STRUCTURAL APPLICATIONS
J. Wardenier, J.A. Packer, X.L. Zhao and G.J. van der Vegte
ISBN 9789072830869
CIDECT, Geneva, Switzerland, 2010
The publisher and authors have made careful efforts to ensure the reliability of the data contained in this
publication, but they assume no liability with respect to the use for any application of the material and
information contained in this publication.
Printed by Bouwen met Staal
Boerhaavelaan 40
2713 HX Zoetermeer, The Netherlands
P.O. Box 190
2700 AD Zoetermeer, The Netherlands
Tel. +31(0)79 353 1277
Fax +31(0)79 353 1278
Email info@bouwenmetstaal.nl
ii
PREFACE
The global construction market requires a worldwide coordination of product, testing, design and execution
standards, so that contracts for delivery of products and for engineering and construction services can be
agreed on a common basis without barriers.
The mission of CIDECT is to combine the research resources of major hollow section manufacturers in order to
create a major force in the research and application of hollow steel sections world wide. This forms the basis of
establishing coordinated and consistent international standards.
For the ease of use of such standards, it is however necessary to reduce their content to generic rules and to
leave more objectoriented detailed rules to accompanying nonconflicting complementary information, that
have the advantage to be more flexible for the adaptation to recent research results and to be useable together
with any international code.
The book by J. Wardenier, J.A. Packer, X.L. Zhao and G.J. van der Vegte "Hollow sections in structural
applications" is such a source, developed in an international consensus of knowledge on the topic. It
incorporates the recently revised design recommendations for hollow sections joints of the International
Institute of Welding, IIW (2009) and CIDECT (2008 and 2009). Both are consistent with each other and are the
basis for the Draft ISO standard for Hollow Section Joints (ISO 14346) and may form the basis for future
maintenance, further harmonisation and further development of Eurocode 3 (EN 199318), AISC (ANSI/AISC
360) and the CISC recommendations.
For the use together with EN 199318 and ANSI/AISC 360, both being based on the previous IIW (1989)
recommendations, the main differences to these rules are highlighted.
The authors are all internationally recognized experts in the field of tubular steel structures, three of them
having been chairmen of the IIWSubcommission XVE on "Tubular Structures" since 1981. This committee is
the preeminent international authority producing design recommendations and standards for onshore tubular
structures.
This book should therefore be an invaluable resource for lecturers, graduate students in structural, architectural
and civil engineering, explaining the important principles in the behaviour of tubular steel structures. It is also
addressed to designers of steel structures who can find in it the special items related to the use of hollow
sections, in particular joints, their failure modes and analytical models as supplements to more general design
codes.
Aachen, Germany, August 2010
Prof. Dr.Ing. Dr.h.c. Gerhard Sedlacek
iii
iv
ACKNOWLEDGEMENTS
This book gives the background to design with structural hollow sections in general and in particular for joints to
hollow sections. For the latter, the recently updated recommendations of the International Institute of Welding
(IIW, 2009) and CIDECT (2008 and 2009) are adopted.
The background to design recommendations with the relevant analytical models is especially important for
students in Structural and Civil Engineering, whereas the design recommendations themselves serve more as
an example. Since the available hours for teaching Steel Structures, and particularly Tubular Structures, vary
from country to country, this book has been written in a modular form. The presentation generally follows
European codes, but the material is readily adapted to other (national) codes.
Since the first edition of this book was used not only by students but also by many designers, this second
edition was needed due to the recent update of the recommendations by IIW and the subsequent revision of
the CIDECT Design Guides Nos. 1 and 3 in 2008 and 2009.
The new IIW (2009) recommendations and the revised CIDECT Design Guides Nos. 1 and 3 (2008 and 2009)
are consistent with each other and are the basis for the Draft ISO standard for Hollow Section Joints (ISO
14346). Although the current Eurocode 3 (EN 199318, 2005) and AISC (2010) recommendations are still
based on the previous IIW (1989) and CIDECT (1991 and 1992) recommendations, it is expected that in the
next revision these will follow the new IIW and CIDECT recommendations presented in this book.
Besides the static design recommendations and background for hollow section joints, information is given for
member design in Chapter 2, composite structures in Chapter 4, and fire resistance in Chapter 5. These
chapters fully comply with the latest versions of the Eurocodes (EN 1993 and EN 1994). Further, fatigue design
of hollow section joints is covered in Chapter 14.
We wish to thank our colleagues from the IIW Subcommission XVE "Tubular Structures" and from the
CIDECT Project Working Group and the CIDECT Technical Commission for their constructive comments during
the preparation of this book.
We are very grateful that Prof. J. Stark and Mr. L. Twilt were willing to check Chapters 4 and 5 respectively on
composite members and fire resistance.
Appreciation is further extended to the authors of CIDECT Design Guides Nos. 1 to 9 and to CIDECT for
making parts of these Design Guides or background information available for this book.
Finally, we wish to thank CIDECT for the initiative to update this book.
Delft, The Netherlands, September 2010
Jaap Wardenier
Jeffrey A. Packer
XiaoLing Zhao
Addie van der Vegte
CONTENTS
1. Introduction 1
1.1 History and developments 1
1.2 Designation 2
1.3 Manufacturing of hollow sections 2
2. Properties of hollow sections 9
2.1 Mechanical properties 9
2.2 Structural hollow section dimensions and dimensional tolerances 10
2.3 Geometric properties 11
2.4 Drag coefficients 14
2.5 Corrosion protection 14
2.6 Use of internal void 15
2.7 Aesthetics 15
3. Applications 29
3.1 Buildings and halls 29
3.2 Bridges 29
3.3 Barriers 29
3.4 Offshore structures 30
3.5 Towers and masts 30
3.6 Special applications 30
4. Composite structures 37
4.1 Introduction 37
4.2 Design methods 37
4.3 Axially loaded columns 37
4.4 Resistance of a section to bending 39
4.5 Resistance of a section to bending and compression 39
4.6 Influence of shear forces 39
4.7 Resistance of a member to bending and compression 39
4.8 Load introduction 41
4.9 Special composite members with hollow sections 41
5. Fire resistance of hollow section columns 49
5.1 Introduction 49
5.2 Fire resistance 50
5.3 Unfilled hollow section columns 52
5.4 Concrete filled hollow section columns 53
5.5 Water filled hollow section columns 55
5.6 Joints 56
6. Design of hollow section trusses 65
6.1 Truss configurations 65
6.2 Joint configurations 65
6.3 Limit states and limitations on materials 66
6.4 General design considerations 67
6.5 Truss analysis 68
7. Behaviour of joints 75
7.1 General introduction 75
7.2 General failure criteria 77
7.3 General failure modes 77
v
7.4 Joint parameters 77
8. Welded joints between circular hollow sections 81
8.1 Introduction 81
8.2 Modes of failure 81
8.3 Analytical models 81
8.4 Experimental and numerical verification 83
8.5 Basic joint strength formulae 83
8.6 Evaluation to design rules 84
8.7 Other types of joints 85
8.8 Design charts 86
8.9 Relation to the previous recommendations of IIW (1989) and CIDECT (1991) 87
8.10 Concluding remarks 87
9. Welded joints between rectangular hollow sections 103
9.1 Introduction 103
9.2 Modes of failure 103
9.3 Analytical models 104
9.4 Experimental and numerical verification 106
9.5 Basic joint strength formulae 106
9.6 Evaluation to design rules 107
9.7 Other types of joints or other load conditions 107
9.8 Design charts 109
9.9 Concluding remarks 109
10. Welded joints between hollow sections and open sections 129
10.1 Introduction 129
10.2 Modes of failure 129
10.3 Analytical models 129
10.4 Experimental verification 131
10.5 Evaluation to design rules 131
10.6 Joints predominantly loaded by bending moments 131
11. Welded overlap joints 141
11.1 Introduction 141
11.2 Modes of failure 141
11.3 Analytical models for RHS overlap joints 141
11.4 Analytical models for CHS overlap joints 143
11.5 Analytical models for overlap joints with an open section chord 143
11.6 Experimental and numerical verification 143
11.7 Joint strength formulae 144
12. Welded I beamtoCHS or RHS column moment joints 151
12.1 Introduction 151
12.2 Modes of failure 151
12.3 Analytical models 151
12.4 Experimental and numerical verification 153
12.5 Basic joint strength formulae 153
12.6 Concluding remarks 154
13. Bolted joints 161
13.1 Flange plate joints 161
13.2 End joints 161
13.3 Gusset plate joints 162
13.4 Splice joints 162
vi
vii
13.5 Beamtocolumn joints 162
13.6 Bracket joints 163
13.7 Bolted subassemblies 163
13.8 Purlin joints 163
13.9 Blind bolting systems 163
13.10 Nailed joints 163
14. Fatigue behaviour of hollow section joints 175
14.1 Definitions 175
14.2 Influencing factors 175
14.3 Loading effects 176
14.4 Fatigue strength 177
14.5 Partial factors 177
14.6 Fatigue capacity of welded joints 177
14.7 Fatigue capacity of bolted joints 179
14.8 Fatigue design 180
15. Design examples 193
15.1 Uniplanar truss of circular hollow sections 193
15.2 Uniplanar truss of square hollow sections 197
15.3 Multiplanar truss (triangular girder) 197
15.4 Multiplanar truss of square hollow sections 199
15.5 Joint check using the joint resistance formulae 199
15.6 Concrete filled column with reinforcement 200
16. References 209
Symbols 221
CIDECT 229
viii
1. INTRODUCTION
Design is an interactive process between the
functional and architectural requirements and the
strength and fabrication aspects. In a good design, all
these aspects have to be considered in a balanced
way. Due to the special features of hollow sections
and their joints, it is here even of more importance
than for steel structures of open sections. The
designer should therefore be aware of the various
aspects of hollow sections.
Many examples in nature show the excellent
properties of the tubular shape with regard to loading
in compression, torsion and bending in all directions,
see Figs. 1.1 and 1.2. These excellent properties are
combined with an attractive shape for architectural
applications (Figs. 1.3 and 1.4). Furthermore, the
closed shape without sharp corners reduces the area
to be protected and extends the corrosion protection
life (Fig. 1.5).
Another aspect which is especially favourable for
circular hollow sections is the lower drag coefficients if
exposed to wind or water forces. The internal void can
be used in various ways, e.g. to increase the bearing
resistance by filling with concrete or to provide fire
protection. In addition, heating or ventilation systems
sometimes make use of the hollow section columns.
Although the manufacturing costs of hollow sections
are higher than those for other sections, leading to
higher unit material cost, economical applications are
achieved in many fields. The application field covers
all areas, e.g. architectural, civil, offshore, mechanical,
chemical, aeronautical, transport, agriculture and
other special fields. Although this book will be mainly
focused on the background to design and application,
in a good design not only does the strength have to be
considered, but also many other aspects, such as
material selection, fabrication including welding and
inspection, protection, erection, in service inspection
and maintenance.
One of the constraints initially hampering the
application of hollow sections was the design of the
joints. However, nowadays design recommendations
exist for all basic types of joints, and further research
evidence is available for many special types of joints.
Based on the research programmes carried out,
CIDECT (Comit International pour le Dveloppement
et l'Etude de la Construction Tubulaire) has published
Design Guides Nos. 1 to 9 for use by designers in
practice. Since these nine Design Guides are all
together too voluminous for educational purposes and
do not give the theoretical background, it was decided
to write this book especially to provide background
information for students and practitioners in Structural
and Civil Engineering.
This book is written in a limit states design format
(also known as LRFD or Load and Resistance Factor
Design in the USA). This means that the effect of the
factored loads (the specified or unfactored loads
multiplied by the appropriate load factors) should not
exceed the factored resistance of the joint or member.
The factored resistance expressions, in general,
already include appropriate material and joint partial
safety factors (
M
) or joint resistance (or capacity)
factors (). This has been done to avoid interpretation
errors, since some international structural steelwork
specifications use
M
values 1,0 as dividers (e.g.
Eurocodes), whereas others use values 1,0 as
multipliers (e.g. in North America and Australia). In
general, the value of 1/
M
is almost equal to .
1.1 HISTORY AND DEVELOPMENTS
The excellent properties of the tubular shape have
been recognised for a long time; i.e. from ancient time,
nice examples are known. An outstanding example of
bridge design is the Firth of Forth Bridge in Scotland
(1890) with a free span of 521 m, shown in Fig. 1.6.
This bridge has been built up from tubular members
made of rolled plates which have been riveted
together, because at that time, other fabrication
methods were not available for these sizes.
In the same century, the first production methods for
seamless and welded circular hollow sections were
developed. In 1886, the Mannesmann brothers
developed the skew roll piercing process
(Schrgwalzverfahren), shown in Fig. 1.7, which made
it possible to roll short thick walled tubulars. This
process, in combination with the pilger process
(Pilgerschrittverfahren, Fig. 1.8), developed some
years later, made it possible to manufacture longer
thinner walled seamless hollow sections.
In the first part of the previous century, an Englishman,
Whitehouse, developed the fire welding of circular
hollow sections. However, the production of welded
circular hollow sections became more important after
the development of the continuous welding process in
1930 by the American, Fretz Moon (Fig. 1.9).
Especially after the Second World War, welding
processes have been perfected, which made it
possible for hollow sections to be easily welded
1
together.
The end cutting required for fitting two circular hollow
sections together was considerably simplified by the
development of a special end preparation machine by
Mller (Fig. 1.10).
For manufacturers who did not have such end cutting
machines, the end preparation of circular hollow
sections remained a handicap.
A way of avoiding the connection problems was the
use of prefabricated connectors, e.g. in 1937
Mengeringhausen developed the Mero system. This
system enabled the fabrication of large space
structures in an industrialized way (Fig. 1.11).
In 1952, the rectangular hollow section was developed
by Stewarts and Lloyds (now Corus Tubes). This
section, with nearly the same properties as the
circular hollow section, enables the connections to be
made by straight end cuttings.
In the fifties, the problems of manufacturing, end
preparation and welding were all solved and from that
point of view the way to a successful story was open.
The remaining problem was the determination of the
strength of unstiffened joints.
The first preliminary design recommendations for
truss connections between circular hollow sections
were given by Jamm in 1951. This study was followed
by several investigations in the USA (Bouwkamp,
1964; Natarajan & Toprac, 1969; Marshall & Toprac,
1974), Japan (Togo, 1967; Natarajan & Toprac,
1968), and Europe (Wanke, 1966; Brodka, 1968;
Wardenier, 1982; Mang & Bucak, 1983; Puthli, 1998;
Dutta, 2002).
Research on joints between rectangular hollow
sections started in Europe in the sixties, followed by
many other experimental and theoretical
investigations. Many of these were sponsored by
CIDECT.
Besides these investigations on the static behaviour,
in the last 25 years much research was carried out on
the fatigue behaviour and other aspects, such as
concrete filling of hollow sections, fire resistance,
corrosion resistance and behaviour under wind
loading.
1.2 DESIGNATION
The preferred designations for structural applications
are:
 Circular hollow sections (CHS)
 Rectangular hollow sections (RHS)
 Square hollow sections (SHS)
In Canada and the USA, it is common to speak about
Hollow Structural Sections (HSS), whereas in Europe
also the term Structural Hollow Sections (SHS) is
used.
1.3 MANUFACTURING OF HOLLOW
SECTIONS
As mentioned, hollow sections can be produced
seamless or welded. Seamless hollow sections are
made in two phases, i.e. the first phase consists of
piercing an ingot and the second step considers the
elongation of this hollow bloom into a finished circular
hollow section. After this process, the tube can go
through a sizing mill to give it the required diameter.
More information about other processes, most of them
based on the same principle, is given by Dutta (2002).
Nowadays, welded hollow sections with a longitudinal
weld are mainly made employing either electrical
resistance welding processes or induction welding
processes, shown in Fig. 1.12. A strip or plate is
formed by rollers into a cylindrical shape and welded
longitudinally. The edges are heated, e.g. by electrical
resistance, then the rollers push the edges together,
resulting in a pressure weld. The weld protrusion on
the outside of the tube is trimmed immediately after
welding.
Rectangular hollow sections are made by deforming
circular hollow sections through forming rollers, as
shown in Fig. 1.13. This forming process can be done
hot or cold, using either seamless or longitudinally
welded circular hollow sections. Although it is
common practice to use longitudinally welded hollow
sections, for the very thick sections, seamless
sections may be used.
Square or rectangular hollow sections are sometimes
made by forming a single strip to the required shape
and closing it by a single weld, preferably in the
middle of a face.
Large circular hollow sections are also made by rolling
plates through a socalled UO press process shown
in Fig. 1.14. After forming the plates to the required
2
shape, the longitudinal weld is made by a submerged
arc welding process.
Another process for large tubulars is to use a
continuous wide strip, which is fed into a forming
machine at an angle to form a spirally formed circular
cylinder, see Fig. 1.15. The edges of the strip are
welded together by a submerged arc welding process
resulting in a socalled spirally welded tube.
More detailed information about the manufacturing
processes and the limitations in sizes can be obtained
from Dutta (2002).
3
Fig. 1.1 Reeds in the wind
Fig. 1.3 Airport Bangkok, Thailand
Fig. 1.2 Bamboo
Fig. 1.4 Ripshorster Bridge, Germany
4
Fig. 1.5 Paint surface for hollow sections vs open
sections
Fig. 1.7 Skew roll piercing process
(Schrgwalzverfahren)
Fig. 1.6 Firth of Forth Bridge, Scotland
Fig. 1.8 Pilger process (Pilgerschrittverfahren)
5
forming rollers
heating
welding rollers
welded CHS
heating
coil
forming rollers
heating
welding rollers
welded CHS
heating
coil
Fig. 1.9 Fretz Moon process
Fig. 1.11 Mero connector
Fig. 1.10 End cutting machine
Pressure rollers
inductor
Welded CHS
Pressure rollers
inductor
Welded CHS
Pressure rollers
inductor
Welded CHS
Fig. 1.12 Induction welding process
6
Fig. 1.13 Manufacturing of rectangular hollow sections
Fig. 1.14 Forming of large CHS
Fig. 1.15 Spirally welded CHS
7
8
2. PROPERTIES OF HOLLOW
SECTIONS
2.1 MECHANICAL PROPERTIES
Hollow sections are made of similar steel as used for
other steel sections, thus in principle there is no
difference in mechanical properties.
Tables 2.1a and 2.2a show, as an example, the
mechanical properties according to the European
standard EN 102101 (2006) for hot finished structural
hollow sections of nonalloy and fine grain structural
steels. The cold formed sections are given in EN
102191 (2006): Cold formed welded structural hollow
sections of nonalloy and fine grain structural steels
(see Tables 2.1b and 2.2b). As shown, the
requirements of EN 102101 and EN 102191 are
almost identical.
Hollow sections can also be produced in special
steels, e.g. high strength steel with yield strengths up
to 690 N/mm
2
or higher, weathering steels and steel
with improved or special chemical compositions, etc.
Generally, the design of members is based on the
yield strength. In this chapter the recommended
M0
and
M1
factors of 1,0 are adopted for the design yield
strength f
yd
.
In statically indeterminate structures, sufficient
deformation capacity or rotation capacity is required
for redistribution of loads. In this case, yielding of
members or yielding in the joints may provide the
required rotation capacity. A tensile member made of
ductile steel can be brittle if a particular cross section
is weakened, e.g. by holes, in such a way that this
cross section fails before the whole member yields. It
is therefore required that yielding occurs first. This
shows that the yieldtoultimate tensile strength ratio is
also important, especially for structures with very
nonuniform stress distributions, which is a situation
that occurs in tubular joints. Some codes, such as
Eurocode 3 (EN 199311, 2005), specify the following
requirement for the minimum ratios:
1 , 1
f
f
yd
u
> (2.1a)
The IIW (2009) recommendations and many offshore
codes require a higher ratio between f
u
and f
yd
:
8 , 0
f
f
or 25 , 1
f
f
u
yd
yd
u
s > (2.1b)
This is only one aspect for ductility. In the case of
impact loading, the steel and members should also
behave in a ductile manner. Hence, Tables 2.1a and
2.2a also give requirements based on the standard
Charpy test to ensure adequate notch toughness.
Nowadays, more refined characterisation methods
exist to describe the ductility of cracked bodies, e.g.
the CTOD (Crack Tip Opening Displacement) method.
These characterisation methods are generally used
for pressure vessels, transport line pipes and offshore
applications, which are beyond the scope of this book.
Another characterisation is sometimes required for
thick walled sections which are loaded in the
thickness direction. In this case, the strength and
ductility in the thickness direction should be sufficient
to avoid cracking, called lamellar tearing, see Fig. 2.1.
This type of cracking is caused by non metallic
manganesesulphide inclusions. Thus, if the sulphur
content is very low or the sulphur is joined with other
elements such as calcium (Ca), such a failure can be
avoided. Indirectly this is obtained by requiring a
certain reduction of area R
AZ
in the tensile test. For
example, R
AZ
= 35 means that in the tensile test the
cross sectional area at failure has been reduced by
35% compared to the original cross sectional area.
In most structural steel specifications the minimum
required yield strength, ultimate tensile strength,
elongation and the Charpy Vnotch values are
specified. Design standards or specifications give
further limitations for the f
u
/f
y
ratio, whereas depending
on the application, more restrictive requirements may
be given related to CTOD values or the properties in
the thickness direction (Z quality).
Another aspect is the effect of cold forming on the
mechanical properties of the parent steel. In the case
of cold forming of hollow sections, the yield strength
and to a lesser extent the ultimate tensile strength are
increased, especially in the corners, as shown in Fig.
2.2. Further, the yieldtoultimate tensile ratio is
increased and the elongation slightly decreased.
If the standards, e.g. EN 102101 and EN 102191,
specify the properties at a particular cross section
location based on the finished product, these
properties have been already partly taken into
account. Thus, this generally applies in Europe.
9
However, some standards outside Europe specify the
material properties of the parent material. In this case,
the increased yield strength can be taken into account
for design. A small corner radius produces a small
cold formed area with a large cold forming effect and
consequently a large increase in yield strength, while
a large corner radius does just the opposite.
According to research work of Lind & Shroff (1971),
the product of area and increase in yield strength can
approximately be taken as constant. Lind & Shroff
assumed that in every corner of 90 the yield strength
of the parent material f
yb
is increased over a length of
7t to the ultimate tensile strength of the parent
material f
u
. The total increase over the section 4(7t)t(f
u

f
yb
) can be averaged over the section, resulting in a
design yield strength f
ya
, as shown in Fig. 2.2.
It is noted that the cold formed sections should satisfy
the requirements for minimum inside corner radius to
guarantee sufficient ductility, see Table 2.3 for fully
aluminum killed steel (steel with limited Si content).
Part 10 of Eurocode 3 (EN 1993110, 2005) specifies
the material selection. Here, a permissible thickness
can be determined based on a reference temperature,
the steel grade and quality and the stress level. The
reference temperature covers, besides the air
temperature, also cold forming effects, strain rate, etc.
However, the current rules cannot be adopted to cold
formed hollow sections because the determination of
the effect of cold forming for cold formed hollow
sections is not yet clearly specified. Based on the data
obtained by Soininen (1996), Kosteski et al. (2003),
Bjrk (2005), Khn (2005), Puthli & Herion (2005) and
Sedlacek et al. (2008), presently a proposal is being
worked out for an amendment of EN 1993110. In
this proposal of CEN/TC 250/SC 3N 1729 (2010), it is
recommended that for cold formed hollow sections
according to EN 10219, the procedure for hot formed
material can be used provided that for the cold
forming effects the reference temperature is reduced
by AT
cf
. For CHS, AT
cf
varies from 0 C to 20 C
depending on the thickness and the d/t ratio. For RHS
with steel qualities according to EN 10219, AT
cf
varies
from 35 C to 45 C depending on the thickness and
the ratio between the inside corner radius and the
thickness. For cold formed hollow sections with
Charpy impact strengths significantly exceeding the
requirements of EN 10219, a lower value of AT
cf
is
allowed.
2.2 STRUCTURAL HOLLOW SECTION
DIMENSIONS AND DIMENSIONAL
TOLERANCES
The dimensions and sectional properties of structural
hollow sections have been standardised in EN (EN
102102, 2006; EN 102192, 2006) and ISO standards
(ISO 65714, 2000; ISO 4019, 2001) for hot finished
and cold formed structural hollow sections
respectively.
The two applicable standards in Europe are EN
102102 (2006) "Hot finished structural hollow
sections of nonalloy and fine grain steels Part 2:
Tolerances, dimensions and sectional properties" and
EN 102192 (2006) "Cold formed welded structural
hollow sections of nonalloy and fine grain steels
Part 2: Tolerances, dimensions and sectional
properties". However, the majority of manufacturers of
structural hollow sections do not produce all the sizes
shown in these standards. It should be further noted
that other sizes, not included in these standards, may
be produced by some manufacturers.
The majority of the tolerances given in EN 102192
are the same as those in EN 102102, see Tables
2.4a and 2.4b.
Internationally, the delivery standards in various
countries deviate considerably with respect to the
thickness and mass tolerances (Packer, 1993). In
most countries besides the thickness tolerance, a
mass tolerance is given, which limits extreme
deviations. However, in some production standards,
e.g. in the USA, the thickness tolerance is not always
compensated by a mass tolerance. This has resulted
in associated design specifications which account for
this, by designating a lower "design wall thickness" of
0,9 or 0,93 times the nominal thickness t. In Eurocode
3, where design is based on nominal thicknesses, the
thickness tolerances in EN 102102 and EN 102192
are (partly) compensated by the mass tolerance. It is
foreseen that in the next revision these tolerances will
be tightened.
Although the circular, square and rectangular hollow
sections are the generallyused shapes; other shapes
are sometimes available. For example, some tube
manufacturers deliver the shapes given in Table 2.5.
Of these, the elliptical hollow sections have become
more popular for architectural designs. These shapes
are not dealt with further in this book. However, more
information about elliptical hollow sections can be
found in Bortolotti et al. (2003), Chan & Gardner
(2008), Choo et al. (2003), MartinezSaucedo et al.
10
(2008), Packer et al. (2009b), Pietrapertosa & Jaspart
(2003), Theofanous et al. (2009), Willibald et al.
(2006) and Zhao & Packer (2009).
2.3 GEOMETRIC PROPERTIES
2.3.1 Tension
The design capacity N
t,Rd
of a member under tensile
loading depends on the cross sectional area and the
design yield strength, and is independent of the
sectional shape. In principle, there is no advantage or
disadvantage in using hollow sections from the point
of view of the amount of material required. The design
capacity is given by:
yd Rd , t
Af N = (2.2)
If the cross section is weakened by bolt holes or slots,
the net cross section should also be checked, in a
similar way as for other sections, e.g. according to
Eurocode 3 (EN 199318, 2005):
9 , 0
f A
N
2 M
u net
Rd , t
= (2.3)
where the partial safety factor
M2
= 1,25.
The factor 0,9 may vary from country to country
depending on the partial factor
M
used. Where ductile
behaviour is required (e.g. under seismic loading), the
yield resistance shall be less than the ultimate
resistance at the net section of fastener holes.
2.3.2 Compression
For centrally loaded members in compression, the
critical buckling load depends on the slenderness
and the section shape.
The slenderness is given by the ratio of the buckling
length
b
and the radius of gyration i.
i
b
= (2.4)
The radius of gyration of a hollow section (in relation
to the member mass) is generally much higher than
that for the weak axis of an open section. For a given
length, this difference results in a lower slenderness
for hollow sections and thus a lower mass when
compared with open sections.
The buckling behaviour is influenced by initial
eccentricities, straightness and geometrical tolerances
as well as residual stresses, nonhomogeneity of the
steel and the stressstrain relationship.
Based on extensive investigations by the European
Convention for Constructional Steelwork (ECCS) and
CIDECT, "European buckling curves" (Fig. 2.3 and
Table 2.7) have been established for various steel
sections including hollow sections. They are
incorporated in Eurocode 3 (EN 199311, 2005).
The reduction factor _ shown in Fig. 2.3 is the ratio of
the design buckling capacity N
b,Rd
to the axial plastic
capacity.
yd
Rd , b
Rd , pl
Rd , b
f
f
N
N
= = _ (2.5)
where:
f
b,Rd
=
A
N
Rd , b
(2.6)
The nondimensional slenderness is determined
by:
E
= (2.7)
where:
y
E
f
E
t = (Euler slenderness) (2.8)
The buckling curves for the hollow sections are
classified according to Table 2.6. Most open sections
fall under curves "b" and "c". Consequently, for the
case of buckling, the use of hot formed hollow
sections generally provides a considerable saving in
material.
Fig. 2.4 illustrates, for a buckling length of 3 m, a
comparison between the required mass of open and
hollow sections for a given load. It shows that in those
cases in which loads are small, leading to relatively
slender sections, hollow sections provide a great
advantage (considerably lower use of material).
However, if loads are higher, resulting in low
slenderness, the advantage (in %) will be less.
The overall buckling behaviour of hollow sections
improves with increasing diameter or widthtowall
11
thickness ratio. However, this improvement is limited
by local buckling. To prevent local buckling, d/t or b/t
limits are given e.g. in Eurocode 3 (EN 199311,
2005), see Table 2.7. In the case of thin walled
sections, interaction between global and local buckling
should be considered.
In addition to the improved buckling behaviour due to
the high radius of gyration and the enhanced design
buckling curve, hollow sections can offer other
advantages in lattice girders. Due to the torsional and
bending stiffness of the members in combination with
joint stiffness, the effective buckling length of
compression members in lattice girders can be
reduced (Fig. 2.5). Eurocode 3 (EN 199311)
recommends an effective buckling length for hollow
section brace members in welded lattice girders equal
to or less than 0,75, in which represents the system
length, see also Rondal et al. (1992).
For chords, 0,9 times the system length for inplane
buckling or 0,9 times the length between the supports
for outofplane buckling, is taken as the effective
buckling length.
These reductions are also based on the fact that the
chord and brace members are generally not fully
optimised. If for example the chord would be fully
utilized with different members for every panel then
these reductions would not be allowed.
Laterally unsupported compression chords of lattice
girders (see Fig. 2.6) have a reduced buckling length
due to the improved torsional and bending stiffness of
the tubular members (Baar, 1968; Mouty, 1981).
These factors make the use of hollow sections in
girders or trusses even more favourable.
2.3.3 Bending
In general, I and H sections are more economical
under bending about the major axis (I
max
larger than
for hollow sections). Only in those cases in which the
design resistance in open sections is largely reduced
by lateral buckling, hollow sections offer an
advantage.
It can be shown by calculations that lateral instability
is not critical for circular hollow sections, square
hollow sections and for the most commonly used
rectangular hollow sections with bending about the
strong axis. Table 2.8 shows allowable spantodepth
ratios for the most commonly used sections (EN
199311, 2005). According to a study of Kaim (2006)
these limits can be taken considerably larger.
It is apparent that hollow sections are especially
favourable compared to other sections if bending
about both axes is present.
Hollow sections used for elements subjected to
bending can be more economically designed by using
plastic design. However, then the sections have to
satisfy more restricted conditions to avoid premature
local buckling. Like other steel sections loaded in
bending, different momentrotation behaviour can be
observed.
Fig. 2.7 shows various momentrotation diagrams for
a member loaded by bending moments.
The momentrotation curve "1" shows a moment
exceeding the plastic moment M
pl
and a considerable
rotation capacity. Momentrotation curve "2" shows a
moment exceeding the plastic moment capacity M
pl
,
but after the maximum, the moment drops
immediately, so that little momentrotation capacity
exists. Momentrotation curve "3" represents a
capacity lower than the plastic moment capacity,
which, however, exceeds the yield moment capacity
M
el
. In the momentrotation curve "4" the capacity is
even lower than the yield moment capacity M
el
and
failure is by elastic buckling. The effect of the
momentrotation behaviour is reflected in the
classification of cross sections as shown in Fig. 2.8
and Table 2.7. The cross section classification is
given in limits for the diameter or flat
widthtothickness ratio.
The limits are based on experiments and can be
expressed as:
yd
f
235
c
t
d
= for CHS (2.9)
f
235
c 3
t
b
yd
= for RHS (2.10a)
yd
f
235
c 3
t
h
= for RHS (2.10b)
with f
yd
in N/mm
2
and c depending on the section
class, the cross section and the loading. For RHS, it is
conservatively assumed that the width of the "flat" is
equal to the external width b or depth h of the RHS
minus 3t.
12
The cross section classes 1 and 2 can develop the
plastic moment capacity up to the given b/t or d/t limits
with bilinear stress blocks, whereas the moment
capacity of the cross section classes 3 and 4 is based
on an elastic stress distribution (see Fig. 2.8). The
difference between the cross section classes 1 and 2
is reflected in the rotation capacity. After reaching the
plastic moment capacity, the cross section class 1 can
keep this capacity after further rotation, whereas the
capacity of the cross section class 2 drops after
reaching this capacity. As a consequence, the
moment distribution in the structure or structural
component should be determined by elastic analysis
for structures made of sections with cross section
classes 2, 3 or 4. For structures made of sections with
cross sections in class 1 a plastic moment distribution
can be adopted, but an elastic moment distribution is
still permissible (and in some countries more
common).
For a class 1 beam fully clamped at both ends and
subjected to a uniformly distributed loading q, the
plastic moment distribution implies that after reaching
the plastic moment capacity at the ends, the beam
can be loaded until a further plastic hinge occurs at
mid span (see Fig. 2.9).
For the class 4 cross section, the maximum stress is
determined by local buckling and the stress in the
outer fibre is lower than the yield strength f
y
.
Alternatively, an effective cross sectional area based
on the yield strength may be determined.
Detailed information about the cross sectional
classification is given by Rondal et al. (1992).
Research by Wilkinson & Hancock (1998) showed
that especially the limits for the side wall slenderness
of RHS need to be reduced considerably. E.g. for
class 1 sections, they suggest reducing the Eurocode
3 limits (EN 199311) for the side wall slenderness to:
6t
) 2r 2t 5(b
70
t
2r) 2t (h
s
(2.11)
with 30
t
r 2 t 2 b
s
For r = t, this can be simplified to:
t
b
83 , 0 77
t
h
s with 34
t
b
s (2.11a)
In the absence of shear forces or if the shear forces
do not exceed 50% of the shear capacity V
pl,Rd
, the
effect of shear may be neglected and the bending
moment capacity about one axis is given by:
yd pl Rd , c
f W M = for cross section classes 1 or 2 (2.12)
yd el Rd , c
f W M = for cross section class 3 (2.13)
yd eff Rd , c
f W M = for cross section class 4 (2.14)
When the shear force exceeds 50% of the shear
capacity, combined loading has to be considered, see
Eurocode 3 (EN 199311).
2.3.4 Shear
The elastic shear stress in circular and rectangular
hollow sections can be determined with simple
mechanics by:
3
f
t I 2
S V
yd
Ed
s = t (2.15)
Fig. 2.10 shows the elastic stress distribution. The
design capacity based on plastic design can be easily
determined based on the HuberHenckyVon Mises
criterion by assuming the shear yield strength in those
parts of the cross section active for shear.
3
f
A V
yd
v Rd , pl
= (2.16)
where:
h b
h
A A
v
+
= for RHS (2.17)
(or just 2
h
t) with V in the direction of h.
A
2
A
v
t
= for CHS (2.18)
2.3.5 Torsion
Hollow sections, especially CHS, have the most
effective cross section for resisting torsional moments,
because the material is uniformly distributed about the
polar axis. A comparison of open and hollow sections
of nearly identical mass in Table 2.9 shows that the
torsional constant of hollow sections is about 200
times that of open sections.
13
The design capacity for torsional moments is
described by:
3
f
W M
yd
t Rd , t
= (2.19)
or circular hollow sections: F
t ) t d (
2 t d
I 2
W
t
t
t
=
= (2.20)
here: w
( ) t t d
4
I
3
t
t
~ (2.21)
or rectangular hollow sections (Marshall, 1971): F
A
m
t
t
A
2 t
I
W
+
= (2.22)
here: w
A
A
3
t
t A 4
3
t
I
2
m
+ ~ (2.23)
) (2.24)
(2.25)
or thin walled rectangular hollow sections, eq. (2.22)
(2.26)
he first term in eq. (2.23) is generally only used for
he exact, more complicated equations for the cross
.3.6 Internal pressure
he design capacity per unit length, shown in Fig.
( ) ( t + = 4 r 2 b h 2
m m m A
( ) t = 4 r h b A
2
m m m m
F
can be approximated by:
t b h 2 W
m m t
=
T
open sections. However, research by Marshall (1971)
showed that the given formula provides the best fit
with the test results.
T
sectional properties are given in EN 102102 (2006)
and EN 102192 (2006).
2
he circular hollow section is most suitable to resist T
an internal pressure p.
T
2.11, is given by:
t 2 d
t 2
f p
yd
= (2.27)
M0
ectional classification,
.4 DRAG COEFFICIENTS
hollow sections,
.5 CORROSION PROTECTION
tructures designed in hollow sections have a 20 to
eq. (2.27), = 1,0, but for transport pipelines, the In
M0
value may be considerably larger than for other
cases, depending on the hazard of the product, the
effect of failure on the environment and inspectability.
The design capacities for RHS sections subjected to
internal pressure are much more complicated;
reference can be made to the Deutscher
Dampfkesselausschu (1975).
2.3.7 Combined loadings
arious combinations of loadings are possible, e.g. V
tension, compression, bending, shear and torsion.
Depending on the cross s
various interaction formulae should be applied.
Reference can be made to Eurocode 3 (EN 199311).
It is beyond the scope of this book to deal with all
possible formulae; however, the interaction between
the various loads in the cross section can be based on
the HuberHenckyVon Mises stress criterion (Roik &
Wagenknecht, 1977). For the member checks, other
interaction formulae apply, see e.g. EN199311.
2
ially circular Hollow sections, espec
have a striking advantage for use in structures
exposed to fluid currents, i.e. air or water. The drag
coefficients are much lower than those of open
sections with sharp edges, see Fig. 2.12 (Schulz,
1970; CIDECT, 1984; Dutta, 2002).
2
tructures made of hollow sections offer advantages S
with regard to corrosion protection. Hollow sections
have rounded corners (Fig. 2.13) resulting in a better
protection than that for sections with sharp corners.
This is especially true for the joints in circular hollow
sections where there is a smooth transition from one
section to another. This better protection increases
the protection period of coatings against corrosion.
S
50% smaller surface to be protected than comparable
structures made of open sections. Many
14
investigations, summarized by Tissier (1978), have
been conducted to assess the likelihood of internal
corrosion. These investigations, carried out in various
countries, show that internal corrosion does not occur
in sealed hollow sections.
Even in hollow sections which are not perfectly
.6 USE OF INTERNAL VOID
he possibilities of using the internal space are briefly
.6.1 Concrete filling
wall thicknesses are not
very important reason for using concrete filled
oncrete filling of hollow sections contributes not only
.6.2 Fire protection by water circulation
nother method for fire protection of buildings is to
he columns are interconnected with a water storage
order to prevent freezing, potassium carbonate
.6.3 Heating and ventilation
he inner voids of hollow sections are sometimes
.6.4 Other possibilities
ometimes hollow section chords of lattice girder
.7 AESTHETICS
rational use of hollow sections leads in general to
sealed, internal corrosion is limited. If there is concern
about condensation in an imperfectly sealed hollow
section, a drainage hole can be made at a point where
water can drain by gravity.
2
he internal void in hollow sections can be used in T
various ways, e.g. to increase the compressive
resistance by filling with concrete, or to provide fire
protection. In addition, heating or ventilation systems
are sometimes incorporated into hollow section
columns.
T
described below.
2
the commonlyavailable If
sufficient to meet the required load bearing resistance,
the hollow section can be filled with concrete. For
example, it may be preferable in buildings to have the
same external dimensions for the columns on every
floor. At the top floor, the smallest wall thickness can
be chosen, and the wall thickness can be increased
with increasing load for lower floors. If the hollow
section with the largest available wall thickness is not
sufficient for the ground floor, the hollow section can
be filled with concrete to increase the load bearing
resistance.
A
hollow sections is that the columns can be relatively
slender. Design rules are given in e.g. Eurocode 4
(EN 199411, 2004).
C
to an increase in load bearing resistance, but it also
improves the fire resistance duration. Extensive test
projects carried out by CIDECT and the European
Coal and Steel Community (ECSC) showed that
reinforced concrete filled hollow section columns
without any external fire protection like plaster,
vermiculite panels or intumescent paint, can attain a
fire life of even 2 hours depending on the cross
section ratio of the steel and concrete, reinforcement
percentage of the concrete and the applied load, see
Fig. 2.14 (Twilt et al., 1994).
2
A
use water filled hollow section columns.
T
tank. Under fire conditions, the water circulates by
convection, keeping the steel temperature below the
critical value of 450 C. This system has economical
advantages when applied to buildings with more than
about 8 storeys. If the water flow is adequate, the
resulting fire resistance time is virtually unlimited.
In
(K
2
CO
3
) is added to the water. Potassium nitrate is
used as an inhibitor against corrosion.
2
T
used for air and water circulation for heating and
ventilation of buildings. Many examples in offices and
schools show the excellent combination of the
strength function of hollow section columns with the
integration of heating or ventilation systems. This
system offers maximization of floor area through
elimination of heat exchangers, a uniform provision of
warmth and a combined protection against fire.
2
S
bridges are used for conveying fluids (pipe bridge). In
buildings, the rain water downpipes may go through
the hollow section columns or in other cases electrical
wiring is located in the columns. The internal space
can also be used for prestressing a hollow section.
2
A
structures which are cleaner and more spacious.
Hollow sections can provide slender aesthetic
columns, with variable section properties but flush
external dimensions. Due to their torsional rigidity,
hollow sections have specific advantages in folded
structures, Vtype girders, etc.
15
16
often made of hollow
sections directly connected to one another without any
stiffener or gusset plate, is often preferred by
architects for structures with visible steel elements.
However, it is difficult to express aesthetic features in
economic comparisons. Sometimes hollow sections
are used only because of aesthetic appeal, see e.g.
Fig. 2.15, whilst at other times appearance is less
important.
Lattice construction, which is
Table 2.1a Hot finished structural hollow sections Nonalloy steel properties (EN 102101, 2006)
Minimum yield strength
(1)
(N/mm
2
)
Minimum tensile
strength
(N/mm
2
)
Longitudinal
(2)
minimum elongation (%)
on gauge
o o
S 65 , 5 L =
Charpy impact
strength
(10 x 10 mm)
Steel
designation
t s 16
mm
16 < t s 40
mm
40< t s 63
mm
t < 3
mm
3 s t s 100
mm
3 < t s 40
mm
40 < t s 63
mm
Temp.
C
J
S235JRH 235 225 215 360510 360510 26 25 20 27
S275J0H
S275J2H
275 265 255 430580 410560 23 22
0
20
27
27
S355J0H
S355J2H
S355K2H
355 345 335 510680 470630 22 21
0
20
20
27
27
40
(3)
(1)
For thicknesses above 63 mm, these values are further reduced.
(2)
In transverse direction 2% lower.
(3)
Corresponds to 27 J at 30 C.
Table 2.1b Cold formed welded structural hollow sections Nonalloy steel (EN 102191, 2006) Steel
properties different from EN 102101 (2006)
Steel designation
Minimum longitudinal elongation (%),
all thicknesses, t
max
= 40 mm
S235JRH 24
(1)
S275J0H
S275J2H
20
(2)
S355J0H
S355J2H
S355K2H
20
(2)
(1)
For t > 3 mm and d/t < 15 or 5 , 12
t 2
h b
<
+
the minimum elongation is reduced by 2 to 22%; for t s 3 mm the minimum
elongation is 17%.
(2)
For d/t < 15 or 5 , 12
t 2
h b
<
+
the minimum elongation is reduced by 2 to 18%.
17
Table 2.2a Hot finished structural hollow sections Fine grain steel properties (EN 102101, 2006)
Minimum yield strength
(N/mm
2
)
Minimum
tensile strength
(N/mm
2
)
Minimum elongation (%)
on gauge
o o
S 65 , 5 L =
t s 65 mm
Charpy impact
strength
(10 x 10 mm)
Steel
designation
t s 16
mm
16 < t s 40
mm
40 < t s 65
mm
t s 65
mm
Long. Trans.
Temp.
C
J
S275NH
S275NLH
275 265 255 370510 24 22
20
50
40
(1)
27
S355NH
S355NLH
355 345 335 470630 22 20
20
50
40
(1)
27
S420NH
S420NLH
420 400 390 520680 19 17
20
50
40
(1)
27
S460NH
S460NLH
460 440 430 540720 17 15
20
50
40
(1)
27
(1)
Corresponds to 27 J at 30 C.
Table 2.2b Cold formed welded structural hollow sections Fine grain steel (EN 102191, 2006) Steel
properties different from EN 102101 (2006)
Feed stock condition M
(1)
Steel designation
Minimum tensile strength
(N/mm
2
)
Minimum longitudinal
elongation (%)
(2)
S275MH
S275MLH
360  510 24
S355MH
S355MLH
450  610 22
S420NH
S420NLH
520  660 19
S460NH
S460NLH
530  720 17
(1)
M refers to thermal mechanical rolled steels.
(2)
For d/t < 15 or
5 , 12
t 2
h b
<
+
the minimum elongation is reduced by 2, e.g. from 24% to 22% for S275MH and S275MLH.
Table 2.3 Minimum inner corner radii of cold formed RHS according to EN 199318 (2005)
Maximum wall thickness t (mm)
General
r/t
Strain due to cold
forming (%)
Predominantly static
loading
Fatigue dominating
Aluminiumkilled steel
(Al > 0,02%)
> 25
> 10
> 3,0
> 2,0
> 1,5
> 1,0
s 2
s 5
s 14
s 20
s 25
s 33
any
any
24
12
8
4
any
16
12
10
8
4
any
any
24
12
10
6
18
Table 2.4a Hot finished structural hollow sections Tolerances (EN 102102, 2006)
Section type Square/rectangular Circular
Outside dimension the greater of 0,5 mm and 1%
(1)
the greater of 0,5 mm and 1% but not
more than 10 mm
Welded 10%
Thickness
Seamless 10% and 12,5% at maximum 25% cross section
Welded 6% on individual lengths
Mass
Seamless 6%; +8%
Straightness 0,2% of the total length and 3 mm over any 1 m length
Length (exact) +10 mm, 0 mm, but only for exact lengths of 2000 to 6000 mm
Out of roundness  2% for d/t s 100
Squareness of sides 90 1 
Corner radii Outside 3,0t maximum 
Concavity/convexity 1% of the side 
Twist 2 mm + 0,5 mm/m
(1)

(1)
For elliptical hollow sections with h s 250 mm, the tolerances are twice the values given in this table.
Table 2.4b Cold formed welded structural hollow sections (EN 102192, 2006) Tolerances different
from EN 102102 (2006)
Section type Square/rectangular Circular
Outside dimension
b < 100 mm: the greater of 0,5 mm and 1%
100 mm s h, b s 200 mm: 0,8%
b > 200 mm: 0,6%
1%, minimum 0,5 mm
maximum 10 mm
Thickness Welded
t s 5 mm: 10%
t > 5 mm: 0,5 mm
for d s 406,4 mm:
t s 5 mm: 10%
t > 5 mm: 0,5 mm
for d > 406,4 mm:
10% with maximum 2,0 mm
Mass 6% 6%
Straightness
0,15% of the total length and 3 mm over any 1 m
length
Outside corner radii
(profile)
t s 6 mm: 1,6 to 2,4t
6 mm < t s 10 mm: 2,0 to 3,0t
t > 10 mm: 2,4 to 3,6t

Concavity/convexity maximum 0,8% with a minimum of 0,5 mm 
Table 2.5 Special shapes available
Triangular Hexagonal Octagonal Flat  oval Elliptical Halfelliptical
Shape
19
Table 2.6 European buckling curves according to manufacturing processes (EN 199311, 2005)
Cross section Manufacturing process Buckling curves
Hot finished
420 N/mm
2
< f
y
s 460 N/mm
2
a
0
Hot finished
f
y
s 420 N/mm
2
a
Cold formed c
h
b
Flange
Web
t
h
b
Flange
Web
h
b
h
b
Flange
Web
t
h
b
Flange
Web
t
h
b
Flange
Web
h
b
h
b
Flange
Web
t
b
t
d
t
d
t
d
t
h
b
Flange
Web
t
h
b
Flange
Web
h
b
h
b
Flange
Web
t
h
b
Flange
Web
t
h
b
Flange
Web
h
b
h
b
Flange
Web
t
b
t
d
t
d
t
d
t
hh
Table 2.7 Limits for b/t, h/t and d/t for cross section classes 1, 2 and 3 (EN 199311, 2005)
Class 1 2 3
f
yd
(N/mm
2
) f
yd
(N/mm
2
) f
yd
(N/mm
2
)
Cross
section
Load type
Considered
element
235 275 355 460 235 275 355 460 235 275 355 460
3
f
235
c
t
b
yd
+ s
c = 33 c = 38 c = 42
RHS
b/t
(1)
Compression Top face
36,0 33,5 29,8 26,6 41,0 38,1 33,9 30,2 45,0 41,8 37,2 33,0
3
f
235
c
t
h
yd
+ s
c = 72 c = 83 c = 124
RHS
h/t
(1)
Bending Side wall
(2)
75,0 69,6 61,6 51,8 86,0 79,7 70,5 62,3 127,0 117,6 103,9 91,6
yd
f
235
c
t
d
s
c = 50 c = 70 c = 90
CHS
d/t
Compression
and/or
bending
t
t
t
t
50,0 42,7 33,1 25,5 70,0 59,8 46,3 35,8 90,0 76,9 59,6 46,0
(1)
For all hot finished and cold formed RHS, it is conservative to assume that the widthtothickness ratio of the "flat" is
3
t
b
t
2r  2t  b
= or 3
t
h
t
2r  2t  h
= .
(2)
Wilkinson & Hancock (1998) suggested reducing the Eurocode limits (EN 199311) for the side wall slenderness of RHS
considerably, e.g. for class 1 in a simplified form to:
t
b
83 , 0 77
t
h
s with 34
t
b
s .
20
Table 2.8 Allowable spantodepth ratios L/(ht) to avoid lateral buckling based on EN 199311 (2005)
s
t h
L
t h
t b
S235 S275 S355 S460
0,5 73,7 63,0 48,8 37,7
0,6 93,1 79,5 61,6 47,5
0,7 112,5 96,2 74,5 57,5
0,8 132,0 112,8 87,4 67,4
0,9 151,3 129,3 100,2 77,3
1,0 170,6 145,8 112,9 87,2
h
b
Flange
b
t
h
b
Flange
b
h
b
h
b
Flange
b
t
h
b
Flange
b
t
h
b
Flange
b
h
b
h
b
Flange
b
t
b
h
t
h
b
Flange
b
t
h
b
Flange
b
h
b
h
b
Flange
b
t
h
b
Flange
b
t
h
b
Flange
b
h
b
h
b
Flange
b
t
b
h
t
Table 2.9 Torsional strength of various sections
Section
Mass
(kg/m)
Torsion constant I
t
(10
4
mm
4
) or (cm
4
)
UPN 200 25,3 11,9
INP 200 26,2 13,5
HEB 120 26,7 13,8
HEA 140 24,7 8,1
140 x 140 x 6 24,9 1475
168.3 x 6 24,0 2017
21
Fig. 2.1 Lamellar tearing
Actual f
y
mean
after cold forming
Actual f
y
mean
after cold forming
Actual f
y
mean
after cold forming
Fig. 2.2 Influence of cold forming on the yield strength for a square hollow section of 100 x 100 x 4 mm
0
1,00
0,75
0,50
0,25
0
0 0,5 1,0 1,5 2,0
00
1,00
0,75
0,50
0,25
0
0 0,5 1,0 1,5 2,0
Fig. 2.3 Eurocode 3 buckling curves (EN 199311, 2005)
22
B
u
c
k
l
i
n
g
s
t
r
e
s
s
(
N
/
m
m
2
)
B
u
c
k
l
i
n
g
s
t
r
e
s
s
(
N
/
m
m
2
)
Fig. 2.4 Comparison of the masses of hollow and open sections under compression in relation to the loading
Fig. 2.5 Restraints for the buckling of a brace member
Fig. 2.6 Bottom chord laterally spring supported by the stiffness of the members, joints and purlins
23
M
pl
M
el
M
e
M
pl
M
el
M
e
Fig. 2.7 Momentrotation curves
Fig. 2.8 Stress distribution for bending
Fig. 2.9 Moment distribution in relation to the cross section classification
24
Fig. 2.10 Elastic shear stress distribution
tf
yd
t f
yd
t f
yd
t
d  2t
Fig. 2.11 Internal pressure
Fig. 2.12 Wind flow for open and circular hollow sections
25
paint layers
steel
steel
corner protection for RHS
and open sections
paint layers
steel
steel
corner protection for RHS
and open sections
paint layers
steel
steel
corner protection for RHS
and open sections
paint layers
steel
steel
corner protection for RHS
and open sections
Fig. 2.13 Painted corners of RHS vs. open sections
Fig. 2.14 Fire resistance of concrete filled hollow sections
RHS 304,8x304,8x9,5
111 min.
14 min.
only
RHS
non
reinforced
concrete
filling
50min.
steel fibre
reinforced
concrete
filling
working load (kN)
f
i
r
e
l
i
f
e
(
m
i
n
.
)
1650. 3150. 3150.
120.
60.
RHS 304,8x304,8x9,5
111 min.
14 min.
only
RHS
non
reinforced
concrete
filling
50min.
steel fibre
reinforced
concrete
filling
working load (kN)
f
i
r
e
l
i
f
e
(
m
i
n
.
)
1650. 3150. 3150.
120.
60.
RHS 304,8x304,8x9,5
111 min.
14 min.
only
RHS
non
reinforced
concrete
filling
50min.
steel fibre
reinforced
concrete
filling
working load (kN)
f
i
r
e
l
i
f
e
(
m
i
n
.
)
1650. 3150. 3150.
120.
60.
26
Fig. 2.15 Aesthetically appealing structures
27
28
3. APPLICATIONS
The applications of structural hollow sections nearly
cover all fields. Hollow sections may be used because
of the beauty of their shape or to express lightness,
while in other cases their geometrical properties
determine their application. In this chapter, examples
are given for the various fields and to show the
possibilities of constructing with hollow sections.
3.1 BUILDINGS AND HALLS
In buildings and halls, hollow sections are mainly used
for columns and lattice girders or space frames for
roofs. In modern architecture, they are also used for
other structural or architectural reasons, e.g. facades.
Fig. 3.1 shows a 10storey building in Karlsruhe,
Germany with rectangular hollow section columns
180 x 100. Special aspects are that the columns are
made of weathering steel and are water filled to
ensure the required fire protection. The columns are
connected with water reservoirs to ensure circulation.
Besides the fire protection, a further advantage is that
due to the water circulation in the columns, the
deformation of the building due to temperature
differences by sunshine is limited.
Fig. 3.2 shows an example of lattice girder trusses
used in a roof of an industrial building. For an optimal
cost effective design, it is essential that the truss joints
are made without any stiffening plates.
An especially appealing application is given in Fig.
3.3, showing a treetype support of the airport
departure hall in Stuttgart, Germany. For the joints,
streamlined steel castings are used.
Fig. 3.4 shows the roof of the terminal of Kansai
International Airport in Osaka, Japan with curved
triangular girders of circular hollow sections.
Fig. 3.5 shows a dome under construction, whereas
Fig. 3.6 illustrates a special application using columns
and beams in the faade for ventilation assuring clean
windows in the swimming pool.
Fig. 3.7 shows a very nice architectural application in
Bush Lane House in the city of London, UK. The
external circular hollow section lattice transfers the
faade loads and the loads on the floors to the main
columns. The hollow sections are filled with water for
fire protection.
Very attractive applications can be found in the halls
and buildings for the Olympic Games in Athens, e.g.
Fig. 3.8.
Elliptical hollow sections are becoming more and
more popular among architects and already several
examples exist, see for example Fig. 3.9, the airport
building in Madrid.
Nowadays, many examples of tubular structures are
found in railway stations (Figs. 3.10 and 3.11) and
roofs of stadia and halls (Figs. 3.12 to 3.14).
Indeed, as stated by one of the former CIDECT vice
presidents, Jim Cran, at the Tubular Structures
Symposium in Delft (1977) "The sky is the limit", whilst
presenting beautiful applications of structural hollow
sections.
3.2 BRIDGES
As mentioned in the introduction, the Firth of Forth
Bridge is an excellent example of using the hollow
section shape for structural applications in bridges.
Nowadays, many modern examples exist (IISI, 1997).
Figs. 1.4, 3.15 to 3.17 and 3.20 show various
examples of pedestrian bridges; some of these are
movable bridges.
Circular hollow sections can also be used as a flange
for plate girders, as shown in Fig. 3.17 for a triangular
box girder.
A very nice example of a roadpedestrian bridge is
illustrated in Fig. 3.18, being a composite
steelconcrete bridge with hollow sections for the arch
and braces and a concrete deck.
Fig. 3.19 shows a railway bridge near Rotterdam, The
Netherlands with circular hollow section arches.
3.3 BARRIERS
There are a few aspects which make hollow sections
increasingly suitable for hydraulic structures, such as
barriers. Due to environmental restrictions, the
maintenance of hydraulic structures requires severe
precautions, making durability an important issue.
Structures of hollow sections are less susceptible to
corrosion due to the rounded corners. Furthermore,
especially circular hollow sections have lower drag
coefficients, leading to lower forces due to wave
loading. Fig. 3.21 shows a barrier with a support
29
structure of circular hollow sections. Fig. 3.22 shows
the storm surge barrier near Hook of Holland with
triangular arms made of circular hollow sections and a
length (250 m) equal to the height of the Eiffel Tower
in Paris.
3.4 OFFSHORE STRUCTURES
Offshore, many application examples are available;
most of them in circular hollow sections. For the
support structure, the jacket or tower, not only is the
wave loading important, but also other aspects are
leading to the use of circular hollow sections. E.g. in
jackets, the circular hollow section piles are often
driven through the circular hollow section legs of the
jacket, thus the pile is guided through the leg.
Sometimes the internal void is used for buoyancy.
Further, the durability and easy maintenance in
severe environments are extremely important.
Hollow section members are used in jackets, towers,
the legs and diagonals in topside structures, cranes,
microwave towers, flare supports, bridges, support
structures of helicopter decks and further in various
secondary structures, such as staircases, ladders, etc.
Figs. 3.23 and 3.24 show two examples.
3.5 TOWERS AND MASTS
Considering wind loading, corrosion protection and
architectural appearance, there is no doubt that hollow
sections are to be preferred. However, in many
countries, electric power transmission towers are
made of angle sections with simple bolted joints.
Nowadays, architectural appearance becomes more
important, while stringent environmental restrictions
make protection and maintenance increasingly
expensive. These factors stimulate designs made of
hollow sections (Figs. 3.25 and 3.26).
3.6 SPECIAL APPLICATIONS
The field of special applications is large, e.g. along the
roads, petrol stations (Fig. 3.27), sound barriers (Fig.
3.28), traffic information gantries (Fig. 3.29), guard
rails, parapets and sign posts.
Further, excellent application examples are found in
radio telescopes (Fig. 3.30), in mechanical
engineering, cranes (Fig. 3.31) and roller coasters
(Fig. 3.32).
In the agricultural field, glass houses (Fig. 3.33) and
agricultural machinery are typical examples. Also in
transport, many examples exist but these are outside
the scope of this book. Indeed, the sky is the limit.
30
Fig. 3.2 Roof with lattice girders
Fig. 3.1 Faade of the Institute for Environment in
Karlsruhe, Germany
Fig. 3.3 Airport departure hall in Stuttgart, Germany
31
Fig. 3.4 Roof of Kansai International Airport in Osaka,
Japan
Fig. 3.6 Faade with ventilation through the RHS
columns and beams, Borkum, Germany
Fig. 3.8 Hall for the 2004 Olympic Games, Athens,
Greece
Fig. 3.5 Dome structure in Gothenburg, Sweden
Fig. 3.7 Bush Lane House in London, UK
Fig. 3.9 Airport Madrid with EHS sections, Spain
32
Fig. 3.10 Railway station in Rotterdam, The
Netherlands
Fig. 3.12 Barrel dome grid for the Trade Fair building
in Leipzig, Germany
Fig. 3.13 Retractable roof for the Rogers Centre in
Toronto, Canada
Fig. 3.11 TGV railway station at Charles de Gaulle
Airport, France
Fig. 3.14 Stadium Australia for the 2000 Olympic
Games, Sydney, Australia
33
Fig. 3.15 Movable pedestrian bridge in RHS, The
Netherlands
Fig. 3.17 Pedestrian bridge in Houdan, France
Fig. 3.19 Railway bridge with CHS arches, The
Netherlands
Fig. 3.21 Eastern Scheldt barrier, The Netherlands
Fig. 3.16 Movable pedestrian bridge in RHS near
Delft, The Netherlands
Fig. 3.18 Composite road bridge in Marvejols, France
Fig. 3.20 Movable pedestrian bridge in CHS near
Delft, The Netherlands
Fig. 3.22 Storm surge barrier, The Netherlands
34
Fig. 3.23 Bullwinkle offshore structure, Gulf of Mexico
Fig. 3.25 Electric power transmission tower
Fig. 3.27 Petrol station, The Netherlands
Fig. 3.24 Amoco P15 offshore platform with jackup,
North Sea
Fig. 3.26 Mast, The Netherlands
Fig. 3.28 Sound barrier, Delft, The Netherlands
35
Fig. 3.29 Traffic information gantry, The Netherlands
Fig. 3.30 Radio telescope
Fig. 3.31 Cranes
Fig. 3.33 Green house, The Netherlands Fig. 3.32 Roller coaster
36
4. COMPOSITE STRUCTURES
4.1 INTRODUCTION
Concrete filled hollow sections (Fig. 4.1) are mainly
used for columns. The concrete filling gives a higher
load bearing capacity without increasing the outer
dimensions. The fire resistance can be considerably
increased by concrete filling, in particular if proper
reinforcement is used.
Due to the fact that the steel structure is visible, it
allows a slender, architecturallyappealing design. The
hollow section acts not only as the formwork for the
concrete, but also ensures that the assembly and
erection in the building process are not delayed by the
hardening process of the concrete.
CIDECT research on composite columns started
already in the sixties, resulting in monographs and
design rules, adopted by Eurocode 4 (EN 199411,
2004). CIDECT Design Guide No. 5 (Bergmann et al.,
1995) provides detailed information for the static
design of concrete filled columns.
To a large extent, this chapter follows the information
given in Design Guide No. 5, but updated with the
latest revisions to Eurocode 4 (EN 199411).
4.2 DESIGN METHODS
In the last decades, several design methods for
composite columns were developed, e.g. in Europe by
Guiaux & Janss (1970), Roik et al. (1975) and Virdi &
Dowling (1976), finally resulting in the design rules
given in Eurocode 4 (EN 199411, 2004).
In this chapter, the design method given is based on
the approach presented in Eurocode 4 (EN 199411).
The design of composite columns has to be carried
out at the ultimate limit state, i.e. the effect of the most
unfavourable combination of actions should not
exceed the resistance of the composite member.
An exact calculation of the load bearing capacity
considering the effect of imperfections and deflections
(second order analysis), the effect of plastification of
the section, cracking of the concrete, etc. can only be
carried out by means of a computer program. With
such a program, the resistance interaction curves as
shown in Fig. 4.2, can be calculated. Based on these
calculated capacities, the following simplified design
methods have been developed.
4.3 AXIALLY LOADED COLUMNS
From the work of Roik et al. (1975), a simplified
design method is given in Eurocode 4 (EN 199411),
similar to the design method adopted for steel
columns, i.e.:
Rd , pl Ed
N N _ s (4.1)
where:
N
Ed
design normal force (including load factors)
reduction factor for the relevant buckling curve,
i.e. curve "a" for
s
s 3% and
curve "b" for 3% <
s
s 6% (see Fig. 2.3)
N
pl,Rd
resistance of the cross section to normal force
according to eq. (4.2)
N
pl,Rd
= A
a
f
yd
+ A
c
f
cd
+ A
s
f
sd
(4.2)
where:
A
a
, A
c
, A
s
cross sectional areas of structural steel,
concrete and reinforcement
f
yd
, f
cd
, f
sd
design strengths of steel, concrete (see
Table 4.1) and reinforcement using the
recommended
M
factors according to
Eurocode 2 (EN 199211, 2004) and
Eurocode 3 (EN 199311, 2005) being
a
=
1,0 for f
y
,
c
= 1,5 for f
c
, and
s
= 1,15 for f
s
The load factors for the actions
F
have to be
determined from EN 1990 (2002).
Concrete classes higher than C50/60 should not be
used without further investigation and classes lower
than C20/25 are not allowed for composite
construction.
In concrete filled hollow sections, the concrete is
confined by the hollow section. Therefore, the
concrete strength reduction factor of 0,85 does not
have to be considered.
The reduction factor follows from the relative
slenderness
eff , cr
Rk , pl
E
N
N
=
= (4.3)
where:
N
pl,Rk
resistance of the cross section to axial load
according to eq. (4.2), however, with f
yd
, f
cd
and
f
sd
replaced by f
yk
, f
ck
and f
sk
N
cr,eff
elastic buckling capacity of the member (Euler
37
critical load)
N
cr,eff
=
2
b
eff
) EI (
2
t
(4.4)
where:
b
buckling length of the column
(EI)
eff
effective stiffness of the composite section
The buckling (effective) length of the column can be
determined by following the rules of Eurocode 3 (EN
199311).
(EI)
eff
= E
a
I
a
+ 0,6 E
c,eff
I
c
+ E
s
I
s
(4.5)
E
c,eff
=
t
Ed
Ed , G
cm
N
N
1
E
+
(4.6)
where:
I
a
, I
c
, I
s
moments of inertia of the cross sectional
areas of structural steel, concrete (with the
area in tension assumed to be uncracked)
and reinforcement, respectively
E
a
, E
cm
, E
s
moduli of elasticity of structural steel,
concrete and reinforcement
E
c,eff
modulus of elasticity of concrete corrected
for creep with E
cm
according to Table 4.1
N
Ed
acting design normal force
N
G,Ed
permanent part of N
Ed
t
creep factor according to Clause 3.1 of
Eurocode 2 (EN 199211)
The calibration factor 0,6 in eq. (4.5) is incorporated to
consider, for example, the effect of cracking of
concrete under moment action due to second order
effects.
4.3.1 Limitations
The reinforcement to be included in the design
calculations should not exceed 6% of the concrete
area. There is no minimum requirement.
The composite column is considered as "composite"
if:
0,2 s s 0,9 (4.7)
where:
Rd , pl
yd a
N
f A
= o (4.8)
If the parameter is less than 0,2, the column shall be
designed as a concrete column following Eurocode 2
(EN 199211). On the other hand, when exceeds
0,9, the column shall be designed as a steel column
according to Eurocode 3 (EN 199311).
To avoid local buckling, the following limits should be
observed for bending and compression loading (EN
199411, 2004):
 For concrete filled rectangular hollow sections (with
h being the greater overall dimension of the
section):
h/t s 52 (4.9)
 For concrete filled circular hollow sections:
d/t s 90
2
(4.10)
The factor accounts for different yield strengths:
=
yd
f
235
(4.11)
with f
yd
in N/mm
2
.
Although the d/t and h/t values given in Table 4.2 are
equal (for CHS) or higher (for RHS) than those of
class 3 for unfilled sections, the plastic resistance of
the section can be used. However, for the analysis of
the internal forces in a structure, an elastic analysis
should be performed. Further discussions on
slenderness limits for unfilled CHS and RHS and the
effect of concrete filling can be found in Zhao et al.
(2005).
4.3.2 Effect of long term loading
The influence of the longterm behaviour of the
concrete on the load bearing capacity of the column is
included by a modification of the concrete modulus of
elasticity, since the load bearing capacity of the
columns may be reduced by creep and shrinkage. As
shown in eq. (4.6) for a load which is fully permanent,
the modulus of elasticity of the concrete will be
considerably reduced.
38
4.3.3 Effect of confinement
For concrete filled circular hollow section columns with
a small relative slenderness < 0,5 (for CHS, this is
approximately /d s 12) and e/d s 0,1, the bearing
capacity is increased due to the impeded transverse
strains. This results in radial compression in the
concrete and a higher resistance to normal stresses,
see Fig. 4.3. Above these values, the confinement
effect is very small.
For concrete filled rectangular hollow sections, any
confinement effect is neglected.
Detailed information can be found in Eurocode 4 (EN
199411).
4.4 RESISTANCE OF A SECTION TO
BENDING
For the determination of the resistance of a concrete
filled section to bending moments, a full plastic stress
distribution in the section is assumed (Fig. 4.4). The
concrete in the tension zone of the section is assumed
to be cracked and is therefore neglected. The internal
bending moment resulting from the stresses and
depending on the position of the neutral axis is the
resistance of the section to bending moments M
pl,Rd
.
4.5 RESISTANCE OF A SECTION TO
BENDING AND COMPRESSION
The resistance of a concrete filled cross section to
bending and compression can be shown by the
interaction curve between the normal force and the
internal bending moment.
Figs. 4.5 to 4.8 show the interaction curves for RHS
and CHS columns in relation to the cross section
parameter . These curves have been determined
without any reinforcement, but they may also be used
for reinforced sections if the reinforcement is
considered in the values and in N
pl,Rd
and M
pl,Rd
respectively.
The interaction curve has some significant points,
shown in Fig. 4.9. These points represent the stress
distributions given in Fig. 4.10. The internal moments
and axial loads belonging to these stress distributions
can be easily calculated if effects of the corner radius
are excluded.
Comparing the stress distribution of point B, where the
normal force is zero, and that of point C with the same
moment as in point B and axial force N
C,Rd
(Fig. 4.10),
the neutral axis moves over a distance 2h
n
. Hence,
the normal force N
C,Rd
can be calculated by the
additional compressed parts of the section with depth
2h
n
. Because the force N
C,Rd
does not contribute to
the moment M
C,Rd
= M
B,Rd
.
Furthermore, the normal force at point C is twice the
value of that at point D: N
C,Rd
= 2N
D,Rd
.
4.6 INFLUENCE OF SHEAR FORCES
The influence of the shear stresses on the normal
stresses does not need to be considered if:
V
Ed
s 0,5
V
pl,Rd
(4.12)
The shear force on a composite column may either be
assigned to the steel profile alone or be divided into a
steel and a reinforced concrete component. The
component for the structural steel can be considered
by reducing the axial stresses in those parts of the
steel profile which are effective for shear (Fig. 4.11).
The reduction of the axial stresses due to shear
stresses may be carried out according to the
HuberHenckyVon Mises criterion or according to
Eurocode 4 (EN 199411). For the determination of
the crosssection interaction, it is easier to transform
the reduction of the axial stresses into a reduction of
the relevant cross sectional areas equal to that used
for hollow sections without concrete filling:
reduced A
v
= A
v
(
(


.

\

2
Rd , pl
Ed
1
V
V 2
1 (4.13)
3
f
A V
yd
v Rd , pl
= (4.14)
For A
v
, see Chapter 2.
4.7 RESISTANCE OF A MEMBER TO
BENDING AND COMPRESSION
4.7.1 Uniaxial bending and compression
Fig. 4.12 shows the principle of the method for the
design of a composite member under combined
compression and uniaxial bending using the
39
crosssection interaction curve. Due to imperfections,
the resistance of an axially loaded member is given by
eq. (4.1) or on the vertical axis in Fig. 4.12.
The moment capacity factor at the level of is defined
as the imperfection moment. Having reached the load
bearing capacity for axial compression, the column
cannot resist any additional bending moment.
The value of
d
resulting from the actual design
normal force N
Ed
(
d
= N
Ed
/N
pl,Rd
) determines the
moment capacity factor
d
for the capacity of the
member. This factor
d
gives the moment capacity
including the imperfection moment, thus the
imperfection moment should be added to the external
moment including second order effects.
The capacity for the combined compression and
bending of the member can now be checked by:
M
,max
s
M
d
M
pl,Rd
(4.15)
where:
M
,max
design bending moment of the column,
including the imperfection moment and second
order effects
M
0,9 for S235 to S335 and 0,8 for S420 and
S460
d
to be obtained from the interaction diagrams in
Figs. 4.5 to 4.8
The additional reduction by the factor
M
accounts for
the assumptions of this simplified design method, e.g.
the interaction curve of the section is determined
assuming full plastic behaviour of the materials with
no strain limitation.
Note: Interaction curves of the composite sections
always show an increase in the bending capacity
higher than M
pl,Rd
. The bending resistance increases
with an increasing normal force, because former
regions in tension are compressed by the normal
force. This positive effect may only be taken into
account if it is ensured that the bending moment and
the axial force always act together. If this is not
ensured, and the bending moment and the axial force
result from different loading situations, the related
moment capacity
d
has to be limited to 1,0.
Columns with equal end moments
The verification procedure for columns with the same
end moments given in Eurocode 4 (EN 199411) is
as follows:
The second order moment M
Ed,
can be approximated
by:
M
Ed,
= k M
Ed
(4.16)
where:
eff , cr
Ed
N
N
1
1
k
= (4.17)
k is the amplification factor to incorporate the second
order effects.
N
cr,eff
can be determined with eq. (4.4), however, with
a modified (EI)
eff,
due to the simplifications mentioned
before:
(EI)
eff,
= 0,9 (E
a
I
a
+ 0,5 E
c,eff
I
c
+ E
s
I
s
) (4.18)
The total moment including the imperfection moment
is:
M
,max
=
eff , cr
Ed
N
N
1
1
(M
Ed
+ N
Ed
e
0
) (4.19)
The capacity can now be checked with eq. (4.15).
Columns with different end moments
If the end moments are not equal (see Fig. 4.13), then
the k factor in eq. (4.17) has to be corrected for the
external moment by a factor :
eff , cr
Ed
N
N
1
k

= (4.20)
where:
= 0,66 + 0,44r but > 0,44 (4.21)
with r being the ratio between the smallest and largest
end moment (Fig. 4.13).
The total moment including the imperfection moment
is now:
eff , cr
Ed
0 Ed
eff , cr
Ed
Ed
max ,
N
N
1
e N
N
N
1
M
M

= (4.22)
This moment has to be used in eq. (4.15). If the first
order moment is larger than M
Ed,
, then this value
40
should be used.
4.9 SPECIAL COMPOSITE MEMBERS
WITH HOLLOW SECTIONS
4.7.2 Biaxial bending and compression
The previous sections consider composite members
consisting of a hollow section at the outer side and
concrete inside. The concrete may be reinforced or
not. However, an alternative is to reinforce the
concrete with steel fibres instead of reinforcing bars
which provide advantages in the extension of the fire
resistance.
A composite member under biaxial bending and
compression has first to be examined for both axes
under uniaxial bending and compression, see Section
4.7.1. Additionally the combined situation has to be
verified. The influence of the imperfection is only
taken into account for the buckling axis which is most
critical.
41
The check can be expressed by the following
condition:
0 , 1
M
M
M
M
Rd , z , pl dz
Ed , z
Rd , y , pl dy
Ed , y
s
(4.23)
Other types of reinforcement used are solid sections
or another hollow section inside a circular or
rectangular hollow section with concrete in between.
Fig. 4.16 shows an example of a CHS with another
CHS member inside. Although many combinations are
possible, the design is in principle similar to that for
the reinforced concrete hollow section columns
described in the previous sections (Zhao et al., 2010).
The values
dy
and
dz
are determined at the level of
d
.
4.8 LOAD INTRODUCTION
In the design of composite columns, a full composite
action of the cross section is assumed. This means
that in the bond area no significant slip can occur
between the steel and the concrete. At locations of
load introduction, e.g. at beamcolumn connections,
this has to be verified. If no calculation is carried out,
the length of load introduction should be assumed to
be the minimum of 2b, 2d or /3, where b or d is the
minimum transverse dimension of the column, and
is the column length.
If the steel is not painted and is free of oil and rust, the
maximum bond stress, based on friction is (EN
199411, 2004):

Rd
= 0,55 N/mm
2
for CHS columns

Rd
= 0,40 N/mm
2
for RHS columns
The shear load transfer can be increased
considerably by shear connectors or steel
components, see Fig. 4.14.
For concentrated loads, a load distribution according
to Fig. 4.15 can be assumed. For such locally loaded
parts of encased concrete, higher design values for
the concrete strength can be used.
Table 4.1 Strength classes of concrete, characteristic cylinder strength and modulus of elasticity for
normal weight concrete
Strength class of concrete f
ck,cyl
/f
ck,cub
C20/25 C25/30 C30/37 C35/45 C40/50 C45/55 C50/60
Cylinder strength f
ck
(N/mm
2
) 20 25 30 35 40 45 50
Modulus of elasticity E
cm
(N/mm
2
) 30000 31000 33000 34000 35000 36000 37000
Note: The recommended values
a
= 1,0,
c
= 1,5 and
s
= 1,15 should be used to determine the design values.
Table 4.2 Limits for wall thickness ratios of concrete filled hollow sections for preventing local buckling
under axial compression (EN 199411, 2004)
Steel grade S235 S275 S355 S460
Rectangular hollow sections
eq. (4.9)
h/t 52,0 48,1 42,3 37,2
Circular hollow sections
eq. (4.10)
d/t 90,0 76,9 59,6 46,0
42
Fig. 4.1 Concrete filled hollow sections with notations
N
Ed
e/M
pl,Rd
1,00
0,75
0,50
0,25
0
0,75 0,50 0,25 0 1,00
S235 / C45
d = 500 mm
t = 10 mm
N
Ed
/N
pl,Rd
N
Ed
e/M
pl,Rd
1,00
0,75
0,50
0,25
0
0,75 0,50 0,25 0 1,00
S235 / C45
d = 500 mm
t = 10 mm
N
Ed
/N
pl,Rd
Fig. 4.2 Bearing capacity of a composite hollow section column
43
Fig. 4.3 Three dimensional confinement effect in concrete filled hollow sections
Fig. 4.4 Stress distribution for the bending resistance of a section
1,0
0,8
0,2
0,4
0,6
0
1,0 0,8 0,2 0,4 0,6 0 1,2 1,4
o = 0,45
0,40
0,275
0,25
0,225
0,20
0,30
0,9
0,8
0,7
0,6
0,5
,
,
N
Ed
/N
pl,Rd
M
Ed
/M
pl,Rd
Rd , pl
yd a
N
f A
parameter = o
0,35
1,0
0,8
0,2
0,4
0,6
0
1,0 0,8 0,2 0,4 0,6 0 1,2 1,4
1,0
0,8
0,2
0,4
0,6
0
1,0 0,8 0,2 0,4 0,6 0 1,2 1,4
o = 0,45
0,40
0,275
0,25
0,225
0,20
0,30
0,9
0,8
0,7
0,6
0,5
,
,
N
Ed
/N
pl,Rd
M
Ed
/M
pl,Rd
Rd , pl
yd a
N
f A
parameter = o
0,35
Fig. 4.5 Interaction curve for rectangular hollow sections with bending about the weak axis, b/h = 0,5
44
1,0
0,8
0,2
0,4
0,6
0
1,0 0,8 0,2 0,4 0,6 0 1,2 1,4
M
Ed
/M
pl,Rd
o = 0,45
0,40
0,35
0,275
0,25
0,225
0,20
0,9
0,8
0,7
0,6
0,5
,
1,0
0,8
0,2
0,4
0,6
0
1,0 0,8 0,2 0,4 0,6 0 1,2 1,4
0,9
0,8
0,7
0,6
0,5
N
Ed
/N
pl,Rd
,
Rd , pl
yd a
N
f A
parameter = o
0,30
1,0
0,8
0,2
0,4
0,6
0
1,0 0,8 0,2 0,4 0,6 0 1,2 1,4
1,0
0,8
0,2
0,4
0,6
0
1,0 0,8 0,2 0,4 0,6 0 1,2 1,4
M
Ed
/M
pl,Rd
o = 0,45
0,40
0,35
0,275
0,25
0,225
0,20
0,9
0,8
0,7
0,6
0,5
0,9
0,8
0,7
0,6
0,5
,
1,0
0,8
0,2
0,4
0,6
0
1,0 0,8 0,2 0,4 0,6 0 1,2 1,4
0,9
0,8
0,7
0,6
0,5
N
Ed
/N
pl,Rd
,
Rd , pl
yd a
N
f A
parameter = o
0,30
Fig. 4.6 Interaction curve for square hollow sections with b/h = 1,0
o = 0,45
0,40
0,275
0,25
0,225
0,20
0,9
0,8
0,7
0,6
0,5
1,0
0,8
0,2
0,4
0,6
0
N
Ed
/N
pl,Rd
M
Ed
/M
pl,Rd 1,0 0,8 0,2 0,4 0,6 1,2 1,4 0
,
Rd , pl
yd a
N
f A
parameter = o
0,30
0,35
o = 0,45
0,40
0,275
0,25
0,225
0,20
0,9
0,8
0,7
0,6
0,5
1,0
0,8
0,2
0,4
0,6
0
N
Ed
/N
pl,Rd
1,0
0,8
0,2
0,4
0,6
0
N
Ed
/N
pl,Rd
M
Ed
/M
pl,Rd 1,0 0,8 0,2 0,4 0,6 1,2 1,4 0
M
Ed
/M
pl,Rd 1,0 0,8 0,2 0,4 0,6 1,2 1,4
M
Ed
/M
pl,Rd 1,0 0,8 0,2 0,4 0,6 1,2 1,4 0
,
Rd , pl
yd a
N
f A
parameter = o
0,30
0,35
Fig. 4.7 Interaction curve for rectangular hollow sections with bending about the strong axis, h/b = 2,0
0,40
0,35
0,275
0,25
0,225
0,30
,
1,0
0,8
0,2
0,4
0,6
0
0,9
0,8
0,7
0,6
0,5
N
Ed
/N
pl,Rd
M
Ed
/M
pl,Rd
1,0 0,8 0,2 0,4 0,6 0 1,2 1,4
1,0
0,8
0,2
0,4
0,6
0
N
Ed
/N
pl,Rd
0,20
,
Rd , pl
yd a
N
f A
parameter = o
o = 0,45
0,40
0,35
0,275
0,25
0,225
0,30
,
1,0
0,8
0,2
0,4
0,6
0
0,9
0,8
0,7
0,6
0,5
N
Ed
/N
pl,Rd
M
Ed
/M
pl,Rd
1,0 0,8 0,2 0,4 0,6 0 1,2 1,4
1,0
0,8
0,2
0,4
0,6
0
N
Ed
/N
pl,Rd
1,0
0,8
0,2
0,4
0,6
0
N
Ed
/N
pl,Rd
0,20
,
Rd , pl
yd a
N
f A
parameter = o
o = 0,45
Fig. 4.8 Interaction curve for circular hollow sections
45
N
E,Rd
N
A,Rd
N
C,Rd
N
D,Rd
N
B,Rd
M
A,Rd
M
C,Rd
M
D,Rd
M
B,Rd
N
E,Rd
N
A,Rd
N
C,Rd
N
D,Rd
N
B,Rd
M
A,Rd
M
C,Rd
M
D,Rd
M
B,Rd
Fig. 4.9 Interaction curve approached by a polygonal connection of the points A to E
N
C,Rd
N
D,Rd
= 0,5N
C,Rd

N
E,Rd
M
E,Rd
N
D,Rd
= 0,5N
C,Rd
M
D,Rd
= M
max,Rd
M
C,Rd
= M
pl,Rd
N
C,Rd
M
B,Rd
= M
pl,Rd
N
pl,Rd
N
C,Rd
N
D,Rd
= 0,5N
C,Rd
N
C,Rd
N
D,Rd
= 0,5N
C,Rd

N
E,Rd
M
E,Rd
N
D,Rd
= 0,5N
C,Rd
M
D,Rd
= M
max,Rd
M
C,Rd
= M
pl,Rd
N
C,Rd
M
B,Rd
= M
pl,Rd
N
pl,Rd
Fig. 4.10 Stress distributions of selected positions of the neutral axis (points A to E)
46
Fig. 4.11 Reduction of the normal stresses due to shear
,
d
,
Rd , pl
Ed
N
N
Rd , pl
Ed
M
M
,
d
,
Rd , pl
Ed
N
N
Rd , pl
Ed
M
M
Fig. 4.12 Design for compression and uniaxial bending
M
Ed
r M
Ed
M
Ed
r M
Ed
Fig. 4.13 Relation between the end moments (1 s r s +1)
47
Fig. 4.14 Load introduction into hollow sections by inserted plates
1:2,5 1:2,5 1:2,5 1:2,5
Fig. 4.15 Load introduction in a composite column
0
d
t
0
concrete
outer tube
inner tube
d
i
t
i
Fig. 4.16 Tubeintube composite column concept
48
5. FIRE RESISTANCE OF
HOLLOW SECTION COLUMNS
5.1 INTRODUCTION
This chapter is a reduced version of CIDECT Design
Guide No. 4 (Twilt et al., 1994), however, updated
with the latest revisions to Eurocodes 3 and 4 on
structural fire design (EN 199312, 2005; EN
199412, 2005).
Unprotected structural hollow sections have an
inherent fire resistance of approximately 15 to 30
minutes. Traditionally, it was assumed that
unprotected steel members fail when they reach
temperatures of about 450 to 550 C. However, the
temperature at which a steel member reaches its
ultimate limit state depends on the massivity of the
section and the actual load level. If the service load
level of a column is less than 50% of its resistance,
the critical temperature rises to over 650 C, which, for
bare steel, means an increase in failure time of more
than 20%.
When hollow steel sections are required to withstand
extended amounts of time in fire, additional measures
have to be considered to delay the rise in steel
temperature.
5.1.1 External insulation of the steel
section
External insulation of the steel section is a type of fire
protection that can be applied to all kinds of structural
elements (columns, beams and trusses). The
temperature development in a protected hollow steel
section depends on the thermal properties of the
insulation material (conductivity), on the thickness of
the insulation material and on the section factor
(massivity) of the steel profile.
External fire protection materials can be grouped as
follows:
 Insulating boards (based mainly on gypsum or
mineral fibre or lightweight aggregates such as
perlite and vermiculite). If board protection is to be
used, care must be taken to ensure the integrity of
joints between the boards.
 Spray coating or plaster (based mainly on mineral
fibre or lightweight aggregates such as perlite and
vermiculite)
 Intumescent coatings (paintlike mixtures applied
directly to the steel surface which, in case of fire,
swell up to a multiple of their original thickness)
 Suspended ceilings (mainly protecting roofs,
trusses)
 Heat radiation shielding
In some countries, intumescent coatings are restricted
to a fire resistance of 30 or 60 minutes, but this
technology is rapidly developing and nowadays
considerably larger protection times are possible.
5.1.2 Concrete filling of the section
Usually, fire protection through concrete filling of the
section is applied to columns only. Filling hollow
sections with concrete is a very simple and attractive
way of enhancing fire resistance. The temperature in
the unprotected outer steel shell increases rapidly.
However, as the steel shell gradually loses strength
and stiffness, the load is transferred to the concrete
core.
Apart from the structural function, the hollow section
also acts as a radiation shield to the concrete core, in
combination with a steam layer between the steel and
the concrete core.
Depending on the fire resistance requirements, the
concrete in the hollow section can be plain concrete
(fire resistance 30 minutes up to 60 minutes) or
concrete with reinforcing bars or steel fibres. New
research aimed at increasing the fire resistance of
concrete filled hollow sections is focused on the use of
high strength concrete.
5.1.3 Water cooling
Water cooling is a type of fire protection that can be
applied to all kinds of hollow sections, but is mostly
used for columns. The hollow section acts both as the
load bearing structure and as the water container.
This protection system is quite sophisticated; it needs
a thorough design and proper hydraulic installations.
The cooling effect consists of the absorption of heat
by water, the removal of heat by water circulation and
its consumption in the vaporization of water. In
practical applications, these effects are combined. A
suitably designed water filled system will limit the
average steel temperature to less than 200 C.
Two different systems can be used: permanently filled
elements or elements filled only when a fire breaks
out. In the latter case, protection depends on a fire
49
detection system and a short water filling time. In
unreplenished systems, the attainable fire resistance
time depends on the total water content (including any
reservoir tank) and on the shape of the heated
structure. In systems where the water is constantly
renewed, the fire resistance is unlimited. Water
cooling by natural flow is mainly used for vertical or
inclined elements in order to ensure the circulation of
the water.
5.2 FIRE RESISTANCE
5.2.1 Concept
Fire safety precautions are specified with the intent of
avoiding any casualties and reducing economic fire
damage to an acceptable level. As far as building
construction is concerned, it is important that the
construction elements can withstand a fire for a
specified amount of time. In this respect, one should
bear in mind that the strength and deformation
properties of the commonly used building materials
deteriorate significantly at the temperatures that may
be expected under fire conditions. Moreover, the
thermal expansion of most building materials is
considerable. As a result, the structural elements and
assemblies may deform or even collapse when
exposed to fire conditions.
The amount of time that a construction element can
resist a fire largely depends on the anticipated
temperature development of the fire itself. This
temperature development depends, among other
aspects, on the type and amount of combustible
materials present, expressed in terms of kg of wood
per m
2
floor surface and called the fire load density,
see Fig. 5.1, and on the fire ventilation conditions.
In practical fire safety design, however, it is
conventional to use a socalled "standard fire curve",
defined in ISO 8341 (1999), which is more or less
representative for post flashover fires in buildings
with relatively small compartments, such as apartment
buildings and offices. Alternative standard fire curves,
with small differences from the ISOcurve, are in use
in the USA.
The amount of time a building component is able to
withstand heat exposure according to the standard fire
curve, is called the "fire resistance". In order to be
able to determine the fire resistance of a building
component, proper performance criteria have to be
determined. These criteria are defined in relation to
the anticipated function of the respective building
element during fire. For more details, see Twilt et al.
(1994).
For building components such as columns, with a load
bearing function, the only relevant performance
criterion is "stability".
As far as the determination of the fire resistance is
concerned, there are basically two possibilities: an
experimental approach and an analytical approach.
The experimental approach, i.e. the determination of
the fire resistance of columns based on standard fire
tests, is the traditional approach. Although employing
different national testing procedures, the concept of
fire testing is, by and large, the same in the various
countries.
The analytical approach is the modern approach and
has become possible by the development of computer
technology. On an international level, calculation rules
for the fire resistance of both steel and composite
steel concrete columns, including concrete filled
hollow section columns, are available. The analytical
approach offers significant advantages, when
compared with the experimental one.
Important factors influencing the fire resistance of
columns are:
 Load level
 Shape and size of the cross section
 Buckling length
 Concrete filling and reinforcement
Bare steel columns (i.e. hollow section columns
without external protection or concrete filling) possess
only a limited fire resistance. Depending on the load
level and the section factor (massivity), a fire
resistance of 15 to 20 minutes is usually attainable. A
30 minutes fire resistance can only be achieved in
more exceptional cases. This situation may be
dramatically improved by applying thermal insulation
to the column. Depending on the type and thickness
of the insulation material, fire resistances of many
hours can be achieved, although most requirements
today are limited to 120 minutes.
Hollow section columns filled with concrete have a
much higher load bearing capacity and a higher fire
resistance than unprotected, empty hollow section
columns. Provided the concrete is of good quality
(over, say, a crushing strength of 20 N/mm
2
) and the
cross sectional dimensions are not too small (not less
than 150 x 150 mm), a fire resistance of at least 30
minutes will be achieved. Sections with larger
50
dimensions will have a higher fire resistance and by
adding additional reinforcement to the concrete, the
fire resistance may be increased to over 120 minutes.
Infinite fire resistance can be achieved by water filling,
provided an adequate water supply is available.
Improved fire performance of hollow section columns
can also be achieved by placing the columns outside
the building envelope an expedient sometimes used
for architectural purposes. By preventing direct flame
impingement on the member, the need for additional
fire protection measures can be significantly reduced
or even become unnecessary.
Since fire safety requirements for columns are
normally expressed in terms of the fire resistance to
be attained, this emphasizes the need to consider the
fire resistance requirements from the beginning in a
structural design project.
5.2.2 Requirements
Fire safety in buildings is based on achieving two
fundamental objectives:
 Reducing the loss of life
 Reducing the property or financial loss in, or in the
neighbourhood of, a building fire
In most countries, the responsibility for achieving
these objectives is divided between the government or
civic authorities who have the responsibility for life
safety via building regulations, and the insurance
companies dealing with property loss through their fire
insurance policies.
The objectives of fire safety may be achieved in
various ways. For example:
 By eliminating or protecting possible ignition
sources (fire prevention).
 By installing an automatic extinguishing device, in
order to prevent the fire from growing into a severe
fire (operational or active measures, e.g. sprinklers).
 By providing adequate fire resistance to the building
components using passive measures to prevent fire
spreading from one fire compartment to adjacent
compartments.
Often a combination of the above measures is
applied.
Requirements with regard to fire resistance clearly
belong to the passive measures. To date, the use of a
conventional fire scenario employing the ISO standard
fire curve (ISO 8341, 1999) is common practice in
Europe and elsewhere. The standard fire test is not
intended to reflect the temperatures and stresses that
would be experienced in real fires, but provides a
measure of the relative performance of elements of
structures and materials within the capabilities and
dimensions of the standard furnaces. In general,
uncertainties about structural behaviour in real fires
are taken into account by making conservative fire
resistance requirements.
Required safety levels are specified in Codes and
normally depend on factors like:
 Type of occupancy
 Height and size of the building
 Effectiveness of fire brigade action
 Active measures, such as vents and sprinklers
An overview of fire resistance requirements as a
function of the number of storeys and representative
for European countries is given in Table 5.1.
The following general features may be identified:
 No specified fire resistance requirements for
buildings with limited fire load density (say, 1520
kg/m
2
) or where the consequences of collapse of
the structure are acceptable.
 Fire resistance for a specified but limited amount of
time, where the time requirement is mainly intended
to allow for safe evacuation of the occupants and
intervention by rescue teams.
 Extended fire resistance of the main structure to
ensure that the structure can sustain a full burn out
of combustible materials in the buildings or a
specified part of it.
Sometimes unprotected steel may be sufficient, for
example for situations where safety is satisfied by
other means (e.g. sprinklers) and/or if requirements
with respect to fire resistance are low (i.e. not over,
say, 30 minutes).
A full fire engineering approach (Natural Fire
Concept), in which compartment and steel
temperature are calculated from a consideration of the
combustible material present, compartment geometry
and ventilation, is nowadays more accepted and has
shown considerable savings in fire protection costs in
specific cases.
5.2.3 Performance criteria
The fundamental concept behind all methods
designed to predict structural stability in fire is that
51
construction materials gradually lose strength and
stiffness at elevated temperatures. The reduction in
the yield strength of structural steel and the
compression strength of concrete with increasing
temperature according to Eurocode 3 (EN 199312)
and Eurocode 4 (EN 199412) is given in Fig. 5.2. It
shows that there is not much difference in the relative
reduction in strength of concrete and steel under high
temperatures. The reason for the difference in the
structural behaviour of steel and concrete elements
under fire conditions is that heat propagates about 10
to 12 times faster in a steel structure than in a
concrete structure of the same massivity, because the
thermal conductivity of steel is higher than that of
concrete.
Normally, the fire resistance design of structures is
based on a similar design approach as used for
design under ambient temperature. In a multistorey
braced frame, the buckling length of each column at
room temperature is usually assumed to be the
column length between floors. However, such
structures are usually compartmented and any fire is
likely to be limited to one storey. Therefore, any
column affected by fire will lose its stiffness, while
adjacent members will remain relatively cold.
Accordingly, if the column is rigidly connected to the
adjacent members, builtin end conditions can be
assumed in the event of fire. Investigations by Twilt &
Both (1991) showed that in the case of fire on one
storey the buckling length of columns in braced
frames is reduced to between 0,5 and 0,7 times the
column length, depending on the boundary conditions,
see Fig. 5.3.
There is an increasing tendency toward assessing the
fire resistance of individual members or
subassembles by analytical fire engineering. The
Eurocodes on structural fire design (EN 199312,
2005; EN 199412, 2005) define three levels of
assessments:
 Level 1: Tabulated data
 Level 2: Simple calculation models
 Level 3: Advanced calculation models
"Advanced calculation models" is the most
sophisticated level. Such calculation procedures
include a complete thermal and mechanical analysis
of the structure and use the values for the material
properties given in the Eurocodes. General calculation
methods enable real boundary conditions to be
considered and take account of the influence of
nonuniform temperature distribution over the section
and therefore lead to more realistic failure times and,
consequently, to the most competitive design.
However, the handling of the necessary computer
programs is quite time consuming and requires expert
knowledge.
For practising engineers and architects not
accustomed to handling specialised computer
programs, "Simple calculation models" have been
developed, which lead to a comprehensive design, but
are limited in application range. They use
conventional calculation procedures and normally
provide adequate accuracy.
"Tabulated data", which provide solutions on the safe
side and allow fast design for restricted application
ranges, forms the lowest level of assessment.
In the following chapters, only the principles of the
design tables and the simple calculation procedures
are shown. In all equations, the recommended
M
and
M,fi
factors in the Eurocodes for steel and concrete
are 1,0.
5.3 UNFILLED HOLLOW SECTION
COLUMNS
The simple calculation rules for the critical
temperature of steel columns discussed hereafter,
hold for classes 1, 2 and 3 cross sections only (as
defined in Eurocode 3 (EN 199311, 2005)) and can
be applied both to protected and unprotected
columns. For columns with a class 4 cross section, a
default value for the critical temperature of 350 C is
to be used.
The critical temperature of an axially loaded steel
column depends on the load N
fi,Ed
which is present
during a fire and the buckling resistance N
Rd
at room
temperature (N
fi,Ed
=
fi,t
N
Rd
).
Note: Instead of the symbols E
fi,d,t
and R
fi,d,t
used in
Eurocode 3 (EN 199312, 2005), for consistency the
symbols N
fi,Ed
and N
fi,Rd
refer here to the design load
and the design resistance in the fire situation.
As a simplification for the calculation, the load N
fi,Ed
which is present during a fire may also be related by
the ratio
fi
to the design load N
Ed
at room
temperature. This ratio depends on the load ratio
between variable and permanent loading, but as a
simplification, generally a ratio
fi
= 0,65 may be
assumed. Only in cases of areas susceptible to
accumulation of goods, the recommended value is
0,70.
52
Hence, generally:
Ed Ed fi Ed , fi
N 65 , 0 N N = q = (5.1)
where:
N
fi,Ed
design load in the fire situation
N
Ed
design load at room temperature
fi
reduction factor for N
Ed
to obtain N
fi,Ed
Now, the following verification has to be carried out:
Rd , pl , y fi Rd , fi Ed , fi
N k N N
u
_ = s (5.2)
where:
N
fi,Ed
design load in the fire situation
N
fi,Rd
buckling resistance in the fire situation
N
pl,Rd
compression resistance of gross cross section
at room temperature
fi
reduction factor for flexural buckling in the fire
design situation, see Fig. 5.4
k
y,
reduction factor for the yield strength of steel at
a steel temperature
a
, see Fig. 5.5
This check is, in principle, similar to that for members
in compression at room temperature. Only the
buckling coefficient at room temperature has to be
replaced by
fi
at temperature
a
in the fire situation
and the yield strength f
yd
at room temperature by k
y,
f
yd
in the fire situation. The load factors are usually
1,0.
For unprotected steel sections, it can be shown that
for standard fire exposure the temperature
development of a steel section depends only on the
relative geometry of the profile. This effect is taken
into account by means of the shape factor, A
m
/V,
where:
A
m
exposed surface area of the member per unit
length [m
2
/m]
V volume of the member per unit length [m
3
/m]
This ratio is equal to the exposed steel perimeter/steel
cross section.
The curves presented in Fig. 5.6 illustrate the effect of
the section factor on the temperature development of
an unprotected steel section when exposed to
standard fire conditions. For commonly used I
sections, shape factors are within a range of, say, 50
to 400 m
1
. For hollow sections exposed to heat from
all sides, the section factor may be approximated by:
A
m
/V = Perimeter/(Perimeter x thickness t) = 1/t [m
1
]
with t being the thickness of the steel hollow section.
For a practical range of the hollow section thickness of
20 to 2.5 mm, this also leads to section factors varying
between 50 and 400 m
1
. However, for equivalent
cross sections, hollow sections have an A
m
/V ratio
which is about 60% of that of comparable open
sections.
For any given critical steel temperature, the fire
resistance of an unprotected steel element
assuming standard fire conditions depends only on
its section factor as illustrated in Fig. 5.7. In many
practical situations, the critical temperature of a steel
member will be approximately 550 C. This figure
shows that an unprotected steel member with a
section factor smaller than approximately 40 m
1
may
have a fire resistance of 30 minutes or beyond.
If external fire insulation is provided, the steel
temperature development depends not only on the
section factor, but also on the type and thickness of
the insulation material.
5.4 CONCRETE FILLED HOLLOW
SECTION COLUMNS
5.4.1 Unprotected columns thermal and
mechanical response
Because of their different locations in the cross
section, the various components of a concrete filled
hollow section column will each have different
timedependent strength reduction characteristics.
The unprotected, directly exposed steel shell will be
rapidly heated and will show a significant strength
reduction within a short time.
The concrete core with its high massivity and low
thermal conductivity will, for some time, maintain a
significantly high proportion of its strength, mostly in
the core area rather than near the surface.
Reinforcement, if used, is normally placed near the
surface, but is protected by typically 2050 mm of
concrete cover. For this reason, it will have a retarded
strength reduction. Fig. 5.8 demonstrates this
characteristic behaviour and describes the fire
performance of the various components of concrete
filled hollow section columns.
The load bearing capacity R of a cross section is the
sum of the load bearing capacities of each of its
53
components r
j
. Under fire conditions, all component
capacities are dependent on the fire endurance time t.
R(t) = r
j
(t) (5.3)
In room temperature design, the steel shell is likely to
be the dominant load bearing component because of
the high strength of the steel and the location of the
profile. However, after a fire time t
1
, only a small
percentage of the original load bearing capacity of the
steel shell can still be activated. This means that in
the case of fire the main part of the load carried by the
steel section will be redistributed to the concrete core,
which loses strength and stiffness more slowly than
the steel section. Therefore:
 The load bearing capacity of the steel shell should
be minimised, which means thin shell thickness and
low steel grade.
 The load bearing capacity of the concrete core
should be optimised, which means higher concrete
strength and reinforcement.
Since the strength reduction of the components is
directly affected by the heating characteristic of the
cross section, a minimum column cross sectional
dimension is often necessary to fulfill a required fire
resistance.
With increasing temperature, the strength and
Young's modulus decrease. Thus, the load bearing
capacity of a structural member decreases with time,
while its deformation increases. In practical fire
design, the influence of the column slenderness has
also to be taken into account.
5.4.2 Assessment methods for
unprotected columns
Levels of assessment
As already explained in Section 5.2.3, Eurocodes 3
and 4 on structural fire design (EN 199312, 2005;
EN 199412, 2005) define three different levels of
assessment. This chapter deals with design
information at levels 1 and 2, i.e. "Tabulated data" and
"Simple calculation models". For more general
calculation models, see Twilt et al. (1994).
Level 1: Tabulated data
The fire rating of unprotected concrete filled hollow
section columns may be classified according to Table
5.2 as a function of:
 The load level
fi,t
 The cross section size (b, h or d)
 The amount (%) of reinforcement (A
s
/(A
c
+ A
s
) 100)
 The minimum cover of the reinforcing bars (u
s
)
The slenderness relation is indirectly incorporated in
the load level
fi,t
which is given by:
fi,t
= N
fi,Ed
/N
Rd
(5.4)
where:
N
fi,Ed
design load in the fire situation
N
Rd
buckling resistance at room temperature
N
Rd
is calculated according to the room temperature
procedures given in Eurocode 4 (EN 199411, 2004).
However, the following limitations apply:
 Irrespective of the actual steel grade, the yield
strength of the hollow sections for fire calculations is
limited to a maximum of 235 N/mm
2
.
 The wall thickness of the steel is limited to a
maximum of 1/25 of the cross sectional dimension
d, b or h.
 Reinforcement ratios higher than 3% are not taken
into account.
 The values given in Table 5.2 are valid for steel
grade S500 used for the reinforcement A
s
.
Level 2: Simple calculation models
At level 2, a computer program has been developed to
model the fire performance of concrete filled hollow
section columns (Grandjean et al., 1980; Twilt & Van
de Haar, 1986). This software tool is based on Annex
H (informative) of Eurocode 4 (EN 199412) for
concrete filled hollow section columns at elevated
temperatures. However, this program will be modified,
because comparisons with other programs revealed
discrepancies on the nonconservative side.
Design charts have been developed in which, for a
standard fire exposure of 30, 60, 90 and 120 minutes,
the design axial buckling load N
fi,Rd
of concrete filled
hollow section columns is given as a function of the
buckling length
s 4,5m
 Width (b or h) or diameter (d): 140 s (b, h or d) s
400 mm
 Concrete strength minimum C20/25 and maximum
C40/50
 Reinforcement ratio s 5%
 Fire resistance R s 120 minutes
For a given buckling length and loading, the fire
54
resistance of concrete filled hollow section columns
mainly depends on the cross sectional dimensions,
the concrete quality and the reinforcement, if any. By
a proper choice of these parameters, practically any
fire resistance can be achieved. If no reinforcement is
used, a fire resistance of 30 minutes can normally be
achieved; 60 minutes, however, is not attainable
unless the load level is significantly decreased. As a
result, the design charts focus on reinforced hollow
section columns with fire resistances of 60 minutes
and more.
The effect of small eccentricities (M/N s 0,5(d or b))
can be taken into account by increasing the axial load
N
fi,Ed
to an equivalent axial load N
equ
:
e s
Ed , fi
equ
N
N
= (5.5)
where:
s
correction coefficient related to the reinforcement,
see Fig. 5.11.
e
correction coefficient related to the eccentricity e,
see Fig. 5.12.
Note: There are some concerns about the Annex H
method. Wang & Orton (2008) pointed out that this
method is rather antiquated. An alternative method,
developed by Wang & Orton (2008), is based on the
well established cold design method for composite
columns in the main part of Eurocode 4 (EN 199411,
2004), but modified to take into account strength and
stiffness degradations of steel and concrete at high
temperatures. A design software package named
"Firesoft" is now available to assist designers, which
has been verified by Wang & Orton (2008).
5.4.3 Technological aspects
Small vent holes (10 to 15 mm diameter) are required
in the hollow section walls, usually in pairs. Such
holes must be provided for each storey length at each
floor level, with a maximum distance of 5,0 m.
between pairs. They must be placed between 100 and
120 mm from each column end. Those holes are
intended to prevent the bursting of the column under
steam pressure from the heating of entrapped water in
the enclosed concrete.
Besides the standard hollow section cross sections, in
the past, a variety of different cross section designs
were developed and successfully applied in building
projects. They are all based either on combinations of
hollow sections (tube inside tube) or on combinations
of hollow sections with other steel sections. The
advantages of such special cross section types are an
increased load bearing capacity without the need to
increase the outer cross sectional dimensions, or
reduced dimensions for a given load capacity.
To fulfill architectural requirements, special steels,
such as weathering steel, can also be used for the
hollow sections of the columns.
Careful design of the top and bottom of a single
column or at the connection of a continuous column is
necessary to ensure that the loading is introduced into
the composite cross section in a proper way.
5.4.4 Externally protected concrete filled
hollow section columns
If an extended fire resistance is desired, in
combination with a high load level and/or a minimised
column cross section, it may be necessary to apply
conventional external protection to a concrete filled
hollow section column.
5.5 WATER FILLED HOLLOW
SECTION COLUMNS
5.5.1 Basic principles
Water filling using natural circulation provides a safe
and reliable fire protection method for hollow section
columns, provided that two conditions are satisfied
(Hnig et al., 1985):
 The system is self activating in fire.
 The system is self controlling.
In a properly designed system, the natural circulation
will be activated when the columns are locally heated
by a fire. The density of warm water is lower than that
of cold water, which produces the pressure
differentials that activate the natural circulation. The
effect will be intensified when localised boiling
commences and steam is formed. As the fire
develops, the rate of steam production will also
increase, thus forcing the cooling effect obtained by
naturally activated circulation.
The following methods of permanent water filling are
available:
1. Unreplenished columns
Simply filling a column with water, with no provision
for replacing any water lost through steam production,
55
will lead to an increased, but limited fire resistance
compared to that of the empty column. In multistorey
columns the water in the topstoreycolumns will be
first evaporated, but the fire resistance can be
increased by externally protecting the top storey
length and using it as a reservoir for the lower storeys.
Heavy steam production may lead to an additional
critical loss of water. Therefore, unreplenished
columns should be used only for lower fire resistance
requirements, up to, say, 60 minutes.
2. Columns with external pipe
This system has a connecting down pipe between the
bottom and top of the columns. The lighter, upward
flowing watersteam mixture must be separated at the
top, so that the water can return down through the
pipe to the bottom. In this manner, an external
naturally forced circulation will be activated. In
addition, the pipe can be connected with a water
storage tank at the top of the building to replace the
water lost from steam production and possibly act as
a common water/steam separating chamber. A group
of individual columns can be connected at their
bottom to a shared connecting pipe as well as with a
connecting pipe at the top. For such a group of
columns, only one down pipe is necessary,
connecting top and bottom of the whole group, see
Fig. 5.13a.
3. Columns with internal pipe
In this system, an internal down tube is used within
each column to provide a supply of cool water to the
bottom of each column. This promotes the internal,
naturally activated circulation of the upward flowing
watersteam mixture and the down flowing water after
steam separation. Thus, each column acts as an
individual member without any connection to the other
columns.
To minimise the number of water storage tanks, the
tops of several columns can be connected by a
common pipe leading to one storage tank for the
whole group, see Fig. 5.13b.
4. Mixed systems
The above mentioned systems can be mixed within a
building and they can be connected to act as a mixed
integrated system. This can be advantageous for
structures containing not only columns, but also water
filled diagonals for bracing, etc.
In the naturally circulating systems described above, a
minimum declination of the diagonals of about 45 is
recommended.
It is not advisable to use any electromechanical
installation, such as pumps, acting against the
naturally produced circulation. This may lead to a
failure of the cooling system and thus to a collapse of
the water filled structure.
5.5.2 Assessment methods
A careful design is necessary to ensure the positive
behaviour of a water filled hollow section column
system. Two main criteria must be fulfilled to ensure
the cooling effect:
 Natural circulation of the water is maintained.
 Water losses due to steam production are replaced.
The mass of the water cooled steel structure as well
as the water within the system can be taken into
account when calculating the time of commencement
of boiling. The loss of water mass by evaporation has
to be estimated only for the time difference between
the start of boiling and the required fire resistance
time. For the characteristic thermal behaviour, refer to
Fig. 5.14.
The maximum temperature reached by the steel can
be estimated from the boiling temperature of the water
filling. The boiling temperature itself depends on the
hydraulic water pressure, i.e. the static head. In
addition, there will be a temperature gradient across
the wall of the hollow section, which will lead to a
slight increase of the temperature of the steel surface
directly exposed to the fire. However, the maximum
external steel surface temperature will normally not
reach a value high enough to significantly affect the
mechanical properties of the steel.
5.6 JOINTS
5.6.1 Unfilled hollow section columns
Normally, the joints of both protected and unprotected
steel structures have a lower local section factor than
the adjacent members and will therefore attain lower
steel temperatures. However, if the section factors are
higher than for the connected members, they may be
the critical elements and the behaviour under fire
conditions has to be considered, especially if catenary
action may occur (Wang & Ding, 2009). When bolted
joints are used for insulated steel members, care must
be taken to ensure that the bolt heads and nuts are as
well protected as the cleat. Normally, this will lead to a
local increase of insulation thickness.
56
57
5.6.2 Concrete filled hollow section
columns
The joints should be designed based on the principle
that the loads can be transferred from the beams to
the columns in such a way that all structural
components structural steel, reinforcement and
concrete contribute to the load bearing capacity
according to their strength. This can be done with a
through plate shown in Fig. 4.14 or with fin plates in
combination with an additional connection between
the hollow section and the concrete e.g. by a through
pin.
Table 5.1 Variations in required fire resistance
Type of building Requirements Fire class
One storey None or low Possibly up to R30
2 or 3 storeys None up to medium Possibly up to R60
More than 3 storeys Medium R60 to R120
Tall buildings High R90 and more
Table 5.2 Minimum crosssectional dimensions, reinforcement ratios and axis distances of the
reinforcing bars for fire resistance classification for various degrees of utilisation levels
fi,t
Standard fire resistance
steel section: b/t > 25 and d/t > 25
A
s
A
c
t
u
s
u
s
u
s
t
b d
h
steel section: b/t > 25 and d/t > 25
A
s
A
c
t
u
s
u
s
u
s
t
b d
h
R30 R60 R90 R120 R180
1 Minimum crosssectional dimensions for load level
fi,t
s 0,28
1.1
1.2
1.3
Minimum dimensions h and b or minimum diameter d (mm)
Minimum ratio of reinforcement A
s
/(A
c
+A
s
) in %
Minimum axis distance of reinforcing bars u
s
(mm)
160
0

200
1,5
30
220
3,0
40
260
6,0
50
400
6,0
60
2 Minimum crosssectional dimensions for load level
fi,t
s 0,47
2.1
2.2
2.3
Minimum dimensions h and b or minimum diameter d (mm)
Minimum ratio of reinforcement A
s
/(A
c
+A
s
) in %
Minimum axis distance of reinforcing bars u
s
(mm)
260
0

260
3,0
30
400
6,0
40
450
6,0
50
500
6,0
60
3 Minimum crosssectional dimensions for load level
fi,t
s 0,66
3.1
3.2
3.3
Minimum dimensions h and b or minimum diameter d (mm)
Minimum ratio of reinforcement A
s
/(A
c
+A
s
) in %
Minimum axis distance of reinforcing bars u
s
(mm)
260
3,0
25
450
6,0
30
550
6,0
40






Note: In Eurocode 4 (EN199412), the thickness of the hollow section "t" is called "e".
58
Time (min.) Time (min.)
1,0
0
0,6
0,4
0,2
0,8
1,0
0
0,6
0,4
0,2
0,8
Fig. 5.1 Natural fire curves and the ISO standard fire
curve (ISO 8341, 1999)
Fig. 5.2 Schematic material strength reduction for
structural steel and concrete
Rigid
core
Fire exposed
column
(a) Section through
the building
(b) Room
temperature
(c) Elevated
temperature
Mode of deformation
top floor :
other floors :
k
0.7
k 0.5
For fire conditions
in Europe e.g. :
For f ire conditions
in Europe e.g.:
top f loor :
= 0,7
other f loors :
= 0,5
Fig. 5.3 Schematic structural behaviour of columns in braced frames
59
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
rel
. (u
a
)
_
f
i
S355
S500
S235
ENV 1993
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
rel
. (u
a
)
_
f
i
S355
S500
S235
ENV 1993
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
rel
. (u
a
)
_
f
i
S355
S500
S235
ENV 1993
0,0 0,5 2,5 1,5 1,0 2,0 3,0
0,8
1,0
0,4
0,6
0,2
0,0
S500
S355
S235
(u
a
)
_
f
i
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
rel
. (u
a
)
_
f
i
S355
S500
S235
ENV 1993
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
rel
. (u
a
)
_
f
i
S355
S500
S235
ENV 1993
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
rel
. (u
a
)
_
f
i
S355
S500
S235
ENV 1993
0,0 0,5 2,5 1,5 1,0 2,0 3,0
0,8
1,0
0,4
0,6
0,2
0,0
S500
S355
S235
(u
a
)
_
f
i
Fig. 5.4 Reduction factor
fi
for flexural buckling in the fire situation for a particular critical temperature
a
(for comparison, the previous ENV 1993 lower bound curve is also shown)
1,0
0,8
0,6
0,4
0,2
0
Reduction factor k
y,u
0
0,2
0,4
0,6
0,8
1
0 200 400 600 800 1000 1200
steel temperature u
a
r
e
d
u
c
t
i
o
n
f
a
c
t
o
r
k
y
,
u
1,0
0,8
0,6
0,4
0,2
0
Reduction factor k
y,u
0
0,2
0,4
0,6
0,8
1
0 200 400 600 800 1000 1200
steel temperature u
a
r
e
d
u
c
t
i
o
n
f
a
c
t
o
r
k
y
,
u
Fig. 5.5 Reduction factor k
y,
for the yield strength of steel at a steel temperature
a
60
Time (min.)
S
t
e
e
l
t
e
m
p
e
r
a
t
u
r
e
u
a
(
o
C
)
Time (min.)
S
t
e
e
l
t
e
m
p
e
r
a
t
u
r
e
u
a
(
o
C
)
Section factor A
m
/V (m
1
)
u
a
T
i
m
e
t
o
r
e
a
c
h
u
a
(
m
i
n
.
)
Section factor A
m
/V (m
1
)
Section factor A
m
/V (m
1
) Section factor A
m
/V (m
1
)
u
a
T
i
m
e
t
o
r
e
a
c
h
u
a
(
m
i
n
.
)
Section factor A
m
/V (m
1
)
Fig. 5.6 Calculated temperature development in an
unprotected steel section as a function of
the shape factor
Fig. 5.7 Time for an unprotected steel section to reach
a given mean temperature under standard
conditions as a function of the section factor
Fig. 5.8 Typical strength reduction characteristics of
the various components of a concrete filled
hollow section column
61
tt
Fig. 5.9 Examples of buckling curves for CHS exposed to fire
ttt
t
Fig. 5.10 Examples of buckling curves for RHS exposed to fire
62
0,0
0,4
1,0
0,9
1,0
0,7
0,8
0,6
0,5
5,0 0,5 3,0 1,5 2,5 2,0 3,5 4,5 4,0
0,0
0,4
1,0
0,9
1,0
0,7
0,8
0,6
0,5
5,0 0,5 3,0 1,5 2,5 2,0 3,5 4,5 4,0 0,0
0,4
1,0
0,9
1,0
0,7
0,8
0,6
0,5
5,0 0,5 3,0 1,5 2,5 2,0 3,5 4,5 4,0 0,0
0,4
1,0
0,9
1,0
0,7
0,8
0,6
0,5
5,0 0,5 3,0 1,5 2,5 2,0 3,5 4,5 4,0
0,0
0,4
1,0
0,9
1,0
0,7
0,8
0,6
0,5
5,0 0,5 3,0 1,5 2,5 2,0 3,5 4,5 4,0
0,0
0,4
1,0
0,9
1,0
0,7
0,8
0,6
0,5
5,0 0,5 3,0 1,5 2,5 2,0 3,5 4,5 4,0 0,0
0,4
1,0
0,9
1,0
0,7
0,8
0,6
0,5
5,0 0,5 3,0 1,5 2,5 2,0 3,5 4,5 4,0 0,0
0,4
1,0
0,9
1,0
0,7
0,8
0,6
0,5
5,0 0,5 3,0 1,5 2,5 2,0 3,5 4,5 4,0
Fig. 5.11 Correction coefficient
s
as a function of the reinforcement (in %)
/b or
/d
/b or
/d
e/b or e/d
/b or
/d
/b or
/d
e/b or e/d
0,4
0,9
1,0
0,7
0,8
0,6
0,5
0,3
0,0 0,10 0,50 0,30 0,20 0,40
e
/b or
/d
Note: In Eurocode 4 (EN 199412), the eccentricity "e" is called "".
Fig. 5.12 Correction coefficient
e
as a function of the eccentricity e
63
Fig. 5.13 Options for columns with external and internal pipes
Time (min.) Time (min.)
Fig. 5.14 Typical temperature development in a water filled hollow section column, exposed to standard fire
conditions
64
6. DESIGN OF HOLLOW
SECTION TRUSSES
6.1 TRUSS CONFIGURATIONS
Various types of trusses are used in practice, see Fig.
6.1. Trusses made of hollow sections should be
designed in such a way that the number of joints and
thus fabrication is minimised. This means that due to
the lower number of joints, a Warren type truss with K
joints (Fig. 6.1a) is preferred to a Pratt type truss with
N joints (Fig. 6.1b).
Vierendeel girders (Fig. 6.1c) are mainly used in those
cases where architectural or functional aspects
require that no diagonals are used.
Trusses are characterised by their length L, depth h,
geometry and the distance between the joints.
The depth is normally related to the span, being about
1/10 to 1/16 L. Considering total economy of a hall
with all costs, a depth of 1/15 L is a common choice.
Whenever feasible, the joints are located at the load
application points, e.g. at the purlin locations.
6.2 JOINT CONFIGURATIONS
6.2.1 Joint classification
Depending on the type of truss, various types of joints
are used (Fig. 6.2), i.e. X, T, Y, N, K or KT. Although
here the designation X, T, Y, etc. is related to the
configuration, the classification is determined by the
loading, see Fig. 6.3. The basic joint types can be
defined as follows:
(a) When the force component normal to the chord in
a brace member is equilibrated by beam shear
(and bending) in the chord member, the joint is
classified as a T joint when the brace is
perpendicular to the chord, otherwise it is
classified as a Y joint.
(b) When the force component normal to the chord in
a brace member is essentially equilibrated (within
20%) by loads in other brace member(s) on the
same side of the joint, the joint is classified as a K
joint. The relevant gap is, in principle, between the
primary brace members whose loads equilibrate.
An N joint is to be considered as a type of K joint
with one brace at 90
o
.
(c) When the force component normal to the chord is
transmitted through the chord member and is
equilibrated by brace member(s) on the opposite
side, the joint is classified as an X joint.
When brace members transmit part of their load as K
joints and part of their load as T, Y, or X joints, the
adequacy of each brace needs to be determined by
linear interaction of the proportion of the brace load
involved in each type of load transfer. However, the
effect of the chord loading should be added to the joint
type with the most unfavourable chord load function.
One K joint, in Fig. 6.3b, illustrates that the brace
force components normal to the chord member may
differ by as much as 20% and still be deemed to
exhibit K joint behaviour. This is to accommodate
slight variations in brace member forces along a
typical truss, caused by a series of panel point loads.
The N joint in Fig. 6.3c, however, has a ratio of brace
force components normal to the chord member of 2:1.
In this case, that particular joint needs to be analysed
as both a "pure" K joint (with balanced brace forces)
and an X joint (because the remainder of the diagonal
brace load is being transferred through the joint), as
shown in Fig. 6.4. For the diagonal tension brace in
that particular joint, one would need to check that:
0 , 1
resistance joint X
N 0,5
resistance joint K
N 0,5
s +
If the gap size in a gapped K (or N) joint becomes
large and exceeds the value permitted by the gap
limit, then the "K joint" should also be checked as two
independent Y joints.
In X joints such as Fig. 6.3e where the braces are
close or overlapping, the joint should be checked as
an X joint, considering both brace load components
perpendicular to the chord.
In K joints such as Fig. 6.3d, where a brace has very
little or no loading, the joint can be treated as a Y joint,
as shown.
6.2.2 Terminology and notation
In principle, the terminology adopted by CIDECT and
IIW to define joint parameters, is used wherever
possible. However, to be consistent with the notation
in Chapters 2 to 5, for the resistance, the Eurocode 3
notations N
i,Rd
and M
i,Rd
are used instead of N
*
or M
*
in
the CIDECT Design Guides.
65
The term "joint" is used to represent the zone where
two or more members are interconnected, whereas
"connection" is used to represent the location at which
two or more elements meet.
The "through member" of a joint is termed the "chord"
and attached members are termed "braces" (although
the latter are also often termed bracings, branch
members or web members).
Fig. 6.5 shows some of the common joint notation for
gapped uniplanar K joints. The numerical subscripts (i
= 0, 1, 2) to symbols shown in Fig. 6.5 are used to
denote the member of a hollow section joint. The
subscript i = 0 designates the chord; i = 1 refers in
general to the brace for T, Y and X joints, or it refers
to the compression brace member for K and N joints; i
= 2 refers to the tension brace member for K and N
joints. For K and N overlap joints, the subscript i is
used to denote the overlapping brace member and j is
used to denote the overlapped brace member (see
Fig. 6.6).
6.2.3 Limitations on geometric
parameters
Most of the joint resistance formulae are subject to a
particular "range of validity". This often represents the
range of the parameters or variables for which the
formulae have been validated, by either experimental
or numerical research. In some cases, it represents
the bounds within which a particular failure mode will
control, thereby making the design process simpler.
Joints with parameters outside these specified ranges
of validity are sometimes allowed, but they may result
in lower joint efficiencies and generally require
considerable engineering judgement and verification.
With reference to Figs. 6.6 and 6.7 for RHS sections,
the gap g or overlap q, as well as the eccentricity e,
may be calculated by eqs. (6.1) and (6.2) (Packer et
al., 1992; Packer & Henderson, 1997):
( )
2
2
1
1
2 1
2 1 0
sin 2
h
sin 2
h
sin sin
sin
2
h
e g
+

.

\

+ = (6.1)
( ) 2
h
sin
sin sin
g
sin 2
h
sin 2
h
e
0
2 1
2 1
2
2
1
1
+


.

\

+ + = (6.2)
For CHS sections replace h
i
by d
i
.
Note that a negative value of the gap g in eq. (6.1)
corresponds to an overlap q. A positive value of the
eccentricity e represents an offset from the chord
centreline towards the outside of the truss.
6.3 LIMIT STATES AND LIMITATIONS
ON MATERIALS
6.3.1 Limit states
As stated in Chapter 1, this book is written in a limit
states design format already including appropriate
material and joint partial safety factors (
M
) or joint
resistance (or capacity) factors (). This means that
the effect of the factored loads (the specified or
unfactored loads multiplied by the appropriate load
factors) should not exceed the factored resistance of
the joint.
Some connection elements, which are not specific to
hollow sections, such as plate material, bolts and
welds, need to be designed in accordance with local
or regional structural steel specifications. Thus,
additional safety or resistance factors should only be
used where indicated.
If allowable stress design (ASD) or working stress
design is used, the joint factored resistance
expressions provided herein should, in addition, be
divided by an appropriate load factor. A value of 1,5 is
recommended by the American Institute of Steel
Construction (AISC, 2005).
6.3.2 Limitations on materials
The recommendations given are applicable to both
hotfinished and coldformed steel hollow sections, as
well as coldformed stressrelieved hollow sections.
The nominal specified yield strength of hollow
sections should not exceed 460 N/mm
2
(MPa). This
nominal yield strength refers to the finished product
and should not be taken larger than 0,8f
u
.
The joint resistances given are for hollow sections
with a nominal yield strength up to 355 N/mm
2
. For
nominal yield strengths greater than this value, the
joint resistances should be multiplied by 0,9. On one
hand, this provision considers the relatively larger
deformations that take place in joints with nominal
yield strengths around 450 to 460 N/mm
2
, when
plastification of the chord face or chord cross section
occurs (for large bracetochord diameter or width
ratios , it may be conservative); on the other hand,
for other joints the deformation or rotation capacity
may be lower with yield strengths exceeding 355
66
N/mm
2
. Furthermore, for any formula, the "design
yield stress" used for computations should not be
taken higher than 0,8 of the nominal ultimate tensile
strength. This provision allows for ample connection
ductility in cases where punching shear failure or
failure due to local yielding of a brace or plate govern,
since strength formulae for these failure modes are
based on the yield stress. For S460 steel hollow
sections, the reduction factor of 0,9, combined with
the limitation on f
y
to 0,8f
u
, results in a total reduction
in joint resistance of about 15%, relative to just directly
using a yield stress of 460 N/mm
2
(Liu & Wardenier,
2004).
Some codes, e.g. Eurocode 3 (EN 1993112, 2007)
give additional rules for the use of steel S690. These
rules prescribe an elastic global analysis for structures
with partialstrength joints. Further, a reduction factor
of 0,8 to the joint capacity equations has to be used
instead of the 0,9 factor which is used for S460.
6.4 GENERAL DESIGN
CONSIDERATIONS
In designing hollow section trusses it is important that
the designer considers the joint behaviour right from
the beginning. Designing members of a truss based
on member loads only may result in undesirable
stiffening of joints afterwards. This does not mean that
the joints have to be designed in detail in the
conceptual design phase. It only means that chord
and brace members have to be chosen in such a way
that the main governing joint parameters provide an
adequate joint strength and an economical fabrication.
Since the design is always a compromise between
various requirements, such as static strength, stability,
economy in fabrication and maintenance, which are
sometimes in conflict with each other, the designer
should be aware of the implications of a particular
choice.
The following guidance is given to arrive at an
optimum design:
 Lattice structures can usually be designed assuming
pin jointed members. Secondary bending moments
due to the actual joint stiffness can be neglected for
static design if the joints have sufficient rotation
capacity. This can be achieved by limiting the wall
slenderness of certain members, particularly the
compression members, which is the basis for some
of the geometric limits of validity. This will be the
case if the joint parameters are within the range of
validity given in the following chapters.
 It is common practice to design the members with
the centre lines noding. However, for ease of
fabrication it is sometimes required to have a certain
noding eccentricity. The effect should be considered
for member and joint design, see Section 6.5.
The gap g (Fig. 6.5) is defined as the distance
measured along the length of the connecting face of
the chord, between the toes of the adjacent brace
members (ignoring welds). In good designs, the
minimum gap between adjacent brace members
should be g > t
1
+ t
2
, to ensure that there is
adequate clearance to form satisfactory welds.
 In overlapped K joints, the inplane overlap should
be large enough to ascertain that the
interconnection of the brace members is sufficient
for adequate shear transfer from one brace to the
other. This can be achieved by ensuring that the
overlap, which is defined in Fig. 6.6, is at least 25%.
Where overlapping brace members are of different
widths, the narrower member should overlap the
wider one. Where overlapping brace members with
the same diameter have different thicknesses and/or
different strength grades, the member with the
lowest f
yi
t
i
value should overlap the other member.
 Gap joints are preferred to partial overlap joints,
since the fabrication is easier with regard to end
cutting, fitting and welding.
 An angle of less than 30 between a brace member
and a chord creates serious welding difficulties at
the crown heel location and is not covered in the
recommendations. However, angles less than 30
may be possible if the design is based on an angle
of 30 and it is shown by the fabricator that a
satisfactory weld can be made.
 In common lattice structures (e.g. trusses), about
50% of the material weight is used for the chords in
compression, roughly 30% for the chord in tension
and about 20% for the web members or braces.
This means that with respect to material weight, the
chords in compression should likely be optimised to
result in thin walled sections. However, for corrosion
protection (painting), the outer surface area should
be minimised. If both chords have to be designed for
compression loads, for example due to uplift wind
loading, the above values may change depending
on the lateral supports.
 Joint strength increases with decreasing chord
67
diameter or widthtothickness ratio. As a result, the
final diameter or widthtothickness ratio for the
chord in compression will be a compromise between
joint strength and buckling strength of the member
and relatively stocky sections will usually be chosen.
For the chord in tension, the diameter or
widthtothickness ratio should be chosen to be as
small as possible.
 Since the joint strength efficiency (i.e. joint strength
divided by the brace yield load A
i
f
yi
) increases with
increasing chordtobrace thickness t
0
/t
i
this ratio
should be chosen as high as possible, preferably
above 2. Furthermore, the weld volume required for
a thin walled brace is smaller than that for a thick
walled brace with the same cross section, if the
welds are to develop the capacity of the connected
brace member.
 The joint strength also depends on the yield stress
ratio between chord and brace, thus the use of
higher strength steel for chords (if available and
practical) may offer economical advantages.
 In principle, multiplanar trusses can be approached
in a similar way as uniplanar trusses, although the
depth can usually be smaller, between 1/15 and
1/18 L.
6.5 TRUSS ANALYSIS
Elastic analysis of hollow section trusses is frequently
performed by assuming that all members are pin
connected. Nodal eccentricities between the centre
lines of intersecting members at panel points should
preferably be kept to e s 0.25d
0
or 0.25h
0
. These
eccentricities produce primary bending moments
which, for a pinned joint analysis, need to be taken
into account in chord member design, e.g. by treating
it as a beamcolumn. This is done by distributing the
panel point moment (sum of the horizontal
components of the brace member forces multiplied by
the nodal eccentricity) to the chord on the basis of
relative chord stiffness on either side of the joint (i.e.
in proportion to the values of moment of inertia divided
by chord length to the next panel point, on either side
of the joint).
If these eccentricity limits are violated, the eccentricity
moment may have a detrimental effect on joint
strength and the eccentricity moment must be
distributed between all members of a joint. If moments
are distributed to the brace members, the joint
capacity must then be checked for the interaction
between axial load and bending moment, for each
brace member.
A rigid joint frame analysis is not recommended for
most planar, triangulated, singlechord, directly
welded trusses, as it generally tends to exaggerate
brace member moments, while the axial force
distribution will still be similar to that for a pin jointed
analysis.
Transverse loads applied to either chord away from
the panel points produce primary moments which
must always be taken into account when designing
the chords.
Computer plane frame programs are regularly used
for truss analysis. In this case, the truss can be
modelled by considering a continuous chord with
brace members pin connected to it at distances of +e
or e from it (e being the distance from the chord
centreline to the intersection of the brace member
centrelines). The links to the pins are treated as being
extremely stiff as indicated in Fig. 6.8. The advantage
of this model is that a sensible distribution of bending
moments is automatically generated throughout the
truss, for cases in which bending moments need to be
taken into account in the design of the chords.
Secondary moments, resulting from end fixity of the
brace members to a flexible chord wall, can generally
be ignored for both members and joints, provided that
there is deformation and rotation capacity adequate to
redistribute stresses after some local yielding at the
connections. This is the case when the prescribed
geometric limits of validity for design formulae given in
Chapters 8 to 11 are followed. Welds in particular
need to have potential for adequate stress
redistribution without premature failure, and this will
be achieved with the recommendations given in
Section 6.5.2. Table 6.1 summarizes when moments
need to be considered for designing CHS or RHS
trusses.
6.5.1 Truss deflections
For the purpose of checking the serviceability
condition of overall truss deflection under specified
(unfactored) loads, an analysis with all members
being pin jointed will provide a conservative
(over)estimate of truss deflections when all the joints
are overlapped. A better assumption for overlap
conditions is to assume continuous chord members
and pin jointed brace members.
68
69
However, for gapconnected trusses, a pin jointed
analysis still generally underestimates overall truss
deflections, because of the flexibility of the joints. At
the service load level, gapconnected hollow section
truss deflections are underestimated by around
510%. Thus, a conservative approach for
gapconnected trusses is to estimate the maximum
truss deflection by 1,1 times that calculated from a pin
jointed analysis.
6.5.2 Weld design
To avoid weld failure it is recommended to design the
welds to be stronger than the connected brace
members, wherever possible. Designing fillet welds in
this way, and adopting Eurocode 3 (EN 199318,
2005), results in the following minimum throat
thickness "a" for fillet welds around brace members,
assuming matched electrodes:
a > 0,92t, for S235 (f
yi
= 235 N/mm
2
)
a > 0,96t, for S275 (f
yi
= 275 N/mm
2
)
a > 1,10t, for S355 (f
yi
= 355 N/mm
2
)
a > 1,42t, for S420 (f
yi
= 420 N/mm
2
)
a > 1,48t, for S460 (f
yi
= 460 N/mm
2
)
For very lightly loaded structures, smaller welds are
allowed, provided care has been taken of the effective
weld lengths around the perimeter (Frater & Packer,
1990; Packer & Wardenier, 1992).
Table 6.1 Moments to be considered for CHS or RHS truss design
Type of moment Primary Primary Secondary
Moments due to
Nodal eccentricity
(e s 0.25d
0
or 0.25h
0
)
Transverse member loading
Secondary effects such as
local deformations
Chord design
Design of braces
Design of joints
Yes
No
Yes, for Q
f
only
Yes
Yes
Yes, influences Q
f
No
No
No, provided parametric
limits of validity are met
Note: For structures subjected to fatigue loading, all primary and secondary bending moments should be
considered, see Chapter 14.
70
b. Pratt truss
c. Vierendeel truss
d. truss with cross braces
a. Warren truss
b. Pratt truss
c. Vierendeel truss
d. truss with cross braces
a. Warren truss
Fig. 6.1 Various types of trusses
X joints T and Y joints
N and K joints KT joints
Fig. 6.2 Basic types of joints
71
gap
u u
N
100%
K
N
(a) (b)
u
N
100%
Y
0
(c) (d)
gap
u u
N 100%
K
N
100%
K
0
+e
(e) (f)
u
N
100%
X
u
N
(g)
Fig. 6.3 Examples of hollow section joint classification
u u
within tolerance
for:
1.2N
100%
K
N
0.2N sinu
within tolerance for:
1,2N
0,2N sin
u
u
100%
K
N
50% K
50% X
+e
0.5N sinu 0,5N sin
u
0.5N sinu 0,5N sin
u
u u
0.5 N/sinu
100%
X
0,5N
/
sin
u
0.5 N/sinu
N
0,5N
/
sin
u
72
0.5N sinu
0.5N sinu
u
N
N cosu
=
0.5N sinu
u
0.5N
0.5N cosu
0.5N sinu
u
+
0.5N
0.5N cosu
0,5Nsin u
0,5Ncos u Ncos u
0,5Nsin u
0,5Nsin u
0,5N
0,5N
0,5Ncos u
0,5Nsin u
0.5N sinu
0.5N sinu
u
N
N cosu
=
0.5N sinu
u
0.5N
0.5N cosu
0.5N sinu
u
+
0.5N
0.5N cosu
0,5Nsin u
0,5Ncos u Ncos u
0,5Nsin u
0,5Nsin u
0,5N
0,5N
0,5Ncos u
0,5Nsin u
Figure 6.4 Checking of a K joint with imbalanced brace loads
b
0
h
0
t
0
u
2
b
2
d
2
h
2
h
1
b
1
d
1
u
1
t
2
t
1
g
+e
N
1
N
2
N
0
1
2
0
Fig. 6.5 Symbols used for K gap joints
Overlap = x 100%
p
q
p
q
e
i = 1 or 2 (overlapping member)
j = overlapped member
i j
Fig. 6.6 Definition of overlap
73
25 , 0
h
e
or
d
e
0 0
s
e < 0
25 , 0
h
e
or
d
e
0 0
s
g
g
e = 0 e > 0
e < 0 e < 0
25 , 0
h
e
or
d
e
0 0
s 25 , 0
h
e
or
d
e
0 0
s
e < 0
25 , 0
h
e
or
d
e
0 0
s
g
g
e = 0 e > 0
e < 0 e < 0
Fig. 6.7 Noding eccentricity
For most
overlap joints
Extremely stiff
members
Pin
Extremely stiff
members
For most gap
joints
For most
overlap joints
Extremely stiff
members
Pin
Extremely stiff
members
For most gap
joints
Fig. 6.8 Plane frame joint modelling assumptions to obtain realistic forces for member design
74
7. BEHAVIOUR OF JOINTS
For a proper understanding of the behaviour of
welded joints between hollow sections it is important
to consider the load path, the internal stiffness
distribution in a joint and the material properties.
7.1 GENERAL INTRODUCTION
7.1.1 Load path
The load path shows the elements through which the
loads have to pass and where failure may occur. For
example, Fig. 7.1 illustrates a welded joint between
plates and a hollow section. The load has to pass via
the following elements:
 Plate
 Weld
 Hollow section face (through thickness)
 Hollow section side wall
In principle, failures can occur in any of these parts. If
the width of the plate b
1
is small compared to the
chord width b
0
, more types of failure can occur in the
chord face. This will be discussed later.
7.1.2 Internal stiffness distribution
The stiffness distribution in the joint determines the
elastic stress distribution. Here, the plate to RHS
chord joint of Fig. 7.1 will be examined again.
Consider the stiffness of the plate and the connected
face of the hollow section.
1. Plate
The plate end remains straight if loaded by a uniform
loading q per unit length. The deformation is
determined by the plate stiffness for axial stresses,
which is high.
2. Hollow section face
First consider a unit load q
1
at a small unit length at
the sides (Fig. 7.2b). The load q
1
can flow directly into
the hollow section side walls. Thus, the deformation is
determined by the stiffness of the hollow section side
wall for axial stresses.
Now consider a unit load q
2
at the centre of the hollow
section face (Fig. 7.2c). The load has to be
transmitted to the side walls by bending. Thus, the
deformation is determined by the bending stiffness of
the top face of the hollow section and the axial
stiffness of the hollow section side walls.
Consequently, the stiffness for a q
2
load is
considerably smaller than for a q
1
load. This is
graphically shown in Fig. 7.3. For loads on the top
face at locations between q
1
and q
2
the behaviour is in
between that for q
1
and q
2
.
The resulting elastic stress pattern in the plate can
now be determined in two ways.
1. Consider the deformations under a uniform stress
For a uniform stress, the plate and hollow section
faces do not have the same deformed shape. To
ensure that the plate and the hollow section face have
the same deformation, the stresses at the centre
should be lower and at the sides higher. Thus,
additional stresses have to be added to the uniformly
distributed loading, as shown in Fig. 7.4b. This
increases the stresses at the sides and reduces the
stresses at the plate centre. Hence, the highest
stresses occur at the stiff parts.
As shown in Fig. 7.4b, the plate remains almost
straight due to the much higher axial stiffness of the
plate compared to the bending stiffness of the chord
top face. Therefore, the plate could have been
assumed to be nearly rigid compared with the
stiffness of the hollow section top face.
2. Assume rigid plate
If the plate is assumed to be rigid, the stress pattern
can be directly determined with Fig. 7.3. For a
deformation
1
, the stress for q
1
is much higher than
for q
2
, resulting in the stress pattern of Fig. 7.4c.
From this evaluation, it is clear that the nonuniformity
largely depends on the b
0
/t
0
ratio. If b
0
/t
0
is very small,
approaching a solid profile, the stress distribution is
uniform if contraction is not considered. If b
0
/t
0
is
large, it may even be that the stress at the centre has
the opposite sign of that at the sides.
7.1.3 Effect of material properties
Fig. 7.5 shows the  diagram of two materials:
(a) Steel with a yield strength f
y
and a strain
hardening part with an ultimate tensile strength f
u
.
(b) A fictitious steel without any deformation capacity,
i.e. it fails immediately after reaching the
maximum stress f
u,b
.
Suppose that failure of the platetoRHS joint is
governed by failure of the plate just before the weld.
75
This means that the stress pattern in the plate (Fig.
7.4c) has to be considered in relation to the material
behaviour.
The stress pattern in Fig. 7.4c is based on an elastic
material behaviour, thus equivalent to material "b". As
soon as the maximum stress at the side (location 1)
reaches the ultimate stress f
u,b
, the material will start
cracking.
If material "a" of Fig. 7.5 had been used, the maximum
stress would first reach the yield stress f
y
. With
increasing load, the material at location (1) yields i.e.
the stress remains constant f
y
and the strain
increases. With further increasing load, the material
just beside location (1) in Fig. 7.4c will yield, etc. At a
certain strain, the material at location (1) will reach the
strain hardening part in the  diagram of Fig. 7.5.
After a further increase of the loading, the stress will
increase until the ultimate stress f
u
, after which the
"actual stress" will still increase, although the
"engineering stress", based on the original cross
section will decrease. At a certain ultimate strain
u
cracking will occur at location (1).
Sometimes cracking occurs at the very stiff locations
and still the loading can be increased due to a more
uniform stress distribution in the remaining cross
section.
The above example shows the importance of yielding
for the load capacity of hollow section joints.
Another aspect which is also extremely important for
static design is the deformation capacity. The
deformation capacity determines if secondary
moments can be redistributed in structures.
For example in a truss, secondary bending moments
exist due to the joint stiffness of the welded joints.
However, these moments are not necessary for
equilibrium of the structure. If the truss is loaded up to
failure and the joint strength is governing instead of
the member strength, at a certain moment, yielding
occurs due to the combination of axial loading and the
(secondary) bending moments. If the deformation
capacity is sufficient, the axial forces in the members
can increase with a decrease of the (secondary)
bending moments due to plastic rotation of the joint. In
the failure stage, the secondary bending moments
may have totally disappeared.
7.1.4 Failure modes
Following the load path (see Section 7.1.1) shows the
possible failure locations, whereas the stiffness
distribution (Section 7.1.2) in combination with the
material behaviour (Section 7.1.3) determine the
failure mode for the various locations. The lowest
failure load for all possible failure modes gives the
governing strength. Now the possible failure modes
for the platetoRHS joint of Fig. 7.1 will be evaluated.
1. Plate
Fig. 7.6a shows the possible stress distribution in the
plate after yielding and after reaching the ultimate
strain at the sides (location 1). If the chord widthto
thickness ratio b
0
/t
0
is low and the material has
sufficient ductility, the yield capacity of the plate can
be attained. In most cases the capacity is lower.
2. Welds
If the strength of the fillet welds (Fig. 7.6b) is lower
than that of the plate, the welds may fail. If plastic
deformation occurs in the welds only, the total
deformation for the joint is small, resulting in a joint
with no deformation capacity (which is generally not
allowed). Therefore, it is recommended that the welds
should preferably be designed to be stronger than the
connected brace members.
Only for very lightly loaded structures, e.g. where
members have been selected based on aesthetical
aspects, are smaller welds allowed, provided the
secondary effects and the effective perimeter are
considered (Frater & Packer, 1990; Packer &
Wardenier, 1992).
3. Chord face
The loading and hence the stresses have to pass via
the top face to the side walls. Especially for thick
material, cracking can occur due to
manganesesulphide (MnS) inclusions, called lamellar
tearing (Fig. 7.6c). To avoid this material problem,
material with good through thickness properties (TTP)
should be used, i.e. steels with low sulphur contents.
If b
1
< b
0
, other failure modes can be obtained for the
chord, i.e. chord face plastification or chord punching
shear.
For the joint with b
1
= b
0
the connecting chord face is
held in position by the plate and the stiff connection to
the side wall. Therefore, chord face yielding with a
distinct yield line pattern can only develop after
excessive yielding of the plate at the sides and/or
excessive yielding of the hollow section side walls
76
under the plate.
Punching shear of the hollow section face can only
occur if the plate width b
1
is smaller than b
0

2t
0
(see
Fig. 7.6d).
4. Chord side wall
All the stresses have to be transmitted through the
side walls over a limited width, thus this may be a
critical failure mode (chord side wall yielding shown in
Fig. 7.6e).
If the loading is compression instead of tension, the
stability of the side wall may be critical.
7.2 GENERAL FAILURE CRITERIA
In general, the static strength can be characterized by
various criteria, i.e.:
 Ultimate load resistance
 Deformation limit
 (Visually observed) crack initiation
The ultimate load capacity is well defined for those
joints which show a maximum in the load deformation
diagram, e.g. for selected joints loaded in
compression. Other joints show an increasing load
capacity with increasing deformation such that the
maximum is obtained at excessive deformation.
Besides the ultimate capacity criterion, a deformation
limit has been defined (Lu et al., 1994) to avoid
deformations which are too large. This limit, being
0,03d
0
or 0,03b
0
, as shown in Fig. 7.7, is based on the
fact that the deformation at serviceability should not
be governing and that crack initiation should not occur
at serviceability either. The deformation limit considers
the local displacement of the chord wall at the
connection of the brace to the chord.
Thus, the ultimate load capacity is defined by the
criterion that is met first, i.e. the maximum capacity or
the load at the deformation limit.
For serviceability, an arbitrary limit of 0,01d
0
or 0,01b
0
is adopted. This 1% limit is the same as the
outofroundness limit and has shown to give
acceptable deformations. However, it must be
mentioned that many formulae in the codes and
design guides were initially developed based on
ultimate load or endoftest data and were later on
evaluated for the 1% d
0
or 1% b
0
rule for
serviceability.
7.3 GENERAL FAILURE MODES
Similar to the platetohollow section chord joint in Fig.
7.1, hollow section joints exhibit, depending on the
loading, joint type and geometric parameters, various
modes of failure.
As an example, Fig. 7.8 illustrates typical failure
modes for a K joint of rectangular hollow sections, i.e.:
 Plastification of the chord face
 Chord punching shear
 Local brace failure (effective width)
 Chord shear failure
 Local buckling of the compression brace
 Local buckling of the chord
If the welds are not strong enough, weld failure can
also occur, or if the material does not have sufficient
through thickness properties (TTP) lamellar tearing is
possible.
The failure modes and associated analytical models
for determination of the strength formulae are
described in detail in the following chapters.
7.4 JOINT PARAMETERS
The geometry of a particular joint is generally defined
by the dimensions given in Fig. 6.5 and by the joint
parameters o, , , t and g shown in Fig. 7.9.
Originally the parameters were related to the radius of
a circular section. Nowadays the diameter, width or
depth are used which explains the factor of 2 in the
definition of the o and ratios.
77
A
1
= b
1
x t
1
1
a
t
0
h
0
b
0
t
1 b
1
A
1
= b
1
x t
1
1
a
t
0
h
0
b
0
t
1 b
1
Fig. 7.1 PlatetoRHS joint
q
q
1
q
2
o
1
o
q
q
1
q
2
o
1
o
Fig. 7.3 Loaddeformation diagram
actual stress
f
u,b
f
u
f
y
c
u
c
b
yield
a
strain hardeni ng
engineeri ng
stress
actual stress
f
u,b
f
u
f
y
c
u
c
b
yield
a
strain hardeni ng
engineeri ng
stress
Fig. 7.5  diagram steel
a. plate
b. RHS loaded
at the sides
c. RHS loaded
at the centre
q q
1
q
1
q
1
q
2
q
1
q
2
a. plate
b. RHS loaded
at the sides
c. RHS loaded
at the centre
q q
1
q
1
q
1
q
2
q
1
q
2
Fig. 7.2 PlatetoRHS chord joint stiffness
plate
RHS
stress deformation
plate
RHS
stress deformation
Fig. 7.4a Stress and resulting deformation
stress
deformation
plate
RHS
stress
deformation
plate
RHS
Fig. 7.4b Compatibility
1 2 1 1 2 1
Fig. 7.4c Resulting stress pattern
78
f
y
f
u
1 1 2
0,5b
e
0,5b
e
f
y
f
u
f
y
f
u
1 1 2
0,5b
e
0,5b
e
max.
3% d
0
N
a
o
3%d
0
or 3%b
0
max.
3% d
0
N
max.
3% d
0
N
a
o
3%d
0
or 3%b
0
Fig. 7.6a Plastic stress pattern and ultimate situation
at failure
Fig. 7.7 Deformation limit
(b) (c) (d)
Fig. 7.6b Weld failure
Fig. 7.6c Lamellar tearing
Fig. 7.6d Punching shear
elastic
plate
difficult to make
a proper weld
2,5:1
f
y
f
u
chord
b
w
plastic
t
1
ultimate
elastic
plate
difficult to make
a proper weld
2,5:1
f
y
f
u
chord
b
w
plastic
t
1
ultimate
Fig. 7.6e Chord side wall failure
79
Fig. 7.8 Failure modes for a K joint of rectangular hollow sections
0
0
0
0
b
2
or
d
2
= o
0
1
0
1
0
1
0
1
b
b
or
b
d
or
d
b
or
d
d
=  for T, Y and X joints
0
2 1
0
2 1
0
2 1
b 2
b b
or
b 2
d d
or
d 2
d d + + +
=  for K and N joints
0
0
0
0
t
b
or
t
d
2 =
0
i
t
t
= t
0
t
g
' g =
0 y
0
f
n
o
=
N
1
N
2
t
0
t
1
N
0p
N
0
0
N
1
u
1 u
1
u
2
N
1
N
2
t
0
t
1
N
0p
N
0
0
N
1
u
1 u
1
u
2
Fig. 7.9 Parameters used for defining the joint geometry
80
8. WELDED JOINTS BETWEEN
CIRCULAR HOLLOW
SECTIONS
8.1 INTRODUCTION
Circular hollow sections can be connected in various
ways, e.g.:
 With special prefabricated connectors (Fig. 8.1)
 With end pieces which allow a bolted joint (Fig. 8.2)
 Welded to a plate (Fig. 8.3)
 Welded directly to the through member (chord)
(Fig. 8.4)
For transport or erection it may be that bolted joints
are preferred or required, whereas for space
structures prefabricated connectors are generally
used. However, the simplest solution is to profile the
ends of the members which have to be connected to
the through member (chord) and weld the members
directly to each other. Nowadays, end profiling does
not give any problem and the end profiling can be
combined with the required bevelling for the welds.
Although the directly welded joint (Fig. 8.4) is the
simplest and cleanest solution, the load transfer is
rather complex due to the nonlinear stiffness
distribution along the perimeter of the connected
braces. The design rules have been based on
simplified analytical models in combination with
experimental evidence, resulting in semiempirical
design formulae.
8.2 MODES OF FAILURE
In Chapter 7 it was already indicated that the ultimate
load capacity is based on either the maximum in the
load deformation diagram (if the chord displacement is
smaller than 0,03d
0
) or the load at a chord
deformation of 0,03d
0
.
In accordance with the procedure described in
Chapter 7, i.e. following the loads, various possible
failure modes (Fig. 8.5) can be expected:
 Local brace failure (yielding, local buckling)
 Weld failure
 Lamellar tearing
 Chord plastification (face/wall, or cross section)
 Chord punching shear failure
 Chord local buckling
 Chord shear failure
As indicated in Chapter 6, to avoid weld failure it is
recommended to design the welds to be stronger than
the connected braces. Prequalified full penetration
welds can be considered always to be stronger than
the connected braces. Partial penetration plus fillet, or
fillet welds alone, can also usually provide a weld
connection as strong as the connected braces.
The material should not be susceptible to lamellar
tearing, i.e. especially for the larger thicknesses, the
sulphur content should be low (TTP quality).
Furthermore, in the current design recommendations,
the d/t ratios have been limited to avoid local buckling.
In addition, limiting the d/t ratio has the effect that local
brace failure is no longer a governing failure mode for
T, Y, X and K gap joints.
Furthermore, within the range of validity of the design
recommendations, it has been shown that the chord
shear criterion can be covered by the formula for
chord plastification.
As a result, the governing modes of failure to be
considered for uniplanar joints have been reduced to:
 Chord plastification
 Chord punching shear
8.3 ANALYTICAL MODELS
For the determination of the influencing joint
parameters, three models are used, i.e.:
 Ring model (for chord plastification)
 Punching shear model (for chord punching shear)
 Chord shear model
8.3.1 Ring model
The ring model, originally developed by Togo (1967),
is based on the assumption that, for example in an X
joint, most of the loading is transferred at the saddles
of the brace, since the chord is stiffest at these parts
of the connection perimeter (see the elastic stress
distribution in Fig. 8.6).
Consequently, the load N
1
in the brace can be divided
into two loads of 0,5
N
1
sin
1
at the saddles of the
brace perpendicular to the chord and at a distance
c
1
d
1
with c
1
< 1,0. These loads will be transferred by
an effective length B
e
of the chord. In the model, the
load 0,5
N
1
sin
1
is now considered as a line load over
the length B
e
, see Fig. 8.7.
81
At failure, the plastic moment capacity will be reached
at the locations A and B (in Fig. 8.8). Neglecting the
influence of axial and shear stresses the plastic
moment capacity per unit length m
p
results in:
0 y
2
0 p
f t
4
1
m = (8.1)
Assuming d
0
 t
0
~ d
0
gives for equilibrium:

.

\

u
=
2
d c
2
d
2
sin N
B m 2
1 1 0 1 1
e p
(8.2)
or:
1
0 y
2
0
1
0 e
1
sin
f t
) c 1 (
d / B 2
N
u 
= (8.3)
The effective width B
e
is determined experimentally
and depends on the ratio, e.g. for = 1,0 the width
B
e
is smaller than for = 0,5 due to the direct load
transfer through the chord. The average value is: B
e
=
2,5d
0
to 3d
0
.
This ring model only considers chord plastification
which is caused by the brace load components
perpendicular to the chord. It may be clear that the
loads in the chord also have an influence on the load
capacity of the joint. The effect of chord stress is given
by a function Q
f
. As a result, the strength equation has
the form:
f
1
0 y
2
0
1
0
1
Q
sin
f t
) c 1 (
c
N
u 
= (8.4)
where c
0
, c
1
and Q
f
are determined using results from
experiments and numerical analyses; see Section 8.4.
For X joints, this model gives good agreement with the
test results, but the formula needs further adjustments
for the more complicated joints, such as K and N
joints.
8.3.2 Punching shear model
Originally, the punching shear model served as a
basis for many design recommendations (Marshall,
1992). The punching shear failure mode is also
caused by the brace load component perpendicular to
the chord, i.e. N
1
sin
1
. The joint resistance is obtained
by multiplying the effective punching shear area and
the punching shear yield stress (Fig. 8.9). Due to the
unequal stiffness distribution, the stress distribution
will be non uniform, even after yielding. However,
tests on CHS joints have shown that within the range
of validity given, the full perimeter can be considered
to be effective.
For joints with
1
= 90 the punching shear area
is
0 1
t d t and the limiting value for the punching shear
stress is 3 / f
0 y
. Thus, the punching shear capacity
is given by:
3
f
t d N
0 y
0 1 1
t = (8.5)
For angles
1
< 90 the component perpendicular to
the chord has to be considered and the joint perimeter
will increase. Projecting the connection perimeter to a
flat surface through the chord crown gives an ellipse
and the ratio between the perimeter of this ellipse and
the circle for
1
= 90 is given by
1
1
sin 2
sin 1
u
u +
, resulting
in:
1
2
1
0 y 0 1 1
sin 2
sin 1
f t d 58 , 0 N
u
u +
t = (8.6)
Tests have shown that the chord stresses have a
minor effect, thus, the chord stress function Q
f
is not
included in eq. (8.6).
8.3.3 Chord shear model
In T and Y joints, failure is governed by a combination
of local failure of the chord cross section due to the
brace load component perpendicular to the chord and
chord failure due to shear, bending and, if present,
axial loading of the chord. This has been worked out
in detail by Van der Vegte & Makino (2006).
K joints with a large value may fail by a shear failure
in the gap location, see Fig. 8.10. The failure mode is
a chord cross section plastification due to shear load,
axial load and, if present, bending.
For compact chords, the chord shear capacity can be
derived from plastic analysis (see Section 2.3.4):
) f 58 , 0 ( A
2
3
f
A V
0 y 0
0 y
v 0 , pl
t
= = (8.7)
82
The axial load capacity of a chord member is given
by:
N
pl,0
= A
0
f
y0
= (d
0
 t
0
) t
0
f
y0
(8.8)
If the bending moments are small, only the interaction
between axial load and shear load has to be
considered, i.e.:
0 , 1
N
N
V
sin N
2
0 , pl
0 , gap
2
0 , pl
i i
s


.

\

+


.

\

u
(8.9)
or:
2
v 0 y
i i
0 y 0 0 , gap
A f 58 , 0
sin N
1 f A N


.

\

u
s (8.10)
If the chord is only loaded by the brace load
components, i.e. N
0p
= 0, then N
gap,0
= N
i
cos
i
, which
is shown in Fig. 8.11.
8.4 EXPERIMENTAL AND NUMERICAL
VERIFICATION
Nowadays, not only does a lot of experimental
evidence exist, but also many results derived with
numerical analyses are available. Makino et al. (1996)
give a good survey of the available experimental data.
The experimental work has been mainly carried out in
Germany, Japan, USA, The Netherlands, UK and
Norway. The joints have been tested in various testing
setups (e.g. Fig. 8.12), primarily on isolated joints.
Only a few tests have been carried out on joints in
complete girders. For example, in the framework of an
offshore programme, Bolt & Billington (2000) have
conducted scale tests on complete frames of jackets
to simulate the interaction between joint and member
behaviour and to calibrate numerical models for frame
behaviour; see also Choo et al. (2005a).
For experiments, attention should be given to the
support and loading conditions to avoid constraint
effects (Liu et al., 1998). For numerical results, it is
important that the FE models used have been
calibrated against experimental data (Van der Vegte
et al., 2010a). The elements and the mesh should be
considered properly (see Fig. 8.13); details are given
by Van der Vegte (1995).
8.5 BASIC JOINT STRENGTH
FORMULAE
8.5.1 T, Y, X joints and K, N gap joints
The analytical ring model approach has served as a
basis for the determination of the joint strength
formulae. Based on the available numerical and test
results for X joints and using eq. (8.4), the values for
c
0
and c
1
have been determined to obtain the function
for the mean strength, see Fig. 8.14. Further, in the
recent IIW (2009) recommendations a minor
correction in the function for and has been
included to reduce the capacity for very low ratios to
be more consistent with the capacity for T joints.
Since for T, Y, K and N joints the load transfer is more
complicated, semiempirical functions for and as
well as for the gap g of K joints have been derived,
resulting in the following, general format:
f
1
2
0 0 y
Rd , 1
Q
sin
t f
) ' g ( f ) ( f ) ( f N
u
 = (8.11)
In the new recommendations of IIW (2009) and
CIDECT Design Guide No. 1 (Wardenier et al.,
2008a), this equation has been presented as:
i
2
0 y0
f u Rd i,
sin
t f
Q Q N = (8.11a)
with Q
u
= f() f() f(g) and Q
f
= f(n), where n is the
ratio between the maximum chord stress at the
connecting face and the chord yield stress:
pl,0
0
pl,0
0
M
M
N
N
n + = (8.12)
In the comparison of the test results with the resulting
formulae for the basic uniplanar joints, it was
concluded that within the range of validity given, the
results can be described by a primary joint strength
function for chord plastification and an additional
check for punching shear. The chord shear criterion
does not have to be checked separately.
In the previous recommendations of IIW (1989) and
CIDECT Design Guide No. 1 (Wardenier et al., 1991),
the chord stress function, called f(n) was based on
the chord prestress which was in contradiction with
the function for joints with square hollow sections,
which used the maximum chord stress as the
governing parameter. This inconsistency has been
83
corrected in the current design recommendations of
IIW (2009) and the 2
nd
edition of CIDECT Design
Guide No. 1 (Wardenier et al., 2008a) adopted in this
book.
8.5.2 K, N overlap joints
For overlap joints, the same approach is adopted for
all types of overlap joints, regardless whether circular
or rectangular braces are used in combination with a
circular, rectangular or open section chord
(Wardenier, 2007; Qian et al., 2007). Only the
effective width parameters depend on the type of
section. The resistance of overlap joints between
circular hollow sections with 25% s Ov s 100%
overlap is based on the following criteria:
(1) Local failure of the overlapping brace
(2) Local chord member yielding at the joint location
based on interaction between axial load and
bending moment
(3) Shear of the connection between the brace(s) and
the chord
Figure 8.16 shows the overlap joint configuration with
the cross sections to be examined for these criteria.
Local failure of the overlapping brace (criterion 1)
should always be checked, while shear between the
braces and the chord (criterion 3) may only become
critical for larger overlaps, i.e. larger than 60% or
80%, depending on whether or not the hidden toe
location of the overlapped brace is welded to the
chord. The check for local chord member yielding
(criterion 2) is, in principle, a member check and may
become critical for larger overlaps and/or larger
ratios.
For 100% overlap joints, similar criteria have to be
checked. Only here, as shown by Qian et al. (2007),
shear of the overlapped brace and chord member
yielding will generally be the governing criteria.
Although an overlap of 100% is given in the
recommendations, in general, the overlap will be
slightly larger to allow proper welding of the
overlapping brace to the overlapped brace.
Joints with overlaps between 0% and 25% should be
avoided because in those cases, the stiffness of the
connection between the overlapping brace and the
overlapped brace is much larger than that of the
overlapping brace to chord connection, which may
lead to premature cracking and lower capacities
(Wardenier, 2007).
More detailed information regarding the design
equations for overlap joints is given in Chapter 11.
8.6 EVALUATION TO DESIGN RULES
In the analysis for the basic joint strength formulae, at
first, functions have been derived which predict the
mean strength with the lowest coefficient of variation.
Considering the scatter in test results, typical
tolerances in dimensions and workmanship and the
variation in yield stress, in the second step,
characteristic joint strength formulae have been
determined with a 5% probability of lower strength
(Kurobane, 1981; Wardenier, 1982; Van der Vegte et
al., 2008).
In the analyses by Van der Vegte et al. (2007, 2008,
2010b) and Zhao et al. (2008) for the new
recommendations of IIW (2009) and CIDECT Design
Guide No. 1 (Wardenier et al., 2008a), these
characteristic formulae have been further divided by a
partial factor
M
of 1,1 for chord plastification and 1,0
for punching shear and 1,0 for chord shear.
Finally, the equations have been slightly simplified to
derive the design resistance formulae.
Table 8.1 summarizes the design resistance formulae
for the basic uniplanar joints. As shown, only the
chord plastification criterion and the general punching
shear criterion should be checked.
Fig. 8.15 shows that a compression stress n < 0 in the
chord may give a considerable reduction of the joint
resistance, whereas for chord tension, the influence
depends on the ratio. The influence functions given
are fully based on experimental and numerical
evidence. For chord compression, a function that
provided the best fit with the test results is used
whereas for chord tension, a lower bound is adopted
for the joints with high ratios. Recent results (Qian et
al., 2008) for low ratios show that the chord stress
function acts more as a mean function for these data.
In the previous recommendations of IIW (1989) and
CIDECT (Wardenier et al., 1991), no reduction for
chord tensile loading was used since at that time, the
available experiments did not show a reduction for
chord tension loads up to about 80% of the chord yield
capacity (Wardenier, 1982).
The joint resistances given in Table 8.1 are for hollow
sections with a nominal yield strength not exceeding
84
355 N/mm
2
. As discussed in Section 6.3, for nominal
yield strengths greater than this value (up to S460),
the joint resistances given have to be multiplied by 0,9
(Liu & Wardenier, 2004).
8.7 OTHER TYPES OF JOINTS
8.7.1 Related types of joints
Large diameter CHS sections are built up from cans
with a maximum length equal to the maximum plate
width from which they are made. In structures with
these large size tubular sections, it is easy to use a
larger can thickness at the joint locations. This is, for
example, commonly used in offshore structures.
However, it can also be applied to other heavily
loaded structures, such as bridges and large span
structures.
X Joints with short can lengths have been numerically
investigated by Van der Vegte (1995). For X joints, it
was shown that the can should have a minimum
length of 2,5d
0
in order to obtain a joint resistance
based on the can thickness. For smaller can lengths,
a linear interpolation can be made between the
resistance of the joint with and without a can.
The joint design resistance for the special types of
joints shown in Table 8.2 can be directly related to
that for the basic type of joints in Table 8.1. In all
cases, the brace load components perpendicular to
the chord have to be considered, since these affect
the chord plastification.
The first and second joints in Table 8.2 have a loading
effect similar to that for an X joint and the design
resistance is therefore related to that for X joints.
The third and fourth joints have a loading comparable
to that for a K joint and the design resistance should
therefore be related to that for K joints.
It is also clear that in the latter case the shear of the
chord is higher than that for a K joint and should be
checked seperately.
8.7.2 Plate to CHS joints
Table 8.3 shows T and X joints with a circular hollow
section chord and various configurations for the brace.
The design resistance for these types of joints can be
related to each other by the following, general
strength function (Wardenier et al., 2009):
f
2
0 y0 1
Q t f ) ( f ) ( f ) ( f N q  = (8.13)
or presented as eq. (8.11a) with:
) ( f ) ( f ) ( f Q
u
q  = (8.14)
where for X joints:
( )
0,15
u
0,4 1
0,7 1
1
2,2 Q +


.

\


 +
= (8.14a)
and for T joints:
( )( ) +  + =
0,2 2
u
0,4 1 6,8 1 2,2 Q (8.14b)
The database for plate to CHS joints mainly includes
tests carried out in Japan and is summarised by
Kurobane (1981), Makino et al. (1991) and Wardenier
(1982).
The capacities of joints between transverse plates, I
sections or RHS sections as braces and a CHS chord,
as illustrated in Table 8.3, are directly related to those
of joints between CHS braces and chords. However, a
detailed analysis (Wardenier et al., 2009) revealed
that for selected cases, large discrepancies exist
between the data sets. Hence, until more evidence is
available, e.g. from Voth (2010), the constant in eqs.
(8.14a) and (8.14b) is taken as 2,2 instead of 2,6
derived for joints between CHS braces and chords. As
a result, the current design capacities are
considerably smaller than those in the previous IIW
(1989) recommendations.
Care has to be taken with regard to the effective
perimeter for punching shear. If, for example, the
flanges of I sections are close to each other, thus is
small, e.g. s 2, then the inner part between the
flanges cannot resist forces and only the outer
perimeter should be considered as effective, similar to
CHS or RHS braces.
8.7.3 Multiplanar joints
In multiplanar joints, two additional effects influence
the joint capacity compared to that for uniplanar joints,
i.e.:
 The geometric effect (stiffening by the braces)
 The loading effect
85
For example, consider the XX joint in Table 8.4. If the
outofplane braces are very small in diameter, and
unloaded, they have hardly any effect on the
deformation of the chord. However, if the diameter
increases (e.g. = 0,6), the chord cross section is
stiffened considerably. As a consequence, for chord
plastification, the geometric effect on the joint capacity
will be minor for small values and significant for
larger values.
The deformation capacity further decreases for
ratios close to 0,7 since the load transfer is
concentrated at the outofplane gap locations
between the braces. Below = 0,7, the braces are not
overlapping.
For the load effects in XX joints, it is clear that, for
chord plastification, loads in the brace planes in the
opposite sense will decrease the joint capacity,
whereas loads in the same sense will increase the
joint capacity. However, for chord punching shear, the
factor should not exceed 1,0.
Although the effects depend on the joint parameters
(Paul, 1992; Van der Vegte, 1995) the influence
function in Table 8.4 for XX joints can be considered
to be a lower bound, especially for loading in both
planes in the same sense.
8.7.4 Joints loaded by brace bending
moments
In principle, the design resistance formulae for joints
loaded by brace inplane or outofplane bending
moments have been determined in a similar way to
that for axially loaded joints, also resulting in two
strength criteria, i.e. chord plastification and chord
punching shear. The design resistance formulae are
based on the analyses by Wardenier (1982) and Van
der Vegte et al. (2010b) and are given in Table 8.5.
For outofplane bending, the loading is concentrated
at the saddles, similar to X joints in the ring model
approach. This explains why the same function is
used for f().
In the case of Vierendeel girders, it is recommended
to choose joints with width ratios close to =1,0 to
provide sufficient stiffness and strength.
For plate, I, H or RHS to CHS chord joints the moment
resistance is given in Tables 8.3a and 8.3b.
8.7.5 Interaction between axial loads and
bending moments
Joints with brace members subjected to combined
loading should satisfy:
0 , 1
M
M
M
M
N
N
Rd , i , op
Ed , i , op
2
Rd , i , ip
Ed , i , ip
Rd , i
Ed , i
s +


.

\

+ (8.15)
This interaction equation is based on the work of
Hoadly & Yura (1985), although the exponents have
been rounded off. Eq (8.15) shows directly that the
influence of axial load and brace outofplane bending,
both with the loading concentrated at the saddles, is
comparable, whereas brace inplane bending, with the
maximum loading more concentrated at the crown,
has a smaller effect in the interaction.
8.8 DESIGN CHARTS
In the design process, it is important that the designer
knows how to design and that he or she can quickly
check if a particular design will be appropriate. Hence,
design graphs have been established (Wardenier et
al., 1991, 2008a) in which the joint resistance is
expressed as an efficiency, i.e. the joint resistance is
given as a fraction of the yield capacity A
i
f
yi
of the
connected brace. This results in the following
efficiency formula:
Efficiency =
i
f
i yi
0 0 y
e
yi i
Rd , i
sin
Q
t f
t f
C
f A
N
u
= (8.16)
In the case of
2 1
d d = for K joints, eq. (8.16) has to be
multiplied by
i
2 1
2d
d d +
, where d
i
is the diameter of the
brace considered.
The efficiency parameter C
e
(C
T
for T joints, C
X
for X
joints and C
K
for K gap joints), see Figs 8.17 to 8.19,
gives the efficiency for a joint with:

i
= 90
 f
y0
t
0
= f
yi
t
i
(identical thickness and yield stress for
brace and chord)
 Q
f
= 1,0
As an example, Fig. 8.19b shows that for a K joint with
g = 2t
0
, 2 ~ 30 and ~ 0,5, a value of C
K
~ 0,4
applies. Thus, for an angle
i
= 45, a 100% efficiency
can be obtained if:
86
77 , 1
t f
t f
i yi
0 0 y
>
Note that this ratio should be slightly larger because
the chord stress effect Q
f
has not been included.
This example shows that for a 100% efficiency, f
y0
t
0
should always be considerably larger than f
yi
t
i
.
The chord stress function Q
f
for CHS joints is given as
a function of the parameter n, thus related to the
maximum stress to yield stress ratio in the chord, see
Figs. 8.11 and 8.15.
For simply supported lattice girders, the influence of
the chord stress function is small (small chord loads)
near the supports, whereas it has a pronounced
influence at the centre of the girder where the chord
loads are high and, in general, the brace loads are
small. The chord stress function is especially
important in the case of continuous or cantilevered
lattice girders.
8.9 RELATION TO THE PREVIOUS
RECOMMENDATIONS OF IIW
(1989) AND CIDECT (1991)
Recently all data have been reanalysed (Van der
Vegte et al., 2007, 2008, 2010b) for the revision of the
recommendations of IIW (2009) and CIDECT Design
Guide No. 1 (Wardenier et al., 2008a). Zhao et al.
(2008) give a detailed overview of the modifications
made for the IIW (2009) recommendations. These
new recommendations are also to be included in the
new ISO standard in this field and will be the basis for
the next revision of the Eurocode 3 recommendations.
For these reanalyses, extensive numerical data have
been obtained. One of the main reasons for the
analyses was to define the chord stress function by
the maximum chord stress, in accord with that for
RHS joints. However, this also influences the basic
joint strength functions, while the same analytical
models apply. Further, the current database now
includes more joints with larger sections where the
size of the welds has a smaller influence on the joint
capacity.
In the new recommendations of IIW (2009) and
CIDECT (Wardenier et al., 2008a), for both chord
compression and tension loading, the maximum chord
load has to be considered for the chord load function
and both give a reduction in joint resistance. In the
previous recommendations, a reduction factor was
only given for chord compression loading and was
based on the chord preload.
For brace inplane bending, only the constant has
been changed resulting in about 10% lower strength.
This reduction is also a result of newly derived data
with larger dimensions and relatively smaller welds.
For brace outofplane bending, the capacity in the
recommendations of IIW (2009) and the 2
nd
edition of
CIDECT Design Guide No. 1 is again related to the
axial load capacity of X joints. On average, for
medium and ratios the joint capacities are about
the same as those in Eurocode 3 (EN 199318,
2005).
For the plate, I or RHS to CHS chord joints, selected
data of Akiyama et al. (1974) were originally excluded
from the database due to unexplained deviations from
other data. In the new analyses these data are
included since no satisfactory explanation could be
obtained for the discrepancies observed. However,
this resulted in a considerable reduction of the joint
capacities.
Currently, a research programme, including both
experiments and numerical analyses, is being carried
out at the University of Toronto (Voth, 2010) to
investigate this type of configuration more in detail.
These results confirm that Akiyamas data are far too
conservative.
In most cases, the updated recommendations of IIW
(2009) and CIDECT (Wardenier et al., 2008a) give
slightly lower capacities than the previous equations,
although there are areas where they give the same or
higher capacities. A comprehensive comparison
between these new recommendations and the
previous equations is given by Wardenier et al. (2008a,
2008b). These references further give a comparison
with the latest update of the API (2007); see also
Pecknold et al. (2007).
Detailed information about the background of
American codes is given by Marshall & Toprac (1974),
Marshall (1984, 1992, 2004, 2006) and Packer et al.
(2010). The background for the ISO 19902 (2007)
offshore code is given by Dier & Lalani (1998) and
Dier (2005).
8.10 CONCLUDING REMARKS
For more detailed information about special types of
87
88
joints such as cropped end joints, flattened end joints,
stiffened joints, slotted CHS gusset plate joints, etc.,
reference is made to the appropriate literature; see
Choo et al. (2004, 2005b), Dutta (2002), Kurobane
(1981), Packer (2006), Packer & Henderson (1997),
Rondal (1990), Thiensiripipat (1979), Wardenier
(1982) and Wardenier et al. (2008a).
Table 8.1 Design axial resistances of welded joints between circular hollow sections
Type of joint Design limit state
T and Y joints Chord plastification
( )
f
0,2 2
1
2
0 y0
Rd 1,
Q 7 , 7 1 6 , 2
sin
t f
N +
u
=
Chord punching shear (for d
1
s d
0
 2t
0
)
u
1
N
1
t
1
d
1
d
0
t
0
N
0
M
0
1
2
1
0 1 y0 Rd 1,
sin 2
sin 1
t d f 0,58 = N
u +
t
X joints Chord plastification
f
15 , 0
1
2
0 0 y
Rd 1,
Q
0,7  1
6 , 2 6 , 2
sin
t f
N


.

\
  +
u
=
Chord punching shear (for d
1
s d
0
 2t
0
)
See chord punching shear equation for T and Y joints
Chord shear (for X joints, if cos
1
> )
u
1
N
1
t
1
d
1
N
1
d
0
t
0
N
0
1
0
0 y Rd , 1
sin
A ) / 2 (
f 58 , 0 N
u
t
=
2
0 0 y
1 Ed , 1
0 y 0 Rd , 0 , gap
A ) / 2 ( f 58 , 0
sin N
1 f A N


.

\

t
u
=
K and N gap joints Chord plastification
f
8 , 0
0
0,3 6 , 1
1
2
0 0 y
Rd 1,
Q
) t / g ( 2 , 1
1
1 ) 2 , 3 1 (1,65
sin
t f
N
(
(
+
+  +
u
=
Rd 1,
2
1
Rd 2,
N
sin
sin
N
u
u
=
Chord punching shear (for d
i
s d
0
 2t
0
)
u
2
N
2
t
2
d
2
N
1
u
1
t
1
d
1
g
+e
d
0
N
0
i
2
i
0 i y0 Rd i,
sin 2
sin 1
t d f 0,58 = N
u +
t
( )
1
C
f
n 1 Q = with
Rd pl,0,
Ed 0,
Rd pl,0,
Ed 0,
M
M
N
N
n + = in connecting face
Function Q
f
Chord compression stress (n < 0) Chord tension stress (n > 0)
T, Y and X joints C
1
= 0,45  0,25
K gap joints C
1
= 0,25
C
1
= 0,20
Range of validity
1,0
d
d
0,2
0
i
s s
0 i
t t s
0,25
d
e
0
s
General
30
i
> u
2 1
t t g + >
y0 yi
f f s
(1)
u y
f 8 , 0 f s
2
y
N/mm 460 f s
Compression class 1 or 2
(2)
and ) 40 /t d : joints X (for 50 /t d
0 0 0 0
s s
Chord
Tension ) 40 /t d : joints X (for 50 /t d
0 0 0 0
s s
Compression class 1 or 2
(2)
and 50 /t d
i i
s
Braces
Tension 50 /t d
i i
s
(1)
For 355 N/mm
2
< f
y0
s 460 N/mm
2
, use a reduction factor of 0,9 for the design resistances.
(2)
Class 1 and 2 limits for d
i
/t
i
are given in Table 2.7.
89
Table 8.2 Design resistance of related types of joints
Type of joint Relationship to the formulae of Table 8.1
All brace member forces act in the same
sense (compression or tension).
N
1,Ed
s N
1,Rd
with N
1,Rd
for X joint given in Table 8.1
All brace member forces act in the same
sense (compression or tension).
N
1,Ed
sin
1
+ N
2,Ed
sin
2
s N
X,Rd
sin
X
with N
X,Rd
sin
X
from X joint given in Table 8.1, being the larger of the values
for brace 1 or 2
Forces in members 1 are in compression
and members 2 in tension.
N
i,Ed
s N
i,Rd
(i = 1 or 2)
with N
i,Rd
from K joint given in Table 8.1, but with the actual chord force
Forces in members 1 are in compression
and members 2 in tension.
N
i,Ed
s N
i,Rd
(i = 1 or 2)
with N
i,Rd
from K joint given in Table 8.1
Note: In a gap joint, the chord cross section in the gap has to be checked for
shear failure:
0 , 1
V
V
N
N
2
Rd , 0 , pl
Ed , 0 , gap
2
Rd , 0 , pl
Ed , 0 , gap
s


.

\

+


.

\

where:
N
gap,0,Ed
= design axial force in gap
0 y 0 Rd , 0 , pl
f A N =
V
gap,0,Ed
= design shear force in gap
t
=
0
0 y Rd , 0 , pl
A 2
f 58 , 0 V
N2
N1
N2
N1
1
N
2
N
90
Table 8.3a Design resistances of welded T joints connecting plates or open sections to CHS chords
Type of T joint Design limit state
Chord plastification
N
1,Rd
M
ip,1,Rd
M
op,1,Rd
t 1
do
b
1
t o
N
1
( )
f
0,2 2
2
0 y0 Rd 1,
Q 15 2 , 2 t f N + = 0 M
Rd , 1 , ip
=
Rd , 1 1 Rd , 1 , op
N b 5 , 0 M =
h
1
d
0
1
t
0
t
N
1
f
2
0 0 y Rd 1,
Q ) 0,4 (1 t f 5 N q + =
Rd , 1 1 Rd , 1 , ip
N h M = 0 M
Rd , 1 , op
=
( )
f
0,2 2
2
0 y0 Rd 1,
Q ) 0,4 (1 15 2 , 2 t f N q + + =
) 4 , 0 1 (
N h
M
Rd , 1 1
Rd , 1 , ip
q +
=
Rd , 1 1 Rd , 1 , op
N b 5 , 0 M =
Chord punching shear (for b
1
s d
0
 2t
0
)
I section brace with s 2 (axial
loading and brace outofplane
bending) and RHS brace
1
0
0 y
1 , op , el
Ed , 1 , op
1 , ip , el
Ed , 1 , ip
1
Ed , 1
t
t
f 58 , 0
W
M
W
M
A
N
s + + (t
1
= flange thickness for I section brace)
All other cases
1
0
0 y
1 , op , el
Ed , 1 , op
1 , ip , el
Ed , 1 , ip
1
Ed , 1
t
t 2
f 58 , 0
W
M
W
M
A
N
s + +
( )
1
C
f
n 1 Q = with
Rd pl,0,
Ed 0,
Rd pl,0,
Ed 0,
M
M
N
N
n + = in connecting face
Brace axial load, brace inplane bending and brace outofplane bending
Function Q
f
Chord compression stress (n < 0): C
1
= 0,25 Chord tension stress (n > 0): C
1
= 0,20
Range of validity
General
1,0
d
b
0,2
0
1
s s
90
1
= u
y0 y1
f f s
(1)
u y
f 8 , 0 f s
2
y
N/mm 460 f s
Compression class 1 or 2
(2)
and ) 40 /t d : joints X (for 50 /t d
0 0 0 0
s s
CHS chord
Tension ) 40 /t d : joints X (for 50 /t d
0 0 0 0
s s
Compression class 1 or 2
(2)
and 40 /t h and 40 /t b
1 1 1 1
s s
RHS braces
Tension 40 /t h and 40 /t b
1 1 1 1
s s
Compression class 1 or 2
(2)
I section braces
Tension none
Plates
Transverse plate: 4 , 0
d
b
0
1
> =  Longitudinal plate: 4
d
h
1
0
1
s = q s
(1)
For 355 N/mm
2
< f
y0
s 460 N/mm
2
, use a reduction factor of 0,9 for the design resistances.
(2)
Section class limitations are given in Table 2.7.
h
1
d
0
t
0
N
1
u
1
h
1
d
0
t
0
N
1
b
1
u
1
91
Table 8.3b Design resistances of welded X joints connecting plates or open sections to CHS chords
Type of X joint Design limit state
Chord plastification
N
1,Rd
M
ip,1,Rd
M
op,1,Rd
t 1
d
0
b
1
t 0
N
1
N
1
f
15 , 0 2
0 0 y Rd 1,
Q
0,7  1
2 , 2 2 , 2
t f N


.

\
  +
= 0 M
Rd , 1 , ip
=
Rd , 1 1 Rd , 1 , op
N b 5 , 0 M =
h
1
d0
1
t 0
t
N
1
N
1
f
2
0 0 y Rd 1,
Q ) 0,4 (1 t f 5 N q + =
Rd , 1 1 Rd , 1 , ip
N h M = 0 M
Rd , 1 , op
=
h
1
d
0
t
0
N
1
N
1
u
1
f
15 , 0
2
0 0 y Rd 1,
Q ) 4 , 0 1 (
0,7  1
2 , 2 2 , 2
t f N q +


.

\
  +
=
) 4 , 0 1 (
N h
M
Rd , 1 1
Rd , 1 , ip
q +
=
Rd , 1 1 Rd , 1 , op
N b 5 , 0 M =
Chord punching shear
See chord punching shear equations for T joints in Table 8.3a
Function Q
f
Same as in Table 8.3a
Validity range Same as in Table 8.3a
(1)
h
1
d
0
t
0
N
1
b
1
N
1
u
1
(1)
For 355 N/mm
2
< f
y0
s 460 N/mm
2
, use a reduction factor of 0,9 for the design resistances.
92
Table 8.4 Correction factors for the design resistance of multiplanar joints
Type of joint Correction factor to uniplanar joint resistance
TT joints General
Members 1 may be either in tension or
compression

N1
g
t N
1

N
1
N
1

N1
g
t N
1

N
1
N
1
= 1,0
XX joints Chord plastification
Ed 1,
Ed 2,
N
N
35 , 0 1+ =
Chord punching shear
0 , 1 but
N
N
35 , 0 1
Ed 1,
Ed 2,
s + =
Members 1 and 2 can be either in compression
or tension
N
1
N
1
N
1
N
1
N
2
N
2
N
1
N
2
N
1
N
2
N
1
N
1
N
1
N
1
N
1
N
1
N
2
N
2
N
1
N
2
N
1
N
2
N
1
N
1
Notes:
 Take account of the sign of N
1,Ed
and N
2,Ed
, with N
2,Ed
 s N
1,Ed

 N
2,Ed
/N
1,Ed
is negative if the members in one plane are in tension and
in the other plane in compression
KK gap joints General
Members 1: compression
Members 2: tension
N
2
N
1

N
1
N
2
A
A
g
t
A
A

N
2
N
1

N
1
N
2
A
A
g
t
A
A
N
2
N
1

N
1
N
2
A
A
g
t
A
A

= 1,0
Note: In a KK gap joint, the chord cross section in the gap has to be
checked for shear failure:
0 , 1
V
V
N
N
2
Rd , 0 , pl
Ed , 0 , gap
2
Rd , 0 , pl
Ed , 0 , gap
s


.

\

+


.

\

where:
N
gap,0,Ed
= design axial force in gap
0 y 0 Rd , 0 , pl
f A N =
V
gap,0,Ed
= design shear force in gap
t
=
0
0 y Rd , 0 , pl
A 2
f 58 , 0 V
Validity range Same as in Table 8.1 and 60 s  s 90
93
Table 8.5 Design moment resistances of welded joints between circular hollow sections
Type of joint brace loading Design limit state
T, Y and X joints inplane bending Chord plastification
f
0,5
1
1
2
0 y0
Rd ip,1,
Q
sin
d t f
4,3 M 
u
=
Chord punching shear (for d
1
s d
0
 2t
0
)
1
2
1 2
1 0 y0 Rd ip,1,
sin 4
sin 3 1
d t f 58 , 0 M
+
=
T, Y and X joints outofplane bending Chord plastification
f
15 , 0
1
1
2
0 0 y
Rd op,1,
Q
0,7  1
3 , 1 3 , 1
sin
d t f
M


.

\
  +
u
=
Chord punching shear (for d
1
s d
0
 2t
0
)
1
2
1 2
1 0 y0 Rd op,1,
sin 4
sin 3
d t f 58 , 0 M
+
=
( )
1
C
f
n 1 Q = with
Rd pl,0,
Ed 0,
Rd pl,0,
Ed 0,
M
M
N
N
n + = in connecting face
Function Q
f
Chord compression stress (n < 0) Chord tension stress (n > 0)
T, Y and X joints C
1
= 0,45  0,25 C
1
= 0,20
Validity range Same as in Table 8.1
(1)
(2)
(1)
For 355 N/mm
2
< f
y0
s 460 N/mm
2
, use a reduction factor of 0,9 for the design resistances.
(2)
The equations in Table 8.5 may also be used for K gap joints, if brace moments have to be considered, by checking that
the brace utilization due to bending plus the brace utilization due to axial load s 0,8. For K overlap joints, no evidence
exists.
94
Fig. 8.1 Example of a prefabricated connector (this
Nodus type is out of production)
Fig. 8.2 Joints with end pieces for bolted joints
Fig. 8.3 Welded CHS slotted gusset plate joints
95
T joint Y joint
d
1
t
1
t
0
N
1
d
0
u
1
d
1
t
1
t
0
N
1
d
0
u
1
N
1
t
0
d
0
N
0
d
1
t
1
u
1
N
1
t
0
d
0
N
0
d
1
t
1
u
1
X joint K joint with gap
t
0
N
1
N
0
N
1
d
0
d
1
t
1
u
1
t
0
N
1
N
0
N
1
d
0
d
1
t
1
u
1
d
0 t
0
N
0
N
2
d
2
t
2
d
1
t
1
N
1
u
1
g
u
2
d
0 t
0
N
0
N
2
d
2
t
2
d
1
t
1
N
1
u
1
g
u
2
N joint with overlap KT joint with gap
t
i
d
0
t
0
N
0
d
i
N
i
t
j
d
j
u
i
N
j
u
j
t
i
d
0
t
0
N
0
d
i
N
i
t
j
d
j
u
i
N
j
u
j
d
2
t
2
d
0
t
0
N
0
N
3
t
3
N
2
u
1
u
2
d
1
t
1
N
1
d
3
g
2
g
1
d
2
t
2
d
0
t
0
N
0
N
3
t
3
N
2
u
1
u
2
d
1
t
1
N
1
d
3
g
2
g
1
Fig. 8.4 Welded CHS joints
96
(a) Brace failure (yielding, local buckling)
(e) Chord punching shear failure
(b) Weld failure
as (a) but failure in the weld
(f) Chord local buckling
(c) Lamellar tearing
see e.g. Fig. 7.6c
(g) Chord shear failure
(d) Chord plastification (face/wall, thus cross section)
or or
Fig. 8.5 Failure modes for joints between circular hollow sections
1,Ed
joint
joint
N
1
N
1
1,Ed
joint
joint
N
1
N
1
Fig. 8.6 Elastic stress distribution in an X joint
97
B
e
2,5 to 3d
0
1
1
sin
2
N
u
1
1
sin
2
N
u
N
1
N
1
c
1
d
1
1
1
sin
2
N
u
e
1 1
B 2
sin N u
B
e
2,5 to 3d
0
1
1
sin
2
N
u
1
1
sin
2
N
u
N
1
N
1
c
1
d
1
1
1
sin
2
N
u
e
1 1
B 2
sin N u
Fig. 8.7 Ring model
Fig. 8.8 Plastic hinges in the ring model at failure
N
0
N
1
V
p
u
1
N
0
N
1
V
p
u
1
Fig. 8.9 Punching shear model
98
N
0
N
2
u
1
u
2
N
gap,0
V
A
A A
A
N
1
N
0
N
2
u
1
u
2
N
gap,0
V
A
A A
A
N
1
Fig. 8.10 Chord shear model
p 0 1 1 gap,0
2
1 i
p 0 i i 0
N cos N N
N cos N N
+ u =
+ u =
=
A
A
u
1
N
1
Fig. 8.11 Chord load N
0
and chord preload N
0p
Fig. 8.12 Test rig for isolated joint tests
N
0
N
2
N
0p
u
2
A
A
u
1
N
1
N
2
u
2
N
0p
N
0
99
N
1
/
(
f
y
0
t
0
2
)
(mm)
 = 0,60 2 = 40,0 N
1
/
(
f
y
0
t
0
2
)
N
1
/
(
f
y
0
t
0
2
)
(mm)
 = 0,60 2 = 40,0
Fig. 8.13 The effect of the element type on numerical results
0
2
4
6
8
0,0 0,2 0,4 0,6 0,8 1,0

f
(
N
1
u
)
X
j
o
i
n
t
s

 +
7 , 0 1
1
Fig. 8.14 Comparison of experiments with the mean joint resistance function (X joints)
100
0,0
0,2
0,4
0,6
0,8
1,0
1,2
1,0 0,8 0,6 0,4 0,2 0,0 0,2 0,4 0,6 0,8 1,0
2 = 63,5
2 = 25,4
2 = 63,5
2 = 50,8
2 = 25,4
N
1
u
s
i
n
u
1
/
(
f
y
0
t
0
2
Q
u
)
N
0
/ N
pl,0
Fig. 8.15 Chord stress function Q
f
brace i = overlapping member; brace j = overlapped member
d
i
d
0
d
j
t
i
t
0
N
i
N
op
N
j
N
o
u
i
u
j
4.8
4.11
1
4.10
4.1
4.10
t
j
brace i =overlapping member; brace j = overlapped member
brace i = overlapping member; brace j = overlapped member
d
i
d
0
d
j
t
i
t
0
N
i
N
op
N
j
N
o
u
i
u
j
4.8
4.10
4.1
4.10
t
j
4.11
1
brace i = overlapping member; brace j = overlapped member
(2) (2)
(1)
(3)
N
0
brace i = overlapping member; brace j = overlapped member
d
i
d
0
d
j
t
i
t
0
N
i
N
op
N
j
N
o
u
i
u
j
4.8
4.10
4.1
4.10
t
j
4.11
1
brace i =overlapping member; brace j = overlapped member
brace i = overlapping member; brace j = overlapped member
d
i
d
0
d
j
t
i
t
0
N
i
N
op
N
j
N
o
u
i
u
j
4.8
4.10
4.1
4.10
t
j
4.11
1
brace i = overlapping member; brace j = overlapped member
(2) (2)
(1)
(3)
d
i
d
0
d
j
t
i
t
0
N
i
N
op
N
j
N
o
u
i
u
j
4.8
4.10
4.1
4.10
t
j
4.11
1
brace i = overlapping member; brace j = overlapped member
d
i
d
0
d
j
t
i
t
0
N
i
N
op
N
j
N
o
u
i
u
j
4.8
4.10
4.1
4.10
t
j
4.11
1
brace i = overlapping member; brace j = overlapped member
brace i =overlapping member; brace j = overlapped member
d
i
d
0
d
j
t
i
t
0
N
i
N
op
N
j
N
o
u
i
u
j
4.8
4.10
4.1
4.10
t
j
4.11
1
brace i = overlapping member; brace j = overlapped member
(1)
(3)
N
0
(2) (2)
brace i = overlapping member; brace j = overlapped member
Fig. 8.16 Checks for overlap joints
101
T joint efficiency
0,0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1,0
0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1

e
f
f
c
i
e
n
c
y
C
T
2=10
2=15
2=20
2=30
2=40
2=50
1
f
1 1 y
0 0 y
T
1 y 1
Rd , 1
sin
Q
t f
t f
C
f A
N
u
=
X joint efficiency
0,0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1,0
0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1

e
f
f
c
i
e
n
c
y
C
X
2=10
2=15
2=20
2=30
2=40
1
f
1 1 y
0 0 y
X
1 y 1
Rd , 1
sin
Q
t f
t f
C
f A
N
u
=
Fig. 8.17 Design charts for CHS T and Y joints
Fig. 8.18 Design charts for CHS X joints
1
f
1 1 y
0 0 y
K
1 y 1
*
1
sin
Q
t f
t f
C
f A
N
u
=
K gap joint efficiency g'=1
0,0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1,0
0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1

e
f
f
c
i
e
n
c
y
C
K
2=10
2=15
2=20
2=30
2=40
2=50
i
2 1
i
f
i yi
0 0 y
K
yi i
Rd , i
d 2
d d
sin
Q
t f
t f
C
f A
N +
u
=
1
f
1 1 y
0 0 y
K
1 y 1
*
1
sin
Q
t f
t f
C
f A
N
u
=
K gap joint efficiency g'=2
0,0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1,0
0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1

e
f
f
c
i
e
n
c
y
C
K
2=10
2=15
2=20
2=30
2=40
2=50
i
2 1
i
f
i yi
0 0 y
K
yi i
Rd , i
d 2
d d
sin
Q
t f
t f
C
f A
N +
u
=
Fig. 8.19a Design charts for CHS K joints with gap
(g = g/t
0
= 1)
Fig. 8.19b Design charts for CHS K joints with gap
(g = g/t
0
= 2)
1
f
1 1 y
0 0 y
K
1 y 1
*
1
sin
Q
t f
t f
C
f A
N
u
=
K gap joint efficiency g'=5
0,0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1,0
0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1

e
f
f
c
i
e
n
c
y
C
K
2=10
2=15
2=20
2=30
2=40
2=50
i
2 1
i
f
i yi
0 0 y
K
yi i
Rd , i
d 2
d d
sin
Q
t f
t f
C
f A
N +
u
=
1
f
1 1 y
0 0 y
K
1 y 1
*
1
sin
Q
t f
t f
C
f A
N
u
=
K gap joint efficiency g'=10
0,0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1,0
0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1

e
f
f
c
i
e
n
c
y
C
K
2=10
2=15
2=20
2=30
2=40
2=50
i
2 1
i
f
i yi
0 0 y
K
yi i
Rd , i
d 2
d d
sin
Q
t f
t f
C
f A
N +
u
=
Fig. 8.19c Design charts for CHS K joints with gap
(g = g/t
0
= 5)
Fig. 8.19d Design chart for CHS K joints with gap
(g = g/t
0
= 10)
102
9. WELDED JOINTS BETWEEN
RECTANGULAR HOLLOW
SECTIONS
9.1 INTRODUCTION
The most economical and common way to connect
rectangular hollow sections is by direct connection
without any intersecting plates or gussets, as shown
in Fig. 9.1. This also gives the most efficient way for
protection and maintenance.
Joints between rectangular hollow sections can be
easily made, since the connecting members have to
be provided with straight end cuts only.
Although the fabrication is simple, the load transfer is
more complex due to the nonuniform stiffness
distribution in the joints. Due to the flat sides, the
difference in stiffness at the side and at the centre of a
face is even greater than for circular hollow sections.
The general philosophy to identify the various failure
modes and to observe the load transfer has been
described in Chapter 7, but will be discussed here in
more detail for the joints between rectangular and
square hollow sections.
Most failure modes can be related to analytical
models to study the impact of the various influencing
parameters. Based on the analytical models and the
tests carried out, design rules have been established.
9.2 MODES OF FAILURE
Similar to circular hollow section joints, the ultimate
load capacity is based on the maximum in the load
deformation diagram (if the chord deformation is less
than 0,03b
0
) or the load at a deformation of 0,03b
0
of
the chord.
As already indicated in Chapter 7 and shown in Fig.
9.2, the following modes of failure can occur:
 (Local) brace failure (cracking or local buckling)
 Weld failure
 Lamellar tearing
 Chord face plastification
 Chord punching shear failure
 Chord side wall yielding or buckling
 Chord local buckling
 Chord shear failure
Similar to joints between circular hollow sections, to
avoid weld failure, the welds should be stronger than
the connected braces and the throat thickness should
satisfy the same requirements as given in Section
6.5.2. Prequalified full penetration welds can always
be considered to be stronger than the connected
brace members. Partial penetration plus fillet, or fillet
welds alone, can also usually provide a weld
connection as strong as the connected braces.
Also here, the steel should not be susceptible to
lamellar tearing. Especially for larger thicknesses (t >
25 mm), a TTP quality with a low sulphur content
should be used for the chords.
In the design recommendations the widthtowall
thickness ratios b/t have been limited to avoid local
buckling and/or to limit deformations.
As a result, the following failure modes still have to be
considered in design:
 Local brace failure (yielding, local buckling)
 Chord face plastification
 Chord punching shear
 Chord side wall failure
 Chord shear failure
Due to the fact that rectangular hollow sections can be
connected with various orientations and in various
combinations, several failure modes have to be
considered, which makes the checking procedure
more complicated.
In the case of joints between square hollow sections
and within a smaller validity range, the failure modes
to be checked can be limited to one or two.
Local brace failure generally occurs for joints with
relatively thin walled braces and is a general failure
mode for overlap joints.
Chord face plastification is the most common type of
failure for T, Y, X and K and N gap joints with width
ratios < 0,85.
Chord punching shear may occur in joints with
relatively low or high ratios. However, to allow shear
of the chord face, the brace width b
i
should satisfy:
b
i
< b
0
 2
t
0
 2(1,4a)
where a is the weld throat thickness.
Chord side wall failure is a common failure mode for
T, Y and X joints with a ratio close or equal to 1,0.
103
Chord shear may occur in K gap joints with a high
ratio or K gap joints with chords with a low h
0
/b
0
ratio.
9.3 ANALYTICAL MODELS
Similar to circular hollow section joints, analytical
models are used to describe the joint behaviour and to
determine the governing parameters for the joint
strength. Sometimes the joint behaviour is too
complicated to cover all influencing parameters and in
combination with the test results semiempirical
formulae have been developed to describe the joint
strength.
9.3.1 Yield line model
The yield line model, originally developed by the
Danish researcher Johansen for plates, is widely used
for joints between rectangular hollow sections. For
joints with medium ratios, the yield line model gives
a good estimate of the chord face plastification
capacity (Packer, 1978; Wardenier & Giddings, 1986;
Yu, 1997; Zhao, 1992), even under low temperatures
(Bjrk et al., 2003). For very small ratios, the
deformation to realise the yield line pattern may be too
high. For high ratios, the model predicts infinite
strengths and other failure modes will be governing,
e.g. chord side wall failure.
In principle, the yield line method is an upper bound
approach. Therefore, various yield line patterns have
to be examined in order to obtain the lowest capacity.
However, the difference in capacity between the
various yield line patterns is relatively small.
Furthermore, local strain hardening effects and
membrane action are ignored. Therefore, the
simplified yield line pattern shown in Fig. 9.3 (model a)
is generally used for T, Y and X joints instead of the
more complicated pattern shown in Fig. 9.3 (model b).
The principle of the yield line method is based on
equating the work done by the external force N
1
over
a deflection and the internal energy dissipated by
the plastic hinge system with yield lines of length
i
and rotation angles
i
.
p i i 1 1
m sin N = o u (9.1)
where:
0 y
2
0 p
f t
4
1
m = per unit length (9.2)
The energy dissipated in the various yield lines is also
indicated in Fig. 9.3. Equating the sum to the external
work gives:


.

\

u
q
+
o

+ o

= u
1
2
0 0 y
1 1
sin tan
) 1 (
tan
1
t f 2
sin N (9.3)
This is a minimum for:
0
d
dN
1
=
o
(9.4)
or:
 = o 1 tan (9.5)
Substitution of eq. (9.5) in eq. (9.3) gives the capacity:
1 1
2
0 0 y
1
sin
1
1 4
sin
2
1
t f
N
u


.

\

 +
u
q

= (9.6)
In this model, some simplifications have been
incorporated, i.e. the thickness of the sections has
been neglected (b
0
 2t
0
~ b
0
). The same applies to the
weld sizes, which have not been considered. Further,
the effect of the chord load has to be included, which
will be done by a chord load function Q
f
.
For K joints, yield line models can also be used.
However, the load transfer is more complicated since
in the gap area, the stress situation in the yield hinge
is largely influenced by membrane stresses, shear
stresses and work hardening. These effects
complicate the analytical models to such an extent
that semiempirical formulae are used for design.
9.3.2 Punching shear model
Similar to joints between circular hollow sections, the
brace can be pulled out of the chord, resulting in
cracking in the chord by shear around the brace
connection perimeter. Since the stiffness along the
perimeter is non uniform, the deformation capacity of
certain parts may not be sufficient to obtain a full
effective perimeter for punching shear; i.e. only certain
parts can be assumed to be effective for resisting the
punching shear. For example, for a T or Y joint (Fig.
9.4) the sides along the chord walls are the stiffest
part. Depending on the b
0
/t
0
ratio of the chord, a larger
or smaller part along the cross walls will be effective,
designated as b
e,p
.
104
Chord punching shear is caused by the brace load
component perpendicular to the chord face, thus the
punching shear criterion is given by:
1
p , e
1
1
0
0 y
1
sin
1
b 2
sin
h 2
t
3
f
N
u


.

\

+
u
= (9.7)
It will be clear that b
e,p
is a function of b
0
/t
0
. The
smaller b
0
/t
0
, the larger b
e,p
. The value for b
e,p
is
determined experimentally (Wardenier, 1982).
For K gap joints, the gap size is extremely important
for the effective punching shear length. For example,
if the gap size is close to zero and the value is low to
medium (Fig. 9.5a), the gap part is relatively very stiff
compared to the other perimeter parts, resulting in:


.

\

u
+ = u
i
i
i 0
0 y
i i
sin
h
c 2 b t
3
f
sin N with c << 1 (9.8)
For a large gap (Fig. 9.5c) a similar situation occurs
as for T, Y and X joints, thus:


.

\

+
u
= u
p , e
i
i
0
0 y
i i
b 2
sin
h 2
t
3
f
sin N (9.9)
For a gap where the stiffness is about the same as at
the sides of the braces (Fig. 9.5b), the punching shear
criterion becomes:


.

\

+ +
u
= u
p , e i
i
i
0
0 y
i i
b b
sin
h 2
t
3
f
sin N (9.10)
Neglecting the thickness and the weld sizes, the gap
value here has to satisfy:
2
b b
2
g
i 0
~ or  ~ 1
b
g
0
Due to the deformation capacity of the material, which
has been proved experimentally, this limit can be
extended to:
( ) (  s s  1 5 , 1
b
g
1 5 , 0
0
) (9.11)
For  values close to 1,0 these limits are impractical
and a minimum gap g = t
1
+ t
2
for welding is required.
9.3.3 Local brace failure model
The model to describe local brace failure (Fig. 9.6)
has a certain relationship with the chord punching
shear model. Due to the non uniform stiffness along
the connection perimeter, both models consider an
effective part, although due to the different
deformation capacities for failure in the brace and
chord punching shear, the values for b
e
and b
e,p
are
different.
Furthermore, chord punching shear is caused by the
brace load component perpendicular to the chord,
whereas for local brace failure, the brace load is
taken. The effect of the angle has not yet been
defined clearly and has been conservatively excluded
up to now.
For a T, Y and X joint, the criterion for local brace
failure can thus be given by:
N
1
= f
y1
t
1
(2h
1
+ 2b
e
 4t
1
) (9.12)
The term 4t
1
has to be included to avoid the corners
being counted twice. Similar to the punching shear
criterion, the effective width b
e
is determined
experimentally (Wardenier, 1982) and becomes larger
if b
0
/t
0
decreases.
For K joints with gap, the same applies as for the
punching shear criterion, i.e. the gap size should
satisfy eq. (9.11) for having a full effective cross wall
of the brace at the gap:
N
i
= f
yi
t
i
(2h
i
+ b
i
+ b
e
 4t
i
) (9.13)
This criterion is also directly applicable to overlap
joints for the overlapping brace, see Chapter 11.
9.3.4 Chord side wall bearing or buckling
model
T, Y and X joints with a high ratio generally fail by
yielding or buckling of the chord side walls, shown in
Fig. 9.8. The model used is similar to that used for
beamtocolumn joints between I sections. For joints
with = 1,0, the capacity can be easily determined by:
1
0
1
1
0 0 y 1
sin
1
t 5
sin
h
t f 2 N
u


.

\

+
u
= (9.14)
For slender walls, the yield stress f
y0
is replaced by a
buckling stress f
k
which depends on the chord web
105
slenderness h
0
/t
0
. This model is simple and gives a
larger margin of safety for high slendernesses.
A model which is in better agreement with the test
results for all slendernesses is based on the "4 hinge
yield line" mechanism, shown in Fig. 9.9 dating back
to the 1970s (Packer, 1978). In the case of
compression of the chord side walls, Yu (1997) also
used a buckling stress, but with a buckling length of
(h
0
 2t
0
)/2.
9.3.5 Chord shear model
Similar to circular hollow section joints, this model,
shown in Fig. 9.10, can be described by the basic
formula for plastic design. The plastic shear load
capacity is given by:
v
0 y
0 , pl
A
3
f
V = (9.15)
In principle, the webs are effective for shear, but if the
gap is small, a part of the top flange may also be
effective, thus:
(
0 0 0 v
t b h 2 A o + = ) (9.16)
The coefficient depends on the g/t
0
ratio and can be
easily determined based on plastic analysis
(Wardenier, 1982).
The remaining cross section has to transmit the axial
force. Using the HuberHenckyVon Mises criterion,
the following interaction formula can be derived:
( )
2
0 , pl
0 , gap
0 y v 0 y v 0 0 , gap
V
V
1 f A f A A N


.

\

+ s (9.17)
This formula is comparable to that for circular hollow
section joints, eq. (8.10).
9.4 EXPERIMENTAL AND NUMERICAL
VERIFICATION
Based on the models and test evidence, either
analytical or semiempirical formulae were developed.
For example, for T, Y and X joints, the yield line model
is used as a lower bound for the test results and also
incorporated in the recommendations (EN 199318,
2005; IIW, 1989, 2009; Packer, 1978; Packer &
Henderson, 1997; Packer et al., 1992, 2009a;
Wardenier, 1982), whereas for K joints with gap, a
semiempirical formula is used. As an example, Fig.
9.11 shows a comparison between the experiments
and the formula for the strength of K gap joints,
established by Wardenier (1982), which was included
in the previous recommendations of IIW (1989) and
CIDECT Design Guide No. 3 (Packer et al., 1992).
In the last 20 years, the results of many numerical
investigations became available, e.g. Lu (1997), Yu
(1997), Liu & Wardenier (2001, 2003, 2006), Kosteski
& Packer (2003), Wardenier et al. (2007a, 2007b).
Furthermore the recommendations had to be
extended to steel S460, which made a revision of the
recommendations and the validity ranges necessary
(Liu & Wardenier, 2004).
9.5 BASIC JOINT STRENGTH
FORMULAE
9.5.1 T, Y, X joints and K, N gap joints
For T, Y and X joints up to = 0,85, the yield line
model for chord face plastification is used as a basic
lower bound formula for the joint resistance (Fig.
9.12).
Above = 0,85, the joint resistance is governed either
by chord side wall failure, brace failure or by chord
punching shear, for b
1
< b
0
 2
t
0
 2(1,4a).
For K gap joints, initially the semiempirical formula of
Wardenier (1982) based on chord face plastification
was used as the primary criterion. However, the
ultimate strength data which served as a basis for this
equation did not consider the deformation limit of 3%
b
0
and only included chord sections of b
0
/t
0
s 35.
Hence, a modification has recently been carried out
(see Section 9.6).
Depending on the joint parameters, various other
criteria may become critical for K gap joints, e.g. local
brace failure, chord punching shear or chord shear.
9.5.2 K, N overlap joints
As mentioned before, for overlap joints the same
approach is adopted for all types of overlap joints,
regardless of whether circular or rectangular braces
are used in combination with a circular, rectangular or
an open section chord (see Chapter 11).
The resistance of overlap joints between rectangular
106
hollow sections with 25% s Ov s 100% overlap is
based on the following criteria (Chen et al., 2005; Liu
et al., 2005; Wardenier & Choo, 2006):
(1) Local failure of the overlapping brace
(2) Local chord member yielding at the joint location
based on interaction between axial load and
bending moment
(3) Shear of the connection between the brace(s) and
the chord
Fig. 9.13 shows the overlap joint configuration with the
cross sections to be examined for these criteria. For K
and N overlap joints, the subscript i is used to denote
the overlapping brace member, while the subscript j
refers to the overlapped brace member.
Local failure of the overlapping brace (criterion 1)
should always be checked, while shear between the
braces and the chord (criterion 3) may only become
critical for larger overlaps, i.e. larger than 60% or
80%, depending on whether or not the hidden toe
location of the overlapped brace is welded to the
chord. The check for local chord member yielding
(criterion 2) is, in principle, a member check and may
become critical for larger overlaps and/or larger
ratios.
Joints with overlaps between 0% and 25% should be
avoided because in those cases, the stiffness of the
connection between the overlapping brace and the
overlapped brace is much larger than that of the
overlapping brace to chord connection, which may
lead to premature cracking and lower capacities
(Wardenier, 2007).
More detailed information regarding the design
equations for overlap joints is given in Chapter 11.
9.6 EVALUATION TO DESIGN RULES
In principle, the evaluation to design rules, e.g. for K
joints with gap, is similar to that described for joints
between circular hollow sections.
For T, Y and X joints, a lower bound analytical yield
line criterion is used for chord face plastification.
Therefore, no statistical evaluation has been carried
out.
Considering Lus deformation limit of 3% b
0
(Lu et al.,
1994), and extending the validity range to class 1 and
2 sections with b
0
/t
0
s 40 and h
0
/t
0
s 40 instead of an
upper limit of 35, for K gap joints, the functions for
and needed to be modified in such a way that the
capacity for large b
0
/t
0
ratios would be reduced.
Changing the original function 8,9
0,5
, used in the
previous editions of IIW (1989) and CIDECT Design
Guide No. 3 (Packer et al., 1992), into 14
0,3
gives
the same capacity for b
0
/t
0
= 20 and a reduction for
large b
0
/t
0
ratios. The new expression, included in IIW
(2009) and CIDECT Design Guide No. 3 (Packer et al.,
2009a) is a reasonable compromise between covering
the data determined with the 3% b
0
deformation limit,
extension of the validity range and backup by previous
analyses (Packer & Haleem, 1981; Wardenier, 1982).
Since in joints of rectangular hollow sections the
sections can have different orientations and
depthtowidth ratios, many configurations are
possible, resulting in a large number of failure modes
and related strength formulae, see Table 9.1.
However, with a smaller range of validity, the design
formulae for joints between square hollow sections
can be reduced to such an extent that only a single
check has to be carried out (see Table 9.2).
Fleischer & Puthli (2008) initiated research on how to
deal with RHS joints with gaps and/or chord
widthtothickness ratios outside the current range of
validity.
9.7 OTHER TYPES OF JOINTS OR
OTHER LOAD CONDITIONS
9.7.1 Related types of joints
As for circular hollow section joints, various joint
configurations exist for which the resistance can be
directly related to the basic types presented in Tables
9.1 and 9.2.
Table 9.3 gives the design resistance for some special
types of RHS uniplanar joints with braces directly
welded to the chord; notice the similarity with Table
8.2 for circular hollow section joints.
9.7.2 Joints between circular braces and
a rectangular chord
As far as chord face plastification is concerned, the
strength of a joint with a circular hollow section brace
with diameter d
i
is about /4 times that of a joint with a
square hollow section brace with a width b
i
= d
i
, see
Fig. 9.14 (Wardenier, 1982; Packer et al., 2007). As a
consequence, the same formulae can be used as for
square hollow section joints, but the resistances have
107
to be multiplied by /4. This also means that the joints
have the same efficiency, i.e. the joint strength divided
by the squash load of the brace.
9.7.3 Joints between plates or I sections
and RHS chords
Joints between plates or I sections and RHS chords
are approached in a similar manner as the rectangular
hollow section joints and in principle the same modes
of failure have to be considered. Within the scope of
this book, these joints are not further discussed, but
reference is made to Lu (1997), Packer et al. (2009a)
and Chapter 12. The design resistances are given in
Table 9.4.
Tee joints to the ends of RHS members
When an axial force is applied to an RHS member, via
a welded Tee joint as shown in Fig. 9.15, the capacity
is determined by local failure of the RHS walls or the
Tee web.
For a commonly used distribution slope of 2.5:1 from
each face of the Tee web (Kitipornchai & Traves,
1989), the dispersed load width is (5t
p
+ t
w
). A
conservative assumption is to use this effective width
at two sides of the RHS member. Thus, the resistance
of the RHS can be computed by summing the
contributions of the parts of the RHS cross sectional
area into which the load is distributed:
1 y 1 p w 1 y1 Rd , 1
f A ) 5t (t t f 2 N s + = (9.18)
A similar load dispersion can be assumed for the
capacity of the Tee web. If the Tee web has the same
width as the width of the cap plate, i.e. (h
1
+ 2s), the
capacity of the Tee web is:
) s 5t , 2 (t t f 2 N
p 1 w yw Rd , 1
+ + = (9.19a)
) 5t (t t f 2
p 1 w yw
+ s (9.19b)
In eqs. (9.18) and (9.19), the size of any weld legs to
the Tee web has been conservatively ignored.
Gusset platetoslotted RHS joints
Single gusset plates, slotted into the ends of hollow
section members and concentrically aligned with the
axis of the member, as shown in Fig. 9.16, are
commonly found in diagonal brace members of steel
framed buildings.
As a consequence of only part of the RHS cross
section being connected, an uneven stress distribution
around the RHS perimeter occurs during load transfer
at the connection. This phenomenon, known as shear
lag, is illustrated in Fig. 9.16.
The possible failure modes for the gusset
platetoslotted RHS joints loaded in tension are
circumferential failure of the RHS and tear out or
"block shear" failure of the RHS. Shear lag is
principally influenced by the weld length, L
w
in relation
to the dimension w which is the distance between the
welds measured from plate facetoplate face, around
the perimeter of the RHS.
For long weld lengths, shear lag effects become
negligible, while for short weld lengths, tear out
governs over circumferential fracture of the RHS.
However, if L
w
= 1,65b for square braces and 1,3d for
circular braces, it can be assumed that the capacity is
equal to that of the connected hollow section or plate.
Detailed design rules are given by Packer et al.
(2009a).
9.7.4 Multiplanar joints
Compared to uniplanar joints, multiplanar joints have
a geometric effect and a loading effect to be
considered.
It is plausible that a multiplanar joint has a geometric
influence only if the value is large, because then the
chord side wall is stiffened, see e.g. Fig. 9.17 for an
XX joint.
For multiplanar joints of rectangular hollow sections,
the tendency of the loading effect is similar but less
pronounced compared to that of joints of circular
hollow sections, see Fig. 9.18.
Extensive analytical and numerical research by Liu &
Wardenier (2001, 2003) showed that the differences
in capacity between uniplanar K gap and multiplanar
KK gap joints are caused by the larger chord force
acting in multiplanar joints. Based on this work, the
following design recommendations are given for
multiplanar KK joints (see Table 9.5).
Multiplanar KK gap joints (Fig. 9.19)
 For chord face plastification (small or medium ),
the strength of the joint can be based on the joint
resistance formulae for uniplanar joints given in
Tables 9.1 and 9.2, and no further multiplanar
correction is necessary, provided that the actual,
total chord force is used for the chord stress function
Q
f
.
108
 For large  ratios or rectangular chord sections, the
strength of a KK gap joint is governed by chord
shear and chord axial force interaction, presented in
Table 9.5. The KK gap joint (with  = 90) is
subjected to a shear force of
Ed gap,0,
V 2 0,5 in each
plane, where V
gap,0,Ed
is the total "vertical" shear
force. The shear force in each plane is resisted by
the two walls of the RHS chord.
109
Multiplanar overlap KK joints
 For multiplanar overlap KK joints, the strength of the
joint is similar to that for uniplanar overlap joints
given in Chapter 11. Thus, compared to the
previous IIW (1989) recommendations, a brace
shear criterion and a local chord yielding criterion
have been added.
9.7.5 Joints loaded by brace bending
moments
The design resistances for joints loaded by brace
bending moments are derived in a similar way to that
for axially loaded joints. To simplify the design,
limitations are also given here for the range of validity
to reduce the criteria to be checked.
For Vierendeel girders it is recommended to choose
joints with  = 1,0 to provide sufficient stiffness and
strength.
The design resistance formulae are based on the
analyses of Wardenier (1982), Mang et al. (1983), Yu
(1997) and Packer et al. (2009a), and are given in
Table 9.6.
9.7.6 Interaction between axial loads and
bending moments
For joints with brace members subjected to combined
loading, the effect of axial load on the joint moment
capacity depends on the critical failure mode, and
hence a complex set of interactions exists.
Consequently, it is conservatively proposed to use a
linear interaction relationship:
0 , 1
M
M
N
N
Rd , 1 , ip
Ed , 1 , ip
Rd , 1
Ed , 1
s + (9.20)
9.8 DESIGN CHARTS
In Figs. 9.20 to 9.23, the joint resistances are
expressed in terms of the efficiency of the connected
braces in a similar way to that for circular hollow
section joints, i.e. the joint resistance is given as a
fraction of the yield capacity A
i
f
yi
of the connected
brace. This results in the following efficiency formula:
i
f
i yi
0 0 y
e
yi i
Rd , i
sin
Q
t f
t f
C
f A
N
Efficiency
u
= = (9.21)
In the case of
2 1
b b = for K joints, eq. (9.21) has to be
multiplied by
i
2 1
2b
b b +
, where b
i
is the width of the
brace considered.
For a detailed explanation, see Section 8.8.
Using the chart of Fig. 9.23 shows that e.g. a K gap
joint with 2 ~ 20 and b
1
= b
2
gives an efficiency
parameter C
K
~ 0,37. Thus, for an angle
i
= 45, a
100% efficiency can be obtained if:
9 , 1
t f
t f
i yi
0 0 y
> (for Q
f
= 1,0)
If the chord load effect Q
f
is included, this ratio should
be slightly larger. Figs. 9.24 and 9.25 show the chord
load effect Q
f
as a function of the parameter n, defined
as the ratio between the maximum stress in the
connecting chord face and the chord yield stress.
9.9 CONCLUDING REMARKS
For more detailed information about joints loaded by
bending moments as well as special types of joints,
reference is made to the appropriate literature, see
Dutta (2002), Korol et al. (1977), Packer & Henderson
(1997), Packer et al. (2009a), Ono et al. (1991), Syam
& Chapman (1996), Wardenier (1982) and Wardenier
& Giddings (1986).
Table 9.1 Design axial resistances of welded joints between RHS or CHS braces and RHS chord
Type of joint Design limit state
T, Y and X joints Chord face plastification (for s 0,85)
f
1 1
2
0 0 y
Rd , 1
Q
1
4
n si ) 1 (
2
sin
t f
N


.

\


+

q
u
=
Local brace failure (general check)
) t 4 b 2 h 2 ( t f N
1 e 1 1 y1 Rd 1,
+ =
Chord punching shear (for b
1
s b
0
 2t
0
)


.

\

+
u
=
p , e
1
1
1
0 y0
Rd 1,
b 2
sin
h 2
sin
t f 0,58
N
Chord shear (for X joints, if cos
1
> h
1
/h
0
)
See chord shear equations for K gap joints, but with V
0,Ed
instead of V
gap,0,Ed
Chord side wall failure (for  = 1,0)
(1)
t
0
t
1
d
1
h
1
b
1
h
0
b
0
N
1
u
1
f 0
1
1
1
0 k
Rd 1,
Q t 10
sin
2h
sin
t f
N


.

\

+
u
=
K gap joints Chord face plastification (general check)
f
i
2
0 y0 3 , 0
Rd i,
Q
sin
t f
4 1 N  =
Local brace failure (general check)
) t 4 b b h 2 ( t f N
i e i i i yi Rd i,
+ + =
Chord punching shear (for b
i
s b
0
 2t
0
)


.

\

+ +
u
=
p , e i
i
i
i
0 y0
Rd i,
b b
sin
h 2
sin
t f 0,58
N
Chord shear (general check)
g
d1
u1 u2
t
0
t
2
2
1
+e
0
h
1
b1
d
2
h
2
b
2
h
0
b
0
t
1
N
0
N
1
N
2
i
v y0
Rd i,
sin
A f 0,58
N = and
2
Rd pl,0,
Ed gap,0,
y0 v y0 v 0 Rd gap,0,
V
V
1 f A f ) A (A N


.

\

+ =
A
v
and V
pl,0,Rd
v y0 Rd pl,0,
A f 0,58 V =
T, Y and X joints
0 0 v
t 2h A =
K gap joints
0 0 0 0 v
t b t 2h A o + =
RHS braces:
) t 3 /( ) g 4 ( 1
1
2
0
2
+
= o
CHS braces: 0 = o
Function Q
f
( )
1
C
f
n 1 Q = with
Rd pl,0,
Ed 0,
Rd pl,0,
Ed 0,
M
M
N
N
n + = in connecting face
Chord compression stress (n < 0) Chord tension stress (n > 0)
T, Y and X joints C
1
= 0,6 0,5
K gap joints C
1
= 0,5 0,5 but > 0,10
C
1
= 0,10
b
e
and b
e,p
i i
i yi
0 0 y
0 0
e
b but b
t f
t f
/t b
10
b s


.

\



.

\

=
i i
0 0
p , e
b but b
/t b
10
b s


.

\

=
Tension:
0 y k
f f = Compression: for T and Y joints, and for X joints
0 y k
f f _ =
1 y0 k
sin f 8 , 0 f _ =
f
k
where _ = reduction factor for column buckling according to e.g. Eurocode 3 (EN 199311, 2005) using
the relevant buckling curve and a slenderness
1 0
0
sin
1
2
t
h
3,46


.

\

=
(1)
For 0,85 <  < 1,0 use linear interpolation between the resistance for chord face plastification at  = 0,85 and the
resistance for chord side wall failure at  = 1,0.
110
Table 9.1 Design axial resistances of welded joints between RHS or CHS braces and RHS chord
(continued)
T, Y, X and K gap joints with CHS
brace
For CHS braces, multiply the above resistances by t/4 (except for chord shear
criterion) and replace b
i
and h
i
by d
i
(i = 1 or 2)
Range of validity
T, Y or X joints K gap joints
RHS braces 0,25 but /t 0,01b 0,1 /b b
0 0 0 i
> + >
Bracetochord
ratio
CHS braces
0 0 0 i
/t 0,01b 0,1 /b d + > and 80 , 0 /b d 0,25
0 i
s s
Compression 2 or 1 class
(2)
and 40 /t h and 40 /t b
0 0 0 0
s s
RHS chord
Tension 40 /t h and 40 /t b
0 0 0 0
s s
Compression 2 or 1 class
(2)
and 40 /t h and 40 /t b
i i i i
s s
RHS braces
Tension 40 /t h and 40 /t b
i i i i
s s
Compression 2 or 1 class
(2)
and 50 /t d
i i
s
CHS braces
Tension 50 /t d
i i
s
Gap N/A
) 1 ( 5 , 1 g/b ) 1 ( 5 , 0
0
 s s 
(3)
and
2 1
t t g + >
Eccentricity N/A
0
h 25 , 0 e s
Aspect ratio 0 , 2 /b h 5 , 0
i i
s s
Brace angle
30
i
> u
Yield stress
y0 yi
f f s
(4)
u y
f 8 , 0 f s
2
y
N/mm 460 f s
(2)
Section class limitations are given in Table 2.7.
(3)
For , check the joint also as two separate T or Y joints. ) 1 ( 5 , 1 g/b
0
 >
(4)
For 355 N/mm
2
< f
y
s 460 N/mm
2
, use a reduction factor of 0,9 for the design resistances.
111
Table 9.2 Design axial resistances of welded joints between square or circular braces and a square
hollow section chord
Joints between square hollow sections Design limit state
T, Y and X joints Chord face plastification
t
0
t
1
d
1
h
1
b
1
h
0
b
0
N
1
u
1
f
1 1
2
0 0 y
Rd , 1
Q
1
4
n si ) 1 (
2
sin
t f
N


.

\


+

q
u
=
K gap joints Chord face plastification
g
d1
u1 u2
t
0
t
2
2
1
+e
0
h
1
b
1
d
2
h
2
b
2
h
0
b
0
t
1
N
0
N
1
N
2
f
i
2
0 y0 3 , 0
Rd i,
Q
sin
t f
4 1 N  =
Function Q
f
Same as in Table 9.1
T, Y, X and K gap joints with CHS brace
For CHS braces, multiply the above resistances by t/4 and replace b
i
by d
i
(i = 1 or 2)
Range of validity
General Same as in Table 9.1 with additional limits given below
T, Y and X joints 85 , 0 /b b
0 1
s
SHS braces
K gap joints 3 , 1 ) )/(2b b (b 6 , 0
i 2 1
s + s 15 /t b
0 0
>
CHS braces K gap joints 3 , 1 ) )/(2d d (d 6 , 0
i 2 1
s + s 15 /t b
0 0
>
112
Table 9.3 Design resistance of related types of joints
Type of joint Relationship to the formulae of Tables 9.1 and 9.2
All brace member forces act in the same
sense (compression or tension).
N
1
N
1
u
1
u
1
N
1,Ed
s N
1,Rd
with N
1,Rd
from X joint given in Table 9.1 or 9.2
All brace member forces act in the same
sense (compression or tension).
N
2
N
1
N
1
N
2
u
2
u
1
N
1,Ed
sin
1
+ N
2,Ed
sin
2
s N
X,Rd
sin
X
with N
X,Rd
sin
X
from X joint given in Table 9.1 or 9.2, being the larger of the
values for brace 1 or 2
Forces in members 1 are in compression
and members 2 in tension.
1
1
N
2
N
2
N
1
N
1
u
2
u
1
N
i,Ed
s N
i,Rd
(i = 1 or 2)
with N
i,Rd
from K joint given in Table 9.1 or 9.2, but with the actual chord force
Forces in members 1 are in compression
and members 2 in tension.
N
i,Ed
s N
i,Rd
(i = 1 or 2)
with N
i,Rd
from K gap joint given in Table 9.1 or 9.2
Note: In a gap joint, the chord cross section in the gap has to be checked for
shear failure:
v y0 Rd , 0 , pl Ed gap,0,
A f 0,58 V V = s
2
Rd , 0 , pl
Ed , 0 , gap
y0 v y0 v 0 Rd , 0 , gap Ed , 0 , gap
V
V
1 f A f ) A (A N N


.

\

+ = s
1
1
N
2
N
1
N
1
N
2
u
2
u
1
113
Table 9.4 Design resistances of uniplanar platetoRHS joints
Type of joint Design limit state
T and X joints transverse plate Chord face plastification (for 0,4 s s 0,85)
f
2
0 y0 Rd , 1
Q
9 , 0 1
8 , 2 2
t f N
(
(

 +
=
Chord punching shear (for 0,85b
0
s b
1
s b
0
 2t
0
)
) 2b t 2 ( t f 0,58 N
p e, 1 0 y0 Rd , 1
+ =
Chord side wall failure (for 1,0)
(1)
f 0 1 0 y0 Rd , 1
Q ) 5t t ( t f 2 N + =
Local plate failure (for all )
t
1
h
0
N
1
t
0
b
1
b
0
t
1
h
0
N
1
N
1
t
0
b
1
b
0
e 1 1 y Rd , 1
b t f N =
T and X joints longitudinal plate Chord face plastification
h
1
h
0
t
1
b
0
t
0
N
1
h
1
h
0
t
1
t
0
b
0
N
1
N
1
f
0
1 2
0 0 y Rd , 1
Q
b
t
1 2 t f 2 N
(
(
+ q =
(1)
For 0,85 <  < 1,0, use linear interpolation between the resistance for chord face plastification at  = 0,85 and the
resistance for chord side wall failure at  = 1,0.
114
Table 9.4 Design resistances of uniplanar platetoRHS joints (continued)
Type of joint Design limit state
T joints longitudinal throughplate Chord face plastification
h
1
h
0
t
1
b
0
t
0
N
1
f
0
1 2
0 0 y Rd , 1
Q
b
t
1 2 t f 4 N
(
(
+ q =
T stub joints stiffened longitudinal plate
h
1
h
0
t
1
t
0
N
1
b
0
bsp
t
sp
*
3
0 sp
e t 5 , 0 t

>
0 0
1 sp *
t b
t b
: with
= 
If t
sp
fulfills the above requirement, the joint can be regarded as an
RHStoRHS T joint. In the design equations for RHStoRHS T joints, the
stiffening plate width b
sp
is then used for the brace width b
1
.
Function Q
f
( )
1
C
f
n 1 Q = with
Rd pl,0,
Ed 0,
Rd pl,0,
Ed 0,
M
M
N
N
n + = in connecting face
Chord compression stress (n < 0) Chord tension stress (n > 0)
Transverse plate C
1
= 0,03 but > 0,10
Longitudinal plate C
1
= 0,20
C
1
= 0,10
b
e
and b
e,p
i i
i yi
0 0 y
0 0
e
b but b
t f
t f
/t b
10
b s


.

\



.

\

=
i i
0 0
p , e
b but b
/t b
10
b s


.

\

=
Range of validity
Compression 2 or 1 class
(2)
and 40 /t h and 40 /t b
0 0 0 0
s s
Tension 40 /t h and 40 /t b
0 0 0 0
s s RHS chord
Aspect ratio 0 , 2 /b h 5 , 0
0 0
s s
Transverse plate 4 , 0 /b b
0 1
> = 
Longitudinal plate 4 /b h 1
0 1
s = q s
Plate angle
o
1
90 ~ u
Yield stress
y0 y1
f f s
(3)
u y
f 8 , 0 f s
2
y
N/mm 460 f s
(2)
Section class limitations are given in Table 2.7.
(3)
For 355 N/mm
2
< f
y
s 460 N/mm
2
, use a reduction factor of 0,9 for the design resistances.
115
Table 9.5 Correction factors for the design resistance of multiplanar joints
Type of joint Correction factor to uniplanar joint strength
TT joints General
Members 1 may be either in tension or
compression
2N
1

N
1 N
1
= 1,0
XX joints Chord face plastification (for  s 0,85)
Ed 1,
Ed 2,
N
N
35 , 0 1+ =
Notes:
 Take account of the sign of N
1,Ed
and N
2,Ed
, with N
2,Ed
 s N
1,Ed

 N
2,Ed
/N
1,Ed
is negative if the members in one plane are in tension and in the
other plane in compression.
Other failure modes
Members 1 and 2 can be either in
compression or tension
N
1
N
2
N
1
N
1
N
1
N
2
= 1,0
KK gap and overlap joints General
Members 1: compression
Members 2: tension
A
A

N
1 N
1
N
1
N
2
=1,0
but in a KK gap joint, the chord cross section in the gap has to be checked
for shear failure according to:
0 , 1
V
V 71 , 0
N
N
2
Rd , 0 , pl
Ed , 0 , gap
2
Rd , 0 , pl
Ed , 0 , gap
s


.

\

+


.

\

where:
Ed gap,0,
N e in gap axial forc design =
Rd pl,0,
N
y0 0
f A =
Ed gap,0,
V in gap force hear s design =
Rd , 0 , pl
V for a square hollow section chord ) A 5 , 0 ( f 58 , 0
0 0 y
=
Range of validity
Same as in Tables 9.1 and 9.2
 90
116
Table 9.6 Design moment resistances of welded joints between rectangular hollow sections
Type of joint brace loading Design limit state
T and X joints inplane bending Chord face plastification (for s 0,85)
f 1
2
0 y0 Rd , 1 , ip
Q
1
1
2
2
1
h t f M


.

\


q
+

+
q
=
Local brace failure (for 0,85 <  s 1,0)
(
=
1 1 1 1
1
e
1 , pl 1 y Rd , 1 , ip
t ) t h ( b )
b
b
1 ( W f M
Chord side wall failure (for  = 1,0)
(1)
t
0
t
1
h
1
b
1
h
0
b
0
1
u
M
ip,1
( )
f
2
0 1 0 k Rd , 1 , ip
Q 5t h t 0,5f M + =
T and X joints outofplane bending
(2)
Chord face plastification (for s 0,85)
f
1
1
1
2
0 y0 Rd , 1 , op
Q
) 1 (
) 1 ( 2
) 1 ( b 2
) 1 ( h
b t f M


.

\

 
 +
+

 +
=
Local brace failure (for 0,85 <  s 1,0)
( ) ] b b t 5 , 0 W [ f M
2
e 1 1 1 , pl 1 y Rd , 1 , op
=
Chord side wall failure (for  = 1,0)
(1)
t
0
t
1
h
1
b
1
h
0
b
0
1
u
M
op,1
f 0 1 0 0 0 k Rd , 1 , op
Q ) 5t (h ) t (b t f M + =
Function Q
f
Same as in Table 9.1
b
e 1 1
1 y1
0 y0
0 0
e
b but b
t f
t f
/t b
10
b s


.

\



.

\

=
Brace inplane bending Brace outofplane bending
T and Y joints X joints T and Y joints X joints
0 y k
f f =
0 y k
f 8 , 0 f _ =
0 y k
f f _ =
0 y k
f 8 , 0 f _ =
f
k
where _ = reduction factor for column buckling according to e.g. Eurocode 3
(EN 199311, 2005) using the relevant buckling curve and a slenderness


.

\

= 2
t
h
46 , 3
0
0
Range of validity Same as in Table 9.1, but with
1
90
(3)
(1)
For 0,85 <  < 1,0, use linear interpolation between the resistance for chord face plastification at  = 0,85 and the
resistance for chord side wall failure at  = 1,0.
(2)
Chord distortion to be prevented for brace outofplane bending.
(3)
The equations are conservative for
1
< 90.
117
T joint Y joint
N
1
b
1
h
1
t
1
u
1
t
0
h
0
b
0
N
1
b
1
h
1
t
1
u
1
t
0
h
0
b
0
N
1
b
0
N
0
h
1
b
1
t
1
u
1
t
0
h
0
N
1
b
0
N
0
h
1
b
1
t
1
u
1
t
0
h
0
X joint K joint with gap
N
1
h
0
b
0
b
1
h
1
N
0
t
1
u
1
t
0
N
1
N
1
h
0
b
0
b
1
h
1
N
0
t
1
u
1
t
0
N
1
N
1
b
0
h
0
N
0
h
2
N
2
b
2
g t
1
u
1
t
0
t
2
u
2
b
1
h
1
N
1
b
0
h
0
N
0
h
2
N
2
b
2
g t
1
u
1
t
0
t
2
u
2
b
1
h
1
N joint with overlap KT joint with gap
b
0
h
0
N
0
N
j
h
i
b
j
h
j
N
i
b
i
t
j
t
i
t
0
u
i
u
j
b
0
h
0
N
0
N
j
h
i
b
j
h
j
N
i
b
i
t
j
t
i
t
0
u
i
u
j
N
1
b
0
h
0
N
0
h
3
t
3
t
2
b
2
h
2
u
2
u
1 t
0
h
1
b
1
t
1
b
3
g
1
g
2
N
3
N
2
N
1
b
0
h
0
N
0
h
3
t
3
t
2
b
2
h
2
u
2
u
1 t
0
h
1
b
1
t
1
b
3
g
1
g
2
N
3
N
2
Fig. 9.1 Welded RHS joints
118
(a) Brace failure
(e) Chord punching shear failure
(b) Weld failure
as (a) but failure in the weld
(f) Chord side wall yielding or buckling
(c) Lamellar tearing
see e.g. Fig. 7.6c
(g) Chord local buckling
(d) Chord face plastification
(h) Chord shear failure
Fig. 9.2 Failure modes for welded RHS joints
119
The total energy dissipated in the yield lines 1 to 5 is as follows:
Yield lines 1:
p
1 0
0
m
cot ) b b (
2
b 2
o
o
=
p
m
) 1 (
tan 4
o

o
Yield lines 2:
p
1 0
1
m
cot ) b b (
2
b 2
o
o
=
p
m
) 1 (
tan 4
o

o 
Yield lines 3:
p
1 0
1 0
1
1
m
) b b (
2
) cot
2
b b
2
sin
h
( 2
o
o
+
u
=
p
1
m cot 4
sin ) 1 (
4
o


.

\

o +
u 
q
Yield lines 4:
p
1 0 1
1
m
) b b (
2
)
sin
h
( 2
o
u
=
p
1
m
sin ) 1 (
4
o


.

\

u 
q
Yield lines 5:
p
5 5
5
m
cot tan
4


.

\

o
o
+
o
o
=
p
m ) cot (tan 4 o o + o
with:
4
t f
m
0
2
0 y
p
=
Total energy E
d
=
)
`
u
q
+
o

+ o

o
1
p
sin tan
) 1 (
tan
) 1 (
m 8
Fig. 9.3 Yield line model for a T, Y and X joint
120
0,5b
e,p
0,5b
e,p
1
1
sin
h
u
) b 2
sin
h
( 2 L
p , e
1
1
eff
+
u
=
0,5b
e,p
0,5b
e,p
1
1
sin
h
u
) b 2
sin
h
( 2 L
p , e
1
1
eff
+
u
=
Fig. 9.4 Chord punching shear model for a T, Y or X joint
,
,
0,5b
e,p 2
2
sin
h
u
0,5b
e,p
,
,
0,5b
e,p 2
2
sin
h
u
0,5b
e,p
Fig. 9.5 Chord punching shear model for a K gap joint (chord face)
121
0,5b
e
0,5b
e
0,5b
e
0,5b
e
Fig. 9.6 Local brace failure model for a T, Y or X joint
N
j
N
i
b
i
h
i
0,5b
e,ov
0,5b
e,ov
N
j
N
i
b
i
h
i
0,5b
e,ov
0,5b
e,ov
Fig. 9.7 Local brace failure model for a 100% overlap joint
122
,
,
,
,
Fig. 9.8 Chord side wall failure model
Fig. 9.9 Four hinge yield line model for chord side wall failure (Packer, 1978)
Fig. 9.10 Chord shear failure model
123
0 100 200 300 400 500 600
500
400
300
200
100
0
0 100 200 300 400 500 600
500
400
300
200
100
0
Fig. 9.11 Comparison between experiments and the mean ultimate joint strength equation for chord
plastification for K gap joints with s 0,85 (Wardenier, 1982)
0 50 100 150 200 250
250
200
150
100
50
0
0 50 100 150 200 250
250
200
150
100
50
0
Fig. 9.12 Comparison between experiments and the analytical yield line criterion for chord plastification for
T, Y and X joints with s 0,85 (Wardenier, 1982)
124
h
j
h
i
b
j
b
i
N
i
N
j
t
i
t
j
i = overlapping member and j = overlapped member
Fig. 9.13 Overlap joint configuration with cross sections to be checked
Fig. 9.14 Comparison of a K joint with a circular brace and an equivalent joint with a square brace
N1
N1
tp
tw
t1
b1
s
h1
s
5tp+tw
2.5
1
Fig. 9.15 Load dispersion for a Tee joint on the end of an RHS member
t
0
N
0p
b
0
u
j
u
i
N
0
h
0
(2) (2)
(3)
(1)
125
Stress trajectory
TOP
SIDE
Stress trajectory
TOP
SIDE
Stress trajectory
TOP
SIDE
Fig. 9.16 Shear lag in gusset platetoslotted RHS joints
1,8
1,6
1,4
1,2
1,0
0,8
0,6
0,4
0,2
0,0
0,0 0,2 0,4 0,6 0,8 1,0
N
1
u
(
J
A
A
=
0
)
/
N
1
u
1,8
1,6
1,4
1,2
1,0
0,8
0,6
0,4
0,2
0,0
0,0 0,2 0,4 0,6 0,8 1,0
N
1
u
(
J
A
A
=
0
)
/
N
1
u
Note: The vertical axis plots the ratio of the ultimate strength N
1u
(J
AA
= 0) of a multiplanar joint with unloaded outofplane
braces and the ultimate strength N
1u
of the corresponding uniplanar joint.
Fig. 9.17 Multiplanar geometry effect for an XX joint of square hollow sections (Yu, 1997)
1,5
0,5
1,0
1,0
1,0 0,0
For  = 0,2
 = 0,4
 = 0,6
 = 0,8
N
1
u
(
J
A
A
)
/
N
1
u
(
J
A
A
=
0
)
1,5
0,5
1,0
1,0
1,0 0,0
For  = 0,2
 = 0,4
 = 0,6
 = 0,8
N
1
u
(
J
A
A
)
/
N
1
u
(
J
A
A
=
0
)
1,0
0,5
1,5
1,0 1,0 0,0
For  = 1,0
N
1
u
(
J
A
A
)
/
N
1
u
(
J
A
A
=
0
)
1,0
0,5
1,5
1,0 1,0 0,0
For  = 1,0
N
1
u
(
J
A
A
)
/
N
1
u
(
J
A
A
=
0
)
Note: The vertical axis plots the ratio of the ultimate strength N
1u
(J
AA
) of a multiplanar joint and the ultimate strength
N
1u
(J
AA
= 0) of the multiplanar joint with unloaded outofplane braces.
Fig. 9.18 Multiplanar loading effect for an XX joint of square hollow sections (Yu, 1997)
126
Fig. 9.19 Multiplanar KK joint of square hollow sections
127
Efficiency T and Y joints in compression
0,0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1,0
0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1

C
T
2=10
2=15
2=20
2=30
2=40
1
f
1 1 y
0 0 y
T
1 y 1
Rd , 1
sin
Q
t f
t f
C
f A
N
u
=
Efficiency X joints in compression
only for u1
= 90
o
and f
y
= 355N/mm
2
0,0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1,0
0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1

C
X
2=10
2=15
2=20
2=30
2=40
1
f
1 1 y
0 0 y
X
1 y 1
Rd , 1
sin
Q
t f
t f
C
f A
N
u
=
Fig. 9.20 Efficiency for T and Y joints of square hollow
sections (brace in compression)
Fig. 9.21 Efficiency for X joints of square hollow
sections (braces in compression)
Efficiency T, Y and X joints in tension
0,0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1,0
0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1

C
X
o
r
C
T
2=10
2=15
2=20
2=30
2=40
1
f
1 1 y
0 0 y
T
1 y 1
Rd , 1
sin
Q
t f
t f
C
f A
N
u
=
Fig. 9.22 Efficiency for T, Y and X joints of square
hollow sections (brace(s) in tension)
Efficiency K gap joints
0,0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1,0
15 20 25 30 35 40
2
C
K
i
2 1
i
f
i yi
0 0 y
K
yi i
Rd , i
b 2
b b
sin
Q
t f
t f
C
f A
N
+
u
=
X and T joints: chord axial stress functions
T joints: chord bending stress function
0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1
1 0,8 0,6 0,4 0,2 0 0,2 0,4 0,6 0,8 1
n
Q
f
=0, 4
=0, 6
=0, 8
=1, 0
X joints: chord bending stress function
0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1
1 0,8 0,6 0,4 0,2 0 0,2 0,4 0,6 0,8 1
n
Q
f
=0, 4
=0, 6
=0, 8
=1, 0
Fig. 9.24 Chord load function for T and X joints of
square hollow sections
K gap joints: chord axial stress functions
0,0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1,0
1 0,8 0,6 0,4 0,2 0 0,2 0,4 0,6 0,8 1
n
Q
f
=0, 25
=0, 4
=0, 6
=0, 81, 0
Fig. 9.23 Efficiency for K gap joints of square hollow
sections
Fig. 9.25 Chord load function for K gap joints of
square hollow sections
128
10. WELDED JOINTS BETWEEN
HOLLOW SECTIONS AND
OPEN SECTIONS
10.1 INTRODUCTION
Hollow sections and open sections are used in various
ways, for example:
 Hollow section braces and open section chords as
shown in Figs. 10.1 and 10.2a
 Open section braces and rectangular hollow section
chords, shown in Fig. 10.2b
 I section beams connected to hollow section
columns, dealt with in Chapter 12
Other combinations, but with bolted connections, are
considered in Chapter 13.
In this chapter, it is shown that in many respects the
behaviour of joints between open sections and hollow
sections is comparable to that of joints between
rectangular hollow sections.
10.2 MODES OF FAILURE
Following the same procedure as described in Chapter
7, the following failure modes (Fig. 10.3) can be
expected and observed for joints between hollow
section braces and I section chords:
 Brace failure (yielding, local buckling)
 Weld failure
 Lamellar tearing
 Chord web failure (yielding, local buckling)
 Chord shear failure
 Chord local buckling
Chord face plastification will not occur here, since this
can only take place after excessive yielding of the
chord web.
As indicated before, the welds should ideally be
stronger than the connected brace members,
assuming the latter are loaded to their limit, thus the
welds should satisfy certain requirements, see Section
6.5.2.
Similar to other hollow section joints, lamellar tearing
should be avoided by choosing proper material.
Local buckling of the members can be avoided by
choosing proper diameter, width and depthto
thickness ratios, thus by limiting the range of validity.
As a result, the governing failure modes to be
considered are:
 Brace failure
 Chord web failure
 Chord shear failure
For joints between hollow section braces and channel
section chords it can be easily shown that with the
same limitations as discussed above, the failure modes
(Fig. 10.4) to be considered in more detail are:
 Brace failure
 Chord face plastification
 Chord punching shear failure
 Chord side wall failure
 Chord shear failure
Since hot rolled channel sections (UNP) have thick
flanges, which act as walls here, chord side wall failure
will not be critical for these sections. In principle, the
other failure modes can be approached in a similar way
to that for joints between rectangular hollow sections
(Wardenier & Mouty, 1979; Wardenier, 1982; Aribert et
al., 1988). If chords of cold formed channel sections are
used (Fig. 10.1), the situation will be different, since the
side walls may deform when the top face deforms,
resulting in lower strengths. For detailed information
about joints with channel section chords, reference can
be made to Wardenier (1982).
The joints of Fig. 10.2 with welded angles or channels
at the sides of a rectangular hollow section are not
different from other joints in open sections. Besides
weld failure, chord shear failure has to be considered.
10.3 ANALYTICAL MODELS
10.3.1 Local brace failure model
The most effective part of the brace is located at the
crossing with the chord web, shown in Fig. 10.5. Here,
the same model can be used as for beamtocolumn
joints of open sections, i.e. for a T, Y or X joint:
N
1
= 2
f
y1
t
1
b
e
(10.1)
where:
0
1 y
0 y
w e
t
f
f
7 r 2 t b + + = (10.2)
If b
e
> b
1
, it is conservatively proposed to follow the
perimeter, as shown in Fig. 10.5.
Similar to joints of circular and rectangular hollow
129
sections, the criterion for local brace failure is also used
for overlap joints. For overlaps Ov > Ov
limit
, the
connection between the braces and the chord has to be
checked for shear.
10.3.2 Chord web failure model
The load from the brace has to be transferred by an
effective area of the chord web, see Fig. 10.6. The
effective areas are located in the chord web at the
location where the brace walls cross the chord web.
Similar to the formulae used for beamtocolumn joints
of open sections:
N
i
sin
i
= f
y0
t
w
b
w
(10.3)
where:
) r t ( 5
sin
h
b
0
i
i
w
+ +
u
= (10.4)
but:
) r t ( 10
sin
t 2
b 2 b
0
i
i
wf w
+ +
u
= s (10.5)
For beamtocolumn joints, a stress dispersion of 2,5:1
is used, which was shown also to be valid for these
joints (Wardenier, 1982).
10.3.3 Chord shear model
Similar to joints with a rectangular hollow section chord,
an interaction formula can be determined for the
combined effect of shear and axial load at the gap
location in the chord, see Fig. 10.7. When the web is
yielding, a part of the flange may also be effective for
shear if the gap is small.
By equilibrium, the moment in the flange is:
2
g V
M
f
f
= (10.6)
The interaction formula for the flange with a rectangular
cross section is:
0 , 1
V
V
M
M
2
f , pl
f
2
f , pl
f
=


.

\

+


.

\

(10.7)
where:
0 y
2
0 0
f , pl
f
4
t b
M = (10.8)
V
pl,f
= b
0
t
0
3
f
0 y
(10.9)
Combination of eqs. (10.6), (10.8) and (10.9) results in:
3 t
g 2
V
V
M
M
0
f , pl
f
f , pl
f
= (10.10)
Substitution in eq. (10.7) gives:
2
0
2
f , pl
f
t 3
g 4
1
1
V
V
+
= (10.11)
Thus, for an I or H section, the active part of the top
flange is:
0 0
t b o (10.12)
where:
2
0
2
t 3
g 4
1
1
+
= o (10.13)
For deep sections the effectiveness of the bottom
flange will be restricted to the area below the web and
corners, thus the following effective shear area is used
for joints with RHS braces:
( ) (
0 w 0 0 0 0 0 v
t r 2 t t b t b 2 A A + ) + o + = (10.14)
If circular hollow section braces are used, the chord
flange is less stiffened at the gap location, and
consequently, is lower. Although here the same area
could be used as in the bottom flange, assuming = 0
in eq. (10.14) gives a better correlation (lower bound)
with the test results.
Thus, shear of the chord cross section at the gap
location has to be checked by:
3
f
A V sin N V
0 y
v 0 , pl i i 0 , gap
= s u = (10.15)
with A
v
according to eq. (10.14).
For the interaction between the axial load and the
130
shear load at the gap location, based on the
HuberHenckyVon Mises criterion, the following
formula applies (see Section 9.3.5):
( )
2
0 , pl
0 , gap
0 y v 0 y v 0 0 , gap
V
V
1 f A f A A N


.

\

+ s (10.16)
131
10.4 EXPERIMENTAL VERIFICATION
The formulae developed in Section 10.3 for joints with
an I section chord have been verified by experiments
(Wardenier & Mouty, 1979; Wardenier, 1982), e.g. the
comparison of eq. (10.15) with test results is given in
Fig. 10.8.
The design criteria for joints with a hot finished channel
chord section have also been checked by tests. It was
shown that for medium to high width ratios , chord
shear failure was the most common failure mode for K
gap joints, whereas chord punching shear or chord
face plastification was governing for low values.
10.5 EVALUATION TO DESIGN RULES
In principle, the same approach has been used as for
the joints discussed in Chapters 8 and 9. However, due
to the limited number of tests, no thorough statistical
analysis has been carried out. The resulting design
resistances are given in Table 10.1.
For overlap joints, the same approach is adopted for all
types of overlap joints, regardless whether circular or
rectangular braces are used in combination with a
circular, rectangular or open section chord. Only the
effective width parameters depend on the type of
section. The resistance of overlap joints between
circular hollow sections with 25% s Ov s 100% overlap
is based on the following criteria:
(1) Local failure of the overlapping brace
(2) Local chord member yielding at the joint location
based on interaction between axial load and
bending moment
(3) Shear of the connection between the brace(s) and
the chord
The cross sections to be examined for these criteria are
the same are those shown in Figures 8.16 and 9.13.
Local failure of the overlapping brace (criterion 1)
should always be checked, while shear between the
braces and the chord (criterion 3) may only become
critical for larger overlaps, i.e. larger than 60% or 80%,
depending on whether or not the hidden toe location of
the overlapped brace is welded to the chord. The check
for local chord member yielding (criterion 2) is, in
principle, a member check and may become critical for
larger overlaps and/or larger ratios.
For 100% overlap joints, similar criteria have to be
checked. Only here, as shown by Chen et al. (2005)
and Qian et al. (2007), shear of the overlapped brace
and chord member yielding will generally be the
governing criteria. Although an overlap of 100% is
given in the recommendations, in general, the overlap
will be slightly larger to allow proper welding of the
overlapping brace to the overlapped brace.
More detailed information regarding the design
equations for overlap joints is given in Chapter 11.
10.6 JOINTS PREDOMINANTLY
LOADED BY BENDING MOMENTS
Here, only beamtocolumn joints with an RHS brace
(or beam) and an I section column are of practical
interest.
The design resistance in Table 10.2 is governed by
local failure of the RHS brace with formulae based on
eqs. (10.1) and (10.2) (see Fig. 10.9) and column web
failure based on eqs. (10.3) to (10.5) (see Fig. 10.10).
Table 10.1 Design resistances of welded joints between RHS or CHS braces and I or H section chords
Type of joint Design limit state
T, Y, X and K gap joints Local brace failure
e i yi Rd , i
b t f 2 N =
Chord web failure
i
w w 0 y
Rd , i
sin
b t f
N
u
=
Chord shear (for K gap joints; for T joints in member
check; for X joints with cos
u
1
> h
1
/h
0
)
b
0
h
0
t
0
t
w
b
1
h
1
t
1
N
1
u
1
d
1
t
1
r
b
0
h
0
t
0
t
w
b
2
h
2
t
2
N
2
u
2
d
2
t
2
r
u
1
b
1
h
1
t
1
N
1
d
1
t
1
g
i
v 0 y
Rd , i
sin
A f 58 , 0
N
u
=
2
Rd , 0 , pl
Ed , 0 , gap
0 y v 0 y v 0 Rd , 0 , gap
V
V
1 f A f ) A A ( N


.

\

+ =
Factors
RHS braces CHS braces
b
e
i i i
yi
0 y
0 w e
t 2 h b but
f
f
t 7 r 2 t b + s + + = ) t d ( 5 , 0 but
f
f
t 7 r 2 t b
i i
yi
0 y
0 w e
t s + + =
b
w
) r t ( 10
sin
t 2
but ) r t ( 5
sin
h
b
0
i
i
0
i
i
w
+ +
u
s + +
u
= ) r t ( 10
sin
t 2
but ) r t ( 5
sin
d
b
0
i
i
0
i
i
w
+ +
u
s + +
u
=
0 w 0 0 0 v
t ) r 2 t ( t b ) 2 ( A A + + o =
A
V
) t 3 /( ) g 4 ( 1
1
2
0
2
+
= o 0 = o
V
pl,0,Rd v 0 y Rd , 0 , pl
A f 58 , 0 V =
Range of validity
X joints T and Y joints K gap joints
Flange class 1 or 2
Compression
Web class 1 and d
w
s 400 mm class 1 or 2 and d
w
s 400 mm
I or H section
chord
Tension none
Compression class 1 or 2
(1)
CHS braces
Tension 50 /t d
i i
s
Compression
class 1 or 2
(1)
Tension 40 /t h and 40 /t b
i i i i
s s RHS braces
Aspect ratio
0 , 2 /b h 5 , 0
i i
s s
Gap N/A
2 1
t t g + >
Eccentricity N/A
0
h 25 , 0 e s
Brace angle
30
i
> u
Yield stress
y0 yi
f f s
(2)
u y
f 8 , 0 f s
2
y
N/mm 460 f s
(1)
Section class limitations are given in Table 2.7.
(2)
For 355 N/mm
2
< f
y
s 460 N/mm
2
, use a reduction factor of 0,9 for the design resistances.
132
Table 10.2 Design moment resistance of uniplanar RHS braces (beams) to I or H section chord joints
Type of joint Design limit state
T joints Local brace failure
z e 1 1 y Rd , 1 , ip
h b t f M =
where h
z
is the distance between the centres of gravity of the
effective parts of the RHS brace (beam)
Chord web failure
h
z
h
1
b
e
b
e
'I' or 'H' Column
RHS Beam
M
I or H chord
RHS brace (beam)
M
ip,1
u
1
h
z
h
1
b
e
b
e
'I' or 'H' Column
RHS Beam
M
I or H chord
RHS brace (beam)
M
ip,1
u
1
) t h ( b t f 5 , 0 M
1 1 w w 0 y Rd , 1 , ip
=
Factors
b
e
1 1 1
1 y
0 y
0 w e
t 2 h b but
f
f
t 7 r 2 t b + s + + =
b
w
) r t ( 10 t 2 but ) r t ( 5 h b
0 1 0 1 w
+ + s + + =
Range of validity Same as in Table 10.1, but
1
90
133
T joint X joint
h
0
b
0
b
1
h
1
d
1
t
1
t
0
t
w
r
t
1
h
0
b
0
b
1
h
1
d
1
t
1
t
0
t
w
r
t
1
h
0
b
0
b
1
h
1
t
0
r
t
w
d
1
t
1
t
1
h
0
b
0
b
1
h
1
t
0
r
t
w
d
1
t
1
t
1
K gap joints
N
1
b
0
h
0
N
2
h
1
h
2
b
1
d
1
d
2
t
1 t
2
t
0
t
w r
u
1
u
2
d
1
b
2
N
1
b
0
h
0
N
2
h
1
h
2
b
1
d
1
d
2
t
1 t
2
t
0
t
w r
u
1
u
2
d
1
b
2
N
1
t
w
b
0
N
2
b
1 b
2
t
1
t
2
t
0
h
0 r
u
2
u
1
h
1
h
2
g = 0,1b
0
N
1
t
w
b
0
N
2
b
1 b
2
t
1
t
2
t
0
h
0 r
u
2
u
1
h
1
h
2
g = 0,1b
0
Fig. 10.1 Welded truss joints between hollow section braces and open section chords
134
Fig. 10.2 Welded truss joints between hollow sections and open sections (Packer & Henderson, 1997)
135
(a) Brace failure
(d) Chord web failure
(b) Weld failure
as (a) but failure in the weld
(e) Chord shear failure
(c) Lamellar tearing
see e.g. Fig. 7.6c
(f) Chord local buckling
Fig. 10.3 Governing failure modes for joints between hollow section braces and I section chords
136
(a) Brace failure
(d) Chord side wall failure
(b) Chord face plastification
(e) Chord shear failure
(c) Chord punching shear failure
Fig. 10.4 Governing failure modes of joints between hollow section braces and a channel section chord
137
(a) (b)
Fig. 10.5 Model for local brace failure
b
wf
b
wf
b
w
1 : 2,5 1 : 2,5
b
wf
b
wf
b
w
1 : 2,5 1 : 2,5
Fig. 10.6 Model for chord web failure
Fig. 10.7 Chord shear model
138
RI joints
700
600
500
400
300
200
100
0
0 100 200 300 400 500 600 700 800
RI joints
700
600
500
400
300
200
100
0
0 100 200 300 400 500 600 700 800
(a) Joints between RHS braces and an I section chord (RI joints)
600
500
400
300
200
100
0
0 100 200 300 400 500 600 700 800
CI joints
600
500
400
300
200
100
0
0 100 200 300 400 500 600 700 800
CI joints
(b) Joints between CHS braces and an I section chord (CI joints)
Fig. 10.8 Comparison between test results and eq. (10.15) for joints between hollow section braces and
an I section chord (Wardenier, 1982)
139
11
Fig. 10.9 Local brace failure criterion for moment loading of RHS beamtocolumn joints
1
b
wf
1
b
wf
Fig. 10.10 Column web failure criterion for moment loading
140
11. WELDED OVERLAP JOINTS
11.1 INTRODUCTION
In the latest revision of the IIW recommendations for
the design of hollow section joints (IIW, 2009), the
design strength formulae for overlap joints have also
been reanalysed based on the research results of
Chen et al. (2005), Liu et al. (2005), Wardenier &
Choo (2006), Wardenier (2007) and Qian et al.
(2007).
The objective was to present design strength
formulae which are, as far as possible, based on a
physical model and which have a good agreement
with available experimental and/or numerical data.
Further, the formulae should be logical, as simple as
possible, consistent for similar types of joints and
easy to understand for designers. Therefore, in the
case of overlap joints, it was preferred that the
approach for CHS and RHS overlap joints would be
consistent and based on the same philosophy.
As a result, for overlap joints, the same approach is
now adopted for all types of overlap joints, regardless
whether circular or rectangular braces are used in
combination with a circular, rectangular or an open
section chord. Only the effective width parameters
depend on the type of section.
11.2 MODES OF FAILURE
In overlap joints the following modes of failure can
occur:
 (Local) brace failure (cracking or local buckling)
 Weld failure
 Lamellar tearing
 Local chord member yielding or local buckling
 Shear failure at the connection between the
brace(s) with the chord
Similar to other welded hollow section joints, to avoid
weld failure, the welds should ideally be stronger than
the connected braces. For fillet welds, the throat
thickness should satisfy the same requirements as
those given in Section 6.5.2.
Also here, the steel should not be susceptible to
lamellar tearing. Especially for larger thicknesses (t >
25 mm), a TTP quality with a low sulphur content
should be used for the overlapped brace and chord
members.
In the design recommendations, the widthtowall
thickness ratios b/t and diametertowall thickness
ratios d/t have been limited to avoid local buckling.
As a result, the following failure modes still have to be
considered in design:
(1) Local brace failure
(2) Local chord member yielding
(3) Shear failure at the connection between the
brace(s) with the chord
The locations of these failure modes are given in Fig.
11.1. Note that for K and N overlap joints, the
subscript i is used to denote the overlapping brace
member, while the subscript j refers to the overlapped
brace member.
In the previous IIW recommendations (IIW, 1989),
local chord member yielding was considered as a
member failure. However, because designers
sometimes forgot to check this criterion, it has now
been included in the joint strength equations.
Although considered as a possible failure mode, in
the experiments of the seventies and eighties, brace
shear failure was not observed. More recently,
numerical studies showed that especially for large
overlaps or rectangular braces with h
j
< b
j
, brace
shear failure could govern. A concentrated local brace
shear may further, in case of chords loaded in
compression, initiate local buckling of the connecting
face of the chord.
11.3 ANALYTICAL MODELS FOR RHS
OVERLAP JOINTS
In the past, various models for overlap joints have
been studied in detail (e.g. Dexter & Lee, 1998;
Kurobane, 1981; Marshall, 1992; Packer, 1978;
Wardenier, 1982). However, in these studies, overlap
joints with CHS, RHS or open section chords were
studied separately, leading to different design
approaches. As shown in this chapter, these joints
can be described by similar models.
11.3.1 Local brace failure model
The local brace failure model, shown in Fig. 11.2 for
50% and 100% overlap joints, is consistent with that
used for other RHS joints, see Chapter 9. Local brace
failure, also referred to as "brace effective width
criterion" is especially critical for overlap joints with
relatively thin walled overlapping braces.
141
For 100% overlap joints in which the brace widths of
the overlapped and overlapping brace are about the
same, three walls of the overlapping brace are fully
effective, while the overlapping brace cross wall on
the overlapped brace is only partly effective by b
e,ov
.
Hence, the capacity of the overlapping brace i can be
given by eq. (11.1):
N
i
= f
yi
t
i
(2h
i
+ b
i
+ b
e,ov
 4t
i
) (11.1)
For overlaps which still sufficiently stiffen the
connection with the chord wall, e.g. for 50% s Ov <
100% (the lower limit of 50% has been verified by
tests), the sides of the overlapping brace are still
assumed to be fully effective and the capacity can be
given by eq. (11.2):
N
i
= f
yi
t
i
(2h
i
+ b
ei
+ b
e,ov
 4t
i
) (11.2)
where b
ei
and b
e,ov
are the effective width parts for the
connected cross walls, being consistent with those for
other joints:
i i
i yi
0 0 y
0 0
ei
b but b
t f
t f
/t b
10
b s


.

\



.

\

= (11.3)
i i
i yi
j yj
j j
ov , e
b but b
t f
t f
/t b
10
b s


.

\



.

\

= (11.4)
For overlaps smaller than 50%, the effective parts of
the side walls are taken in relation to the overlap, i.e.
2h
i
(Ov/50). Overlaps smaller than 25% are not
recommended due to the possible large difference in
stiffness between the overlapping connection and the
connection to the chord, which may lead to premature
cracking and lower capacities (Wardenier, 2007).
11.3.2 Local chord member failure
At the joint location, the chord should always be
checked for local chord member failure based on an
interaction between axial load and bending moment.
Various interaction formulae exist, e.g. for RHS eq.
(11.5) with c = 1,5 which is based on Roik &
Wagenknecht (1977). In the experiments of Sopha et
al. (2006), the capacity according to eq. (11.5) with c
= 1,5 could just be reached. However, the numerical
results showed that at the joint location it would be
better to use a linear interaction, i.e. with c = 1,0:
0 , 1
M
M
N
N
0 , pl
0
c
0 , pl
0
s +


.

\

(11.5)
This criterion becomes critical for joints with large
overlaps, thus with a large eccentricity.
11.3.3 Shear failure of the connection
between brace(s) and chord
Shear failure at the connection between the brace(s)
and the chord may occur for large overlaps or in
cases where h
j
< b
j
. As shown in Fig. 11.3 for 100%
overlap joints, three sides of the overlapped brace are
fully effective while the cross wall at the heel is only
effective for a part b
ej
. Thus, the yield capacity can be
given by eq. (11.6):
j
j ej j j yj
j j i i
sin
t ) b b h 2 (
3
f
cos N cos N
u
+ +
s u + u (11.6)
For joints with overlaps Ov < 100%, the effective parts
of both braces with relevant thicknesses and steel
grade have to be considered, see Wardenier & Choo
(2006). Further, the effective parts depend on
whether or not the hidden location of the overlapped
brace has been welded to the chord.
For partially overlapped joints with the hidden toe of
the overlapped brace not welded, the parts effective
for shear are shown in Fig. 11.3 and the yield
capacity is given by eq. (11.7):
= u + u
j j i i
cos N cos N
j
j ej j yj
i
i ei i
yi
sin
t ) b h 2 (
3
f
sin
t b h 2
100
Ov 100
3
f
u
+
+
u
(
+ 
.

\

(11.7)
where:
b
ei
effective width at the connection between the
overlapping brace cross wall and the chord
according to eq. (11.3)
b
ej
effective width at the connection between the
overlapped brace cross wall and the chord
according to eq. (11.8):
j j
j yj
0 0 y
0 0
ej
b but b
t f
t f
/t b
10
b s


.

\



.

\

= (11.8)
The overlap limit for which this criterion may become
decisive depends on various geometrical parameters.
142
11.4 ANALYTICAL MODELS FOR CHS
OVERLAP JOINTS
For CHS joints, the same principles are followed as
for RHS joints and the criteria are directly related to
those of RHS joints (Wardenier, 2007; Qian et al.,
2007).
For local failure of the overlapping brace, the criteria
for overlap joints with RHS braces are multiplied by
/4 to obtain the capacity of CHS brace joints, since
/4 is the ratio of cross sectional areas of thin walled
CHS and RHS braces with d = b = h. Further all "b"
and "h" dimensions in the formulae are replaced by
"d".
Since the local stiffness of the joints between CHSto
CHS members is more uniform than that for RHSto
RHS members, the constant in the effective width
terms of eqs. (11.3), (11.4) and (11.8) is increased by
20%, i.e. changed from 10 to 12. This increase is also
found when comparing the efficiency of CHS X joints
to that of RHS X joints. Adopting these modifications
in eqs. (11.1) and (11.2) results in the functions given
in Tables 11.1 and 11.4. Numerical data showed that
due to the more uniform stiffness distribution in CHS
overlap joints, a single expression, related to the
expression for RHS joints with 50% s Ov < 100%, can
be used to describe the local brace failure of CHS
joints with overlaps 25% s Ov < 100%.
For the local chord yield criterion, eq. (11.5) with c =
1,7 for CHS sections is given, based on Roik &
Wagenknecht (1977).
11.5 ANALYTICAL MODELS FOR
OVERLAP JOINTS WITH AN
OPEN SECTION CHORD
Overlap joints with open section chords behave in
principle similar to those with an RHS chord section.
Only the effective width terms b
ei
and b
ej
at the
connection of the braces with the chord differ.
However, these terms are in agreement with the
expression used for T, X and K gap joints, see eq.
(10.2).
For local chord member failure, eq. (11.5) is used with
c = 1,0, similar to the equation for overlap joints with
an RHS chord.
The brace shear criterion is also consistent with
overlap joints with an RHS chord, although the
effective width terms b
ei
and b
ej
at the connection of
the braces with the chord should again be based on
eq. (10.2).
11.6 EXPERIMENTAL AND
NUMERICAL VERIFICATION
The equations in Sections 11.3 to 11.5 for local brace
failure have been verified with experiments and/or
numerical data; for RHS overlap joints in Wardenier
(1982), Chen et al. (2005), Liu et al. (2005) and
Wardenier & Choo (2006) and for CHS joints in
Wardenier (2007) and Qian et al. (2007). For joints
with open section chords, the local brace failure
criterion was previously evaluated in Wardenier
(1982).
As an example, Fig. 11.4 shows, for all experimental
and numerical data for CHS overlap joints with 25% s
Ov < 100% in the screened database of Makino et al.
(1996), the ratio of the actual capacity and the
predicted value using the local brace failure criterion,
see Table 11.1. This evaluation gives a mean value of
1,11 with a coefficient of variation (COV) of 6,0%.
Due to the nonuniform stiffness distribution in joints
with small overlaps, and depending on the geometric
parameters, the joint capacity may drop for small
overlaps, leading to a large scatter. Therefore, it is
proposed to limit the validity of equations for local
brace failure to 25% s Ov < 100%.
As already mentioned in Section 11.3.2, for local
chord failure, the experiments by Sopha et al. (2006)
just reached the capacity according to eq. (11.5) with
c = 1,5. However, the numerical results showed that
at the joint location it would be better to use a linear
interaction, i.e. with c = 1,0, which is also adopted for
open section chords.
The available experimental results and numerical
data (Chen et al., 2005; Qian et al., 2007) showed
that the brace shear criteria, i.e. eqs. (11.6) and
(11.7) are too conservative if based on the yield
strength f
y
. Since fracture does not occur but the
shear deformation has to be limited to avoid local
chord deformation leading to premature chord local
buckling, it was proposed to base this strength
criterion on the ultimate strength f
u
of the parts
effective for shear.
In 1994, Davies & Crocket (1994) already
investigated the effect of the welded or nonwelded
hidden toe of the overlapped brace for RHS joints.
They found, for their numerically investigated joints,
143
144
no effect. However, Dexter & Lee (1998) observed for
CHS joints with the hidden location welded, a
capacity which was about 10% higher than that for
joints with no weld. The recent reanalysis for the IIW
(2009) recommendations showed that the effect
depends on the overlap and the governing criterion. If
local brace failure or the local chord yielding is
decisive, then welding of the hidden toe of the
overlapped brace has no effect on the joint strength.
However, if the brace shear criterion is governing, the
capacity of the joints with the hidden location welded
is about 10% higher than that of the joints with no
weld, which agrees with the above results found in
the literature.
Fig. 11.5 shows, for CHS joints, that the local brace
failure criterion and the brace shear criterion with the
hidden location welded or unwelded give about the
same capacity for an overlap of 60% if the hidden toe
location is unwelded and for an overlap of 80% if the
hidden toe location is welded. From this comparison,
the limits decisive for checking the brace shear
criterion are determined. This also explains why, in
general, no brace shear failures were observed in the
experiments: in most cases the hidden location was
welded and the overlap was smaller.
For joints with braces with h
j
< b
j
, the brace shear
criterion can become critical for smaller overlaps.
11.7 JOINT STRENGTH FORMULAE
Using the equations developed in Sections 11.3 to
11.5 with modifications based on the experimental
and/or numerical data discussed in Section 11.6, the
final joint strength equations shown in Table 11.1
have been developed.
The range of validity is given in Table 11.2, while
details of the parameters used in Table 11.1 are
provided in Tables 11.3 and 11.4.
Designers who feel the equations for overlap joints
with RHS braces are too complicated could, as a
(conservative) simplification, neglect the effective
width terms (b
ei
+ b
e,ov
 4t
i
) for the local brace failure
criterion and b
ei
and b
ej
for the brace shear criterion,
thus reducing many parameters in Table 11.4 to zero
or to a certain conservative value. Similar comments
apply for d
ei
, d
e,ov
and d
ej
for overlap joints with CHS
braces.
Table 11.1 Design axial resistance of uniplanar overlap joints with a CHS, RHS, I or H section chord
Type of joint Design limit state
Axially loaded overlap joints Local failure of overlapping brace
. eff , b i yi Rd , i
t f N =
Local chord member yielding
c = 1,7 for CHS chord
0 , 1
M
M
N
N
Rd , 0 , pl
Ed , 0
c
Rd , 0 , pl
Ed , 0
s +


.

\

c = 1,0 for RHS or I section chord
Brace shear (for Ov
limit
< Ov s 100%)
(1)
t
0
N
j
u
j
d
j
t
j
u
i
d
i
N
i
t
i
t
0
N
j
u
j
u
i
N
i b
j
h
j
t
j
b
i
i
t
i
h
Rd , s j Ed , j i Ed , i
N cos N cos N s u + u (see Table 11.3)
b,eff.
(2)
CHS braces RHS braces
25% s Ov < 50%
i ov , e ei i . eff , b
t 4 b b h 2 )
50
Ov
( + + =
50% s Ov < 100%
) t 4 d d d 2 (
4
i ov , e ei i . eff , b
+ +
t
=
i ov , e ei i . eff , b
t 4 b b h 2 + + =
Ov = 100% ) t 4 d 2 d 2 (
4
i ov , e i . eff , b
+
t
=
i ov , e i i . eff , b
t 4 b b h 2 + + =
General note
The efficiency (i.e. design resistance divided by the yield load) of the overlapped
brace j shall not exceed that of the overlapping brace i, i.e.


.

\

s
yi i
yj j
Rd , i Rd , j
f A
f A
N N
(1)
Ov
limit
= 60% if hidden toe of the overlapped brace is not welded.
Ov
limit
= 80% if hidden toe of the overlapped brace is welded.
(2)
The expressions for d
ei
, d
e,ov
, b
ei
and b
e,ov
are given in Table 11.4.
145
Table 11.2 Design axial resistance of uniplanar overlap joints with a CHS, RHS, I or H section chord
Range of validity
30 and
j i
> u u
General
20 , 0 /d d and /d d
0 j 0 i
>
25 , 0 /b b and /b b
0 j 0 i
>
25 , 0 /b d and /b d
0 j 0 i
>
75 , 0 /d d
j i
>
75 , 0 /b b
j i
>
0 j i
t t and t s
j i
t t s
% 25 Ov >
0 y yj yi
f f and f s
u y
f 8 , 0 f s
2
y
N/mm 460 f s
(2)
Compression class 1 or 2
(1)
and 50 /t d
0 0
s
CHS
Tension 50 /t d
0 0
s
Compression class 1 or 2
(1)
and 40 /t h and 40 /t b
0 0 0 0
s s
Tension 40 /t h and 40 /t b
0 0 0 0
s s RHS
Aspect ratio 0 , 2 /b h 5 , 0
0 0
s s
Flange class 1 or 2
Compression
Web class 1 or 2 and d
w
s 400 mm
Chord
I or H
section
Tension none
CHS or RHS chord I or H section chord
Compression class 1 or 2
(1)
and 50 /t d
1 1
s class 1
CHS
Tension 50 /t d
2 2
s
Compression class 1 or 2
(1)
and 40 /t h and 40 /t b
1 1 1 1
s s class 1
Tension 40 /t h and 40 /t b
2 2 2 2
s s
Braces
RHS
Aspect ratio 0 , 2 /b h 5 , 0 and 0 , 2 /b h 5 , 0
j j i i
s s s s 0 , 1 /b h and 0 , 1 /b h
j j i i
= =
(1)
Section class limitations are given in Table 2.7.
(2)
For 355 N/mm
2
< f
y
s 460 N/mm
2
, use a reduction factor of 0,9 for the design resistances.
Table 11.3 Design brace shear resistance of uniplanar overlap joints with a CHS, RHS or I section chord
N
s,Rd
for brace shear criterion
(1)
(only to be checked for Ov
limit
< Ov s 100%)
(2)
Ov
limit
< Ov < 100%
]
sin
t ) d c d 2 (
f 58 , 0
sin
t d d 2
100
Ov 100
f 58 , 0 [
4
N
j
j ej s j
uj
i
i ei i
ui Rd , s
u
+
+
u
(
+ 
.

\

t
=
CHS braces
Ov = 100%
j
j ej j
uj Rd , s
sin
t ) d d 3 (
4
f 58 , 0 N
u
+
t
=
Ov
limit
< Ov < 100%
j
j ej s j
uj
i
i ei i
ui Rd , s
sin
t ) b c h 2 (
f 58 , 0
sin
t b h 2
100
Ov 100
f 58 , 0 N
u
+
+
u
(
+ 
.

\

=
RHS braces
Ov = 100%
j
j ej j j
uj Rd , s
sin
t ) b b h 2 (
f 58 , 0 N
u
+ +
=
(1)
The expressions for d
ei
, d
ej
, b
ei
and b
ej
are given in Table 11.4.
(2)
Ov
limit
= 60% and c
s
= 1 if hidden toe the overlapped brace is not welded.
Ov
limit
= 80% and c
s
= 2 if hidden toe of the overlapped brace is welded.
In the case of overlap joints with h
i
< b
i
and/or h
j
< b
j
, the brace shear criterion shall always be checked.
146
Table 11.4 Effective width factors (b
e
and d
e
) used in Tables 11.1 and 11.3
Factors for CHS braces to CHS chords
CHS braces
Overlapping CHS brace to CHS chord
i i
i yi
0 0 y
0 0
ei
d but d
t f
t f
/t d
12
d s


.

\



.

\

=
Overlapped CHS brace to CHS chord
j j
j yj
0 0 y
0 0
ej
d but d
t f
t f
/t d
12
d s


.

\



.

\

=
Overlapping CHS brace to overlapped CHS brace
i i
i yi
j yj
j j
ov , e
d but d
t f
t f
/t d
12
d s


.

\



.

\

=
Factors for CHS or RHS braces to RHS chords
CHS braces RHS braces
Overlapping CHS brace to RHS chord
i i
i yi
0 0 y
0 0
ei
d but d
t f
t f
/t b
10
d s


.

\



.

\

=
Overlapped CHS brace to RHS chord
j j
j yj
0 0 y
0 0
ej
d but d
t f
t f
/t b
10
d s


.

\



.

\

=
Overlapping CHS brace to overlapped CHS brace
i i
i yi
j yj
j j
ov , e
d but d
t f
t f
/t d
12
d s


.

\



.

\

=
Overlapping RHS brace to RHS chord
i i
i yi
0 0 y
0 0
ei
b but b
t f
t f
/t b
10
b s


.

\



.

\

=
Overlapped RHS brace to RHS chord
j j
j yj
0 0 y
0 0
ej
b but b
t f
t f
/t b
10
b s


.

\



.

\

=
Overlapping RHS brace to overlapped RHS brace
i i
i yi
j yj
j j
ov , e
b but b
t f
t f
/t b
10
b s


.

\



.

\

=
Factors for CHS or RHS braces to I section chords
CHS braces RHS braces
Overlapping CHS brace to I section chord
i
yi
0 y
0 w ei
d but
f
f
t 7 r 2 t d s + + =
Overlapped CHS brace to I section chord
j
yj
0 y
0 w ej
d but
f
f
t 7 r 2 t d s + + =
Overlapping CHS brace to overlapped CHS brace
i i
i yi
j yj
j j
ov , e
d but d
t f
t f
/t d
12
d s


.

\



.

\

=
Overlapping RHS brace to I section chord
i
yi
0 y
0 w ei
b but
f
f
t 7 r 2 t b s + + =
Overlapped RHS brace to I section chord
j
yj
0 y
0 w ej
b but
f
f
t 7 r 2 t b s + + =
Overlapping RHS brace to overlapped RHS brace
i i
i yi
j yj
j j
ov , e
b but b
t f
t f
/t b
10
b s


.

\



.

\

=
147
t
0
N
0p
h
j
b
j
t
j
b
0
h
0
b
i
h
i
t
i
N
0
u
j
u
i
N
i
N
j
(2) (2)
(3)
(1)
i = overlapping member and j = overlapped member
Fig. 11.1 Overlap joint with cross sections to be checked (for other sections, the same locations apply)
0,5b
e,ov
0,5b
e,ov
b
0
0,5b
ei
t
0
u
i
u
i
u
j
u
j
h
0
Fig. 11.2 Local failure of the overlapping brace for RHS joints with 50% and 100% overlap
h
j
b
j
t
j
t
i
h
i
b
i
u
j
u
i
0,5b
ej
0,5b
ei
h
j
/ sinu
j
0,5h
i
/ sinu
i
t
i
t
j
h
j
b
j
t
j
t
i
h
i
b
i
u
j
u
i
0,5b
ej
0,5b
ei
h
j
/ sinu
j
0,5h
i
/ sinu
i
t
i
t
j
h
j
b
j
t
j
t
i
h
i
b
i
u
j
u
i
0,5b
ej
b
j
h
j
/ sinu
j
t
j
h
j
b
j
t
j
t
i
h
i
b
i
u
j
u
i
0,5b
ej
b
j
h
j
/ sinu
j
t
j
Fig. 11.3 Effective shear area for RHS joints with 50% (no hidden weld) and 100% overlap
148
CHS 25100% Overlap joints (excl. brace yielding)
0,0
0,2
0,4
0,6
0,8
1,0
1,2
1,4
0,0 0,2 0,4 0,6 0,8 1,0

N
e
x
p
.
/
N
l
o
c
a
l
b
r
a
c
e
f
a
i
l
u
r
e
Washio '63
Kurobane '64
Togo '67
Wardenierde Koning '77
Kurobane '80
Ochi '81
De KoningWardenier '81
Kurobane '82
Dexter '94
Fig. 11.4 Ratio of the experimental joint capacity (database Makino et al., 1996) and the capacity based on
the local brace failure criterion (25% s Ov < 100%)
Local brace failure vs. brace shear criteria
0,0
0,2
0,4
0,6
0,8
1,0
1,2
0 20 40 60
d
0
/t
0
E
f
f
i
c
i
e
n
c
y
Local brace failure
Brace shear; hidden toe
welded; Ov = 80%
Brace shear; hidden toe
not welded; Ov = 60%
Fig. 11.5 Comparison between the criteria for local brace failure and brace shear with the hidden location
welded or unwelded (for d
i
/t
i
= 25, d
i
= d
j
, t
i
= t
j
, f
yi
= f
yj
, f
ui
= f
uj
= 1,25f
yi
,
i
=
j
= 45)
149
150
12. WELDED I BEAMTOCHS OR
RHS COLUMN MOMENT
JOINTS
12.1 INTRODUCTION
Connections in beamtocolumn joints can be welded
or bolted. This chapter will focus mainly on unstiffened,
welded joints between CHS or RHS columns and I
section beams, as shown in Fig. 12.1. Examples of
some stiffened joints, especially used in earthquake
prone regions, e.g. Japan, are shown in Fig. 12.2.
In Chapters 8 and 9 it is already stated that the design
formulae for hollow section joints loaded by bending
moments can be derived in a similar way to that for
axially loaded joints. This also applies to beamto
column joints. For more detailed information,
reference can be made to Dutta (2002), Kamba &
Tabuchi (1994), Kurobane et al. (2004), Lu (1997),
Packer & Henderson (1997) and De Winkel (1998).
12.2 MODES OF FAILURE
In a similar way to that described in Chapter 7, various
modes of failure (Fig. 12.3) can be identified by
following the load transfer through the joint:
 Local beam flange failure (yielding, local buckling)
 Weld failure
 Lamellar tearing
 Column plastification (face, wall or cross section)
 Column punching shear
 Column local buckling
 Column shear failure
As indicated in Chapter 7, several modes of failure
can be avoided, e.g.:
 Weld failure, by making the welds stronger than the
connected beam, i.e. for double fillet welds the
throat thickness "a" should meet at least 0,5 times
the value given in Section 6.5.2.
 Lamellar tearing, by choosing a material quality
which is not susceptible to lamellar tearing (TTP
quality).
 Local buckling, by limiting the widthtothickness
and/or the diametertothickness ratio.
As a result, the following modes of failure have to be
considered for design:
 Local failure of the beam flange
 Column plastification (face, wall or cross section)
 Column punching shear
 Column shear failure
12.3 ANALYTICAL MODELS
12.3.1 Local failure of the beam flange
The effective width for the beam flange b
e
shown in
Fig. 12.4 can be determined from platetoCHS or
platetoRHS joints (see Sections 8.7.2 and 9.7.3 with
more detailed information in Section 9.3) because the
flange connections are governing.
The moment capacity can be expressed by:
M
1,Rd
= N
1,Rd
(h
1
 t
1
) (12.1)
where N
1,Rd
is the flange capacity for axial loading
based on local failure of the beam flange determined
in a similar way to that for the joints between hollow
sections.
Tests on platetoCHS column joints showed that,
within the range of validity of the formulae, local failure
of the plate was not critical compared to the other
criteria.
For platetoRHS column joints, the capacity for local
failure of the plate is given by:
M
1,Rd
= f
y1
t
1
b
e
(h
1
 t
1
) (12.2)
where b
e
is similar to that for joints between
rectangular hollow sections, see Table 9.1.
12.3.2 Column plastification
The plastification of I beamtoCHS or RHS column
joints is not only determined by the connection at the
flanges but also by the column depth. The web of the I
beam forces the chord face of an RHS column into a
different yield line pattern than that which would be
observed by two flanges at a certain distance apart,
see Fig. 12.5.
If the web of the beam was not present, the capacity
of the flange could be given according to eq. (12.1)
where N
1,Rd
is the flange capacity based on the
column plastification criterion. For example, for an
RHS column (with
1
= 90 and = t
1
/b
0
which is very
small), eq. (9.6) would result in:
) t h (
1
4
t f M
1 1
2
0 0 y Rd , 1


.

\


~ (12.3)
151
If the influence of the web is incorporated, the
equation becomes considerably more complicated
and reference can be made to Lu (1997).
For the chord side wall plastification criterion, similar
rules can be used to those for beamtocolumn joints
between I sections and those used for RHS joints with
=1,0, i.e.
) t h ( b t f 2 M
1 1 wf 0 0 y Rd , 1
= (12.4)
where:
2
t 5 h
but t 5 t b
0 1
0 1 wf
+
s + = (12.5)
For I beamtoCHS column joints, the strength of the
flange plate connection can be based on the ring
model (see Chapter 8) with the resistances given in
Table 8.3. However, for moment loading with the
beam web included, the formulae become rather
complicated and have to be calibrated with test
results, resulting in semiempirical formulae (De
Winkel, 1998).
12.3.3 Column punching shear
The column punching shear strength of I beamto
CHS or RHS joints can be directly determined from
platetoCHS or platetoRHS joints (Wardenier et al.,
2008a; Packer et al., 2009a). For more detailed
information, see Wardenier (1982) and Voth (2010).
Here, similar to the criterion for local failure of the
beam flange, the flanges are governing because the
webs are located at the softest part of the column face
and are generally not effective.
As shown in Fig. 12.6, the capacity is given by:
) t h )( t 2 b 2 ( t
3
f
M
1 1 1 p , e 0
0 y
Rd , 1
+ = (12.6)
For platetoCHS column joints, it was shown that,
within the range of application given in Table 8.3, b
e,p
can be taken as b
1
.
For platetoRHS column joints, the same b
e,p
can be
taken as that given in Table 9.4.
12.3.4 Column shear failure
If the beamtocolumn joints only have a moment
loaded beam on one side, or alternatively the beam
moments on either side of the joint do not balance
each other, shear forces will act in the column, which
may cause shear failure of the column. The cross
section of the column has to be checked here for the
combined actions of axial load, shear load and
bending moment. For class 1 and class 2 sections,
the interaction can be based on the
HuberHenckyVon Mises criterion (Wardenier, 1982)
or a suitable, possible stress distribution can be
assumed, as e.g. shown in Fig. 12.7b. According to
the HuberHenckyVon Mises criterion, the following
condition applies for a side wall of an RHS column:
2 2 2
y
3 f t + o = (12.7)
or:
2
y
2
y
3 / f
f
1


.

\

t
+


.

\

o
= (12.7a)
or:
1
V
V
M
M
2
Rd , 0 , pl
Ed , 0
2
Rd , 0 , pl
Ed , 0
=


.

\

+


.

\

(12.7b)
1
V
V
N
N
2
Rd , 0 , pl
Ed , 0
2
Rd , 0 , pl
Ed , 0
=


.

\

+


.

\

(12.7c)
or:
2
Rd , 0 , pl
Ed , 0
Rd , 0 , pl Rd , 0 , V , pl
V
V
1 M M


.

\

= (12.8)
2
Rd , 0 , pl
Ed , 0
Rd , 0 , pl Rd , 0 , V , pl
V
V
1 N N


.

\

= (12.9)
Adding the flange parts (which are not reduced by
shear) and conservatively assuming that the effective
shear area is 2 h
m
t
0
results in:
2
Rd , 0 , pl
Ed , 0
0 y 0
2
m 0 y 0 m m Rd V,0, pl,
V
V
1 f t h 5 , 0 f t h b M


.

\

+ =
(12.10)
152
2
Rd , 0 , pl
Ed , 0
0 y 0 m 0 y 0 m Rd , 0 , V , pl
V
V
1 f t h 2 f t b 2 N


.

\

+ =
(12.11)
The formulae (12.10) and (12.11) show the plastic
capacities for axial loading and moment, reduced by
the effect of shear.
In a similar way, the interaction between axial load
and bending moment can be derived (Wardenier,
1982). By introducing N
pl,V,0,Rd
and M
pl,V,0,Rd
instead of
N
pl,0,Rd
and M
pl,0,Rd
, the full interaction can be obtained.
In the standards (EN 199311, 2005) these formulae
are approximated by simpler formulae. Also, the effect
of small shear loads has been neglected, e.g. for V
0,Ed
s 0,5 V
pl,0,Rd
.
12.4 EXPERIMENTAL AND
NUMERICAL VERIFICATION
Most of the initial tests on platetoCHS and I
beamtoCHS joints were carried out in Japan (Makino
et al., 1991; Kamba & Tabuchi, 1994). A good survey
of all existing evidence on beamtoCHS column
joints, including many tests on stiffened joints, is given
by Kamba & Tabuchi (1994). Later work by De Winkel
(1998) concentrated on a numerical parameter study
with experiments carried out for validation of the
numerical models. This study not only dealt with
simple, unstiffened joints, but also with I beamtoCHS
column joints in which the column was filled with
concrete and/or combined with a composite
steelconcrete floor.
A similar investigation to that carried out by De Winkel
(1998) was done by Lu (1997) for plate and I
beamtoRHS column joints.
Stiffened joints with strips welded at the sides of the
flanges were investigated by Shanmugam et al.
(1993) i.e. the cross section of the flange at the
connection with the RHS column also had an I shape.
12.5 BASIC JOINT STRENGTH
FORMULAE
In the studies mentioned in Section 12.4 on
unstiffened RHS joints, strength formulae have been
determined using analytical models as a basis. These
formulae have been modified to fit the numerical data.
Since the strength data have been based on a
deformation criterion of 3% of the chord width, the
resulting formulae for platetoRHS column joints give,
for low ratios, lower strengths than those based on
the yield line method.
At present, only strength functions are established for
uniplanar and multiplanar joints, but no formulae have
been developed for the stiffness, although the
stiffness is extremely important for the determination
of the moment distribution in unbraced frames.
However, a large number of momentrotation
diagrams is available for a wide range of parameters.
From these momentrotation diagrams (Lu, 1997; De
Winkel, 1998) indications can be obtained for the
stiffness.
The effect of the joint stiffness on the elastic moment
distribution is represented in Figs. 12.8 and 12.9. It is
shown that with semirigid joints the elastic moment
distribution can be influenced considerably.
If a rigidplastic analysis is used, not only is the joint
moment resistance of primary importance, but also
the rotation capacity. For example, if the stiffness of
the beamtocolumn joints in Fig. 12.8 is very low, the
plastic moment capacity of the beam at midspan
M
pl,Rd
may be reached first. The moment capacity of
the end joints M
j,Rd
can only be attained if the beam
has sufficient rotation capacity at the location of the
plastic moment. In the case of joints with a very low
stiffness, this might not be the case, e.g. see curve "e"
in Fig. 12.10.
If the stiffness of the joint is high, the (partial) strength
capacity of the beamtocolumn joints (e.g. curve "b"
in Fig. 12.10) may be reached first. Now, these joints
should have sufficient deformation capacity to allow
the plastic moment capacity of the beam at midspan
to be attained. The joints with behaviour "a" or "c" are
stronger than the connected beam, thus the beam
should have sufficient deformation capacity if the
plastic moment is reached first at the ends.
Hence, for an accurate analysis of frames with
semirigid joints, an appropriate description of the
momentrotation behaviour is required as well as
evidence regarding:
 Stiffness (at serviceability and at the ultimate limit
state)
 Strength (ultimate limit state)
 Rotation capacity
However, all this information is not yet generally
available for beamtotubular column joints, although
research has been initiated by Weynand et al. (2006).
153
154
Alternative options are to use joints with such a
stiffness that the joints can be classified as (nearly)
rigid or (nearly) pinned. For both cases, limits can be
defined. However, the deflections can only be
determined properly if the joint stiffnesses are known
(Fig. 12.11).
Part 8 of Eurocode 3 (EN 199318, 2005) provides
stiffness classifications, see Fig. 12.12. A possible
joint modelling approach is given in Fig. 12.13.
For I beamtotubular column joints, the factors c
1
and
c
2
still have to be defined. Another complication is that
the axial loads and moments in the column do not
only affect the strength but also the stiffness, as
shown by the curve in Fig. 12.13 for N
0
0.
12.6 CONCLUDING REMARKS
The design recommendations in CIDECT Design
Guide No. 9 (Kurobane et al., 2004) give strength
functions for beamtoCHS and beamtoRHS column
joints.
This chapter intends to give basic background
information without going too much into detail in the
resulting design formulae. The design of frames with
semirigid joints is not typical for tubular structures
and is therefore not described here in detail.
Fig. 12.1 Unstiffened I beamtoCHS and I beamtoRHS column joints
B
f
d
0
t
0
B
f
b
1
t
p
b
0
t
0
t
p
b
f
B
f
d
0
t
0
B
f
b
1
t
p
b
0
t
0
t
p
b
f
Fig. 12.2 Flange diaphragm I beamtoCHS and I beamtoRHS column joints
155
a. Beam flange failure
b. Weld failure
Joint configuration
c. Lamellar tearing
d
1
. Column face plastification
d
2
. Column wall plastification (side view)
f. Column local wall buckling (side view)
e. Column punching shear
g. Column shear failure
Fig. 12.3 Modes of failure for I beamtoRHS column joints
156
Fig. 12.4 Stress distribution in beam flange
Fig. 12.5 RHS column face plastification
t
1
0,5b
e,p
t
1
0,5b
e,p
Fig. 12.6 RHS column punching shear
157
Fig. 12.7 Column shear failure
Fig. 12.8 Beam with various end conditions
158
0,67
0,50
0,33
0,67
0,50
0,33
Fig. 12.9 Variation of elastic moment distribution with joint stiffness
Fig. 12.10 Various M characteristics
159
1,0
0,8
0,6
0,4
0,2
0
1,0
0,8
0,6
0,4
0,2
0
Fig. 12.11 Variation of midspan deflection with joint stiffness
,
,
,
,
Fig. 12.12 Boundaries for stiffness classification of beamtocolumn joints according to Eurocode 3 (EN
199318)
Ed Ed
Fig. 12.13 M modelling
160
13. BOLTED JOINTS
The calculation methods used for bolted joints
between, or to, hollow sections are basically not
different from those used for any other type of joint in
conventional steel construction.
Most details given in this chapter are presented
without (detailed) design formulae.
13.1 FLANGE PLATE JOINTS
13.1.1 Flange plate joints to CHS under
axial tension load
For the flange plate joints shown in Fig. 13.1, various
investigations were carried out (Kato & Hirose, 1984;
Igarashi et al., 1985; Cao & Packer, 1997).
Economical joints under tension load can be obtained
if prying force is permitted at the ultimate limit state,
with the connection proportioned on the basis of a
yielding failure mechanism of the flange plates. In
CIDECT Design Guide No. 1 (Wardenier et al., 2008a)
formulae and tables are given, based on the work of
Igarashi et al. (1985). In the context of this book, only
the failure modes are presented (Fig. 13.2). It is
preferable to design primary structural joints on the
basis of the yield resistance of the circular hollow
section.
13.1.2 Flange plate joints to RHS under
axial tension load
Research by Birkemoe & Packer (1986) and Packer et
al. (1989) on bolted RHS flange plate joints with bolts
on two sides of the RHS only, see Fig. 13.3, showed
that in principle the strength of these joints can be
analysed on the basis of the traditional prying model
developed for Tstubs by Struik & De Back (1969).
The location of the plastic hinge lines may be adjusted
for greater accuracy, i.e. the distance b in Fig. 13.4 is
adjusted to b' according to:
i
t
2
d
b ' b + = (13.1)
Detailed formulae are given by Packer & Henderson
(1997) and Packer et al. (2009a).
Many tests have been carried out on RHS flange plate
joints with bolts on 4 sides of the RHS, as shown in
Fig. 13.3. A thorough study of this type of bolted joint
has been undertaken by Willibald et al. (2002, 2003a).
It was revealed that RHS flange plate joints bolted on
all four sides could still be proportioned on the basis of
the twodimensional Tstub prying model of Struik &
De Back (1969), with some minor modifications.
Following the procedure for bolted RHS flange plate
joints with bolts on two sides, the inner yield lines in
the flange plate can now be expected adjacent to the
RHS outer face and hence the term t
i
should be
deleted from eq. (13.1). The bolt pitch to be used is
the minimum of p from both sides. The dimension p,
the plate width or depth divided by the number of bolts
in that direction, is illustrated in Fig. 13.3. This
"minimum p" value is then used in the joint analysis
based of a twodimensional prying model. In order for
this design model to be valid, the centres of the bolt
holes should not be positioned beyond the corners of
the RHS (as illustrated in Fig. 13.3).
Detailed information can be found in CIDECT Design
Guide No. 3 (Packer et al., 2009a).
13.1.3 Flange plate joints to CHS or RHS
under axial tension load and
moment loading
Design methods for bolted flange plate joints to date
have generally been developed for axial tension
loading. Frequently, however, hollow sections are
subjected to both axial tension load (N
i
) and bending
moment (M
i
). In such cases, a hypothetical "effective"
axial load can be computed (Kurobane et al., 2004)
for use with the flange plate joint design procedures
given in Sections 13.1.1 and 13.1.2:
i
i
i
i
i
A
W
M
A
N
axial Effective


.

\

= (13.2)
where:
A
i
cross sectional area of the CHS or RHS
W
i
elastic (or plastic) section modulus of the CHS or
RHS
This procedure will be conservative, especially for
CHS, as it computes the maximum tensile normal
stress in the CHS or RHS and then applies this to the
whole member cross section.
13.2 END JOINTS
Some bolted end joints are shown in Fig. 13.5. The
flange of the tee in Fig. 13.5d, as well as the other
flange plates perpendicular to the CHS or RHS
161
section, must be sufficiently thick to effectively
distribute the load to the cross section (Wardenier et
al., 2008a; Packer et al., 2009a), see also Section
9.7.3.
13.3 GUSSET PLATE JOINTS
For bolted gusset plate joints, the design can be
based on the various possible failure modes, e.g. for a
tension member:
 Yielding of the cross section
 Rupture of the net area
 Rupture of the effective net area reduced for shear
lag
Similar to other bolted joints, the total net area is the
sum of individual net areas along a potential critical
section of a member or gusset plate, see Fig. 13.6. If
such a critical section comprises net areas loaded in
tension and segments loaded in shear, the shear
segments should be multiplied by the shear strength
and the tension areas by the ultimate strength.
Eurocode 3 (EN 199311, 2005) specifies a
M
factor
of 1,0 for yielding and 1,25 for ultimate strength
(rupture).
A failure mode of the gusset plate which also must be
checked is yielding across an effective dispersion
width of the plate, which can be calculated using the
Whitmore (1952) effective width concept illustrated in
Fig. 13.7. For this failure mode (for one gusset plate),
the strength is given by:
( )
M
o
p yp
1
p ) 30 (tan 2 g t f N
Rd , i
E + = (13.3)
where the term Ep represents the sum of the bolt
pitches in a bolted connection or the length of the
weld in a welded connection, and
M
=1,1.
If the member is in compression, buckling of the
gusset plate must also be prevented.
Fig. 13.8 shows some examples of bolted gusset plate
joints. It must be borne in mind that fitting of these
connections is very sensitive with regard to
dimensional tolerances and to deformations of the
welded gusset due to weldinduced distortions. Thus,
care has to be taken to ensure fitting at site.
When a member is connected by some, but not all
parts of its cross section elements and if the net
section includes elements which are not connected,
the net area perpendicular to the load has to be
multiplied by a shear lag factor which depends on the
shape of the section, the number of connected faces
and the number of transverse rows of fasteners.
Such a case is illustrated in Fig. 13.8b where bolting
plates are welded to the sides of the RHS brace
member. For welds parallel to the direction of load (as
the four flare groove welds would be in Fig. 13.8b,
along the four corners of the RHS), the shear lag
factor is a function of the weld lengths and the
distance between them. For the RHS, the shear lag
reduction factors can be applied to each of the four
sides (two of width w = b
i
 t
i
, and two of width w = h
i

t
i
), to produce a total effective net area of the RHS
reduced by shear lag. Suggested shear lag reduction
factors for these four element areas, in terms of the
weld length L
w
, are (CSA, 2009):
 1,00 when the weld Iengths (L
w
) along the RHS
corners are > 2b
i
(or 2h
i
as applicable)
 (0,5 + 0,25L
w
/b
i
) when the weld lengths along the
RHS corners are b
i
s L
w
< 2b
i
, or
 (0,5 + 0,25L
w
/h
i
) when the weld lengths along the
RHS corners are h
i
s L
w
< 2h
i
 0,75L
w
/b
i
when the weld lengths along the RHS
corners are L
w
< b
i
(or h
i
as applicable)
13.4 SPLICE JOINTS
Fig. 13.9 shows a splice joint for circular hollow
sections. This type of connection can, for example, be
executed with four, six or eight strips welded
longitudinally on the periphery of the hollow sections
and connected by double lap plates, one on each
side.
Lightly loaded splice joints in tension can be made as
shown in Fig. 13.10 and for architectural appearance
the bolts can be hidden. Using one plate on each side,
instead of the solution in Fig. 13.10, provides a more
fabricationfriendly solution. Such an eccentric joint,
however, may have little stiffness and resistance to
outofplane flexure under compression loading, thus
the designer should be confident that such a condition
has been considered. Experimental and numerical
research on this RHS joint type, under tension loading,
has been conducted by Willibald et al. (2003b).
13.5 BEAMTOCOLUMN JOINTS
Bolted beamtocolumn joints can be designed in
various ways, mainly depending on the type of load
that has to be transmitted. In general, shear joints are
simpler to fabricate than moment joints. Typical joints
162
163
are given in Figs. 13.11 to 13.15 without detailed
description.
13.6 BRACKET JOINTS
Some typical joints for lightly loaded beams are shown
in Fig. 13.16.
13.7 BOLTED SUBASSEMBLIES
Lattice structures are often connected to columns by
bolted flanges, plates or Tee profiles. Some examples
are shown in Fig. 13.17.
13.8 PURLIN JOINTS
Fig. 13.18 shows some examples of purlin joints for
trusses with CHS or RHS chords.
13.9 BLIND BOLTING SYSTEMS
Due to the closed nature of hollow sections, in many
cases additional welded plates are used for bolted
joints. However, solutions are then not aesthetically
appealing. Nowadays, bolting systems are available
which can be used when only one side of the
connection is accessible. Blind bolting systems make
use of either special types of bolts or inserts or special
drilling systems.
13.9.1 Systems using bolts and inserts
Special types of bolts and systems allow one to bolt
from one side of a hollow section. A number of
patented blind bolting systems is available, e.g. Huck
"Ultra Twist Blind Bolt" and Lindapter "HolloFast" and
"HolloBolt". The latter, which uses a special insert and
a standard bolt, has been investigated by CIDECT
(Sidercad & British Steel, 1996; Yeomans, 1998) with
regard to its axial, shear and bending capacity (see
Fig. 13.19).
The systems are based on the principle that after
bringing them in from one side, the bolts are torqued
and a "bolt head" forms on the inside of the connected
plies.
The design rules for blind bolting systems are based
on typical failure modes, i.e.
 Punching shear of the fastener through the column
face
 Yielding of the column face (yield line pattern
around the bolts)
 Bolt failure in shear, tension or a combination of
both
13.9.2 Drilling system
The Flowdrill system, see Fig. 13.20, is a special
patented method for extruded holes. CIDECT has
carried out research (Yeomans, 1994; British Steel,
1996) to assess the load bearing capacity of this type
of joint in structural hollow sections.
Flowdrilling is a thermal drilling process (Fig. 13.21) to
make a hole through the wall of a hollow section by
bringing a tungsten carbide bit into contact with the
hollow section wall and generating sufficient heat by
friction to soften the steel. As the bit moves through
the wall, the metal flows to form an internal bush. In
the next step, the bush is threaded using a roll tap.
Conventional bolts are then used in this tapped hole.
Bolting to hollow sections with wall thicknesses up to
12,5 mm can be recommended by using the Flowdrill
method, see Yeomans (1994).
13.10 NAILED JOINTS
As an alternative to bolting or welding, steel circular
hollow sections can be nailed together to form reliable
structural joints. Up to now, this method of connection
has only been verified for splice joints between two
coaxial tubes (see Fig. 13.22). In such a joint, one
tube can fit snugly inside the other, in such a way that
the outside diameter of the smaller equals the inside
diameter of the larger. Nails are then shot fired and
driven through the two wall thicknesses and arranged
symmetrically around the tube perimeter.
As an alternative, two tubes of the same outside
diameter can be joined by means of a tubular collar
over both tube ends; in this case nails are again
inserted by driving them through the two tube walls.
Research to date has covered a range of tube sizes
with various diametertothickness ratios, tube wall
thickness and lack of fit (Packer, 1996). The observed
failure modes were nail shear failure, tube bearing
failure, and net section fracture of the tube. These
failure modes have been identified for both static and
fatigue loading. Simple design formulae, derived from
bolted and riveted joints, have been verified for both
these load cases.
Fig. 13.1 Bolted CHS flange plate joint
Fig. 13.2 Failure modes for bolted CHS flange plate joints
164
p p p
p
p
p p p
p
p
Fig. 13.3 Bolted RHS flange plate joints
Fig. 13.4 RHS flange plate joint with bolts at two sides of the RHS
165
Fig. 13.5 Bolted end joints
Bolt hole diameter d
Shear segments
Inclined segments
Tension segment
Bolt hole diameter d
Shear segments
Inclined segments
Tension segment
Total net area for critical section AA is the sum of the individual segments:
For tension segment : A
n
= (g
1
 d/2)
t
For shear segment : A
gv
= L t
For each inclined segment : A
n
= (g
2
 d)
t + (s
2
/4g
2
)
t
Fig. 13.6 Calculation of total net area for a gusset plate
166
,,
Fig. 13.7 Whitmore criterion for gusset plate yielding
Fig. 13.8 Some examples of bolted gusset plate joints
Fig. 13.9 Bolted splice joint for CHS
167
Fig. 13.10 Hidden bolted splice joint
IPE or HE cut off IPE or HE cut off
Fig. 13.11 I section beamtoCHS column joints
168
a
b
c
d
e
f
Fig. 13.12 I section beamtoRHS column simple shear joints
169
a
b
c
d
Fig. 13.13 Moment joints between open section beams and CHS or RHS columns
170
Fig. 13.14 RHS sections connected to I section columns
Fig. 13.15 Knee joint assemblies for portal frames
Fig. 13.16 Bracket joints
171
a
b
c
d
e
f
Fig. 13.17 Bolted joints for lattice girder supports
172
a
b
c
d
e
f
Fig. 13.18 Purlin joints
Fig. 13.19 Lindapter "HolloFast" connection
173
Fig. 13.20 Flowdrill connection for joining end plates or angles to RHS
Fig. 13.21 Flowdrill process
Fig. 13.22 Nailed CHS joint
174
14. FATIGUE BEHAVIOUR OF
HOLLOW SECTION JOINTS
Fatigue is a mechanism whereby cracks grow in a
structure under fluctuating stress. Final failure
generally occurs when the reduced cross section
becomes insufficient to carry the load without rupture.
Generally the fatigue cracks start at locations with
high stress peaks.
High stress peaks may occur at local notches, e.g. at
welds (Fig. 14.1). However, geometric peak stresses
may also occur due to the geometry, e.g. at holes or
in hollow section joints due to the nonuniform
stiffness distribution at the perimeter of the connection
(Fig. 14.2).
Thus the fatigue behaviour is largely influenced by the
loading and the way the members are connected. In a
fatigue analysis the loading and the loading effects
should be carefully evaluated and compared with the
fatigue resistance.
Sometimes it is difficult to determine the loading
effects accurately, e.g. the secondary bending
moments in lattice girders. In such cases, simplified
approaches can be used.
14.1 DEFINITIONS
Stress range
The stress range (shown in Fig. 14.3) is the
difference between the maximum and the minimum
stress in a constant amplitude loading regime.
Stress ratio R
The stress ratio R is defined as the ratio between the
minimum stress and the maximum stress in a stress
cycle of constant amplitude loading (Fig. 14.3).
AN or Whler line
The relation (on a loglog scale) between the stress
range and the number of cycles N to failure is
presented in a socalled AN or Whler line (Fig.
14.4).
Fatigue strength
The fatigue strength of a welded component is defined
as the stress range , which causes failure of the
component after a specified number of cycles N.
Fatigue life
The number of cycles N to a defined failure is known
as the endurance or fatigue life.
Fatigue limit
The fatigue limit is defined as the stress range below
which it is assumed that no fatigue failure occurs for a
constant amplitude loading, see Fig. 14.4. For
Eurocode 3 (EN 199319, 2005) and the IIW
recommendations for hollow section joints (IIW, 1999),
for example, this occurs at N = 5 x 10
6
cycles. Note:
IIW (2008) for fatigue design of welded joints and
components has recently changed this limit to 10
7
cycles.
Cut off limit
The cut off limit is defined as the stress range below
which it is assumed that the stress ranges of a
variable amplitude loading do not contribute to the
fatigue damage. For Eurocode 3 (EN 199319), for
example, this occurs at N = 10
8
cycles (see Fig. 14.5).
Note: The recently revised recommendations of IIW
(2008) no longer give a cut off limit for variable
amplitude loading.
Geometric stress
The geometric stress, also called hot spot stress, is
defined as the extrapolated principal stress at a
specified location at the weld toe. The extrapolation
must be carried out from the region outside the
influence of the effects of the weld geometry and
discontinuities at the weld toe, but close enough to fall
inside the zone of the stress gradient caused by the
global geometrical effects of the joint. The
extrapolation is to be carried out on the brace side
and the chord side of each weld (see Fig. 14.6).
Generally the geometric stress (or the hot spot stress)
can be determined by considering the stress normal to
the weld toe since the orientation of the maximum
principal stress is normal or almost normal to the weld
toe.
Stress concentration factor
The stress concentration factor (SCF) is the ratio
between the geometric peak stress, or hot spot stress,
excluding local effects, at a particular location in a
joint and the nominal stress in the member due to a
basic member load which causes this geometric
stress.
14.2 INFLUENCING FACTORS
The fatigue behaviour can be determined either by
N methods or with a fracture mechanics approach.
The various N methods are based on experiments
resulting in N graphs (Fig. 14.4) with a defined
175
stress range on the vertical axis and the number of
cycles N to a specified failure criterion on the
horizontal axis.
The relation between the number of cycles to failure N
and the stress range can be given by:
) ( C N
m
o A = (14.1)
or:
( o A = log m C log N log ) (14.2)
On a loglog scale, this gives a straight line with a
slope of m, see Fig. 14.4.
The fracture mechanics approach is based on a
fatigue crack growth model and is not further
discussed in the context of this book.
Due to the appearance of residual stresses, the stress
ratio R =
min
/
max
(see Fig. 14.3) is not taken into
account in modern fatigue design. Only if the structure
is fully stress relieved, it might be advantageous to
take the stress ratio into account.
Fig. 14.7 shows an example illustrating the influence
of residual stresses on a plate with f
y
= 240 N/mm
2
. A
nominal stress range
nom
= 120 N/mm
2
is applied on
the plate, resulting in a nominal average stress range
in the net cross section at the hole of:
2
nom
N/mm 160 120
60
80
= = o A
With a geometric stress concentration factor SCF = 3
for this detail, this nominal stress range results in a
theoretical geometric stress range:
geom
= SCF
nom
= 3 x 160 = 480 N/mm
2
However, at 240 N/mm
2
the material is yielding,
resulting in stress pattern "b". Unloading to zero
(loading in the opposite sense) means that the elastic
stress pattern "a" has to be subtracted from "b"
resulting in residual stress pattern "c". The next cycle
will thus fluctuate between patterns "c" and "b", i.e.
starting from a residual stress equal to the yield
stress.
In welded structures, residual stresses and
discontinuities exist at welds too. Therefore, it is
assumed that locally the stress will fluctuate between
the yield stress and a lower limit (f
y

geom
).
The example in Fig. 14.7 with a tensile loading results
in a compressive residual stress at the hole, which is
favourable. However, it also shows that a
compressive loading would have resulted in a tensile
residual stress at the hole, which is unfavourable. In
this latter case, it would result in fatigue failure under
an external nominal compression stress range . In
practical situations, various stress ranges occur and
the residual stress ranges are not known. That is why,
in general, no difference is made between tensile and
compressive loading.
For simple details, e.g. butt welded, endtoend
connections, cover plates, attachments, etc., N
lines can be determined. These N lines (Figs. 14.4
and 14.5) already take into account the effect of the
welds and the geometry of the connection.
For more complicated geometries, such as the directly
welded joints between hollow sections, the peak
stresses and thus the peak stress ranges depend on
the geometrical parameters. This means that for every
joint, a separate N line exists. Hence, in modern
design, the fatigue behaviour of hollow section joints
is related to basic N curves, in which
represents a geometrical or hot spot stress range
taking into account the effect of the geometry of the
joint. This geometrical stress range can also be
calculated by multiplying the nominal stress range
with a relevant stress concentration factor SCF. As a
consequence of this method, stress concentration
factors should be available for joints with various
geometries. The effect of the weld is included in the
basic N lines. As shown later, the thickness of the
sections is also an influencing parameter.
14.3 LOADING EFFECTS
As stated before, the stress range is a governing
parameter for fatigue.
For constant amplitude loading, the N line is
generally cut off between N = 2 x 10
6
to 10
7
cycles
depending on the code. CIDECT (Zhao et al., 2001)
and IIW (1999) follow Eurocode 3 (EN 199319,
2005) and use 5 x 10
6
cycles (see Fig. 14.4).
For random loading, also called spectrum loading, the
smaller cycles may also have an effect and the cut off
is taken at 10
8
to 2 x 10
8
cycles, depending on the
code. Eurocode 3 (EN 199319) uses a cut off at N =
10
8
cycles. Between 5 x 10
6
to 10
8
cycles, a N line
is used with a shallower slope of m = 5.
176
Although the sequence of the stress ranges affects
the damage, the simplest and best rule available up to
now to determine the cumulative damage, is the
PalmgrenMiner linear damage rule, i.e.
0 , 1
N
n
i
i
s E (14.3)
in which n
i
is the number of cycles of a particular
stress range
i
and N
i
is the number of cycles to
failure for that particular stress range.
If vibration of the members occurs, the nominal stress
ranges may be considerably increased. This will be
the case if the natural frequency of a structural
element is close to the frequency of the loading. It is
therefore essential to avoid this.
For very high stress ranges and consequently a low
number of cycles, low cycle fatigue may occur. In this
case, no N curves are given in the codes since the
fatigue is mainly governed by the strains. Research by
Van der Vegte et al. (1989) showed that the various
N curves can be used if translated into strain, i.e.
N curves. In this case, the loading should be
evaluated based on the resulting strain range .
14.4 FATIGUE STRENGTH
In IIW (1999), CIDECT (Zhao et al., 2001) and
Eurocode 3 (EN 199319, 2005) the fatigue strength
for members with butt welded or fillet welded
endtoend connections, with plates or members with
attachments, etc. is given as the stress range at 2 x
10
6
cycles. This classification is in line with the
classifications given for other members such as I
sections, etc. Table 14.1 shows some of these
classifications, adopted by Eurocode 3 (EN 199319,
2005), which are based on the analyses of Noordhoek
et al. (1980) and Wardenier (1982).
Table 14.1 shows that the classification of 160 or 140
N/mm
2
for plain sections without attachments or
connections is much higher than the classification for
details.
Table 14.1 further makes clear that a butt welded
endtoend connection has a better fatigue behaviour
than a fillet welded endtoend connection with an
intersecting plate.
For welded details, the classification is independent of
the steel grade. Higher strength steels are more
sensitive to notches, which reduces the favourable
effect of the higher strength. However, if improved
welding methods are used or post weld improvements
(grinding, TIG dressing, plasma dressing, ultrasonic
impact treatment (UIT), etc.) are carried out, higher
fatigue strengths can be obtained, especially for
higher strength steels (e.g. for S460 and S690).
14.5 PARTIAL FACTORS
If fatigue actions are included in a design, it should be
noted that the stress ranges produced by the
(unfactored) loading have to be multiplied by a partial
(safety) factor which depends on the type of structure
(failsafe or non failsafe) and the possibility for
inspection and maintenance. The recommended
partial (safety) factors according to CIDECT (Zhao et
al., 2001) are given in Table 14.2. For example, a
partial factor of
M
= 1,25 and m = 3 reduces the
design fatigue life by about a factor of (1,25)
3
~ 0,5.
14.6 FATIGUE CAPACITY OF WELDED
JOINTS
In Section 14.4 the basic parameters influencing the
strength of members with attachments or with
endtoend connections are discussed. It has already
been stated that in welded joints between hollow
sections the stiffness around the intersection is
nonuniform, resulting in a geometrical nonuniform
stress distribution as shown in Fig. 14.2 for an X joint
of circular hollow sections. This nonuniform stress
distribution depends on the type of brace loading
(axial, inplane bending, outofplane bending), the
joint type and the geometry. Therefore, fatigue design
of hollow section joints is, in general, different from,
for example, that of simple welded connections
between plates.
14.6.1 Geometrical stress approach
Since peak stresses determine the fatigue behaviour,
in modern design methods, the fatigue design is
related to the geometrical stress range of a
connection. This geometrical stress range includes
the geometrical influences but excludes the effects
related to fabrication such as the configuration of the
weld (flat, convex, concave) and the local condition at
the weld toe (radius of weld toe, undercut, etc.). Since
the geometrical peak stress (also called the hot spot
stress) can only be determined with finite element
methods or by measurements on actual specimens,
stress concentration factors have been developed for
177
the basic types of joints and basic loadings. These
stress concentration factors are defined as the ratio
between the geometrical (peak) stress and the
nominal stress causing the geometrical stress, for
example for an axially loaded X joint without chord
loading:
brace the in stress nominal
stress l geometrica
SCF
k , j , i
k , j , i
= (14.4)
where:
i chord or brace
j location, e.g. crown, saddle or in between for CHS
joints
k type of loading
Thus, several SCFs have to be determined for various
likely critical locations (Fig. 14.6). The maximum SCF
and the location depend on the geometry and loading.
This is particularly important for combined loadings.
Thus, for determining the fatigue life the geometrical
(peak) stress range has to be calculated for the
various locations:
) SCF (
k , j , i nom geom
k , j , i k , j , i
o A E = o A (14.5)
The N line to be considered is also based on the
geometrical stress range.
Initially, it was expected that one N curve could be
used. However, for the same geometrical stress the
strain gradient may be different, resulting in a different
fatigue life. Since the gradient is steeper for thin
walled members than for thick walled members, this
effect is accounted for by a thickness correction.
The geometrical N curves in IIW (1999) and
CIDECT (Zhao et al., 2001) are related to fatigue
class 114 (at 2 x 10
6
cycles) for a wall thickness of 16
mm. For joints with wall thicknesses other than 16 mm,
a wall thickness function should be applied.
The N curves with the thickness correction
included are given in Fig. 14.8. It should be noted that
no further thickness effect should be used for
thicknesses less than 4 mm, since the weld
performance may overrule the geometrical influence,
which may sometimes result in lower fatigue strengths
(Wardenier, 1982; Mashiri et al., 2007).
The curves presented in Fig.14.8 slightly deviate from
the classifications previously given in CIDECT Design
Guide No. 6 (Wardenier et al., 1995). However, the
reanalysis by Van Wingerde et al. (1997) showed that
one set of curves with a single, general thickness
correction factor can be conservatively used for all
joint types of CHS and RHS.
For a designer it is important to have an insight into
the parameters which determine the stress
concentration factors. Optimal design requires the
stress concentration factors to be as low as possible.
As an indication, the stress concentration factors are
given for various joint configurations. Fig. 14.9 shows
the stress concentration factors for axially loaded X
joints of circular hollow sections at four locations, i.e.
for the chord and brace, at the crown and saddle.
The following conclusions can be made:
For the chord:
 Generally the highest SCF occurs at the saddle
position
 The highest SCFs at the saddle are obtained for
medium ratios
 SCFs decrease with decreasing value
 SCFs decrease with decreasing 2 value
For the brace:
 Generally the same applies as for the chord;
however, a decrease in value results in an
increase in SCF at the crown position (in some
cases the graphs are only given here for = 1,0).
 SCF in the brace may become critical compared to
that in the chord for small values; however, the
brace thickness is then smaller than the chord
thickness. Considering the thickness effect, it is
most likely that the chord location is still the critical
position.
For X joints of square hollow sections (Van Wingerde,
1992), the SCFs are given for various locations in Fig.
14.10. As mentioned before, similar observations can
be made as for circular hollow section X joints:
 For = 1,0 the highest SCFs generally occur in the
chord at locations B and C
 The highest SCFs are found for medium ratios
 The lower the 2 ratio, the lower the SCF
 The lower the ratio, the lower the SCF in the
chord, whereas has a small influence on the SCF
in the brace.
Fig. 14.11 shows the SCFs for an axially loaded K gap
joint of circular hollow sections with g = 0,1d
0
. Here,
the same observations apply as for X joints, but the
SCFs are considerably lower due to the stiffening
effect of both braces with opposite loadings connected
to the chord face. Here, only maximum SCF values for
chord and brace are given (i.e. no differentiation is
178
made between the saddle and crown positions).
Although the effect of the brace angle is not included
in the figures, a decrease in the angle between brace
and chord results in a considerable decrease in SCF.
If the chord is loaded, the geometrical or hot spot
stress range at the chord locations at the crown (CHS)
or at the locations C and D (RHS) (see Fig. 14.6) has
to be increased by the chord nominal stress range
multiplied by the stress concentration factor produced
by the chord stresses. As shown in Fig. 14.12 for RHS
T and X joints, the SCF for chord loading varies
between 1 and 3, depending on the loading, joint type
and geometrical parameters (Van Wingerde, 1992).
Another aspect to be considered is multiplanar
loading. For the same type of joint and the same
geometry, different SCFs can be obtained for different
loading conditions (see Fig. 14.13).
All the SCFs are based on measurements taken at the
toe of the weld, since this location is generally critical.
However, for very low SCFs, crack initiation can start
at the weld root. Therefore a minimum SCF = 2,0 is
recommended.
Further, the SCF values have been determined for
butt or groove welded joints. Fillet welds cause slightly
lower SCFs in the chord, but due to local wall bending
effects, result in considerably higher SCFs in the
braces. It is therefore recommended to increase the
SCFs in RHS braces of T and X joints by a factor of
1,4 when fillet welds are used (Van Wingerde, 1992).
Considering all these aspects, it can be concluded
that optimal design can be achieved if the SCFs are
as low as possible. Thus the following guidelines can
be given:
 Avoid medium ratios. Ratios close to = 1,0 give
the lowest SCFs.
 Make the wall thickness of the brace as low as
possible (low ratio).
 Take relatively thick walled chords (low 2 ratio).
In this way, SCFs of about 2 to 4 are possible,
resulting in an economic design.
If the bending moments in girders have not been
determined by finite element analyses or other
methods, e.g. a rigid frame analysis, the effect of the
bending moments can be incorporated in a simplified
manner as discussed in Section 14.6.2.
More detailed information about stress concentration
factors can be obtained for CHS joints from Efthymiou
(1988) and Romeijn (1994) and for RHS joints from
Herion (1994) and Van Wingerde (1992).
14.6.2 Classification method based on
nominal stress ranges
To simplify design methods, it would be easier if the
SCFs could already be incorporated in the design
class, taking account of the main influencing
parameters. However, this is impossible for T, Y and
X joints, since the variation in SCF is considerable.
For K joints, classification is possible to a certain
extent. Table 14.3 shows the detail categories for
lattice girder joints given in Eurocode 3 (EN 199319,
2005), based on the analyses of Noordhoek et al.
(1980) and Wardenier (1982). The thickness effect is
indirectly included. The method can only be used for
thin walled sections (see recommended range of
validity). As shown in Table 14.3, the classification
only depends on the gap, overlap and ratio. It must
be stated that the design classes are based on test
results and on an independent analysis, and do not
comply with the geometric stress method. Further, the
design classes given clearly show the advantage of
using a low ratio or high t
0
/t
i
.
Since it might be difficult for designers to determine
the secondary bending moments, the CIDECT and
IIW recommendations allow a calculation based on
the assumption that the members are pin connected.
However, the stress ranges in chord and braces
caused by the axial loading in the braces have to be
multiplied by the factors given in Tables 14.4 and 14.5
to account for the secondary bending moments due to
joint stiffness. These multiplication factors are based
on measurements in actual girders (De Koning &
Wardenier, 1979). In the case of eccentricities or
loads in between the joints, a calculation based on the
model in Fig. 6.8 is recommended. Although the
effects of the eccentricities and loads in between the
joints are then covered, the effects of joint stiffness
still have to be added in a similar way as described
here before.
14.7 FATIGUE CAPACITY OF BOLTED
JOINTS
Joints with pretensioned high strength friction grip
(HSFG) bolts have a favourable fatigue behaviour
compared to joints with nonpretensioned bolts. Joints
loaded in tension can, for example, be designed in
179
such a way that the fatigue load does not critically
affect the joints. Joints with pretensioned high strength
bolts subjected to shear or friction loads can bear
higher fatigue loads than welded joints, which means
that, in general, these bolted joints in shear do not
represent the most critical elements. Bolted joints
under fatigue loading should be designed in such a
way that there is no play and no slip between the
faying surfaces.
14.7.1 Bolted joints under tensile load
Theoretical investigations and experimental results
(Bouwman, 1982) show that the fatigue behaviour of
joints with pretensioned bolts is significantly
influenced by the way in which the load is transmitted.
A few examples of bolted joints in tension are shown
in Fig. 14.14.
For configurations a
1
and a
2
(Fig. 14.14) the contact
face pressure is, by design, coaxial with the external
load. In this case, as the load increases, at first there
is a significant decrease in the contact pressure. Only
when the applied load exceeds the contact pressure,
the load in the high strength bolts begins to increase
appreciably. For configurations c
1
and c
2
, the load in
the high strength bolts increases from the very
beginning with growing applied load due to prying.
Thus, connection types a
1
and a
2
exhibit a much
better fatigue performance than the other types.
For the a
1
and a
2
configurations, the stiffness of the
flanges is distinctly higher than that for the axial load
transmission through the bolts accompanied by flange
bending. As a result, for configurations a
1
and a
2
under fatigue load, only a small stress increase in the
bolts is observed. This change in stresses due to
fatigue loading can generally be neglected, when the
applied load per bolt is smaller than the bolt preload.
Should the load transmitting parts not be located on
the same plane, the connections must then be
designed in such a manner that the load transmission
is mainly effected through a reduction of contact
loads. The arrangements shown in Fig. 14.15 are
recommended for bolted ring flange joints between
hollow sections. These proposed flange connections
require expensive machining and finishing. However,
the same level of load transmission can be achieved
simply with the help of packing plates (shims). It is
further recommended that the high strength bolts be
located as closely as possible to the load carrying
structural parts.
Snugtight bolted connections without pretensioning
should be avoided, since the fatigue behaviour is bad.
14.7.2 Bolted joints under shear load
The fatigue behaviour of high strength pretensioned
bolted joints under shear load is generally better than
that of welds connecting hollow sections to end plates.
As shown in Fig. 14.16, the stress distribution in joints
with pretensioned bolts is significantly better than that
in bolted joints without preload. The reason is that part
of the external load is transmitted by friction around
the bolt hole. After a high strength pretensioned bolted
joint has slipped, a more unfavourable stress
distribution develops than before, since part of the
force is now transmitted by pressure on the face of the
hole (see Fig. 14.16c).
Nonpretensioned, nonfitted bolts should be avoided
for structural parts subjected to fatigue loading.
Recommendations for the design class of high
strength bolted joints are, for example, given in
Eurocode 3 (EN 199319, 2005), see Table 14.6.
It is noted that, depending on the conditions, partial
(safety) factors have to be used; see e.g. Table 14.2.
14.8 FATIGUE DESIGN
In the previous sections the fatigue resistance of
hollow section joints has been discussed in relation to
the geometry. However, when a designer starts with
the design process, the geometry is not known and he
or she has first to determine the geometry, e.g. for a
truss. Here, several steps have to be followed:
Step A:
For the determination of the geometry, use has to be
made of the knowledge obtained in the previous
sections. For a good performance of joints, the
following points have to be considered to obtain the
best fatigue behaviour:
 Select joints with high or low values, and avoid
intermediate values. Considering fabrication, = 0,8
is preferred to = 1,0.
 Choose the chord diameter or chord widthto
thickness ratio 2 to be as low as possible; e.g. for
tension chords about 20 or lower and for
compression chords about 25 or lower.
 Design the braces to be as thin as possible to
achieve preferable bracetochord thickness ratios
s 0,5 and to minimise welding.
180
If the chord is also subjected to large loads, a larger
cross section should be taken to account for these
chord loads.
 Design girders in such a way that the angles
between braces and chord are preferably about 40.
Considering the above, joints with relatively low stress
concentration factors can be achieved. For example,
for a K joint of circular hollow sections (Fig. 14.11)
with = 0,8; 2 = 20; = 0,5 and = 40:
Step G:
The configuration now determined should be checked
for ease of fabrication, inspection and the validity
range for joints.
SCF chord ~ 2,1
If the design satisfies the requirements, the final check
can be carried out, now starting from the loading and
the known geometry.
SCF brace ~ 2,0
Step B:
Based on the required lifetime, the number of cycles N
has to be determined.
The procedure is then as follows:
(1) Determine loads and moments in members and
from these, the stresses in braces and chord, e.g.
assuming pin ended braces to continuous chords
if nodal eccentricities between the centrelines of
intersecting members at the joints should be
included, see Fig. 14.17.
Step C:
For the number of cycles N (from B), the geometric
stress range
geom
can be obtained from the N
line assuming a certain thickness, see Fig. 14.8.
(2) Determine from (1) the nominal stress ranges
nom,brace
in the brace and
nom,chord
in the chord.
Step D:
Depending on the inspection frequency and the type
of structure ("failsafe" or "non failsafe") the partial
factor
M
can be determined (Table 14.2).
(3) Multiply the nominal stress ranges by the partial
factor
M
, the factors C to account for secondary
bending moments (step E) and the SCFs to obtain
the maximum geometric stress ranges in chord
and braces.
Step E:
Determine the additional multiplication factors C to
account for secondary bending moments in chord and
braces (Tables 14.4 and 14.5).
(4) Determine the fatigue life (number of cycles) from
the N line for geometric stress for the relevant
thickness.
Step F:
If the calculated fatigue life meets or exceeds the
required fatigue life, the design satisfies the
requirements. Otherwise modifications have to be
made.
Based on steps A, C, D and E, the allowable nominal
stress range for the braces can be calculated:
) SCF ( C
M
geom
brace , nom
o A
= o A
If a spectrum (variable amplitude) loading is acting,
the spectrum can be divided into stress blocks and for
each stress range the number of cycles to failure can
be determined. Using the PalmgrenMiner rule, given
in eq. (14.3), will result in the cumulative damage
which should not exceed 1,0. It may sometimes be
easier to determine first an equivalent constant
amplitude stress range with the corresponding
number of cycles.
For a particular R ratio,
max
can be determined with:
min max
max
min
and R o o = o A
o
o
=
) R 1 (
max
o = o A
Thus:
SCF ) R 1 ( C
M
geom
brace , nom max,
o A
= o
With this maximum brace stress, the cross section of
the braces can be determined and, with the
parameters selected under step A, the chord
dimensions and the joint layout selected.
181
Table 14.1 Detail categories for hollow sections and simple joints according to Eurocode 3 (EN 199319,
2005), IIW (1999) and CIDECT (Zhao et al., 2001)
Details loaded by nominal normal stresses
Detail
category
m = 3
Constructional detail Description
160
Rolled and extruded products
Nonwelded elements
Sharp edges and surface flaws to be improved by grinding
140
Continuous longitudinal welds
Automatic longitudinal welds with no stopstart positions, proven free of
detectable discontinuities
71
Transverse butt welds
Butt welded endtoend connection of circular hollow sections
Requirements:
 Height of the weld reinforcement less than 10% of weld with smooth
transitions to the plate surface
 Welds made in flat position and proven free of detectable discontinuities
 Details with wall thickness greater than 8 mm may be classified two
detail categories higher ( 90)
56
Transverse butt welds
Butt welded endtoend connection of rectangular hollow sections
Requirements:
 Height of the weld reinforcement less than 10% of weld with smooth
transitions to the plate surface
 Welds made in flat position and proven free of detectable discontinuities
 Details with wall thicknesses greater than 8 mm may be classified two
detail categories higher ( 71)
71
Welded attachments (non loadcarrying welds)
Circular or rectangular hollow section, fillet welded to another section
Section width parallel to stress direction s 100 mm
50
Welded connections (loadcarrying welds)
Circular hollow sections, endtoend butt welded with intermediate plate
Requirements:
 Welds proven free of detectable discontinuities
 Details with wall thicknesses greater than 8 mm may be classified one
detail category higher ( 56)
45
Welded connections (loadcarrying welds)
Rectangular hollow sections, endtoend butt welded with intermediate
plate
Requirements:
 Welds proven free of detectable discontinuities
 Details with wall thicknesses greater than 8 mm may be classified one
detail category higher ( 50)
40
Welded connections (loadcarrying welds)
Circular hollow sections, endtoend fillet welded with intermediate plate
Requirements:
 Wall thickness less than 8 mm
36
Welded connections (loadcarrying welds)
Rectangular hollow sections, endtoend fillet welded with intermediate
plate
Requirements:
 Wall thickness less than 8 mm
182
Table 14.2 Partial factors
M
according to IIW (1999) and CIDECT (Zhao et al., 2001)
Inspection and access
"Failsafe" structures
(redundant)
Non "failsafe" structures
(nonredundant)
Periodic inspection and maintenance
Accessible joint detail
M
= 1,00
M
= 1,25
(1)
Periodic inspection and maintenance
Poor accessibility
M
= 1,15
M
= 1,35
(1)
In Eurocode 3 (EN 199319):
M
= 1,15
Table 14.3 See next page
Table 14.4 Multiplication factors to account for secondary bending moments in CHS lattice girder joints
(EN 199319, 2005), IIW (1999) and CIDECT (Zhao et al., 2001)
Type of joint Chords Verticals Diagonals
K 1,5  1,3
Gap joints
N 1,5 1,8 1,4
K 1,5  1,2
Overlap
joints
N 1,5 1,65 1,25
Table 14.5 Multiplication factors to account for secondary bending moments in RHS lattice girder joints
(EN 199319, 2005), IIW (1999) and CIDECT (Zhao et al., 2001)
Type of joint Chords Verticals Diagonals
K 1,5  1,5
Gap joints
N 1,5 2,2 1,6
K 1,5  1,3
Overlap
joints
N 1,5 2,0 1,4
Table 14.6 Fatigue classes for various pretensioned bolted joints according to Eurocode 3 (EN 199319)
Constructional detail Detail category m
Bolts (d s 30 mm) loaded in tension based on tensile
stress area
50 3
Onesided, slipresistant joints, e.g. splices or cover
plates (based on stress in gross section)
90 3
Doublesided, slipresistant joints, e.g. splices or
cover plates (based on stress in gross section)
112 3
183
Table 14.3 Detail categories for lattice girder joints based on nominal stresses according to Eurocode 3
(EN 199319, 2005) and CIDECT (Zhao et al., 2001)
Detail categories for lattice girder joints based on nominal stresses
Detail
category
(1) (2)
m = 5
Constructional details Description
90 t
0
/t
i
> 2,0
45 t
0
/t
i
= 1,0
Circular hollow sections
K and N gap joints
71 t
0
/t
i
> 2,0
36 t
0
/t
i
= 1,0
Rectangular hollow sections
K and N gap joints
Requirements:
 0,5(b
0

b
i
) s g s 1,1(b
0

b
i
)
 g > 2t
0
71 t
0
/t
i
> 1,4
56 t
0
/t
i
= 1,0
K overlap joints
Requirements:
 30% s Ov s 100%
71 t
0
/t
i
> 1,4
50 t
0
/t
i
= 1,0
N overlap joints
Requirements:
 30% s Ov s 100%
General requirements
35 s
i
s 50
4 s t
0
s 8 mm b
0
s 200 mm 0,40 s b
i
/b
0
s 1,0 (b
0
/t
0
)(t
0
/t
i
) s 25
(3)
0,5h
0
s e s 0,25h
0
4 s t
i
s 8 mm d
0
s 300 mm 0,25 s d
i
/d
0
s 1,0 (d
0
/t
0
)(t
0
/t
i
) s 25
(3)
0,5d
0
s e s 0,25d
0
Outofplane eccentricity: s 0,02b
0
or s 0,02d
0
Fillet welds are permitted for braces with wall thicknesses s 8 mm
(1)
Note that the detail category is based on the stress range in the braces.
(2)
For intermediate t
0
/t
i
values, use linear interpolation between nearest detail categories.
(3)
This formulation is based on test data and slightly deviates from the IIW (1999) recommendations.
184
Maximum
geometrical
peak stress
Stress increase due to weld toe effects
Stress increase due to weld geometry
Geometrical stress
weld
notch
Chord wall
B
r
a
c
e
w
a
l
l
a b
Maximum
geometrical
peak stress
Stress increase due to weld toe effects
Stress increase due to weld geometry
Geometrical stress
weld
notch
Chord wall
B
r
a
c
e
w
a
l
l
a b
Fig. 14.1 Peak stress due to weld discontinuity
o
peak in chord
o
nominal
o
peak in brace
o
peak in chord
o
nominal
o
peak in brace
Fig. 14.2 Geometrical stress distribution in an axially loaded X joint of circular hollow sections
Stress
R > 0
R = 1
R = 0
Ao
Ao
Ao
Stress
R > 0
R = 1
R = 0
Ao
Ao
Ao
Fig. 14.3 Stress range A and stress ratio R
185
Number of stress cycles N
S
t
r
e
s
s
r
a
n
g
e
A
o
(
N
/
m
m
2
)
m = 3
Detail category
Constant amplitude
fatigue limit
1000
500
100
50
10
10
4
5 10
7
5 10
5
5 10
8
5 10
6
Number of stress cycles N
S
t
r
e
s
s
r
a
n
g
e
A
o
(
N
/
m
m
2
)
m = 3
Detail category
Constant amplitude
fatigue limit
1000
500
100
50
10
10
4
5 10
7
5 10
5
5 10
8
5 10
6
Fig. 14.4 A N curves for classified details and constant amplitude loading (IIW, 1999; CIDECT, 2001;
EN 199319, 2005)
Number of stress cycles N
S
t
r
e
s
s
r
a
n
g
e
A
o
(
N
/
m
m
2
)
m = 3
Detail category
Constant amplitude
fatigue limit
1000
500
100
50
10
10
4
5 10
7
5 10
5
5 10
8
5 10
6
m = 5
2
Cutoff limit
Number of stress cycles N
S
t
r
e
s
s
r
a
n
g
e
A
o
(
N
/
m
m
2
)
m = 3
Detail category
Constant amplitude
fatigue limit
1000
500
100
50
10
10
4
5 10
7
5 10
5
5 10
8
5 10
6
m = 5
2
Cutoff limit
Fig. 14.5 A N curves for classified details and variable amplitude loading (IIW, 1999; CIDECT, 2001;
EN 199319, 2005)
186
Brace
Crown
Brace
Chord
Chord
Saddle
Brace
Crown
Brace
Chord
Chord
Saddle
Fig. 14.6 Locations of extrapolation of geometric peak stresses for a T joint
30 20 30
Ao = 120 N/mm
2
Ao
average
= 160 N/mm
2
f
y
=240 N/mm
2
theoretical
stress
at Ao
max
actual
stress
at Ao
max
residual
stress
at Ao= 0
240
240
a c
b
A B
B
B
A
A 30 20 30
Ao = 120 N/mm
2
Ao
average
= 160 N/mm
2
f
y
=240 N/mm
2
theoretical
stress
at Ao
max
actual
stress
at Ao
max
residual
stress
at Ao= 0
240
240
a c
b
A B
B
B
A
A
Fig. 14.7 Plate with a hole
Number of stress cycles N
1000
500
100
50
10
10
4
5 10
7
10
5
10
8
10
6
G
e
o
m
e
t
r
i
c
a
l
s
t
r
e
s
s
r
a
n
g
e
A
o
(
N
/
m
m
2
)
10
9
t = 4
t = 8
t = 12,5
t = 16
t = 25
(t in mm)
,
, ,
,,
,
For 10
3
< N < 5 x 10
6
log(Ao
geom
) = 1/3 [(12,476  log(N)] + 0,06 log(N) log(16/t)
For 5 x 10
6
< N < 10
8
log(Ao
geom
) = 1/5 [(16,327  log(N)] + 0,402 log(16/t)
402 , 0
) mm 16 t ( geom
) t ( geom
t
16

.

\

=
o A
o A
=
Number of stress cycles N
1000
500
100
50
10
10
4
5 10
7
10
5
10
8
10
6
G
e
o
m
e
t
r
i
c
a
l
s
t
r
e
s
s
r
a
n
g
e
A
o
(
N
/
m
m
2
)
10
9
t = 4
t = 8
t = 12,5
t = 16
t = 25
(t in mm)
,
, ,
,,
,
For 10
3
< N < 5 x 10
6
log(Ao
geom
) = 1/3 [(12,476  log(N)] + 0,06 log(N) log(16/t)
For 5 x 10
6
< N < 10
8
log(Ao
geom
) = 1/5 [(16,327  log(N)] + 0,402 log(16/t)
402 , 0
) mm 16 t ( geom
) t ( geom
t
16

.

\

=
o A
o A
=
Fig. 14.8 Basic A
geom
 N design curves for the geometrical stress method for hollow section joints (IIW, 1999;
CIDECT, 2001)
187
0 0,2 0,4 0,6 0,8 1,0
0 0,2 0,4 0,6 0,8 1,0
24
20
16
12
8
4
0
40
36
32
28
24
20
16
12
8
4
0
5
4
3
2
1
0
8
7
6
5
4
3
2
1
0
0 0,2 0,4 0,6 0,8 1,0
0 0,2 0,4 0,6 0,8 1,0




S
C
F
S
C
F
S
C
F
S
C
F
Brace saddle
Brace crown
Chord saddle
Chord crown
2 = 15 2 = 30 2 = 50
t = 0,5
t = 1,0
t = 1,0
t = 1,0
t = 1,0
t = 1,0
t = 1,0
t = 1,0
t = 1,0
t = 1,0
Brace crown independent of t
t = 1,0
0 0,2 0,4 0,6 0,8 1,0
0 0,2 0,4 0,6 0,8 1,0
24
20
16
12
8
4
0
40
36
32
28
24
20
16
12
8
4
0
5
4
3
2
1
0
8
7
6
5
4
3
2
1
0
0 0,2 0,4 0,6 0,8 1,0
0 0,2 0,4 0,6 0,8 1,0




S
C
F
S
C
F
S
C
F
S
C
F
Brace saddle
Brace crown
Chord saddle
Chord crown
2 = 15 2 = 30 2 = 50
t = 0,5
t = 1,0
t = 1,0
t = 1,0
t = 1,0
t = 1,0
t = 1,0
t = 1,0
t = 1,0
t = 1,0
Brace crown independent of t
t = 1,0
Fig. 14.9 SCFs for axially loaded circular hollow section X joints (IIW, 1999; CIDECT, 2001)
188
32
28
24
20
16
12
8
4
0
S
C
F
/
t
0
,
7
5
0 0,2 0,4 0,6 0,8 1,0
32
28
24
20
16
12
8
4
0
S
C
F
/
t
0
,
7
5
Line B
0 0,2 0,4 0,6 0,8 1,0


S
C
F
/
t
0
,
7
5
Line D
20
16
12
8
4
0
0 0,2 0,4 0,6 0,8 1,0
Line A, E
20
16
12
8
4
0
S
C
F
2 = 15 2 = 30 2 = 50
t = 0,25 1,0
Symbol size ~ t
Brace
Chord
Line C
0 0,2 0,4 0,6 0,8 1,0


32
28
24
20
16
12
8
4
0
S
C
F
/
t
0
,
7
5
0 0,2 0,4 0,6 0,8 1,0
32
28
24
20
16
12
8
4
0
S
C
F
/
t
0
,
7
5
Line B
0 0,2 0,4 0,6 0,8 1,0


S
C
F
/
t
0
,
7
5
Line D
20
16
12
8
4
0
0 0,2 0,4 0,6 0,8 1,0
Line A, E
20
16
12
8
4
0
S
C
F
2 = 15 2 = 30 2 = 50
t = 0,25 1,0
Symbol size ~ t
Brace
Chord
Line C
0 0,2 0,4 0,6 0,8 1,0


Notes:
 For a T joint, the effect of chord bending due to the axial brace load should be separately included in the analysis.
 For fillet welded joints: multiply SCFs for the brace by 1,4.
 A minimum SCF = 2,0 is recommended to avoid crack initiation from the root.
Fig. 14.10 SCFs for butt welded T and X joints of square hollow sections, loaded by an axial force on the brace
(parametric formulae compared with FE calculations (Van Wingerde, 1992))
189
8
7
6
5
4
3
2
1
0
0 0,2 0,4 0,6 0,8 1,0
S
C
F
Brace saddle / crown

8
7
6
5
4
3
2
1
0
0 0,2 0,4 0,6 0,8 1,0
S
C
F
Chord saddle / crown

t = 0,5
t = 1,0
2 = 15 2 = 30 2 = 50
8
7
6
5
4
3
2
1
0
0 0,2 0,4 0,6 0,8 1,0
S
C
F
Brace saddle / crown

8
7
6
5
4
3
2
1
0
0 0,2 0,4 0,6 0,8 1,0
S
C
F
Chord saddle / crown

t = 0,5
t = 1,0
2 = 15 2 = 30 2 = 50
Fig. 14.11 Maximum SCFs for axially loaded K joints of circular hollow sections with gap g = 0,1d
0
Line D
4
3
2
1
0
0 0,2 0,4 0,6 0,8 1,0

S
C
F
/
t
0
,
2
4
Line C
4
3
2
1
0
0 0,2 0,4 0,6 0,8 1,0

S
C
F
/
t
0
,
1
9
Line D
4
3
2
1
0
0 0,2 0,4 0,6 0,8 1,0

S
C
F
/
t
0
,
2
4
Line C
4
3
2
1
0
0 0,2 0,4 0,6 0,8 1,0

S
C
F
/
t
0
,
1
9
Fig. 14.12 SCFs for T and X joints of rectangular hollow sections (chord locations C and D of Fig. 14.6 only),
loaded by an inplane bending moment or an axial force on the chord
Max. SCF 4,7 7,0 10,6
 =0,5 2 = 24 t = 0,5
Max. SCF 4,7 7,0 10,6
 =0,5 2 = 24 t = 0,5
Fig. 14.13 Effect of multiplanar loading on the SCF
190
F
F
F
F
F
F
a
1
b
1
c
1
a
2
b
2
c
2
 circular hollow section
Good Bad
F
F
F
F
F
F
a
1
b
1
c
1
a
2
b
2
c
2
 circular hollow section
Good Bad
Fig. 14.14 Examples of bolted joints (with deformed flanges) in tension
Fig. 14.15 Recommended bolted ring flange joint for fatigue loading
191
a. No preload b. Preloaded c. Preloaded with slip a. No preload b. Preloaded c. Preloaded with slip
Fig. 14.16 Possible stress distributions in bolted shear joints
Noding condition
for most
overlap joints
Extremely
stiff members
Pin
Extremely
stiff members
Noding condition
for most
gap joints
Noding condition
for most
overlap joints
Extremely
stiff members
Pin
Extremely
stiff members
Noding condition
for most
gap joints
Fig. 14.17 Plane frame joint modelling assumptions
192
15. DESIGN EXAMPLES
15.1 UNIPLANAR TRUSS OF
CIRCULAR HOLLOW SECTIONS
In this example (Wardenier et al., 2008a), the design
principles of Chapter 6 are illustrated as well as the
joint design methods.
Truss layout and member loads
A Warren type truss with low brace member angles is
chosen to limit the number of joints, see Fig. 15.1.
The trusses are spaced at 12 m intervals and the top
chord is considered to be laterally supported at each
purlin position at 6 m centretocentre. The span
todepth ratio is 15, which is approximately the optimal
limit considering service load deflections and overall
costs.
For this example, hot finished members are selected
and the member resistances are calculated according
to Eurocode 3 (EN 199311, 2005), assuming a
partial factor
M
= 1,0 (this factor may be different for
various countries).
The factored design load P from the purlins including
the weight of the truss has been calculated as P = 108
kN. A pin jointed analysis of the truss gives the
member forces shown in Fig. 15.2.
Design of members
In this example, the chords are made from steel S355
with a yield stress of 355 N/mm
2
and the braces from
steel S275 with a yield stress of 275 N/mm
2
.
For member selection, use can be made of either
member resistance tables with the applicable effective
length or the applicable strut buckling curve. The
availability of the member sizes selected has to be
checked. Since the joints at the truss ends are
generally decisive, the chords should not be too thin
walled. As a consequence, a continuous chord with
the same wall thickness over the whole truss length is
often the best choice.
Top chord
Use a continuous chord with an effective inplane and
outofplane buckling length
b
(see Chapter 2) of:
b
= 0,9 x 6000 = 5400 mm
Maximum chord force N
0
= 1148 kN (compression).
Possible section sizes are shown in Table 15.1, along
with their compressive resistances.
From a material point, the sections 244,5 x 5,6 and
219,1 x 7,1 are most efficient. However, for the
supplier considered in this example, these two
dimensions are not available from stock (only
deliverable from the mill). These dimensions can only
be used if a large quantity is required, which is
assumed in this example.
Bottom chord
For the capacities of the joints, it is best to keep the
tension chord as compact and stocky as possible.
However, to allow gap joints and to keep the
eccentricity within the limits, a larger diameter may be
needed. Possible section sizes are given in Table
15.2.
Diagonals
Try to select members (see Chapters 6 and 8) which
satisfy:
i yi
0 0 y
t f
t f
> 2,0
i.e. 0 , 2
t 275
1 , 7 355
i
>
or t
i
s 4,5 mm
For the braces loaded in compression, use an
effective buckling length
b
(see Chapter 2) of:
m 88 , 2 0 , 3 4 , 2 75 , 0 75 , 0
2 2
b
= + = =
The possible sizes for the compression diagonals are
given in Table 15.3 and for the tension diagonals in
Table 15.4.
Member selection
The number of sectional dimensions depends on the
total tonnage to be ordered. In this example, for the
braces only two different dimensions will be selected.
Comparison of the members suitable for the tension
members and those suitable for the compression
members shows that the following sections (see Fig.
15.3) are most convenient:
 Braces: 139,7 x 4,5
88,9 x 3,6
 Top chord: 219,1 x 7,1
 Bottom chord: 193,7 x 6,3
193
(These chord sizes allow gap joints; no eccentricity is
required.)
It is recognized that the d
0
/t
0
ratios of the chords
selected are high. This may give joint strength
problems in joints 2 and 5. The check for joint
resistance is given in Table 15.5.
Joint strength checks, commentary and revision
General
In Table 15.5, all joints are treated as K joints, initially
neglecting the additional X joint action in the joints 2, 3
and 4. In this case, there should be a larger margin
between the design and the acting efficiency as will be
shown in the more detailed, accurate evaluation given
below.
Joint 1
If, in joint 1, a gap g = 2t
0
is chosen between the cap
plate and the brace, nearly no eccentricity exists for
the bolted joint of the cap plate. This joint is checked
as a K(N) joint (see Table 15.5) because the load
transfer is similar as in an N joint (the reaction in the
cap plate is upwards and the diagonal loading is
downwards). In the example, is conservatively
based on the diameter of the brace.
Joint 2
Table 15.5 shows that the strength of joint 2 with g =
12,8t
0
(and eccentricity e = 0 mm) is not sufficient.
The easiest way to obtain adequate joint strength will
be to decrease the gap from 12,8t
0
to 3t
0
, resulting in
a larger C
K
= 0,39
and a slightly lower Q
f
. However,
this means that a (negative) eccentricity of e = 28 mm
is introduced resulting in an eccentricity moment of:
M
0
= (878  338) x 28 x 10
3
= 15,12 kNm.
Since the length and the stiffness El of the top chord
members between joints 12 and 23 are the same
(see Fig. 15.3), this moment can be equally distributed
over both members, i.e. both members have to be
designed additionally for M
0,Ed
= 7,56 kNm.
Including the chord bending moment effect gives the
following values for the chord stress parameter n in
the connecting face at the left and right side of the
joint.
Left side:
113,3
7,56
0,355 4728
338
M
M
f A
N
n
Rd pl,0,
Ed 0,
y0 0
Ed 0,
+
= + =
0,13 0,067 0,20 = + =
Right side:
113,3
7,56 
0,355 4728
878
M
M
f A
N
n
Rd pl,0,
Ed 0,
y0 0
Ed 0,
+
= + =
0,59 0,067 0,52 = =
The right side with n = 0,59 is decisive, resulting in Q
f
= 0,80. In combination with C
K
, this gives a joint
strength efficiency (see Table 15.5):
0,82
f A
N
0,84
f A
N
1 y 1
Ed 1,
1 y 1
Rd 1,
= > = (o.k.)
The chord members between joints 12 and 23,
which are in compression, should also be checked as
a beamcolumn for buckling. From these, chord
member 23 is most critical. This check depends on
the national code to be used. In general, the criterion
to be verified has the following format:
1,0
M
M
k
f A
N
Rd pl,0,
Ed 0,
y0 0
Ed 0,
s +
_
where:
reduction factor for column buckling (see
Table 15.1 for the values of for the possible
chord sections)
k amplification factor for second order effects
depending on slenderness, section
classification and moment diagram (in this
case use a triangular shape)
M
pl,0,Rd
plastic moment resistance (W
pl,0
f
y0
) of the
chord (class 1 or 2 sections)
113,3
7,56
k
1189
878
M
M
k
f A
N
Rd pl,0,
Ed 0,
y0 0
Ed 0,
+ = +
_
0 , 1 k 067 , 0 74 , 0 < + =
Independent of the code used, this will not be critical.
More accurate calculation based on combined K and
X joint actions
As already mentioned, this joint actually has a
combination of K joint and X joint actions and should
be substituted by a K joint and an X joint, see Chapter
6.2.1.
194
0 404
259 259
=
338 474
173
108
+
338 878
432 259
108
0 404
259 259
0 404
259 259
=
338 474
173
108
338 474
173
108
+
338 878
432 259
108
338 878
432 259
108
Joint 2 K joint action:
0,24
0,355 4728
404
f A
N
n
0 y 0
Ed , 0
=
= = compression
93 , 0 ) 0,24 (1 Q
0,25
f
= =
For the modified configuration with g = 3t
0
, = 0,52
and 2 = 30,9:
C
K
= 0,39 (see Fig. 8.19)
For brace 1:
0,97 0,82
0,625
0,93
2,04 0,39
f A
N
y1 1
Rd 1,
= =
Due to acting load:
49 , 0
5 , 525
259
f A
N
1 y 1
Ed , 1
= =
Hence, the utilization ratio for K joint action is:
50 , 0
97 , 0
49 , 0
N
N
Rd , 1
Ed , 1
= =
For brace 2:
1,0 1,29
0,625
0,93
2,55 0,39
f A
N
y2 2
Rd 2,
> =
The actual efficiency is:
1,0 0,98
f A
N
y2 2
Ed 2,
< = (o.k.)
Joint 2 X joint action:
0,28
0,355 4728
474
f A
N
n
0 y 0
Ed , 0
=
= = compression
Including the above mentioned bending moment in the
chord M
0,Ed
gives:
n = 0,28  0,067 = 0,35
Thus, for 0,64
219,1
139,7
d
d
0
1
= = :
( )
88 , 0 ) 0,35 (1 Q
0,64 0,25  0,45
f
= =
For 0,64
219,1
139,7
d
d
0
1
= = and 2 = 30,9:
C
X
= 0,20 (see Fig. 8.18)
0,57
0,625
0,88
2,04 0,20
f A
N
y1 1
Rd 1,
= =
Due to acting load:
33 , 0
5 , 525
173
f A
N
1 y 1
Ed , 1
= =
Hence, the utilization ratio for X joint action is:
58 , 0
57 , 0
33 , 0
N
N
Rd , 1
Ed , 1
= =
The combined acting efficiencies for brace 1 due to K
joint and X joint action are:
0,50 + 0,58 = 1,08 > 1,0 (not o.k.)
Thus, the joint is still not o.k.
Note: Based on the check as a K joint only (see Table
15.5 for the evaluation of joint 2 with g/t
0
= 3,0), this
utilization ratio was 98 , 0
84 , 0
82 , 0
N
N
Rd , 1
Ed , 1
= = which would
have been about 10% overoptimistic.
Further decreasing the gap will not help because the
design efficiency as a K joint is already close to 1,0.
Hence, the effect of the X joint action should be
decreased. This can be done as follows:
(1) By using a section for brace 1 with about the
same cross sectional area but a lower thickness
(e.g. 168,3 x 3,6), which increases the design
efficiency. However, this increases the number of
section types for the braces to three.
(2) By increasing the thickness of the top chord and
choosing 219,1 x 8,0.
An additional type of section can increase the costs,
but increasing the chord thickness also increases
195
material costs. The choice will be made after checking
the other joints.
Joint 3
oint 3 K joint action:
J
0,08
0,355 4728
134
f A
N
n
0 y 0
Ed , 0
=
= = compression
98 , 0 ) 0,08 (1 Q
0,25
f
= =
or g = 12,8t
0
with = 0,52 and 2 = 30,9:
K
= 0,34 (see Fig. 8.19)
F
C
0,89 0,82
0,625
0,98
2,04 0,34
f A
N
y1 1
Rd 1,
= =
ue to acting load: D
16 , 0
5 , 525
86
f A
N
1 y 1
Ed , 1
= =
ence, the utilization ratio for K joint action is: H
18 , 0
89 , 0
16 , 0
N
N
Rd , 1
Ed , 1
= =
or brace 2: F
1,0 1,29
0,625
0,98
2,55 0,34
f A
N
y2 2
Rd 2,
> =
he actual efficiency is: T
1,0 0,32
f A
N
y2 2
Ed 2,
< = (o.k.)
oint 3 X joint action: J
0,60
0,355 4728
1014
f A
N
n
Thus, for 0,64
219,1
139,7
d
d
0
1
= = :
( )
77 , 0 ) 0,60 1 ( Q
0,64 0,25  0,45
f
= =
0 134
86 86
=
878 1014
173
108
+
878 1148

108
0 y 0
Ed , 0
=
= = compression
For 0,64
219,1
139,7
d
d
0
1
= = and 2 = 30,9:
C
X
= 0,20 (see Fig. 8.18)
0,50
0,625
0,77
2,04 0,20
f A
N
y1 1
Rd 1,
= =
Due to acting load:
33 , 0
5 , 525
173
f A
N
1 y 1
Ed , 1
= =
Hence, the utilization ratio for X joint action is:
66 , 0
50 , 0
33 , 0
N
N
Rd , 1
Ed , 1
= =
The combined acting efficiencies due to K joint and X
joint actions are:
0,18 + 0,66 = 0,84 < 1,0 (o.k.)
Thus, the criteria are satisfied.
Note: Based on the check as a K joint only (see Table
15.5 for the evaluation of joint 3), this utilization ratio
was 72 , 0
68 , 0
49 , 0
N
N
Rd , 1
Ed , 1
= = which would have been about
14% overoptimistic.
Joint 4
At joint 4, a site joint will be made consisting of two
plates which also transfer the purlin load to the chord.
This means that joint 4 behaves as two N joints.
Assuming no eccentricity at the bolted joint and cap
plates of 15 mm, the gap between the toe of the brace
and the cap plate will be (see eq. (6.1)):
259 86
108
0 134
86 86
0 134
86 86
=
878 1014
173
878 1014
173
108
+
108 108
878 1148 878 1148
259 86 259 86
86
1148 1148
86
=
1148 1148
86 86
108
54 54
86
1148 1148
86
=
1148 1148
86 86
108
54 54
1148 1148
86
=
1148 1148
86 86
108
54 54
196
15
38,7 sin
88,9
38,7 sin
38,7) sin(2
2
219,1
0,5 g
2


.

\

=
0
7,1t mm 50,6 = =
The check in Table 15.5 shows that the joint is o.k.
Evaluation
The joint checks showed that joint 2 is not o.k.
Considering the options mentioned, in this example,
the top chord section will be changed from 219,1 x
7,1 to 219,1 x 8,0. Recalculating joint 2 for e = 0
(with g = 12,8t
0
) gives a utilization ratio of 0,49 for K
joint action and 0,45 for X joint action, thus a
combined utilization of 0,94 < 1,0.
Compared to the selected members in Fig. 15.3, only
the top chord is changed to 219,1 x 8,0 and all
joints can be made without any eccentricity.
The above extensively workedout example shows
that checking as a K joint only is much faster than
using the combined K and X joint actions.
A fast alternative would be to check as a K joint with C
and Q
f
factors selected between the values for K and
X joints depending on the contributions.
Purlin joints
Depending on the type of purlins, various purlin joints
are possible. If corrosion will not occur, a cutout of a
channel section welded on top of the chord at the
purlin support location and provided with bolt stubs
gives an easy support, see Fig. 15.6. Table 8.3
provides evidence for the design of platetotube joints.
The joints in Table 8.3 are not exactly similar to those
between open U sections and a CHS chord but the
capacity may be based on the design resistance for
an RHStoCHS joint. Since no cross plates are
present, only the sides are effective; therefore a very
conservative reduction factor to be applied is
h
1
/(h
1
+b
1
).
For the purlin joint at the centre, another alternative
has to be used to allow a site bolted truss joint. If the
top chord parts are provided with cap plates, a Tstub
for purlin support can be fitted in between the cap
plates.
Site bolted flange joints
This book does not give complete design procedures
for bolted flange joints. However, in Wardenier et al.
(2008a) this example is worked out further, e.g.
resulting in a joint according to Fig. 13.1 with 13 bolts
22 10,9 with an end plate thickness of 20 mm (f
y
=
355 N/mm
2
) for the bottom tensile chord joint. To
avoid displacements in the joint it is recommended to
pretension the bolts. For fatigue loaded joints, the
bolts have to be pretensioned.
For the top chord joint, the compression loading is
transferred through contact pressure. The number of
bolts required depends on the erection loads which
can be tensile, and the national code requirements
with regard to the minimum joint strength related to
the member tensile strength.
15.2 UNIPLANAR TRUSS OF SQUARE
HOLLOW SECTIONS
In Packer et al. (2009a), a truss with the same
configuration and loading has been designed with
square hollow sections, all with a yield stress of 355
N/mm
2
. In principle, the approach is similar, resulting
in the member dimensions shown in Fig. 15.7.
15.3 MULTIPLANAR TRUSS
(TRIANGULAR GIRDER)
For an easy comparison, for this example a
multiplanar truss (Fig. 15.8) is chosen with the side
elevation dimensions equivalent to the uniplanar truss
discussed in Section 15.1.
Member loads
The member loads can be determined in a similar way
as for the uniplanar truss, assuming pin ended
members.
The load in the bottom chord follows by dividing the
relevant moment by the girder depth. Since two top
chords are used, the load at the top has to be divided
by 2. The loads in the braces follow from the shear
forces V in the girder (Fig. 15.9).
The top chords should be connected in the top plane
for equilibrium of loading, see Fig. 15.10. This can be
achieved by a bracing system which connects the
loading points. Connection of the loading points only
in the horizontal plane results in a triangular truss
which has no torsional rigidity. A combination with
diagonals in the horizontal plane, thereby completing
a Pratt type truss in the horizontal plane, gives
torsional resistance. It is also possible to use the
purlins or the roof structure as the connecting parts
197
between the loading points. Once the loads in one
plane are known, the design can be treated in a
similar way as for uniplanar trusses.
Joints
The joints can also be treated in a similar way as for
uniplanar joints, however, taking account of the larger
chord loads. This means a larger Q
f
reduction factor
for the joints with the bottom chord. From a fabrication
point of view, it is better to avoid overlaps of the
intersecting braces from both planes. Sometimes this
may result in an eccentricity in the two planes, also
called an offset (see Fig. 15.11).
The offset has to be incorporated in member design
and joint capacity verification. For the chords, the
moments due to this offset have to be distributed over
the chord members, affecting the chord stress
function Q
f
and hence, the joint capacity.
Design calculation
Assume P = 187 kN (at ultimate limit state).
This means that the loads acting in the side planes of
the triangular truss (see Fig. 15.12) are:
kN 108
30 cos 2
P
o
=
This is equal to the purlin loads used in the design
example for the uniplanar truss in Section 15.1. As a
consequence, the top chord and the diagonals can be
the same to those for the uniplanar truss, provided the
same steel grades are used (see Fig. 15.13).
For the bottom chord, the required cross section
should be twice that required for the uniplanar truss,
i.e. 219,1 x 11,0 with A
0
= 7191 mm
2
and W
pl,0
=
476,8 x 10
3
mm
3
. (This section may have a longer
delivery time.)
A detailed check of the members is already given in
Section 15.1 and is the same here. However, the
eccentricity moment should be taken into account
both for member design and joint strength verification
(i.e. the effect of the chord moment on the chord
stress factor Q
f
).
The braces between the top chords are determined by
the horizontal loads of 54 kN at each purlin support or
by loads resulting from unequally distributed loading
on the roof. Since transport is simpler for Vtrusses
than for triangular trusses, it is also possible to use
the purlins as connection between the top chords. A
simple bolted connection, as given in Fig. 15.6, can
easily be designed to transfer the shear load of 54 kN.
However, in this way the truss has no torsional rigidity
and cannot act as horizontal wind bracing for the roof.
If this is required, braces between the top chords
should be used.
Joint strength check
As mentioned, the initial difference with the joint
strength checks for the uniplanar truss in Section 15.1
is that the effect of noding eccentricity has to be
incorporated. A joint without any eccentricity would
result in an overlap of the braces in the two planes,
see Fig 15.14a. To allow welding, an outofplane gap
of 22,5 mm is chosen which results in an eccentricity
of 50 mm (inplane eccentricity = 43 mm s 0,25d
0
). As
a consequence, the inplane gap increases, resulting
in slightly lower C
K
values.
Besides the joint capacity checks carried out in
Section 15.1, the multiplanar joint has to be checked
for chord shear, see Table 8.4. The joint with
maximum shear in the gap is joint 5 with:
V
gap,0,Ed
= 2,5 P = 2,5 x 187 = 467,5 kN
Further (see Fig. 15.2):
N
gap,0,Ed
= 0,5 x (2 x 675) = 675 kN
M
gap,0,Ed
= 675 x 0,05 = 33,75 kNm
kN 2552 355 , 0 7191 f A N
y0 0 Rd pl,0,
= = =
0,26
2552
675
N
N
Rd pl,0,
Ed gap,0,
= =
t
=
t
=
7191 2
355 , 0 58 , 0
A 2
f 58 , 0 V
0
0 y Rd pl,0,
kN 943 =
0,50
943
467,5
V
V
Rd pl,0,
Ed gap,0,
= =
3 3
y0 pl,0 Rd pl,0,
10 355 , 0 10 8 , 476 f W M
= =
kNm 3 , 169 =
0,20
169,3
33,75
M
M
Rd pl,0,
Ed gap,0,
= =
198
A conservative, linear interaction gives:
= + +
Rd pl,0,
Ed gap,0,
Rd pl,0,
Ed gap,0,
Rd pl,0,
Ed gap,0,
M
M
V
V
N
N
1,0 0,96 0,20 0,50 0,26 s = + +
The exact interaction is more complicated (Wardenier,
1982). In general, this chord shear check becomes
critical for larger ratios.
15.4 MULTIPLANAR TRUSS OF
SQUARE HOLLOW SECTIONS
The approach for a multiplanar truss of square hollow
sections is similar to that in Section 15.3. Generally,
the braces in the two side planes are connected at
different faces of the bottom chord, giving no
problems with outofplane overlaps as would be
possible for circular hollow sections.
Working out the example used in Section 15.3 for
square hollow sections (all members with f
y
= 355
N/mm
2
), results in the same dimensions for the top
chords and braces as those given in Fig. 15.7. For the
bottom chord, a section with twice the cross sectional
area has to be selected.
15.5 JOINT CHECK USING THE JOINT
RESISTANCE FORMULAE
The joints in the previous examples have been
checked using the efficiency parameters C
e
from the
design graphs. However, the joints can also be
checked, using the formulae given in Chapters 8 and
9. Here, as an example, only joint 5 of the uniplanar
(RHS) truss shown in Fig. 15.7 will be checked using
the resistance formulae.
The dimensions of the sections and the yield stresses
are presented in Fig. 15.7. All other information
remains similar to that given in Figs. 15.1 and 15.2.
Since all members are square hollow sections, Table
9.2 applies.
Check of validity range
According to Table 9.1, the gap g has to satisfy:
) 1,5(1 g/b ) 0,5(1
0
 s s 
Thus, for b
0
= 150 mm and 8 , 0
150 2
120 2
b 2
b b
0
2 1
=
=
+
=  :
0,8) 1,5(1 g/150 0,8) 0,5(1 s s
or:
15 s g s 45
The eccentricity (e) corresponding to the minimum
gap of 15 mm, giving the minimum value for e, can be
calculated with:
( ) 2
h
sin
sin sin
g
sin 2
h
sin 2
h
e
0
2 1
2 1
2
2
1
1
+


.

\

+ + =
( )
mm 8
2
150
38,7 38,7 sin
38,7 sin 38,7 sin
15
38,7 sin
120
+ =
+

.

\

+ =
8 , 0
150
120
b
b
0
i
= = 0,34
6,3
150
0,01 0,1 = + > (o.k.)
s = = 24
5
120
t
b
1
1
class 2 limit (= 33,9) and s 40 (o.k.)
8 , 23
3 , 6
150
t
b
0
0
= = 15 s 23,8 s 40 (o.k.)
0 , 1
120 2
120 2
b 2
b b
i
2 1
=
=
+
0,6 s 1,0 s 1,3 (o.k.)
i
= 38,7 > 30 (o.k.)
Check for chord plastification
f
i
2
0 y0 3 , 0
Rd i,
Q
sin
t f
4 1 N  =
For tension:
0,10
f
) n (1 Q =
8 , 0
150 2
120 2
b 2
b b
0
2 1
=
=
+
= 
9 , 11
3 , 6 2
150
=
=
199
355 , 0 192000
8 675
355 , 0 3480
675
M
M
N
N
n
Rd , 0 , pl
Ed , 0
Rd , 0 , pl
Ed , 0
= + =
46 , 0 08 , 0 54 , 0 = =
94 , 0 ) 0,46 (1 Q
0,10
f
= =
94 , 0
7 , 38 sin
3 , 6 355 , 0
) 9 , 11 ( 8 , 0 14 N
2
3 , 0
Rd i,
=
= 498,7 kN > N
i,Ed
= 432 kN (o.k.)
The approach using design charts, given in an
example in CIDECT Design Guide No. 3 (Packer et al.,
2009a), results in approximately the same efficiency.
If rectangular hollow sections had been used, the joint
resistance check would have been considerably more
complicated, which is evident if Table 9.1 is compared
to Table 9.2. The extra checks involve:
 Chord shear
 Local brace failure
 Chord punching shear
15.6 CONCRETE FILLED COLUMN
WITH REINFORCEMENT
Here, the axial compressive capacity is calculated for
a concrete filled circular hollow section with a cross
section and reinforcement as shown in Fig. 15.15 and
with the factors according to Table 4.1.
Concrete C20 with
c
= 1,5
CHS S275 with
a
= 1,0
Reinforcement S500 with
s
= 1,15
Assumptions for the analysis
b
= 3,6 m
N
Ed
= 6000 kN
N
G,Ed
= 0,5N
Ed
t
= 3,0
Strength
f
yd
= 275/1,0 = 275 N/mm
2
f
sd
= 500/1,15 = 435 N/mm
2
f
cd
= 20/1,5 = 13,3 N/mm
2
Cross sectional areas
A
a
( 406,4 x 8,0) = ( )
2
mm 10013 0 , 8 0 , 8 4 , 406 = t
A
s
(16 25) =
2 2
mm 7854 25
4
16 =
t
A
c
=
2 2
mm 111850 7854 10013 4 , 406
4
=
t
Reinforcement ratio
10013 4 , 406
4
7854
2
s
t
= = 6,6% > 6%
The ratio of reinforcement
s
has to be limited to 6%
for the calculation (see Section 4.3.1). This may be
achieved by:
 Using 14 25
 Considering only reinforcing bars which lie in the
most favourable position of the section for bending
so that
s
s 6% (neglect two centre bars). In the
current design example, this option is selected.
 Reducing the diameters of the reinforcing bars to
such an extent that
s
= 6%
2
s
mm 7140 7854
6 , 6
6
A = s
2 2
c
mm 112564 7140 10013 4 , 406
4
A =
t
=
Check of concrete filled column
kN 7357
3 , 13 112564 435 7140 275 10013
f A f A f A N
sd s cd c yd a Rd , pl
=
+ + =
+ + =
0,2 < 37 , 0
7357
275 , 0 10013
N
f A
Rd , pl
yd a
=
=
kN 8575 N 10 8575
20 112564 500 7140 275 10013
f A f A f A N
3
sk s ck c yk a Rk , pl
= =
+ + =
+ + =
2
b
eff
eff , cr
) EI (
N
2
t
=
(EI)
eff
= E
a
I
a
+ 0,6 E
c,eff
I
c
+ E
s
I
s
200
2
t
Ed
Ed , G
cm
eff , c
N/mm 12000
3 5 , 0 1
30000
N
N
1
E
E =
+
=
+
=
2 9 4 5
a a
Nmm 10 41727 10 19870 10 1 , 2 I E = =
( )
2 9
4
c eff , c
Nmm 10 8210
8 2 4 , 406
64
12000 6 , 0 I E 6 , 0
=
t
=
Neglecting the two centre bars gives:
2 9 2 2
2 2 5
s s
Nmm 10 19812 ) 0 , 155 2 2 , 143 4
6 , 109 4 3 , 59 4 (
16
7854
10 1 , 2 I E
= + +
+ =
(EI)
eff
= 41727 + 8210 + 19812 = 69749 kNm
2
kN 53117
6 , 3
69749
N
eff , cr
=
t
=
2
2
40 , 0
53117
8575
N
N
eff , cr
Rk , pl
= = =
The reinforcement ratio
s
> 3%, thus use curve "b"
(see Fig. 2.3):
= 0,93
kN 6842 7357 93 , 0 kN 6000 N
Rd , pl Ed
= = N _ s = (o.k.)
Check for local buckling
9 , 76
275
235
90 90 8 , 50
0 , 8
4 , 406
t
d
2
= = c < = = (o.k.)
Note: The increase in compression capacity caused
by confinement effects is here neglected, but would
give an increase in capacity due to the low
ratio.
201
Table 15.1 Possible section sizes for top (compression) chord
f
y0
(N/mm
2
)
N
0
(kN)
b
(mm)
Possible sections
(mm)
A
0
(mm
2
)
d
0
/t
0
(1)
(1)
f
y0
A
0
(kN)
355 1148 5400
193,7 x 10,0
219,1 x 7,1
219,1 x 8,0
244,5 x 5,6
244,5 x 6,3
5771
4728
5305
4202
4714
19,4
30,9
27,4
43,7
38,8
1,09
0,94
0,95
0,84
0,84
0,61
0,71
0,71
0,78
0,78
1245
1189
1329
1159
1298
(1)
Buckling curve "a" of Eurocode 3 (EN 199311, 2005).
Table 15.2 Possible section sizes for bottom (tension) chord
f
y0
(N/mm
2
)
N
0
(kN)
Possible sections
(mm)
A
0
(mm
2
)
d
0
/t
0
f
y0
A
0
(kN)
355 1215
168,3 x 7,1
177,8 x 7,1
193,7 x 6,3
3595
3807
3709
23,7
25,0
30,7
1276
1351
1317
Table 15.3 Possible section sizes for compression diagonals
f
yi
(N/mm
2
)
N
i
(kN)
b
(m)
Possible sections
(mm)
A
i
(mm
2
)
(1)
(1)
f
yi
A
i
(kN)
432 2,881
168,3 x 3,6
139,7 x 4,5
1862
1911
0,57
0,69
0,90
0,85
462
448
259 2,881 114,6 x 3,6 1252 0,85 0,77 266
275
86 2,881 88,9 x 2,0
(2)
546 1,08 0,61 92
(1)
Buckling curve "a" of Eurocode 3 (EN 199311, 2005).
(2)
Wall thickness is rather small for welding, outside the validity range.
Table 15.4 Possible section sizes for tension diagonals
f
yi
(N/mm
2
)
N
i
(kN)
Possible sections
(mm)
A
i
(mm
2
)
f
yi
A
i
(kN)
432 133,3 x 4,0 1621 445
259 88,9 x 3,6 964 265 275
86 48,3 x 2,3 332 91
202
Table 15.5 Joint strength check, assuming K joint action only
Member sizes Joint parameters Chord load
Joint Chord (mm) Braces (mm) d
0
/t
0
g/t
0
n
1 219,1 x 7,1
Plate
139,7 x 4,5
0,64 30,9 2,0 0,20
2 219,1 x 7,1
139,7 x 4,5
88,9 x 3,6
0,52 30,9 12,8 0,52
2a
Additional analysis of joint 2 with
g/t
0
= 3,0 and e = 28 mm
0,52 30,9 3,0 0,59
3 219,1 x 7,1
139,7 x 4,5
88,9 x 3,6
0,52 30,9 12,8 0,68
4 219,1 x 7,1
88,9 x 3,6
88,9 x 3,6
0,41 30,9 7,1 0,68
5 193,7 x 6,3
139,7 x 4,5
139,7 x 4,5
0,72 30,7 2,9 0,51
6 193,7 x 6,3
88,9 x 3,6
139,7 x 4,5
0,59 30,7 9,4 0,82
7 193,7 x 6,3
88,9 x 3,6
88,9 x 3,6
0,46 30,7 15,8 0,92
Actual efficiency Joint strength efficiency Check
Joint
yi i
Ed , i
f A
N
C
K
i yi
0 0 y
t f
t f
Q
f
i
sin
1
u
i
2 1
d 2
d d +
yi i
Rd , i
f A
N
Ed , i Rd , i
N N >
1

0,82
0,41

2,04
0,95
1,60
> 1,0
> 1,0
o.k.
2
0,82
0,98
0,34
2,04
2,55
0,83 1,60
0,82
1,29
0,76
> 1,0
not o.k.
o.k.
2a
0,82
0,98
0,39
2,04
2,55
0,80 1,60
0,82
1,29
0,84
> 1,0
o.k.
o.k.
3
0,49
0,32
0,34
2,04
2,55
0,75 1,60
0,82
1,29
0,68
> 1,0
o.k.
o.k.
4
0,32
0,32
0,35
2,55
2,55
0,75 1,60
1,0
1,0
> 1,0
> 1,0
o.k.
o.k.
5
0,82
0,82
0,41
1,81
1,81
0,87 1,60
1,0
1,0
> 1,0
> 1,0
o.k.
o.k.
6
0,98
0,49
0,37
2,26
1,81
0,71 1,60
1,29
0,82
> 1,0
0,62
o.k.
o.k.
7
0,32
0,32
0,32
2,26
2,26
0,60 1,60
1,0
1,0
0,69
0,69
o.k.
o.k.
Note: Joints 14 discussed in detail in text.
203
L = 6 x 6000 = 36000 mm
tan = 2,4/ 3 = 0,8 = 38,7
Bolted joint
L = 6 x 6000 = 36000 mm
tan = 2,4/ 3 = 0,8 = 38,7
Bolted joint
Fig. 15.1 Truss layout
  
  
  
  
Fig. 15.2 Truss member axial loads
219,1 x 7,1 88,9 x 3,6
139,7 x 4,5
193,7 x 6,3
219,1 x 7,1 88,9 x 3,6
139,7 x 4,5
193,7 x 6,3
Fig. 15.3 Initially selected member dimensions and joint numbers for CHS truss.
(In the final design, the top chord is changed to 219,1 x 8,0)
2t
0
t
0
2t
0
t
0
Fig. 15.4 Joint 1
204
3t
0
t
0
338 kN
139,7 x 4,5
219,7 x 7,1
878 kN
88,9 x 3,6
3t
0
t
0
338 kN
139,7 x 4,5
219,7 x 7,1
878 kN
88,9 x 3,6
M
0,Ed
= 7,56 kNm
M
0,Ed
1 2 3
M
0,Ed
= 7,56 kNm
M
0,Ed
1 2 3
Fig. 15.5 Joint 2 (with M
0,Ed
= 7,56 kNm in chord on both sides of the joint)
Fig. 15.6 Purlin joint
180 x 180 x 8,0 80 x 80 x 3,0
120 x 120 x 5,0
150 x 150 x 6,3
Site bolted joint
180 x 180 x 8,0 80 x 80 x 3,0
120 x 120 x 5,0
150 x 150 x 6,3
Site bolted joint
Fig. 15.7 Member dimensions and joint numbers for RHS truss (f
y0
= f
yi
= 355 N/mm
2
)
Fig. 15.8 Triangular truss
205
2

i
i
i
sin
2
cos 2
V
N
u 
.

\
 
=
2

i
i
i
sin
2
cos 2
V
N
u 
.

\
 
=
2
P
2
P
2
P
2
P
Fig. 15.9 Shear force Fig. 15.10 Horizontal loads
Offset s 0,25d
0
> t
i
Offset s 0,25d
0
> t
i
Fig. 15.11 Gap and offset
2
4
0
0 93,5 kN
54 kN
108 kN
2
4
0
0 93,5 kN
54 kN
108 kN
Fig. 15.12 Cross section of the triangular truss with circular hollow sections
206
219,1 x 8,0
219,1 x 11,0 139,7 x 4,5
88,9 x 3,6
Chord: f
y0
= 355 N/mm
2
Diagonals: f
yi
= 355 N/mm
2
219,1 x 8,0
219,1 x 11,0 139,7 x 4,5
88,9 x 3,6
Chord: f
y0
= 355 N/mm
2
Diagonals: f
yi
= 355 N/mm
2
Fig. 15.13 Member dimensions and steel grades
Diagonals: 139,7 x 4,5
Chord: 219,1 x 11
22,5
Diagonals: 139,7 x 4,5
Chord: 219,1 x 11
22,5
Diagonals: 139,7 x 4,5
Chord: 219,1 x 11
22,5
Fig. 15.14 Connection of the diagonals to the bottom chord
207
5
9
,
3
1
5
5
,
0
1
0
9
,
6
8,8
406,4
16 25, S500
S275
C20
1
4
3
,
2
5
9
,
3
1
5
5
,
0
1
0
9
,
6
8,8
406,4
16 25, S500
S275
C20
1
4
3
,
2
Fig. 15.15 Concrete filled column
208
16. REFERENCES
AISC, 2005: Specification for structural steel buildings. ANSI/AISC 36005, American Institute of Steel
Construction, Chicago, USA.
AISC, 2010: Specification for structural steel buildings. ANSI/AISC 36010, American Institute of Steel
Construction, Chicago, USA.
Akiyama, N., Yayima, M., Akiyama, H., & Otake, F., 1974: Experimental study on strength of joints in steel
tubular structures. Journal of Society of Steel Construction, JSSC, Vol. 10, No. 102, pp. 3768, (in Japanese).
API, 2007: Recommended practice for planning, designing and constructing fixed offshore platforms Working
stress design. API RP 2A, 21
st
Edition, Suppl 3, American Petroleum Institute, Dallas, USA.
Aribert, J.M., Ammari, F., & Lachal, A., 1988: Influence du mode d'application d'une charge de compression
locale sur la rsistance plastique de l'me d'un profil cas des assemblages tubulaires. Construction Mtallique
No. 2, pp. 330.
Baar, S., 1968: Etude thorique et exprimentale du dversement des poutres membrures tubulaires.
Collection des publications de la Facult des Sciences Appliques de Universit de Lige, No. 10, Lige,
Belgium.
Bergmann, R., Matsui, C., Meinsma, C., & Dutta, D., 1995: Design guide for concrete filled hollow section
columns under static and seismic loading. CIDECT Series "Construction with hollow steel sections" No. 5,
Verlag TV Rheinland, Kln, Germany, ISBN 3824902982.
Birkemoe, P.C., & Packer, J.A., 1986: Ultimate strength design of bolted tubular tension connections.
Proceedings Conference on Steel Structures Recent Research Advances and their Applications to Design,
Budva, Yugoslavia, pp. 153168.
Bjrk, T., Marquis, G., Kemppainen, R., & Ilvonen, R., 2003: The capacity of coldformed rectangular hollow
section K gap joints. Proceedings 10
th
International Symposium on Tubular Structures, Madrid, Spain, Tubular
Structures X, Swets & Zeitlinger, Lisse, The Netherlands, pp. 227234.
Bjrk, T., 2005: Ductility and ultimate strength of coldformed rectangular hollow section joints at subzero
temperatures. Ph.D. Thesis, Lappeenranta University of Technology, Lappeenranta, Finland.
Bolt, H.M., & Billington, C.J., 2000: Results from ultimate load tests on 3D jackettype structures. Proceedings
Offshore Technology Conference, OTC 11941, Houston, USA.
Bortolotti, E., Jaspart, J.P., Pietrapertosa, C., Nicaud, G., Petitjean, P.D., Grimmault, J.P., & Michard, L., 2003:
Testing and modelling of welded joints between elliptical hollow sections. Proceedings 10
th
International
Symposium on Tubular Structures, Madrid, Spain, Tubular Structures X, Swets & Zeitlinger, Lisse, The
Netherlands, pp. 259264.
Bouwkamp, J.G., 1964: Concept of tubularjoint design. Journal of the Structural Division, American Society of
Civil Engineers, Vol. 90, No. ST2, pp. 77101.
Bouwman, L.P., 1982: Bolted connections dynamically loaded in tension. Journal of the Structural Division,
American Society of Civil Engineers, Vol. 108, No. 9, pp. 21172129.
British Steel, 1996: Flowdrill jointing system. CIDECT Project 6F Final Reports 6F13A/96 and 6F13B/96,
British Steel Tubes & Pipes, Corby, UK.
209
Brodka, J., 1968: Stahlrohrkonstruktionen. Verlagsgesellschaft Rdolf Mller, KlnBraunsfeld, Germany, ISBN
9783481109912.
Cao, J., & Packer, J.A., 1997: Design of tension circular flange joints in tubular structures. Engineering Journal,
American Institute of Steel Construction, Vol. 34, First Quarter, pp. 1725.
CEN/TC 250/SC 3N 1729, 2010: Choice of steel material to avoid brittle fracture for hollow section structures.
Report prepared for amending EN 1993110. Rev. 4, CEN/TC 250 CIDECT  Liaison, Aachen, Germany.
Chan, T.M., & Gardner, L., 2008: Bending strength of hotrolled elliptical hollow sections. Journal of
Constructional Steel Research, Vol. 64, Issue 9, pp. 971986.
Chen, Y., Liu, D.K., & Wardenier, J., 2005: Design recommendations for RHSK joints with 100% overlap.
Proceedings 15
th
International Offshore and Polar Engineering Conference, Seoul, Korea, Vol. IV, pp. 300307.
Choo, Y.S., Liang, J.X., & Lim, L.V., 2003: Static strength of elliptical hollow section Xjoint under brace
compression. Proceedings 10
th
International Symposium on Tubular Structures, Madrid, Spain, Tubular
Structures X, Swets & Zeitlinger, Lisse, The Netherlands, pp. 253258.
Choo, Y.S., Liang, J.X., & Vegte, G.J. van der, 2004: An effective external reinforcement scheme for circular
hollow section joints. Proceedings ECCSAISC Workshop "Connections in Steel Structures V", Bouwen met
Staal, Zoetermeer, The Netherlands, pp. 423432.
Choo, Y.S., Qian, X.D., & Foo, K.S., 2005a: Nonlinear analysis of tubular space frame incorporating joint
stiffness and strength. Proceedings 10
th
International Conference on Jackup Platform Design, Construction
and Operation, City University, London, UK.
Choo, Y.S., Vegte, G.J. van der, Zettlemoyer, N., Li, B.H., & Liew, J.Y.R., 2005b: Static strength of T joints
reinforced with doubler or collar plates  I: Experimental investigations. Journal of Structural Engineering,
American Society of Civil Engineers, USA, Vol. 131, No. 1, pp. 119128.
CIDECT, 1984: Construction with hollow steel sections. British Steel Plc., Corby, Northants, UK, ISBN 0
951006207.
CSA, 2009: Design of steel structures. CSAS1609, Canadian Standards Association, Toronto, Canada.
Davies, G., & Crockett, P., 1994: Effect of the hidden weld on RHS partial overlap K joint capacity. Proceedings
6
th
International Symposium on Tubular Structures, Melbourne, Australia, Tubular Structures VI, Balkema,
Rotterdam, The Netherlands, pp. 573579.
Deutscher Dampfkesselausschu, 1975: Glatte Vierkantrohre und Teilkammern unter innerem berdruck.
Technische Regeln fr Dampfkessel (TRD 320), Vereinigung der Technischen berwachungsvereine e.V.,
Essen, Germany.
Dexter, E.M., & Lee, M.M.K., 1998: Effect of overlap on the behaviour of axially loaded CHS Kjoints.
Proceedings 8
th
International Symposium on Tubular Structures, Singapore, Tubular Structures VIII, Balkema,
Rotterdam, The Netherlands, pp. 249258.
Dier, A.F., & Lalani, M., 1998: New code formulations for tubular joint static strength. Proceedings 8
th
International Symposium on Tubular Structures, Singapore, Tubular Structures VIII, Balkema, Rotterdam, The
Netherlands, pp. 107116.
Dier, A.F., 2005: Tubular joint technology for offshore structures. International Journal of Steel Structures, Vol.
5, No. 5, pp. 495502.
210
Dutta, D., Wardenier, J., Yeomans, N., Sakae, K., Bucak, ., & Packer, J.A., 1998: Design guide for fabrication,
assembly and erection of hollow section structures. CIDECT Series "Construction with hollow steel sections"
No. 7, TVVerlag, Kln, Germany, ISBN 3824904438.
Dutta, D., 2002: Structures with hollow sections. WileyVCH Verlag GmbH, Berlin, Germany, ISBN 9783433
014585, (also available in German).
Eekhout, M., 1996: Tubular structures in architecture. 1
st
Edition, Delft University of Technology, Delft, The
Netherlands, CIDECT.
Eekhout, M., 2010: Tubular structures in architecture. 2
nd
Edition, Delft University of Technology, Delft, The
Netherlands, CIDECT.
Efthymiou, M., 1988: Development of SCF formulae and generalised influence functions for use in fatigue
analysis. Proceedings Offshore Tubular Joints Conference, Surrey, UK.
EN 1990, 2002: Eurocode Basis of structural design. European Committee for Standardization, Brussels,
Belgium.
EN 199211, 2004: Eurocode 2: Design of concrete structures Part 11: General rules and rules for buildings.
European Committee for Standardization, Brussels, Belgium.
EN 199311, 2005: Eurocode 3: Design of steel structures Part 11: General rules and rules for buildings.
European Committee for Standardization, Brussels, Belgium.
EN 199312, 2005: Eurocode 3: Design of steel structures Part 12: General rules  Structural fire design.
European Committee for Standardization, Brussels, Belgium.
EN 199318, 2005: Eurocode 3: Design of steel structures Part 18: Design of joints. European Committee
for Standardization, Brussels, Belgium.
EN 199319, 2005: Eurocode 3: Design of steel structures Part 19: Fatigue. European Committee for
Standardization, Brussels, Belgium.
EN 1993110, 2005: Eurocode 3: Design of steel structures Part 110: Material toughness and through
thickness properties. European Committee for Standardization, Brussels, Belgium.
EN 1993112, 2007: Eurocode 3: Design of steel structures Part 112: Additional rules for the extension of
EN 1993 up to steel grades S700. European Committee for Standardization, Brussels, Belgium.
EN 199411, 2004: Eurocode 4: Design of composite steel and concrete structures Part 11: General rules
and rules for buildings. European Committee for Standardization, Brussels, Belgium.
EN 199412, 2005: Eurocode 4: Design of composite steel and concrete structures Part 12: General rules 
Structural fire design. European Committee for Standardization, Brussels, Belgium.
EN 102101, 2006: Hot finished structural hollow sections of nonalloy and fine grain steels Part 1: Technical
delivery conditions. European Committee for Standardization, Brussels, Belgium.
EN 102102, 2006: Hot finished structural hollow sections of nonalloy and fine grain steels Part 2:
Tolerances, dimensions and sectional properties. European Committee for Standardization, Brussels, Belgium.
EN 102191, 2006: Cold formed welded structural hollow sections of nonalloy and fine grain steels Part 1:
Technical delivery conditions. European Committee for Standardization, Brussels, Belgium.
211
EN 102192, 2006: Cold formed welded structural hollow sections of nonalloy and fine grain steels Part 2:
Tolerances, dimensions and sectional properties. European Committee for Standardization, Brussels, Belgium.
Fleischer, O., & Puthli, R., 2008: Extending existing design rules in EN 199318 (2005) for gapped RHS K
joints for maximum chord slenderness (b
0
/t
0
) of 35 to 50 and gap size g to as low as 4t
0
. Proceedings 12
th
International Symposium on Tubular Structures, Shanghai, China, Tubular Structures XII, Taylor & Francis
Group, London, UK, pp. 293301.
Frater, G.S., & Packer, J.A., 1990: Design of fillet weldments for hollow structural section trusses. CIDECT
Report 5AN/290/7, University of Toronto, Toronto, Canada.
Grandjean, G., Grimault, J.P., & Petit, L., 1980: Dtermination de la dure au feu des profiles creux remplis de
bton. Cometube, Paris, France, (also published as ECSC Report No. 7210SA/302).
Guiaux, P., & Janss, J., 1970: Comportement au flambement de colonnes constitues de tubes en acier
remplis de bton. Centre de Recherches Scientifiques et Techniques de l'Industrie des Fabrications
Mtalliques, MT 65, Brussels, Belgium.
Herion, S., 1994: Rumliche KKnoten aus RechteckHohlprofilen. Ph.D. Thesis, University of Karlsruhe,
Karlsruhe, Germany.
Hoadley, P.W., & Yura, J.A., 1985: Ultimate strength of tubular joints subjected to combined loads. Proceedings
Offshore Technology Conference, OTC 4854, Houston, USA.
Hnig, O., Klingsch, W., & Witte, H., 1985: Baulicher Brandschutz durch wassergefllte Sttzen in
Rahmentragwerken (Fire resistance of water filled columns). Research Report, Studiengesellschaft fr
Stahlanwendung e.V., Forschungsbericht p. 86/4.5, Dsseldorf, Germany.
Igarashi, S., Wakiyama, K., lnoue, K., Matsumoto, T., & Murase, Y., 1985: Limit design of high strength bolted
tube flange joint Parts 1 and 2. Journal of Structural and Construction Engineering, Transactions of AIJ,
Department of Architecture Reports, Osaka University, Osaka, Japan.
IISI, 1997: Innovation in steel Bridges around the world. International Iron and Steel Institute.
IIW, 1989: Design recommendations for hollow section joints Predominantly statically loaded. 2
nd
Edition,
International Institute of Welding, Subcommission XVE, Annual Assembly, Helsinki, Finland, IIW Doc. XV
70189.
IIW, 1999: Recommended fatigue design procedure for welded hollow section joints, Part 1: Recommendations
and Part 2: Commentary. International Institute of Welding, Subcommission XVE, IIW Docs. XV103599/XIII
180499.
IIW, 2008: Recommendations for fatigue design of welded joints and components. International Institute of
Welding, IIW Docs. IIW182307/XIII2151r407/XV1254r407.
IIW, 2009: Static design procedure for welded hollow section joints Recommendations. 3
rd
Edition,
International Institute of Welding, Subcommission XVE, Annual Assembly, Singapore, IIW Doc. XV132909.
ISO 65714, 2000: Hotrolled steel sections Part 14: Hotfinished structural hollow sections Dimensions and
sectional properties. International Organization for Standardization, Geneva, Switzerland.
ISO 8341, 1999: Fire resistance tests Elements of building construction Part 1: General requirements.
International Organization for Standardization, Geneva, Switzerland.
212
ISO 4019, 2001: Structural steels Coldformed, welded, structural hollow sections Dimensions and sectional
properties. International Organization for Standardization, Geneva, Switzerland.
ISO 19902, 2007: Petroleum and natural gas industries Fixed steel offshore structures. International
Organization for Standardization, Geneva, Switzerland.
Jamm, W., 1951: Form strength of welded tubular connections and tubular structures under static loading.
(Translation from German). Schweissen und Schneiden, Vol. 3, Germany.
Kaim, P., 2006: Buckling of members with rectangular hollow sections. Proceedings 11
th
International
Symposium on Tubular Structures, Quebec City, Canada, Tubular Structures XI, Taylor & Francis Group,
London, UK, pp. 443449.
Kamba, T., & Tabuchi, M., 1994: Database for tubular column to beam connections in moment resisting frames.
IIW Doc. XVE94208, Dept. of Architecture, Kobe University, Kobe, Japan.
Kato, B., & Hirose, A., 1984: Bolted tension flanges joining circular hollow section members. CIDECT Report
8C84/24E.
Kitipornchai, S., & Traves, W.H., 1989: Welded tee end connections for circular hollow tubes. Journal of
Structural Engineering, American Society of Civil Engineers, Vol. 115, No. 12, pp. 31553170.
Koning, C.H.M. de, & Wardenier, J., 1979: Tests on welded joints in complete girders made of square hollow
sections. Stevin Report 6794, Delft University of Technology, Delft, The Netherlands.
Korol, R.M., ElZanaty, M., & Brady, F.J., 1977: Unequal width connections of square hollow sections in
Vierendeel trusses. Canadian Journal of Civil Engineering, Vol. 4, No. 2, pp. 190201.
Kosteski, N., & Packer, J.A., 2003: Longitudinal plate and through platetoHSS welded connections. Journal of
Structural Engineering, American Society of Civil Engineers, Vol. 129, No. 4, pp. 478486.
Kosteski, N., Packer, J.A., & Puthli, R.S., 2003: Notch toughness of cold formed hollow sections. CIDECT
Report 1B2/03. University of Toronto, Toronto, Canada.
Khn, B., 2005: Beitrag zur Vereinheitlichung der Europischen Regelungen zur Vermeidung von Sprdbruch.
Ph.D. Thesis, RWTH Aachen, Aachen, Germany.
Kurobane, Y., 1981: New developments and practices in tubular joint design (+ Addendum). International
Institute of Welding, Annual Assembly, Oporto, Portugal, IIW Doc. XV48881.
Kurobane, Y., Packer, J.A., Wardenier, J., & Yeomans, N., 2004: Design guide for structural hollow section
column connections. CIDECT Series "Construction with hollow steel sections" No. 9, TVVerlag, Kln,
Germany, ISBN 3824908026.
Lind, N.C., & Shroff, D.K., 1971: Utilization of cold work in light gage steel. Proceedings 1
st
Specialty
Conference on ColdFormed Steel Structures, University of Missouri, Rolla, USA, pp. 1013.
Liu, D.K., Yu, Y., & Wardenier, J., 1998: Effect of boundary conditions and chord preload on the strength of
RHS uniplanar gap Kjoints. Proceedings 8
th
International Symposium on Tubular Structures, Singapore,
Tubular Structures VIII, Balkema, Rotterdam, The Netherlands, pp. 223230.
Liu, D.K., & Wardenier, J., 2001: Multiplanar influence on the strength of RHS multiplanar gap KK joints.
Proceedings 9
th
lnternational Symposium on Tubular Structures, Dsseldorf, Germany, Tubular Structures IX,
Swets & Zeitlinger, Lisse, The Netherlands, pp. 203212.
213
Liu, D.K., & Wardenier, J., 2003: The strength of multiplanar KKjoints of square hollow sections. Proceedings
10
th
International Symposium on Tubular Structures, Madrid, Spain, Tubular Structures X, Swets & Zeitlinger,
Lisse, The Netherlands, pp. 197205.
Liu, D.K., & Wardenier, J., 2004: Effect of the yield strength on the static strength of uniplanar Kjoints in RHS
(steel grades S460, S355 and S235). IIW Doc. XVE04293, Delft University of Technology, Delft, The
Netherlands.
Liu, D.K., Chen, Y., & Wardenier, J., 2005: Design recommendations for RHSK joints with 50% overlap.
Proceedings 15
th
International Offshore and Polar Engineering Conference, Seoul, Korea, Vol. IV, pp. 308315.
Liu, D.K., & Wardenier, J., 2006: Effect of chord loads on the strength of RHS uniplanar gap Kjoints.
Proceedings.11
th
International Symposium on Tubular Structures, Quebec City, Canada, Tubular Structures XI,
Taylor & Francis Group, London, UK, pp. 539544.
Lu, L.H., Winkel, G.D. de, Yu, Y., & Wardenier, J., 1994: Deformation limit for the ultimate strength of hollow
section joints. Proceedings 6
th
International Symposium on Tubular Structures, Melbourne, Australia, Tubular
Structures VI, Balkema, Rotterdam, The Netherlands, pp. 341347.
Lu, L.H., 1997: The static strength of Ibeam to rectangular hollow section column connections. Ph.D. Thesis,
Delft University of Technology, Delft, The Netherlands.
Makino, Y., Kurobane, Y., Paul, J.C., Orita, Y., & Hiraishi, K., 1991: Ultimate capacity of gusset platetotube
joints under axial and in plane bending loads. Proceedings 4
th
International Symposium on Tubular Structures,
Delft, The Netherlands, Delft University Press, Delft, The Netherlands, pp. 424434.
Makino, Y., Kurobane, Y., Ochi, K., Vegte, G.J. van der, & Wilmshurst, S.R., 1996: Database of test and
numerical analysis results for unstiffened tubular joints. IIW Doc. XVE96220, Kumamoto University,
Kumamoto, Japan.
Mang, F., & Bucak, ., 1983: Hohlprofilkonstruktionen. StahlbauHandbuch, Bd. I, StahlbauVerlag, Kln,
Germany.
Mang, F., Bucak, ., & Wolfmuller, F., 1983: The development of recommendations for the design of welded
joints between steel structural hollow sections (T and Xtype joints). Final Report on ECSC Agreement 7210
SA/l 09 and CIDECT Programme 5AD, University of Karlsruhe, Karlsruhe, Germany.
Marshall, J., 1971: Torsional behaviour of structural rectangular hollow sections. The Structural Engineer, The
Institution of Structural Engineers, London, UK, Vol. 49, Issue 8, pp. 375379.
Marshall, P.W., & Toprac, A.A., 1974: Basis for tubular joint design. Welding Journal, Vol. 53, No. 5, pp. 192
201, (also ASCE preprint no. 2008, April 1973).
Marshall, P.W., 1984: Connections for welded tubular structures. IIW Houdremont Lecture, Proceedings 2
nd
International Conference on Welding of Tubular Structures, Boston, USA. Pergamon Press, Oxford, UK, pp. 1
54.
Marshall, P.W., 1992: Design of welded tubular connections. Elsevier, Amsterdam, The Netherlands.
Marshall, P.W., 2004: Review of tubular joint criteria. Proceedings ECCSAISC Workshop "Connections in
Steel Structures V", Bouwen met Staal, Zoetermeer, The Netherlands, pp. 457467.
Marshall, P.W., 2006: Punching shear and hot spot stress  Back to the future? Kurobane Lecture, Proceedings
11
th
International Symposium on Tubular Structures, Quebec City, Canada, Tubular Structures XI, Taylor &
Francis Group, London, UK, pp. 287299.
214
MartinezSaucedo, G., Packer, J.A., & Zhao, X.L., 2008: Static design of elliptical hollow section end
connections. Proceedings Institution of Civil Engineers, Structures & Buildings 161, Issue SB2, pp. 103113.
Mashiri, F.R., Zhao, X.L., Hirt, M., & Nussbaumer, A., 2007: Size effect of welded thinwalled tubular joints.
International Journal of Structural Stability and Dynamics, Vol. 7, No. 1, pp. 101127.
Mouty, J., 1981: Effective lengths of lattice girder members. CIDECT Monograph No. 4, CIDECT.
Natarajan, M., & Toprac, A.A., 1968: Studies on tubular joints in Japan: Review of research reports. Structures
Fatigue Research Laboratory, The University of Texas, Austin, USA.
Natarajan, M., & Toprac, A.A., 1969: Studies on tubular joints in USA: Review of research reports. Structures
Fatigue Research Laboratory, The University of Texas, Austin, USA.
Noordhoek, C., Wardenier, J., & Dutta, D., 1980: The fatigue behaviour of welded joints in square hollow
sections Part 2: Analysis. Stevin Report 6804, TNOIBBC Report BI8010/0063.4.3821, Delft University of
Technology, Delft, The Netherlands.
Ono, T., Iwata, M., & Ishida, K., 1991: An experimental study on joints of new truss system using rectangular
hollow sections. Proceedings 4
th
International Symposium on Tubular Structures, Delft, The Netherlands, Delft
University Press, Delft, The Netherlands, pp. 344353.
Packer, J.A., 1978: Theoretical behaviour and analysis of welded steel joints with RHS chord sections. Ph.D.
Thesis, University of Nottingham, Nottingham, UK.
Packer, J.A., & Haleem, A.S., 1981: Ultimate strength formulae for statically loaded welded HSS joints in lattice
girders with RHS chords. Proceedings Canadian Society for Civil Engineering, Annual Conference, Fredericton,
Canada, Vol. 1, pp. 331343.
Packer, J.A., Bruno, L., & Birkemoe, P.C., 1989: Limit analysis of bolted RHS flange plate joints. Journal of
Structural Engineering, American Society of Civil Engineers, Vol. 115, No. 9, pp. 22262242.
Packer, J.A., & Wardenier, J., 1992: Design rules for welds in RHS K, T, Y and X connections. Proceedings IIW
International Conference on Engineering Design in Welded Constructions, Madrid, Spain, pp. 113120.
Packer, J.A., Wardenier, J., Kurobane, Y., Dutta, D., & Yeomans, N., 1992: Design guide for rectangular hollow
section (RHS) joints under predominantly static loading. 1
st
Edition, CIDECT Series "Construction with hollow
steel sections" No. 3, Verlag TV Rheinland, Kln, Germany, ISBN 3824900890.
Packer, J.A., 1993: Overview of current international design guidance on hollow structural section connections.
Proceedings 3
rd
International Offshore and Polar Engineering Conference, Singapore, Vol. 4, pp. 17.
Packer, J.A., 1996: Nailed tubular connections under axial loading. Journal of Structural Engineering, American
Society of Civil Engineers, Vol. 122. No. 8, pp. 458467.
Packer, J.A., & Henderson, J.E., 1997: Hollow structural section connections and trusses A design guide. 2
nd
Edition, Canadian Institute of Steel Construction, Toronto, Canada.
Packer, J.A., 2006: Tubular brace member connections in braced steel frames. Houdremont Lecture,
Proceedings 11
th
International Symposium on Tubular Structures, Quebec City, Canada, Tubular Structures XI,
Taylor & Francis Group, London, UK, pp. 314.
Packer, J.A., Mashiri, F.R., Zhao, X.L., & Willibald, S., 2007: Static and fatigue design of CHStoRHS welded
connections using a branch conversion method. Journal of Constructional Steel Research, Vol. 63, Issue 1, pp.
8295.
215
Packer, J.A., Wardenier, J., Zhao, X.L., Vegte, G.J. van der, & Kurobane, Y., 2009a: Design guide for
rectangular hollow section (RHS) joints under predominantly static loading. 2
nd
Edition, CIDECT Series
"Construction with hollow steel sections" No. 3, CIDECT, ISBN 9783938817049.
Packer, J.A., Wardenier, J., Choo, Y.S., & Chiew, S.P., 2009b: Elliptical steel tubes. Steel News and Notes,
Singapore Structural Steel Society, SN&N 25
th
Anniversary Issue, pp. 8690.
Packer, J., Sherman, D., & Lecce, M., 2010: Hollow structural section connections. Steel Design Guide No. 24,
American Institute of Steel Construction, Chicago, USA.
Paul, J.C., 1992: The ultimate behaviour of multiplanar TT and KK joints made of circular hollow sections. Ph.D.
Thesis, Kumamoto University, Kumamoto, Japan.
Pecknold, D.A., Marshall, P.W., & Bucknell, J., 2007: New API RP2A tubular joint strength design provisions.
Journal of Energy Resources Technology, American Society of Mechanical Engineers, USA, Vol. 129, No. 3,
pp. 177189.
Pietrapertosa, C., & Jaspart, J.P., 2003: Study of the behaviour of welded joints composed of elliptical hollow
sections. Proceedings 10
th
International Symposium on Tubular Structures, Madrid, Spain, Tubular Structures
X, Swets & Zeitlinger, Lisse, The Netherlands, pp. 601608.
Puthli, R.S., 1998: Hohlprofilkonstruktionen aus Stahl nach DIN V ENV 1993 (EC 3) und DIN 18 800 (11.90).
Werner Verlag GmbH & Co. KG., Dsseldorf, Germany, ISBN 3804129757.
Puthli, R.S., & Herion, S., 2005: Welding in coldformed areas of rectangular hollow sections. CIDECT Report
1A1/05. University of Karlsruhe, Karlsruhe, Germany.
Qian, X.D., Wardenier, J., & Choo, Y.S., 2007: A uniform approach for the design of 100% CHS overlap joints.
Proceedings 5
th
International Conference on Advances in Steel Structures, Singapore, Vol. II, pp. 172182.
Qian, X.D., Choo, Y.S., Vegte, G.J. van der, & Wardenier, J., 2008: Evaluation of the new IIW CHS strength
formulae for thickwalled joints. Proceedings 12
th
International Symposium on Tubular Structures, Shanghai,
China, Tubular Structures XII, Taylor & Francis Group, London, UK, pp. 271279.
Roik, K., Bergmann, R., Bode, H., & Wagenknecht, G., 1975: Tragfhigkeit von ausbetonierten
Hohlprofilsttzen aus Baustahl. Technisch Wissenschaftliche Mitteilungen, Institut fr Konstruktiven
Ingenieurbau, TWMHeft Nr. 754, RuhrUniversitt Bochum, Bochum, Germany.
Roik, K., & Wagenknecht, G., 1977: Traglastdiagramme zur Bemessung von Druckstben mit
doppelsymmetrischem Querschnitt aus Baustahl. Technisch Wissenschaftliche Mitteilungen, Institut fr
Konstruktiven Ingenieurbau, Heft 27, RuhrUniversitt Bochum, Bochum, Germany.
Romeijn, A., 1994: Stress and strain concentration factors of welded multiplanar tubular joints. Heron, Vol. 39,
No. 3, Delft University of Technology, Delft, The Netherlands.
Rondal, J., 1990: Study of maximum permissible weld gaps in connections with plane end cuttings (5AH2);
Simplification of circular hollow section welded joints (5AP). CIDECT Report 5AH2/5AP90/20.
Rondal, J., Wrker, K.G., Dutta, D., Wardenier, J., & Yeomans, N., 1992: Structural stability of hollow sections.
CIDECT Series "Construction with hollow steel sections" No. 2, Verlag TV Rheinland, Kln, Germany, ISBN
3824900750.
Schulz, G., 1970: Der Windwiderstand von Fachwerken aus zylindrischen Stben und seine Berechnung.
Internationaler Normenvergleich fr die Windlasten auf Fachwerken. CIDECT Monograph No. 3, CIDECT.
216
Sedlacek, G., Feldmann, M., Khn, B., Tschickardt, D., Hhler, S., Mller, C., Hensen, W., Stranghner, N.,
Dahl, W., Langenberg, P., Mnstermann, S., Brozetti, J., Raoul, J., Pope, R., & Bijlaard, F., 2008: Commentary
and worked examples to EN 1993110 Material toughness and through thickness properties and other
toughness oriented rules in EN 1993. JRC 47278, EUR 23510 EN, Luxembourg.
Shanmugam, N.E., Ting, L.C., & Lee, S.L., 1993: Static behaviour of Ibeam to boxcolumn connections with
external stiffeners. The Structural Engineer, The Institution of Structural Engineers, London, UK, Vol. 71, Issue
15, pp. 269275.
Sidercad, & British Steel, 1996: Hollow section connections using HolloFast expansion bolts. CIDECT Project
6G Final Report 6G14E/96, British Steel Tubes & Pipes, Corby, UK.
Soininen, R., 1996: Fracture behaviour and assessment of design requirements against fracture in welded steel
structures made of coldformed rectangular hollow sections. Ph.D. Thesis, Lappeenranta University of
Technology, Lappeenranta, Finland.
Sopha, T., Chiew, S.P., & Wardenier, J., 2006: Test results for RHS Kjoints with 50% and 100% overlap.
Proceedings 11
th
International Symposium on Tubular Structures, Quebec City, Canada, Tubular Structures XI,
Taylor & Francis Group, London, UK, pp. 133138.
Struik, J.H.A., & Back, J. de, 1969: Tests on bolted Tstubs with respect to a bolted beamtocolumn
connection. Stevin Report 66913, Delft University of Technology, Delft, The Netherlands.
Syam, A.A., & Chapman, B.G., 1996: Design of structural steel hollow section connections. 1
st
Edition,
Australian Institute of Steel Construction, Sydney, Australia.
Theofanous, M., Chan, T.M., & Gardner, L., 2009: Flexural behaviour of stainless steel oval hollow sections.
ThinWalled Structures, Vol. 47, Nos. 67, pp. 776787.
Thiensiripipat, N., 1979: Statical behaviour of cropped web joints in tubular trusses. Ph.D. Thesis, University of
Manitoba, Winnipeg, Canada.
Tissier, P., 1978: Resistance to corrosion of the interior of hollow steel sections. Acier, Stahl, Steel 2/1978.
Togo, T., 1967: Experimental study on mechanical behaviour of tubular joints. Ph.D. Thesis, Osaka University,
Osaka, Japan, (in Japanese).
Twilt, L., & Haar, P.W. van de, 1986: Harmonization of the calculation rules for the fire resistance of concrete
filled SHScolumns. CIDECT Project 15F86/70; IBBCTNO Report B86461, Delft, The Netherlands.
Twilt, L., & Both, C., 1991: Technical notes on the realistic behaviour and design of fire exposed steel and
composite structures. Final Report ECSC 7210SA/112, Activity D: "Basis for Technical Notes", TNO Building
and Construction Research, BI91069, The Netherlands.
Twilt, L., Hass, R., Klingsch, W., Edwards, M., & Dutta, D., 1994: Design guide for structural hollow section
columns exposed to fire. CIDECT Series "Construction with hollow steel sections" No. 4, Verlag TV
Rheinland, Kln, Germany, ISBN 3824901714.
Vegte, G.J. van der, Back, J. de, Wardenier, J., 1989: Low cycle fatigue of tubular T and Xjoints. Proceedings
3
rd
International Symposium on Tubular Structures, Lappeenranta, Finland, pp. 266277.
Vegte, G.J. van der, 1995: The static strength of uniplanar and multiplanar tubular T and X joints. Ph.D. Thesis,
Delft University of Technology, Delft, The Netherlands.
217
Vegte, G.J. van der, & Makino, Y., 2006: Ultimate strength formulation for axially loaded CHS uniplanar Tjoints.
International Journal of Offshore and Polar Engineering, ISOPE, Vol. 16, No. 4, pp. 305312.
Vegte, G.J. van der, Makino, Y., & Wardenier, J., 2007: New ultimate strength formulation for axially loaded
CHS Kjoints. Proceedings 5
th
International Conference on Advances in Steel Structures, Singapore, Vol. II, pp.
218227.
Vegte, G.J. van der, Wardenier, J., Zhao, X.L., & Packer, J.A., 2008: Evaluation of new CHS strength formulae
to design strengths. Proceedings 12
th
International Symposium on Tubular Structures, Shanghai, China,
Tubular Structures XII, Taylor & Francis Group, London, UK, pp. 313322.
Vegte, G.J. van der, Wardenier, J., & Puthli, R.S., 2010a: FE analysis for welded hollow section joints and
bolted joints. Proceedings Institution of Civil Engineers, Structures & Buildings (to be published).
Vegte, G.J. van der, Wardenier, J., Qian, X.D., & Choo, Y.S., 2010b: Reevaluation of the moment capacity of
CHS joints. Proceedings Institution of Civil Engineers, Structures & Buildings (to be published).
Virdi, K.S., & Dowling, P.J., 1976: A unified design method for composite columns. International Association for
Bridge and Structrural Engineering (IABSE), Zrich, Switzerland, Mmoires Vol. 36II, pp. 165184.
Voth, A.P., 2010: Branch platetocircular hollow structural section connections. Ph.D. Thesis, University of
Toronto, Toronto, Canada.
Wang, Y.C., & Orton, A.H., 2008: Fire resistant design of concrete filled tubular steel columns. The Structural
Engineer, The Institution of Structural Engineers, London, UK, Vol. 86, Issue 19, pp. 4045.
Wang, Y.C., & Ding, J, 2009: The robustness of joints between steel beams and concrete filled tubular (CFT)
columns under fire conditions. CIDECT Final Report 15S14/09, The University of Manchester, Manchester, UK.
Wanke, J., 1966: Stahlrohrkonstruktionen. Springer Verlag, Vienna, Austria, ISBN 9783211807910.
Wardenier, J., & Mouty, J., 1979: Design rules for predominantly statically loaded welded joints with hollow
sections as bracings and an I or Hsection as chord. Welding in the World, Vol. 17, No. 9/10.
Wardenier, J., 1982: Hollow section joints. Delft University Press, Delft, The Netherlands.
Wardenier, J., & Giddings, T.W., 1986: The strength and behaviour of statically loaded welded connections in
structural hollow sections. CIDECT Monograph No. 6, CIDECT.
Wardenier, J., Kurobane, Y., Packer, J.A., Dutta, D., & Yeomans, N., 1991: Design guide for circular hollow
section (CHS) joints under predominantly static loading. 1
st
Edition, CIDECT Series "Construction with hollow
steel sections" No. 1, Verlag TV Rheinland, Kln, Germany, ISBN 3885859750.
Wardenier, J., Dutta, D., Yeomans, N., Packer, J.A., & Bucak, ., 1995: Design guide for structural hollow
sections in mechanical applications. CIDECT Series "Construction with hollow steel sections" No. 6, Verlag
TV Rheinland, Kln, Germany, ISBN 3824903024.
Wardenier, J., 2002: Hollow sections in structural applications. 1
st
Edition, Bouwen met Staal, Zoetermeer, The
Netherlands.
Wardenier, J., & Choo, Y.S., 2006: Recent developments in welded hollow section joint recommendations.
Advanced Steel Construction, The Hong Kong Institute of Steel Construction, Vol. 2, No. 2, pp. 109127.
Wardenier, J., 2007: A uniform effective width approach for the design of CHS overlap joints. Proceedings 5
th
International Conference on Advances in Steel Structures, Singapore, Vol. II, pp. 155165.
218
Wardenier, J., Vegte, G.J. van der, & Liu, D.K., 2007a: Chord stress function for rectangular hollow section X
and T joints. Proceedings 17
th
International Offshore and Polar Engineering Conference, Lisbon, Portugal, Vol.
IV, pp. 33633370.
Wardenier, J., Vegte, G.J. van der, & Liu, D.K., 2007b: Chord stress functions for K gap joints of rectangular
hollow sections. International Journal of Offshore and Polar Engineering, ISOPE, Vol. 17, No. 3, pp. 225232.
Wardenier, J., Kurobane, Y., Packer, J.A., Vegte, G.J. van der, & Zhao, X.L., 2008a: Design guide for circular
hollow section (CHS) joints under predominantly static loading. 2
nd
Edition, CIDECT Series "Construction with
hollow steel sections" No. 1, CIDECT, ISBN 9783938817032.
Wardenier, J., Vegte, G.J. van der, Makino, Y., & Marshall, P.W., 2008b: Comparison of the new IIW (2008)
CHS joint strength formulae with those of the previous IIW (1989) and the new API (2007). Proceedings 12
th
International Symposium on Tubular Structures, Shanghai, China, Tubular Structures XII, Taylor & Francis
Group, London, UK, pp. 281291.
Wardenier, J., Vegte, G.J. van der, & Makino, Y., 2009: Joints between plates or I sections and a circular
hollow section chord. International Journal of Offshore and Polar Engineering, ISOPE, Vol. 19, No. 3, pp. 232
239.
Weynand, K., Busse, E., & Jaspart, J.P., 2006: First practical implementation of the component method for
joints in tubular construction. Proceedings 11
th
International Symposium on Tubular Structures, Quebec City,
Canada, Tubular Structures XI, Taylor & Francis Group, London, UK, pp. 139145.
Whitmore, R.E., 1952: Experimental investigation of stresses in gusset plates. Bulletin 16, Engineering
Experiment Station, The University of Tennessee, Knoxville, USA.
Wilkinson, T., & Hancock, G., 1998: Compact or class 1 limits for rectangular hollow sections in bending.
Proceedings 8
th
International Symposium on Tubular Structures, Singapore, Tubular Structures VIII, Balkema,
Rotterdam, The Netherlands, pp. 409416.
Willibald, S., Packer, J.A., & Puthli, R.S., 2002: Experimental study of bolted HSS flangeplate connections in
axial tension. Journal of Structural Engineering, American Society of Civil Engineers, Vol. 128, No. 3, pp. 328
336.
Willibald, S., Packer, J.A., & Puthli, R.S., 2003a: Design recommendations for bolted rectangular HSS flange
plate connections in axial tension. Engineering Journal, American Institute of Steel Construction, Vol. 40, First
Quarter, pp. 1524.
Willibald, S., Packer, J.A., & Puthli, R.S., 2003b: Investigation on hidden joint connections under tensile
loading. Proceedings 10
th
International Symposium on Tubular Structures, Madrid, Spain, Tubular Structures X,
Swets & Zeitlinger, Lisse, The Netherlands, pp. 217225.
Willibald, S., Packer, J.A., & MartinezSaucedo, G., 2006: Behaviour of gusset plate connections to ends of
round and elliptical hollow structural section members. Canadian Journal of Civil Engineering, Vol. 33, pp. 373
383.
Wingerde, A.M. van, 1992: The fatigue behaviour of T and Xjoints made of square hollow sections. Heron,
Vol. 37, No. 2, Delft University of Technology, Delft, The Netherlands.
Wingerde, A.M. van, Delft, D.R.V. van, Wardenier, J., & Packer, J.A., 1997: Scale effects on the fatigue
behaviour of tubular structures. Proceedings IIW International Conference on Performance of Dynamically
Loaded Welded Structures, San Francisco, USA, pp. 123135.
219
220
Winkel, G.D de, 1998: The static strength of Ibeam to circular hollow section column connections. Ph.D.
Thesis, Delft University of Technology, Delft, The Netherlands.
Yeomans, N.F., 1994: IBeam/rectangular hollow section column connections using the Flowdrill system.
Proceedings 6
th
International Symposium on Tubular Structures, Melbourne, Australia, Tubular Structures VI,
Balkema, Rotterdam, The Netherlands, pp. 381388.
Yeomans, N.F., 1998: Rectangular hollow section column connections using the Lindapter HolloBolt.
Proceedings 8
th
International Symposium on Tubular Structures, Singapore, Tubular Structures VIII, Balkema,
Rotterdam, The Netherlands, pp. 559566.
Yu, Y., 1997: The static strength of uniplanar and multiplanar connections in rectangular hollow sections. Ph.D.
Thesis, Delft University of Technology, Delft, The Netherlands.
Zhao, X.L., 1992: The behaviour of cold formed RHS beams under combined actions. Ph.D. Thesis, The
University of Sydney, Sydney, Australia.
Zhao, X.L., Herion, S., Packer, J A., Puthli, R.S., Sedlacek, G., Wardenier, J., Weynand, K., Wingerde, A.M.
van, & Yeomans, N.F., 2001: Design guide for circular and rectangular hollow section welded joints under
fatigue loading. CIDECT Series "Construction with hollow steel sections" No. 8, TVVerlag, Kln, Germany,
ISBN 3824905655.
Zhao, X.L., Wilkinson, T., & Hancock, G.J., 2005: Coldformed tubular members and connections. Elsevier
Science, London, UK, ISBN 9780080441016.
Zhao, X.L., Wardenier, J., Packer, J.A., & Vegte, G.J. van der, 2008: New IIW (2008) static design
recommendations for hollow section joints. Proceedings 12
th
International Symposium on Tubular Structures,
Shanghai, China, Tubular Structures XII, Taylor & Francis Group, London, UK, pp. 261269.
Zhao, X.L., & Packer, J.A., 2009: Tests and design of concretefilled elliptical hollow section stub columns.
ThinWalled Structures, Vol. 47, Nos. 67, pp. 617628.
Zhao, X.L., Han, L.H., & Lu, H., 2010: Concretefilled tubular members and connections. Taylor & Francis
Group, London, UK, ISBN 9780415435000.
SYMBOLS
Abbreviations of organisations
AISC American Institute of Steel Construction
AWS American Welding Society
CEN European Committee for Standardization
CIDECT Comit International pour le Dveloppement et l'Etude de la Construction Tubulaire
CSA Canadian Standards Association
IIW International Institute of Welding
ISO International Organization for Standardization
Other abbreviations
CHS circular hollow section
CTOD crack tip opening displacement
FE finite element
RHS rectangular or square hollow section
SCF stress concentration factor
SHS square hollow section
TTP through thickness properties
General symbols
A cross sectional area
A
a
cross sectional area of structural steel in a composite column
A
c
cross sectional area of concrete in a composite column
A
gv
gross area in shear for block failure
A
i
cross sectional area of member i (i = 0, 1, 2, 3)
A
m
exposed surface area of a steel member/unit length under fire conditions; cross section
parameter for torsion
A
n
net section in a bolted joint
A
net
net cross sectional area
A
s
cross sectional area of reinforcement in a composite column
A
v
shear area
B
e
effective chord length in ring model
B
f
width parameter of diaphragm plate in beamtocolumn joints
C constant; constant in N relationship; multiplication factor to account for secondary bending
moments
C
e
efficiency parameter for joints (general)
C
K
efficiency parameter for K joints
C
T
efficiency parameter for T joints
C
X
efficiency parameter for X joints
C
1
constant used in chord stress function
E modulus of elasticity
E
a
modulus of elasticity of steel section in a composite column
E
cm
modulus of elasticity of concrete in a composite column
E
c,eff
modulus of elasticity of concrete in a composite column, corrected for creep
E
d
total energy dissipated in yield line model
E
fi,d,t
design load under fire conditions, at fire exposure time t
E
s
modulus of elasticity of reinforcement in a composite column
(EI)
eff
effective stiffness of a composite column
221
(EI)
eff,
stiffness of a composite column used for a second order analysis
F force
I moment of inertia
I
a
moment of inertia of steel section in a composite column
I
b
moment of inertia of a beam
I
c
moment of inertia of concrete in a composite column
I
s
moment of inertia of reinforcement in a composite column
I
t
torsional moment of inertia
J
AA
ratio of outofplane axial load and inplane axial load in a multiplanar joint
K
b
stiffness of a beam
L length; span; length in a bolted (gusset plate) joint
L
b
length of a beam
L
eff
effective length
L
o
measured length of a tensile test specimen
L
w
weld length in a gusset plate joint
M moment
M
b
bending moment in a beam at midspan
M
c,Rd
design resistance of a cross section to bending moment
M
e
elastic moment capacity
M
el
elastic yield moment capacity
M
Ed
design bending moment
M
Ed,
second order bending moment
M
f
bending moment in a flange
M
gap,0,Ed
design bending moment in the cross section of a chord at the gap
M
i
applied bending moment in member i (i = 0, 1, 2, 3)
M
i,Rd
design resistance of a joint, expressed in terms of bending moment in member i (i = 0, 1, 2, 3)
M
ip,i
applied inplane bending moment in member i (i = 0, 1, 2, 3)
M
ip,i,Ed
design applied inplane bending moment in member i (i = 0, 1, 2, 3)
M
ip,i,Rd
design resistance of a joint, expressed in terms of inplane bending moment in member i (i = 0,
1, 2, 3)
M
j
applied bending moment in a beamtocolumn joint
M
j,Ed
applied design bending moment in a beamtocolumn joint
M
j,Rd
design moment resistance of a beamtocolumn joint
M
max,Rd
maximum design resistance of a composite cross section to bending moment
M
op,i
applied outofplane bending moment in member i (i = 0, 1, 2, 3)
M
op,i,Ed
design applied outofplane bending moment in member i (i = 0, 1, 2, 3)
M
op,i,Rd
design resistance of a joint, expressed in terms of outofplane bending moment in member i
(i = 0, 1, 2, 3)
M
pl
plastic moment capacity
M
pl,f
plastic moment capacity of a flange
M
pl,V,0,Rd
design value of the plastic moment capacity of a chord cross section, reduced by shear
M
pl,Rd
design value of the plastic moment capacity of a (composite) cross section
M
pl,y,Rd
design resistance of a (composite) cross section to bending moment about the yy axis
M
pl,z,Rd
design resistance of a (composite) cross section to bending moment about the zz axis
M
pl,0
plastic moment capacity of a chord cross section
M
pl,0,Rd
design value of the plastic moment capacity of a chord cross section
M
Rd
design resistance of a (composite) cross section to bending moment
M
t,Rd
design resistance of a cross section to torsional moment
M
y,Ed
design bending moment about the yy axis
M
z,Ed
design bending moment about the zz axis
M
0
bending moment in a chord
M
0,Ed
design bending moment in a chord
M
1
brace or beam bending moment
M
,max
design bending moment of a member including imperfection and second order effects
N axial load; number of cycles (to failure)
222
N
b,Rd
design buckling resistance of a member
N
cr,eff
buckling capacity used to determine the second order effects
N
Ed
(acting) design axial load (at room temperature)
N
equ
equivalent design axial load under fire conditions including effect of eccentricities
N
fi,Ed
design axial load under fire conditions
N
fi,Rd
buckling resistance of a member under fire conditions
N
G,Ed
permanent part of the acting design force in a composite column
N
gap,0
axial load in the cross section of a chord at the gap
N
gap,0,Ed
design axial load in the cross section of a chord at the gap
N
gap,0,Rd
reduced resistance to axial load, due to shear, in the cross section of a chord at the gap
N
i
applied axial force in brace i (i = 1, 2, 3); applied axial force in overlapping brace
N
i
number of cycles to failure (in PalmgrenMiner rule)
N
i,Ed
design applied axial load in member i (i = 0, 1, 2, 3)
N
i,Rd
design resistance of a joint, expressed in terms of axial load in member i (i = 0, 1, 2, 3)
N
j
applied axial load in overlapped brace
N
j,Ed
design applied axial load in overlapped brace
N
pl
axial load capacity of a cross section
N
pl,V,0,Rd
design value of the axial load capacity of a chord cross section, reduced by shear
N
pl,Rd
design resistance of a (composite) cross section to axial load (at room temperature)
N
pl,Rk
characteristic resistance of a (composite) cross section to axial load
N
pl,0
axial yield capacity of a chord cross section
N
pl,0,Rd
design axial yield capacity of a chord cross section
N
Rd
design resistance of a (composite) cross section to axial load; buckling resistance of a member
at room temperature
N
s,Rd
design value of the shear resistance of the brace(s) at the connection with the chord
N
t,Rd
design tensile capacity of a cross section
N
0
chord load
N
0,Ed
design value of the chord load
N
0p
chord preload
N
1u
ultimate axial load capacity of a joint based on the load in member 1
N
1u
(J
AA
) ultimate strength of a multiplanar joint with load ratio J
AA
N
1u
(J
AA
= 0) ultimate strength of a multiplanar joint with unloaded outofplane braces
P factored design load
Ov overlap, Ov = q/p x 100%
Ov
limit
limit for overlap, decisive for brace shear check
Q
f
chord stress function
Q
u
function in the design strength equations accounting for the effect of geometric parameters
R stress or load ratio
R
AZ
reduction of area in a tensile test (in %)
R
fi,d,t
design resistance under fire conditions, at fire exposure time t
R(t) resistance under fire conditions
S static moment to neutral axis
S
j
rotational stiffness of a joint
S
j,ini
initial rotational stiffness of a joint
S
o
cross sectional area of a standard tensile test specimen (in mm
2
)
V volume of a steel member/unit length; shear load
V
Ed
(acting) design shear load
V
f
shear load in a flange
V
gap,0,Ed
design shear load in the cross section of a chord at the gap
V
i
shear load in brace i (i = 1, 2, 3)
V
p
punching shear stress
V
pl,f
plastic shear capacity of a flange
V
pl,Rd
design plastic shear yield capacity
V
pl,0
plastic shear yield capacity of a chord cross section
V
pl,0,Rd
design plastic shear yield capacity of a chord cross section
223
V
0
shear load in a chord
V
0,Ed
design shear load in a chord
W
eff
effective section modulus
W
el
elastic section modulus
W
el,ip,i
elastic section modulus for inplane bending moment in member i
W
el,op,i
elastic section modulus for outofplane bending moment in member i
W
i
elastic section modulus of member i (i = 0, 1, 2, 3)
W
pl
plastic section modulus
W
pl,i
plastic section modulus of member i (i = 0, 1, 2, 3)
W
t
section modulus for torsion
a throat thickness of a weld; edge distance of bolt
a edge distance of plastic hinge line in bolted end plate connection
b external width of a rectangular hollow section; width of a plate; distance from bolt to RHS face
b
e
effective width of a plate, a beam flange or a brace member
b
ei
effective width of an overlapping RHS brace member at the chord connection
b
ej
effective width of an overlapped RHS brace member at the chord connection
b
e,ov
effective width of an overlapping RHS brace member at the connection to the overlapped
brace
b
e,p
effective punching shear width
b
f
minimum width of a diaphragm plate in beamtocolumn joints
b
i
external width of brace i (i = 1, 2, 3)
b
j
width of overlapped brace
b
m
average width of an RHS (bt)
b
sp
width of a stiffening plate
b
w
effective width of a web
b
wf
effective width of a web under a brace wall, a plate or a beam flange
b distance between plastic hinge lines in bolted end plate connection
b
0
external width of a chord
b
1
width of a plate or width of an I section or RHS brace
c constant in chord MN interaction equation
c, c
0
, c
1
, c
2
coefficients
c
s
factor considering the condition (welded or unwelded) at the hidden toe of the overlapped
brace
d external diameter of a hollow section; bolt diameter
d
ei
effective width of an overlapping CHS brace member at the chord connection
d
ej
effective width of an overlapped CHS brace member at the chord connection
d
e,ov
effective width of an overlapping CHS brace member at the connection to the overlapped
brace
d
i
external diameter of brace i (i = 1, 2, 3); diameter of overlapping brace; external diameter of
inner tube in composite column
d
j
diameter of overlapped brace
d
w
depth of the web of an I section chord (d
w
= h
0
 2t
0
 2r)
d bolt hole diameter
d
0
external diameter of a chord
e, e
0
eccentricity
e
1
distance from bolt to CHS face
e
2
edge distance of bolt
f
b,Rd
design buckling stress
f
c
strength of concrete in a composite column
f
cd
design strength of concrete in a composite column
f
ck
characteristic concrete cylinder strength (in N/mm
2
)
f
ck,cub
characteristic concrete cube strength (in N/mm
2
)
f
ck,cyl
characteristic concrete cylinder strength (in N/mm
2
)
f
k
buckling stress for chord side wall failure (general)
224
f
s
yield strength of reinforcement in a composite column
f
sd
design strength of reinforcement in a composite column
f
sk
characteristic strength of reinforcement in a composite column
f
u
specified ultimate tensile strength
f
u,b
ultimate tensile strength of material "b"
f
ui
ultimate tensile strength of overlapping brace
f
uj
ultimate tensile strength of overlapped brace
f
u0
ultimate tensile strength of a chord
f
y
yield strength; design yield strength in joint strength equations
f
ya
design yield strength based on the average yield strength of an RHS section
f
yb
yield strength of parent material
f
yd
specified design yield strength
f
yi
design yield strength of brace i (i = 1, 2, 3); design yield strength of overlapping brace
f
yj
design yield strength of overlapped brace
f
yk
characteristic yield strength of hollow steel section in a composite column
f
yp
design yield strength of a plate
f
yw
design yield strength of a web
f
y0
design yield strength of a chord
g, g
1
, g
2
gap between the braces (ignoring welds) of a K, N or KT joint at the face of the chord
g bolt gauge
g
1
bolt edge distance
g
2
bolt distance
g gap divided by chord thickness: g = g/t
0
h depth of a rectangular hollow section
h
i
external depth of brace i (i = 1, 2, 3)
h
j
depth of overlapped brace
h
m
average depth of an RHS (ht)
h
n
parameter in a composite column
h
z
moment arm
h
0
external depth of a chord
h
1
depth of a plate or depth of an I section or RHS brace
i radius of gyration
k amplification factor to incorporate second order effects
k
y,
reduction factor for the yield strength of steel at steel temperature
a
l, length
A
circumferential parameter for torsion
b
buckling length
b,eff.
effective perimeter for local brace failure
i
length of yield line i; length of member i (i = 0, 1, 2, 3)
M
reduction factor for the plastic moment capacity of a composite column
parameter for the effect of the end moments on member buckling
diameter or width ratio between braces and chord:
0
1
0
1
0
1
0
1
b
b
or
d
b
or
b
d
or
d
d
(T, Y or X joints);
0
2 1
0
2 1
0
2 1
2b
b b
or
2b
d d
or
d 2
d d
(K or N joints);
0
3 2 1
0
3 2 1
3b
b b b
or
d 3
d d d
(KT joints)
*
width parameter used for T stub joints
reduction factor for buckling according to EN 199311 (2005) or any equivalent national
buckling curve
d
ratio between actual design force and the axial capacity of a composite section
fi
reduction factor for buckling in the fire design situation
,
1
deformation
section parameter (contribution ratio of steel section in a composite column)
section class parameter accounting for different steel grades; engineering strain; logarithmic
strain
u
ultimate strain
joint resistance (or capacity) factor; angle between the planes in a multiplanar joint
,
j
rotation of a beamtocolumn joint
half diameter or half widthtothickness ratio of the chord: = d
0
/2t
0
or b
0
/2t
0
a
partial (safety) factor for steel section in a composite column
c
partial (safety) factor for concrete in a composite column
F
load factor for the action
M
,
M0
,
M1
,
M2
material or joint partial (safety) factor
s
partial (safety) factor for reinforcement in a composite column
brace member depthtochord diameter or width ratio: = h
i
/d
0
or h
i
/b
0
fi
ratio between the design load present under fire conditions and the design load at room
temperature
fi,t
ratio between the design load present under fire conditions and the buckling resistance at
room temperature
angle used in ring model
226
e
correction coefficient related to the eccentricity
i
rotation in a yield line i
s
correction coefficient related to the reinforcement
t
creep factor
slenderness of a member under compression
E
Euler slenderness
relative slenderness ratio
correction factor to uniplanar joint resistance to account for multiplanar effect
d
related bending capacity in a composite column
dy
related bending capacity about the y axis
dz
related bending capacity about the z axis
temperature; angle between brace and chord
a
steel temperature
i
acute angle between brace member i (i = 1, 2, 3) and the chord; acute angle between
overlapping brace and the chord
j
acute angle between overlapped brace and the chord
s
reinforcement ratio of a composite column (in %)
axial or bending stress; engineering stress; logarithmic stress
a
stress in steel section
c
stress in concrete
joint
stress at the intersection of brace and chord
max
maximum stress
min
minimum stress
nom
nominal stress in a member
peak
peak stress at the weld toe of a joint
r
stress in radial direction
0
chord stress
1,Ed
nominal stress in brace
shear stress
bracetochord wall thickness ratio: = t
i
/t
0
Rd
design bond stress
midspan deflection of a beam under a uniformly distributed load
T
cf
temperature shift in the notch impact energy  temperature diagram due to cold forming (cf)
strain range
stress range
geom
geometrical stress range
nom
nominal stress range
Subscripts
a referring to structural steel in a composite column
c referring to concrete in a composite column
e, eff effective
el elastic
fi referring to fire conditions
i subscript used to denote the member of a hollow section joint. Subscript i = 0 designates the
chord (or through member); i = 1 refers in general to the brace for T, Y and X joints, or it
refers to the compression brace for K, N and KT joints; i = 2 refers to the tension brace for K, N
and KT joints; i = 3 refers to the vertical brace for KT joints. For K and N overlap joints, the
subscript i is used to denote the overlapping brace member; subscript used to indicate the
member for which the SCF is given
ip inplane
227
228
j subscript used to denote the overlapped brace member for K and N overlap joints; subscript
used to indicate the location where the SCF is given
k subscript used to indicate the type of loading for which the SCF is given
max maximum
min minimum
n net
nom nominal
op outofplane
p plate; preload
pl plastic
s referring to reinforcement in a composite column
t tension; torsion; time
u ultimate
v shear
w web; weld
y yield
Ed design value of action
Rd design value of resistance
Rk characteristic value of resistance
Symbols not shown here are specifically defined at the location where they are used.
In all calculations, the nominal (guaranteed minimum) mechanical properties should be used.
Comit International pour le Dveloppement et IEtude 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 worldwide.
The CIDECT website is http://www.cidect.com
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 congresses, etc.
To cooperate with organisations concerned with specifications, practical design recommendations,
regulations or standards at national and international levels.
Technical activities
The technical activities of CIDECT have centred on the following research aspects of hollow 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 joints
Fatigue resistance of welded joints
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.
229
CIDECT Publications
The current situation relating to CIDECT publications reflects the ever increasing emphasis 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 (1
st
edition 1991 and
2
nd
edition 2008)
2. Structural stability of hollow sections (1992, reprinted 1996)
3. Design guide for rectangular hollow section (RHS) joints under predominantly static loading (1
st
edition 1992
and 2
nd
edition 2009)
4. Design guide for structural hollow section columns exposed to fire (1994, 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 (2001)
9. Design guide for structural hollow section column connections (2004)
In addition, as a result of the everincreasing interest in steel hollow sections in internationally acclaimed
structures, two books have been published, i.e. "Tubular Structures in Architecture" by Prof. Mick Eekhout (1
st
edition 1996 and 2
nd
edition 2010) and "Hollow Sections in Structural Applications" by Prof. Jaap Wardenier (1
st
edition 2002) and this 2
nd
edition by Prof. Jaap Wardenier et al. (2010).
Copies of the Design Guides, the architectural book and research papers may be obtained through the CIDECT
website: http://www.cidect.com
"Hollow Sections in Structural Applications" by Prof. Jaap Wardenier et al. (2010) is available in hard copy
colour print from the publisher:
Bouwen met Staal
Boerhaavelaan 40
2713 HX Zoetermeer, The Netherlands
P.O. Box 190
2700 AD Zoetermeer, The Netherlands
Tel. +31(0)79 353 1277
Fax +31(0)79 353 1278
Email info@bouwenmetstaal.nl
CIDECT Organisation (2010)
President: P. Ritakallio, Finland
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 Promotion Committee meeting at least once a year and directly responsible
for the research work and technical promotion work.
Chairman Technical Commission: G. Iglesias, Spain
Chairman Promotion Committee: J. Krampen, Germany
230
Present members of CIDECT are:
Atlas Tube, Canada/USA
OneSteel Australian Tube Mills, Australia
Borusan Mannesmann Boru, Turkey
Corus Tubes, United Kingdom
Grupo Condesa, Spain
Industrias Unicon, Venezuela
Rautaruukki Oyj, Finland
Robor Steel Services, South Africa
Sidenor SA, Greece
Vallourec & Mannesmann Tubes, Germany
VoestAlpine Krems, Austria
Acknowledgements for photographs
The authors express their appreciation to the following firms and persons for making available some of the
photographs used in this book:
Bouwdienst, Rijkswaterstaat, The Netherlands
CORUS Tubes, UK
Delft University of Technology, The Netherlands
HGG Profiling Equipment BV, Wieringerwerf, The Netherlands
Instituto para la Construccin Tubular (ICT), Spain
University of Toronto, Canada
Vallourec & Mannesmann Tubes, Germany
Prof. Y.S. Choo, National University of Singapore, Singapore
Mr. D. Dutta, Germany
Mr. Flix Escrig, Spain
Mr. Jos Snchez, Spain
Disclaimer
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 misinterpretations of information
contained in this book or in its use.
231
232
Viel mehr als nur Dokumente.
Entdecken, was Scribd alles zu bieten hat, inklusive Bücher und Hörbücher von großen Verlagen.
Jederzeit kündbar.