9

DESIGN GUIDE
FOR STRUCTURAL HOLLOW SECTION
COLUMN CONNECTIONS

Y. Kurobane, J. A. Packer, J. Wardenier, N. Yeomans

TÜV-Verlag
 Bibliografische Information der Deutschen Bibliothek

Die Deutsche Bibliothek verzeichnet diese Publikation in der

Deutschen Nationalbibliografie; detaillierte bibliografische Daten
sind im Internet über [Link] abrufbar.

ISBN 3-8249-0802-6

© by TÜV-Verlag GmbH,
TÜV Rheinland Group, Köln
Entirely printed by: TÜV-Verlag GmbH, Köln
Printed in Germany 2004
Preface
Steel structural hollow sections, circular, square and rectangular, are some of the most effi-
cient structural sections under compression loading. This design guide has been written
to give the design engineer the information one needs to design hollow section column
connections in the most efficient and economic way. Steel structural hollow sections are
unique in the world of structural steel sections, because their geometry is such that their
mass is distributed away from their longitudinal axis, making them ideal for use as
columns.

This design guide is the 9th in a series that CIDECT has published under the general series
heading “Construction with Hollow Steel Sections“. The previously published design
guides in the series, which are all available in English, French, German and Spanish, are:

1. Design guide for circular hollow section (CHS) joints under predominantly static load-
ing (1991)
2. Structural stability of hollow sections (1992, reprinted 1996)
3. Design guide for rectangular hollow section (RHS) joints under predominantly static
loading (1992)
4. Design guide for structural hollow section columns exposed to fire (1995, reprinted
1996)
5. Design guide for concrete filled hollow section columns under static and seismic
loading (1995)
6. Design guide for structural hollow sections in mechanical applications (1995)
7. Design guide for fabrication, assembly and erection of hollow section structures (1998)
8. Design guide for circular and rectangular hollow section welded joints under fatigue
loading (2000)

CIDECT’s sincere and grateful thanks go to the authors of this design guide, Prof. Dr.
Yoshiaki Kurobane of Sojo University in Japan, Prof. Dr. Jeffrey Packer of University of
Toronto in Canada, Prof. Dr. Jaap Wardenier of Delft University of Technology in The
Netherlands and Mr. Noel Yeomans of Corus Tubes in The United Kingdom, for their thor-
oughness and professionalism in writing the various chapters and to CIDECT and its mem-
bers for their useful comments and support.

CIDECT
2004

5
CONTENTS

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.1 Design philosophy ..............................................9

2 Advantages of hollow section columns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.1 Plain columns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2 Concrete filled columns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.3 Fire protection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.3.1 External fire protection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.3.2 Internal fire protection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

3 Single sided bolting systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

3.1 Flowdrill drilling system . ......................................... 17
3.2 Lindapter HolloBolt insert ........................................ 19
3.3 Huck Ultra-Twist bolt . . . ......................................... 19
3.4 Stud welding . . . . . . . . . ......................................... 20
3.5 Other methods . . . . . . . ......................................... 20

4 Connection classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . .......... . . . . . . . . . . . . . . . . . . 23
4.1.1 Elastic behaviour . . . . . . . . . . . . . . . . . . . .......... . . . . . . . . . . . . . . . . . . 23
4.1.2 Plastic behaviour . . . . . . . . . . . . . . . . . . . .......... . . . . . . . . . . . . . . . . . . 25
4.2 Semi-rigid connection design according to Eurocode 3 . . . . . . . . . . . . . . . . . . 27
4.2.1 Classification of connections . . . . . . . . . . .......... . . . . . . . . . . . . . . . . . . 28
4.2.2 Moment capacity . . . . . . . . . . . . . . . . . . . .......... . . . . . . . . . . . . . . . . . . 29
4.2.3 Rotational stiffness . . . . . . . . . . . . . . . . . .......... . . . . . . . . . . . . . . . . . . 30
4.2.4 Rotational capacity . . . . . . . . . . . . . . . . . .......... . . . . . . . . . . . . . . . . . . 31
4.2.5 Conceptual design . . . . . . . . . . . . . . . . . .......... . . . . . . . . . . . . . . . . . . 31

5 Simple shear connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
5.2 Limit states for simple shear connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
5.3 Single shear plate connections (shear tabs, fin plates) . . . . . . . . . . . . . . . . . . . 34
5.3.1 Connection to RHS column design example . . . . . . . . . . . . . . . . . . . . . . . . . . 36
5.3.2 Connections to CHS columns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
5.3.3 Single shear plate connections to RHS column corner . . . . . . . . . . . . . . . . . . 40
5.4 “Through-Plate” connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
5.5 End plate connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
5.6 Tee connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
5.7 Single and double angle connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
5.8 Unstiffened seat connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
5.9 Stiffened seat connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
5.10 Hollow section beams to hollow section columns . . . . . . . . . . . . . . . . . . . . . . 50
5.11 Use of through-bolts to hollow section columns . . . . . . . . . . . . . . . . . . . . . . . 50
5.12 Influence of concrete slabs on behaviour of connections . . . . . . . . . . . . . . . . . 51

6
6 Semi-rigid connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
6.1 Types of semi-rigid connections with hollow section members . . . . . . . . . . . . 53
6.2 Welded hollow section beam and column connections . . . . . . . . . . . . . . . . . . 54
6.2.1 CHS beam and column members . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
6.2.2 RHS beam and column members . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
6.2.3 CHS and RHS beam and column members . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
6.3 Welded I-beam-to-hollow section column connections . . . . . . . . . . . . . . . . . . . 64
6.3.1 I-beam-to-CHS column connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
6.3.2 I-beam-to-RHS column connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
6.4 Bolted hollow section beam and column connections . . . . . . . . . . . . . . . . . . . 78
6.4.1 CHS beam-to-column connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
6.4.2 RHS beam-to-column connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
6.5 Bolted I-beam-to-hollow section column connections . . . . . . . . . . . . . . . . . . . 79
6.5.1 I-beam-to-CHS column connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
6.5.2 I-beam-to-RHS column connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
6.6 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
Example 1: CHS beams and columns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Example 2: RHS beams and columns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
Example 3: I-beams and CHS columns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
Example 4: Bolted I-beam-to-RHS column connection . . . . . . . . . . . . . . . . . . . 94

7 Special requirements for seismic loading . . . . . . . . . . . . . . . . . . . . . . . . . . 97

7.1 Dissipative and non-dissipative structural behaviours . . . . . . . . . . . . . . . . . . . 98
7.2 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
7.3 Structural types and behaviour factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
7.4 Joints in dissipative zones . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
7.5 Strong column-weak beam design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
7.6 Beam-to-column moment connections (rigid and full-strength connections) . 103
7.7 Column web panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106

8 Rigid (full strength) connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109

8.1 Connections with through diaphragms for shop welding application . . . . . . . 109
8.2 Bolted through diaphragm connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
8.2.1 Design example for bolted through diaphragm connection . . . . . . . . . . . . . . 119
8.3 Connections with through diaphragms for field welding application . . . . . . . . 123
8.3.1 Connections with improved details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
8.3.2 Connections for ordinary moment frames . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
8.3.3 Reinforced connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
8.4 Reduced beam section (RBS) connections . . . . . . . . . . . . . . . . . . . . . . . . . . 129
8.5 Connections with internal diaphragms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
8.6 Connections with external diaphragms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
8.7 End plate connections with blind bolts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
8.8 Rigid connections for structures in low seismicity zones . . . . . . . . . . . . . . . . 143

9 Connections to concrete filled columns . . . . . . . . . . . . . . . . . . . . . . . . . . . 145

9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
9.2 Simple shear connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
9.2.1 Load entry to the column . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
9.2.2 Connection design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
9.3 Semi-rigid connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
9.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147

7
9.3.2 Unreinforced welded hollow section beam and column connections . . . . . . . 147
9.3.3 Unreinforced welded I-beam-to-hollow section column connections . . . . . . . 149
9.3.4 Bolted hollow section beam and column connections . . . . . . . . . . . . . . . . . . 151
9.3.5 Bolted I-beam-to-hollow section column connections . . . . . . . . . . . . . . . . . . 151
9.3.6 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
9.4 Rigid (full strength) connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
9.4.1 Shear strength of column web panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
9.4.2 Flexural strength of beam-to-column connections . . . . . . . . . . . . . . . . . . . . . 154

10 Bracing and truss connections to columns . . . . . . . . . . . . . . . . . . . . . . . . 159

10.1 Bracing connections to RHS columns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
10.1.1 Longitudinal plate-to-RHS columns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
10.1.2 Longitudinal “through-plate”-to-RHS columns . . . . . . . . . . . . . . . . . . . . . . . . 165
10.1.3 Stiffened longitudinal plate (T-stub)-to-RHS columns . . . . . . . . . . . . . . . . . . . 165
10.1.4 Transverse plate-to-RHS columns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
10.2 Bracing connections to CHS columns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
10.2.1 Longitudinal plate-to-CHS columns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
10.2.2 Longitudinal “through-plate”-to-CHS columns . . . . . . . . . . . . . . . . . . . . . . . . 168
10.2.3 Stiffened longitudinal plate (T-stub)-to-CHS columns . . . . . . . . . . . . . . . . . . . 169
10.2.4 Transverse plate-to-CHS columns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
10.3 Bracing connections to RHS and CHS columns under seismic loading . . . . . 169
10.4 Truss connections to columns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169

11 Column splices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171

11.1 Plain columns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
11.1.1 Bolted end plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
11.1.2 Bolted side plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
11.1.3 Welding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
11.1.4 Welded column splices in seismic areas . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
11.2 Concrete filled columns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
11.3 Nailing of poles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
11.4 Design example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
11.4.1 Bolted end plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180

12 List of symbols and abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183

12.1 Abbreviations of organisations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
12.2 Other abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
12.3 General symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
12.4 Subscripts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
12.5 Superscripts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185

13 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187

Annex A: Investigation into through diaphragms . . . . . . . . . . . . . . . . . . . . . . . . . . 201

A.1 Summary of tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
A.2 Evaluation of rotation capacity of beams . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
A.3 Flexural strength of beam-column connections . . . . . . . . . . . . . . . . . . . . . . . 205
A.4 Definition of cumulative plastic deformation factor . . . . . . . . . . . . . . . . . . . . 206

CIDECT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209

8
1 Introduction

Steel structural hollow sections, whether they are circular, square or rectangular, are inher-
ently more efficient as compression members than any other structural steel section, e.g.
I-, H- or L-sections, etc., due to their geometric shape.

In order to obtain a technically secure, economic and architecturally pleasing structure,

both the architect and design engineer must, from the very beginning, be aware of the
effects of their design decisions on the fabrication, the assembly and the erection of the
structure.

Architects, design engineers, fabricators and erectors are all experts in their own particu-
lar fields, but traditionally have worked separately. The architect and the design engineer,
between them, are responsible for the conceptual lay-out, the sizing of the members and,
often with tubular construction, some initial detailing of the connections. All of these are
generally aimed at the reduction of the material weight of the structure, often with too lit-
tle thought to the fabrication, assembly and erection. This lack of communication between
the various disciplines and, subsequently, an inadequate knowledge of the interaction
between them, often leads to a situation where the impact of the design on the fabrication
and erection, and vice-versa, is not properly taken into account.

The practice of designing for minimum material weight is very often counter-productive as
an overall solution, because of the additional costs caused by complex fabrication and site
erection imposed by the initial conceptual design. This can only be avoided by an effec-
tive dialogue between all of the disciplines involved, with each having some knowledge of
the other’s requirements for the realisation of a satisfactory and cost effective structure.

A properly designed steel construction using structural hollow sections, taking into
account all of the foregoing, will nearly always be lighter in terms of material weight than a
similar construction made with open section profiles. As a result, although structural hol-
low sections are more expensive than open section profiles on a per tonne basis, the over-
all weight saving that can be gained by using them will very often result in a much more
cost effective and therefore economic construction.

1.1 Design philosophy

The design philosophy, requirements and terminology can be considerably different from
country to country, e.g.

• limit states or allowable stress design;

• requirements or not for robustness (also called structural integrity);
• material yield strength, tensile strength or a combination of both;
• the methodology and specific value of partial safety factors (or resistance factors) for
both load and capacity;
• design details;
• the symbols used vary not only from country to country, but in some cases within the
same country.

Design method: This design guide is written in terms of a limit states format, unless
specifically stated otherwise. However, if the information given in this design guide is to be

9
used in an allowable stress design, it is suggested that a safety factor of about 1.5 should
be used on the capacities given in this design guide.

Robustness: In many countries the building codes and regulations have a robustness or
structural integrity requirement. This requirement is that all connections, even simple shear
ones with only vertical shear loads, must also have the ability to carry specified, nominal
horizontal forces. This is to ensure that if accidental horizontal forces are present in a build-
ing, it and the individual connections will remain intact and will not collapse.

Material strength: Structural hollow sections are manufactured in many countries and
product specifications can be quite different from one country to another. These differ-
ences can include method of manufacture (hot finished or cold formed), yield and tensile
strength, elongation and impact properties, and also differences in the geometric proper-
ties and tolerances of the hollow section.

There are also differences in the definition of shear yield strength; some use 0.6 times the
tensile yield strength and others the tensile yield strength divided by 3. In this design
guide the latter, tensile yield strength divided by 3, has been generally used.

Partial safety factors: Different building codes use different philosophies for partial safe-
ty factors for resistance. Some codes, such as Eurocode No. 3 (CEN 1992), use partial (g)
safety factors for resistance which generally have values above or equal to 1.0 and are
used as dividers, i.e. (nominal strength)/gM. Others, especially in North America and
Australia, use resistance or capacity (f) factors with values equal to or less than 1.0 and
are used as multipliers, i.e. f (nominal strength). However, this difference in design
methodology will usually make little difference to the design since the value of 1/gM is gen-
erally very nearly the same as f.

In this design guide, as with all previous CIDECT design guides, all the design expressions
relating to the actual hollow section column already have the relevant partial safety (or
resistance) factor incorporated (g = 1/f) and as a result no further partial safety or resis-
tance factors should be applied to them. However, for the beam members, plates, bolts,
etc. the partial safety or capacity factors relevant to the design code being used by the
design engineer should always be used. Thus, gM or f factors should only be used in this
Design Guide where indicated.

Design details: Different codes and specifications use different design details for such
items as bolt spacing, edge distances from bolt centres, effective lengths of welds, etc.
This design guide does not attempt to lay down specific values for any of these and the
values specified in the relevant code being used by the design engineer should always be
applied. In some examples in this design guide, the authors have, for completeness,
undertaken detail design using a local design code. These occasions are explicitly stated,
and designers should still use their own local code when making similar checks.

Symbols: A wide variety of symbols are used around the world for the same thing. For
example, in different codes, the symbol used for yield strength may be Fy or fy or py or Ys
or Re, etc. A list of symbols used in this design guide is given in Chapter 12.

10
2 Advantages of hollow section columns

The radius of gyration, especially about the minor axis, of a structural hollow section is sig-
nificantly higher than that of an open section of a similar size and area. This results in a
much lower slenderness ratio for the same effective length, and hence a higher compres-
sion capacity. Any residual stresses that may be in the section due to the method of manu-
facture are generally also distributed in a much more favourable way than those in open
sections because of the different shape characteristics and this can also result in an
increase in the compression capacity.

Structural hollow sections are generally available in lengths up to 12 or 15 m (40 or 50 ft),

but in some circumstances longer lengths, up to 20 m, may be available. This means that
for buildings of up to about 4 storeys only one length per column is required.

An additional benefit of structural hollow sections is that for any given section size the out-
side dimensions remain the same irrespective of the thickness, unlike H-section columns,
where the inside dimensions remain the same and the external dimensions change. This
means that even if the column cross sectional area is reduced in higher storeys, the beam
lengths can remain the same for the full height of the building, which should result in
reduced beam fabrication and erection times and therefore reduced overall costs.

2.1 Plain columns

In most countries of the world, the current design codes and standards either are, or will
be in the near future, based on a limit states design method. The majority of these use what
are described as multiple column curves for the design of compression members. The
designation of these curves varies. In Europe, for example, Eurocode 3 (CEN 1992) uses
the designations a, b, c, etc. while others use 1, 2, 3, etc. However, in all cases hot finished
structural hollow sections have been assigned to the highest curve (i.e. curve a or 1). In
Eurocode 3, but not necessarily world-wide, cold formed structural hollow sections, based
on the enhanced mechanical properties of the finished cold formed product, have been
assigned to the third curve (i.e. curve c or 3). A graph of the buckling curves given in
Eurocode 3 is shown in figure 2.1. This can result in either a much higher capacity or a con-
siderable weight saving if a structural hollow section is used instead of an open structural
column section. In addition, if columns are subject to moment loading about both axes,
structural hollow sections generally have a higher moment of inertia and section modulus
about the minor axis than a comparable H-section. The design of structural hollow section
compression members is described in much more detail in the CIDECT Design Guide on
Structural Stability of Hollow Sections (Rondal et al. 1992).

11
 1.0
Curve 'a' - Hot finished
Curve 'b'
0.8 Curve 'c' - Cold formed
Curve 'd'
Yield strength reduction factor

0.6

0.4

0.2

0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0

Non-dimensional slenderness ratio

Figure 2.1 – Eurocode 3 column buckling curves

An example is given in figures 2.2 and 2.3. This comparison has been made based on an
effective length of 5 m and designing to Eurocode 3, with the requirements of the UK
national application document, DD ENV 1993 (BSI 1992). The sections used are a British
universal column (UC, H-section), BS 4 (BSI 1993), and two European hot finished struc-
tural hollow sections, one square (RHS) and one circular (CHS), EN 10210 (CEN 1997), with
all the sections having a nominal yield strength of 275 N/mm2.

3500

3000
Compression capacity - kN

2500

2000

1500

1000

500

0
203x203x60 323.9x8.0 254x254x107 355.6x12.5
250x250x8.0 350x350x10

H-section RHS CHS H-section RHS CHS

a) mass ~ 60 kg/m b) mass ~ 106 kg/m
Figure 2.2 – Comparison of compression capacity for sections of equal mass

12
Based on the concept of equal masses, figure 2.2 shows that for section masses of about
60 kg/m a structural hollow section has a capacity almost twice that of a universal column
and for masses of about 106 kg/m the capacity is about 50% higher. The converse of this
is shown in figure 2.3, where for equal capacities a mass saving of nearly 40% can be
achieved for a capacity of about 1000 kN and a saving of between 30% and 35% on a
capacity of about 2100 kN.
120

100
Section mass - kg/m

80

60

40

20

0
203x203x60 200x200x6.3 244.5x6.3 254x254x107 250x250x10 355.6x8.0
H-section RHS CHS H-section RHS CHS
a) capacities ~ 950 kN b) capacities ~ 2100 kN

Figure 2.3 – Comparison of section masses for equal compression capacities

2.2 Concrete filled columns

Because of the hole in its centre a structural hollow section can be easily filled with con-
crete, either with or without rebar, to create a steel/concrete composite section, without
the need for the temporary shuttering or formwork associated with composite columns
made from open sections. Generally concrete with cylinder strengths in the range from
20 N/mm2 to 50 N/mm2 (cube strengths of 25 to 60 N/mm2) has been used. It is possible
to use higher strength concrete, but at the present time research work in this area is still
underway and no definitive CIDECT design guidance is available. Concrete filled hollow
section columns are much more ductile than a plain or reinforced concrete column and
connections for beams, etc. can generally be designed and constructed using straight-for-
ward steel design criteria. The ductility and rotation capacity of concrete filled hollow sec-
tion columns is much better than that of other types of composite column because the
concrete is contained within the steel shell and cannot split away even if the ultimate
strength of the concrete is reached.

Figure 2.4 gives a comparison of the capacities of the same sections as those shown in
figure 2.2.a), but also includes those for the two structural hollow sections when filled with
concrete having a cube strength of 40 N/mm2. The capacities of the hollow sections have
been increased considerably and are now about 170% and 220% higher than that of the
universal column section.

13
 3500

3000
Compression capacity – kN

2500

2000

1500

1000

500

0
203x203x60 250x250x8.0 323.9x8.0 250x250x8.0 323.9x8.0
H-section Square Circular Square Circular
Empty hollow sections Concrete filled hollow sections

Figure 2.4 – Compression capacities for sections of equal mass (about 60 kg/m) with concrete filling

Most countries, for example Australia, Canada and those in Europe, now use limit states
methods for the design of composite steel/concrete columns, although some, notably
Japan, still use an allowable stress approach. The design of concrete filled structural hol-
low sections is fully described in the CIDECT Design Guide on Concrete Filled Hollow
Section Columns (Bergmann et al. 1995).

2.3 Fire protection

Structural hollow sections are unique among structural steel profiles in that they can be
protected from fire damage by using either internal or external methods of protection. As
with other structural steel sections, in some cases where the required fire resistance time
is quite short, about 15 to 30 minutes, it is possible that no fire protection of any type is
needed.

CIDECT Design Guide No. 4 (Twilt et al. 1995) gives detailed information on the design
requirements for both external and internal methods of fire protection for structural hollow
sections.

2.3.1 External fire protection

This type of fire protection can be applied to all types of structural steel profiles. The
degree of fire protection depends upon the properties and thickness of the insulation
material, the shape factor (heated surface periphery divided by cross sectional area) of the
steel profile and the load being carried.

If a sprayed or profile following external protection material is to be used, a structural hol-

low section will generally require a smaller volume of fire protection material than an equiv-
alent H-section, because of its smaller exposed surface area. For example, consider the
structural sections shown in figure 2.3 for a capacity of about 1000 kN. All have a shape
factor of about 160 and will, therefore, all require about the same thickness of fire protec-

14
tion material. However, both of the hollow sections have a surface area about 35% less
than the H-section, so the volume of fire protection material required will also be about
35% less.

2.3.2 Internal fire protection

The hole down the centre of a structural hollow section can be used to great effect as a
means of providing the required fire protection to the section and still retain its original
external dimensions. Two types of internal fire protection can be used: concrete filling and
water filling.

Concrete filling of structural hollow sections has previously been described (section 2.2) to
produce a composite steel/concrete column, but it can also be used as a method of fire
protection. In a fire the temperature distribution in a concrete filled hollow section is sig-
nificantly different to that in an empty hollow section. The combination of materials with
markedly different thermal conductivities produces extreme transient heating behaviour
and high temperature differentials across the section. As a result of these differentials rein-
forced concrete filled hollow section columns can be designed to have a fire resistance of
up to 120 minutes, or more, without any external fire protection. In this situation the basic
idea is that the steel plus reinforced concrete are designed to carry the normal factored
loads under a no-fire situation, and the reinforced concrete is designed to carry the much
lower service loads that need to be taken into account in a fire.

Water filling, using natural circulation, provides a safe and reliable fire protection method
for structural hollow section columns provided that the system is self activating in a fire
and that the system is also self controlling. In a properly designed system the natural cir-
culation will be activated when the columns are locally heated by a fire. The lower density
of the heated water, compared to that of the remaining cooler water, produces pressure
differentials which cause natural circulation. As the fire develops this behaviour increases,
which in turn increases the cooling effect and the system becomes self-controlling. Several
methods of designing a water filled system are described in CIDECT Design Guide No. 4
(Twilt et al. 1995).

15
16
3 Single sided bolting systems

There are two main methods of making site connections: bolting and welding. Bolting is
nearly always the preferred method, unless special circumstances dictate otherwise. Using
standard bolts and nuts to make connections to structural hollow sections is difficult
because there is normally no access to the inside of the section to tighten them. Unless
on-site welding has been adopted, this has usually meant that some form of additional fab-
rication, and therefore cost, has been necessary to overcome the problem.

Although a number of single sided, or blind, bolting systems have been in existence for a
number of years, they have not normally been used in general steel construction mainly
because they have been too small in diameter for structural applications. There had, as a
result, been very few investigations into their structural strength and behaviour. In recent
years, however, a number of blind bolting systems have become available in structural
sizes (up to M20 or even M24) and strengths (ISO grade 8.8, ASTM A325, etc.). Blind bolt-
ing systems make use of either special types of bolts or inserts or special drilling systems.
As the name implies, these can be used when only one side of the connection is accessi-
ble, and, therefore, access to both sides is not necessary. This allows, for example, bolt-
ed beam to structural hollow section column connection details to be designed in a simi-
lar way to a beam to open section column connection.

As these blind bolting systems have become available, CIDECT and others have carried
out various research and development projects, in conjunction with the system manufac-
turers. These projects have been used to determine the requirements for the design of
connections to structural hollow section columns incorporating these different systems.
Although other systems may be available, these research projects have concentrated on
the following systems: the Flowdrill drilling system, the Lindapter HolloBolt insert and the
Huck Ultra-Twist bolt, which are described in the following sections of this chapter.

There is no intrinsic reason why these systems cannot be applied to both rectangular and
circular hollow section columns. However, direct bolting to rectangular hollow section
columns is an accepted procedure, but direct bolting to circular hollow section columns is
not so usual because curved saddle plates, instead of flat ones, are required.

The following sections 3.1 to 3.5 describe these methods/systems and their capacities as
individual bolts in a structural hollow section. In most connections incorporating a group
of bolts loaded in tension the connection capacity will almost always be controlled by the
deformation or yielding capacity of the face of the structural hollow section and not that of
the individual bolt. The design methods and details for these practical connections are
given in section 6.5.2.

3.1 Flowdrill drilling system

The Flowdrill system is a patented method for the extrusion of holes using a four lobed
tungsten-carbide friction drill. Details of the drilling tools and procedure are available from
the manufacturer – Flowdrill b.v. at [Link].

The tungsten-carbide drill bit forms a truncated cone on the far side of the workpiece and
a small upset on the near side, which can automatically be removed by a milling cutter
incorporated into the drill bit. The hole can then be threaded using a roll (or forging) tap,
rather than a cutting tap, to produce a threaded hole, which has an effective thread length

17
of 1.5 to 2.0 times the material thickness. The Flowdrill process is shown schematically in
figure 3.1.

The advantages of this system are that the specialist equipment is fabrication shop based,
only standard fully threaded bolts are used (no nuts are needed), virtually standard beam
and column bolt hole layouts can be used and no specialist equipment is required on site.

Figure 3.1 – Schematic of the Flowdrill process

The results of a series of tests on individual flowdrilled holes and on connections made
using the Flowdrill system (Yeomans 1996a and 1996b) have shown that they are suitable
for structural applications. These tests have shown that:

• flowdrilled holes can be produced in both hot finished and cold formed hollow
sections from 5.0 to 12.5 mm thick;
• threaded roll tapped holes with M16, M20 and M24 ISO course thread profiles can be
made;
• the full tension capacity of grade 8.8 (similar to ASTM A325) bolts can be carried by
flowdrilled and roll tapped holes, provided that the RHS thickness is equal to or greater
than the minimum thickness shown in Table 3.1 and the RHS has a nominal yield
strength in the range 275 to 355 N/mm2;

Table 3.1 – Minimum RHS thickness for full grade 8.8 bolt tension capacity

• the shear and bearing capacities of the hole and bolt can be calculated in the normal
manner;
• in most applications in which the bolts are loaded in tension, the deformation or yield-
ing of the RHS face will determine the overall connection capacity and not the capa-
city of each individual bolt. The design criteria for this are given in section 6.5.2.

18
3.2 Lindapter HolloBolt insert

The HolloBolt is a three part pre-assembled unit consisting of a main body, a threaded
truncated cone and a standard grade 8.8 bolt and is shown in figure 3.2. A five part sys-
tem is also available. Details of dimensions, hole tolerances, torque requirements, etc. are
available from the insert manufacturer – Lindapter International plc at [Link].

Figure 3.2 – The Lindapter HolloBolt insert

The operating principle of the HolloBolt insert is that once placed in the hole, through the
materials being joined, the tightening of the bolt draws the tapered cone into the legs
of the body. As this happens the legs of the body are splayed out and provide the mechan-
ical interlock necessary to prevent the insert being pulled out. The tension and shear
capacities of the insert are at least equal to that of the corresponding grade 8.8 bolt, but
it is suggested that the grade 8.8 bolt capacities should be used for design purposes
(Occhi 1996).

As with the Flowdrill system in connections in which the bolt, or bolts, are loaded in ten-
sion the RHS face deformation (or yielding) capacity will usually be the determining factor,
and not that of the individual insert (Yeomans 1998) unless the hollow section face is
reinforced. The design criteria for this are given in section 6.5.2.

3.3 Huck Ultra-Twist bolt

The Ultra-Twist bolt is a pre-assembled unit manufactured by Huck International Inc. at

[Link]/industrial, from whom details of dimensions, tolerances, torque require-
ments etc. are available. An exploded view of the bolt is shown in figure 3.3. The Ultra-
Twist bolt is installed using an electric bolting wrench in holes 2 mm larger than the out-
side diameter of the bolts, which provides conventional clearances for fit-up.

These bolts have tensile strengths, installed tensions and shear capacities meeting the
requirements of ASTM A325 bolts (equivalent to ISO grade 8.8, Sadri 1994 and Korol et al.
1993), so that the tension, shear and bearing capacities of individual fasteners can be cal-
culated in the normal way. However, as stated previously, in applications where a group of
bolts are used in tension the deformation or yielding of the hollow section face will nearly
always be the determining factor in design (see section 6.5.2) unless the hollow section
face is reinforced in some way.

19
Figure 3.3 – Exploded view of Huck Ultra-Twist bolt

3.4 Stud welding

Threaded studs welded to structural hollow section columns can also be used to produce
connections. Various types of studs are available, from many manufacturers, who should
be consulted concerning requirements for their installation and their capacities.

Some research has been carried out (Maquoi et al. 1985) to investigate welding param-
eters and connection capacities. Provided that the weld is adequate and the studs are cer-
tified in a similar manner to bolts, the capacity of individual studs can be based on
normal bolt and nut design methods, but with additional checks for punching shear and
tear out of the hollow section. If studs are to be welded onto the hollow sections in the
fabrication workshop, then special care is needed to prevent damage during transit to site.

Again, in connections containing a group of studs in tension deformation or yielding of the

face of the hollow section will nearly always be the determining criterion (see chapters 5,
6 and 8) unless the face of the hollow section is reinforced.

3.5 Other methods

There are several other methods available for making bolted connections, which can be
fixed from one side only. Two of these are briefly described below.

The first method is simply drilling and tapping the hollow section, but this generally needs
a wall thickness of 16 mm or more to generate enough pull out capacity.

Another method (Kato 1988) is to drill holes in the hollow section large enough for a nut of
the required size to be inserted and then to weld the nut to the hollow section flush with
the outside surface (see figure 3.4).

20
 Attached plate

Hollow section

Figure 3.4 – Nuts welded into hollow section wall

21
22
4 Connection classification

This chapter gives some general background information on the classification of connec-
tions. The subsequent chapters 5, 6 and 8 give actual design guidance on connections
to hollow section columns for simple shear (pinned), semi-rigid and rigid connections
respectively. Chapter 9 contains specific design guidance on connections to concrete filled
columns.

4.1 Introduction

In the past, most designers have designed beam-to-column connections either as pinned
or as rigid. However, in reality, the actual stiffness of a connection will nearly always be
somewhere between these two extremes, i.e. the connection will behave in a semi-rigid
manner. Also the capacity of an unstiffened connection might be less than that of the
connected beam, in which case it is termed “partial strength”.

The use of semi-rigid connections may offer a considerable reduction in overall frame
costs, because they generally have either no stiffeners or much fewer stiffeners than rigid
connections. Cost calculations for semi-rigid frames made of I- or H-beams show reduc-
tions in costs of 10 to 20% over rigid frames, depending on the structural arrangement and
the sections used. However, it is not only the costs of material and labour for columns,
beams and connections that should be taken into consideration, additional effects, such
as beam depth or avoidance of temporary bracing, should also be considered.

Note: in this design guide the terms “joint” and “connection” use the definition given in
AISC (1997), and not that in Eurocode 3: Annex J (CEN 1992) which uses them the other
way around.

4.1.1 Elastic behaviour

The effect of the connection stiffness on the elastic moment distribution for a beam with a
uniformly distributed load is represented in figures 4.1 and 4.2. Figure 4.1 shows the elas-
tic distribution in the beam for the pin end, the fixed end and the semi-rigid end conditions.
It can be seen that with semi-rigid connections the elastic moment distribution can be
influenced considerably.

23
 q

Lb
q L2b q L2b
M= M=
12 12

a. Simply supported
q L2b
M=
8

q L2b
M=
12 b. Fixed

q L2b
M=
24

M*j q L2b M*j c. Semi-rigid

8
M*b
Mb

Figure 4.1 – Beam with various end conditions

The joint rotation f j is given by:

q · L3b Mj · Lb ............................................................................................. 4.1
fj = -
24Elb 2Elb

with Mj = Sj · f j and Kb = 2Elb /L b ............................................................................................. 4.2

q · L3b Sj · f j · L b
............................................................................................. 4.3
or f j = -
24Elb 2Elb

Mj q · L3b Mj · L b
and = - ............................................................................................. 4.4
Sj 24Elb 2Elb

q L b2 Sj
............................................................................................. 4.5
or Mj = ·
12 (Kb + Sj)

q · L 2b
Mb = - Mj ............................................................................................. 4.6
8

24
Based on these relationships, in figure 4.2 the elastic moment at the beam centre Mb and
the moments at the connections Mj are given for different joint stiffnesses Sj.

qqLL2b2b
M* = Mb* + M*j =
1.00 88

Kb = 2EIb/Lb
M*
M

0.67
Mj
0.50
Mb
0.33

0 1 2 3 4 5
Sj / Kb

Figure 4.2 – Variation of elastic moment distribution with connection stiffness (Anderson et al. 1997)

4.1.2 Plastic behaviour

If a rigid-plastic analysis is used, the moment capacity of the connections is of primary

importance, but the rotation capacity is also important. For example, if the stiffness of the
connections of the beam in figure 4.1 is very low, the plastic moment capacity of the beam
at mid-span Mpl may be reached first. As a result the moment capacity of the end con-
nections Mj can only be reached if the beam has sufficient rotation capacity at the location
of the plastic hinge. In the case of connections with a very low stiffness this might not be
the case, e.g. see connection “e” in figure 4.3.

c
a
Mpl
b d e
M

Mpl = plastic moment capacity beam

φ
f

Figure 4.3 – Various M-f characteristics

25
If the stiffness of the connection is high, the (partial) strength capacity of the end connec-
tions (e.g. connection “b” in figure 4.3) may be reached first. Now these connections
should have sufficient deformation capacity to develop, with increasing load, the plastic
moment capacity of the beam at mid-span.

Thus, for a proper analysis of frames with semi-rigid connections, a description of the
moment-rotation behaviour is required. Thus, evidence is required regarding:

• stiffness (serviceability and at the ultimate limit state),

• strength (ultimate limit state) and
• rotation capacity.

However, all this information is not yet generally available for tubular beam-to-column
connections. Other options are that the stiffness is such that the connections can be
classified as (nearly) rigid or (nearly) pinned as discussed in other chapters. For both cases,
limits can be given. However, the deflections can only be determined properly if the joint
stiffness is available.

5q · L b4 Mj · L b2 ............................................................................................. 4.7
d= -
384El b 8Elb

Combined with equation 4.5 gives:

d=
5q L b4
384El b 3 1-
4Sj
5(Kb + Sj) 4 ............................................................................................. 4.8

Figure 4.4 shows this relationship between the mid-span deflection of the beam of figure
4.1 and the connection stiffness Sj.

5 q L 4b
!"=
384 EIb
1.0
!
!

0.8 2EI b
Kb =
0.6 Lb

0.4

0.2

0
0 1 2 3 4 5
Sj
Kb

Figure 4.4 – Variation of mid-span deflection with connection stiffness (Anderson et al. 1997)

26
4.2 Semi-rigid connection design according to Eurocode 3

In this section, the analysis method used in Eurocode 3: Annex J (CEN 1992) for semi-
rigid connections between I- and H-sections is briefly presented. The method is generally
known as the “component method” and is used to determine the strength and stiffness of
semi-rigid connections. These connections are principally for moment-resisting connec-
tions and can be welded or bolted (with end plates or angle cleats).

Note: At the time that this design guide was being written, CEN was reorganising and
updating Eurocode 3 and it is expected that sometime between 2002 and 2004 Eurocode
3: Annexes J and K will become part of Eurocode 3: Part 1.8: chapters 6 and 7 respec-
tively.

Connections between hollow sections are dealt with in Eurocode 3: Annex K. This deals
primarily with the ultimate strength of axially loaded connections, however, formulae are
also given for some types of moment connections between circular or rectangular hollow
sections. No information is given for the stiffness.

In Annex J, the types of connections are distinguished as indicated in figure 4.5

Method of Types of connections

global analysis
Elastic Nominally pinned Rigid Semi-rigid

Rigid-Plastic Nominally pinned Full-strength Partial-strength

Elastic-Plastic Nominally pinned Rigid and Semi-rigid and

full-strength partial-strength
Semi-rigid and
full-strength
Rigid and full-
strength
Type of joint model Simple Continuous Semi-continuous

Figure 4.5 – Types of connections according to Eurocode 3: Annex J

For an elastic global analysis, the connections are classified according to their stiffness,
for a rigid plastic analysis the connections are classified according to their strength and for
an elastic-plastic analysis the connections are classified according to both stiffness and
strength.

For elastic and elastic-plastic analyses the rotational stiffness of a semi-rigid joint is
needed. A simplified method is to use the initial rotational stiffness Sj,ini up to 2/3 Mj* and
Sj,ini /h, for larger values as shown in figure 4.6. The value for h is between 2 and 3.5,
depending on the type of joint. An even more simplified method is to use the stiffness value
for Mj* for all values of Mj.

27
 actual
M*j NC = 0
modelled

Sj,ini / η
2 *
Mj
3
Mj

Sj,ini

φ
f
Figure 4.6 – M-f modelling according to Eurocode 3: Annex J

4.2.1 Classification of connections

The classification by stiffness is given in figure 4.7. All connections in the zone marked with
“semi-rigid” should be classified as semi-rigid. The two other zones may optionally be
treated as semi-rigid, if convenient.

rigid, if Sj,ini 8EIb/Lb rigid, if Sj,ini 25EIb/Lb

Mj

Mj

nominally pinned, nominally pinned,

if Sj,ini< 0.5EIb/Lb if Sj,ini< 0.5EIb/Lb

semi-rigid
semi-rigid

φf φf

a) Braced frames b) Unbraced frames

Figure 4.7 – Boundaries for stiffness classification of beam-to-column connections (EC3, Annex J)

The classification by strength is as follows:

Full strength – if the moment design capacity of the connection is such that the plas-
tic moment capacities are first reached in the connecting beam(s) or column(s).
Pinned – if the design moment capacity of the connection does not exceed 25% of the
design moment capacity required for a full strength connection.
Partial strength – if the connection moment capacities are between the above limits or
alternatively for all connection moment design capacities less than the plastic moment
capacity of the connecting beam.

28
4.2.2 Moment capacity

The moment capacity of the connection is based on the strength of all components, which
may fail, see figure 4.8. For example, the failure behaviour of an I-beam-to-I-column con-
nection (capacity and stiffness) may be translated to that of an equivalent T-stub length.
The combination of all these stiffnesses and capacities result in the behaviour of the con-
nection. Thus for each component in the connection a reference is given to determine the
capacity, stiffness and where available the rotation capacity.

Grotmann (1997) analysed the behaviour of some welded and bolted connections between
I- or H-section beams and rectangular hollow section columns on the basis of the com-
ponent method. In principle, he used a methodology comparable to figure 4.8 and similar
to that used by Togo (1967) for tubular connections known as the ring model. For RHS
columns it is not a ring, but a frame with a certain effective length (see figure 4.9).

The equivalent effective length can be determined based on a yield line mechanism for the
flange to RHS column connection resulting in a similar capacity. In particular cases, he
obtained a good agreement with the actual moment rotation curves. However, in other
cases large deviations occurred and further evaluation is necessary before this method can
be used for the design of hollow section column connections.

Mode 1: Complete yielding of the flange

Mode 2: Bolt failure with yielding of the flange

Mode 3: Bolt failure

Figure 4.8 – Failure modes of actual components and equivalent T-stub flanges for bolted beam-to-
column connections (I-beams)

29
Figure 4.9 – Simplified frame for component behaviour

4.2.3 Rotational stiffness

The rotational stiffness of a connection is determined from the flexibilities of its basic
components. An advanced model (Jaspart 1997) is shown in figure 4.10. However, in
Eurocode 3: Annex J only linear springs for each component of connections between open
sections are given.

k3,1 k4,1 k5,1 k7,1

k5,2 k7,2
h1 k1 k2 k3,2 k4,2
h2 φj

a) Mj

keff,1 keq
keff,2
k1 k2 k1 k2 Z
φj φj

b) Mj c) Mj

P Parallel springs P Springs in series

1
1 2
P1+P2 2

P2

P1

δ δ1 δ2 δ1+δ2 δ

Figure 4.10 – Examples of spring models used for a bolted beam-to-column connection with an end
plate (Jaspart 1997)

30
In figure 4.10 the springs signify the behaviour of the following components:

k1: column web (compression) k2: beam flange (compression)

k3: column web (tension) k4: column flange (tension)
k5: bolt (tension) k7: flange plate (bending)

The springs k1 to k2 and k3 to k7 work in series whereas the results of k3,1 to k7,1 with k3,2
to k7,2 work in parallel.

For springs in series the deformations are added for the same force whereas for parallel
springs the forces are added for the same deformation, as indicated in figure 4.10.

4.2.4 Rotational capacity

With regard to the available rotation capacity, some indications are given in Eurocode 3:
Annex J, however, research is still underway to determine the required deformation
capacity for various systems and to determine the available rotation capacity for various
joint configurations (Boender et al. 1996).

4.2.5 Conceptual design

In the conceptual design of steel frames the actual dimensions of the connections are not
known, but assumptions have to be made for the stiffness and strength of the connections.
Steenhuis et al. (1994, 1996) and Jaspart (1997) give guidance for stiffness values for
various types of semi-rigid connections between open sections. Later on, the actual
stiffness values, Sj,act, have to be checked and should not deviate by more than the
following limits (Steenhuis et al. 1994):

for braced frames:

8El b 8S j, app Elb 10S j, app El b

for Sj, app< then ≤ Sj, act ≤ ........................... 4.9
Lb 10El b + S j, app L b 8El b - S j, app L b

for unbraced frames:

25El b 24S j, app Elb 30S j, app El b

for Sj, app< then ≤ Sj, act ≤ ......................... 4.10
Lb 30El b + S j, app L b 24Elb - S j, app L b

If these limitations are satisfied the load capacity of the frame between that with the
applied value (Sj, app) in the calculations and that with the actual stiffness (Sj, act) will
differ by less than 5%.

31
32
5 Simple shear connections

5.1 Introduction

The ends of members with simple shear connections are assumed to be rotationally
unrestrained or free to rotate under load. However, simple shear connections do actually
possess some rotational restraint. This is discussed further in chapter 4, which gives the
rigid, semi-rigid, and pinned joint classifications based on initial joint rotational stiffness
according to Eurocode 3 (CEN 1992). This small amount of moment resistance is usually
neglected and the joint is idealised to be completely flexible. Hence, simple shear
connections are sized only for the end reaction or shear force from the supported beam.
However, simple shear connections must still provide flexibility to accommodate the
required end rotations of the supported beam. To accomplish this, inelastic action at the
specified (unfactored) load levels in the joint is permitted. Thus, for most simple framing
systems the connection moment-rotation response (as shown in figure 4.7) remains linear
only in the initial stages of loading.

In some countries the building codes have a structural integrity/robustness requirement

that all shear connections be capable of carrying, in addition to the vertical shear load
parallel to the column, a further nominal horizontal load acting normal to the column. This
is to ensure that if accidental horizontal forces are present in a building, then the connec-
tions will remain intact and the building will not collapse. Assuming that the nominal
horizontal load on the connection occurs under the same load combination that produces
the maximum beam end reaction (shear) on the connection, then the resultant force on the
connection will be inclined to the axis of the column. This is similar to the case of an
inclined brace member connected to a column, which is covered in chapter 10.

When members are designed with simple shear connections, provision must be made to
stabilise the frame for gravity loads and also to resist lateral loads. Many of the familiar
simple (shear) connections that are used to connect I-section beams to I-section columns
can be used with hollow section columns. These include single and double angles (cleats),
unstiffened and stiffened seats, single shear plates (also termed “shear tabs” or “fin
plates”) and tee connections (Packer and Henderson 1997, AISC 1997, SCI 1991). One
additional connection type that is unique to hollow section connections is the through-
plate. One should note that this alternative is seldom required for structural reasons and it
incurs a significant cost penalty when a single shear plate connection would otherwise
suffice. Variations in attachments are more limited with hollow section columns since the
connecting element will typically be shop-welded to the hollow section column and
bolted to the supported beam. Except for seated connections, the bolting will be to the
web of an I-section (or other open profile) beam. Beam coping is generally not required
except for bottom-flange copes (removal of the bottom flanges) with double angle
connections, because of practical erection considerations (the beam is usually lowered
vertically down with its web between the angles.)

Simple beam-to-column connections could also be made to RHS columns by direct

bolting to the column wall. With such connections a beam, typically with a shop welded
flush end-plate, would be site bolted to a column using “blind bolts” or regular bolts in
flowdrilled holes. These fastening methods are described in chapter 3. One advantage of
bolting directly to the RHS column is that there are no protruding attachments to the
column, thereby requiring less care during transportation and erection. Testing of simple
shear beam-to-RHS column connections using single-sided bolting systems has not iden-

33
tified any special failure modes, so these shear connections can be designed using
normal practice (Yeomans 1996, Korol et al. 1993, Sherman 1995, France et al. 1999).

5.2 Limit states for simple shear connections

There are a number of limit states associated with the bolts, connecting elements (plates,
angles, tees), welds and beam webs that are applicable to the design of all shear connec-
tions, whether using hollow or open section columns, and the applicable national or
regional structural steelwork specifications should be followed for such design criteria. In
addition to these limit states, the following potential failure modes should also be checked
for shear connections to hollow section columns (AISC 1997):

(i) shear yield strength of the tube wall adjacent to a weld (for all connection types);
(ii) punching shear through the tube wall (for single shear plate connections only);
(iii) plastification of the tube wall, using a yield line mechanism (for stiffened seat
connections to RHS columns only).

Elaboration of the above three failure modes is provided in the following discussion on
various connection types. Mode (iii) above, representing a flexural failure of the hollow
section face, is not a limit state (with the one exception as noted) because the end
rotation of a beam supported at both ends is limited and is insufficient to develop a yield
line mechanism in the column connecting face (AISC 1997). However, Sputo and Ellifritt
(1991) performed tests on stiffened seat connections to the webs of I-section columns and
found that a yield line mechanism may be an applicable limit state. Since this situation
(connecting to a plate element that is supported for a long length on two opposite edges)
is similar to that for a RHS column face, the yield line mechanism is considered a possible
limit state for stiffened seat connections (AISC 1997). The inclusion of this failure mode
acknowledges that all “simple shear connections” will still have some end rotation of the
connected beam.

5.3 Single shear plate connections (shear tabs, fin plates)

When selecting the type of connection, one should bear in mind that RHS columns may
likely have a smaller width than the equivalent I-section column flange or web, which
thereby restricts the width of a connecting angle leg or flange of a tee. Moreover, the
factored shear load to be transmitted at a connection is often low so a single shear plate
connection, shown in figure 5.1, is frequently a logical and economic choice. One of the
earliest experimental studies on simple shear connections to RHS columns was done by
White and Fang (1966), but thereafter the topic received little attention for over 20 years.
Sherman and Ales (1991) and Sherman (1995) have investigated a large number of simple
framing connections between I-section beams and RHS columns, in which the load
imposed on the column was predominantly shear. In all, nine different types of simple
framing connections to RHS columns were considered structurally and with a relative cost
review. The latter showed that the single shear plate and single angle connections were the
cheapest. Double angle and fillet-welded tee connections were more expensive, while
through-plate and flare-bevel-welded tee connections were among the most expensive
(Sherman 1995).

34
 Section A – A

Figure 5.1 – Single shear plate connection

Single shear plate connection tests were performed with bolts both snug tight and fully
pretensioned. The connections with snug tight bolts had the same ultimate capacities and
eccentricities as those with pretensioned bolts. (The eccentricity is the distance from the
column face to the point of contraflexure in the beam, or the distance from the column face
to the line of action of the beam shear reaction.) However, at working loads pretensioned
bolts produced larger eccentricities (to the contraflexure point/inflection point where
negative moment changed to positive moment) and hence larger end moments in the
columns. It was found that the local distortion that does occur in the RHS wall (for
connections on one or both sides of the RHS) has negligible influence on the column
resistance provided the RHS is not thin-walled or slender. The definition of “slender” used
herein is a width-to-thickness ratio for the flat of the RHS connecting face exceeding
1.4 (E/fc,y) (AISC 1997). An extrapolation of this provision was also made by AISC (1997)
for CHS columns, wherein single shear plate connections were permitted for non-slender
CHS under axial load, which was defined by dc/tc ≤ 0.114E/fc,y. Thus, providing the
column wall is not slender (according to the above limits), which is normally the case for
most practical columns, there is no advantage to using through-plates (Sherman 1995).
A possible failure mode for the single shear plate connection is warping of the shear plate
due to twisting of the beam. It is therefore recommended that long unbraced beams
attached by shear plate connections be provided with lateral support in the vicinity of the
connection. Alternatively, avoid shear plate connections in such situations.

Over a wide range of connections tested by Sherman (1995, 1996), only one limit state was
identified for the RHS column. This was a punching shear failure related to end rotation of
the beam when a thick shear plate was joined to a relatively thin-walled RHS. Two con-
nections failed when the shear plate pulled out from the RHS wall at the top of the plate
around the perimeter of the welds. A simple criterion to avoid this failure mode is to ensure
that the tension resistance of the plate under axial load (per unit plate length) is less than
the shear resistance of the RHS wall along two planes (per unit plate length). Thus
(Sherman 1995, AISC 1997),
f 1 f,p,y tp · (unit length) < 2 f 2 (0.60fc,u) tc · (unit length) .....................................................5.1

35
In the above inequality the left hand side, the tensile strength of the plate, is multiplied (for
limit states design) by a resistance factor of f1 = 0.9 for yielding. The right hand side of
the inequality, the shear strength of the RHS wall, (for which the ultimate shear stress is
taken to be 0.6 of the ultimate tensile stress), is multiplied by a resistance factor of f2 =
0.75 for punching shear failure (AISC 1997).

Hence tp < (fc,u /fp,y) tc ........................................................................................................5.2

5.3.1 Connection to RHS column design example

The following example demonstrates all the typical limit states that need to be checked for
a simple I-section beam shear plate connection to a RHS column, along with the unique
criterion given by equation 5.2. To do this, it is necessary to conform to a particular limit
states structural steel specification and the Canadian standard CAN/CSA-S16.1-94 (CSA
1994) is used in this instance.

Connect a W410 x 39 Grade 350W beam via a single shear plate to a HSS 203 x 203 x 8.0
Grade 350W Class C column, to develop the capacity of the beam in shear. (An I-section
beam approximately 410 mm deep and weighing 39 kg/metre, with a yield stress of
350 N/mm2, is to be joined to a cold-formed square RHS measuring 203 mm x 203 mm x
8 mm, also with a yield stress fc,y = 350 N/mm2 and minimum ultimate stress fc,u = 450
N/mm2.)

Shear capacity of beam:

V* = 484 kN = required shear capacity of connection CSA Specification

Shear plate thickness:

Slenderness of the flat RHS face = (bc – 4tc)/tc
= (203 – 4(7.95))/7.95
= 21.5 < 1.4 (E/fc,y) = 33.5

Hence the RHS is not “slender” and equation 5.2 is applicable. Use Grade 300W plate with
fp,y = 300 N/mm2 and fp,u = 450 N/mm2.
tp < (fc,u /fp,y) tc = (450/300)7.95 = 11.93 mm eqn. 5.2
So choose 10 mm thick plate.

Bolts required:
Exclude the bolt threads from the shear plane, the bolts are in single shear, so try
4 M22 ASTM A325 bolts in punched holes.
Total bolt shear resistance, Vb* = 4 · (127) = 508 kN > 484 kN.
With bolts in punched holes, the effective hole diameter = bolt diameter + 4 mm
= 26 mm (24 mm punch for 22 mm diameter holes + 2 mm allowance for damage
to the edge of the hole caused by punching).

36
Bearing resistance:
Both beam web (thickness = 6.4 mm) and shear plate (tp = 10 mm) have steels
with an ultimate stress of 450 N/mm2, so bearing will be critical on the thinner
material (beam web).
B* = 3 f3 tb,w db n fb,w,u CSA Specification
= 3(0.67)(6.4)(22)(4)(0.450) = 509 kN > 484 kN.
In the above a resistance factor of f3 = 0.67 has been used for failure associat-
ed with a connector (equivalent to a partial safety factor of 1.5).

Plate length:
The clear distance between the beam root fillets for the W410 x 39 section
= 348 mm CISC Handbook
So choose a plate length, Lp = 340 mm.

Shear yield strength of tube wall adjacent to welds:

V* = 2 f1 Lp tc (0.6fc,y) AISC Manual
= 2(0.9)(340)(7.95)(0.6)(0.350) = 1,022 kN > 484 kN.
[The nominal RHS wall thickness is 7.95 mm.]

Net section fractures of shear plate:

The four bolts will be arranged in one bolt line, similar to the connection shown in
figure 5.2. All possible failure paths should, in general, be checked. After laying
out the bolts as shown in figure 5.2, two possible failure paths as illustrated in that
figure will be checked.
For figure 5.2(a):
V* = 0.85 f1 Ae fp,u CSA Specification
= 0.85(0.9)(340 - 4(26))(10)(0.6)(0.450)
= 487 kN > 484 kN.
For figure 5.2(b):
V* = 0.85 f1 Ae fp,u CSA Specification
= 0.85(0.9)[3(70-26)(0.6) + (65-13)(0.6) + (65-13)](10)(0.450)
= 559 kN > 484 kN.
This is clearly less critical than the failure path in figure 5.2(a) because the length
of the failure line is still the same but one part is now in tension rather than shear.

Net section fractures of beam web:

Non-critical unless the beam is coped.

37
Gross section yielding of shear plate:
V* = f1 Ag fp,y CSA Specification
= 0.9(340)(10)(0.300) = 918 kN > 484 kN.

Fillet welds:
A multipurpose electrode is chosen with an ultimate strength of 480 N/mm2. By
welding along the full length of the plate, on both sides, a weld shear resistance
of 0.762 kN/mm is provided by a 5 mm weld (CISC Handbook).
Hence, V* = 2(340)(0.762) = 518 kN > 484 kN.
So choose a fillet weld (leg) size of 5 mm. Generally, this weld would be carried
all around the plate. This design procedure has neglected the bending moment
on the weld caused by the eccentricity of the line of action of the shear force from
the RHS face, as this bending moment is small.

Tensile Rupture
65

70

Shear 70 Shear
Rupture Rupture

70

65

65 65

(a) (b)

Figure 5.2 – Two possible failure paths for net section fracture in shear plate

5.3.2 Connections to CHS columns

This is a popular form of connection (see figures 5.3 and 5.4) because connecting elements
do not need to be rounded or saddle-cut. Instead, the vertical shear plate can just be
fillet welded all around to the CHS column face. As noted in section 5.3, this type of
connection would be permitted for CHS columns that are not “slender”; i.e. dc/tc ≤
0.114E/fc,y. Aside from this provision, the design procedure would be the same as
described in section 5.3.1 for a connection to a RHS column.

38
Figure 5.3 – Single shear plate connection to CHS column

Figure 5.4 – Shear plate connection to CHS column

39
In figures 5.1 and 5.3 the connections are detailed such that the single shear plane of the
bolted connection aligns with the centre line of the column. Although this is a common
practice, an alternative might be to align the centre line of the beam with the centre line of
the column. It is believed that the capacity of the connection will be practically identical
with either detailing arrangement.

5.3.3 Single shear plate connections to RHS column corner

A variation on the connection shown in figure 5.1 can be made if the plate is connected to
the corner of the RHS column, as shown in figure 5.5. The plate is then connected to a
much stiffer part of the column cross-section, which thereby avoids any consideration of
the RHS wall slenderness as described in section 5.3.

Section D– D

Figure 5.5 – Single shear plate connection to corner of rectangular column

Such connections have been tested by White and Fang (1966) and no special failure limit
states have been noted. However, it should be emphasized that if cold-formed RHS
columns are used caution should be exercised if heavy welding is planned, as the tube
material will have a lower ductility in the corners.

5.4 “Through-plate” connections

With the through-plate connection shown in figure 5.6, two opposite faces of the column
(either RHS or CHS) are slotted so that the single plate can be passed completely through
the hollow section column. The plate is then welded to both faces of the RHS or CHS column.

40
 Section E – E

Figure 5.6 – Through-plate connection

The plate does act as reinforcement to the tube face, so this type of connection is
preferable to the single shear plate connection if a single plate is still preferred and the
column is a “slender” section (see section 5.3). However, the through-plate connection is
considerably more expensive than the single shear plate connection, so the latter should
be used if it suffices.

When a connection is made on both sides of the column, by using a long or extended
through-plate, the portion of the plate inside the hollow section is subject to a uniform
bending moment. For long connections this part of the plate may be liable to buckle in a
lateral-torsional mode prior to yielding, unless the depth of the column is small (AISC
1997).

5.5 End plate connections

A flexible end plate connection, generally with a plate thickness of only 8 or 10 mm, can
be partial depth and welded only to the beam web, to achieve a simple or pinned joint.
Tests on such connections to RHS columns, using flowdrill connectors, by France et al.
(1999, 1999a) have shown that these connections meet the EC3 criterion for pinned joints
(see figure 4.7). A common practice has been to use a full depth end plate and to weld this
both to the beam web and flanges, making what is commonly called a flush end plate
connection, but tests have confirmed that this joint type is semi-rigid by the EC3 criterion.
Semi-rigid joints are discussed in chapter 6. France et al. (1999, 1999a) found that the end
plate depth, end plate width, end plate thickness, bolt locations and column wall thickness
all affected the joint stiffness and strength, as may be expected, but no special connection
limit states were observed in their tests beyond those for conventional bolted shear con-
nections. Several connections were tested for the influence of column compression load,
with all the RHS columns being “non-slender” according to the limit given in section 5.3.
For RHS in this category, column axial stresses of up to 50% of yield had little
influence on the behaviour as a simple shear connection. Sherman’s (1995) tests on
connections with web end plates, bolted to RHS columns with flowdrilled connectors, also
confirm these recommendations.
41
A disadvantage of the end plate connection is that it will require site-bolting to the column
using a single-sided (or “blind”) bolting system (see Chapter 3). This type of connection –
like all the following connection types presented – is also only suitable for RHS columns,
not CHS columns.

5.6 Tee connections

With this connection, shown in figure 5.7, the flange of the tee is shop-welded to the RHS
column and the web of the tee is site-bolted to the beam web. Sherman (1995) has
performed tests on these connections with the tee flange narrower than the RHS, with
vertical fillet welds, and with the tee flange wider than the RHS, with flare-bevel groove
welds to the tube corners. Both details performed well but fillet welding to the flat of the
RHS is a more economical alternative.

Figure 5.7 – Tee connection

White and Fang (1966) originally proposed that the width to thickness ratio of the tee flange
be ≥ 10 in order to provide desired flexibility. Subsequent research by Astaneh and Nader
(1990) on tee connections to heavy I-section columns concluded that a tee flange width to
thickness ratio ≥ 13 provides sufficient flexibility for the joints to be considered as simple
(or pin-jointed). This has since been verified by shear tests on tee connections to RHS
columns by Dawe and Mehendale (1995). There is little difference in capacity, whether the
tee is centred or offset (to allow the beam to be on the column centreline).

AISC (1997) recommends that, in order to ensure rotational ductility, the tee web (or stem)
has a thickness ≤ db/2 + 2 mm. This same criterion could also be applied to a single shear
plate or through-plate. As noted in section 5.2, the only limit state unique to the RHS wall
to be checked is the shear yield strength of the tube wall adjacent to the vertical welds
(assuming the tee flange is welded to the flat of the RHS).

42
5.7 Single and double angle connections

A single angle connection (or angle cleat), see figure 5.8, is made with an angle on one side
of the beam web with the angle shop-welded to the RHS column. An L-shaped weld is rec-
ommended to provide adequate joint flexibility, with welding along the angle toe and
across the bottom of the angle, plus just a small weld return at the top of the angle (see
figure 5.9).

Figure 5.8 – Single angle connections to RHS column

43
 Section A – A

Section B – B

Figure 5.9 – Double angle connection

Welding across the entire top of the angle should be avoided as it would inhibit flexibility
(AISC 1997). A 100 mm x 75 mm angle is often selected, with the 75 mm leg welded to the
RHS. A minimum angle thickness of 10 mm (for M20 and M22 bolts) or 12 mm (for M24
bolts) is also recommended by AISC (1997). If fillet-welding the angle toe to the flat of the
RHS is desired, and the centre of the beam web is to be kept in line with the centre of the
RHS, then columns with a connecting face dimension of 200 mm or greater will typically
be needed. Alternatively, single angles can be welded to narrow RHS with a flare-bevel
groove weld. Assuming the former (fillet) welding procedure is used, the only limit state
unique to the RHS wall to be checked is the shear yield strength of the tube wall adjacent
to the vertical weld.

A double angle connection (or double angle cleat), as shown in figure 5.9, is one of the
most traditional simple shear connections. Pairs of angles are shop-welded along the
angle toes, with a small weld return at the top of the angle (see figure 5.9), then field-bolt-
ed to the beam web. This connection is sensitive to shop fabrication tolerances, and the
two angles may need to be pried apart to allow entry of the beam web on site. It is pru-
dent to cope the bottom of the beam (see figure 5.9) so that erectors can place the beam
by lowering it between the angles from above. If the beam is coped, block shear rupture
failure of the beam web (a tearout of the beam web, with the failure path passing through
the bolt holes) should be checked. Double angle connections provide the strength of bolts
in double shear combined with good flexibility and, being symmetrical, the connection
avoids any lateral torsion. Fabricators can prepare standard detail angles from stock,
rather than prepare special components such as tees, and many steel design handbooks
will give standard “pre-engineered connection designs” for this connection type. Sherman
(1995) has verified the adequacy of double angle connections to RHS columns and,
assuming the angle toes are welded to the flat of the RHS, the only limit state unique to

44
the RHS wall to be checked is the shear yield strength of the tube wall adjacent to the
vertical welds (AISC 1997).

5.8 Unstiffened seat connections

An unstiffened seated connection is made with a seat angle and a top angle, as illustrated
in figure 5.10.

Figure 5.10 – Unstiffened seat connection

Seated connections are common for connections with light loads and for applications such
as open web steel joints. While the seat is assumed to carry the entire end reaction of the
supported beam, the top angle (typically 100 mm x 100 mm x 100 mm long) must be placed
as shown, or in the alternative side location, for satisfactory performance and stability. To
provide adequate flexibility for the connection, only the toe of the top angle is welded to
the RHS. The thickness of the top angle ought to be 6 mm or greater to accommodate the
minimum size fillet weld to the RHS or beam flange. Even if there is no calculated horizontal
shear force transfer between the beam flanges and the seat angles, two M20 Grade 8.8 (or
ASTM A325) bolts are recommended for the bottom seat angle. Two bolts may also be
used to connect the top angle to the beam flange, or a fillet weld may be used across the
toe of the top angle. Again, the only limit state unique to the RHS wall to be checked
is the shear yield strength of the tube wall adjacent to the two vertical welds to the lower
seat angle (AISC 1997).

5.9 Stiffened seat connections

A stiffened seated connection is made in the same manner as an unstiffened seated con-
nection except the seat angle is replaced by a tee (either a structural tee or comprised of
two plates), wherein the web (or stem) of the tee is vertical and the flange of the tee (on
which the beam sits) is horizontal (see figure 5.11).

45
 ≥ 0.2hp

Figure 5.11 – Stiffened seat connection

The seat is again assumed to carry the entire end reaction of the supported beam and the
comments given above for the top angle of the unstiffened seated connection are again
applicable here. The supported beam must be bolted to the seat plate (tee flange) with two
bolts of at least M19 Grade 8.8 (ASTM A325) capacity, to account for prying action caused
by the rotation of the joint at ultimate load. Welding the beam to the seat plate is not
recommended. Also, the distance (Lb) from the RHS column face to the centreline of the
bolts should be not greater than the larger of {half the length of the seat plate (Lp)
measured normal from the RHS column face; and 67 mm}, for practical size beams (AISC
1997).

The thickness of the horizontal seat plate (or tee flange) should be at least 10 mm. Welds
connecting the two plates should have a strength not less than the horizontal welds to the
support under the seat. It is also a conservative recommendation that the thickness of the
tee web (or stem) tp (see figure 5.11) be (AISC 1997):

tp ≥ (fb,y /fp,y) tb,w and also ≥ cw .....................................................................5.3

where w is the weld (leg) size and c = 1.5 for fp,y of 350 N/mm2, with the welds being
assumed to be made of electrode having an ultimate strength of 480 N/mm2. Alternatively,
if the tee web (or stem) material has fp,y of 250 N/mm2, but the same (overmatching)
electrode is still used, then c can be taken as 2.

As mentioned in section 5.2, there are two limit states for the RHS face to be checked:

(i) shear yield strength of the tube wall adjacent to the two vertical welds along the tee
web (or stem). This failure mode has been cited many times and sample calculations
are given in section 5.3.1.
(ii) plastification of the tube wall, using a rotational yield line mechanism. A limit states
design resistance for the RHS connecting face under in-plane moment loading is given
in section 6.1.2 of CIDECT Design Guide No. 3 (Packer et al. 1992), for an RHS-to-RHS
tee joint. That yield line failure mode is deemed to only be applicable for joint width
ratios (the ratio in this case of the seat flange width to the RHS column width) less than
46
 Section A – A

Section B – B
Figure 5.12 – Simple shear connections with hollow section beams

47
Figure 5.12 – Simple shear connections with hollow section beams

48
Figure 5.12 – Simple shear connections with hollow section beams

49
or equal to 0.85. In applying that yield line solution, the depth of the stiffened seat
(hp) replaces the term for the branch member depth (h1), the width of the seat flange (which
is recommended to be at least 0.4hp) replaces the term for the branch
member width (b1) and the bending moment applied to the RHS column face is the beam
shear force reaction multiplied by the effective eccentricity (e) of this load from the column
face. AISC (1997) takes this effective eccentricity, for this connection type, as 0.8Lp (see
figure 5.11). In section 6.1.2 of CIDECT Design Guide No. 3 it can be seen that this RHS
column face moment resistance is also reduced by the effect of the axial compression load
in the column.

5.10 Hollow section beams to hollow section columns

I-section beams are the usual choice in pin-jointed (simply connected) frames, and these
have been presumed in the preceding sections. However, there are instances where
hollow sections are used for both the columns and beams. Detail material such as tees,
angles and plates is again generally used and site-bolted connections are typical. Some
examples are given in figure 5.12.

Figure 5.12(a) shows a double tee connection, with the tees either built up from plates or
cut from rolled sections. Central alignment of the beam and column can be maintained by
offsetting the tees. The column tee should be welded only along the vertical edges, and
should include a short weld return around the top corners. Design guidance for propor-
tioning the column tee is given in section 5.6. Figure 5.12(b) shows a pair of angles that
provide double shear loading on the bolts. Welding of these angles to the column is the
same as for the tee in figure 5.12(a).

Figure 5.12(c) illustrates a relatively narrow beam framing into a wide RHS column. Angles,
welded near the column corners, are used on either side of the beam that has the bottom
flange cut back to provide access for bolting. Eccentric loading in the plane of the column
face upon the angle welds may limit the capacity of this arrangement. Hence, this uncon-
ventional connection type should only be used for lightly-loaded beams. Figure 5.12(d)
shows a better match of beam and column widths where two shear plates are used. This
would be a relatively stiff connection with the plates welded near the column corners.
Again, a cutout to the beam bottom flange is used, to facilitate conventional bolting. A
slight variation of this connection is shown in figure 5.12(e), where the members have equal
width. Some of these connection arrangements have potential for difficult fitting at the site
due to welding-induced distortions. It may be necessary to spread pairs of plates slightly
with jacks after the welds cool, because welding contraction will tend to deflect and pull
the plates together.

Figure 5.12(f) portrays a popular end plate connection, with bolting done easily beyond the
width of the members. This connection accommodates both RHS and CHS beams and
allows the hollow section beam to be sealed. A variation of this connection concept is
shown in figure 5.13.

5.11 Use of through-bolts to hollow section columns

Rather than using “blind bolts” (or a “single-sided bolting system”) when bolting an
element such as a beam end plate to a RHS column, it is possible to use long bolts or
threaded rods which pass all the way through the RHS column member. Thus, both the
head and the nut of the bolt (or threaded rod) are on the outside of the RHS. This is accept-

50
able in a shear connection if the bolts are only snug tight (i.e. not fully pretensioned).
Single shear plate connections have been tested by Sherman (1995) with bolts snug tight
and fully pretensioned, and both have performed adequately. The connections with snug
tight bolts had the same ultimate capacities and eccentricities as those with pretensioned
bolts. However, at working loads, pretensioned bolts produced larger eccentricities (to the
contraflexure point where negative moment changed to positive moment).

In many connections the bolts will be fully pretensioned, especially if the bolts are liable to
go into tension. With full pretensioning, through bolts should only be used if the bolt inside
the RHS walls passes through a spacer tube or pipe. This spacer tube should have a
length equal to the inside depth of the RHS so that when the bolt is tensioned the spacer
tube is placed in compression, hence preventing the flexible faces of the RHS from being
pulled inwards and thereby being deformed during the erection process. Section 3.5.6 of
CIDECT Design Guide No. 6 (Wardenier et al. 1995) gives some fabrication advice for such
joints.

5.12 Influence of concrete slabs on behaviour of connections

Composite floor systems are now commonplace in “steel frames”. A concrete slab around
the RHS column increases the rotational stiffness of the beam-to-column connection.
Thus, a simple shear connection is actually transformed into a composite semi-rigid
connection. More information on semi-rigid connections is provided in chapters 6 and 9.

Figure 5.13 – Double channel beams bolted to plates on RHS column faces

51
52
6 Semi-rigid connections

6.1 Types of semi-rigid connections with hollow section members

Moment connections in tubular structures can be designated as indicated in figure 6.1.

Figure 6.1 – Designation of moment connections with hollow section members

All investigations up to now concern mainly the determination of the moment capacity.
Only formulae for unstiffened CHS-to-CHS connections exist for the (initial) joint stiffness,
e.g. Efthymiou (1985).

For unstiffened CHS-to-CHS, RHS-to-CHS and I-beam-to-CHS connections formulae for

the moment capacity are given in Eurocode 3: Annex K (CEN, 1992), IIW-XV-E (1989) and
CIDECT Design Guide No. 1 (Wardenier et al. 1991) mainly based on Kurobane (1981).
A detailed study including multiplanar connections is given in Van der Vegte (1995).

Formulae for the moment capacity of CHS-to-CHS connections are also given in API
(1997), AWS (2002) and in the draft ISO standard (ISO, 1997), however these standards use
a different database and give different recommendations.

Recently a reanalysis has been carried out by Ariyoshi and Makino (2000) on plate-to-CHS
connections. This information can also be used for RHS to CHS and I-beam-to-CHS con-
nections.

53
De Winkel (1998) gives information for the strength of uniplanar and multiplanar
I-beam-to-CHS connections, but the stiffness can only be derived from the moment rota-
tion diagrams.

For unstiffened RHS-to-RHS and I-beam-to-RHS connections, design formulae are given
in Eurocode 3: Annex K (CEN, 1992), IIW-XV-E (1989) and the CIDECT Design Guide
(Packer et al., 1992) mainly based on Wardenier (1982). Yu (1997) has recently carried out
a very detailed study on uniplanar and multiplanar RHS-to-RHS connections with various
loading combinations.

Lu (1997) gives information for the strength of uniplanar and multiplanar I-beam-to-RHS
connections, but, again, the stiffness can only be derived from the moment
rotation diagrams. For unstiffened I-beam-to-CHS and I-beam-to-RHS, the moment
capacity is based on evidence given in Wardenier (1982) and Packer et al. (1992).

Bolted connections are now becoming more popular due to the use of blind bolting sys-
tems. Test results and design recommendations are given by Yeomans (1996, 1996a, 1998)
and summarised by Packer and Henderson (1997). Tanaka et al. (1996) describe an inter-
esting new development where the columns have a locally increased wall thickness, which
is produced by using local induction heating and axial pressure.

For frames, the I-beam-to-CHS and I-beam-to-RHS connections are most frequently used,
since I- or H-sections are primarily used for beams and circular or rectangular hollow sec-
tions have advantages for columns. Section 6.2 will concentrate on moment connections
between hollow sections (CHS-to-CHS and RHS-to-RHS) and section 6.3 will concentrate
on unstiffened I-beam-to-CHS or RHS connections.

6.2 Welded hollow section beam and column connections

Beam to hollow section column connections behave in a similar way to the Vierendeel or
moment connections dealt with in the CIDECT Design Guides Nos. 1 and 3. However, for
completeness they will be dealt with here again. The usual combinations are members of
the same type, i.e. CC or RR connections. Other combinations like CR or RC are very rare
for moment connections.

6.2.1 CHS beam and column members

The strength of moment connections between CHS beams and columns is based on the
chord plastification and the chord punching shear criterion. The design strength criteria are
given in figure 6.2. Although out-of-plane moments do not occur in plane frames, the
strength criteria for out of plane moments are also given for completeness and for the
analysis of three-dimensional frames.

54
 Type of connection Factored connection resistance

T,Y,X Chord plastification

db
f (n' )
Mb,ip
Mb* ,ip % 4.85 " fc,y " t c2 " $ 0.5 " # " db "
!b tb sin !b

dc
tc

T,Y,X Chord plastification

2. 7 f (n' )
Mb* ,op % fc,y ! t c2 ! ! ! db
1 $ 0.81! # sin "b
!b

Mb,op

General Punching shear check

fc,y 1 # 3 " sin !b

Mb* ,ip $ " t c " db2 "
Punching shear check for 3 4 " sin2 !b
db # d c " 2 ! t c fc,y 3 # sin !b
Mb* ,op $ " t c " db2 "
3 4 " sin2 !b

Functions

f (n' ) $ 1 # 0.3 " n'!0.3(n' ) 2 for n' ! 1.0

f (n' ) ! 1.0 for n' " 1.0

n' ! f c,p / f c,y

Range of validity

db Class 2 and 30! ! "b ! 90!

0.2 " ! 1.0
dc db e " ! 25
! 25 0.55 " ! 0.25
2 " tb dc
# " 20 ( X ! jo int s)

Figure 6.2 – Design recommendations for CC connections loaded by primary bending moments

55
The function f(n’) represents the influence of the axial and/or bending stress in the column
on the resistance according to the chord plastification criterion. As shown in CIDECT
Design Guide No. 1 (Wardenier et al. 1991) the design strength for in-plane bending mo-
ments is considerably larger than that for out-of-plane moments. For a good efficiency it
is recommended to choose the diameter to thickness ratio dc / tc of the column preferably
below 25 and the yield stress-thickness ratio between column and beam fc,y tc > 2 · fb,y · tb

For stiffness reasons the diameter ratio b between beam and column should be large,
preferably close to 1.0.

Combinations of axial loads and bending moments have to be checked for interaction
according to:

3 4
2
Nb M b, ip M b, op
+ + ≤ 1.0 ............................................................................................6.1
Nb* Mb, ip* Mb, op*

In which Nb , Mb,ip and Mb,op are the loads acting, and Nb*, Mb,ip*and Mb,op*are the
design capacities.

The rotational stiffness C (moment per radian) for Vierendeel connections with 0.3 ≤ b ≤
0.8 is given by Efthymiou (1985).

For T- and Y-connections under in-plane moment loading the connection stiffness is:

Cb,ip = 1.3 E (dc/2)3 b(2.25+g/125) g-1.44 /sin(b+0.4) (u) .....................................................................6.2

For T- and Y-connections under out-of-plane moment loading the connection stiffness is:

Cb,op = 2.3 E (dc/2)3 b(2.12) g(0.7(0.55-b)2-2.2) /sin(b+1.3) (u) ...............................................................6.3

These equations can be used within the following validity range:

0.3 ≤ b ≤ 0.8 10 ≤ g ≤ 30 0.3 ≤ t ≤ 0.8 35°≤ u ≤ 90°

It should be noted that the stiffness can be affected considerably by the presence of axial
and/or bending stresses in the column, however insufficient test evidence is available to
quantify this influence in more detail.

Van der Vegte (1995) investigated, among other aspects, the geometrical and the loading
effect in CHS multiplanar connections. In this study all kinds of loading situations were
considered on the in-plane and out-of-plane members. However, there are so many com-
binations of loading and the interactions are too complicated for routine design. Therefore,
these interaction formulae have not been included in this design guide, but information can
be obtained from the given reference.

56
Figure 6.3 – CHS knee connections

Some special types of connections are given in figure 6.3 for frame corners. These knee
connections have especially been investigated at the University of Karlsruhe (Karcher and
Puthli, 2001 and Choo et al., 2001). They recommend designing these connections based
on the following requirements for both members:

N M
+ ≤a ...........................................................................................................................6.4
Npl Mpl

The term a is a stress reduction factor, which can be taken 1.0 for mitre connections with
stiffening plates. For the mitre connections without stiffening plates it is a function of the
cross sectional dimensions and is given in figure 6.4 and equation 6.5. The S grade in fig-
ure 6.4 corresponds to the nominal yield stress fy.

a = (0.05 d/t + 0.77)-1.2 (235/fy)0.5 .....................................................................................6.5

Based on previous work (Mang et al., 1991), it is recommended that for connections with-
out stiffening plates the shear force V and the axial force N should not exceed:

V/Vpl ≤ 0.5 and N/Npl ≤ 0.2

For mitre knee connections with angles u > 90° the same recommendations as for u = 90°
can be adopted (Karcher and Puthli, 2001).

Although the unstiffened connections have been investigated for 10 ≤ d/t ≤ 100, it is rec-
ommended that for structural applications d/t is restricted to class 1 sections according to
EC 3 (CEN, 1992).

57
 1.6
1.4
stiffened
1.2
reduction factor α

1.0
0.8 S 235
S 355
0.6 S 460
S 690
0.4 unstiffened
S 890
0.2
0.0
0 10 20 30 40 50 60 70 80 90 100
d
t
Figure 6.4 – Stress reduction factor a, for unstiffened mitred CHS knee connections

The connections with a stiffening plate can be considered to be rigid, whereas the stiffness
behaviour of the unstiffened connections depends on the d/t ratio.

The unstiffened connections can only be assumed to be rigid for very low d/t ratios. No
formulae for the joint stiffness are available.

For those structural applications where a reasonable strength, stiffness and rotational
capacity are required it is recommended that a stiffened connection using class 1 sections
is used. For other structural applications it is recommended to use the unstiffened con-
nections only if the sections satisfy at least the plastic design requirements.

The stiffening plate thickness should satisfy tp ≥ 1.5 tb and not be taken smaller than
10 mm. An additional requirement is that d/t ≤ 50/(235/fy)0.5.

6.2.2 RHS beam and column members

The strength of welded moment connections of RHS beams and columns without stiffen-
ers is based on various failure modes, i.e.:

• column face yielding (plastification);

• cracking of the column face (chord punching shear);
• cracking in the beam (effective width);
• yielding or crippling of the column side walls;
• column shear.

These failure modes and the strength criteria are dealt with in detail in Wardenier (1982)
and in CIDECT Design Guide No. 3 (Packer et al. 1992) and are here summarised in figure
6.5 for a connection of square hollow sections with u = 90°. The strength criteria for

58
out-of-plane moments are also given for completeness and for three-dimensional frames.
However, it should be noted that the formulae for out-of-plane loading should only be used
if distortion of the chord cross section is prevented, e.g. by stiffeners located close to the
connection. These design recommendations have also been adopted for Eurocode 3,
Annex K (CEN 1992).

The moment capacity of connections with low to moderate b values (0.85) can be deter-
mined with a yield line model. The function f(n) is a function to allow for the reduction in
moment capacity due to the presence of compression stresses in the column face. For
values b > 0.85, depending on the geometry parameters, several failure modes may be
critical. As shown in figure 6.5, the beam effective width criterion and the column side wall
failure criterion have to be checked. For a better understanding these criteria are illustrat-
ed in figure 6.6.

Punching shear was not observed in the tests and not given as a separate check in figure
6.5 but it is recommended to design the beams with a thickness tb < 0.6tc or avoid
connections with bb ≈ bc - 2tc where punching shear may occur. More detailed information
is given in CIDECT Design Guide No. 3 (Packer et al., 1992).

From the expressions in figure 6.5 it can be seen that full width (b = 1.0) unstiffened RHS
Vierendeel connections are capable of developing the full moment capacity of the beam,
providing bc/tc is sufficiently low. For hc = bc= hb = bb and bc/tc< 16 the chord side wall
crippling is given by Wardenier (1982):

M*b,ip ≈ 8 . f c,y t c
•

........................................................................................................6.6
M pl bc/tc f b,y t b
•

Thus, for beam to column connections of square sections with a b ≈ 1, a column width to
thickness ratio bc/tc = 16 and a column to beam thickness ratio tc/tb= 2 the moment
capacity will be equal to the beam plastic moment capacity. This agrees with findings from
Korol et al. (1977).

The previous expressions for the moment capacity are based on moment loading only,
however axial loads in the beams may also exist. The interaction between axial loads and
bending moments depends on the failure criterion. A conservative approximation is to use
a linear relationship:

Nb + M b, ip + M b, op ≤ 1.0 ..........................................................................................6.7
Nb* M*b, ip M*b, op

Yu (1997), in a similar way to van der Vegte (1995) for CHS connections, investigated the
geometrical and the loading effect in RHS multiplanar connections. In her study all kinds
of loading situations were considered on the in-plane and out-of-plane members.
However, there are so many combinations of loading and the interactions are too compli-
cated for routine design. Therefore, these interaction formulae have not been included in
this design guide, but information can be obtained from the given reference. The work of
Yu confirmed that the CIDECT formulae for moment loaded RHS to RHS connections in
figure 6.5 give a lower bound for the FE results based on the load or moment capacity at
a local deformation of 3% of the column width bc.

59
 Type of connection Factored connection resistance
T and X connection under in- " ! 0.85 basis: chord face yielding
plane bending moments

(" 1 2 h / b %"
M b* ,ip , f c,y ! t c2 ! h b ! ' + + b c $ ! f (n )
"& 2 ! h b / b c 1 * ) (1 * )) "#
Mb,ip
0.85< " ! 1.0 basis: effective width
!b
'! / b , $!
Mb* ,ip 0 f b,y ) &Wb,pl ( --1 ( e ** ) b b ) t b ) (h b ( t b )#
!% . bb + !"
Mb,ip Mb,ip
0.85< " ! 1.0 basis: chord side wall failure
!b
M b* ,ip # 0.5 ! fk ! t c ! (h b " 5 ! t c ) 2

T and X connection under out-

" ! 0.85 basis: chord face yielding
of-plane bending moments

Mb,op (" h ! (1 + )) 2 ! b c ! b b ! (1 + )) %"

Mb* ,op , f c,y ! t c2 ! ' b + $ ! f (n)
"& 2 ! (1 * )) (1 * )) "#

0.85< " ! 1.0 basis: effective width

!
M b* ,op # f b,y " Wb,pl ! 0.5 " t b " (b b ! b e ) 2 "
Mb,op
0.85< " ! 1.0 basis: chord side wall failure

M b* ,op $ f k " t c " (h b # 5 " t c ) " (b c ! t c )

Mb,op

Function
f(n) = 1.0 for n ! 0(tension) 10 f c,y " t c
be # " " bb ! bb
0.4 b c / t c f b,y " t b
f(n) =1.3 + ! n for n < 0
"
but ! 1.0 fk ! f c,y for T connections
Nc Mc f k " 0.8 ! f c,y for X connections
n# "
A c ! fc,y Wc,el ! fc,y

Range of validity

Braces: class 2 sections bc h

and c ! 35
tc tc

"b ! 90!

Figure 6.5 – Design recommendations for RHS-to-RHS connections loaded by primary bending
moments

60
 hb
Mb,ip

tb

bc

tc
hc

bb

!
a. Yield line mechanism for chord face yielding under in-plane bending
hb

0.5be
0.5be
Mb,ip

o
! = 90

b. Brace effective width criterion for T, Y and X joints

Mb,ip
hb + 5 tc
hb
fk
fk
o
! = 90 0.5hb + 2.5 tc
tc

hb + 5 tc
hc

c. Chord side wall bearing or buckling failure model under in-plane bending

Figure 6.6 - Failure criteria for RHS-to-RHS moment connections loaded by in-plane bending

61
The design capacities for axial loading Nb* can be obtained from the CIDECT Design Guide
No. 3 (Packer et al., 1992) and are not reproduced here again.

The connections between rectangular hollow sections with ratios b < 1.0 are not stiff
enough to be used as moment connections. However, they can be stiffened by plates or
haunches.

Figure 6.7 shows some knee connections for Vierendeel girders or for frame corners. These
knee connections have been investigated at the University of Karlsruhe (Mang et al., 1991)
and at the University of Sydney (Wilkinson and Hancock,1998). Based on the research evi-
dence it is recommended in CIDECT Design Guide No. 3 to design these connections based
on the following requirements for both members:

N M
+ ≤a ...............................................................................................6.8
Npl Mpl

with V/Vpl ≤ 0.5 and N/Npl ≤ 0.2 ...........................................................................................6.9

Here N, M and V refer to the acting axial force, the acting bending moment and the acting
shear force in a connecting member at the connection, whereas Npl, Mpl and Vpl are the
capacities of the connecting member with

Vpl = 2h t fy/ 3 .............................................................................................6.10

bc= bb

bb= bc Section A - A Detail C Detail D

D tb
o o
60 a=t 60
A hb t

tc A C o
θ = 90
hc

bb= bc Section B - B Detail E Detail F

bc= bb
F tb a=t
t
B hb
tp
tc B E o
θ = 90
hc

Figure 6.7 – RHS knee connections

62
Similar to the approach for CHS-to-CHS knee connections the term a is a stress
reduction factor, which can be taken as 1.0 for mitre connections with stiffening plates. For
the mitre connections without stiffening plates it is a function of the cross sectional dimen-
sions and is shown in figures 6.8 and 6.9. If mitre knee connections are used with an angle
u > 90° between the members use conservatively the same design checks as for u = 90°.

h/b !=
1.0 0.9 0.85 0.8 0.78 0.76 0.74
3.5
0.72

0.7
3.0
0.68
b
2.5 0.66
0.64 h
2.0 0.62 t
0.6
0.58
1.5 0.56
0.54
0.52
1.0
5 10 15 20 25 30 35 40 45
b/t

Figure 6.8 – Stress reduction factor a, for 90° unstiffened mitred RHS knee connections subjected to
bending about the major axis

!=
b/h
1.0 0.9 0.8 0.70.65 0.6 0.56 0.53 0.50.48 0.460.44 0.42
3.0

2.5

b
2.0
h
t
1.5

1.0
5 10 15 20 25 30 35 40 45
b/t

Figure 6.9 – Stress reduction factor a, for 90° unstiffened mitred RHS knee connections subjected to
bending about the minor axis

63
Since the rotation capacity of the unstiffened connections might be rather low, it is also
recommended here to use a stiffened connection for those structural applications where a
reasonable rotational capacity is required. For other structural applications it is recom-
mended to use the unstiffened connections only if the sections satisfy at least the plastic
design requirements.

The stiffening plate thickness should satisfy tp ≥ 1.5t and not be taken smaller than
10 mm.

Additional requirements are that the welds should be at least equal to the connected wall
thickness and that the factor a used in design should be:

a < 0.84 for fy = 235 N/mm2

a < 0.71 for fy = 355 N/mm2

The connections with a stiffening plate can be considered to be rigid whereas the stiffness
behaviour of the unstiffened connections especially depends on the b/t and h/t ratio. Only
for very low b/t ratios can the connection be assumed to be rigid. No formulae for the joint
stiffness are available.

An alternative form of connection reinforcement is a haunch of the same width as the con-
nected RHS members on the inside of the knee. However, insufficient test evidence is
available to quantify the properties, especially the rotational capacity, of this connection
type.

6.2.3 CHS and RHS beam and column members

Connections with a CHS beam and a RHS column are very rare and not efficient in
transferring moments. Connections with a RHS beam and a CHS column are not
frequently used because of the end preparation required, however, with the current end
cutting machines, the end preparation is not a problem anymore. For moment loading no
test results are available, but based on the research on plate and I-beam-to-CHS column
connections recommendations are given in CIDECT Design Guide No. 1 which have been
adopted for Eurocode 3 and will also be adopted here.

The design formulae are given together with the I-beam-to-CHS column connections in
figure 6.11.

6.3 Welded I-beam-to-hollow section column connections

The first investigations on unstiffened connections between plates or I-beam and CHS
hollow sections have been carried out in Japan. The work of Akiyama, Kamba, Kanatani,
Kurobane, Makino, Sasagawa, Suzuki, Tabuchi, Taguchi, Tanaka and Wakabayashi, main-
ly published in Japanese papers, has been first summarised and analysed by Kurobane
(1981). Later reanalyses have been given by Wardenier (1982), Makino (1984), Kamba and
Taclendo (1998) and recently by Ariyoshi and Makino (2000).

Unstiffened moment connections between plates or I-beams and RHS hollow sections
have initially been investigated by Kanatani et al. (1980).

64
Ting et al. (1991) and Shanmugan et al. (1993) investigated numerically the effect of vari-
ous types of external stiffeners for I-beam-to-RHS columns.

The rigid diaphragm stiffened connections have been extensively studied in Japan and
summarised by Kurobane (1981) and Kamba et al. (1995, 1998). Most of the other research
is related to simple shear connections (using shear tabs, plates or cleats) or rigid moment
connections using straps or diaphragm plates or other reinforcing plates, see chapters 5
and 8.

In the nineties, an extensive programme has been carried out by Lu (1997) and de Winkel
(1998) to investigate the behaviour of unstiffened uniplanar and multiplanar connections
between I- or H-section beams and circular or square hollow section columns (see figure
6.10).

axially loaded plate-to-

CHS column

axially loaded I-beam-

to-CHS column

I-beam-to-CHS column
loaded by inplane
bending

I-beam-to-CHS column
with a floor

Figure 6.10 - I-beam-to-CHS or RHS column connections investigated by de Winkel (1998) and by Lu
(1997); only the I-beam-to-CHS column connections are shown here.
65
Within this programme, the following aspects have been investigated both for circular hol-
low section columns and for square hollow section columns.

1. Behaviour of plate to tubular column connections

Multiplanar geometrical effect
Multiplanar load effect
2. Interaction of two plates at different distances
Effect of beam web
3. Behaviour of I- or H-beam to tubular column connections loaded by in-plane bending
moments
Multiplanar geometrical effect
Multiplanar load effect
4. Effect of a steel plate floor (offshore)
5. Effect of a composite steel-concrete floor (buildings)
6. Influence of concrete filling of the column for the various load conditions
7. Influence of column loading or the moment capacity

The programme was set up in such a way that information could be obtained for
particular parts, i.e. flange and web and the various parameter influences investigated.
The intention was to simulate the behaviour of the more complicated connections by
combination of the separate effects. For example, the moment rotation diagram for a beam
to column connection with a composite floor can be built up from the load-deformation
behaviour of the connection of the bottom flanges, the web, the bolts, the studs between
beam and concrete and the reinforcement of the concrete. However, such a component
method could not be presented in a simple and sufficiently accurate way.

With numerical simulations many load deformation and moment-rotation curves have been
established. The work concentrated on the strength formulation but information for the
stiffness can be obtained from the many load (moment)-deformation (rotation) diagrams.

All models used in the parameter study were carried out for columns with a diameter or
width of 300 mm and varying thickness and beam dimensions. All welds were modelled as
butt (groove) welds which results in somewhat lower results than specimens with fillet
welds.

The ultimate load capacity was defined as the peak in the load-deformation curve or
moment-rotation curve, or, if reached earlier, the load or moment at which a local deflec-
tion of 3% bc or 3% dc occured in the column wall (Lu et al. 1994, Lu 1997).

6.3.1 I-beam-to-CHS column connections

As mentioned in section 6.3, recent analyses for gusset plate-to-CHS connections have
been given by Kamba and Taclendo (1998) and by Ariyoshi and Makino (2000). However,
these formulae for the yield and ultimate strength need further evaluation and modification
to design strengths and further simplification. This may be done for the next revision of the
IIW-XV-E and CIDECT design recommendations. In principle the same applies to the work
of de Winkel (1998) on unstiffened I-beam-to-CHS column moment connections.

In this design guide the recommendations are consistent with the recommendations in
CIDECT Design Guide No. 1 which have been based on Kurobane (1981), Wardenier (1982)
and other reanalyses for the IIW-XV-E committee and Eurocode 3 and later confirmed by

66
Makino et al. (1991). These formulae have also been adopted for Eurocode 3. Where
required, the recommendations here have been extended or modified based on the men-
tioned recent investigations.

The recommended formulae for the design strength of a plate, an I-beam and a RHS-to
CHS column connection are based on the ring model approach for chord plastification with
a statistical curve fitting. They are given in figure 6.11.

Design strength for XP- and TP-joints

Axial loading
Type of N* = f (β) ⋅ f (η) ⋅ f (n' ) ⋅ fc, y ⋅ t c2 Bending Bending
connection in plane out-of-plane
f (β) f (η) f (n' ) f c,y ⋅ t c2

XP-1/TP-1
dc
5.0
1 f (n' ) f c,y ⋅ t c2 ------ M b* ,op = 0.5 ⋅ b b ⋅ N( XP−1)
1 − 0.81 ⋅ β

tc
bb

XP-4/TP-4
dc
5.0
θi 1 + 0.25 ⋅ η f (n' ) f c,y ⋅ t c2 M b* ,ip = h b ⋅ N(*XP−1) M b* ,op = 0.5 ⋅ b b ⋅ N( XP− 4 )
1 − 0.81 ⋅ β
hb η≤4

tc
bb

XP-5/TP-5
dc

5.0
θi 1 + 0.25 ⋅ η f (n' ) f c,y ⋅ t c2 Mb* ,ip = hb ⋅ N(*XP−5 ) M b* ,op = 0.5 ⋅ b b ⋅ N( XP−5 )
hb 1 − 0.81 ⋅ β
η≤2 η≤2

tc
bb

General Punching shear check

Punching shear fb ⋅ t b, f ≤ 1.16 ⋅ f c, y ⋅ t c for XP-1 / TP-1 (general)
check for
for XP-4 / TP-4 (bending in plane)
b b ≤ dc − 2 ⋅ t c fb ⋅ t b, f ≤ 0.58 ⋅ f c, y ⋅ t c for other cases

Function
2
f (n' ) = 1 + 0.3 ⋅ n'−0.3 ⋅ (n' ) for n' ≤ 1.0
f (n' ) = 1.0 for n' > 1.0

fc,p
n' =
fc,y

Range of validity
dc
θ i = 90 ο ≤ 40 beams: class 2
tc

Figure 6.11 – Design strength formulae for uniplanar RC and IC connections.

67
The formulae for the moment capacity of the I-beam to column connection are based on
the strength of the plate connections. In principle the connections with one beam at one
side of the column (indicated as TP) behave somewhat different from those with a beam
at both sides of the column (indicated as XP) and in the mentioned literature different
formulae are given. From the statistical evaluation the function f(b) for the plate-to-CHS
connections is as follows:

for XP-1 connections with two plates:

f(b) = 5/(1 – 0.81b) ...........................................................................................6.11

for TP-1 connections with one plate:

f(b) = 4 + 20b2 ...........................................................................................6.12

However, due to the statistical curve fitting procedures the resulting formulae are not cor-
rect for small b values, i.e. the XP connection becomes stronger than the TP connection
which is physically incorrect. Therefore, here it is recommended to use the formulae of
figure 6.11.

From the work of Ariyoshi and Makino (2000) indications can be given for the axial stiff-
ness of flange plate connections (TP and XP connections). The formulae for the initial stiff-
ness of the flange plate connections have been simplified here and are given by equations
6.13 and 6.14:

for XP-1 connections with two plates:

KXP-1 = 6.8 E tc b (2g)-1.3 ..........................................................................6.13

for TP-1 connections with one plate:

KTP-1 = 1.9 E tc b1.3 (2g)-0.7 ........ .................................................................6.14

The original and the simplified equations are presented in figures 6.12 and 6.13

For the moment rotation behaviour the following equations apply:

2d 2N
Mb, ip = Np (hb - tb,f) = Cb, ip f = Cb, ip = Cb, ip ....................6.15
(hb - tb,f) K(hb - tb,f)

or Cb,ip = 0.5 K(hb - tb,f)2 .....................................................................6.16

Thus multiplying the axial flange plate connection stiffness K by 0.5 (hb- tb,f) 2 gives an
approximation for the initial rotation Cb, ip for IC moment connections, however the effect
of the beam web is neglected.

68
 0.3
Equation for the
stiffness of
XP connections

0.2 Original equation

Etc
k

2! = 15
2! = 25
2! = 35
0.1 2! = 50

Simplified equation

0.0
0.0 0.2 0.4 0.6 0.8 1.0
!
Figure 6.12 – Stiffness of XP-1 connections

0.3
Equation for the
stiffness of
TP connections

0.2 Original equation

Etc
k

2! = 15
2! = 25
2! = 35
0.1 2! = 50

Simplified equation

0.0
0.0 0.2 0.4 0.6 0.8 1.0
!
Figure 6.13 – Stiffness of TP-1 connections

69
From the work of de Winkel (1998) the following conclusions can be drawn:

– Comparing the multiplanar connections with the uniplanar connections shows that the
geometrical stiffening effect becomes only significant for b values close to 0.7; for b ≤
0.5 it is negligible. However, for b values close to 0.7 the rotation capacity may also
decrease since cracking in the column between the beam flanges may be the critical
failure mode.

– Positive load or moment ratios J (i.e. the ratio of the load or moment out-of-plane divi-
ded by that in-plane) generally show some beneficial effect, whereas negative load
ratios J show a considerable decrease of the capacity.

This effect is here somewhat simplified compared to the original very complicated formu-
la and given by:

Mj/Mj=0 = 1 + J (b – 0.4 b2 – 0.1) for J < 0 ..............................................................6.17a

Mj/Mj=0 = 1 for J ≥ 0 ..............................................................6.17b

The simplified equations generally are conservative except for low b values in
combination with J > 0. Figures 6.14 to 6.16 give a graphical presentation of the multi-pla-
nar effect and the effect of the simplification.

– The influence of the load ratio is independent of the beam depth.

– Axially loaded I-beam-to-CHS column connections have, for h < 2, a capacity which is
less than twice that of the plate-to-CHS connection.

– If the web is not present at the intersection with the column, the connection strength is
reduced by only 2–12%.

– Column prestressing decreases the connection capacity considerably. The prestress-

ing function based on the maximum column stress, in a simplified form by excluding
the b effect, is given by:

f(n) = 1 – 0.25n2 (2g)0.3 .................................................................6.18

The original and the simplified equation is, for b = 0.45, given in figure 6.17.

– The use of a steel plate floor, as used for offshore decks, does not enhance the con-
nection capacity.

– Concrete filling of the CHS columns increases the stiffness and capacity considerably
but limits the deformation capacity of the connections.

– The connection strength and stiffness can be significantly increased by using a com-
posite floor. However, if the concrete reinforcement is governing for the connection
capacity, the rotation capacity is small if cold formed (low ductility) reinforcement is
used.

70
 1.2

1.0
β = 0.25
M(J) / M(J=0)

0.8 Original equation

0.6 2γ = 15

0.4
2 γ = 50

Simplified equation
0.2

0.0
-0.8 -0.4 0.0 0.4 0.8
J
Figure 6.14 – Multiplanar effect of I-beam-to-CHS column connections loaded by in-plane-bending
moments (b= 0.25)

1.2

1.0
β = 0.45
M(J) / M(J=0)

0.8 Original equation

0.6 2γ = 15

0.4
2γ = 50

Simplified equation
0.2

0.0
-0.8 -0.4 0.0 0.4 0.8
J
Figure 6.15 – Multiplanar effect of I-beam-to-CHS column connections loaded by in-plane-bending
moments (b = 0.45)

71
 1.2

1.0
β = 0.65
M(J) / M(J=0)

0.8 Original equation

0.6 2 γ = 15

0.4
2γ = 50

Simplified equation
0.2

0.0
-0.8 -0.4 0.0 0.4 0.8
J
Figure 6.16 – Multiplanar effect of I-beam-to-CHS column connections loaded by in-plane-bending
moments (b = 0.65)

1.0

β = 0.45
0.8
Original equation
f(n)

0.6
2γ = 15
0.4
2γ = 50

0.2 Simplified equation

0.0
-1.0 -0.8 -0.6 -0.4 -0.2 0.0
n
Figure 6.17 – Influence of column stress on I-beam-to-CHS column connections loaded by in-plane-
bending moments (b = 0.45)

72
6.3.2 I-beam-to-RHS column connections

Initial investigations on I-beam-to-RHS connections have been carried out by Kanatani et

al. (1980) and further flange plate to RHS connections have been investigated by
Wardenier (1982) and Davies and Packer (1982). For I-beam-to-RHS column connections,
Lu (1997) has recently carried out similar investigations as de Winkel (1998) did for I-beam
to-CHS column connections.

For consistency the formulae given here are in principle consistent with those given in the
CIDECT Design Guide No. 3 (Packer et al., 1992) and in Eurocode 3 Annex K (CEN 1992).
However, they have been checked with recent research and extended where required.

Comparison of the design strength formulae for the various failure criteria of flange plate-
to-RHS column connections (Wardenier 1982, Packer et al. 1992) shows that for tp ≤ tc the
plate effective width criterion (equation 6.20) generally is critical compared to punching
shear, column face plastification and column side wall failure (see figure 6.18).

Np* = fp,y tp be ...................................................................................................................6.19

with
10 f c,y • t c
be = • • bp ≤ bp ......................................................................................6.20
bc f p,y • t p
/
tc

The moment capacity for I-beam-to-RHS column connections follows then by multiplying
the flange plate connection strength Np* = Nb,f* for axial loading with the depth (hb - tb, f).

Mb* = Nb, f*(hb - tb, f) ............................................................................................................6.21

Figure 6.18 – Comparison of the FE results based on the 3% of the column width bc deformation
criterion and the column face yield line criterion, the punching shear criterion, the plate effective width
and column side wall criterion (Lu, 1997)
73
However, for highly loaded columns the design strength for column face plastification is
reduced by a function f(n) which may result in this criterion governing. Further, Lu (1997)
showed that the yield line mechanism for column face plastification occurs at deformations
which exceed the deformation limit of 3% of the column width bc. As a
consequence she proposed based on the deformation limit of 3% of the column width bc,
the following modified criterion for column face plastification of I-beam-to-RHS column
connections:
4 fc,y tc2
Mb* = (0.5 + 0.7b) (hb - tb,f) f(n) .................................................................6.22
(1 - 0.9b)0.5

with (not simplified) and based on the maximum column stress:

2)
f(n) = 1 + 1.48 · n · (2g)-0.33 - 0.46 · n1.5 · (2g)(0.33 - 0.11b .....................................................6.23

but f(n) ≤ 1.0

The function f(n) is given in figure 6.19. Thus, the minimum moment resistance calculated
using equations 6.19 to 6.23 governs. The criterion for column face plastification for the
plate-to-RHS connection is not given in CIDECT Design Guide No. 3.

In the case of axial loads it should be noted that the axial load capacity of an I-beam-to-
RHS column connection will only be two times the axial load capacity of one flange if
h > 2√(1-b), see Lu (1997).

In that case for the column face plastification criterion there is no interaction between
one flange and the other (if two separate flanges would be present). Thus for values h < 2
several strength criteria may have to be considered.

From the work of Lu (1997) the following conclusions can be drawn:

– All multiplanar connections with a load ratio J = 0 and b ≤ 0.73 behave like uniplanar
connections. It should be noted that for higher b ratios up to 1.0, a positive geo-
metrical effect is expected, in line with the findings of Yu (1997).

– As shown in figure 6.20 for the investigated width ratios 0.15 < b < 0.75, negative load
ratios (J < 0) decrease the connection capacity considerably, whereas positive load
ratios (J > 0) generally have a small beneficial effect. Simplified, this effect (see fig. 6.20)
is given by f(J) = 1 + 0.4J, but ≤ 1.0. This lower bound can also be used for axially
loaded I-beam- to-RHS connections. Based on the work of Yu (1997), it is expected
that for b ratios close to 1.0 that positive load ratios may also have a negative effect on
the load, therefore the validity range is limited to 0.2 ≤ b ≤ 0.8.

– The use of a steel plate floor does not enhance the connection capacity significantly if
based on a load deformation of 3% bc.

– Concrete filling of the RHS columns increases the stiffness and capacity considerably.

– The connection strength and stiffness can be significantly increased by using a com-
posite floor. However, if the concrete reinforcement is governing for the connection
capacity, the rotation capacity is small if low ductility reinforcement is used (see chapter 9).

74
 1.0

β = 0.50
0.8
Original equation
f(n)

0.6
2γ = 15
0.4 2γ = 25
2γ = 35
0.2
Simplified equation

0.0
-1.0 -0.8 -0.6 -0.4 -0.2 0.0
n

1.0

β = 0.90
0.8
Original equation
f(n)

0.6
2γ = 15
0.4 2γ = 25
2γ = 35
0.2
Simplified equation

0.0
-1.0 -0.8 -0.6 -0.4 -0.2 0.0
n
Figure 6.19 – Effect of column loading on the connection moment resistance

75
 1.2

1.0 2γ= 25

0.8
f(J)

β = 0.30
0.6
β = 0.50
0.4
β = 0.70

0.2

0.0
-0.8 -0.4 0.0 0.4 0.8
J
Figure 6.20 – Multiplanar loading effect for I-beam-to-RHS column moment connections

The previous mentioned strength criteria have been summarised in figure 6.21.

76
Figure 6.21 - Design strength formulae for I-beam-to-RHS column connections

77
6.4 Bolted hollow section beam and column connections

Bolted connections between hollow section members can be made using flange plates,
gusset plates, angles, cleats or cut-outs of open sections. Most bolted connections are
designed as shear connections, tension loaded splices or stiffened moment connections.

6.4.1. CHS beam-to-column connections

The only common types of bolted moment connections between CHS members are shown
in figure 6.22.

(a) (b) (c)

Figure 6.22 – Bolted knee assemblies of CHS or RHS members for portal frames

No detailed evidence is available for the stiffness of these connections in relation to the
plate dimensions and the bolt locations. It is therefore recommended to use a plate thick-
ness such that the connection can be assumed to be rigid. The bolts should preferably be
designed for the moment capacity of the connected hollow section.

6.4.2 RHS beam-to-column connections

For the bolted knee connections, shown in figure 6.22, the same remarks can be made as
for the connections with CHS members. Welding a haunch between the bottom flange of
the RHS beam and the flange plate stiffens the knee assembly of figure 6.22 (c). Another
bolted assembly, which may be designed to transfer moments, is shown in figure 6.23.

flange

shim plate, if needed

Figure 6.23 – Bolted flange plate connection between RHS members

78
6.5 Bolted I-beam-to-hollow section column connections

Bolted connections between I-beams and hollow section columns can be distinguished in
connections with continuous beams as shown in figure 6.24, through plate connections,
shown in figure 6.25, and connections with continuous columns, shown in figure 6.26.

Figure 6.24 – Bolted continuous beam-to-column connections (Packer and Henderson, 1997)

The strength and stiffness of bolted connections with a continuous beam is directly depend-
ent on the thickness of the cap plates, the reinforcement of the beam and the bolts. Since
no evidence is available for the stiffness, it is recommended to design these connections
as rigid moment connections with relatively thick cap plates.

Figure 6.25 – Bolted through-plate moment connections (Packer and Henderson, 1997)

79
The through-plate connections of figure 6.25 allow a direct load transfer from beam to
beam or to a column whereas the shear is transferred by shear tabs or angles welded to
the column web. Here the flexibility for the beam connection depends mainly on the bolts
loaded in shear (prestressed or not); for the top column connection with an interrupted
column it is similar to that of the connections in figure 6.24.

The bolted connection in figure 6.26(a) is in principle a welded beam-to-column connec-

tion, as discussed in section 6.3.1, with a bolted splice. This type of connection is very
common and easy to handle on site.

flange

shim plate, if needed

(a) (b)
Figure 6.26 - Bolted beam-to-column connection with a continuous column

6.5.1 I-beam-to-CHS column connections

Most of the bolted moment connections between I-beams and CHS columns are stiffened
with plates and can be designed as rigid connections. These are further dealt with in
chapter 8.

6.5.2 I-beam-to-RHS column connections

Besides the bolted I-beam-to-RHS column connections with extended plates shown in
figures 6.23 and 6.26(b), nowadays it is also possible to connect directly to the face of the
RHS column. In this case single sided (also called blind) bolting is used. The systems
currently used, i.e. the Flowdrill system, the HolloBolt and the Huck Ultra-Twist Bolt
system, are described in chapter 3. Figure 6.27 shows two examples for moment connec-
tions: (a) with an extended end plate, and (b) with a flush end plate.

80
 tp bp
g
e1

e2
p1
p2
PH hp
p3

tc b
PV
(a)

tp bp
g
e1

p1
p2
PH hp
p3

tc b
PV
(b)

Figure 6.27 - Blind bolted I-beam-to RHS column connections

As discussed in chapter 3, these systems can be treated as normal bolted connections

provided the limitations for bolt diameter in relation to the RHS column thickness are taken
into account. However, for moment connections not only the deformation of the end plate
has to be considered but also the flexible face of the RHS column.

Thus, the following criteria have to be considered (Yeomans 1996, 1996a):

- bolts (tension and shear and bolt bearing for plate and column face);
- stripping of the bolt threads;
- column face - punching shear of the bolt through the column face;
- column face plastification (yield line pattern);
- column side wall crippling;
- plate plastification (yield line pattern).

81
For the bolt design the following well known criteria can be used:
– bolt shear capacity;
– bolt bearing capacity;
– bolt tensile capacity;
– combination shear and tension.

Stripping of the threads of the bolts has to be checked for the bolts in flowdrilled
connections. The bolt thread strip capacity is:

Fts = 0.6 fc,y p db (tc + 8 mm) .................................................................................................6.24

For the punching shear criterion the diameter to be considered depends on the type of
system being used. For example for punching shear of the flowdrilled extrusion from the
RHS:

Fps = 0.6 fc,y p tc (db + tc ) .................................................................................................6.25

For a hollowbolt connection the punching shear capacity is just given by:

Fps = 0.6 fc,y p tc db .................................................................................................6.26

As shown in figure 6.28, the column face plastification criterion depends on the plate
dimension in relation to the width of the RHS column. For small plate widths and stiff end
plates the compression area will be pushed in and the tension area of the connection will
be pulled out giving a yield line pattern in the column face (case a). However, for moment
loaded connections it is recommended to have the plate width the same as the column
width, which increases the stiffness and the moment capacity (case b). In this case a yield
line pattern will be formed in the tensile area only if the crippling strength of the column
walls is not governing. The column face plastification criterion for a bolt pattern with four
bolts in tension given by Yeomans (1994, 1998) is:

Npl = fc,y tc2 [2(hb - db)/b’ + 4(1 - c/b’)0.5 ]/(1-c/b’) f(n) .................................................6.27

where b’ = bc - tc and c = g - db

f(n) = 1 + n ≤ 1.0 ..................................................................................................6.28

In those cases where the plate thickness is smaller, the plate has to be checked in a
similar way as for beam-to-column connections of I-beams, i.e. considering the model
shown in figure 6.29. Here the plate plastification (Zoetemeyer 1974, Eurocode 3-Annex J)
has to be checked for complete end plate yielding (case c1) and end plate yielding with
bolt failure (case c2) similar to column connections of I-beams. However, depending on
the stiffness and thus the deformation of the column face the prying force action may be
different.

82
Figure 6.28 – Column face plastification yield line patterns

83
 A -– A
A A
A

c1: complete end c2: end plate yielding c3: bolt failure
plate yielding + bolt failure
case c

Figure 6.29 - Plate plastification models

The wall bearing or crippling capacity for columns with bc/tc < 35 can be given by:

Nc,w = 2 fc,y tc bw ...............................................................................................................6.29

The bearing width bw can be taken as:

bw = (tb,f + 2tp + 5tc) .............................................................................................................6.30

Apart from the bolted part the welds connecting the end plate and the I-beam have to be
checked. It should also be noted that for one sided connections the shear capacity of the
column has to be checked.

The moment capacity of the connection follows by multiplying the minimum governing
axial load capacity by the beam depth (hb - tb,f).

At present no information in formulae or graphs is available about the stiffness, thus a real
semi-rigid analysis is not yet possible for these connections.

6.6 Examples

In the design of semi-rigid connections the following procedures can be followed:

1. Assume rigid connections and after determination of the member sizes check if the
connection stiffness meets the minimum stiffness requirement given in figure 4.7.

a. If not, the joint parameters and thus the sections should be changed in such
a way that the stiffness requirement is met, or

b. the actual stiffness has to be used in the design calculations and it has to be
checked whether the structure still meets the strength and stiffness requirements.

2. Assume pin-ended connections and after determination of the member sizes check if
the connection stiffness does not exceed the maximum stiffness requirement given in
figure 4.7 for a pin-ended condition.

84
 a. If the requirement is not met, the joint parameters and thus the sections should
be changed in such a way that the stiffness requirement is met, or

b. the actual stiffness has to be used in the design calculations and it has to be
checked whether the strength and stiffness requirements are met.

3. Design the frame based on a rigid plastic frame analysis and check if the stiffness of
the connections and the rotation capacity allow the assumed redistribution of moments.

4. If the designer has knowledge about the connection parameters it is also possible
to determine the associated connection stiffness and use it in the design; after deter-
mination of the final member sizes it should be checked that the actual connection
stiffness does not deviate too much from the assumed stiffness. If so, the design is
O.K., otherwise the analyses should be done again with the actual connection stiffness.

Method 1 is especially appropriate for connections with a large stiffness e.g. with a low 2g
ratio, a large b ratio and/or a low h ratio.

Method 2 is more appropriate for connections with a low stiffness, e.g. with a high 2g ratio,
a low b ratio and/or a low h ratio.

Method 3 is a very simple approach for those cases where the stiffness of the connections
is less important, e.g. braced frames.

Example 1: CHS beams and columns

Figure 6.30 gives a braced frame of CHS members for which the connections between the
circular hollow sections have been assumed to be rigid.

Detail A
4000
4000

A
4000

6000

Figure 6.30 – Frame of CHS beams and columns

85
The steel grade is S355 with a yield stress fy = 355 N/mm2.

Assume that based on the frame analysis with rigid connections the following sections
have been selected:

columns 298.5 x 10 : beams 298.5 x 6.3

(Note: These sections are not available at every tube supplier.)

Check if the stiffness is sufficient to assume a rigid connection.

Eurocode 3 (CEN1992) gives the following requirement for braced frames (see figure 4.7):
Sj,ini ≥ 8EIb/Lb

For the beam 298.5 x 6.3 the following properties apply:

lb = 6175 x 104 mm4, Wel = 414x 103 mm3, E = 2.1x 105 N/mm2, Lb = 6000 mm

Hence the beam stiffness is:

EIb/Lb = (2.1 x 105 ) (6175 x 104)/6000 = 2161 x 106 Nmm/rad = 2161 kNm/rad

And the required stiffness for braced frames is: 8 x 2161 = 17288 kNm/rad

The stiffness for connections between CHS members is given by equation 6.2; however,
this equation is only valid for 0.3 ≤ b ≤ 0.8.

If, however, the stiffness would be sufficient assuming b = 0.8, then it will also be sufficient
for b = 1.0 because the stiffness increases with b.

3 4
3
dc 1
Cb,ip = 1.3E b(2.25+g/125) g-1.44
2 sin (b+0.4)u

= (1.3) (2.1 x 105) (3.32 x 106) (0.589) (0.02) x 1=10.89 x 109 Nmm/rad
= 10890 kNm/rad < 17288 kNm/rad

Thus the available stiffness for b = 0.8 is not sufficient to assume a rigid connection.

For example, it can be checked now if relevant test evidence is available.

Van der Vegte (1995) did numerical calculations for T-connections with b = 1.0 and loaded
by bending in-plane, however the column diameter was 406.4 mm. Thus, the stiffness
given by van der Vegte should be corrected for the influence of the column diameter, or
the influence of the b parameter between 0.8 and 1.0 has to be estimated.

In (van der Vegte, 1995) the stiffness can only be determined from the figures with test
results. To avoid all kinds of recalculations to account for the different dimensions it is
easier to determine the influence of the parameter b between 0.8 and 1.0.

From the results it can be concluded that the stiffness for b = 1.0 is about 60% higher than
for b = 0.8. Equation 6.2, which is graphically shown in Wardenier et al. (1991) and here in

86
figure 6.31, gives for a conservative linear extrapolation from b = 0.8 to b = 1.0 an increase
of 50 to 60%. Thus, an increase of 60% seems to be acceptable.

This results in a connection stiffness of: 1.6 x 10890 = 17424 kNm/rad.

70
dc/tc
60

15
50
4
x 10

40
3
Cb,ip

c E d

20
30
25
20 30
35
40
10 45
50

0
0.0 0.2 0.4 0.6 0.8 1.0
!
Figure 6.31 - Connection stiffness for T-connections between CHS members

Because 17424 > 17288, the connection may be assumed to be rigid for braced frames.

Note 1: Instead of doing all these exercises it could also have been checked with equa-
tion 4.9 if the initially calculated value of Cip = 10890 kNm/rad (for b = 0.8) would not result
in a frame capacity which is more than 5% lower than intended. The limits given would
result in 5793 ≤ 10890 ≤ 36782 which means that the stiffness of 10890 kNm/rad would be
acceptable. This also shows that even for large deviations in stiffness the influence on the
frame capacity is small.

Note 2: If the frame in this example had been an unbraced frame the required stiffness
would have been:

Sj,ini ≥ 25EIb/Lb = 25 x 2161 = 54025 kNm/rad

Selecting sections with a 20% larger thickness and a 20% smaller diameter has, accord-
ing to equation 6.2, the following effect on the stiffness:

3 4
-1.44
0.8
(0.8)3 = 0.92
1.2

87
The favourable effect of reducing the g ratio is compensated by the reduction in diameter
and the resulting stiffness is nearly the same.

Keeping the member diameters the same and increasing the column thickness has a con-
siderable effect on the connection stiffness. However, this would result in more material
costs, thus from an economical point of view it is better here to adopt the previously
determined stiffness in the frame analysis.

Check of the connection capacity.

The formulae for the connection capacity are given in figure 6.2. It is also possible to use
the design graph of CIDECT Design Guide No. 1 (Wardenier et al., 1991) and here given in
figure 6.32.

Figure 6.32 – Design graph for CHS connections loaded by in-plane bending moments

From the design graph:

dc
for = 29.85: C b,ip = 0.6
tc

fc,y · tc
for = 1.6: M*b,ip = 0.6 · 1.6 · f(n’) · Mb,pl = 0.96f(n’) · Mb,pl
fb,y · tb

Suppose the compression stress in the column is 0.6 fc,y then n’ = - 0.6 and with
f(n’) = 1 + 0.3n’ - 0.3(n’)2 = 0.71

M*b,ip = 0.68 · Mb,pl

88
Thus for braced frames this connection is a rigid partial strength connection and for
unbraced frames it is a semi-rigid partial strength connection.

Example 2: RHS beams and columns

Figure 6.33 shows an X-connection of RHS beam and column members. For these con-
nections no formulae are available for the determination of the stiffness. However, indica-
tions can be obtained from literature, e.g. Yu (1997). Compared to connections of CHS
sections here the parameter h = hb/bc also has to be considered because the sections can
be rectangular instead of square.

200 x 200 x 8
column 200 x 200 x 8

beams 200 x 120 x 6.3

steel grade fy = 355 N/mm2

200 x 120 x 6.3

Figure 6.33 – Connection of RHS members

Check if the connection can be assumed to be pin ended.

Eurocode 3 (CEN 1992) states that the connection should be assumed to be pin-ended if
the following requirement is satisfied (see figure 4.7):

Sj, ini < 0.5Elb/Lb

For the beam 200 x 120 x 6.3 the following properties apply:

Ib = 2065 x 104 mm4, Wb,el = 207 x 103 mm3 , Wb,pl = 253 x 103 mm3

E = 2.1 x 105 N/mm2, Lb = 4000 mm

Hence 0.5EIb/Lb = 0.5(2.1 x 105) (2065 x 104)/4000 = 542 x 106 Nmm/rad

= 542 kNm/rad

For the connection stiffness the following connection parameters are important:

b = 120/200 = 0.6 ; 2g = 200/8 = 25 ; h = 200/200 = 1.0

89
In Yu (1997) a graphical presentation of test results (see figure 6.34) is given for this type
of connection for:

b = 0.6 ; 2g = 24 ; h = 2b = 1.2

h = 2.0 • b = 1.2
h = 1.0 • b ^
= 0.6
h = 0.5 b =^0.3
• ^

Figure 6.34 – Numerical results of Yu (1997) for X-connections loaded by in-plane bending moments

The parameters are nearly the same as for the connection being considered, only the
dimensions in the tests were different, i.e. bc = 150 mm instead of 200 mm. Thus, the
influence of bc has to be incorporated in the results of Yu and the effect of h should be
included by interpolation between h = 0.6 and h = 1.2.

As shown in the figure the moment rotation curve is strongly bi-linear. The initial stiffness
fi is given by:

fi = 2di/hb

For h = 1.67b = 1.0 and for Mb,ip = 10 · fc,y · tc2 · bc : di = 2.8 mm

90
thus, fi = 2 x 2.8/150 = 0.037 for

Mb,ip = 10 x 355 x (6.25)2 x 150 = 20.8 x 106 Nmm = 20.8 kNm

Cb,ip = Mb,ip/fI = 562 kNm/rad

The local indentation is a function of bc4 ; for the rotation the indentation is divided by the
beam depth, which is related to bc, thus the rotation is a function of bc3.

Consequently the initial stiffness is given by:

3

Cb,ip = 562 3 4
200
150
= 1332 kNm/rad > 542 kNm/rad

Thus, it can be concluded that according to Eurocode 3 the connections (considering the
3% bc limit) cannot be assumed to be pin-ended.

Considering the secant stiffness at the moment the capacity of the connection (3% bc) is
reached, gives:

di = 6.5 mm instead of 2.8 mm for h = 1.67b = 1.0 and for Mb,ip =10 fc,y tc2 bc

consequently the rotational stiffness drops to:

2.8
x 1332 = 573 kN/rad > 542 kNm/rad
6.5
This is marginally higher than the limit, and based on stiffness the connection could be
classified as semi-rigid.

Note: However, if the connection has sufficient rotation capacity, there is no

problem to assume a pin-ended connection.

Check of the connection capacity

The formulae for the connection capacity are given in figure 6.5.

4
200

3
1 2 200
M*b,ip = 355 x 8 x 200
2 + + f(n) = 28 x 106 f(n) Nmm
200 (1 - 0.6)0.5 (1 - 0.6)
2
200

= 28 x f(n) kNm

The beam capacity is:

Mb,pl = (253 x 103) (355 x 10-6) = 89.8 kNm >> 28 x f(n) kNm

(Note: both sections can be classified as class 1 sections)

Thus, this connection should be classified as a semi-rigid partial strength connection.

91
Example 3: I-beams and CHS columns

Figure 6.35 gives a frame with I-beams at both sides welded to CHS columns.

Columns: 273 x 6 beams: IPE 360

For the IPE 360 beam the following properties apply:

Ib = 16270 x 104 mm4 ; Wb,el = 904 x 103 mm3 ; Wb,pl = 1020 x 103 mm3

E = 2.1 x 105 N/mm2 ; Lb = 6000 mm

The steel grade of the beams and columns is S235 with a specified minimum yield stress
fy = 235 N/mm2.

The I-beam has originally been designed for a uniform distributed loading assuming
pin-ended connections.

Check whether based on a rigid plastic analysis the loading on the IPE 360 beams
can be increased.

!"273 x 6 !"273 x 6 !"273 x 6

IPE 360 IPE 360 IPE 360 IPE 360

6000 6000

!"273 x 6

IPE 360

Figure 6.35 – Welded connection between the IPE 360 and the CHS 273 x 6

For the connections with the columns 273 x 6 the connection parameters are:

b = 170/273 = 0.62 ; 2g = 273/6 = 45.5

92
Check of the connection capacity

The connection moment capacity is given in figure 6.11.

5 360
M*b,ip = 360 (1 + 0.25 ) x 235 x 62 x f(n’) = 40.675 x 106 f(n’) Nmm
1 -0.81b 273
= 40.7 f(n’) kNm

Suppose that n’ = -0.7, then according to figure 6.11 : f(n’) = 1 + 0.3 n’- 0.3n’2 = 0.64

(N.B.: according to equation 6.18 and assuming that n ~

– n’ :
f(n) = 1 - 0.25n2 x 45.50.3 = 1 - 0.79 n2 = 0.61)
Thus M*b,ip = 40.7 x 0.61 = 26 kNm.

Check punching shear.

fb tb,f ≤ 1.16 fc,y · tc or fb ≤ 1.16 (6/12.7) · fc,y

hence:
fb ≤ 0.55 fc,y and fb ≤ 129 N/mm2.
M*b,ip = Wb,el x fb = (904 x 103) x 129 = 116.4 x 106 Nmm = 116.4kNm > 26 kNm.

Thus punching shear is not governing.

The plastic moment capacity of the IPE 360 is:

Wb,pl · fb,y = (1020 x 103) x (0.235) = 240 x 103 kNmm = 240 kNm.

This means that the total capacity of the beam and the connection is:

M*b,ip + Wb,pl fb,y = 26 + 240 = 266 kNm,

and the connection capacity gives in this case an increase of only 10.6% in the total ca-
pacity provided the requirements regarding stiffness and/or rotation capacity are satisfied.

Check if the connection stiffness is sufficient to reach the connection moment

capacity before the beam reaches at the centre the rotation capacity.

The connection stiffness is given by equations 6.13 and 6.16:

Cb,ip = 6.8 E tc b (2g)-1.3 (hb-tb)2/2

Cb,ip = 6.8 (2.1 x 105) (6) (0.62) (45.5)-1.3 (360 - 12.7)2/2 = 2240 x 106 Nmm/rad
= 2240 kNm/rad

The limit for assuming a pin-ended condition is (see figure 4.7):

Sj,ini < 0.5EIb/Lb

0.5EIb/Lb = 0.5 (2.1 x 105) (16.270x104) /6000 = 2847 x 106 Nmm/rad
= 2847 kNm/rad
Cb,ip = 2240 < 2847, thus the original assumption of a pin-ended connection is correct.

93
With this low connection stiffness the plastic moment capacity at the centre of the beam
will be reached first; thus it should be checked whether the connection capacity is reached
before the beam reaches the rotation capacity.

Suppose the beam has a rotation capacity R = 3, which means that according to the def-
inition of R the beam can rotate over (R + 1) fpl = 4fpl.

If it is assumed that in the plastic hinge, when the plastic moment is reached,
yielding takes place in the outer beam fibres at both sides of the hinge over a distance
equal to the beam depth, thus over a total distance of 2hb, then the elongation DL in the
outer fibres is equal to 2hb · ey.

The rotation in the plastic hinge of the beam is then:

2hb · ey fb,y
f = 4 DL = 4 = 16 = 18 x 10-3 rad
0.5hb 0.5hb E

Assuming a rigid plastic situation, thus neglecting the elastic part, the connection will
rotate over:
0.5f = 9 x 10-3 rad.

For this rotation the connection moment resistance will be:

0.5f · Cb,ip = (9 x 10-3) 2240 = 20.2 kNm

This is smaller than the connection moment capacity M*b,ip = 26 kNm at the ultimate limit
state and the actual increase in capacity by including the connection capacity is 20.2 kNm
resulting in a total capacity of the connection and the beam:

20.2 + 240 = 260.2 kNm

Note: In this example the column rotation has been neglected.

Example 4: Bolted I-beam-to-RHS column connection

A flowdrilled connection is considered for various I-beams and RHS columns.

For these connections the stiffness is not only influenced by the stiffness of the column
face but also by the geometry of the end plate. The simplest way is to design the connec-
tions to be pin-ended, which can for example be achieved by using partial depth and rela-
tively thin end plates for the beams. Otherwise the stiffness has to be obtained from the
tests in literature or numerical calculations have to be carried out. As an example, some
tests carried out by France et al. (1999, 1999a) are shown in figure 6.36.

The columns were 200 x 200 with the thickness varying between 6.3 and 12.5 mm. The
steel grade used was S275, however the actual column yield stresses varied from 300 to
340 N/mm2.

94
 10 160 160 160
30 100 30 30 100 30 30 100 30

20 20 20
60 60 60
80 180 80 280 80
80 80 340
80
254 x 146 x 31 UB
356 x 171 x 45 UB

(a) 457 x 152 x 52 UB

10 160 (160)* 160

30 100 30 (30)* (100)* (30)* 30 100 30

60 60 60
100 250 100 100
350
100 100 450
100
254 x 146 x 31 UB
356 x 171 x 45 UB
457 x 152 x 52 UB
(b)
Figure 6.36 – I-beam-to-RHS column connection with partial depth and flush endplates (France et
al.1999, 1999a)

Figure 6.37 shows a comparison between the moment rotation curves of a flush and a par-
tial depth endplate for a connection between a 457 x 152 x 52 UB section and a 200 x 200
x 8 RHS column. Also, the pin-end requirement for a span of 7.5 m is indicated. The flush
end plate connection is classified as semi-rigid and the partial depth end plate connection
as pin-ended.

80
Semi-rigid

60
Moment (kN m)

.
Pinned

40

20 Test 1: Partial depth end-plate

Test 2: flush end-plate
EC3 classification
0
0 10 20 30 40 50 60 70
Rotation (milli-rads)
Figure 6.37 – Moment rotation curves of a flush and a partial depth endplate connection

95
From this investigation it can also be concluded that if the flush endplate thickness is
about 1.5 times the thickness of the RHS section it only marginally contributes to the
deformation of the connection; i.e. the deformation of the RHS face is most important.

The influence of the RHS thickness is shown in figure 6.38. Thus, in a similar manner to
example 2, use can be made of available test data for design.

140

120

100
Moment (kN m)

. 80 200 x 200 x 12
200 x 200 x 8
60 200 x 200 x 6

40

20

0
0 10 20 30 40 50 60 70
Rotation (milli-rads)

Figure 6.38 – Moment-rotation curves of flush endplate connections between a 356 UB and RHS
columns with varying thickness

96
7 Special requirements for seismic loading

The seismic load varies with the energy dissipation (or absorption) capacity of the struc-
ture. This fact is explained by simple relationships between the response shear load V
(response acceleration multiplied by mass) and response displacement d when a structure
is subjected to an impulsive load at its base as shown in figure 7.1. If the structure behaves
elastically, the structure sustains the shear load Vel and the displacement del. If, however,
the structure yields and achieves the ultimate shear capacity at Vu, the structure sustains
displacement du. The energies dissipated by the two structures, namely the areas under
the two load-displacement curves, are roughly the same.

Vel
Response shear force

Vu

δ el δu
Response displacement

Figure 7.1 – Load displacement relationships of elastic and inelastic structure under impulsive loading

The above fact makes seismic design different from ordinary ultimate limit state design
which considers other loads like gravity loads or wind force. If a wind load exceeds the
capacity Vu, the structure collapses. But this is not the case with seismic load. Instead,
seismic design requires that the structure should not collapse even if the maximum
displacement reaches du.

Thus, the existing seismic codes specify design earthquake loads as a function of the
energy dissipation capacity of structures. Specifically, the behaviour factor q in Eurocode
8 (CEN 1994), the reduction factor R in Uniform Building Code (ICBO1997) and the struc-
tural characteristics factor Ds in the Japanese Building Code (BCJ 1997) similarly play the
role of reducing the elastic response spectrum to obtain the design response spectrum,
taking into account the different dissipation characteristics of the various types of struc-
tures. Furthermore, all these codes specify detailing rules for structural elements and
frames to ensure that the structure can dissipate a certain amount of energy correspond-
ing to the reduction factor.

Another point that makes seismic design different from ordinary ultimate limit state design
is the fact that the impulsive load is applied not once but cyclically, although the number

97
of cycles of major impulses is very small, say 2 or 3 cycles. Nevertheless, portions of the
structure are strained well into the strain-hardening region cyclically. This cyclic cold work-
ing quickly deteriorates material toughness, which may cause a non-ductile tensile failure
of structural elements, frequently starting from critical points in welded connections. Non-
ductile failures are undesirable and should be avoided. If the structure is designed to
remain nearly elastic, even under rare intense earthquakes, the material deterioration due
to cold working is avoided. However, keeping an ordinary building structure nearly elastic
to provide for the probability of such a rare occurrence is grossly uneconomical and not
usually attempted unless the structure is isolated from ground shaking by using special
devices.

The following part of this chapter discusses special requirements for the earthquake-
resistant design of beam-to-column connections, additional to those required for ordinary
ultimate limit state design. The descriptions generally follow the Eurocode 8 format,
although design procedures adopted in Eurocode 8 are similar to those adopted in the
other codes like the Uniform Building Code or the Japanese Building Code (the Building
Standard Law of Japan and its subsidiary laws and regulations issued by the Ministry of
Land, Infrastructure and Transportation). Hereafter, these are referred to as the Japanese
Building Code.

7.1 Dissipative and non-dissipative structural behaviours

Eurocode 8 recommends the following two design concepts:

a) dissipative structural behaviour;

b) non-dissipative structural behaviour;

In concept a) the capability of parts of the structure (called dissipative zones) to resist
earthquake loads beyond their elastic region is taken into account. Members and joints in
dissipative zones sustain yielding or local buckling and participate in dissipating input
energy during earthquakes by hysteretic behaviour. When assuming design earthquake
loads, the behaviour factor q is taken greater than 1.0 in accordance with the energy
dissipation capacity of the structure. Values of the behaviour factor are referred to later in
section 7.3.

In concept b) a frame analysis is based on an elastic analysis without taking into account
non-linear material behaviour. When assuming design earthquake loads, the behaviour
factor (namely the reduction factor) q is taken as 1.0. For structures designed using
concept b) the resistance of members and connections can be evaluated in accordance
with the rules presented in Eurocode 3 (CEN 1992), without having to satisfy the ductility
requirements presented in this chapter. The design concept b) may be used for structures
in low seismicity zones, slender trussed structures or isolated structures and will not be
discussed any more in this chapter.

7.2 Materials

The 1994 Northridge and 1995 Kobe earthquakes both took structural engineering profes-
sionals by surprise in that many of the welded connections in modern steel building
frames sustained brittle fractures. These fractures most frequently occurred in regions
around beam bottom flange groove welds. Especially in Northridge, brittle fractures initiat-

98
ed at a very low level of plastic demand and, in some cases, while structures remained
elastic. Low toughness of weld metal produced by electrodes designated as AWS E70T-4
and by high deposition rate welding procedures was found to have played an important
role in inducing brittle fractures (Fisher 1997). The recent trend is for seismic codes to
impose more stringent toughness requirements on steel to be used in dissipative zones.

Eurocode 8 specifies that steel in dissipative zones should conform to EN 10025 (CEN
1993). The minimum required Charpy V Notch (CVN) toughness varies from 27 Joules at
20 ˚C to 40 Joules at – 20 ˚C depending on the grade of steel. The IIW and CIDECT re-
commendations for fatigue design of hollow section connections (IIW 1999, Zhao et al.
2000), recently revised, specify ISO 630 steel as well as hot-finished and cold-formed hol-
low sections designated as EN 10210-1(HF) and EN 10219-1(CF) (CEN 1994a, 1997a),
which both require the same minimum CVN toughness as EN 10025, see the CIDECT
design guide on fabrication, assembly and erection for more details (Dutta et al. 1998). The
FEMA design criteria (2000) are recommending 27 J at 21 ˚C for base metal and 27 J at –
29 ˚C and 54 J at 21 ˚C for weld filler metal as the minimum required CVN values. It is also
worth noting the fact that most of the experimental investigations, which formed the basis
for these recommendations, used materials that showed toughness properties much supe-
rior to the minimum requirement of 27 J at 0 ˚C currently recommended by Japanese build-
ing authorities.

The other important mechanical properties required for dissipative zones are yield strength
and yield/tensile strength ratio. These requirements are to ensure that the structure shows
the same collapse mechanism during earthquakes as that anticipated at the design stage
and will be discussed in later parts of this chapter. Eurocode 8 recommends that the vari-
ation range of yield and tensile strengths of steel used in fabrication should be specified.

7.3 Structural types and behaviour factors

Steel building frames resist horizontal earthquake loads by moment resisting frames or by
braced frames. When moment resisting frames and braced frames are used in combina-
tion, those frames are called dual structures in Eurocode 8 and also in Uniform Building
Code. Moment resisting frames resist horizontal loads by members acting in an essential-
ly flexural manner. In these structures the dissipative zones are mainly located in plastic
hinges near the beam-column connections and energy is dissipated by means of cyclic
bending. Braced frames resist horizontal loads by axial forces in the bracings. In these
frames the dissipative zones are mainly located in tension and/or compression bracings.
Braced frames are in general much stiffer and stronger than moment resisting frames.
However, moment resisting frames show much greater deformation capacity than braced
frames. Eurocode 8 recommends the values of the behaviour factor depending on the type
of structures as shown in figure 7.2, although these values are applicable only when the
detailing rules shown in chapters 7 – 8 are met.

Note that tension bracings have greater energy dissipation capacity than compression
bracings. Values of the q factor in Eurocode 8 are smaller than the R factors in Uniform
Building Code but greater than the 1/Ds factors in the Japanese Building Code. These
differences, however, have no influence on detailing rules described in this chapter.

99
 Moment resisting frames
αu/α1≈1.2

αu/α1≈1.1 αu
q=5
α1

dissipative zones (plastic hinges)

Concentric braced frames

Diagonal bracings

q=4

dissipative zones (tension diagonals only)

V-bracings

q=2

dissipative zones (tension & compression diagonals)

Note: αu /α1 denotes the ratio of the seismic load at which a number of sections,
sufficient for development of overall structural instability, reach their plastic moment
resistance to the seismic load at which the most strained cross section reaches its
plastic resistance. αu /α1 should be limited to 1.6.

Figure 7.2 – Structural types and behaviour factors according to Eurocode 8

100
7.4 Joints in dissipative zones

Eurocode 8 defines the following criteria for seismic design:

1. Structural parts of dissipative zones should have adequate ductility and

resistance until the structure sustains sufficient deformation without failing due
to overall instability.

2. Non-dissipative parts of dissipative structures and the connections of the

dissipative parts to the rest of the structure should have sufficient overstrength
to allow the cyclic yielding of the dissipative parts.

To ensure the sufficient overstrength of connections Eurocode 8 specifies the following

detailing rules:

1. The value of the yield strength of the steel actually used in the fabrication
should not exceed by more than 10% the value fy used in the design.

2. Connections of dissipative parts made by means of complete joint-penetration

(CJP) groove welds (full-penetration butt welds) are considered to satisfy the
overstrength criterion.

3. For fillet-welded or bolted connections the following requirements should be

met. These are also applicable to connections at the ends of bracings.

a) (resistance of the connection according to clause 6 of part 1-1 of

Eurocode 3) ≥ 1.2 x (plastic resistance of the connected part)

b) For bolted shear connections bearing failure should precede bolt

shear failure.

It has been shown after the Northridge and Kobe earthquakes that the detailing rule on
butt-welded joints mentioned above is not always correct. Further detailing rules to fulfil
the sufficient overstrength criterion are discussed in section 7.6.

Eurocode 8 allows connections that are designed to contribute significantly to the energy
dissipation capability inherent in the chosen q-factor. The overstrength conditions need
not apply for these connections. But these connections have to use experimentally-
verified special devices and, therefore, are not suitable to ordinary design office work. The
only exception to these difficult connections is the column web panel, which is described
in section 7.7.

7.5 Strong column-weak beam design

The formation of hinges in columns, as opposed to beams, is undesirable, because this

may result in the formation of a storey mechanism (see figure 7.3), in which damage
concentrates on a few storeys, and relatively few elements participate in energy dissipa-
tion. In addition, such a mechanism may result in local damage to the columns that are
critical gravity load bearing elements.

101
Backing bars were fillet-welded to the beam flanges at points around a quarter of the beam
width in accordance with the AIJ Recommendations (AIJ 1995) in all the conventional and
improved specimens. This was because past test results as well as damage to connec-
tions due to the Kobe earthquake indicated that an existence of fillet welds right in front of
the beam cope was harmful and frequently induced brittle fractures starting from the toes
of the copes.

Weld tabs used were of two kinds, steel tabs and flux tabs. Although steel tabs are com-
monly used everywhere, flux tabs are special to Japanese fabricators. Flux tabs are a kind
of weld dams made of ceramics but are called flux tabs. When steel tabs were used at the
corners of the through diaphragms and beam flanges and left as they were after welding,
cracks frequently started at notch roots formed by the unfused regions between the weld
tabs and beam flanges. These cracks deteriorated significantly the rotation capacity of
beams. Two different welding procedures were specified for fabricating specimens, which
were called multi-passes per each layer and single pass per each layer. Cross sections of
weld beads are compared between these two in figure A.3. Herein, the former welding pro-
cedure is called stringer passes, while the latter one is called weave passes. The weave
passes provide higher weld metal deposition rate and greater heat input, which promotes
grain growth in the HAZ and attendant low notch toughness.

7 4
5 6
3
3 4
2 2
1 1

STRINGER BEADS WEAVE BEADS

Figure A.3 – Cross section of weld beads

Of 86 specimens, 8 specimens were tested in dynamic loading with the rate varying
between 1 Hz and 0.6 Hz while 12 specimens were tested at a temperature of -23 °C. The
remainders of the specimens were subjected to slowly applied cyclic loading at room tem-
peratures. All the specimens were tested in principle under the loading sequences prede-
termined as follows: at least 2 cycles of reversed loading in an elastic region and, subse-
quently, displacement controlled cyclic loading with the amplitude increased as 2upl, 4upl,
6upl, ….. up to failure, where upl signifies the elastic beam rotation at the full plastic
moment Mpl (see sect. A.4). Two cycles of loading were applied at each displacement
increment.

Of 86 specimens, 70 specimens failed by brittle fracture or ductile tensile tear while the
remaining 16 specimens reached the maximum loads owing to local buckling of plate ele-
ments at the beam ends. The tensile failure modes can be subdivided into 2 large groups:
those that fractured due to cracks starting at the toes of beam copes and the others that
fractured from weld metal or heat-affected zones (HAZ) of CJP groove welds at the beam
flange ends. Most fractures belonging to the latter failure mode started at the terminations
(starting and stopping ends) of groove welds between the beam flanges and through
diaphragms.

A.2 Evaluation of rotation capacity of beams

Among several factors that would deteriorate the performance of connections, these
research results revealed that the following 4 factors significantly participated in reducing

203
the rotation capacity of beam-column assemblies. These factors include the conventional
beam cope, steel weld tab, weave beads, and unskilled welding operation. One of the
other important factors is the material factor, whose effects were unable to be evaluated
from these research results. However, it should be noted that the toughness properties of
both base and weld materials used in these tests were higher than the specified minimum
value of 27 J at 0 °C, which were referred to in section 7.2, according to Charpy V impact
test results.

The reduction in the cumulative plastic deformation factor h is denoted by R with a sub-
script showing the cause for the reduction. Each value of R is calculated as the difference
in h between the two opposite cases, namely the cases with conventional and improved
details. The following are the evaluated R values.

If failure was governed by a fracture from the toe of the beam cope,

RCONVENTIONAL COPE = 56.7 - 40.1 > 17 ........................................................ A.1

If failure was governed by a fracture from the welded joints, the following 3 different reduc-
tions were found significant.

RSTEEL TAB = 59.5 - 37.8 > 22 ......................................................................... A.2

The above reduction is due to the difference between flux and steel tabs but is not applic-
able to the field welded connections with the improved beam copes used in these tests.
This is because the improved field-welded connections had straight edges at the termina-
tions of groove welds.

RWEAVE BEADS = 52.8 - 39.2 > 14 ................................................................... A.3

RUNSKILLED AT FLUX TAB = 59.5 - 43.2 > 16 ..................................................... A.4

The last reduction (equation A.4) is applicable only to field-welded connections so far as
the present test results are concerned. This is because the welders who specialise in field-
welding were found to be unaccustomed to using flux tabs.

Of 86 specimens, 24 specimens had improved details in the profile of beam copes and in
weld tabs, and were welded using stringer beads by skilled welders. These specimens
require no reduction in h. These specimens showed an average h value of 67.3. Thus, the
estimate of the cumulative plastic deformation factor he can be given by the smaller of
equations A.5, A.6 and A.7.

If cracks start at the toes of the beam copes,

he = 67 - RCONVENTIONAL COPE ..................................................................... A.5

If cracks start at the CJP groove welds at the beam ends,

he = 67 - RWEAVE BEADS - RSTEEL TAB ............................................................. A.6

If cracks start at the CJP groove welds at the beam ends and for field-welded connections,

he = 67 - RWEAVE BEADS - RUNSKILLED AT FLUX TAB ......................................... A.7

204
For connections with improved details, the R values corresponding to the improvement are
taken to be null in the above 3 equations.

The cumulative plastic deformation factors are calculated for all the specimens included in
the large-scale investigation. The ratios of observed h to predicted h distribute as shown
in figure A.4.

33

mean = 1.01
18 COV = 0.376

14
mean = 0.93
COV = 0.409
9

2 2 2 2
1 1 1 1

0.2 0.6 1 1.4 1.8 2.2 2.6 0.2 0.6 1 1.4 1.8
TEST/PREDICTION TEST/PREDICTION

SHOP-WELDING FIELD- WELDING

Figure A.4 – Histograms showing distributions of test to predicted ratios

The above equations slightly overestimate the deformation capacity of field-welded

connections. Thus, the mean hm and standard deviation sh of the cumulative plastic defor-
mation factor are given as:

for shop-welded connections

hm = 1.01 he and sh = 15.9 ........................................................................... A.8

and for field-welded connections

hm = 0.93 he and sh = 22.3 ........................................................................... A.9

A.3 Flexural strength of beam-column connections

No definite correlation was found in the test results between the maximum moment
carried by the beams and the connection details. The maximum moments were greater
than the fully plastic moments of the beams. As described in section 8.1, the overstrength
factors are plotted against the cumulative plastic deformation factors in figure A.5.

Although scatter is large, the overstrength factor increases linearly with h and can be rep-
resented by the following equation.

Mcf,max
= 0.0025h + 1.18 ......................................................................... A.10
Mpl

205
The data for dynamically loaded specimens are omitted for evaluating the above regres-
sion equation, because material properties under dynamic loads were not reported.

1.7

1.6

1.5

1.4

1.3
Mcf,max
Mpl

1.2

1.1

1 Cyclic loading
Dynamic loading
0.9
Linear regression
0.8
0 20 40 60 80 100 120
h

Figure A.5 – Overstrength factor compared with cumulative plastic deformation factor (AIJ Kinki 1997)

A.4 Definition of cumulative plastic deformation factor

Several parameters have so far been used as the measure representing the performance
of beam-column assemblies. The large-scale investigation used the cumulative plastic
deformation factor as defined below.

An example of hysteresis loops of the flexural moment at the column face, Mcf versus the
rotation of the beam segment between the loading point and column face Ucf (see figure
A.6) is shown in figure A.7.

V Mcf =V L

θcf

Figure A.6 – Definition of beam rotation and moment at column face

206
 η +θ pl
i

Mcf Mcf
Mpl

W i+
θ cf θ cf

θ pl

W i-

ηi-θpl

Figure A.7 – Definition of cumulative plastic deformation factor

The elastic beam rotation Upl at the fully-plastic moment Mpl is defined as the elastic
component of beam rotation at Mcf = Mpl (see figure A.7). Mpl is calculated using the mea-
sured yield stresses of beam materials and measured dimensions of beam sections. The
plastic components of beam rotation at the i-th half cycle, non-dimensionalised by divid-
ing it by Upl, are denoted by hi+ and hi-, in which the + and - symbols distinguish
positive and negative moments (see figure A.7). The cumulative plastic deformation
factor is defined as the sum of hi+ and hi- sustained by the specimen until failure occurs
and is written as:

S ( hi + hi )
+ -
h= .......................................................................................... A.11
i
The alternative definition of the cumulative plastic deformation factor is the sum of plastic
energies dissipated during all the cycles, non-dimensionalized by dividing the energy by
MplUpl. According to the latter definition hi+ and hi- are written as:
+ -
SWi SWi
hi+ = i
and hi- = i
.................................................................. A.12
MplUpl MplUpl

where Wi denotes the energy absorbed at the i-th half cycle (see figure A.7).

The FEMA criteria (2000) are using the inter-storey drift angle as a performance parameter.
The drift angle can be expressed as a function of h and the overstrength factor given by
equation A.10 by following the loading sequences adopted by the large-scale investiga-
tion. Assuming Upl = 0.09, which is equal to the average value of Upl for the specimens
used in these tests, the cumulative plastic deformation factor can be converted to the drift
angle by the following equation:

U = 0.0081h0.46 .............................................................................................. A.13

207
208
 Comité International pour le
Développement et l’Étude de la
Construction Tubulaire

International Committee
for the Development and Study
of Tubular Structures
CIDECT, founded in 1962 as an international association, joins together the research
resources of the principal hollow steel section manufacturers to create a major force in the
research and application of hollow steel sections world-wide.
The CIDECT web site is [Link]

The objectives of CIDECT are:

• to increase the knowledge of hollow steel sections and their potential application by
initiating and participating in appropriate research and studies.
• to establish and maintain contacts and exchanges between producers of hollow steel
sections and the ever increasing number of architects and engineers using hollow steel
sections throughout the world.
• to promote hollow steel section usage wherever this makes good engineering practice
and suitable architecture, in general by disseminating information, organising congres-
ses, etc.
• to co-operate with organisations concerned with specifications, practical design recom-
mendations, regulations or standards at national and international levels.

Technical activities

The technical activities of CIDECT have centred on the following research aspects of hol-
low steel section design:
• Buckling behaviour of empty and concrete filled columns
• Effective buckling lengths of members in trusses
• Fire resistance of concrete filled columns
• Static strength of welded and bolted connections
• Fatigue resistance of welded connections
• Aerodynamic properties
• Bending strength of hollow steel section beams
• Corrosion resistance
• Workshop fabrication, including section bending
• Material properties

The results of CIDECT research form the basis of many national and international design
requirements for hollow steel sections.

209
CIDECT publications

The current situation relating to CIDECT publications reflects the ever increasing empha-
sis on the dissemination of research results.
The list of CIDECT Design Guides, in the series “Construction with Hollow Steel Sections”,
already published, is given below. These design guides are available in English, French,
German and Spanish.
1. Design guide for circular hollow section (CHS) joints under predominantly static loading
(1991)
2. Structural stability of hollow sections (1992, reprinted 1996)
3. Design guide for rectangular hollow section (RHS) joints under predominantly static loa-
ding (1992)
4. Design guide for structural hollow section columns exposed to fire (1995, reprinted 1996)
5. Design guide for concrete filled hollow section columns under static and seismic loading
(1995)
6. Design guide for structural hollow sections in mechanical applications (1995)
7. Design guide for fabrication, assembly and erection of hollow section structures (1998)
8. Design guide for circular and rectangular hollow section welded joints under fatigue
loading (2000)

In addition, as a result of the ever increasing interest in steel hollow sections in internation-
ally acclaimed structures, two books “Tubular Structures in Architecture” (sponsored by
the European Community) and “Hollow Sections in Structural Applications” (published by
Bouwen met Staal) have been published.

Copies of the design guides, the architectural book and research papers may be obtained
from:

The Steel Construction Institute

Silwood Park
Ascot
Berkshire SL5 7QN
England
Tel: +44-(0)1344-23345 Fax: +44-(0)1344-22944
e-mail: publications@[Link] Website: [Link]

“Hollow Sections in Structural Applications” is available from the publisher:

Bouwen met Staal
PO Box 29075
NL-3001 GB Rotterdam
The Netherlands
Tel: +31-10-411 5070 Fax: +31-10-412 1221
e-mail: info@[Link]

210
CIDECT organisation (2004)
• President: B. Lugtenberg – United Kingdom
• Treasurer/secretary: R. Murmann – United Kingdom
• A General Assembly of all members meeting once a year and appointing an Executive
Committee responsible for administration and execution of established policy
• A Technical Commission and a Promotions Commission meeting at least once a year
and directly responsible for the research work and technical promotion work

Present members of CIDECT are:

• Aceralia Transformados, Spain

• Arcelor Tubes, France
• Borusan Muhendislik, Turkey
• Corus Tubes, United Kingdom
• IPSCO Inc., Canada
• Onesteel, Australia
• Rautaruukki Oy, Finland
• Sidenor, Greece
• Smorgon Steel Tube Mills, Australia
• Tata Tubes, India
• Vallourec & Mannesmann Tubes, Germany
• Voest Alpine Krems, Austria

Acknowledgements for photographs:

The authors express their appreciation to the following companies for making available the
photographs used in this design guide:

IPSCO Inc.
Corus Tubes

Care has been taken to ensure that all data and information herein is factual and that
numerical values are accurate. To the best of our knowledge, all information in this book is
accurate at the time of publication.
CIDECT, its members and the authors assume no responsibility for errors or misinterpre-
tations of information contained in this book or in its use.

211
212
213
214
215