 3 
2
+ 
 (1 − Φ ) Av (1 − Φ )C t
y

Equation 7-40
  V y , Sd 
2
 V y , Sd TSd  
2
 TSd
f y ,reduced = MIN  f y2 − 3  −  , f y − 3
2
+ 
  (1 − Φ ) Av (1 − Φ )C t   Av C t  
 
Equation 7-41
As with the individual actions, the reduced area method for resistance
calculation for combined actions requires a displacement based serviceability
criterion to be useful in design.
100%
Bending capacity allowing for coexistent shear and torsion
% of shear capacity of perforated section 90% (elastic or plastic)
80% % of bending capacity of perforated section
70%
Shear force
60%
50%
40%
30%
20%
10%
99% 95% 90% 80% 70% 60% 50% 40% 0%
0%
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Torque
% of torsional capacity of perforated section
Figure 7-56: Contours of bending capacity with coexistent shear and torsion
% of bending capacity of unperforated section
100%
90% Bending failure
80%
70%
Bending strength
Shear-bending failure
60%
or
50% Torsion-bending failure
40%
30%
20%
10% Shear or torsion failure
0%
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Shear force
(% of shear capacity of unperforated section)
or Torque
(% of torsional capacity of unperforated section)
Figure 7-57: Bending capacity with coexistent shear or torsion
A number of design recommendations have been proposed which have been
developed and justified by the findings of a parametric FE study. Expressions for
resistance based on a simplified reduced area approach have been shown to be
reliable when instability of the cross-section is not critical. Empirical expressions
for the prediction of concentration factors for elastic stress have been developed for
when all plasticity is to be avoided.
The recommendations are intended to form the basis for the development of
design rules, but the following areas require additional investigation in order for the
recommendations to be useful for design:
1) An extended fully plastic parametric FE study to cover the full range of web
slenderness. The identification of stability limits and suitable models for
Vierendeel mechanisms in shear and torsion. The identification of sections with
unstable cross-sections at elastic loads is particularly important.
2) An extended fully plastic parametric FE study to cover combinations of actions
to confirm the recommendations for the treatment of perforated RHS beams with
coexistent shear and torsion.
3) Additional laboratory testing to calibrate the FE models where instability of the
cross-section and Vierendeel mechanisms are critical.
4) The development of reliable expressions for displacement and twist (including
localised deformation) for elastic and plastic loads (serviceability).
5) An experimental and FE study to develop rules for checking the stability of
perforated RHS members (e.g. lateral torsional buckling).
The following extensions to the scope would produce a wider and more generally
applicable set of design rules:
1) The inclusion of axial forces, bending about the minor axis and shear force
parallel to the flanges.
2) The inclusion of alternative common shapes for openings (e.g. rectangular and
extended circular).
3) Provision for closely spaced multiple perforations.
4) Provision for staggered and eccentric openings.
It may also be useful to consider the design of reinforcement and stiffeners
for situations where the reduction in resistance or stiffness due to an opening is
unacceptable. Possible schemes for reinforcement and stiffening of perforated RHS
are shown in Figure 7-58. The schemes are based on the reinforcement of circular
openings in I-sections (see EC3 Annex N clause N.2.1.5) although the closed shape
of the RHS presents the additional practical problem of limited access for welding.
z z
(a) (b)
x x
x y y x y y
z z
z z
(c) (d)
x x
x y y x y y
z z
(a) Longitudinal stiffener or reinforcement (b) Ring stiffener

(c) Circular reinforcement (d) Transverse reinforcement
Figure 7-58: Possible stiffening or reinforcement of web openings
8 Conclusions
8.1 Introduction
The primary aim of this project was to study the behaviour of Rectangular
Hollow Sections (RHS) with web openings, and to develop the basis for a set of
design rules for perforated RHS edge beams. To this end, the fundamental structural
behaviour of RHS with web openings was examined by an investigation that
combined analytical study with both small and full-scale testing. Representative
laboratory tests were conducted to provide sufficient data for Finite Element (FE)
models to be calibrated. The FE models were then used to study the behaviour in
greater detail, over a wider range of parameters. This Chapter summarises the
various stages of the project and ends with a synopsis of the conclusions and a brief
critical analysis of the project as a whole.
8.2 Laboratory testing and calibration of Finite Element models
8.2.1 Four-point bending (bending)
A series of 13 full-scale four-point bending tests were conducted on two
uniform bending. During the tests, the moment and average curvature were
measured. In 12 tests, strains on the external surface of the flanges and webs were
also measured.
The results of the tests showed that the reduction in the bending capacity due
to the web openings was as much as 30%, and that the openings caused no
measurable reduction in elastic stiffness. The reductions in bending capacity were
comparable to the reductions in the elastic and plastic section moduli at the location
of the hole. In some tests, the web openings were observed to decrease the plastic
stability of the cross-section causing a reduction in the rotation capacity.
in the elastic range. The experimental moment-curvature relationships matched
those predicted by the FE models, although the FE tended to under-predict the
bending resistance by around 10%.
8.2.2 Three-point bending (shear)
A series of six full-scale three-point bending tests were conducted on two
combined shear and bending. During the tests, the shear force, bending moment and
displacement were measured. In all six tests, strains on the external surface of the
flanges and webs were also measured.
The results of the tests showed that the reduction in the shear capacity due to
the web openings was as much as 85%, and that the openings caused a significant
reduction in the shear stiffness of the perforated zone. The reductions in shear
capacity were comparable to the reductions in the cross-sectional area of the webs at
the location of the hole.
in the elastic range. The experimental shear force-displacement relationships closely
matched those predicted by the FE models with a maximum difference of only 5%.
8.2.3 Full-scale torsion
A series of 14 full-scale torsion tests were conducted on two sizes and grades
of RHS to assess the effect of web openings upon structural performance in torsion.
During the tests, the torque and angle of twist were measured. In six tests, strains on
the external surface of the webs were also measured.
The experimental results showed that large holes cause a significant
reduction in both torsional capacity (up to 60%) and stiffness (up to 40%). The
reduction in stiffness was due to the perforated zone being much more flexible than
the unperforated beam, and severe deformations in this region were observed. The
reductions in capacity were comparable to the reductions in the cross-sectional area
of the perforated web.
The experimental results were compared with FE predictions and good
agreement was observed in terms of displaced shapes, elastic stiffness and strains in
elastic range. Experimental torque-rotation relationships matched closely those
predicted by the FE models in some cases, but were significantly different in others.
The largest difference between the FE predictions of torsional resistance and
experimental results occurred for the specimens without web openings. Theoretical
predictions of the torsional resistance of the unperforated sections were close to the
FE predictions.
The experimental method was re-examined in detail and the experimental
measurements were found to be valid. One explanation (that appears to fit most of
the available evidence) is that the steel does not always comply with the von Mises
failure criterion. Evidence of similar behaviour is present in the results of full-scale
testing of hot-finished RHS by Marshall (1972). Further experimental investigation
is required to investigate this phenomenon as it has important implications for
design.
8.2.4 Small-scale torsion
A series of 35 small-scale torsion tests was conducted on one size of cold-
formed steel tube to assess the effect of web openings of various diameters upon
structural performance in torsion. During the tests, the torque and angle of twist
were measured.
The experimental results showed that large web openings caused a
significant reduction in both torsional capacity and stiffness (as observed in the full-
scale torsion tests). The holes with the greatest diameter caused the biggest
reduction in resistance and stiffness and resulted in plastic deformation at lower
twist. Small holes caused only a small reduction in the torsional resistance and
stiffness, but reduced the torsional ductility of the member.
Various plastic collapse mechanisms were observed that were dependent on
the size and arrangement of holes, as it was possible to apply larger rotations with
the small-scale torsion testing apparatus than the full-scale apparatus.
A small number of tests showed how two holes in the same web were able to
interact if close enough together. The interaction between two openings caused a
further reduction in resistance than that resulting from one hole alone.
The experimental results were compared with FE predictions and good
agreement was observed in terms of displaced shapes and collapse mechanisms.
Experimental torque-rotation relationships matched closely those predicted by the
FE models in all cases (including specimens without holes), but there were large
differences between measurements and predictions of elastic stiffness due, in part, to
the testing method. The FE models also showed that, for unsymmetrical cases, the
boundary conditions (at the ends) have a large influence upon the torsional
behaviour of short members.
8.3 The parametric Finite Element study
8.3.1 Bending
Parametric FE models of three different sizes of RHS, with web openings of
various diameters, were used to study the structural performance of RHS in pure
bending. Moment-curvature relationships were obtained and stress distributions
were observed. In all cases, the RHS were thick enough to prevent elastic instability
of the flange and web.
The presence of the web opening was observed to cause an increased von
Mises stress in the webs above the opening, due to the reduced section moduli and
the stress concentrating effect of the change in cross-section. For small holes, the
increased stress above the hole balanced the effect of the reduction in elastic section
modulus and the moment at yield was not reduced. For large holes, the raised stress
was higher than the stress in the flanges and the moment at first yield was reduced
by the presence of the hole. The structural performance of the member was seen to
be unaffected by small areas of yielding material and, for design purposes, some
plasticity at working loads may be permissible. An empirical formula for the
prediction of maximum von Mises stress was developed and formulae for the
calculation of elastic and plastic moments, based on reduced area calculations, were
proposed.
The long yield plateau, characteristic of hot finished steel, prevented strain
hardening in the unperforated sections. In contrast, the high strains present in
perforated sections were able to generate sufficiently strain hardened stresses to
maintain increasing resistance at high curvatures. The ultimate resistance of
perforated members was observed to be increased by strain hardening and limited by
the stability of the compression flange and compression cee above the hole.
The stiffness of a member with a single perforated zone was seen to be
reduced slightly by the presence of large openings, but the reduction is small enough
to be neglected in design with no compromise in safety.
8.3.2 Shear
various diameters, were used to study the structural performance of RHS in high
shear. Shear force-deformation relationships were obtained and stress distributions
were observed. In all cases, the RHS were thick enough to prevent elastic instability
of the flange and web.
As with bending, the presence of the web opening was observed to cause an
increased von Mises stress in the webs at the edge of the opening, in this case due to
the reduced shear area and the stress concentrating effect of the change in cross-
section. The shear capacity and stiffness were seen to be reduced by all but the
smallest of openings, and the consequences of the increased flexibility of the
perforated zone for serviceability were seen to be important for design. For shear,
the structural performance of the member was seen to be governed by small areas of
yielding material and, for design purposes, plasticity at working loads is likely to be
undesirable. An empirical formula for the prediction of maximum von Mises stress
was developed and a formula for the calculation of the shear force at yield, based on
a reduced area calculation, was proposed. The high shear resistance of unperforated
members means that large reductions in shear resistance may not be critical in
design.
The high strains present in perforated sections were able to generate
sufficiently strain hardened stresses to maintain increasing shear resistance beyond
the first yield.
8.3.3 Torsion
various diameters, were used to study the structural performance of RHS in uniform
torsion. Torque-twist relationships were obtained and stress distributions were
observed. In all cases, the RHS were thick enough to prevent elastic instability of
the flange and web.
The presence of the web opening was observed to cause an increased von
Mises stress in the webs at the edge of the opening, due to the reduced area of cross-
section and the stress concentrating effect of the change in cross-section. The
torsional resistance and stiffness were seen to be reduced by all but the smallest of
openings, and the consequences of the increased flexibility of the perforated zone for
serviceability were seen to be important for design. The structural performance of
the member was seen to be governed by small areas of yielding material and, for
design purposes, plasticity at working loads may be undesirable. An empirical
formula for the prediction of maximum von Mises stress was developed and a
formula for the calculation of the torque at yield, based on a reduced area
calculation, was proposed.
The long yield plateau, characteristic of hot finished steel, prevented strain
hardening in the unperforated sections. In contrast, the high strains present in
perforated sections were able to generate sufficiently strain hardened stresses to
maintain increasing resistance beyond the twist at first yield.
In all but one category of tests, good agreement was achieved between
experimentally measured quantities and the corresponding FE predictions, allowing
parametric investigations to be conducted using the calibrated FE models.
Preliminary design advice was formulated based on the results of the parametric
study and laboratory tests. Aspects requiring further investigation or development
were identified (Sections 5.7 and 7.9).
The full-scale torsion tests produced some unexpected results with measured
capacities significantly lower than those predicted by FE and torsion theory.
Attempts were made to determine the cause of this behaviour and a number of
possibilities were eliminated. Although the anomalous results have not yet been
fully explained, evidence of similar behaviour in previous research was discovered.
The primary aim of the project was to develop the basis for a set of design
rules for perforated RHS in combined torsion, bending and shear. The preliminary
design rules were formulated, but problems with a testing machine and unexpected
torsion test results meant that more experimental work was conducted than
originally planned.
8.5 Lists of conclusions
8.5.1 Bending of perforated RHS (without elastic instability)
• For both perforated and unperforated sections, the results of four-point
bending tests can be adequately modelled using the FE technique (resistance,
stiffness, displaced shapes, collapse mechanisms, and strains in the elastic
range). [Section 3.5]
• Unperforated sections behave as predicted by conventional (Euler-Bernoulli-
Parent) beam bending theory (resistance, stiffness and strains in the elastic
• Perforations cause an increased von Mises stress in the webs above the
opening, due to the reduced section moduli and the stress concentrating effect
of the change in cross-section. [Section 7.5.2]
• If the hole is smaller than a certain size (the 'critical hole size'), the increased
stress above the hole balances the effect of the reduction in the elastic section
modulus, and the moment at yield is not reduced. [Section 7.5.3]
• For holes larger than the critical hole size, the raised stress is higher than the
stress in the flanges, and the moment at first yield is reduced by the presence
of the hole. [Section 7.5.3]
• The critical hole size and the maximum von Mises stress can be approximated
by empirical equations. [Section 7.5.2]
• Bending capacity can be conservatively predicted by an approximate method
based on the reduced elastic and plastic section moduli at the location of the
hole. [Section 7.5.3]
• The ultimate resistance of perforated members is increased by strain hardening
and is limited by the stability of the compression flange and compression cee
above the hole. [Section 7.5.3]
• The bending resistance of a section with two web openings is lower than that
of a section with one opening of the same diameter. [Section 7.5.3]
• The stiffness of a member with a single perforated zone is reduced slightly by
the presence of large openings, but the reduction is small enough to be
neglected in design with no compromise in safety. [Sections 3.5 and 7.5.4]
• In the elastic range, the zone of influence extends a distance in the order of
one hole diameter on either side of the edge of the hole. [Section 7.5.5]
8.5.2 Shearing of perforated RHS (without elastic instability)
• For both perforated and unperforated sections, the results of three-point
bending tests can be adequately modelled using the FE technique (resistance,
• Perforations cause an increased von Mises stress in the webs around the
opening, due to the reduced shear area and the stress concentrating effect of
the change in cross-section. [Section 7.6.2]
• The stress distribution in the web of an RHS with one hole is similar to that in
the web of an RHS with two holes, although the maximum stress is slightly
lower, particularly for large holes. [Section 7.6.2]
• The maximum von Mises stress can be approximated by an empirical
equation. [Section 7.6.2]
• In shear, the structural performance of the member is governed by small areas
of yielding material and, for design purposes, plasticity at working loads is
likely to be undesirable. [Section 7.6.3]
• Asymmetry induces internal torsional and lozenging forces that, for high
plastic shear loads in longer spans, can cause distortion of the cross-section at
some considerable distance from the hole. [Sections 4.4, 4.5 and 7.6.3]
• Shear capacity can be conservatively predicted by an approximate method
based on the reduced shear area at the location of the hole. [Section 7.6.3]
• The ultimate resistance of perforated members is increased by strain hardened
stresses (due to extreme deformations in the perforated zone). [Section 7.6.3]
• The shear capacity and stiffness are reduced by all but the smallest of
openings, and the consequences of the increased flexibility of the perforated
zone for serviceability are important for design. [Sections 7.6.4 and 7.6.6]
• The high shear resistance of unperforated members means that large
reductions in shear resistance may not be critical in design. [Section 7.6.1]
one hole diameter on either side of the edge of the hole. [Section 7.6.5]
8.5.3 Torsion of perforated RHS (without elastic instability)
• In some cases, for both perforated and unperforated sections, the results of
torsion tests can be adequately modelled using the FE technique (resistance,
range). However, the FE models used in this project did not produce accurate
predictions for some of the laboratory tests. [Sections 5.5 and 6.4]
• In terms of resistance, unperforated sections do not appear to behave as
predicted by conventional thick wall torsion theory, although stiffness and
elastic shear can be correctly predicted. [Sections 5.5 and 5.6.9]
• Perforations cause an increased von Mises stress in the webs around the
opening, due to the reduced cross-sectional area and the stress concentrating
effect of the change in cross-section. [Sections 6.5 and 7.7.2]
• The stress distribution in the web of an RHS with one hole is similar to that in
the web of an RHS with two holes, although maximum stresses are slightly
lower, particularly for large holes. [Sections 6.5 and 7.7.2]
• The maximum von Mises stress can be approximated by an empirical
equation. [Section 7.7.2]
• In torsion, the structural performance of the member is governed by small
areas of yielding material and, for design purposes, plasticity at working loads
is likely to be undesirable. [Section 7.7.3]
• Asymmetry induces internal lozenging forces that, for high plastic torques, can
cause distortion of the cross-section at some considerable distance from the
hole. [Sections 6.4.2 and 7.5.3]
• Torsional capacity (in FE at least) can be conservatively predicted by an
approximate method based on the reduced web area at the location of the hole.
[Section 7.7.3]
• The ultimate resistance of perforated members is increased by strain hardened
stresses (due to extreme deformations in the perforated zone). [Section 7.7.3]
• The torsional capacity and stiffness are reduced by all but the smallest of
openings, and the consequences of the increased flexibility of the perforated
zone for serviceability are important for design. [Sections 7.7.3 and 7.7.4]
two hole diameters on either side of the edge of the hole. [Section 7.7.5]
Rectangular Hollow Sections with circular web openings I
Appendices
Merely corroborative detail, intended to give artistic

verisimilitude to an otherwise bald and unconvincing narrative.
W.S. Gilbert, 1885

The Mikado, (Act II)
Rectangular Hollow Sections with circular web openings II
Table of contents
Table of Contents II
List of Figures III
List of Tables IV
A. Calibration of the Avery torsion testing machine VI
A.1 Introduction VI
A.2 Details of the testing machine VII
A.3 Measurement of torque VII
A.4 Measurement of twist XIII
B. Sensitivity of sectional properties XVII
C. Formulae for the calculation of the properties of cross-section XXV
D. Material properties XXIX
D.1 Introduction XXIX
D.2 Material properties RHS 250x250x10 grade S275J2H bar number 1 XXX
D.3 Material properties RHS 250x250x10 grade S275J2H bar number 2 XXXII
D.4 Material properties RHS 250x250x10 grade S275J2H bar number 3 XXXIV
D.5 Material properties RHS 200x100x8 grade S275J2H bar number 1 XXXVI
D.6 Material properties RHS 200x100x8 grade S275J2H bar number 2 XXXVIII
D.7 Material properties RHS 200x100x8 grade S275J2H bar number 3 XL
D.8 Material properties RHS 150x150x6.3 grade S275J2H bar number 1 XLII
D.9 Material properties RHS 200x100x8 grade S355J2H bar number 1 XLIV
D.10 Material properties RHS 150x150x6.3 grade S355J2H bar number 1 XLVI
D.11 Statistical analysis of material properties XLVIII
E. Stress concentration factors (Finite Element) LII
F. Calibration of testing equipment LVIII
F.1 Introduction LVIII
F.2 The Instron testing machine LVIII
F.3 The Zwick testing machine LIX
F.4 The Denison testing machine LIX
Rectangular Hollow Sections with circular web openings III
List of figures
Figure A-1: Test piece dimensions XI

Figure A-2: Torque-rotation relationship XII
Figure A-3: The material model and tensile test results XII
Figure A-4: Experimental measurements of elastic torsional stiffness XVI
Figure B-1: Sensitivity to section width (RHS 200x100x8) XVIII
Figure B-2: Sensitivity to section width (RHS 250x250x10) XVIII
Figure B-3: Sensitivity to section height (RHS 200x100x8) XIX
Figure B-4: Sensitivity to section height (RHS 250x250x10) XIX
Figure B-5: Sensitivity to section thickness (RHS 200x100x8) XX
Figure B-6: Sensitivity to section thickness (RHS 250x250x10) XX
Figure B-7: Sensitivity to external corner radius (RHS 200x100x8) XXI
Figure B-8: Sensitivity to external corner radius (RHS 250x250x10) XXI
Figure B-9: Sensitivity to internal corner radius (RHS 200x100x8) XXII
Figure B-10: Sensitivity to internal corner radius (RHS 250x250x10) XXII
Figure B-11: Sensitivity to yield stress (any section) XXIII
Figure B-12: Sensitivity to Young's modulus (any section) XXIII
Figure B-13: Sensitivity to Poisson's ratio (any section) XXIV
Figure D-1: Typical values of percentage energy error XLIX
Figure D-2: Typical values of plateau ratio XLIX
Figure D-3: Typical values of post yield stiffness L
Figure D-4: Typical values of elliptical strain hardening factor L
Figure D-5: Typical values of strain at maximum force LI
Figure D-6: Plateau ratio and post yield stiffness LI
Rectangular Hollow Sections with circular web openings IV
List of tables
Table A-1: Torsion test results summary X

Table A-2: Tension test results summary X
Table A-3: Summary of comparison with theoretical and FE predictions X
Table A-4: Stiffness of the testing machine (Kends) XV
Table D-1: Tensile test results for RHS 250x250x10 grade S275J2H bar #1 XXX
Table D-2: Material model for RHS 250x250x10 grade S275J2H bar #1 XXXI
Table D-3: ABAQUS material model (RHS 250x250x10 grade S275J2H bar #1) XXXI
Table D-4: Tensile test results for RHS 250x250x10 grade S275J2H bar #2 XXXII
Table D-5: Material model for RHS 250x250x10 grade S275J2H bar #2 XXXIII
Table D-6: ABAQUS material model (RHS 250x250x10 grade S275J2H bar #2) XXXIII
Table D-7: Tensile test results for RHS 250x250x10 grade S275J2H bar #3 XXXIV
Table D-8: Material model for RHS 250x250x10 grade S275J2H bar #3 XXXV
Table D-9: ABAQUS material model (RHS 250x250x10 grade S275J2H bar #3) XXXV
Table D-10: Tensile test results for RHS 200x100x8 grade S275J2H bar #1 XXXVI
Table D-11: Material model for RHS 200x100x8 grade S275J2H bar #1 XXXVII
Table D-12: ABAQUS material model (RHS 200x100x8 grade S275J2H bar #1) XXXVII
Table D-13: Tensile test results for RHS 200x100x8 grade S275J2H bar #2 XXXVIII
Table D-14: Material model for RHS 200x100x8 grade S275J2H bar #2 XXXIX
Table D-15: ABAQUS material model (RHS 200x100x8 grade S275J2H bar #2) XXXIX
Table D-16: Tensile test results for RHS 200x100x8 grade S275J2H bar #3 XL
Table D-17: Material model for RHS 200x100x8 grade S275J2H bar #3 XLI
Table D-18: ABAQUS material model (RHS 200x100x8 grade S275J2H bar #3) XLI
Table D-19: Tensile test results for RHS 150x150x6.3 grade S275J2H bar #1 XLII
Table D-20: Material model for RHS 150x150x6.3 grade S275J2H bar #1 XLIII
Table D-21: ABAQUS material model (RHS 150x150x6.3 grade S275J2H bar #1) XLIII
Table D-22: Tensile test results for RHS 200x100x8 grade S355J2H bar #1 XLIV
Table D-23: Material model for RHS 200x100x8 grade S355J2H bar #1 XLV
Table D-24: ABAQUS material model (RHS 200x100x8 grade S355J2H bar #1) XLV
Table D-25: Tensile test results for RHS 150x150x6.3 grade S355J2H bar #1 XLVI
Table D-26: Material model for RHS 150x150x6.3 grade S355J2H bar #1 XLVII
Table D-27: ABAQUS material model (RHS 150x150x6.3 grade S355J2H bar #1) XLVII
Table E-1: FE stress concentration factors for bending (part 1 of 2) LII
Table E-2: FE stress concentration factors for bending (part 2 of 2) LIII
Rectangular Hollow Sections with circular web openings V
Table E-3: FE stress concentration factors for shear (part 1 of 2) LIV

Table E-4: FE stress concentration factors for shear (part 2 of 2) LV
Table E-5: FE stress concentration factors for torsion (part 1 of 2) LVI
Table E-6: FE stress concentration factors for torsion (part 2 of 2) LVII
Rectangular Hollow Sections with circular web openings VI
A. Calibration of the Avery torsion testing machine
A.1 Introduction
The torsion testing machine used in the small-scale study (Chapter 6) is
employed principally for undergraduate teaching and, unlike the equipment used in
the full-scale experimental work (Chapters 3, 4 and 5), is not calibrated regularly to
National Standards (Appendix F).
The original role of the small-scale torsion tests was that of exploratory
study, preliminary to the full-scale testing programme. At the time of testing, the
uncertain accuracy of the results was not regarded as significant, because the
principal aim was to gain a qualitative understanding of the torsional behaviour of
perforated RHS. However, the results proved to be particularly informative and the
matter of accuracy became important after the tests were complete. This Appendix
contains a short description of the methods employed to verify the accuracy of the
experimental results.
Rectangular Hollow Sections with circular web openings VII
A.2 Details of the testing machine
Description Avery “Reverse torsion testing machine” (Figure 6-1)

Type 6609 CGG Number E61981
Load ranges 0 - 15000 inch lb in 50 inch lb divisions (calibrated)
(Dial readout) 0 - 7500 inch lb in 20 inch lb divisions (not calibrated)
0 - 3000 inch lb in 10 inch lb divisions (not calibrated)
0 - 1500 inch lb in 5 inch lb divisions (not calibrated)
Note: 1 inch lb is equivalent to 0.113 Nm
Speed ranges 90 degrees per minute (not checked)
(Both directions) 30 degrees per minute (not checked)
10 degrees per minute (checked)
31/3 degrees per minute (checked)
Date of calibration 17th July 1997
A.3 Measurement of torque
To investigate the accuracy of the torsion rig torque gauge, two (nominally
identical) steel test pieces of solid circular cross-section were tested at angular
velocities of 31/3 degrees and 10 degrees per minute. Torque-rotation histories of the
two tests were compared with numerical predictions based on established torsion
formulae and a simple Finite Element (FE) model. Tensile tests provided values for
the material properties used in the calculations.
The approximate dimensions of the torsion test pieces are given in Figure
A—1a. The driving flats on each specimen were initially mutually perpendicular to
restrict transmission of any bending that may result due to any accidental asymmetry
of test pieces.
Both specimens were initially tested at an angular velocity of 31/3 degrees per
minute and then unloaded at the same speed. The specimens were then reloaded at
Rectangular Hollow Sections with circular web openings VIII
10 degrees per minute until a steady value of torque was obtained. The 15,000 inch
lb. (1700 Nm) scale was used for both tests. The torsion tests on the two specimens
are referred to as ‘test A’ and ‘test B’.
The approximate dimensions of the tension test pieces are given in Figure
A—1b. Tensile tests were conducted on a 200 kN Zwick machine with a 50 mm
gauge length extensometer. The tension tests on the two specimens are referred to
as ‘test 1’ and ‘test 2’.
Tensile test piece dimensions are specified in accordance with BS EN
10002-1:1990. The tensile test pieces were turned from the same steel bar as the
torsion specimens.
Table A—1 gives a summary of the torsion test results. Complete torque-
rotation histories can be seen in Figure A—2. Table A—2 gives a summary of the
tension test results.
The torsion tests were modelled using an ABAQUS v5.5.1 (Section 2.3) FE
model consisting of a single three-noded quadratic beam element (B32) of circular
cross-section4. The material model is shown in Figure A—3. The results were also
compared with predictions based on torsion theory (Section 2.5). The torque at
yield (Tel) was calculated using Equation A—1. The maximum torque (Tmax) was
calculated using Equation A—2. The elastic stiffness was calculated using Equation
A—3.
4
The formulation of the beam element used in the FE study is also based on torsion theory, and is
able to model plastic yielding as it moves inwards from the outside of the cross-section.
Rectangular Hollow Sections with circular web openings IX
1
Tel = πr 3 f y
2 3
Equation A—1
2
Tmax = πr 3 f u
3 3
Equation A—2
T Eπr 4
=
θ 4 L(1 + υ )
Equation A—3
Where
E = Young’s modulus (208.6 kN/mm2)
fy = Yield stress (580 N/mm2)
fu = Tensile strength (620 N/mm2)
L = Gauge length (108 mm)
r = Radius of bar (10 mm)
T = Torque
Tel = Torque at yield
Tmax = Maximum torque
υ = Poisson’s ratio (assumed to be 0.3)
The torque-rotation histories from the torsion tests are compared with the FE
and theoretical predictions in Figure A—2. The comparisons are summarised in
Table A—3. The measurements of maximum torque and torque at yield are
sufficiently close to the predictions to conclude that the testing machine is correctly
Rectangular Hollow Sections with circular web openings X
indicating the torque. The experimental measurement of stiffness is lower than the
predictions, due to the flexibility of the testing machine (Section A.4).
Table A—1: Torsion test results summary

Test Initial stiffness Torque at yield Maximum torque
Tel Tmax
(Nm per degree) (Nm) (Nm)
Test A 118.8 491.6 728.9
Test B 115.3 474.6 748.1
Mean 117.0 483.1 738.5
Standard deviation 2.5 12.0 13.6
(% of mean) (2.1%) (2.5%) (1.8%)
Table A—2: Tension test results summary

Test Young’s modulus 0.1% proof stress 0.2% proof stress Tensile strength
E fu
2 2 2
(kN/mm ) (N/mm ) (N/mm ) (N/mm2)
Test 1 209.8 576.0 579.2 626.5
Test 2 207.2 573.4 581.9 623.6
Mean 208.5 574.7 580.6 625.1
Standard deviation 1.8 1.8 1.9 2.1
(% of mean) (0.9%) (0.3%) (0.3%) (0.3%)
Table A—3: Summary of comparison with theoretical and FE predictions

Test Initial stiffness Torque at yield Maximum torque
Tel Tmax
(Nm per degree) (Nm) (Nm)
Theoretical 202.2 523.2 745.8
(% of theoretical) (100.0%) (100.0%) (100.0%)
Finite Element 195.0 500.0 747.5
(% of theoretical) (96.4%) (95.6%) (100.2%)
Average experimental 117.0 483.1 758.5
(% of theoretical) (57.9%) (92.3%) (101.7%)
Rectangular Hollow Sections with circular web openings XI
230 mm
(a) Torsion test specimen 20 mm diameter ± 0.2 mm
108 mm
27 mm
flats 10 mm
31 mm diameter
236 mm
(b) Tension test specimen 10 mm diameter
60 mm
20 mm diameter 13 mm
Figure A—1: Test piece dimensions
Rectangular Hollow Sections with circular web openings XII
900
800 Theoretical maximum torque
700
600
Torque (Nm), T
500
Theoretical torque at yield
400
300
See Section A.4 for an
Test A
200 explanation of the
differences in elastic Test B
100 stiffness Finite Element
0
0 10 20 30 40 50 60 70 80 90 100
Angle of twist (degrees), θ
Figure A—2: Torque-rotation relationship
700
600
Test A
2
500 Test B
Material model
400 Engineering stress (N/mm )
2
Schematic
620
300
580
200
100 0.278 2.922 Engineering strain (%)
0
0 1 2 3 4 5 6 7 8
Figure A—3: The material model and tensile test results
Rectangular Hollow Sections with circular web openings XIII
A.4 Measurement of twist
The angle of twist in the small-scale torsion tests was measured by timing
the machine while it applied a rotation at the constant angular velocity of 31/3
degrees per minute. This method allowed the tests to be conducted quickly, by a
single operator, but it lacked the accuracy that could be achieved by measuring the
twist of the specimen directly. Similar tests, such as those conducted by Marshall
(1972) and Moore and Clark (1952), made use of optical devices5, and twists were
measured very accurately, but to the detriment of the simplicity of the experimental
set-up and rate of testing.
The largest error that resulted as a consequence of the chosen method of
twist measurement was that the total rotation also included the rotation that occurred
outside the gauge length of the specimen. Rotation occurred outside the gauge
length due to a combination of the deformation of the devices that serve to hold the
specimen in place and the rotation that occurred at the ‘fixed’ end, where some
movement was necessary in order to measure the torque. Consequently, the
uncorrected experimental results appeared to show that the specimens were
significantly less stiff than predicted by torsion theory and FE analysis. The circular
bars tested in order to calibrate the torque measurement showed a similar error
(Table A—3), in this case due entirely to the rotation at the ‘fixed’ end where the
torque is transmitted to the mechanical torque measuring device. Since the elastic
portion of these two experimental measurements is linear (Figure A—2), it can be
5
See Fenner (1965) for a description of an optical twist meter.
Rectangular Hollow Sections with circular web openings XIV
deduced that the rotation of the torque measuring device is proportional to the
torque.
A retrospective correction was applied to the original experimental results to
remove the error due to the rotational flexibility of the testing apparatus. The factor
of correction was obtained by measuring the flexibility experimentally, as described
below.
If the rotation stiffness of the testing apparatus (Kends) is assumed to be
elastic and proportional to the applied torque (T) then the total twist (θ) can be
described by Equation A—4. In the equation, L is the gauge length, G is the shear
modulus of the steel and It is the torsional inertia constant of the material.
TL T
θ= +
I t G K ends
Equation A—4
Since it was not possible to test a specimen with a zero gauge length, the
stiffness of the testing apparatus was obtained by testing specimens of various gauge
lengths. Angles of twist for different lengths were compared at different values of
elastic torque and, at each, a value of Kends was obtained by a least squares fit (Table
A—4). The final value of Kends was obtained by taking the average of the
measurements of Kends at the different torques.
The results were a good match with the model described by Equation A—4,
and this was borne out by the strong linear fit with the experimental results
(correlation coefficient = 0.93). However, the retrospective altering of results is an
additional source of error that is best avoided whenever possible. Ideally, the twist
over the gauge length would have been measured directly, but this would have
Rectangular Hollow Sections with circular web openings XV
required a complicated experimental set-up, and unfortunately, was not a practical
approach for this investigation.
The elastic torsional stiffness of the cold-formed tube was very difficult to
measure experimentally. This was due partly to the errors in the measurement of
twist, but also due to difficulties in reading the torque scale accurately when the
needle was moving, and the difficulties in measuring the Young’s modulus of the
thin tube material. Following the application of the correction factor, the
experimental measurements of torsional stiffness were 50% higher than the
theoretical predictions (Figure A—4). However, comparisons of the corrected
experimental results and the FE predictions for the perforated specimens were
extremely good (Section 6.4.3).
Table A—4: Stiffness of the testing machine (Kends)

Torque (Nm) Twist at the ends (o) Kends (Nm/o)
100 0.4361 229.31
200 0.8722 229.31
300 1.3084 229.29
400 1.7445 229.29
500 2.1806 229.29
600 2.6167 229.30
Average 229.30
Rectangular Hollow Sections with circular web openings XVI
180
160
Elastic stiffness (Nm /deg), GI t
140 Average experimental value
120
2
100
80 Theoretical value
60
40
20
0
0 50 100 150 200 250 300 350
Gauge length (mm), L
Figure A—4: Experimental measurements of elastic torsional stiffness
Rectangular Hollow Sections with circular web openings XVII
B. Sensitivity of sectional properties
The following figures (Figure B—1 through Figure B—13) illustrate the
sensitivity of various sectional properties (unperforated) to small changes in the
dimensions of the cross-section and the properties of the material.
The figures show that accurate measurements of the principal dimensions
(width, height and thickness) are necessary for accurate calculation of the sectional
properties. The dimensions of the corners are much less important, and accurate
measurements (which are difficult to make) are not necessary for accurate
calculation of the sectional properties. However, the internal radius of the corner is
very important when calculating stress concentrations in the case of torsion (Section
2.5).
Of all the principal material properties, the yield stress is the most important
as it has a direct influence on all the sectional properties. In this investigation, the
yield stress was measured accurately using multiple direct tensile tests on coupons in
accordance with BS 10002-1:1990 (Appendix D). Young’s modulus is more
difficult to measure accurately in a tensile test, but only influences stiffness.
Poisson’s ratio is very difficult to measure, but only influences torsional stiffness
slightly. In this investigation, the Poisson’s ratio was assumed to be 0.3 (a typical
value for steel) and no attempt was made to measure it directly.
Rectangular Hollow Sections with circular web openings XVIII
Sensitivity to width (RHS 200x100x8 grade S275J2H)
8
Moment at first yield
6 Curvature at first yield
Fully plastic moment
4 Pure shear capacity
Change in property (%)
Rotation at yield
2 Torque at yield
Theoretical values
0
calculated from formulae
-6 -4 -2 0 2 4 6 in EN10210-2:1997.
-2
Shear capacity calculated
-4 in accordance with
BS5950 & EC3.
-6
Positive values are
-8 increases.
Change in section width (%)
Figure B—1: Sensitivity to section width (RHS 200x100x8)
Sensitivity to width (RHS 250x250x10 grade S275J2H)
8
4 Pure shear capacity
Rotation at yield
2 Torque at yield
Theoretical values
0
-6 -4 -2 0 2 4 6 in EN10210-2:1997.
-2
-4 in accordance with
BS5950 & EC3.
-6
Positive values are
-8 increases.
Change in section width (%)
Figure B—2: Sensitivity to section width (RHS 250x250x10)
Rectangular Hollow Sections with circular web openings XIX
Sensitivity to height (RHS 200x100x8 grade S275J2H)
10
6
Pure shear capacity
4 Rotation at yield
2 Torque at yield
Theoretical values
0
-6 -4 -2 -2 0 2 4 6 in EN10210-2:1997.
-4
in accordance with
-6 BS5950 & EC3.
-8
Positive values are
-10 increases.
Change in section height (%)
Figure B—3: Sensitivity to section height (RHS 200x100x8)
Sensitivity to height (RHS 250x250x10 grade S275J2H)
10
6
Pure shear capacity
4 Rotation at yield
2 Torque at yield
Theoretical values
0
-6 -4 -2 -2 0 2 4 6 in EN10210-2:1997.
-4
in accordance with
-6 BS5950 & EC3.
-8
Positive values are
-10 increases.
Change in section height (%)
Figure B—4: Sensitivity to section height (RHS 250x250x10)
Rectangular Hollow Sections with circular web openings XX
Sensitivity to thickness (RHS 200x100x8 grade S275J2H)
10
6
Pure shear capacity
4 Rotation at yield
2 Torque at yield
Theoretical values
0
-10 -5 -2 0 5 10 in EN10210-2:1997.
-4
in accordance with
-6 BS5950 & EC3.
-8
Positive values are
-10 increases.
Change in thickness (%)
Figure B—5: Sensitivity to section thickness (RHS 200x100x8)
Sensitivity to thickness (RHS 250x250x10 grade S275J2H)
10
6
Pure shear capacity
4 Rotation at yield
2 Torque at yield
Theoretical values
0
-10 -5 -2 0 5 10 in EN10210-2:1997.
-4
in accordance with
-6 BS5950 & EC3.
-8
Positive values are
-10 increases.
Change in thickness (%)
Figure B—6: Sensitivity to section thickness (RHS 250x250x10)
Rectangular Hollow Sections with circular web openings XXI
Sensitivity to external corner radius (RHS 200x100x8 grade S275J2H)
1.2
1.0 Curvature at first yield
0.8 Fully plastic moment
0.6 Pure shear capacity
0.4 Rotation at yield

Torque at yield
0.2
Theoretical values
0.0
-10 -5 -0.2 0 5 10 in EN10210-2:1997.
-0.4 Shear capacity calculated
-0.6 in accordance with
BS5950 & EC3.
-0.8
-1.0 Positive values are
-1.2 increases.
Change in external radius (%)
Figure B—7: Sensitivity to external corner radius (RHS 200x100x8)
Sensitivity to external corner radius (RHS 250x250x10 grade S275J2H)
1.2
0.8 Fully plastic moment
0.4 Rotation at yield

Torque at yield
0.2
Theoretical values
0.0
-10 -5 -0.2 0 5 10 in EN10210-2:1997.
-0.4 Shear capacity calculated
BS5950 & EC3.
-0.8
-1.0 Positive values are
-1.2 increases.
Change in external radius (%)
Figure B—8: Sensitivity to external corner radius (RHS 250x250x10)
Rectangular Hollow Sections with circular web openings XXII
Sensitivity to internal corner radius (RHS 200x100x8 grade S275J2H)
0.20
Rotation at yield
0.05 Torque at yield
Theoretical values
0.00
-10 -5 0 5 10 in EN10210-2:1997.
-0.05
BS5950 & EC3.
-0.15
Positive values are
-0.20 increases.
Change in internal radius (%)
Figure B—9: Sensitivity to internal corner radius (RHS 200x100x8)
Sensitivity to internal corner radius (RHS 250x250x10 grade S275J2H)
0.20
Rotation at yield
0.05 Torque at yield
Theoretical values
0.00
-10 -5 0 5 10 in EN10210-2:1997.
-0.05
BS5950 & EC3.
-0.15
Positive values are
-0.20 increases.
Change in internal radius (%)
Figure B—10: Sensitivity to internal corner radius (RHS 250x250x10)
Rectangular Hollow Sections with circular web openings XXIII
Sensitivity to yield stress (any section grade S275J2H)
6
All
Curvature at first yield
4 Fully plastic moment
Pure shear capacity
2 Rotation at yield
Torque at yield
Theoretical values
0
-6 -4 -2 0 2 4 6 in EN10210-2:1997.
-2 Shear capacity calculated
in accordance with
BS5950 & EC3.
-4
Positive values are
-6 increases.
Change in yield stress (%)
Figure B—11: Sensitivity to yield stress (any section)
Sensitivity to Young's modulus (any section)
15
Pure shear capacity
5 Rotation at yield
Torque at yield
All others
Theoretical values
0
-10 -5 0 5 10 in EN10210-2:1997.
in accordance with
Rotation at yield BS5950 & EC3.
-10
& curvature at yield
Positive values are
-15 increases.
Change in Young's modulus (%)
Figure B—12: Sensitivity to Young’s modulus (any section)
Rectangular Hollow Sections with circular web openings XXIV
Sensitivity to Poisson's ratio (any section)
3
Rotation at yield only
Pure shear capacity
1 Rotation at yield
Torque at yield
All others
Theoretical values
0
-10 -5 0 5 10 in EN10210-2:1997.
in accordance with
BS5950 & EC3.
-2
Positive values are
-3 increases.
Change in section Poisson's ratio (%)
Figure B—13: Sensitivity to Poisson’s ratio (any section)
Rectangular Hollow Sections with circular web openings XXV
C. Formulae for the calculation of the properties of
cross-section
The following formulae for geometric properties are given in Annex A of BS
EN 10210-2:1997. Some of the symbols have been changed to maintain consistency
with the main body of this Thesis. Dimensions of the cross-section are as defined in
Figure 1-9 and Figure 1-10.
Nominal corner radii
ro = 1.5t
Equation C—1
ri = 1.0t
Equation C—2
Cross-sectional area
(
A = 2t (b + h − 2t ) − (4 − π ) ro2 − ri 2 )
Equation C—3
Second moment of area (major axis)
bh 3 (b − 2t ) (h − 2t )
3
Iy =
12
−
12
( ) (
− 4 I ζζ + Aζ hζ2 + 4 I ξξ + Aξ hξ2 )
Equation C—4
Rectangular Hollow Sections with circular web openings XXVI
Radius of gyration (major axis)
Iy
iy =
A
Equation C—5
Elastic section modulus (major axis)
2I y
Wel , y =
h
Equation C—6
Plastic section modulus (major axis)
bh 2 (b − 2t ) (h − 2t )
2
W pl , y = − − 4 Aζ hζ + 4 Aξ hξ
4 4
Equation C—7
Torsional inertia constant
hc
It = t 3 + 2 KAh
3
Equation C—8
Note: The above equation for torsional inertia constant is the same as the equation
given in Appendix B of BS 5950-1:1990 (lateral torsional buckling) with an
allowance for the radii of the corners.
Torsional modulus constant
It
Ct =
t+K t
Equation C—9
Rectangular Hollow Sections with circular web openings XXVII
Where
Aζ = (1 − π 4 ) ro2
Equation C—10
Aξ = (1 − π 4 ) ri 2
Equation C—11
h  10 − 3π 
hζ = − ro
2  12 − 3π 
Equation C—12
h − 2t  10 − 3π 
hξ = − ri
2  12 − 3π 
Equation C—12
1 π 1  4
Iζζ =  − − ro
 3 16 3(12 − 3π ) 
Equation C—13
1 π 1  4
Iξξ =  − − ri
 3 16 3(12 − 3π ) 
Equation C—14
Mean perimeter
hc = 2((b − t ) + (h − t )) − 2 Rc (4 − π )
Equation C—15
Rectangular Hollow Sections with circular web openings XXVIII
Area enclosed by mean perimeter
Ah = (b − t ) (h − t ) − Rc2 (4 − π )
Equation C—16
2 Aht
K=
hc
Equation C—17
ro + ri
Rc =
2
Equation C—18
Roark (Young 1989) gives an alternative formula for calculation of the
torsional inertia constant for an RHS with sharp re-entrant corners (Equation C—
19). The formula is strictly only accurate for infinitely thin sections and, for typical
RHS, is between 5% and 1% lower than the BS EN 10210 value (Figure 2-21).
2t (b − t ) (h − t )
2 2
I t ,no radii , Roark =
b + h − 2t
Equation C—19
Roark (Young 1989) also gives an alternative formula for calculation of the
torsional modulus constant for an RHS with sharp re-entrant corners (Equation C—
20). The formula is only accurate for infinitely thin sections and, for typical RHS, is
between 5% and 25% higher than the BS EN 10210 value (Figure 2-22).
C t ,no radii = 2t (h − t ) (b − t )
Equation C—20
Rectangular Hollow Sections with circular web openings XXIX
D. Material properties
D.1 Introduction
The properties of the nine hot finished bars, used in the full-scale testing
component of this research, are summarised in Tables D—1 to D—27. The
properties were measured by uniaxial tensile testing in accordance with BS EN
10002–1:1990, with the exception of the four coupons indicated below. Material
models for the bars were determined by averaging the results of at least four
coupons from each (two from each end of the bar), using the scheme outlined in
Section 2.2.
One coupon was tested at a very low rate of strain (0.5% strain per minute)
in order to verify that the yield stress and subsequent plateau were unaffected by a
slow strain rate. No differences were observed in the stress-strain behaviour of the
coupon with the low strain rate, and the other coupons taken from the same bar (See
coupon code 200b in Table D—16).
Three coupons with shortened parallel lengths were tested to confirm that the
steel in the Rectangular Hollow Section (RHS) was isotropic. The parallel lengths
of the three coupons were inclined at 0.0o, 22.5o and 45.0o to the longitudinal axes of
the RHS. No differences were observed in the stress-strain behaviour of the three
coupons (See coupon codes Ang00, Ang22 and Ang 45 in Table D—19).
Rectangular Hollow Sections with circular web openings XXX
D.2 Material properties RHS 250x250x10 grade S275J2H bar number 1
Table D—1: Tensile test results for RHS 250x250x10 grade S275J2H bar #1
Coupon ID code B1 B2 B3 B4 B5 B6 Mean
[ ]
Parallel length 125 125 125 125 125 125 -
(mm)
Thickness 9.70 9.71 9.77 9.64 9.67 9.68 9.70
(mm) [0.5%]
Width 8.04 8.13 7.95 7.95 7.92 7.93 -
(mm)
Proportionality factor 5.66 5.63 5.67 5.71 5.71 5.71 -
Upper yield 349 359 349 364 402 363 364

2
(N/mm ) ReH [5.4%]
Lower yield 342 352 348 359 366 327 349
(N/mm2) ReL [3.9%]
Suitability of fit 0.15 0.14 0.20 0.15 0.21 0.23 0.18
(% energy error) K [21%]
Yield stress 342.2 357.1 346.6 351.4 362.0 327.1 347.7
2
(N/mm ) fy1 [3.6%]
Young’s modulus 211.0 210.8 202.3 212.5 208.2 215.1 210.0
2
(kN/mm ) E [2.1%]
Plateau ratio 9.5 9.1 16.3 26.0 13.8 13.8 14.8
λplateau [42%]
Post yield stiffness 0.61 -0.02 0.83 1.61 -0.51 0.59 0.52
(kN/mm2) Eplateau [140%]
Elliptical strain 0.21 0.17 0.27 0.42 0.14 0.21 0.24
Hardening factor λESHF [42%]
Tensile strength 461.2 469.6 463.5 476.7 460.9 450.7 463.8
2
(N/mm ) fu [1.9%]
Strain at maximum force 14.8 14.5 15.7 16.2 16.2 16.4 15.6
(%) εu [5.2%]
Strain at fracture 22.1 22.4 25.3 25.7 26.2 26.8 24.8
(%) [8.1%]
Note 1: Refer to Section 2.2
Note 2: denotes the standard deviation as a percentage of the mean value
Rectangular Hollow Sections with circular web openings XXXI
Table D—2: Material model for RHS 250x250x10 grade S275J2H bar #1
2 2 2 2
Grade (%) (N/mm ) (kN/mm ) (kN/mm ) (N/mm ) (%)
RHS 250x250x10 1.48 347.7 210.0 14.8 0.52 0.24 463.8 15.6
grade S275J2H
Table D—3: ABAQUS material model (RHS 250x250x10 grade S275J2H bar #1)
*ELASTIC
210520.0 , 0.3
*PLASTIC
348.28 , 0
369.77 , 0.023990
392.82 , 0.030203
435.08 , 0.048735
487.43 , 0.084932
536.24 , 0.142637
565.22 , 0.195143
Rectangular Hollow Sections with circular web openings XXXII
Coupon ID code La Lb Loe1 Loe2 Low1 Low2 Mean
[ ]
Parallel length 125 125 125 125 125 125 -
(mm)
Thickness 9.66 9.66 9.66 9.57 9.74 10.17 9.74
(mm) [2.2%]
Width 7.86 7.91 7.89 7.93 7.84 7.97 -
(mm)
Upper yield 365 335 374 384 412 345 369

2
(N/mm ) ReH [7.6%]
Lower yield 331 320 351 358 357 334 342
(N/mm2) ReL [4.6%]
Yield stress 327.9 323.2 356.3 366.0 357.3 328.9 343.2
2
(N/mm ) fy1 [5.4%]
Young’s modulus 199.7 199.6 210.1 198.0 202.5 175.9 197.6
2
(kN/mm ) E [5.8%]
Plateau ratio 15.3 12.5 9.2 11.4 16.0 13.0 12.9
λplateau [19%]
Post yield stiffness 0.36 0.19 0.47 -0.56 0.18 0.61 0.21
2
(N/mm ) fu [2.6%]
(%) εu [3.8%]
(%) [5.9%]
Rectangular Hollow Sections with circular web openings XXXIII
2 2 2 2
RHS 250x250x10 3.16 343.2 197.6 12.9 0.21 0.22 468.0 15.9
grade S275J2H
*ELASTIC
198159.6 , 0.3
*PLASTIC
343.84 , 0
356.30 , 0.022041
383.54 , 0.028487
431.47 , 0.047738
489.69 , 0.085326
542.65 , 0.145171
572.49 , 0.198557
Rectangular Hollow Sections with circular web openings XXXIV
Coupon ID code Lne1 Lne2 Lnw1 Lnw2 Mean
[ ]
Parallel length 125 125 125 125 -
(mm)
Thickness 10.16 10.09 10.46 10.17 10.22
(mm) [1.6%]
Width 7.99 8.11 8.00 7.97 -
(mm)
Proportionality factor 5.55 5.53 5.47 5.55 -
Upper yield 383 368 375 388 378

2
(N/mm ) ReH [2.4%]
Lower yield 355 352 362 374 361
(N/mm2) ReL [2.8%]
Suitability of fit 0.25 0.22 0.18 0.27 0.23
Yield stress 360.9 350.2 361.1 384.2 364.1
2
(N/mm ) fy1 [3.9%]
Young’s modulus 210.1 200.0 206.2 199.5 204.0
2
(kN/mm ) E [2.5%]
Plateau ratio 10.6 9.8 8.9 14.7 11.0
λplateau [24%]
Post yield stiffness -0.44 0.67 0.82 0.17 0.31
Elliptical strain 0.28 0.30 0.29 0.20 0.27
Tensile strength 488.8 486.8 493.5 491.1 490.0
2
(N/mm ) fu [0.6%]
Strain at maximum force 15.0 14.6 14.5 15.9 15.0
(%) εu [4.3%]
Strain at fracture 24.0 24.4 26.1 27.7 25.6
(%) [6.7%]
Rectangular Hollow Sections with circular web openings XXXV
2 2 2 2
RHS 250x250x10 1.23 364.1 204.0 11.0 0.31 0.27 490.0 15.0
grade S275J2H
*ELASTIC
204505.8 , 0.3
*PLASTIC
364.75 , 0
378.05 , 0.019353
403.20 , 0.025509
450.97 , 0.043881
510.10 , 0.079796
563.60 , 0.137105
593.03 , 0.187853
Rectangular Hollow Sections with circular web openings XXXVI
Coupon ID code A1 A3 A4 Ca Cb S Mean
[ ]
Parallel length 85 85 85 85 85 85 -
(mm)
Thickness 7.70 7.70 7.81 7.89 7.84 7.69 7.77
(mm) [1.1%]
Width 9.87 9.98 10.07 10.05 9.98 9.90 -
(mm)
Upper yield 377 363 352 377 448 346 377

2
(N/mm ) ReH [9.8%]
Lower yield 346 345 340 336 350 344 343
(N/mm2) ReL [1.4%]
Yield stress 341.1 336.6 371.7 374.7 358.7 349.7 355.4
2
(N/mm ) fy1 [4.4%]
Young’s modulus 185.7 205.9 193.6 205.2 204.6 202.0 199.5
2
(kN/mm ) E [4.1%]
Plateau ratio 16.7 14.0 10.6 13.4 12.3 13.2 13.4
λplateau [15%]
Post yield stiffness 1.11 0.54 -1.66 -0.64 -0.53 0.78 -0.07
(kN/mm2) Eplateau [high]
Hardening factor λESHF [high]
2
(N/mm ) fu [1.5%]
(%) εu [1.6%]
(%) [3.0%]
Rectangular Hollow Sections with circular web openings XXXVII
2 2 2 2
RHS 200x100x8 1.21 355.4 199.5 13.4 -0.07 0.17 477.4 16.2
grade S275J2H
*ELASTIC
200037.3 , 0.3
*PLASTIC
356.03 , 0
362.90 , 0.023456
393.12 , 0.029921
442.24 , 0.049268
500.97 , 0.087044
554.64 , 0.147170
585.53 , 0.201217
Rectangular Hollow Sections with circular web openings XXXVIII
Coupon ID code A2 Soe1 Soe2 Sow1 Sow2 Mean
[ ]
Parallel length 85 125 125 125 125 -
(mm)
Thickness 7.85 7.66 7.73 7.73 7.66 7.73
(mm) [1.0%]
Width 10.01 9.89 9.95 9.87 9.93 -
(mm)
Proportionality factor 5.64 5.74 5.70 5.72 5.73 -
Upper yield 405 362 342 339 335 356

2
(N/mm ) ReH [8.2%]
Lower yield 351 343 339 334 325 338
(N/mm2) ReL [2.9%]
Suitability of fit 0.29 0.23 0.23 0.34 0.28 0.27
Yield stress 345.3 325.4 343.1 337.8 323.7 335.1
2
(N/mm ) fy1 [3.0%]
Young’s modulus 181.8 180.6 183.5 162.8 171.6 182.0
2 (ignored) (ignored)
(kN/mm ) E [0.8%]
Plateau ratio 12.1 12.3 7.2 12.9 13.7 11.7
λplateau [22%]
Post yield stiffness -0.08 0.26 0.65 0.53 1.16 0.50
Elliptical strain 0.09 0.20 0.26 0.20 0.28 0.21
Tensile strength 471.1 457.9 486.9 482.5 478.9 475.5
2
(N/mm ) fu [2.4%]
Strain at maximum force 16.1 16.1 15.0 16.7 16.2 16.0
(%) εu [3.9%]
Strain at fracture 28.3 28.3 26.2 26.7 26.9 27.3
(%) [3.5%]
Note 2: Values of Young’s modulus for coupons Sow1 and Sow2 ignored (extensometer fault)
Rectangular Hollow Sections with circular web openings XXXIX
2 2 2 2
RHS 200x100x8 2.53 335.1 182.0 11.7 0.50 0.21 475.5 16.0
grade S275J2H
*ELASTIC
182458.1 , 0.3
*PLASTIC
335.67 , 0
353.91 , 0.021088
383.61 , 0.027588
434.77 , 0.047034
496.30 , 0.085019
551.62 , 0.145492
582.07 , 0.199061
Rectangular Hollow Sections with circular web openings XL
Coupon ID code Sne1 Sne2 Snw1 Snw2 200a 200b Mean
[ ]
Parallel length 125 125 125 125 125 125 -
(mm)
Thickness 7.73 7.79 7.79 7.83 7.88 7.88 7.82
(mm) [0.8%]
Width 9.91 9.91 9.85 9.90 12.48 12.46 -
(mm)
Upper yield 347 344 344 355 348 350 348

2
(N/mm ) ReH [1.2%]
Lower yield 344 337 341 341 321 313 333
(N/mm2) ReL [3.8%]
Yield stress 344.2 338.8 344.1 350.5 327.6 319.9 337.5
2
(N/mm ) fy1 [3.4%]
Young’s modulus 198.2 192.3 187.9 202.5 181.9 186.5 191.6
2
(kN/mm ) E [4.0%]
Plateau ratio 14.8 14.1 11.7 14.1 13.7 12.8 13.5
λplateau [8.1%]
Post yield stiffness 1.01 1.63 0.34 -0.21 -0.12 -0.20 0.42
2
(N/mm ) fu [2.1%]
(%) εu [4.4%]
(%) [5.9%]
Note 2: Coupon 200b tested at low strain rate (0.5% strain per minute)
Rectangular Hollow Sections with circular web openings XLI
2 2 2 2
RHS 200x100x8 0.96 344.4 195.2 13.7 0.71 0.25 476.2 16.2
grade S275J2H
*ELASTIC
195757 , 0.3
*PLASTIC
344.97 , 0
370.77 , 0.023636
395.89 , 0.030102
442.39 , 0.049390
499.98 , 0.087039
553.14 , 0.146972
583.92 , 0.200968
Rectangular Hollow Sections with circular web openings XLII
D.8 Material properties RHS 150x150x6.3 grade S275J2H bar number 1
Table D—19: Tensile test results for RHS 150x150x6.3 grade S275J2H bar #1
Coupon ID code 150a 150b 150c 150d 150e 150f Ang00 Ang22 Ang45 Mean
[ ]
Parallel length 125 125 125 125 125 125 65 65 65 -
(mm)
Thickness 6.08 6.11 6.10 6.07 6.09 6.10 5.96 5.95 5.94 6.04
(mm) [1.2%]
Width 12.44 12.38 12.41 12.42 12.41 12.44 12.49 12.49 12.47 -
(mm)
Proportionality factor 5.75 5.75 5.75 5.76 5.75 5.74 5.80 5.80 5.81 -
Upper yield 368 372 381 328 315 353 339 332 330 347
2
(N/mm ) ReH [6.6%]
Lower yield 329 336 328 308 312 316 323 321 324 322
(N/mm2) ReL [2.8%]
Suitability of fit 0.29 0.37 0.37 0.33 0.28 0.43 0.25 0.26 0.24 0.31
Yield stress 339.6 345.8 344.0 315.9 312.6 335.0 325.3 321.8 323.8 329.3
2
(N/mm ) fy1 [3.7%]
Young’s modulus 196.3 196.2 188.3 191.0 208.0 208.1 208.5 189.6 186.8 191.9
2
(kN/mm ) E [4.1%]
Plateau ratio 15.8 19.6 18.1 16.4 15.1 15.1 11.7 13.8 12.6 15.3
λplateau [16%]
Post yield stiffness -0.26 -0.19 -0.35 0.28 0.38 -0.82 -0.08 0.24 0.03 -0.09
Elliptical strain 0.11 0.15 0.11 0.11 0.14 0.20 0.14 0.18 0.17 0.15
hardening factor λESHF [22%]
Tensile strength 456.3 462.3 459.2 451.6 454.9 457.8 469.8 468.5 470.7 461.2
2
(N/mm ) fu [1.5%]
Strain at maximum force 17.5 18.5 18.7 18.1 17.5 17.1 19.1 19.0 19.2 18.3
(%) εu [4.3%]
Strain at fracture 29.2 29.5 29.1 29.0 28.4 28.0 32.9 33.4 32.7 30.3
(%) [7.0%]
Note 2: Coupon Ang00 inclined at 0.0o to longitudinal axis of tube
Rectangular Hollow Sections with circular web openings XLIII
Table D—20: Material model for RHS 150x150x6.3 grade S275J2H bar #1
2 2 2 2
RHS 150x150x6.3 1.02 329.3 195.2 16.7 -0.16 0.14 457.0 17.9
grade S275J2H
Table D—21: ABAQUS material model (RHS 150x150x6.3 grade S275J2H bar #1)
*ELASTIC
195678.2 , 0.3
*PLASTIC
329.86 , 0
336.62 , 0.027677
369.97 , 0.034711
421.48 , 0.055755
482.65 , 0.096743
538.72 , 0.161711
571.40 , 0.220436
Rectangular Hollow Sections with circular web openings XLIV
Coupon ID code Tt5 Tt6 Tt7 Tt8 Mean
[ ]
(mm)
Thickness 8.00 7.73 7.95 8.03 7.93
(mm) [1.7%]
Width 10.06 10.10 10.05 10.11 -
(mm)
Upper yield 384 380 395 377 384

2
(N/mm ) ReH [2.1%]
Lower yield 371 372 371 369 371
(N/mm2) ReL [0.3%]
Yield stress 381.3 380.9 382.7 382.7 381.9
2
(N/mm ) fy1 [0.3%]
Young’s modulus 219.6 211.6 200.6 200.7 208.1
2
(kN/mm ) E [4.5%]
Plateau ratio 7.5 8.9 6.4 5.7 7.1
λplateau [19%]
Post yield stiffness 0.23 0.20 -0.37 -0.41 -0.09
hardening factor λESHF [8.8%]
2
(N/mm ) fu [0.7%]
(%) εu [1.3%]
(%) [4.5%]
Rectangular Hollow Sections with circular web openings XLV
2 2 2 2
RHS 200x100x8 0.71 380.9 203.1 7.0 -0.09 0.24 523.7 12.9
grade S355J2H
*ELASTIC
203631.8 , 0.3
*PLASTIC
381.65 , 0
385.43 , 0.013002
415.80 , 0.018471
470.63 , 0.034870
535.91 , 0.067075
591.45 , 0.118765
636.53 , 0.192000
Rectangular Hollow Sections with circular web openings XLVI
D.10 Material properties RHS 150x150x6.3 grade S355J2H bar number 1
Table D—25: Tensile test results for RHS 150x150x6.3 grade S355J2H bar #1
Coupon ID code Tt1 Tt2 Tt3 Tt4 Mean
[ ]
(mm)
Thickness 5.78 6.05 5.69 5.65 5.79
(mm) [3.1%]
Width 13.02 12.77 12.66 12.64 -
(mm)
Upper yield 411 404 414 413 411

2
(N/mm ) ReH [1.1%]
Lower yield 403 396 404 401 401
(N/mm2) ReL [0.9%]
Yield stress 406.0 405.7 405.8 401.2 404.7
2
(N/mm ) fy1 [0.6%]
Young’s modulus 172.5 230.8 199.9 205.2 202.1
2
(kN/mm ) E [12%]
Plateau ratio 6.9 9.2 6.8 5.8 7.2
λplateau [20%]
Post yield stiffness 0.01 -0.29 0.01 1.05 0.20
Hardening factor λESHF [2.6%]
2
(N/mm ) fu [0.5%]
(%) εu [3.0%]
(%) [8.3%]
Rectangular Hollow Sections with circular web openings XLVII
Table D—26: Material model for RHS 150x150x6.3 grade S355J2H bar #1
2 2 2 2
RHS 150x150x6.3 0.98 404.7 199.9 7.1 0.13 0.34 544.7 13.1
grade S355J2H
Table D—27: ABAQUS material model (RHS 150x150x6.3 grade S355J2H bar #1)
*ELASTIC
200524.8 , 0.3
*PLASTIC
405.50 , 0
413.24 , 0.014222
438.97 , 0.019725
491.95 , 0.036169
558.53 , 0.068438
616.18 , 0.120229
659.19 , 0.187487
Rectangular Hollow Sections with circular web openings XLVIII
D.11 Statistical analysis of material properties
In order to determine a suitable material model for use in the parametric
Finite Element study (Chapter 7), a statistical analysis of the tensile test results was
performed. Figures D—1 to D—6 show typical values of the quantities defined in
Section 2.2. Figures D—1 to D—5 are for grade S275J2H steel, while Figure D—6
also includes data for grade S355J2H steel. The material model used in the
parametric Finite Element study is shown in Table 7-3.
Additional data were added following tests on hot finished RHS provided
kindly by Moayad Omair (The School of the Built Environment, Coventry
University, UK), and analysis of tensile test data provided kindly by Tim Wilkinson
(The Department of Civil Engineering, University of Sydney, Australia).
Rectangular Hollow Sections with circular web openings XLIX
35
30.6
30 27.4
25
Frequency (%)
20
15
11.3
9.7
10
6.5 6.5
5 3.2 3.2
1.6
0
less than 0.15 to 0.20 to 0.25 to 0.30 to 0.35 to 0.40 to 0.45 to More
0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 than 0.50
Percentage energy error (%), K
Figure D—1: Typical values of percentage energy error
18 17.0
16 15.1
14 13.2 13.2
12 11.3
Frequency (%)
10 9.4
8
5.7 5.7 5.7
6
3.8
4
0
less 9 to 10 10 to 11 11 to 12 12 to 13 13 to 14 14 to 15 15 to 16 16 to 17 More
than 9 than 18
Plateau ratio, λ plateau
Figure D—2: Typical values of plateau ratio
Rectangular Hollow Sections with circular web openings L
35
29.0
30
25.8
25
21.0
Frequency (%)
20
15
11.3
10 8.1
5 3.2
1.6
0.0
0
less than - -1.50 to - -1.00 to - -0.50 to 0.00 to 0.50 to 1.00 to More than
1.50 1.00 0.50 0.00 0.50 1.00 1.50 1.50
2
Post yield stiffness (kN/mm ), E plateau
Figure D—3: Typical values of post yield stiffness
24 22.6 22.6 22.6

22
20
18
16 14.5
Frequency (%)
14
12
9.7
10
8
6
3.2
4
1.6 1.6 1.6
2 0.0
0
less 0.00 to 0.05 to 0.10 to 0.15 to 0.20 to 0.25 to 0.30 to 0.35 to More
than 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 than
0.00 0.40
Elliptical strain hardening factor, λ ESHF
Figure D—4: Typical values of elliptical strain hardening factor
Rectangular Hollow Sections with circular web openings LI
40 37.1
35
30
Frequency (%)
25
19.4
20
16.1
15
9.7 9.7
10
4.8
5 3.2
0
less than 14 14 to 15 15 to 16 16 to 17 17 to 18 18 to 19 More than
19
Strain at maximum force (%), ε u
Figure D—5: Typical values of strain at maximum force
4 RHS 200x100x8 Grade S275 RHS 250x250x10 Grade S275

RHS 150x150x6.3 Grade S275 RHS 80x40x3 Grade S275
RHS 80x80x4 Grade S275 (Omair) RHS 60x60x4 Grade S275 (Omair)
3 RHS 200x150x6.3 Grade S275 (Wilkinson) RHS 250x150x6.3 Grade S275 (Wilkinson)
RHS 200x100x8 Grade S355 RHS 150x150x6.3 Grade S355
Post yield stiffness (kN/mm )
2
0
0 5 10 15 20 25 30
-1
-2
Plateau ratio, λ plateau
Figure D—6: Plateau ratio and post yield stiffness
Rectangular Hollow Sections with circular web openings LII
E. Stress concentration factors (Finite Element)
Table E—1: FE stress concentration factors for bending (part 1 of 2)

Section Number of holes Hole size ratio, Φ Stress concentration factor, ψbending
FE Fit Difference (Fit – FE)
150x150x12.5 1 0.70 1.09 1.09 0.00
150x150x12.5 1 0.85 1.46 1.52 0.06
150x150x12.5 2 0.78 1.29 1.32 0.03
250x250x10 1 0.10 1.00 1.00 0.00
250x250x10 1 0.30 1.00 1.00 0.00
250x250x10 1 0.50 1.00 1.00 0.00
250x250x10 1 0.55 1.00 1.00 0.00
250x250x10 1 0.70 1.39 1.41 0.02
250x250x10 1 0.90 2.23 2.23 0.00
250x250x10 2 0.10 1.00 1.00 0.00
250x250x10 2 0.20 1.00 1.00 0.00
250x250x10 2 0.30 1.00 1.00 0.00
250x250x10 2 0.40 1.00 1.00 0.00
250x250x10 2 0.50 1.01 1.00 -0.01
250x250x10 2 0.60 1.10 1.00 -0.10
250x250x10 2 0.70 1.38 1.41 0.02
250x250x10 2 0.80 1.75 1.82 0.07
250x250x10 2 0.90 2.20 2.23 0.03
350x350x12 1 0.78 1.73 1.80 0.07
350x350x12 2 0.63 1.21 1.15 -0.07
350x350x12 2 0.69 1.39 1.41 0.02
350x350x12 2 0.85 2.05 2.11 0.06
350x350x12 2 0.90 2.30 2.33 0.03
120x80x8 1 0.55 1.00 1.00 0.00
120x80x8 1 0.85 1.67 1.67 0.00
120x80x8 2 0.70 1.21 1.19 -0.03
120x80x8 2 0.90 1.83 1.83 0.00
300x200x12 1 0.10 1.00 1.00 0.00
300x200x12 1 0.30 1.00 1.00 0.00
300x200x12 1 0.50 1.00 1.00 0.00
300x200x12 1 0.55 1.00 1.00 0.00
300x200x12 1 0.70 1.39 1.41 0.01
300x200x12 1 0.90 2.28 2.23 -0.05
Rectangular Hollow Sections with circular web openings LIII
Table E—2: FE stress concentration factors for bending (part 2 of 2)

Section Number of holes Hole size ratio, Φ Stress concentration factor, ψbending
300x200x12 2 0.10 1.00 1.00 0.00
300x200x12 2 0.30 1.00 1.00 0.00
300x200x12 2 0.50 1.01 1.00 -0.01
300x200x12 2 0.55 1.01 1.00 -0.01
300x200x12 2 0.58 1.05 1.00 -0.05
300x200x12 2 0.70 1.40 1.41 0.01
300x200x12 2 0.90 2.27 2.23 -0.04
200x120x6 1 0.80 1.91 1.93 0.02
200x120x6 2 0.60 1.16 1.02 -0.14
200x120x6 2 0.70 1.47 1.47 0.00
200x120x6 2 0.85 2.17 2.15 -0.02
140x80x4 1 0.70 1.48 1.47 -0.01
140x80x4 2 0.90 2.51 2.39 -0.12
200x100x10 1 0.78 1.61 1.61 0.00
200x100x10 2 0.63 1.13 1.06 -0.08
200x100x10 2 0.69 1.31 1.28 -0.03
200x100x10 2 0.85 1.93 1.87 -0.06
200x100x10 2 0.90 2.16 2.06 -0.10
200x100x8 1 0.10 1.00 1.00 0.00
200x100x8 1 0.30 1.00 1.00 0.00
200x100x8 1 0.50 1.01 1.00 -0.01
200x100x8 1 0.70 1.41 1.41 -0.01
200x100x8 1 0.90 2.33 2.23 -0.10
200x100x8 2 0.10 1.00 1.00 0.00
200x100x8 2 0.20 1.00 1.00 0.00
200x100x8 2 0.30 1.00 1.00 0.00
200x100x8 2 0.40 1.00 1.00 0.00
200x100x8 2 0.50 1.01 1.00 -0.01
200x100x8 2 0.55 1.01 1.00 -0.01
200x100x8 2 0.60 1.11 1.00 -0.11
200x100x8 2 0.70 1.41 1.41 0.00
200x100x8 2 0.80 1.81 1.82 0.01
200x100x8 2 0.90 2.34 2.23 -0.11
Rectangular Hollow Sections with circular web openings LIV
Table E—3: FE stress concentration factors for shear (part 1 of 2)

Section Number of holes Hole size ratio, Φ Stress concentration factor, ψshear
250x250x16 2 0.50 3.62 3.70 0.09
250x250x16 2 0.80 7.56 7.20 -0.36
250x250x12.5 2 0.50 3.74 3.82 0.08
250x250x12.5 2 0.80 8.35 7.68 -0.67
250x250x10 1 0.10 2.36 2.24 -0.12
250x250x10 1 0.20 2.59 2.54 -0.05
250x250x10 1 0.30 2.84 2.90 0.06
250x250x10 1 0.40 3.22 3.36 0.14
250x250x10 1 0.50 3.76 3.96 0.19
250x250x10 1 0.60 4.51 4.77 0.26
250x250x10 1 0.70 5.67 5.99 0.32
250x250x10 1 0.80 7.66 8.22 0.55
250x250x10 1 0.90 10.97 14.37 3.39
250x250x10 2 0.10 2.36 2.24 -0.12
250x250x10 2 0.20 2.59 2.54 -0.05
250x250x10 2 0.30 2.86 2.90 0.05
250x250x10 2 0.40 3.25 3.36 0.11
250x250x10 2 0.50 3.83 3.96 0.12
250x250x10 2 0.60 4.68 4.77 0.09
250x250x10 2 0.70 6.13 5.99 -0.14
250x250x10 2 0.80 8.98 8.22 -0.76
250x250x10 2 0.90 14.37 14.37 -0.01
250x250x8 2 0.20 2.64 2.58 -0.06
250x250x8 2 0.50 3.91 4.13 0.22
250x250x8 2 0.60 4.78 5.02 0.24
250x250x8 2 0.70 6.34 6.39 0.04
250x250x8 2 0.80 9.50 8.89 -0.60
250x250x8 2 0.90 14.58 15.89 1.31
250x250x6 2 0.50 3.98 4.41 0.43
250x250x6 2 0.80 10.05 10.02 -0.03
300x200x12 1 0.10 2.42 2.24 -0.18
300x200x12 1 0.20 2.66 2.54 -0.12
300x200x12 1 0.30 2.91 2.90 -0.01
300x200x12 1 0.40 3.27 3.36 0.09
300x200x12 1 0.50 3.79 3.96 0.16
300x200x12 1 0.60 4.49 4.77 0.28
300x200x12 1 0.70 5.49 5.99 0.51
300x200x12 1 0.80 7.10 8.22 1.12
300x200x12 1 0.90 9.44 14.37 4.93
Rectangular Hollow Sections with circular web openings LV
Table E—4: FE stress concentration factors for shear (part 2 of 2)

Section Number of holes Hole size ratio, Φ Stress concentration factor, ψshear
300x200x12 2 0.10 2.42 2.24 -0.18
300x200x12 2 0.20 2.66 2.54 -0.12
300x200x12 2 0.30 2.93 2.90 -0.02
300x200x12 2 0.40 3.31 3.36 0.05
300x200x12 2 0.50 3.89 3.96 0.07
300x200x12 2 0.60 4.72 4.77 0.05
300x200x12 2 0.70 6.11 5.99 -0.11
300x200x12 2 0.80 8.88 8.22 -0.66
300x200x12 2 0.90 14.15 14.37 0.22
200x100x16 2 0.10 2.16 2.20 0.05
200x100x16 2 0.20 2.45 2.45 0.00
200x100x16 2 0.40 3.25 3.14 -0.11
200x100x16 2 0.50 3.54 3.62 0.08
200x100x16 2 0.70 5.15 5.20 0.06
200x100x16 2 0.80 6.71 6.86 0.16
200x100x16 2 0.90 9.31 11.32 2.01
200x100x12.5 2 0.50 3.72 3.71 0.00
200x100x12.5 2 0.70 5.56 5.43 -0.14
200x100x8 1 0.10 2.48 2.24 -0.24
200x100x8 1 0.20 2.73 2.54 -0.19
200x100x8 1 0.30 2.99 2.90 -0.09
200x100x8 1 0.40 3.36 3.36 0.00
200x100x8 1 0.50 3.90 3.96 0.06
200x100x8 1 0.60 4.62 4.77 0.15
200x100x8 1 0.70 5.72 5.99 0.28
200x100x8 1 0.80 7.35 8.22 0.87
200x100x8 1 0.90 9.88 14.37 4.49
200x100x8 2 0.10 2.48 2.24 -0.24
200x100x8 2 0.20 2.73 2.54 -0.19
200x100x8 2 0.30 2.99 2.90 -0.09
200x100x8 2 0.40 3.37 3.36 -0.01
200x100x8 2 0.50 3.95 3.96 0.01
200x100x8 2 0.60 4.77 4.77 0.00
200x100x8 2 0.70 6.12 5.99 -0.12
200x100x8 2 0.80 8.83 8.22 -0.62
200x100x8 2 0.90 15.06 14.37 -0.69
200x100x6 2 0.50 4.05 4.18 0.13
200x100x6 2 0.70 6.40 6.52 0.12
200x100x4 2 0.10 2.71 2.32 -0.39
200x100x4 2 0.20 2.89 2.71 -0.18
200x100x4 2 0.40 3.53 3.81 0.28
200x100x4 2 0.50 4.16 4.63 0.48
200x100x4 2 0.70 6.70 7.57 0.87
200x100x4 2 0.80 10.24 10.92 0.68
200x100x4 2 0.90 20.46 20.45 -0.01
Rectangular Hollow Sections with circular web openings LVI
Table E—5: FE stress concentration factors for torsion (part 1 of 2)

Section Number of holes Hole size ratio, Φ Stress concentration factor, ψtorsion
250x250x10 1 0.10 1.85 1.97 0.11
250x250x10 1 0.30 2.39 2.27 -0.11
250x250x10 1 0.50 3.19 3.17 -0.02
250x250x10 1 0.70 5.22 5.08 -0.14
250x250x10 1 0.90 12.17 11.74 -0.44
250x250x10 2 0.10 1.85 1.97 0.11
250x250x10 2 0.30 2.39 2.27 -0.12
250x250x10 2 0.50 3.24 3.17 -0.07
250x250x10 2 0.70 5.35 5.08 -0.27
250x250x10 2 0.90 12.76 11.74 -1.03
38x38x1.43 1 0.03 1.72 1.98 0.26
38x38x1.43 1 0.06 1.93 1.97 0.04
38x38x1.43 1 0.15 2.22 2.00 -0.21
38x38x1.43 1 0.30 2.55 2.29 -0.26
38x38x1.43 1 0.44 3.14 2.90 -0.24
38x38x1.43 1 0.59 4.15 3.97 -0.19
38x38x1.43 1 0.74 6.28 5.91 -0.37
38x38x1.43 1 0.89 12.29 11.40 -0.89
38x38x1.43 2 0.03 1.72 1.98 0.26
38x38x1.43 2 0.06 1.93 1.97 0.04
38x38x1.43 2 0.15 2.22 2.00 -0.21
38x38x1.43 2 0.30 2.55 2.29 -0.26
38x38x1.43 2 0.44 3.15 2.90 -0.25
38x38x1.43 2 0.59 4.19 3.97 -0.23
38x38x1.43 2 0.74 6.30 5.91 -0.39
38x38x1.43 2 0.89 12.29 11.40 -0.89
250x250x6 2 0.10 2.19 2.04 -0.15
250x250x6 2 0.20 2.36 2.21 -0.15
250x250x6 2 0.30 3.04 2.54 -0.50
250x250x6 2 0.40 2.94 3.06 0.12
250x250x6 2 0.60 4.46 4.89 0.43
250x250x6 2 0.70 6.10 6.56 0.46
250x250x6 2 0.80 9.37 9.53 0.17
250x250x6 2 0.90 17.43 17.44 0.01
300x200x12 1 0.10 1.83 1.97 0.14
300x200x12 1 0.30 2.41 2.27 -0.14
300x200x12 1 0.50 3.08 3.17 0.10
300x200x12 1 0.70 5.03 5.08 0.05
300x200x12 1 0.90 11.25 11.74 0.49
300x200x12 2 0.10 1.83 1.97 0.14
300x200x12 2 0.30 2.35 2.27 -0.08
300x200x12 2 0.50 3.15 3.17 0.02
300x200x12 2 0.70 5.21 5.08 -0.13
300x200x12 2 0.90 11.74 11.74 -0.01
Rectangular Hollow Sections with circular web openings LVII
Table E—6: FE stress concentration factors for torsion (part 2 of 2)

Section Number of holes Hole size ratio, Φ Stress concentration factor, ψtorsion
250x150x8 2 0.20 2.26 2.11 -0.15
250x150x8 2 0.40 2.79 2.79 0.00
250x150x8 2 0.60 4.21 4.30 0.09
250x150x8 2 0.80 8.40 7.95 -0.46
250x150x8 2 0.90 13.93 13.87 -0.06
200x100x12.5 1 0.60 3.31 3.43 0.12
200x100x12.5 1 0.80 5.76 5.63 -0.13
200x100x12.5 2 0.10 1.84 1.93 0.09
200x100x12.5 2 0.20 2.02 1.97 -0.05
200x100x12.5 2 0.30 2.18 2.13 -0.05
200x100x12.5 2 0.40 2.38 2.41 0.03
200x100x12.5 2 0.50 2.71 2.83 0.12
200x100x12.5 2 0.70 4.20 4.28 0.08
200x100x12.5 2 0.85 6.41 6.74 0.33
200x100x12.5 2 0.90 7.41 8.66 1.25
200x100x8 1 0.10 1.82 1.97 0.14
200x100x8 1 0.30 2.34 2.27 -0.06
200x100x8 1 0.50 3.06 3.17 0.11
200x100x8 1 0.70 5.04 5.08 0.04
200x100x8 1 0.90 11.50 11.74 0.23
200x100x8 2 0.10 1.82 1.97 0.14
200x100x8 2 0.30 2.33 2.27 -0.06
200x100x8 2 0.50 3.12 3.17 0.05
200x100x8 2 0.70 5.13 5.08 -0.05
200x100x8 2 0.90 10.88 11.74 0.85
200x100x6 1 0.60 4.15 4.42 0.27
200x100x6 1 0.80 8.36 8.26 -0.09
200x100x6 2 0.50 3.38 3.49 0.11
200x100x6 2 0.90 13.94 14.59 0.64
Rectangular Hollow Sections with circular web openings LVIII
F. Calibration of testing equipment
F.1 Introduction
The load indicating devices of the testing machines used in the course of this
research project are calibrated annually in accordance with National Standards. The
classifications at the time of testing are listed in the following Sections.
The Instron machine was used in the three and four-point loading tests
(compression), and the Zwick machine was used for tensile testing of coupons. The
Denison machine was used to calibrate the load cells used in the full-scale torsion
tests. The accuracy of the Avery torsion testing machine used for the small-scale
tests is discussed in Appendix A.
F.2 The Instron testing machine
2000 kN Instron servo hydraulic testing machine with autoranging facility

Machine type 8500 Serial Number H0468
Verified in compression as a grade 0.5 testing machine from 2000 kN to 50 kN
In accordance with BS 1610-1:1992 and NAMAS Technical Policy Statement 10
Edition 2 (May 1996) using equipment calibrated to BS 1610-2:1985.
Rectangular Hollow Sections with circular web openings LIX
F.3 The Zwick testing machine
200 kN Zwick testing machine with autoranging facility

Machine type 1484 Serial Number 116431
Verified in tension as a grade 0.5 testing machine from 200 kN to 2 kN
In accordance with BS EN 10002-2:1992 and NAMAS Technical Policy Statement
10 Edition 2 (May 1996) using equipment calibrated to BS 1610-2:1985.
F.4 The Denison testing machine
3000 kN Denison compression testing machine with analogue scale

Machine type T60CT Serial Number 29411
Verified in tension as a grade 1.0 testing machine from 6 kN to 3000 kN
In accordance with BS 1610-1:1992 and NAMAS Technical Policy Statement 10
Edition 2 (May 1996) using equipment calibrated to BS 1610-2:1985 and BS EN
10002-3:1995
Rectangular Hollow Sections with circular web openings LX
References
Abramyan, B.L., “Torsion and bending of prismatic rods of hollow rectangular section”, National
Advisory Committee for Aeronautics, Technical Memorandum 1319, November 1951.
ASM, “Materials handbook”, 9th Edition, Volume 8 “Mechanical testing”, American Society for
Metals, 1985.
Brendt, R., “Kritische bemerkungen zur drehungselastizität”, Zeitschrift des Vereines Deutscher
Ingenieure, Vol. 40, pp 785-790 and pp 785-817, 1896.
Byrne, J.G., and Carré, B.A., “Torsional stress concentrations at rounded corners of Rectangular
Hollow Sections”, Journal of Mechanical Engineering Science, Vol. 4, No. 4, pp 334-340, 1962.
Cook, R.D, Malkus, D.S. and Plesha, M.E., “Concepts and applications of finite element analysis”,
Wiley, 3rd Edition, 1988.
Coulomb, C. A. de, “Histoire de l’Académie 1784”, Paris, pp 229-269, 1787.
Dally, J.W. and Riley, W.F., “Experimental stress analysis”, McGraw-Hill, 3rd Edition, 1991.
Dwyer, T.J. and Galambos, T.V., “Plastic Behaviour of Tubular Beam-Columns”, American
Society of Civil Engineers, Journal of the Structural Division, Vol. 91, No. ST4, pp 153-168, 1965.
El Darwish, I.A. and Johnson, B.G., “Torsion of structural shapes”, American Society of Civil
Engineers, Journal of the Structural Division, Vol. 91, No. ST1, pp 203-27, 1965.
Fenner, A.J., “Mechanical testing of materials”, George Newnes, 1965.
Griffith, A.A. and Taylor, G.I., “The use of soap films in solving torsion problems”, Technical
Reports of the British Advisory Committee for Aeronautics, Vol. 3, No. 333, pp 901-938, June 1917.
Rectangular Hollow Sections with circular web openings LXI
Hasan, S.W. and Hancock, G.J., “Plastic Bending Tests of Cold-Formed Rectangular Hollow
Sections”, Research Report, No. R586, School of Civil and Mining Engineering, The University of
Sydney, Australia, 1988.
Hohenmser, G.I, and Prager, W., “The mechanics of the plastic deformation of mild steel”, R.T.P.
Translation number 2468, British Ministry of Aircraft Production, 1932.
Huber, M.T., “Czasoprismo Technizne”, Lwóv, 1904.
Isida, M., “On the Bending of an Infinite Strip with an Eccentric Circular Hole”, Proceedings of the
2nd Japanese Congress of Applied Mechanics, 1952.
Joannides, F., Weller, A. and Gwynn, J., “Structural Steelwork, basic design to ENV 1993-1-
1:1992 (prestandard) Eurocode 3, Design of Steel Structures, Part 1.1 General rules and rules for
buildings”, Parthenon Press, Dublin, 1993.
Just, D.J. and Walley, J., “Torsion of solid and hollow rectangular beams”, American Society of
Civil Engineers, Journal of the Structural Division, Vol. 105, No. 9, pp 1789-1804, September 1979.
Kármán, T. von, and Chien, W.Z., “Torsion with variable twist”, J Aero Sci, Vol. 13, No. 10, pp
503-510, 1946.
Kazic, M. and Dong, S.B., “Analysis of restrained torsion”, American Society of Civil Engineers,
Journal of Engineering Mechanics, Vol. 116, No. 4, pp 870-891, April 1990.
Korol, R.M. and Hudoba, J., “Plastic Behaviour of Hollow Structural Sections”, American Society
of Civil Engineers, Journal of the Structural Division, Vol. 98, No. ST5, pp 1007-1023, 1972.
Lawson, R.M. and Mullett, D.L.,“Slim floor construction using deep decking”, Steel Construction
Institute, Ref. SCI-P-127, 1993.
Lawson, R.M., Mullett, D.L. and Rackham, J.W.,“Design of asymmetric Slimflor beams using
deep composite decking”, Steel Construction Institute, Ref. SCI-P-175, 1997.
Lode, W., Z. Physik, Vol 36, p 913, 1926.
Lode, W., “Der Einfluss der mittleren Hauptspannung auf das Fliessen der Metalle”. Forsch. –Arb.
Geb. Ing. –Wes., VDI-Verlag G.M.b.H, Berlin, Heft 303, 1928.
Rectangular Hollow Sections with circular web openings LXII
Lyse, I. and Johnson, B.G., “Structural beams in torsion”, Transactions of the American Society of
Civil Engineers, Vol. 101, pp 857-926, 1936.
Marciniak, Z., “Aspects of metal formability”, Metal working research group, McMaster University,
Ontario, Canada, pp 84-91, 1973.
Marin, J., “Working stresses for members subjected to compound loads”, Product Engineering, Vol
8 no 5, p 169, 1937.
Marshall, J., “Derivation of Torsional Formulas for Multiply Connected Thick-Walled Rectangular
Sections”, Journal of Applied Mechanics, pp 399-402, June 1970.
Marshall, J., “Torsional behaviour of structural rectangular hollow sections”, The Structural
Engineer, No. 8, Vol. 49, pp 375-379, August 1971.
Marshall, J., “Aspects of torsion of structural rectangular hollow sections”, Ph.D. thesis, University
of Strathclyde, 1972.
Megson, T.H.G., “Aircraft structures for engineering students”, Edward Arnold, 2nd Edition, 1990.
Mises, R. von, “Gottingen Nachrichten”, Math. – Phys. Klasse, p 582, 1913.
Miyauchi, K., “Stress-strain relationships in simple shear testing on in-plane deformation for
various steel sheets”, 13th Congress of the International deep drawing research group, Melbourne,
Australia, pp 360-371, February 1984.
Moore, R.L. and Paul, D.A., “Torsion tests of 24S-T Aluminum Alloy Noncircular Bar and Tubing”,
National Advisory Committee for Aeronautics, Technical Note 885, 1943.
Moore, R.L. and Paul, D.A., “Observations on the behavior of some noncircular aluminum alloy
sections loaded to failure in torsion”, National Advisory Committee for Aeronautics, Technical Note
1097, March 1947.
Moore, R.L. and Clark, J.W., “Torsion, compression and bending tests of tubular sections
machined from 75S-T6 rolled round rod”, National Advisory Committee for Aeronautics, Research
Memorandum 52125, November 1952.
Mullett, D.L., “Slim floor design and construction”, Steel Construction Institute, Ref. SCI-P-110,
1992.
Rectangular Hollow Sections with circular web openings LXIII
Mullett, D.L., Ogden, R.G. and Kerry, J., “Design of structural hollow sections for edge beams in
Slimflor construction”, Report to British Steel (Tubes and Pipes), Steel Construction Institute, Doc
RT481, Draft 01, 1995.
Mullett, D.L., “Design of RHS Slimflor edge beams”, Steel Construction Institute, Ref. SCI-P-169,
1997.
Nádai, A., “Plasticity – A mechanics of the state of matter”, 6th (English) edition, Engineering
societies monographs, McGraw-Hill, pp 93-96, 1931.
Navier, L. M. H., “Résumé des leçons données à l’école des ponts et chaussées sur l’application de
la mécanique à l‘établissement des constructions et des machines”, 1st Ed, Paris, 1826.
Navier, L. M. H., “Résumé des leçons données à l’école des ponts et chaussées sur l’application de
la mécanique à l‘établissement des constructions et des machines”, 3rd Ed, Paris, with notes and
appendices by Saint-Venant, 1864.
Nethercot, D.A., Salter, P.R. and Malik, A.S., “Design of members subjected to combined bending
and torsion”, Steel Construction Institute, Ref. SCI-P-057, 1989.
Nitzche, R.N. and Miller, R.E., “Torsion and flexure of closed thin-walled beams”, American
Society of Civil Engineers, Journal of the Engineering Mechanics Division, Vol. 102, No. 1, pp 189-
193, September 1979.
Peterson, R.E., “Stress Concentration Factors”, John Wiley and Sons, 1974.
Prandtl, L., “Zur torsion von prismatischen stäben”, Physik. Z. Vol. 4, p 758, 1903. See also
Anthes, Dinglers Polytech. J. p 342, 1906.
Ridley-Ellis, D.J., Davies, G. and Owen, J.S., “Fundamental behaviour of rectangular hollow
sections with web openings”, Presented at the 8th International Symposium on Tubular Structures,
Singapore, August 1998.
Roark, R.J., “Formulas for stress and strain”, McGraw-Hill, 1938.
Rockey, K.C, Evans, H.R, Griffiths, D.W. and Nethercot, D.A., “The finite element method - a
basic introduction for engineers”, Collins, 2nd Edition, 1986.
Saint-Venant, A. J. C. B. Count de, “De la Torsion des Prismes – Extrait du Tome XIV des
Mémoires Presentés par divers Savants a l’Académie des Sciences”, Vol. 14, 1855.
Rectangular Hollow Sections with circular web openings LXIV
Savin, G.N., “Stress concentration around holes”, Pergamon, 1961.
Schulz, M. and Filippou, F.C., “Generalized warping torsion formulation”, American Society of
Civil Engineers, Journal of Engineering Mechanics, Vol. 124, No. 3, pp 339-347, March 1998.
Sedlacek, G., Dahl, N., Stranghöner, B., Kalinowski, B., Rondal, J. and Boeraeve, Ph.,
“Investigation of the rotation behaviour of hollow section beams”, European Commission report
(technical steel research) EUR 17994, 1998.
Shanley, F.R., “Strength of Materials”, McGraw-Hill, 1957.
Smith, F.A., Thomas, F.M. and Smith, J.O., “Torsional analysis of heavy box beams in structures”,
American Society of Civil Engineers, Journal of the Structural Division, Vol. 96, No. ST3, pp 613-
635, 1970.
Stang, A.H., Ramberg, W. and Back, G., “Torsion tests of tubes”, National Advisory Committee
for Aeronautics, Technical Report 601, 1937.
Taylor, G.I. and Quinney, H., “The plastic distortion of metals”, Phil. Trans. Roy. Soc., Ser. A. Vol.
230, London, pp 323-362, November 1931.
Thomas, F.M., “An analysis of the effects due to restrained warping of thick walled box beams and
I-beams subject to torsion”, Ph.D. thesis, University of Ilinois at Urbana, 1969.
Timoshenko, S.P., “Strength of materials, Part 1: Elementary theory and problems”, D. Van
Nostrand, 2rd Edition, 1941(a).
Timoshenko, S.P., “Strength of materials, Part 2: Advanced theory and problems”, D. Van
Nostrand, 2rd Edition, 1941(b).
Timoshenko, S.P., “Strength of materials, Part 1: Elementary theory and problems”, D. Van
Nostrand, 3rd Edition, 1955.
Timoshenko, S.P. and Goodier, J.N., “Theory of elasticity”, McGraw-Hill, 3rd Edition, 1970.
Trahair, N.S. and Bradford, M.A., “The behaviour and design of structures to AS 4100”, E & FN
Spon, 3rd Edition, 1998.
Rectangular Hollow Sections with circular web openings LXV
Wilkinson, T. and Hancock, G.J., “Tests to investigate the web slenderness limits for cold-formed
RHS in bending”, The Mechanics of Structures and Materials, Proceedings of the 15th Australasian
Conference on the Mechanics of Structures and Materials, Melbourne, Australia, (Balkema, publ.),
(Grzebieta, Al-Mahaidi & Wilson editors), pp 559-564, December 1997.
Willems, N. Easley, J.T. and Rolfe, S.T., “Strength of materials”, McGraw-Hill, p 316, 1981.
Young, W.C., “A practical approach to theoretical stress analysis – or where do you find it in Roark
and how do you use it effectively”, Society for experimental mechanics, 1982.
Young, W.C., “Roark’s formulas for stress and strain”, McGraw-Hill, 6th Edition, 1989.
Zhao X.L. and Hancock G.J., “T-Joints in Rectangular Hollow Sections Subject to Combined
Actions”, American Society of Civil Engineers, Journal of Structural Engineering, Vol. 117, No. 8, pp
2258-2277, 1990.
Zhao X.L. and Hancock G.J., “Tests to Determine Plate Slenderness Limits for Cold-Formed
Rectangular Hollow Sections of Grade C450”, Steel Construction, Journal of Australian Institute of
Steel Construction, Vol. 25, No. 4, pp 2-16, November 1991.
Zhao X.L. and Hancock G.J., “Square and Rectangular Hollow Sections Subject to Combined
Actions”, American Society of Civil Engineers, Journal of Structural Engineering, Vol. 118, No. 3, pp
648-668, 1992.
Zienkiewicz, O.C. and Taylor, R.L., “The finite element method”, McGraw-Hill, 4th Edition, 1994.
Rectangular Hollow Sections with circular web openings LXVI
Standards and technical design documents
British Steel publication TD 365, “SHS Design to BS 5950 Part 1”, British Steel Tubes and Pipes,
1994.
BS 1610-1:1992, “Materials testing machines and force verification equipment”, Part 1:

“Specification for the grading of the forces applied by materials testing machines when used in the
compression mode”, British Standards Institution, 1992.
BS 1610-2:1985, “Materials testing machines and force verification equipment”, Part 2:

“Specification for the grading of equipment used for the verification of the forces applied by
materials testing machines”, British Standards Institution, 1985.
BS 4360:1990, “Specification for weldable structural steels”, British Standards Institution, 1990.
BS 5950-1:1990, “Structural use of steelwork in building”, Part 1: “Code of practice for design in
simple and continuous construction, hot rolled sections”, British Standards Institution, 1990.
BS EN 10002-1:1990, “Tensile testing of metallic materials”, Part 1: “Method of test at ambient

temperatures”, British Standards Institution, 1990.
BS EN 10002-2:1992, “Tensile testing of metallic materials”, Part 2: “Verification of the force

measuring system of the tensile testing machine”, British Standards Institution, 1992.
BS EN 10002-3:1992, “Tensile testing of metallic materials”, Part 3: “Calibration of force proving

instruments used for the verification of uniaxial testing machines”, British Standards Institution,
1995.
BS EN 10025:1990, “Specification for hot rolled products of non-alloy structural steels and their
technical delivery conditions”, British Standards Institution, 1990.
BS EN 10025:1993, “Hot rolled products of non-alloy structural steels. Technical delivery

conditions”, British Standards Institution, 1993.
Rectangular Hollow Sections with circular web openings LXVII
BS EN 10210-1:1994, “Hot finished structural hollow sections of non-alloy and fine grain structural
steels”, Part 1: “Technical delivery requirements”, British Standards Institution, 1994.
BS EN 10210-2:1997, “Hot finished structural hollow sections of non-alloy and fine grain structural
steels”, Part 2: “Tolerances, dimensions and sectional properties”, British Standards Institution,
1997.
Constrado document Cl/SfB 1968, “Holes in beam webs. Allowable stress design”, Constructional
Steel Research and Development Organisation, 1977.
Eurocode 3, “Design of Steel Structures”, Part 1.1: “General Rules and Rules for buildings”, ENV
1993-1-1:1992, European Committee for Standardization, UK version, British Standards Institution,
1992.
Rectangular Hollow Sections with circular web openings LXVIII
Papers and project reports
Ridley-Ellis D. J., Owen J. S. and Davies G., “Flexural behaviour of Rectangular Hollow Sections
with circular side wall openings.” Submitted for presentation at the 9th International Symposium on
Tubular Structures, Düsseldorf (Germany), April 2001.
Ridley-Ellis D. J., Davies G. and Owen J. S., “Fundamental behaviour of rectangular hollow
sections with web openings.” Presented at the 8th International Symposium on Tubular Structures,
Singapore, August 1998.
Ridley-Ellis D. J., Davies G. and Owen J. S., “Rectangular Hollow Sections as Slim Floor edge
beams.” Trends and developments in building, TEMPUS JEP 11297, Iasi (Romania), May 1998.
(Further papers in preparation)
Ridley-Ellis D. J., Owen J. S. and Davies G., “Recommendations for the design of hot-finished steel
Rectangular Hollow Sections with circular web openings; Torsion, bending and shear.” Report to
Corus Tubes, Report no: SC2000-011, School of Civil Engineering, The University of Nottingham,
July 2000.
Ridley-Ellis D. J., Owen J. S. and Davies G., “Torsion-testing of Rectangular Hollow Sections with
web openings; A summary of experimental procedure, apparatus and results and a quantitative
comparison of experimental observations with theoretical and Finite Element predictions.” Report to
Corus Tubes, Report no: SC2000-010, School of Civil Engineering, The University of Nottingham,
March 2000.
Ridley-Ellis D. J., Owen J. S. and Davies G., “A mathematical approximation of the stress-strain
relationship of steel from hot finished Rectangular Hollow Sections.” Internal Report, Report no:
SC2000-008, School of Civil Engineering, The University of Nottingham, February 1999.
Rectangular Hollow Sections with circular web openings LXIX
Errata

Thesis Example

Uploaded by

Copyright:

Available Formats

Thesis Example

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Thesis Example

Uploaded by

Copyright:

Available Formats

Rectangular Hollow Sections with circular

Fundamental behaviour in torsion, bending and shear

by D.J.Ridley-Ellis, B.Eng. M.Res.

Thesis submitted to the University of Nottingham

for the degree of Doctor of Philosophy, October 2000

Since men were of an imitative and teachable nature, they

Vitruvius, circa 27 B.C.E.

Table of contents iii

2.4.4.3 Brittle coatings 2-38

5.2.5 Testing procedure 5-12

7.7.2 Concentration factors for elastic stress 7-62

Figure 1-1: Slim floor and conventional construction 1-3

Figure 2-24: Warping of RHS in torsion (theory and FE) 2-52

Figure 4-4: Measurement of vertical displacement with linear potentiometers 4-5

Table 3-1: Specimens tested 3-11

Table 6-1: Specimens tested 6-6

Abstract (long version)

building construction is becoming increasingly common. Edge beams are subject to

offers an ideal opportunity for improved service integration, the influence of an

situations where it may be advantageous to cut holes in load bearing tubular

members of rectangular cross-section.

loading) laboratory testing.

found to be comparable to the reduction in the cross-sectional area of the perforated

causing a reduction in the rotation capacity.

between experimentally measured quantities (such as capacities, deflections and

strains) and the corresponding Finite Element predictions, allowing parametric

investigations to be conducted with calibrated analytical models. Generally, the

quantities and predictions of elastic stiffness within 20%.

Preliminary design recommendations are presented based on the results of

Aspects of behaviour requiring further investigation have been identified and

Torsion tests on full-scale RHS without web openings yielded some

Abstract (short version)

situations where it may be advantageous to cut holes in load bearing tubular

members of rectangular cross-section.

loading) laboratory testing.

between experimentally measured quantities (such as capacities, deflections and

strains) and the corresponding Finite Element predictions, allowing parametric

investigations to be conducted with calibrated analytical models.

Preliminary design recommendations are presented based on the results of

Aspects of behaviour requiring further investigation have been identified and

their continuing patience.

University of Nottingham and by British Steel Tubes and Pipes plc.

Daniel James Ridley-Ellis

Daniel DN: cn=Daniel

Notation and symbols

Iζζ Intermediate quantity (BS EN 10210-2:1997)

T Torque (torsional moment)

κe Curvature at elastic moment

Φ Hole size ratio

List of abbreviations and trademarks

ASB Asymmetric Slimflor Beam

ABAQUS is a registered trademark of Hibbit, Karlsson and Sorrensen Inc.

1 Introduction and background

This Thesis describes an investigation into the fundamental behaviour of

predictions, in order to develop the basis for design recommendations. The

research is fundamental in nature, the results have a more general applicability.

1.2.1 An outline of the slim floor method of construction

Slim floor is a method of construction for multi-storey steel-framed buildings

in a significant reduction in building height and a corresponding reduction in

or a false ceiling with no obstructions for service runs.