Untitled PDF

Timber trussed arch for long span
Master’s Thesis in the International Master’s Programme in Structural Engineering
AUDREY FARREYRE
JEAN-BERNARD JOURNOT
Department of Civil and Environmental Engineering
Division of Structural Engineering
Steel and Timber Structures
CHALMERS UNIVERSITY OF TECHNOLOGY
Göteborg, Sweden 2005 Master’s Thesis 2005:103
MASTER’S THESIS 2005:103
AUDREY FARREYRE

CHALMERS UNIVERSITY OF TECHNOLOGY
Göteborg, Sweden 2005

AUDREY FARREYRE
©AUDREY FARREYRE & JEAN-BERNARD JOURNOT, 2005
Master’s Thesis 2005:103103

Chalmers University of Technology
SE-412 96 Göteborg
Sweden
Telephone: + 46 (0)31-772 1000
Cover:
Elevation of Hamar Olympiahall, Niels Torp arkitekter
FE model of the trussed arch in ABAQUS,
Hamar Olympiahall during construction, Biong Arkitekter
Hamar Olympiahall inside view, Niels Torp arkitekter.

Göteborg, Sweden 2005
AUDREY FARREYRE
ABSTRACT
Scandinavia has the largest timber reserves in western Europe and it is one of the
largest manufacturers of timber products. Consequently, timber is commonly used as
a building material in roof structures, for houses and larger buildings. The
development of glued laminated timber (Glulam) technology at the beginning of the
twentieth century has enlarged the possibilities of timber. All kind of geometries and
spans are thereby feasible.
In the case of long span roof structures, timber arches are one of the best solutions
from both structural and aesthetical point of view. Not only are arches slender
structures and require no intermediate support to achieve long span, but they also use
efficiently the compressive strength of timber. For spans over 100m, trussed arches
are usually preferred since they are more convenient to manufacture and to assembly.
For instance, several buildings own such roof structures in Scandinavia.
This master’s thesis deals with Glulam trussed arches for long span. It especially
focuses on an arch of 100m span and 20m height. It introduces different technical
solutions and explains the design of such a structure.
At first, a finite element model is performed with the ABAQUS software in order to
study the influence of boundary conditions and geometrical parameters e.g. depth of
the truss and number of diagonal elements. This analysis leads to an optimum static
system, which is considered later on.
Furthermore, as the connections are the weakest point in timber structures, they have
to be investigated from the preliminary design. Hence, different joints are examined
and the best one is retained to be calculated.
Finally, the stiffness of the connections are introduced in the finite element model to
make it more realistic. The design forces resulting from the new analysis are used to
design the members and to study the stability of the structure, according to Eurocodes.
At last, the partitioning of the trussed arch for transportation is overviewed. The arch
is partitioned in six pieces and the location of the cuts is studied.
The recommendations made in this thesis are aimed to be used as references for the
design of similar structures.
Key words: Timber arch, trussed arch, glulam, timber connections, multiple steel
plates connection.
CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2005:103 I

II CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2005:103
Arche triangulée de longue portée
Thèse de fin d’étude du Master international « Structural Engineering »
AUDREY FARREYRE
Département de génie civil et environnement
RESUMÉ FRANÇAIS
Les vastes forêts recouvrant les pays scandinaves garantissent une production
quasiment illimitée de bois de construction. Il est principalement employé dans les
charpentes, pour les habitations et les bâtiments publics. L’apparition du lamellé-collé
au début du vingtième siècle a largement accru les possibilités d’utilisation du bois.
En effet, toutes les géométries et toutes les longueurs peuvent être produites. Ainsi
pour la conception d’une charpente de longue portée, la réalisation d’une arche en
lamellé-collé est incontournable, car d’une part elle utilise de façon optimale la
capacité en compression du bois et d’autre part, elle présente des avantages
esthétiques.
Ce mémoire traite des arches triangulées de longues portées en lamellé-collé. Il
introduit les possibilités de réalisations et présente la conception et le
dimensionnement d’une telle structure.
Ce projet repose plus particulièrement sur l’analyse d’une arche triangulée de 100m
de portée entre les supports et 20m de haut. Les conditions d’appuis et les paramètres
géométriques comme le nombre de diagonales et la hauteur du treillis sont déterminés
selon une étude comparative entre différents modèles. L’analyse statique des
différentes alternatives est réalisée grâce à la méthode des éléments finis avec le
logiciel ABAQUS. Cette comparaison aboutit enfin à un système statique optimum
d’un point de vue structurel.
De plus, les connections étant un point délicat dans la construction en bois, elles
doivent être considérées dès la première phase de conception. Ainsi, différents types
de connecteurs sont examinés avant de retenir le préférable. Le système BSB, formé
de broches et de plaques d’acier noyées dans la section, est ici le plus adéquat. Le
nombre de broches et de plaques est calculé de telle façon à obtenir une rupture
plastique de la connexion.
Afin d’améliorer, la modélisation de l’arche, la raideur des connections est par la suite
introduite dans le modèle en éléments finis. Les forces qui résultent de l’analyse
statique de ce modèle, sont utilisées pour effectuer le dimensionnement de la
charpente et d’en étudier sa stabilité suivant les critères donnés par l’Eurocode 5.
Enfin, la découpe de l’arche en plusieurs parties pour le transport est abordée. L’arche
est sectionnée en 6 parties. L’influence de l’emplacement des coupures est analysée
grâce à la méthode des éléments finis.
Les recommandations faites dans cette thèse ont pour but de servir de support lors de
la conception d’une structure similaire.
CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2005:103 III

IV CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2005:103
Contents
ABSTRACT I
CONTENTS V
PREFACE IX
NOTATIONS AND CONVENTIONS X
1 INTRODUCTION 1
1.1 Background 1
1.2 Aims of the thesis and limitations 2
1.3 Outlines 3
2 GLUED LAMINATED TIMBER- GLULAM 4

2.1 Production process 4
2.1.1 Manufacturing 4
2.1.2 Glue types 5
2.1.3 Standard shapes 5
2.1.4 Transportation 5
2.2 Material properties 6
2.2.1 Advantages of Glulam 6
2.2.2 Composed glulam GL32C 6
2.3 Example of current structures 7
2.3.1 Hamar Olympiahall, Norway 7
2.3.2 Håkons Hall, Norway 9
2.3.3 Sørlandshallen and Telemarkshallen, Norway 10
2.3.4 Conclusions 12
3 OPTIMUM STATIC SYSTEM 13

3.1 Theory of arches 13
3.2 Simple arch and trussed arch 15
3.3 Investigated model 16
3.4 Finite Element model 19
3.4.1 General modelling 19
3.4.2 Influence of 2nd order effects 20
3.5 Influence of the boundary conditions 21
3.5.1 Description of the models 21
3.5.2 Normal forces in case of triangular snow load 23
3.5.3 Buckling of the lower chord 29
3.5.4 Bending moment in case of triangular snow load 30
3.5.5 Case of uniform snow load 32
3.6 Influence of the number of diagonal elements 32
3.6.1 Model I1 34
CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2005:103 V

3.6.2 Model I2 36
3.6.3 Comparison between models BC3, I1 and I2 38
3.6.4 Risk of local buckling 39
3.7 Influence of the depth of the truss 40
3.7.1 Model H1 41
3.7.2 Model H2 43
3.7.3 Comparison of models H1, I2 and H2 45
4 PRELIMINARY STRUCTURAL ANALYSIS 47

4.1 Design load 47
4.1.1 Permanent loads 47
4.1.2 Variable loads 47
4.1.3 Load combinations 49
4.2 Finite Element model 50
4.2.1 Geometry 50
4.2.2 Glulam properties 50
4.2.3 Loads 50
4.2.4 Initial imperfections of the geometry 51
4.2.5 FE Analysis 52
4.3 Results 53
4.3.1 Forces and moments 53
4.3.2 Comparison with hand calculations 56
4.3.3 Uplift of the trussed arch 58
4.3.4 Influence of initial imperfections 58
5 CONNECTIONS 60
5.1 Hinges at the abutments and at the crown 60
5.1.1 Hinges at the abutment 60
5.1.2 Hinge at the crown 61
5.2 Possible joining systems in the truss 62
5.2.1 Connection with glued-in rods 62
5.2.2 Tube joint 64
5.2.3 Multiple steel plates connection 65
5.2.4 Choice of the optimum connection 67
5.3 Theory of multiple shear steel-to-timber joints 67
5.3.1 Double shear planes connection 67
5.3.2 Multiple shear planes connection 68
5.4 Design of the connection 71
5.4.1 Optimum number of steel plates 71
5.4.2 Required number of dowels 74
5.4.3 Tension perpendicular to the grain 76
5.4.4 Design of the steel plates 78
5.4.5 Stiffness of the connections 82
VI CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2005:103

6 ANALYSIS OF THE STRUCTURE AND VERIFICATION 85
6.1 Finite Element analysis including connection stiffness 85
6.1.1 FE model 85
6.1.2 Results of the analysis 87
6.2 Verification of the members 90
6.2.1 Chord members 91
6.2.2 Diagonal members 93
6.3 Verification of the connections 94
6.4 Lateral bracing of the lower chord 97
6.5 Partitioning of the arches 98
7 CONCLUSION AND RECOMMENDATIONS 101
8 REFERENCES 103
APPENDIX A: COMPARISON BETWEEN ABAQUS RESULTS AND HAND
CALCULATIONS 104
APPENDIX B: NUMBER OF DOWELS IN THE CONNECTIONS 107
APPENDIX C: CHECKING OF THE STEEL PLATE 112
APPENDIX D: VERIFICATION OF THE MEMBERS 116
CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2005:103 VII

VIII CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2005:103
Preface
This thesis is the culmination of the International Master’s Programme in Structural
Engineering at Chalmers University of Technology, Sweden. It has been carried out
between June and December 2005 at the Division of Structural Engineering,
Department of Civil and Environmental Engineering.
The work was proposed and supervised by Dr. Eng. Roberto Crocetti, designer at
MOELVEN Töreboda, Sweden. It was also overseen by Assistant Professor Marie
Johansson at the department.
This master’s thesis was a great opportunity to learn more about timber structures. It
was especially interesting to complete this thesis in Sweden where timber is a
common building material.
We would like to take the opportunity to express our gratitude towards those involved
in the project:
Roberto Crocetti for his advice and his guidance during the project,
Marie Johansson for her everyday support and all the literature borrowed from her,
Robert Kliger for his general advice,
Julia Meronk and Agnieszka Gilun, our opponent group, for their comments on the
thesis.
This Master Program gave us the chance to get an international experience and to
meet people from all over the world. We would like to thank all these people for the
great time we had together.
Our gratefulness also goes to our parents and relatives for their support during all our
stay in Sweden.
Göteborg December 2005
Audrey Farreyre
Jean-Bernard Journot
CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2005:103 IX

Notations and Conventions
Roman upper case letters
A Area of the cross-section
Cp,e External pressure coefficient
Cp,i Internal pressure coefficient
E0,mean Mean value of modulus of elasticity
Fn Normal force
Fv,Ed Design shear force per shear plane of fastener
G Shear modulus
Ip Polar moment of inertia
Kf Translational stiffness
Kser Slip modulus
Ku Instantaneous slip modulus for ultimate limit state
Kθ Rotational stiffness
LC Load combination
M Moment
My,k Characteristic value for the yield moment
Roman lower case letters
a1 Spacing of dowels within one row parallel to the grain

a2 Spacing of dowels within one row perpendicular to the grain
a3 End distance
a4 Edge distance
d Diameter
e End distance in the steel plate
fc,0,d Design compressive strength along the grain
fh,o,d Design embedment strength along the grain
fh,α,d Design embedment strength at an angle to the grain
fm,d Design bending strength
ft,0,d Design tensile strength perpendicular along the grain
ft,90,k Characteristic tensile strength perpendicular to the grain
fu,k Characteristic tensile strength of bolts
h Depth of the truss
he Embedment depth
kcrit Factor used for lateral buckling
kc,z Instability factor
kmod Modification factor for duration of load and moisture content
lef Buckling length
sk Characteristic snow load
t Thickness
vk Characteristic wind load
X CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2005:103

Greek lower case letters
α Angle
γ Safety coefficient
γM Partial factor for material properties
δ Deflection
ρ Density
σc,0,d Design compressive stress along the grain
σm Design bending stress
σt,0,d Design tensile stress
υ Poison coefficient
φ Diameter
Sign conventions
Forces and stresses
- compression
+ tension
Bending moments
CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2005:103 XI

XII CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2005:103
1 Introduction
1.1 Background
Arches have been built for centuries. This is mainly due to the fact that the only
materials available in the Middle-Ages, apart from timber, were stones and bricks. As
these materials are not able to withstand tensile stresses, it has resulted in a
widespread use of masonry arches and vaults. Cathedrals and old stone bridges are the
most relevant examples of this period.
Nowadays, concrete and steel are common building materials. All types of structures
can be built with these materials. As a consequence, arches are less used. However,
arches should not be underestimated because they are good alternatives to achieve
very long span. Moreover, from an aesthetical point of view, arches are thin and
slender structures.
With the development of the glued laminated timber (Glulam) technology at the
beginning of the twentieth century, arches were again used as roof structure. A well-
known example in Sweden is the central station in Stockholm, see Figure 1.1.
Figure 1.1 Hall of the central station in Stockholm, Carling, Svenskt limträ AB
(2001)
This thesis deals with arches in Glulam. It especially focuses on 100m circular trussed
arches. Such structures are often built to roof large buildings.
This subject was proposed by Roberto Crocetti, who is an engineer at Moelven

Töreboda.
CHALMERS, Structural Engineering and Mechanics, Master’s Thesis 2005:103 1

1.2 Aims of the thesis and limitations
The aim of this thesis is to study the overall behaviour of a timber circular trussed
arch with a large span and to design it. Our analysis focuses on a three-pinned trussed
arch with a 100m span and 20m height. These are common dimensions for a roof
structure of a soccer field or ice-skating rink.
Crown
20m
Abutment Abutment
100m
Figure 1.2 Sketch of a trussed arch
At first, an optimisation of the static system is performed. Several static systems are
studied in order to observe the influence of some relevant parameters such as location
of the hinges, depth of the truss, and number of diagonal elements. Then, particular
attention is paid on the choice of connections. Their design is done in such a way that
their modes of failure are under control. Finally, the design and all the controls of the
structural members are accomplished.
It should be noticed that this thesis only deals with the timber structure. Hence, the
foundations are not considered in this thesis. They are supposed to be in concrete but
no specific calculations have been worked out. The design of the bracing units is also
not investigated. However, a short assessment about the location of the bracing units
is done. Moreover, as the seismic activity is not important in Sweden, the earthquake
resistance of the structure has not been investigated in this thesis
All the calculations are made with the commercial finite element ABAQUS software.
The model is also checked by hand calculations.
The structure is designed according to the European standards. The loads applied on
the structure are calculated with Eurocode 1. The design of the members and the
connections is done according to Eurocode 5. However, other literature was used
when Eurocodes were unclear. The references are given in the chapter 8.
2 CHALMERS, Structural Engineering and Mechanics, Master’s Thesis 2005:103

005:103
1.3 Outlines
This thesis is divided in five parts. Chapter 2 presents Glulam as a building material.
The production process is briefly explained. The standards for Glulam elements are
given and the material properties are introduced. Examples of current buildings
owning trussed arches are thereafter described.
The main purpose of chapter 3 is to find an optimum static system. Different

parameters, which can have an influence on the behaviour of the trussed arch, are
studied. First, several boundary conditions are investigated. The hinges of the
abutments and the crown are either located at the upper chord or the lower chord. The
most efficient system regarding the force path is adopted. Then, the influence of the
number of diagonal members and the effect of the depth of the truss are studied. An
optimum static system results from these studies.
The next step presented in chapter 4 consists in performing a preliminary analysis of

the structure. Several load combinations, calculated according to Eurocode 1, are
applied on the static system from chapter 3. Imperfections are also included in this
model, according to Eurocode 5. Hence, a brief explanation about the FE computation
is given. Finally, forces and moments resulting from the different load combinations
are presented. This study leads to the design forces, which will be used to calculate
the connections.
The chapter 5 especially deals with the connections between the chord elements and
the diagonals. After a presentation of different alternatives, multiple steel plates joints
are adopted. The required number of plates and dowels are determined so that the
failure of the connection is ductile. Then the stiffness of each connection is calculated.
In chapter 6, a more realistic finite element model including the stiffness of the
connections is computed in ABAQUS. The results are presented and compared with
the previous one. Thus, structural members, as well as the connections are verified
according to Eurocode 5 under relevant loads. Finally, the problem of the buckling of
the lower chord and the partitioning of the arch for transportation are investigated.

2 Glued laminated timber- Glulam
Glulam technology was developed in Germany around 1900. It was first used in
Scandinavia around 1920. Among the first structures in Glulam built in Sweden are
the central stations in Stockholm, Göteborg and Malmö.
2.1 Production process

2.1.1 Manufacturing
The way to produce Glulam can change from one country to another, but the main
steps of the production are presented in the Figure 2.1.
Figure 2.1 Glulam production process, extract from Carling, Svenskt limträ AB
(2001)
Most of the time, Glulam is made of spruce. The first step of the production is the
preparation of the planks. They are dried in order to get suitable moisture content for
the glue. Indeed, the moisture content should not exceed 12%. Then, the planks are
pre-planed and strength graded.
To achieve great length, the planks are placed end-to-end and glued together with
finger joints. Then, the laminations are planned again and the glue is applied on the
surface.
The laminations are placed on the top of each other and pressed together. This
operation has to be completed before the glue cures. At this point of the process, if it
is necessary, a curvature can be given to the beam. Finally, the Glulam beam is
planned to remove the surplus of glue so that the surface is smooth.
At this level, additional finishing works like drilling holes are done.

2.1.2 Glue types
Nowadays, two types of glue are used at Moelven Töreboda. The common one is the
melamine. This glue is white but may become darker later. It is increasingly used in
Europe. However, to fulfil the requirements of the Japanese market, another glue is
used at Moelven Töreboda: the phenol-resorcinol-formaldehyde. This glue is black; as
a consequence, finger joints and laminations are visible.
2.1.3 Standard shapes

Different cross-sections are attainable. It mainly depends on the mechanical
equipment of the manufacture. However, rectangular cross-sections are usually used.
The standard depth of a lamination is 45mm. The nominal width varies between a
range of 90mm to 215mm. It is possible to achieve greater width by edge gluing.
To take advantage of the laminations and to exploit timber in an efficient way, it is
also possible to match the laminations quality. It is usual to put laminations with a
higher strenght in the outer part of the cross-sections because high stresses may
appear at this level. This “combined Glulam” is more economical, especially for
important structures.
Figure 2.2 Repartition of the lamination within the cross-section from Carling,
Svenskt limträ AB (2001)
2.1.4 Transportation
Transport can be a capital part of a project, especially when the members have a
considerable size.
Ordinary lorry can transport members up to 9m. Members up to 30m can be
transported as well by means of trailers. However, permissions from the authorities
are necessary and the way followed by the truck has to be decided before. Special
transport is usually required if the width exceeds 2,5m or the total height 4,5m, which
is often the case of frames or arch structures. In this case, the design of the arch has to
take into account that the structure must be divided into a certain number of pieces.

Figure 2.3 Transport requirements, from Carling, Svenskt limträ AB (2001)
2.2 Material properties

2.2.1 Advantages of Glulam
Glulam predominates solid timber in many points. First of all, the strength properties
are better due to the even distribution of the defects, such as knots, within the volume
of the beam. The risk of failure is limited and for a lower quality timber, a higher
strength and stiffness can be reached. In relation to its self–weight, Glulam has a high
strength to weight ratio. As a consequence, Glulam beams can reach large span with a
minimum of support.
The second advantage is the great number of possibilities offered by the production
process. Long beam can be manufactured thanks to the finger joints. Concerning the
beam shapes, I- T- or L-sections as well as cambered beams can be produced. Before
gluing the planks all together, it is always possible to curve the members in order to
build arches for instance.
Figure 2.4 Example of Glulam cross-sections

The control of the moisture content of the planks during the manufacturing process
reduces significantly the risk of distortion and deformation occurring during the
drying process in the construction. Glulam also have a high resistance to fire, which is
very important, especially for public buildings.
2.2.2 Composed glulam GL32C

Glulam members are mainly used in bending. Hence, the stresses are not uniformly
distributed over the cross-section. Therefore, in order to utilize the material in an
efficient way, sections with different laminations quality are used.
In this thesis, the “composed glulam”: GL32C has been chosen. The outer laminations
are made of C35 whereas the inner laminations are made of C27. The repartition of
the laminations will be as it is presented in Figure 2.2. To compare, an homogeneous
beam in GL32 is made only of C35.

2.3 Example of current structures
Some structures similar to the one studied in this thesis already exist. The best
examples of long span trussed arches have been found in Norway. It is probably due
to the abundance of spruces in the Norwegian forest (see Figure 2.5) and the
important number of timber factories. Moreover, two of the examples presented below
were built for the special occasion of the Olympic Games in Lillehammer in 1994.
Figure 2.5 Norwegian forest
2.3.1 Hamar Olympiahall, Norway

This hall has been built for the Olympic games in Lillehammer in 1994. To shape the
arena, the architects were inspired by the “Oselver”, a type of boat that has been built
in Norway for a thousand years.
Figure 2.6 Hamar olympiahall

The roof consists of trussed Glulam arches with different spans. The largest span is
96,4m. In order to give the shape of the boat to the roof, a dorsal arch has been added.
This arch has mainly an aesthetical purpose and is supported by the other arches.

Figure 2.7 Drawings of the arena
Table 2.1: Main information about the Hamar Olympic Hall
Name of the project Hamar Olympic Hall-Viking Ship
Architects NIELS TORP
Total height 37m
Max. span 96,4m
Depth of the truss 3,45m (from center line to center line)
Angle between the diagonals 40º
Dimension of the cross-section
Diagonal members 420x300mm2
Chord members 570x566mm2
Type of connections Multiple steel plates connections
Location of the hinges
Abutment Lower chord
Crown Lower chord

2.3.2 Håkons Hall, Norway
This hall has also been built in Lillehammer for the Olympic games in 1994. The
building has a special shape due to the unsymmetrical arch.
Figure 2.8 Håkons Hall
Figure 2.9 Drawings of the Håkons Hall

Table 2.2 Main information about the Håkons Hall
Name of the project Håkons Hall
Architects Østgaard Arkitkter A.S
Max. span 85,8m
Depth of the truss 3,45m (from center line to center line)
Crown Upper chord
2.3.3 Sørlandshallen and Telemarkshallen, Norway

These two halls are quite similar. They were built as sport halls, to cover soccer fields.
Their roof consists in a trussed Glulam arches with the same shape.
Figure 2.10 Sørlandshallen

80m
Figure 2.11 Drawings of the Telemarkshall
Table 2.3 Main information about the Sørlandshallen and Telemarkshall
Name of the project Sørlandshallen
Architects BIONG &BIONG AS
Total height 24m
Max. span 80m
Crown Lower chord

2.3.4 Conclusions
The four examples presented above give a good illustration of the utility of long span
arch structures.
However, such structures are not spread in other countries due to the higher cost of
timber compared to other building materials. The choice of timber should therefore be
done in the architectural design stage.
From the different examples, it can be observed that many possibilities can be chosen
for the location of the hinges in the structure, the depth of the truss and also the
number of diagonal elements. The influence of these parameters will be studied later
on.

3 Optimum static system
3.1 Theory of arches
When a cable is subjected to a load, it deforms following the funicular shape. The
shape of the cable when it is submitted only to its self-weight is called catenary. It is
often approximated by an ellipse or an arc of circle.
catenary
Figure 3.1 The shape of a rope under different loads

Depending on the length of the cable, the shape changes. All the possible shapes
under a load case are a family of funicular shapes.
Figure 3.2 A family of funicular shape

The forces in the cable are only tensile forces. Let’s now imagine that the cable is up
side down. If the same load is applied, the cable will sustain the same force magnitude
but in compression. In this case, it will be called an arch and will be also a funicular
of this load case.
Figure 3.3 Funicular arch under a concentrated load
The great property of an arch is to be able to transfer the load to the support only with
compressive forces. This characteristic has made the success of the arches in
construction because it allowed the use of stones, which basically cannot carry tension
(see Figure 3.4).
A bending moment appears in the arch when the load is not the same than the one
defined by the funicular shape but this moment does not necessarily imply tensile
stresses if the section is high enough.
CHALMERS, Structural Engineering and Mechanics, Master’s Thesis 2005:103

005:103 13
Figure 3.4 Pont du Gard, France built 2000 years ago under the Roman Empire and
Cathédrale de Chartres, France built in the 12th century.
The arch structures have been decreasingly used with the coming of the modern
building materials like steel and reinforced concrete, which have a tensile capacity.
Today, arches are mainly used in large structures that require a long span with no
support. Moreover, regarding to building roofs, timber is commonly used, as it is light
and easy to shape. The possibilities of Glued Laminated Timber (Glulam) solve the
problem of the limited length imposed by natural dimensions of trees.
Furthermore, the number of hinges of the arch is of great importance. The most
common solutions are either 2-pinned or 3-pinned arches. The 3-pinned arch is
preferable since it is isostatic contrary to the 2-pinned arch. It is then less sensitive to
the eventual support movement.
Figure 3.5 2-pinned and 3-pinned arches

The production of circular arches is more economical when the arches have to be cut
in pieces since each piece is shaped with the same radius of curvature. Even if the
circular shape is not a perfect funicular of a load case, it will often be used. The total
height of the arch is often taken as 0,15 to 0,30 times the span. The horizontal thrust
in the foundation increases when the height of the arch decreases.
Natterer (2000) gives orders of magnitude of the different dimensions that can be used
in the preliminary design.

Figure 3.6 Range of size for arches according to Natterer et al. (2000)
3.2 Simple arch and trussed arch

The design of very large structures is ambitious and the choice of the roof has to be
done wisely since it often represents a large part of the building cost. In that way, this
thesis is an assessment of many possible solutions.
The question of whether to build a simple arch or a trussed arch is coming in the early
stage.
Figure 3.7 Sketches of a simple arch and a trussed arch
In order to compare the structural behaviour of the simple arch and of the trussed
arch, a static analysis is worked out in both cases, under a uniform load of 10kN/m.
The normal forces and the bending moments are depicted in Figures 3.8 and 3.9.
Figure 3.8 Maximum forces and moments in the trussed arch under uniform load of
10kN/m (the repartition is symmetrical)

Figure 3.9 Forces and moments in the simple arch under uniform load of 10kN/m
(the repartition is symmetrical)
This comparison highlights that the forces and especially the bending moments are
larger in the simple arch than in the trussed one. This is not the only reason that makes
the trussed arch more competitive. The construction and assembly process are very
important when considering a 100m span arch. Indeed, it implies that the structure
will be divided in several parts, then transported and finally assembled on the
construction site. It is very hard to achieve a very stiff connection in timber that could
ensure the continuous stiffness of the chord. This problem is less important when
considering a trussed arch since it is not such a problem to have a semi-rigid
connection.
3.3 Investigated model

The purpose of this chapter is to study trussed arches and especially to investigate the
influence of boundary conditions and some relevant geometrical parameters. In order
to go further in details, the main shape of the arch and the cross-section of the
members are decided.
Therefore, the main structure consists in a 100m-span and a 20m-height (see Figure
3.10). The arches are parallel to each other and the distance between them is set at
12m. The upper chord is assumed to be laterally stabilized by the roofing.
20m
100m
Figure 3.10 Illustration of the model studied in this thesis

The model is built with members of two different sections. The sections are taken
using European standard dimensions.
Table 3.1 Sections of the members used in the FE Model.
Chords Diagonal members
3×215mm 3×215mm
270mm
630mm
A=0,406m2 A=0,174m2
All the elements will be made of Glulam GL32 C. In order to simplify the analysis,
isotropic properties will be used to model the behaviour of the timber in the beam
element:
ρ = 400kg/m3
ν = 0,2
Ε0,mean = 13,5GPa
Although the main dimensions of the structure remain the same, the boundary
conditions, the number of diagonal elements and the depth of the truss are thereafter
changed.
These different models are compared under snow load since it is the dominant load in
the case of a long span arch. According to Eurocode 1, two cases should be
considered for the snow load: uniform load and triangle-shaped load (see Figure
3.11). In order to have a better view of the behaviour of the structure, both cases
loaded uniformly and non-uniformly have been investigated.
1,6kN/m2
4,4kN/m2
2,2kN/m2
25m 25m 25m 25m
Figure 3.11 Snow load according to Eurocode 1

005:103 17
The summary of the selection process is shown in Figure 3.12
Model BC1 Model BC2 Model BC1

Study of the influence of
the support conditions
Analysis and evaluation
Model I1 Model I2 Best Model

the number of diagonals
Model H1 Best Model Model H2

the depth of the trussed
arch
Best Model
Figure 3.12 Selection process
At each stage, an evaluation will be done with the following criteria:
- Structural efficiency: the forces and moments are compared for the different
models.
- Stability and risk of buckling: this criterion is hard to compare when the total
load acting on the structure is not known, but discussions are made about that.
- Economy: even if no economical comparison is done in this thesis, this
parameter has been discussed with the supervisor of the thesis and evaluated
according to his experience.
- Production and transportation: limitations are done, so that the solution is
feasible and does not require excessive equipment.

005:103
3.4 Finite Element model
3.4.1 General modelling
In order to study the structural behaviour of the arch, a Finite Element (FE) model is
created using the commercial ABAQUS software. After several tries, it has been
decided to carry out the comparison with the model described below.
The trussed arch is modelled with beam and truss elements. As the chords are
continuous, made of long Glulam members, they are modelled with beam elements.
They are subjected to normal forces and moments. The diagonal elements are
connected to the chords by hinges. Thus, truss elements are used; they carry only
normal forces.
The hinges at the crown and at the abutments are considered as perfect hinges.
Diagonal elements:
Upper/Lower chord:
truss elements (hinges
continuous members
at the 2 ends)
beam elements
H α
i
Figure 3.13 Detail of the links between the different members
The load is applied as a line load along the upper chord of the arch. The magnitude is
determined on each element by a projection of the snow load on the circular shaped
arch.
The size of the meshing is optimised. Chord parts are divided in 10 elements to
describe accurately the moment. Only one element per diagonal is necessary to
compute the normal force since it is a truss element.
Beam elements Crown

(10 / member)
20m
Truss elements
(1 / member)
Abutment Abutment
100m
Figure 3.14 Model created with ABAQUS

005:103 19
3.4.2 Influence of 2nd order effects
The linear elastic theory used to solve a static problem is based on the principle of
superposition: If the displacement of, and stresses at, all points of a structure are
proportional to the loads causing them, then the total displacements and stresses
resulting from the application of several loads will be the sum of the displacements
and stresses caused by these loads when applied separately, extract from Kinney
(1957).
Two assumptions must be correct in order to make this principle valid. Firstly, a linear
relationship between stresses and strains should exist in the range of working stresses.
Secondly, the change of shape in the loaded structure is neglected and the variables
are computed in the original shape.
When a long span timber arch is studied, it appears that the second assumption may
be invalid. At mid span, it is obvious that the arch will deflect of several decimetres.
Therefore, it may be inaccurate to solve the stresses and strains with elastic theory. A
method called deflection theory can be used to consider the effects of large
deflections.
Figure 3.15 Illustration of the deflection theory

Figure 3.15 is an example in which the deflection affects the stresses. The arch is
shown in the original shape and in deformed shape. If the elastic theory is applied, the
moments’ equilibrium at the crown leads to underestimate the horizontal thrust H
since the considered lever arm is h1 instead of h2. This problem is solved by the
iterative method of the deflection theory. The iterations are briefly described below:
1. Computation of the deflection with the elastic theory (using h1)
2. Re-calculation considering the deflection and computation of the new
deflection
… same process until the deflection is considered correct
3. Final computation of the strains, stresses and deflection
This method is available in the FE program ABAQUS.
A comparison has been performed to know whether it basically influences the results
in the case of the trussed arch. The models have been loaded with a uniform load of
40kN/m, which is approximately the magnitude of the design load.

Table 3.2 Influence of the large deflection on the forces and bending moments in a
trussed arch during a static analysis
1st order 2nd order
elastic theory deflection theory
Max. compressive force in the chords -2980kN -2990kN
Max. positive moment in the chords +163kN +166kN
Max. negative moment in the chords -212kN -234kN
Maximum deflection 5,0cm 5,0cm
This evaluation leads to say that the results of the 1st order theory are accurate enough
for the design of the trussed arch. As a consequence, the comparisons between the
different models are thereafter performed with elastic theory (1st order).
3.5 Influence of the boundary conditions

For the sake of this thesis, the results of the two load cases shown in Figure 3.11 are
not all described in this part. Only the results under unsymmetrical snow loading are
depicted since the difference between the models is accentuated.
First, a comparison between the different models is accomplished regarding the
normal forces and the bending moments. The study of normal forces illustrates the
force path in the structure (Section 3.5.2). Then the moment distributions in the
chords are studied (Section 3.5.4). Finally, the summary of the analysis is presented.
3.5.1 Description of the models

The position of the hinges at the abutments and at the crown basically changes the
behaviour of a trussed arch. In order to compare the different solutions, a static
analysis is performed to figure out which model is structurally the best. Three models
are studied.

Table 3.3 Three models used to study the influence of the boundary conditions.
3,9m
20m
Model 3,5m
BC1
100m
3,9m
20m 3,5m
Model
BC2
100m
3,9m
20m 3,5m
Model
BC3
100m
The distance between the two chords is set to 3,5m. The truss diagonals are built as
equilateral triangles. The elements in the chords have a length of 3,9m.

005:103
3.5.2 Normal forces in case of triangular snow load
• Model BC1
52,8kN/m
26,4kN/m
Total load = 1968kN
Figure 3.16 Applied load on model BC1
1148kN
820kN
δmax=62mm
1229kN 1229kN
Figure 3.17 Deformed shape and reaction forces – Model BC1
highest compression
Figure 3.18 Compressed members – Model BC1

005:103 23
highest tension
Figure 3.19 Tensioned members – Model BC1
-540 -1330 -1165 -957

-2537 -2288
-2246 -911
+37 +20 -125
-1705 +1123 -333 -1330
+790 -499
+41
+37 -208 +240
Figure 3.20 Normal forces in the members (kN) – Model BC1

• Model BC2
52,8kN/m
26,4kN/m
Total load = 1977kN
1160kN 817kN
δmax=50mm
1185kN
1185kN
highest compression

:103 25
highest tension
+144
-196 +321
-1139
+426
-839
-74 -1080 +204

-106 -1257
-1519
-1740
-755
+254 -1818
-78 -1620
-1528 +215

• Model BC3
52,8kN/m
26,4kN/m
Total load = 1968kN
1163kN
806kN
δmax=44mm
946kN 946kN
highest compression

:103 27
highest tension
-1048 -877
-1830 -569
-1901
-208
-1485
-208 +203 -141

-453 +58
+898
-981
+79
-266 -403 +37
-1352 -1260
+362
• Comparison and discussions
Table 3.4 Comparison of models BC1, BC2 and BC3
Model BC1 Model BC2 Model BC3
Horizontal thrust 1229kN 1185kN 946kN
Max. compressive -2537kN -1818kN -1901kN

force
Max. tensile force +1123kN +254kN +898kN
Max. displacement 62mm 50mm 44mm
The horizontal thrust is an important parameter for the foundations of the arch. As it is
said in section 3.1, it depends on the ratio height/span of the arch. The model BC3 is
the best regarding this criterion.

In the different models, the force path through the trussed arch is very different from
one model to another. In model BC1, the upper chord is highly compressed but the
lower chord is not subjected to important forces. At the contrary, in model BC2, the
lower chord is very compressed whereas the lower chord is moderately loaded. Model
BC3 has a different load path and works like if the load is transferred from the upper
chord to the lower chord. Therefore, the maximum force in compression is lower than
in the other models.
3.5.3 Buckling of the lower chord

This study is based on the assumption that the upper chord is laterally stabilized by
the roof. In this way, the buckling of the upper chord is prevented but the problem of
buckling remains important in the lower chord.
Figure 3.31 Illustration of the buckling in the lower chord

Regarding this issue, models BC1 and BC3 are better. The risk of buckling in the
lower chord is significantly reduced since the lower chord is partly tensioned.
Moreover, it may be unnecessary to have bracing in such models. A striking example
is the Håkons hall in Lillehammer, which does not have any lateral bracing in the
lower chord.
At the contrary, the model BC2 is obviously the worst case because its lower chord is
highly compressed. Consequently, the buckling may happen very easily. If this model
is used, lateral bracings will certainly be needed on the lower chord. Such bracings
have been used in the Hamar’s Olympiahall where the arches are built on the model
BC2.
Lateral bracing units
Figure 3.32 Hamar Olympiahall – notice the use of lateral bracings

005:103 29
Figure 3.33 Håkons hall, Lillehammer – Notice the absence of lateral bracing in the
lower chord
Model BC2 will be skipped in the following comparison because of the buckling
problem in the lower chord. The comparison is now accomplished between model
BC1 and BC3 regarding the bending moment.
3.5.4 Bending moment in case of triangular snow load

• Model BC1
Mmin= -13kNm
Mmax= +70kNm
Figure 3.34 Bending moment in the upper chord – Model BC1
Mmax= +45kNm
Figure 3.35 Bending moment in the lower chord – Model BC1

• Model BC3
Bending moment in N.m
Mmin= -25kNm
Mmax= +48kNm
X-distance in m
Figure 3.36 Bending moment in the upper chord – Model BC3
Mmin= -30kNm
Mmax= +45kNm
X-distance in m
Figure 3.37 Bending moment in the lower chord – Model BC3
The bending moment distribution in the upper chord may be compared to the one of a
continuous beam. It works like the continuous chord is supported at the nodes. Thus,
extremum moments appear either between two nodes or at the nodes. This
phenomenon is not observed on the lower chord of BC1 and in this case, the bending
moment looks like the one of a simple arch.
Moreover, the bending moment magnitude is the same in the lower and upper chord
of model BC3. However, the upper chord of model BC1 is much more subjected to
moment than the lower chord. Thus, the maximum positive moment in a node of the
quarter arch reaches 70kNm.
To conclude, this comparison shows that model BC3 is more favourable.

:103 31
3.5.5 Case of uniform snow load
The previous study, in case of a triangular snow load leads to say that model BC3 is
preferable. However, it is necessary to understand the behaviour of the structure under
the uniform snow load to ensure that it is really the optimum model.
Table 3.5 Normal forces and bending moments in the case of uniform snow load
Model BC1 Model BC3

Maximum FORCES
chords -1470kN No tension -1456kN No tension
diagonals -132kN +132kN -177kN +130kN
Maximum MOMENTS
upper chord -8kNm +30kNm -30kNm +18kNm
lower chord 0 +15kNm -30kNm +15kNm
The magnitude of the normal forces is the same in the two models and the distribution
of the bending moments is comparable to the case of triangular loading.
As a conclusion, it is still valid to assert that model BC3 is the optimum solution and
these boundary conditions will be kept afterwards.
3.6 Influence of the number of diagonal elements

After finding the optimum boundary conditions for the trussed arch, a study is
performed in order to find the optimum number of diagonal elements in the trussed
arch. As in Section 3.5, the comparison is done under the two cases of snow load
described in Eurocode 1.
At the first stage, the number of diagonal elements has been taken so that they form
approximately equilateral triangles. This configuration requires a large number of
diagonal members, which seems to be too conservative and consequently, too
expensive. In order to see the influence of the number of diagonal elements on the
overall behaviour of the structure, two more models have been created and compared
to the previous model BC3.

H α
i
Figure 3.38 Geometry of the truss
Table 3.6 Three models used to study the influence of the number of diagonal
elements.
α~60°
Model BC3 i ~ 4m
27 diagonals/half-
arch
α~45°
Model I1 i ~ 7m
15 diagonals/half-
arch
α~40°
Model I2 i ~ 8m
13 diagonals/half-
arch
Only the results of the calculations about models I1 and I2 will be shown. The results
of model BC3 can be found in the previous section 3.5.

005:103 33
3.6.1 Model I1
Table 3.7 Reaction forces and deformed shape in model I1
H = 1248kN
Uniform load
Vleft = 982kN
Vright= 982kN
H = 1249kN
Triangular load
Vleft = 1212kN
Vright= 862kN
Table 3.8 Maximum forces in the chords and in the diagonals in model I1
154kN
Uniform load
-285kN
-1480kN
-1480kN
-2224kN
Triangular load
506kN
919kN
-563kN

005:103
Table 3.9 Bending moment in the chords in model I1
Upper chord:
Mmin= -63kNm
Mmax=50kNm
Uniform load
Lower chord:
Mmin= -82kNm
Mmax= 47kNm
Upper chord:
Mmin= -51kNm
Mmax= 80kNm
Triangular load
Lower chord:
Mmin= -85kNm
Mmax= 59kNm

005:103 35
3.6.2 Model I2
Table 3.10 Reaction forces and deformed shape in model I2
H = 1258kN
Uniform load
Vleft = 1009kN
Vright= 1009kN
H = 1230kN
Triangular load
Vleft = 1194kN
Vright= 843kN
Table 3.11 Maximum forces in the chords and in the diagonals in model I2
+145kN
Uniform load
-352kN
-1457kN -1457kN
-2228kN
Triangular load
-620kN
+875kN
+541kN

005:103
Table 3.12 Bending moment in the chords in model I2
Upper chord:
Mmin= -89kNm
Mmax= 82kNm
Uniform load
Lower chord:
Mmin= -107kNm
Mmax= 65kNm
Upper chord:
Mmin= -64kNm
Mmax=92kNm
Triangular load
Lower chord:
Mmin=-108kNm
Mmax=67kNm

005:103 37
3.6.3 Comparison between models BC3, I1 and I2
Table 3.13 Comparison of the forces and moments between models BC3, I1 and I2
Model BC3 Model I1 Model I2

Total vertical
forces (kN) 1868 1965 2018
Horizontal thrust
Uniform Loading
(kN) 1179 1248 1258
FORCES (kN)
-in the diagonals -177 +130 -285 +154 -352 +145
-in the chords -1456 -1480 -1457
MOMENTS(kNm)
-in the upper chord -30 +18 -63 +50 -89 +82
-in the lower chord -30 +15 -82 +47 -107 +65
Total vertical
forces(kN) 1968 2074 2037
Horizontal thrust
Triangular Loading
(kN) 946 1249 1230
FORCES (kN)
-in the diagonals -403 +362 -563 +506 -620 +541
-in the chords -1901 +898 -2224 +919 -2228 +875
MOMENTS(kNm)
Before comparing the values of stresses, it has to be noticed that the models I1 and I2
are slightly more loaded than the model BC3. This difference happened because of the
discretization of the load over the span.
In the three models, the force path is almost similar. However, the less diagonals
elements there are, the more important are the forces in these members. At the
contrary, the forces in the chords are changing only slightly.
The moment in the elements increases with the length. However, the maximum and
minimum moments are not located in the same element of the chord from one model
to another.
The results correspond to what we could expect: when the number of diagonal
elements is reduced, the forces in the diagonals increase, and the bending moments in
the chords increase too. However, it also shows that the normal forces in the chords
are not very dependent on the number of diagonals.
The results of this comparison are not sufficient to decide which model is the
optimum. The local buckling of the chord and of the diagonal elements has not been

taken into account so far. It will be important to know if models I1 and I2 present a
risk of buckling in order to make the final choice.
3.6.4 Risk of local buckling

The decision has been taken to check the buckling in the worst case (model I2 under
triangular snow load) in the most loaded element. However, at this point of the study,
the final design normal forces and bending moments are not known. The values found
in Table 3.13 are only values under the snow loads. As the design values are given by
the load combination: Qd = 1,35 ⋅ Gself −weight + 1,5 ⋅ Qsnow , it is reasonable to assume that
Qd ≈ 2 ⋅ Qsnow . Hence, it has been chosen to check the buckling under forces and
moments equal to two times the values from Table 3.13.
Thus, the local buckling has been checked for a chord section (as shown in Table 3.1)
with a buckling length of 8m. This chord element was subjected to a normal force of
2600kN and a bending moment of 216kNm. This checking has been performed
according to Eurocode 5.
Most loaded element
N=2600kNm
M=216kNm
Figure 3.39 Illustration of the buckling estimation

Finally it results from the calculations that no buckling is likely to occur. It is not
possible at this point to ensure that the model I2 is safe from local buckling, but it can
be asserted that the risk is low.
Because of this fact, the decision has been taken, with Mr. Crocetti the supervisor of
this project, to keep the model I2 for the following studies. This decision is also
influenced by the fact that the trussed arches in Hamar and Lillehammer were built
with diagonals at an angle ~ 40º. This model is more economical than the others since
it saves material and decreases the number of connections.

005:103 39
3.7 Influence of the depth of the truss
The last step consists in studying the influence of the truss depth. The features of the
arch selected before are kept: boundary conditions and number of diagonals. In order
to have an idea of the influence of the depth of the truss, two more models have been
created and compared under the two snow load cases described in the previous
section.
H 40°
i
Figure 3.40 Geometry of the truss
Table 3.14 Three models used to study the influence of the truss depth
H=3m
Model
H1 15 diagonals/half- i=7,2m
arch
H=3,5m
Model
I2 13 diagonals/half- i= 8m
arch
H=4m
Model
H2 11 diagonals/half- i=9,6m
arch

005:103
3.7.1 Model H1
Table 3.15 Reaction forces and deformed shape in model H1
H = 1281kN
Uniform load
Vleft = 1004kN
Vright= 1004kN
H = 1216kN
Triangular load
Vleft = 1179kN
Vright= 833kN
Table 3.16 Maximum forces in the chords and in the diagonals in model H1
Uniform load
-278kN
+133kN
-1537kN -1537kN
-2363kN
Triangular load
+505kN
+1077kN
-551kN

005:103 41
Table 3.17 Bending moment in the chords in model H1
Upper chord:
Mmin=-71kNm
Mmax=52kNm
Uniform load
Lower chord:
Mmin=-86kNm
Mmax=45kNm
Upper chord:
Mmin=-54kNm
Mmax=80kNm
Triangular load
Lower chord:
Mmin=-84kNm
Mmax=70kNm

005:103
3.7.2 Model H2
Table 3.18 Reaction forces and deformed shape in model H2
H = 1309kN
Uniform load
Vleft=1004kN
Vright=1004kN
Triangular loading
H = 1198kN
Vleft=1167kN
Vright=823kN
Table 3.19 Maximum forces in the chords and in the diagonals in model H2
Uniform load
-396kN
+176kN
-1481kN
-1985kN
Triangular loading
+507kN
+730kN
-666kN

005:103 43
Table 3.20 Bending moment in the chords in model H2
Upper chord:
Mmin= -105kNm
Mmax= 105kNm
Uniform load
Lower chord:
Mmin= -140kNm
Mmax= 70kNm
Upper chord:
Mmin= -85kNm
Mmax= 120kNm
Triangular load
Lower chord:
Mmin= -125kNm
Mmax= +80kNm

3.7.3 Comparison of models H1, I2 and H2
Table 3.21 Comparison of the models H1, I2 and H2
Model H1 – H=3m Model I2– H=3,5m Model H2– H=4m

Total vertical forces
(kN) 2008 2018 2008
Horizontal thrust
(kN) 1281 1258 1309
Uniform Loading
FORCES (kN)
-in the diagonals -278 +133 -352 +145 -396 +176
-in the chords -1537 -1457 -1481
MOMENTS (kNm)
Total vertical forces

(kN) 2012 2037 1990
Horizontal thrust
Triangular Loading
(kN) 1216 1230 1198
FORCES (kN)
-in the diagonals -551 +505 -620 +541 -666 +507
-in the chords -2363 +1077 -2228 +875 -1985 +730
MOMENTS (kNm)
Before comparing the models, it has to be noticed that the model I2 is slightly more
loaded than model H1 and model H2.
As it was expected, the normal forces in the chords are reduced when the depth of the
truss increases, but the difference is not so high. At the contrary, the bending moments
in the chords increase since the length of the members is more important. As a result,
the chord is subjected to smaller forces but larger moment. The reduction of the
normal force is not so important and this fact leads to say that it is not relevant to have
a very large depth like model H2. The problem of buckling is still kept in mind to
make this choice. When the members are longer, this risk increases. The
transportation requirements have to be considered as well.
Finally, I2 is chosen as the optimum system and kept for the rest of the thesis. This
choice has been taken together with the experienced supervisor of this project.
As a conclusion of paragraph 3, Figure 3.41 summarizes the process followed to
choose the best trussed arch system.

Model BC2 Model BC3
Model BC1
High risk of instability in the lower

chord, lateral bracings needed.
Evaluation
Model BC3 is structurally better than Model BC1: Maximum Forces

and moments are lower in BC3
Model BC3 Model I1 Model I2

α = 60º α = 45º α = 40º
Too many diagonals

Too expensive
Evaluation
Risk of local buckling of the members: I1 and I2 should not be

affected by the local buckling according to preliminary calculations
Model H1 Model I2 Model H2

H = 3m H = 3,5m H = 4m
Evaluation
Forces decrease when the truss is higher but moments increase in

the members since they are longer
When the depth of the truss increases, more material is required
Model I2 is optimum
Figure 3.41 Conclusion of the optimum solution selection

4 PRELIMINARY STRUCTURAL ANALYSIS
In this chapter, a preliminary analysis of the model selected in chapter 3 is performed.
Several load combinations will be studied and the resulting design forces will be
worked out. These forces will be used in the next chapter 5 to design the connections
between the chords and the diagonals.
4.1 Design load

In order to design the structure, a survey of the actions that may be applied to the
construction has to be done. The trussed arch will be subjected to two types of load:
the permanent actions and the variable actions, which are calculated according to
Eurocode 1.
4.1.1 Permanent loads

The permanent actions are due to the self-weight of the structural members and the
roofing. According to Eurocode 1, the density of the Glulam GL32C is taken equal to
400kg/m3. The load due to the roofing is set to 0,8kN/m2.
4.1.2 Variable loads

4.1.2.1 Snow load
The snow load applied on the structure is calculated according to Eurocode 1, part1-
3(2003). Two load arrangements have to be considered depending on the distribution
of the snow on the roof. The assumed snow load is sk=2,0kN/m2, which is a common
value for Sweden. For instance, it can correspond to a building located in Stockholm.
Then, for a ratio h/b=0,2 in the arch, µ3=2.
Figure 4.1 Snow load distribution from Eurocode 1

The final values are presented in the Figure 4.2 below. It should be borne in mind that
the distance between the arches is 12m.
Figure 4.2 Snow load distribution
4.1.2.2 Wind load

The wind load applied on the structure is calculated according to Eurocode 1, part1-4
(2004). The assumed wind load is vk= 0,7kN/m2. The distribution of the external
pressure is non-uniform. Hence, the roof is subjected either to suction or to pressure.
The internal pressure coefficient Cp,i is either taken to +0,2 or -0,3 depending on the
worst case. Thus, cumulating the internal and the external pressure coefficient, it
results into two wind distributions.
Wind
Figure 4.3 Wind load distribution W1, with Cp,i= +0,2

005:103
Wind
Figure 4.4 Wind load distribution W2, with Cp,i= +0,2
4.1.3 Load combinations

Several load arrangements have to be considered in order to get the design forces in
the structure.
Considering the snow load dominant, two load cases are studied:
LC1 : 1,35 ⋅ G + 1,5 ⋅ S1

LC 2 : 1,35 ⋅ G + 1,5 ⋅ S 2
Then, considering the wind load dominant, two other cases have to be carried out.
From these load cases, the uplifting of the structure will be checked.
LC 3 : 1,0 ⋅ G + 1,5 ⋅ W1
LC 4 : 1,0 ⋅ G + 1,5 ⋅ W2
Finally, the snow load and the wind load are combined, taking the snow load as the
dominant load:
LC 5 : 1,35 ⋅ G + 1,5 ⋅ S1 + 0,6 ⋅ 1,5 ⋅ W1

LC 6 : 1,35 ⋅ G + 1,5 ⋅ S 2 + 0,6 ⋅ 1,5 ⋅ W2
Thus, the design load will result from those six different cases.

4.2 Finite Element model
4.2.1 Geometry
The geometry of the trussed arch results from the choice made in chapter 3.
Crown
20m
Abutment Abutment
100m
Figure 4.5 Geometry of the trussed arch
26 diagonals are used and the depth of the truss is taken equal to 3,5m. The arches are
spaced by 12m.
4.2.2 Glulam properties

The Glulam is assumed to be a homogeneous, isotropic, elastic material. Although the
assumption is not realistic, this choice has been done in order to simplify the analysis.
Thus, the characteristics for GL32C set in ABAQUS are:
ρ= 400kg/m3
E0,mean=13,5GPa
ν=0,2
4.2.3 Loads
The structure is studied under six different load cases. Each case is carried out in a
file. The self-weight of the structure is modelled in ABAQUS by using the load type:
gravity. The self-weight of the roof is applied on the top chord of the arch as a line
load. The snow load and the wind load are modelled in ABAQUS by means of the
line load type. The wind load is applied perpendicular to the members whereas the
snow load remains vertical.

005:103
4.2.4 Initial imperfections of the geometry
At the production stage, irregularities in the geometry of the members can appear. As
the structure is slender, the internal extra forces and moments due to initial
imperfections should be taken into account. The imperfect shape of the structure
should be assumed to be similar to the deformed shape under a symmetric load or
unsymmetrical load. Eurocode 5 proposes magnitudes for the imperfections, which
should be implemented to the two-pinned arch, see Figure 4.6.
Figure 4.6 Assumed initial deviations in the geometry of a two-pinned arch

corresponding to a symmetrical load and unsymmetrical. (Eurocode5)
In the studied case of the 3-pinned arch, the imperfect shape should be different from
Figure 4.6. This imperfect shape is determined in the way described below.
First, a buckling analysis of the “perfect” model of the structure is carried out. Some
of the in-plane buckling modes are similar to the assumed initial deviation of the arch
under a symmetrical load and a non-symmetrical load. From this analysis, the lengths
l1 and l2 (see Figure 4.6) are found and thus, the values of the imperfection at some
points are determined. Then, a scale factor associated to each mode is calculated in
order to get a suitable imperfection magnitude.
These imperfections are set in the FE model in order to compute the static analysis.
0,075 m
0,15 m
Figure 4.7 Initial imperfections applied to the model in case of symmetrical loading

005:103 51
0,1 m
0,1 m
Figure 4.8 Initial imperfections applied to the model in case of unsymmetrical

loading
Depending on the load case, the load distribution can be symmetrical or

unsymmetrical. In case of symmetrical loading (LC1), initial imperfections are
defined like Figure 4.7. For an unsymmetrical load pattern (LC2, 3, 4, 5, 6), initial
imperfections are similar to Figure 4.8.
4.2.5 FE Analysis
To accomplish an analysis of the structure, including the imperfections, the only way
proposed by ABAQUS is to perform a static “Riks” analysis. This analysis is a load-
displacement analysis, which consists in incrementing the load as well as the
displacements and realising the equilibrium for each increment. Only the results from
the step, where the incremental load factor is equal to one, are used. In this step, the
total load applied on the structure is equal to the design load.
Moreover, this analysis is done taking into account the 2nd order effect of large
displacements. This choice has been done because larger deflections are expected.
To summarize, this FE-analysis is performed taking into account:
- initial imperfections of the structure as described in section 4.2.4
- 2nd order effect of large displacements by solving the problem with the
iterative method presented in section 3.4.2

4.3 Results
4.3.1 Forces and moments
First and foremost, let’s consider only the diagonal members. The following table
present the maximum tensile and compressive forces for the different load cases.
Table 4.1 Maximum force in the diagonals
Max. compressive Max tensile force

force
LC1 -809kN +388kN
LC2 -1188kN +934kN
LC3 -325kN +284kN
LC4 -400kN +332kN
LC5 -1190kN +954kN
LC6 -1260kN +984kN
The forces in the structure are completely different from one load case to another. The
location of the maximum forces is also not the same. However, it can be observed that
the maximum compression always appears in the members near the support whereas
the location of the maximum tensile force depends on the load arrangement, see
Figure 4.9. The load case 6 is determinant for the forces in the diagonal members, as it
gives the biggest forces.
Location of the maximum

compressive force
LC1, LC2, LC5, LC6
LC3, LC4
LC1
LC1 Location of the maximum

LC2, LC5, LC6 tensile force
LC3, LC4
Figure 4.9 Location of the maximum force in the diagonals depending on the load
case

005:103 53
Considering now the chord elements, the maximum tensile and compressive forces
are shown in Table 4.2.
Table 4.2 Maximum forces in the chord
Max. compressive Max. tensile force

force
LC1 -3783kN No Tension
LC2 -4168kN +1116kN
LC3 -703kN +540kN
LC4 -1000kN +343kN
LC5 -4029kN +1362kN
LC6 -4265kN +1315kN
The forces in the chords are higher than the forces in the diagonals. The load case 6
gives the maximum compressive force, whereas the maximum tensile force appears in
the load combination LC5.

LC4 LC2, LC5, LC6 compressive force
LC3
LC1
LC2
LC5, LC6
LC3, LC4
tensile force
Figure 4.10 Location of the maximum forces in the chord depending on the load case

005:103
As the chord is assumed to be continuous, the moments have to be considered. These
moments induce additional stresses.
Table 4.3 Maximum and minimum moment in the chord
Maximum Maximum
negative moment positive moment
LC1 -192kNm +299kNm
LC2 -171kNm +266kNm
LC3 -50kNm +39kNm
LC4 -77kNm +50kNm
LC5 -176kNm +226kNm
LC6 -184 kNm +248kNm
The distributions of the bending moment are similar to the ones presented in section
3.6.2 depending to which snow load is considered (uniform or triangular shape).
Location of the maximum positive

LC3
LC5, LC6 moments
LC1
LC4 LC2
LC3
LC4
Location of the maximum negative
moments
LC1, LC2, LC5, LC6
Figure 4.11 Location of the maximum moments depending on the load case

005:103 55
4.3.2 Comparison with hand calculations
In order to verify the model in ABAQUS, hand calculations have been done under
load combination LC1 because it is a symmetrical loading.
Let’s check first the reaction forces. Regarding to the permanent loads, the self-weight
of the structure and the roof can easily be approximated. The details of the
calculations can be found in the Appendix A.
G selfweight = G structure + Groof = 445,8 + 1070 = 1515,8kN
The variable action, which is considered for LC1, is the uniform snow load.
Qsnow = 19,2 ⋅ l span = 19,2 ⋅ 100 = 1920kN
Design load:
Qd = 1,35 ⋅ G se lfweight + 1,5 ⋅ Qsnow = 1,35 ⋅ 1515,8 + 1,5 ⋅ 1920 = 4926.3kN
As the system is symmetrical, it is reasonable to assume that the reaction forces are
equal to half of the load:
4926,3
Rleft = Rright = = 2463kN
2
Qd ⋅ l
The horizontal thrust is given by: H = , where f is the height of the arch.
8⋅ f
4926,3 ⋅ 100
H= = 3078kN
8 ⋅ 20
The reaction forces given by ABAQUS are: Rleft=Rright= 2520kN and H=3211kN. The
difference is mainly due to the incremental method used by ABAQUS to perform the
equilibrium.
The model is also checked by making the equilibrium in the two first nodes of the
trussed arch.
Figure 4.12 Location of node 1 and note 2

005:103
The calculations can be found in the Appendix A. The comparison with the results
from ABAQUS is done in Tables 4.4 and 4.5.
Equilibrium at node 1: Equilibrium at node 2:
FA
75,88º FB
FC
188kN 31,86º
34,7º
2
1
HA=3079kN FD 6,5º
RA=2463kN
14,12º
FA= 414kN
Table 4.4 Comparison with ABAQUS results
NODE 1 Hand calculation ABAQUS Error
FA 414kN 378kN 9%
FB 3620kN 3783kN 4%
RA 2463kN 2520kN 2%
HA 3078kN 3211kN 4%
Table 4.5 Comparison with ABAQUS results
NODE 2 Hand calculation ABAQUS Error
FA 414kN 378kN 9%
FC 360kN 428kN 18%
FD 206kN 210kN 2%
The error between the hand calculations and ABAQUS results is quite small.
The hand calculations are done with a truss model with concentrated forces applied on
the nodes of the truss. As a result, the bending moments are not taken into account. It
is the main reason of the error.
It is reasonable to assume that the model in ABAQUS is realistic.

005:103 57
4.3.3 Uplift of the trussed arch
The wind load is taken as the dominant load for both cases LC3 and LC4. These
combinations are decisive for the uplift of the trussed arch. Using ABAQUS, the
uplift is easily checked by considering the Y-component of reaction forces, see Table
4.6.
Table 4.6 Reaction forces at the abutments
REACTION FORCES
Left support Right support
LC3 356kN 219kN
LC4 666kN 529kN
The Y-component of the reaction forces is positive (directed upward) in the both
cases, which means that there is no risk of uplift.
4.3.4 Influence of initial imperfections

This section presents the difference of the results between the perfect model and the
model including the imperfections in the geometry. The influence of the initial
imperfections has been studied in the cases where the structure is highly loaded i.e.
LC1 and LC2. The results are presented in the Table 4.7 and Table 4.8.
Table 4.7 Load combination LC1
Without
With imperfection Difference
imperfection
Max compressive
-3569kN -3783kN 5%
force in the chord
Max tensile force

+112kN NO TENSION -
in the chord
Max moment +274kNm +298kNm 8%
Min moment -174kNm -192kNm 1%

Table 4.8 Load combination LC2
Without
With imperfection Difference
imperfection
Max compressive
-4089kN -4168kN 1%
force in the chord
Max tensile force

+1048kN +1117kN 6%
in the chord
Max moment +259kNm +266kNm 2%
Min moment -165kNm -172kNm 4%
When the imperfections are introduced in the model, the forces and the bending
moments increase. The rise is rather low and stays below 10%.
For the load combination LC1, the imperfections add compression forces in the lower
chord, which cancel the tensile forces.

5 Connections
In this chapter, possible joining systems for the abutments, the crown and the truss are
presented. The design of the connection in the truss is performed with the results of
the computation done in chapter 4.
5.1 Hinges at the abutments and at the crown

5.1.1 Hinges at the abutment
At the abutment, the connection should be designed to behave almost like a perfect
hinge. For arches with a small span, it is more economical to use imperfect hinge
since the moments and the support forces are not so large. In this case, external
fishplates, nail plates or flat steel are used, see Figure 5.1.
Figure 5.1 Springing point of arch with flat steel and screws/bolts. Principles.
Carling, Svenskt limträ AB (2001)
For middle span arches, it is of main importance to achieve a system, which behaves
as the intended static system since the forces are important. The more common
solution is a welded support fixture with hinged connection, see Figure 5.2.
Figure 5.2 Welded support fixture with hinged connections. Principles. Carling,
Svenskt limträ AB (2001)
Axial and normal forces from the arch are transferred by contact pressure to the steel
shoe and then through the hinge down to the concrete foundation.

In the case of very long span arches, embedded steel plates in the chord and dowels
are used to fix the hinge. This type of connection has been used for the Håkons Hall in
Lillehammer. It is presented in the Figures 5.3.
Figure 5.3 Abutment connections, Håkons Hall in Lillehammer, Carling, Svenskt

limträ AB (2001)
5.1.2 Hinge at the crown

At the crown, where the two parts of the arch are joined, the third hinge is achieved.
The two parts end with a rectangular steel box, which can fit into each other, and are
bolted together. The steel boxes are fixed to the chord by internal steel plates fastened
with dowels, see Figure 5.5.
Figure 5.4 Example of connection at the crown in the lower chord, MOELVEN
brochure
Figure 5.5 Example of connection at the crown in the upper chord

5.2 Possible joining systems in the truss
The solution chosen to achieve the connections between the diagonals and the chord is
of the main importance for the whole structure. It will affect the structure in different
ways:
- the strength : The connection is often the weakest point of the structure. The
joints have to transfer the forces between the different members. Moreover, it
should not induce a large tension perpendicular to the grain because of the
high risk of splitting in timber.
- the stiffness : For a long span arch, the deformation is often critical in the
serviceability limit state. The stiffness of the connection will significantly
affect the stiffness of the whole structure.
- the ductility : The risk of collapse mainly depends on the ability of the
structure to redistribute the forces. This redistribution is only possible when
the members have enough ductility. In this way, it is preferable to have a
ductile failure of the connection.
With the help of these 3 criteria, the system of connection has to be chosen,
considering also economical and aesthetic requirements.
The chosen solution has to be reliable and tested before. Indeed, it seems improbable
that the designer will carry full-scale tests before erecting the arch. It is better to rely
on well-known techniques when an ambitious 100m span arch is to be built.
After a literature survey, it has been found that only few types of joints are able to
carry important forces as in the case of the trussed arch. Hence, 3 connections capable
to fulfil the “strength requirement” are presented.
5.2.1 Connection with glued-in rods

Glued-in rods connectors were apparently studied first at Chalmers, 20 years ago
according to Madsen (2000). This assembly technique uses steel rods glued in the
timber members. It can carry important tensile or compressive loads and can also be
used to achieve moment resisting connections. Until now, different solutions have
been studied and tested but this joint has not been introduced in the codes yet.
Some examples of connections using glued-in rods are shown in the next Figures.

Figure 5.6 Beam splice joint with glued-
in rods, Madsen (2000)
Figure 5.9 Moment resisting joint for a

beam to column connection, Madsen (2000)
Figure 5.7 Column to foundation joint

with glued-in rods, Madsen (2000)
Figure 5.10 Knee joint with outer plates

Figure 5.8 Knee joint with steel plates and rods at an angle to the grain, Madsen
and rods parallel to the grain, Madsen (2000)
(2000)

No example of truss connection using glued-in rods has been found in the literature.
One could imagine a solution for the connections in the trussed arch. However, it is
very hard to rely on it since it hasn’t been tested. A sketch is proposed in Figure 5.11.
rods
Steel plates
Figure 5.11 Sketch of a proposed solution for a joint using glued-in rods
The design of this joint should be made so that the failure would occur by yielding of
the rods. In this case, the failure would be ductile and the connection more reliable.
One of the main drawbacks of this joint is the production. The holes in the members
have to be drilled perfectly in order to have a correct joint. This perfect alignment
needs to be performed with special equipment.
5.2.2 Tube joint

The tube joint is a new technique developed in the 90’s by Dr. Leijtens in Delft
University. This connection is achieved by using steel tubes and it is associated with
densified veneer wood as shown in Figure 5.12.
Figure 5.12 Assembly process of a tube joint from Madsen (2000)

005:103
Figure 5.13 Ductile failure of the tube joint from Madsen (2000)
This connection can carry very important loads. It has to be designed to have a ductile
failure. The tube joint is often used to achieve moment resisting connections.
A design method is proposed by Leijten in Madsen (2000). However, no reference to

this kind of joint is done in Eurocode 5.
In the case of the trussed arch, the use of such a joint implies that the diagonals are
split in 2 parts. This is not favourable for the stability of the members and the risk of
local buckling is more important.
Figure 5.14 Example of truss connection with a tube joint
5.2.3 Multiple steel plates connection

The multiple steel plates connection with dowel fasteners has been developed in the
40’s and has been successfully applied to many structures. It is also called BSB
system (Blumer System Binder). This joint uses steel plates, which are slotted into
timber and connected with dowels, as depicted in Figure 5.15.
It is compact and ideal from an aesthetic point of view since all the mechanical parts
are invisible from outside. The ductility of the joint can be chosen in design and a
good stiffness can be obtained. To achieve this goal, the designer should find an
optimum for the distance between the steel plates, the dowel spacing and end-
distances.

Figure 5.15 Detail of multiple shear steel to timber connection from Madsen (2000)
In the case of the trussed arch, the connections should use V-type steel plates as
shown in Figure 5.16.
Figure 5.16 Components of the BSB connection from Madsen (2000)
The design should follow the requirements of Eurocode 5, part steel-to-timber

connections. It follows Johansen equations, dividing the whole joint in 3 members
units (timber-steel-timber).

5.2.4 Choice of the optimum connection
As a conclusion of this paragraph, a choice has to be made on the connection that will
be used in design. The choice is based on the 3 criteria (strength, stiffness, ductility)
and on the reliability of the system.
The Multiple steel plates connection is estimated as being the best solution. The main
advantage is the reliability of this joint, which has been used in many trussed arches.
(see Section 2.3)
5.3 Theory of multiple shear steel-to-timber joints

Before starting the description, one has to be careful about the meaning of different
terms used in this section. The joint called “multiple steel plates connection” in
section 5.2.3 is also called later “BSB connection” or “Multiple shear planes
connection” or “Multiple shear steel-to-timber connection”.
5.3.1 Double shear planes connection

The design of the connection is based on the theoretical Johansen model, which is
introduced in Eurocode 5. This theory is based on the analysis of all the possible
modes of failure in the connection. Originally developed for timber-to-timber
connection, this theory has been extended to steel-to-timber joints.
Let’s consider a single slotted steel plate as in Figure 5.17. This assembly is also
called double shear planes connection because two faces of the steel plate are in
contact with timber.
steel plate
dowel
timber
Figure 5.17 Double shear planes connection with a dowel fastener

This connection can fail in different ways, depending on the strength of steel and
timber and on the dimensions of the components. The possible modes of failure are
described in Jorissen (1998) and presented in Table 5.1.

Table 5.1 Basic failure modes and equations for a double shear planes connection
with dowels, Figures from Jorissen (1998)
Failure mode 3
Failure mode 1 Failure mode 2
3 plastic hinges in the
failure of timber 1 plastic hinge in the dowel
dowel
Load carrying capacity per shear plane according to Jorissen (1998)
4⋅M y
F1 = d ⋅ t1 ⋅ f h F2 = d ⋅ f h ⋅ t1 2+ 2
−1 F3 = 4 ⋅ M y ⋅ d ⋅ f h
d ⋅ f h ⋅ t1
Load carrying capacity per shear plane according to Eurocode 5
4⋅M y
F1 = d ⋅ t1 ⋅ f h F2 = d ⋅ f h ⋅ t1 2+ 2
−1 F3 = 2,3 ⋅ M y ⋅ d ⋅ f h
d ⋅ f h ⋅ t1
It is visible that equation 3 in Eurocode 5 is different from the original model. The
design of the connection is done following Eurocode’s equations.
The failure is brittle when mode 1 appears whereas it is ductile in mode 2 or 3.
5.3.2 Multiple shear planes connection

When the connection is made of several steel plates, it becomes a multiple shear
planes connections. In this case, it is necessary to divide the joint in a number of
double shear planes units, in order to calculate the load carrying capacity of the whole
connection.

A B C …
Figure 5.18 Multiple shear planes connection, Jorissen (1998) at left, and illustration
of the incompatibility of modes 1 and 2 in the inner parts at right.
A difference has to be done between the outer units (A) and the inner ones (B and C),
since the dimensions t1 and t2 are different (see Figure 5.18 left). Moreover, it is
important to consider the compatibility between the modes, which satisfies the
continuous deformation of the dowel. For instance, mode 1 is not compatible with
mode 2 as shown in the inner parts in Figure 5.18 right. As a result, only mode 1 or
mode 3 can happen.
According to Jorissen (1998), eight failure modes are compatible and have to be
considered for this type of connections. Equations (5.1) to (5.4) represent the load
carrying capacity per shear plane and per dowel for the different modes:
F1 / outer = d ⋅ f h ⋅ t1 (5.1)
F1 / inner = d ⋅ f h ⋅ 1 / 2 ⋅ t 2 (5.2)
4⋅M y
F2 = d ⋅ f h ⋅ t1 ⋅ 2+ −1 (5.3)
d ⋅ f h ⋅ t12
F3 = 2,3 ⋅ M y ⋅ d ⋅ f h (5.4)
The load carrying capacity per steel plate and per dowel will be calculated as a
combination of compatible modes, as described in the Table 5.2.
Table 5.2 Combination of compatible failure modes
Outer units: Fouter= Inner units : Finner=
Lowest value of: Lowest value of:
F1/outer+F1/inner F1/inner+F1/inner
F1/outer+F3 F3+F3
F2+F1/inner
F3 +F1/inner
F3+F2
F3+F3

The compatible failure modes are depicted in Figure 5.3.
Table 5.3 Basic failure modes and equations for a multiple shear planes connection
with dowels, Figures from Jorissen (1998)
F1/outer+F1/inner F1/outer+F3 F2+F1/inner
F3 +F1/inner F3+F2 F3+F3
F1/inner+F1/inner F3/inner+F3 /inner
Finally, the load carrying capacity per dowel of the whole connection is obtained by:
F = nouter _ units ⋅ Fouter + ninner _ units ⋅ Finner (5.5)

5.4 Design of the connection
5.4.1 Optimum number of steel plates
In the design, it is interesting to optimise the number of steel plates used in the
connection. In the journal entitled Joints, Connections and Substructures for Timber
Bridges published by the Timber Nordic Council, a method is proposed to optimise
this kind of joints.
The first step is to assume that the two possible failure modes in the inner units
happen at the same time. In that way, it is possible to choose the optimum thickness
between two steel plates (t2). This is obtained by setting: F1/inner+F1/inner =F3+F3. In
this way, it can be deduced that:
2,3 ⋅ M y ⋅ d ⋅ f h
t 2 _ optimum = (5.6)
0.5 ⋅ d ⋅ f h
In a second time, the thickness t1_optimum of the outer units can be chosen so that the
failure happens at the same time in the inner and the outer units.
When this method is applied to a truss connection, it appears that the optimum
number of steel plates is not the same in the two members connected. Indeed, the
embedding strength of timber changes with the orientation of the force. The
embedding strength of timber in the chord is reduced due to the fact that the resulting
force acting on the steel plate is not parallel to the grain.
Chord
Steel plate
Diagonal
Figure 5.19 Equilibrium of the steel plate

As a result, a choice has to be made, on which part of the connection should be
optimised. In the case of the trussed arch, the number of steel plates has been chosen
as the optimum between the steel plate and the chord.

005:103 71
This method is applied in our case using the following values:
Glulam GL32C Steel Fe 510
ρtimber=400kg/m3 φdowel=12mm
fh,0,k=28,86MPa fu,k=510MPa
My,k=97,85kNmm
Plate thickness=8mm
When designing this connection, the yielding of the steel should be estimated in an
accurate way. The strengths given in Eurocodes are often on the safe side. Then, it
could be wise to decrease this value in order to have a right optimum with the
predicted mode of failure.
It is found that the optimal values are t1= 82mm and t2=78mm. This pair of values
has to be modified to fit in the geometry of the chord. Two solutions are possible:
Configuration with 6 steel plates Configuration with 7 steel plates
85 83 70 81
630 630
645 645
t1=85mm – t2=83mm t1=81mm – t2=70mm
In order to decide which solution is better, the load carrying capacity is calculated in
both cases.

005:103
Table 5.4 Load carrying capacity (L.c.c) of the connection with 6 and 7 steel plates
Configuration with 6 steel plates Configuration with 7 steel plates
t1=85mm – t2=83mm t1=81mm – t2=70mm
F1/outer= 28749N F1/outer= 28056N

F1/outer= 14721N F1/outer= 12123N
F2= 13542N F2= 13294N
F3= 13390N F3= 13390N
Outer part Inner part Outer part Inner part
F1/outer+F1/inner = F1/outer+F1/inner =
43469N 40179N
F1/inner+F1/inner
F1/outer+F3 = 42138N =29441N F1/outer+F3 = 41446N F1/inner+F1/inner
=24246N
F1/inner+F2 = 28263N F1/inner+F2 = 25417N
F1/inner+F3 = 28111N F3+F3=26780N F1/inner+F3 = 25513N
F3+F3=26780N
F2+F3= 26932N F2+F3= 26684N
F3+F3= 26780N F3+F3= 26780N
Min= 26780N Min= 26780N Min= 25417N Min= 24246N
R0,k = 2 ⋅ 26780 + 4 ⋅ 26780 R0,k = 2 ⋅ 25417 + 5 ⋅ 24246
= 160679N per dowel = 172063N per dowel
The difference in the load carrying capacity is not very important between the two
solutions. Therefore, it seems more economical to make a connection with 6 steel
plates.
When 7 plates are used, the mode of failure is 1&2 in the outer part and 1 in the inner
part of the joint. This means that the timber is crushing, which will result in a brittle
failure.
When 6 plates are used, the mode of failure is 3 for both inner and outer part of the
connection. This mode corresponds to the yielding of the dowel and will result in a
ductile failure due to the plastic behaviour of steel.
It is much better to have a ductile failure since it will allow a redistribution of loads in
the structure before the collapse.
Finally, the alternative with 6 steel plates is chosen.

5.4.2 Required number of dowels
From the 6 different load combinations studied in chapter 4, the maximum force
applied on each connection is found and used in the design.
The required number of dowels can be obtained by dividing this force by the design
load carrying capacity per dowel. The design load carrying capacity per dowel is
k ⋅R 0,8 ⋅ 160679
given by: R0,d = mod 0,k = = 98879 N /dowel.
γM 1,3
Results of the calculations are shown in Table 5.4

Table 5.5 Required number of dowels in the connection between diagonals and steel
plates
Left arch Right arch

element critical load maximum nb of dowel element critical load maximum nb of dowel
number combination force kN required number combination force kN required
1 6 934 9,5 1 1 210 2,1
12 6 -1260 12,7 12 1 -809 8,2
23 6 -1024 10,4 23 1 -377 3,8
34 6 984 10,0 34 1 338 3,4
55 2 -1009 10,2 55 1 -709 7,2
76 2 693 7,0 76 1 388 3,9
87 1 -636 6,4 87 1 -636 6,4
98 1 -480 4,9 98 1 -480 4,9
109 1 363 3,7 109 1 363 3,7
110 6 482 4,9 110 6 -346 3,5
111 5 500 5,1 111 1 -308 3,1
122 5 -305 3,1 122 1 274 2,8
123 5 -409 4,1 123 6 338 3,4
Figure 5.20 Numbering of the diagonals on the arch

In the upper and lower chords, the resulting force is calculated by making the static
equilibrium of the steel plate. In the FE model the chords are continuous, which
implies that the moment is not zero at the joints. As a result, when the static
equilibrium is done in this point, the resulting force is not parallel to the chord.
Since this resulting force is applied at an angle to the grain α, the embedding strength
of timber ( fα,h ) is modified taking into account that angle. Then, the design load
carrying capacity per dowel can be calculated and used to define the minimum
number of dowels.
R
α

The details of the calculations can be found in Appendix B. The results of the
calculations are shown in Table 5.5 and Table 5.6. Some connections need to be
checked in several load cases because from a load combination to another, the
diagonals are tensioned or compressed.
Table 5.6 Required number of dowels in the connection between upper chord and
steel plates
Critical Force nb Critical Force nb

f α,h of dowels f α,h of dowels
node load magnitude α (deg) node load magnitude α (deg)
(MPa) (MPa)
comb. (kN) required comb. (kN) required
1 6 1516 13,5 27,8 16,3 7 1 444 5,5 28,6 4,8
2 6 1751 15,3 27,5 18,9 8 1 864 2,3 28,8 9,3
3 2 1338 16,8 27,3 14,4 9 1 829 5,3 28,7 8,9
4 1 795 18,7 26,9 8,6 10 1 756 5,8 28,6 8,1
1 600 18,7 26,9 6,5 11 1 571 5,8 28,6 6,1

5
5 341 3,8 28,7 3,7 1 330 2,5 28,8 3,6
12
6 5 708 14,8 27,6 7,6 6 473 15,1 27,5 5,1

005:103 75
Table 5.7 Required number of dowels in the connection between lower chord and
steel plates
Critical Force Critical Force
α f α,h nb α f α,h nb
node load magnitude node load magnitude
(deg) (MPa) of dowels (deg) (Mpa) of dowels
comb. (kN) comb. (kN)
13 6 1732 6,6 28,6 18,6 19 1 882 24,9 25,7 9,5
6 1543 0,0 28,8 16,6 20 1 852 15,6 27,5 9,2
14
2 1524 1,7 28,8 16,4 21 1 818 10,8 28,1 8,8
2 1009 3,5 28,7 10,9 22 1 665 6,2 28,6 7,2
15
1 818 10,8 28,1 8,8 23 1 457 2,5 28,8 4,9
16 1 665 6,2 28,6 7,2 24 6 536 0,5 28,8 5,8
17 5 643 10,8 28,1 6,9
5 700 3,6 28,7 7,5

18
6 700 3,7 28,7 7,5
Figure 5.22 Node numbering on the arch
5.4.3 Tension perpendicular to the grain

Timber is very sensitive to tension perpendicular to the grain. Indeed, the
characteristic resistance is really low ( f t ,90,k = 0,45MPa for GL32). A force applied
on a member with an angle to the grain can involve splitting and drive the structure to
the failure. When it is not possible to avoid it, it is necessary to check the risk of
splitting.
Eurocode 5 proposes a criterion to check the splitting when a member is tensioned at
an angle to the grain.

Splitting
b
FV,ed1 he FV,ed2
Fed
Figure 5.23 Risk of splitting in a joint, Eurocode 5
The component Fed ⋅ sin α produces splitting of the beam. It has to be checked that
the depth he is sufficient to avoid splitting. That means that the distance between the
loaded edge and the last fastener is determinant for the splitting.
The resistance is calculated as following:
he
F90, R ,k = 14 ⋅ b ⋅ w ⋅ with w = dowel type fastener (5.7)
(1 − he / h)
F90, R ,k .k mod
F90, R ,d = (5.8)
γM
FV ,ed 1
The resistance has to be greater than: max
FV ,ed 2
In the case of the trussed arch, the splitting of the chords is checked for all load
combinations. The shear forces in the chords FV,ed1 and FV,ed2 can easily be found in
the results of the Finite Element computation. Table 5.7 shows the maximum shear
forces for the different load cases.
Table 5.8 Maximum shear forces in the chords of the trussed arch
Load FV,ed1 FV,ed2 At node
combination (kN) (kN) See Fig5.22
1 209 145 13 or 19
2 140 188 19
3 38 27 5
4 53 33 3
5 120 160 19
6 130 175 19

The most important risk of splitting occurs in the first load combination. The splitting
would occur at first on nodes 13 and 19 in the lower chord.
At node 13 or 19:
FV,ed =209kN
he F90, R ,k .0,8
F90, R ,k = 14 ⋅ 645 ⋅ 1 ⋅ and F90, R ,d =
(1 − he / 630) 1,3
and then, he,min = 440mm

This value will be used for all the steel plates on the safe side.
630mm
he,min = 440mm
Figure 5.24 Minimum spacing to prevent splitting
5.4.4 Design of the steel plates

In order to make the execution easier and to optimise the production, it has been
decided to use only four different connections in the whole structure. Two types of
steel plates will be set in upper chord and two others in the lower chord. The number
of dowels has been chosen in order to fulfil the requirements, see paragraph 5.4.2.
However, the spacing between the dowels and the spacing of the holes in the steel
plates have to be calculated according to Eurocodes so that the final pattern of the
dowels is suitable. All those characteristics are presented in the following parts.
5.4.4.1 Fastener spacings
The dowels have to be spaced at an appropriate distance to fulfil the requirements of
both Eurocode 5 and Eurocode 3. Indeed, the spacing between the dowels should be
suitable in order to avoid splitting in the timber member. Moreover, the distance
between the holes in the steel plates should also fulfil some requirements.
The spacing requirements in the timber member are calculated according to Eurocode
5. Here, a distinction has to be done between the diagonals members and the chord.
The end distance a3 and the edge distance a4 have thereby different values in the
diagonal members and in the chord members. This is due to the fact that the angle of
the force is not the same in these members: αdiag=0 and αchord,max=19º.
Eurocode 3 is used to calculate the spacing between the holes in the steel plate.

005:103
Figure 5.25 Definition of the distance in the diagonals
Table 5.9 Minimum spacings and distances for dowels in the diagonals αdiag=0,
according to Eurocode 5.
a1 Parallel to the grain (3+2.cosα).d 60mm

a2 Perpendicular to 3.d 36mm
the grain
a3,t αdiag=0 max(7.d ;80mm) 84mm
a3,c max (a3,t.sinα.d ; 3.d) 36mm
a4,t αdiag=0 max ((2+2sinα).d ; 36mm
a 4,c 3.d) 36mm
3d
Figure 5.26 Definition of the distance in the chord

005:103 79
Table 5.10 Minimum spacings and distances for dowels in the chord αchord,max=19,
according to EC5.
a1 Parallel to the grain (3+2.cosα).d 60mm

a2 Perpendicular to the grain 3.d 36mm
a4,t αchord,max=19 max ((2+2sinα).d ; 3.d) 36mm
a 4,c 3d 36mm
Compared to the spacing requirements between the holes in the plate, the values given
in Table 5.8 are much higher. This is due to the higher strength of the steel.
However, the end distance and the edge distance in the steel plate still have to be
determined:
Minimum end distance: 1,2 × d 0 ≤ e1 ≤ 40 + 4 × t ⇒ 15,6mm ≤ e1 ≤ 72mm
Minimum edge distance: 1,5 × d 0 ≤ e2 ≤ 40 + 4 × t ⇒ 19,5mm ≤ e2 ≤ 150mm
In the final design of the connection, the following distance were used:
a1 = 100mm - a3,t = 86mm - a4 = 50mm - e1 = 50mm - e2 = 50mm

The different connections and their location are presented in Table 5.11.
Table 5.11 Four connection types used in the arch.
UP1: 21 dowels in the chord UP2: 9 dowels in the chord

14 dowels in the diagonals 8 dowels in the diagonals
LO1: 21 dowels in the chord LO2: 13 dowels in the chord


UP2 UP2
UP2 UP2
UP2 UP2
UP1 LO2 LO2 UP1
LO2 LO2
LO2 LO2 UP1
UP1 LO2
LO2
UP1 LO1 UP1
LO1
LO1 LO1
Figure 5.27 Location of the connections
5.4.4.2 Resistance of the steel plate

The pressure between the dowel and the steel plate may imply failure of the plate
Thus, according to Eurocode 3, the bearing resistance of the plate, the resistance of
the net section and also the tensile failure of the plate have to be checked.
As an example, the checking of the connections UP1 in the upper chord is presented
in the tables below. The design values come from the ABAQUS results. The details of
the calculations can be found in Appendix C.
H
V
Table 5.12 Resistance of the plate in the diagonal
Resistance of the plate in the diagonal
Bearing resistance Fb,Rd = 94,79kN/dowel > Fd = 90kN/dowel O.K
Resistance of the Ft,Rd = 3771kN > Ft= 934kN O.K

net section
Tensile failure of Ft,Rd =614kN/plate > Ft =164kN /plate O.K

the plate
In the chord, the two components of the resultant force acting on the steel plate in the
chord have to be checked.

Table 5.13 Resistance of the plate in the chord regarding to the horizontal component
Resistance of the plate in the chord regarding to the horizontal component
Bearing resistance Fb,Rd = 106kN/dowel > Fd = 70kN/dowel O.K

net section
Tensile failure of Ft,Rd = 1105kN/plate > Ft =245kN /plate O.K

the plate
Table 5.14 Resistance of the plate in the chord regarding to the vertical component
Resistance of the plate in the chord regarding to the vertical component
Bearing resistance Fb,Rd = 106kN/dowel > Fd = 22kN/dowel O.K

net section
Tensile failure of Ft,Rd = 1032kN/plate > Ft =77,3 kN/plate O.K

the plate
Thus, the connection fulfils the requirements of Eurocode 3 and 5 and is strong
enough to withstand the forces.
5.4.5 Stiffness of the connections

In the preliminary analysis performed in chapter 4, the connections in the truss are
assumed to be pinned. This assumption is not realistic since the multiple steel plates
connection is pretty stiff. Thus, it has been decided to create a more realistic FE model
including the stiffness of the joints. This model may lead to different results. In this
section, the stiffness of the joints are estimated. They will be used in chapter 6 when
the new model will be created.
In the joint, each group of dowels can be modelled by two springs: one rotational
spring and one translational spring. With the slip modulus Kser, calculated with
Eurocode 5, the stiffness are worked out as:
2 (5.9)
Kf = K ser ⋅ n
3
2 (5.10)
K θ = K ser ⋅ ∑ x 2 + y 2
3 n
where n is the number of dowels, and (x ,y) the coordinates of the dowels.

005:103
The connection consists in three groups of dowels, one in each diagonal and one in
the chord. As the connection is symmetrical, the group of dowels in the diagonals
have the same rotational stiffness and translational stiffness. However, it is reasonable
to assume that the rotation of the diagonal members is negligible since it is prevented
by the contact with the chord element.
Finally, an appropriate model for the connections consists in one rotational spring in
the chord and two translational springs, which link the chord and the diagonals, see
Figure 5.29.
Kθ
Kf
Kf
Figure 5.29 Joint model.

The stiffness of the translational spring of one steel plate should take into account the
stiffness of the group of dowels in the chord (kf1) and the stiffness of the group of
dowels in the diagonal (kf2). This leads to a unique spring, which has a stiffness of:
k f1 ⋅ k f 2
Kf = . (5.14)
k f1 + k f 2
Then, the final values of translational and rotational stiffness of the connections are
given by multiplying the Kf and Kθ with the number of plates. In the previous section,
four different connections have been designed with different shapes and different
numbers of dowels see Table 5.10. For each connection, the characteristic values of
the springs are calculated. The calculations for the connection UP1 are presented
below. All the stiffness are given in Table 5.15.
UP1 consists in 21 dowels in the chord and 14 in the diagonals. Knowing that the
diameter of the dowels is 12mm and the density of the Glulam is 400kg/m3, the
stiffness are:
1
Kser = 2 ⋅ ⋅ ρ 1,5 ⋅ d = 8348N/mm
23
2
Ku= ⋅ K ser =5565N/mm
3
2
Kf chord per plate= K ser ⋅ nchord =116870N/mm
3
2
Kf diag per plate= K ser ⋅ ndiagonal =77913N/mm
3
k fchord ⋅ k fdiag
K f FINAL PER PLATE = =46748N/mm
k fchord + k fdiag

005:103 83
K f FINAL= 6. K f FINAL PER PLATE =280,5kN/mm
Ip= ∑ x 2 + y 2 =340000mm2
2
K θ per plate= K ser ⋅ ∑ x 2 + y 2 =1,89.10+9kN/rad
3 n
K θ FINAL= 6. K θ per plate=1,14.10+10 kN/rad
Table 5.15 Ttranslational and rotational stiffness
Kf (kN/m) Kθ (kN/rad)
+3
UP1 280.10 1,14.10+10
UP2 141.10+3 2,0.10+9
LO1 280.10+3 1,14.10+10
LO2 165.10+3 5,4.10+9

6 Analysis of the structure and verification
In the previous chapter 5, the connections of the trussed arch have been designed. The
design was based on the results of computation presented in chapter 4.
In this chapter, a more advanced model will be done. It will take into account the
stiffness of the connections by introducing spring connectors between the members.
The results of this computation will be more accurate than the previous one. They will
be used to check the strength of the chords and the diagonals. The connection will be
checked as well. The possibility of optimising the section will be discussed.
At the end, two other investigations will be done in order to study the need for lateral
bracing and the question of partitioning the arches for transport.
6.1 Finite Element analysis including connection stiffness

6.1.1 FE model
Eurocode 5 states that the trusses should normally be computed using pinned
connections, as it has been done in chapter 4. However, the choice has been made to
use multiple steel plates joints, which are pretty stiff. Indeed, a further investigation
has been done in order to see the influence of their stiffness.
The model has the same geometry than before, but is not composed with the same
elements. The chords and the diagonals are made of beam elements, because they are
subjected to normal forces and bending moments.
Continuous chord:
beam elements
20m
Diagonals:
beam elements
100m
Figure 6.1 Illustration of the FE model
The connections are semi-rigid and have the stiffness calculated in section 5.4.5. The
stiffness is considered to be linear, because no information has been found to estimate
the behaviour of this kind of joint more accurately.

005:103 85
Figure 6.2 Illustration of the model of the connections in the trussed arch
Table 6.1 Stiffness of the connectors
Kf ( kN/m) Kθ (kN/rad)
UP1 280 10+3 1,1 10+10
UP2 141 10+3 2,0 10+9
LO1 280 10+3 1,1 10+10
LO2 165 10+3 5,4 10+9
The model is still not perfect since the diagonals are not prevented from extending
over the chord. To solve this problem, a limit position has been defined in the
ABAQUS model in order to prevent the diagonal to go out of its position. That means
that the translation of the diagonal is possible only when the member is in tension. In
compression, the diagonal can only rotate.
The model is analysed in load cases 1,2,5 and 6 because they are the critical ones in
ultimate limit state.

005:103
6.1.2 Results of the analysis
The results of this analysis are presented in the same way as in chapter 3 and the
difference between each value is shown.
Table 6.2 Maximum force in the diagonals
Max. Difference with Max tensile Difference with

compressive previous model force previous model
force
LC1 -902kN + 11% +436kN + 12%
LC2 -1365kN + 15% +1034kN + 10%
LC5 -1400kN + 17% +1089kN + 14%
LC6 -1473kN + 17% +1152kN + 17%

compressive force
LC1, LC2, LC5, LC6

LC1
LC1, LC2, LC5, LC6 LC1

tensile force
Figure 6.3 Location of the maximum normal forces in the diagonals
The forces in the diagonals increase significantly when the stiffness of the
connections is introduced.

005:103 87
Table 6.3 Maximum forces in the chords
Maximum Difference with Max tensile force Difference with

compressive previous model previous model
force
LC1 -3839kN + 1% No tension =
LC2 -4737kN + 13% +1086kN - 3%
LC5 -4688kN + 16% +1388kN + 2%
LC6 -4892kN + 14% +1326kN + 1%

compressive force
LC2, LC5, LC6
LC1 LC1
LC2, LC5, LC6

tensile force
Figure 6.4 Location of the maximum normal forces in the chords
The maximum compressive forces are rising in this new model but the tensile forces
are almost not influenced.

005:103
Table 6.4 Maximum moments in the chords
Max. negative Difference Max positive Difference with

moment with previous moment previous model
model
LC1 -237kNm + 23% +171kNm - 43%
LC2 -266kNm + 55% +211kNm - 20%
LC5 -231kNm + 31% +225kNm - 1%
LC6 -253kNm + 37% +232kNm - 6%

positive moments
LC2
LC5, LC6
LC1 LC1
LC1, LC2, LC5, LC6 LC1

negative moments
Figure 6.5 Location of the maximum moments in the chords
The magnitude of the bending moments changes radically. The maximum negative
moment is rising up to 55% whereas the positive moment is decreasing.
By these results, it has been demonstrated that the computation of the forces and
moments using a pinned-connected truss is on the unsafe side.
A bending moment is transferred to the diagonal elements by the effect of the
rotational springs. The magnitude of this moment is relatively low (up to 30kNm).
The values are not presented because the effect of this moment is limited.

005:103 89
It has been shown that the forces and the bending moments rise when semi rigid
connections are used in the model. The magnitude of the deflection of the trussed arch
is now shown and compared with the previous model created in chapter 3.
Table 6.5 Deflection of the trussed arch, comparison between preliminary model and
model with semi rigid connections
Deflection Deflection
preliminary model with semi
model, chapter 3 rigid connections
LC1 6,8cm 10,4cm
LC2 9,6cm 15,7cm
LC5 9,4cm 15,9cm
LC6 9,8cm 16,2cm
It appears that the deflection is more important in the model with semi rigid
connections. However, one could expect that the deflection should decrease when the
stiffness of the truss connections increases since the rotational stiffness has a good
effect on the deflection. Hence, it seems reasonable to assert that the deflection
increases because of the reduction of the translational stiffness.
6.2 Verification of the members

All the members in the trussed arch are subjected to either compression or tension,
combined with bending. Both chords and diagonal elements have to be checked.
Special cares have to be taken regarding the buckling of the members.
For the most loaded members, the verification will be done in two locations:
- Between the nodes, to check instability.
- At the level of the connection, taking the reduction of the section into account.
The buckling length of a member depends mainly on its support conditions. The
stiffer are the connections, the smaller is the buckling length. The buckling length is
determined in the next sections.
At the level of the connections, the steel plates slots reduce the cross-section. This
means that the width should be decreased by 6 × 9mm = 54mm (number of steel
plates×width of the slots).
The most subjected members are verified according to Eurocode 5. Their locations are
shown in Figure 6.6.

C
D
12 B
76
34
A
Figure 6.6 Location of the critical members
6.2.1 Chord members

The chord members have a cross-section of 645×630 mm2. At the level of the
connection, the reduced area is 591×630mm2. The curvature of the members can be
neglected since the radius of the arch is important (~80m). It has been checked that
the influence of the curvature on the maximum stress is less than 1%.
The chord elements can be seen as continuous beam supported on several supports.
Hence, lateral-torsional buckling should be verified between the nodes. In this case, it
has been assumed that the buckling length is equal to the distance between two nodes
times 0,7. This value has been chosen as an approximation of the real case,
considering that the chord is continuous but not totally rigid.
N N N
lef = l lef = l/ 2 lef = l/2
Figure 6.7 Buckling length for different boundary conditions
Thus, the stresses in the members subjected to combined compression and bending
should verify (6.1) between the nodes.
2
σ m ,d σ c ,d
+ ≤1 (6.1)
k crit ⋅ f m ,d k c , z ⋅ f c , 0 ,d
At the level of the connection, there is no risk of buckling. In case of compression, the
stresses should then verify (6.2) considering the reduced cross-sectional area.
2
σ c , 0 ,d σ m , y ,d
+ ≤1 (6.2)
f c , 0 ,d f m , y ,d

For the chord members subjected to combined tension and bending, the stresses
should verify (6.3) considering also the reduced cross-sectional area.
σ t , 0 ,d σ m , y ,d
+ ≤1 (6.3)
f t , 0 ,d f m , y ,d
The characteristics of the chord members are presented in the table below:
Table 6.6 Parameters of a chord member
In the mid-span at the node level

b 0,645 m b 0,591 m
h 0,63 m h 0,27 m
l 8,23 m f m,d 20,5 MPa
f c,0,d 18,6 MPa
leff 5,761 m
k crit 1 f t,0,d 15,4 MPa
kc,z 0,922
f m,d 20,5 MPa
fc,0,d 18,6 MPa
ft,0,d 15,4 MPa
The checking is presented for member A in this section. The other members are
verified in Appendix D.
Table 6.7 Verification of the member A in the mid-span
Combined bending and compression-Member A
Max Compressive Max Bending Combined

force Stresses moment stresses stress
LC1 3840kN 9,45MPa 237kNm 5,55MPa 0,63 <1
LC2 3920kN 9,63MPa 266kNm 6,23MPa 0,66 <1
LC5 3310kN 8,15MPa 231kNm 5,41MPa 0,55 <1
LC6 3650kN 8,97MPa 253kNm 5,93MPa 0,61 <1

Table 6.8 Verification of the member A at the node level
Combined bending and compression Member A
Compressive Bending Combined

Max moment
Max force Stresses stresses stress
LC1 3840kN 10,3MPa 171kNm 4,37MPa 0,52 <1
LC2 3920kN 10,5MPa 162kNm 4,14MPa 0,52 <1
LC5 3310kN 8,90MPa 135kNm 3,45MPa 0,40 <1
LC6 3650kN 9,79MPa 149kNm 3,81MPa 0,46 <1
These results show that in all the critical load combinations, the stresses in the
member A are acceptable. However, it can be noticed that the cross-section is always
stressed at less than 70% of its capacity.
6.2.2 Diagonal members

The diagonal members have a cross-section of 645×270mm2. At the level of the
connections, the area is reduced to 591×270mm2. The stability of these members
should also be checked. As for the chord members, it is assumed that the buckling
length is 0,7.lef.
For the diagonal members subjected to combined bending and compression, the
stresses should verify (6.4) so that there is no risk of buckling:
σ m ,d σ c , 0 ,d
+ ≤1 (6.4)
f m ,d k c , z ⋅ f c , 0 ,d
At the level of the connection, there is no risk of instability. The stresses should verify
(6.5), considering the reduced cross-sectional area.
2
σ c , 0 ,d σ m , y ,d
+ ≤1 (6.5)
f c , 0 ,d f m , y ,d
In the case of combined tension and bending, the stress should verify (6.6)
considering also the reduced cross-sectional area.
σ t , 0 ,d σ m , y ,d
+ ≤1 (6.6)
f t , 0 ,d f m , y ,d
All the verifications can be found in Appendix D. It appears that there is no instability
problem. However, the cross-section is not used in an efficient way since the stress
never exceeds 60% of the capacity.

6.3 Verification of the connections
The magnitude of the forces in the diagonals being more important, the connections
have to be checked again taking into account this rise. The same method as in the
chapter 5 is used to calculate the required number of dowels. The results are presented
in the tables before, showing the differences with the previous stage.
Table 6.9 New estimation of the necessary number of dowels between the steel
plates and the diagonal members
Required number of dowels
Node With new Previous Difference

nb model stage
1 10,4 9,5 +9 %
12 14,9 12,7 +17 %
23 11,8 10,4 +13 %
34 11,6 10,0 +16 %
55 12,1 10,2 +18 %
76 7,9 7,0 +12 %
87 7,3 6,4 +13 %
98 7,7 4,9 +57 %
109 3,7 3,7 0%
110 4,9 4,9 0%
111 5,3 5,1 +3 %
122 3,2 3,1 +3 %
123 4,3 4,1 +4 %
Figure 6.8 Numbering of the diagonals on the arch

Table 6.10 New estimation of the necessary number of dowels between steel plates
and chords
Required number of dowels
Node nb With new Previous Difference

model stage
1 21,1 16,3 +29 %
2 25,4 18,9 +34 %
3 19,4 14,4 +34 %
4 9,9 8,6 +15 %
5 7,2 6,5 +10 %
6 9,2 7,6 +21 %
13 24,6 18,6 +32 %
14 22,9 16,6 +36 %
15 14,6 10,9 +34 %
16 10,6 7,2 +47 %
17 8,3 6,9 +20 %
18 8,9 7,5 +18 %
Figure 6.9 Node numbering on the arch
The results shown in the tables before confirm that the calculations performed with
the previous model were on the unsafe side. The connections designed before would
not fulfil the Eurocode 5 requirements. As a result, it has been decided to draw new
steel plates.

Table 6.11 Steel plates resisting the forces computed at the second analysis
UP1’ : 29 dowels in the chord UP2’ : 10 dowels in the chord

LO1’ : 25 dowels in the chord LO1’ : 17 dowels in the chord

UP2’ UP2’
UP2’ UP2’
UP2’ UP2’
UP1’ LO2’ LO2’ UP1’
LO2’ LO2’
LO2’ LO2’ UP1’
UP1’
LO2’ LO2’
UP1’ LO1’ UP1’
LO1’
LO1’ LO1’
Figure 6.10 Location of the connections
These new steel plates were checked in the same way than in chapter 5, including the
splitting of the chords.

005:103
6.4 Lateral bracing of the lower chord
The problem of buckling of the lower chord has already been discussed in chapter 3,
when an optimum model had to be chosen. As said before, the upper chord is assumed
to be stabilized by the roofing. That means that the upper chord has no risk of
buckling. However, the lower chord has to be checked. If the risk is not acceptable,
lateral bracing units should be designed like in Hamar Olympiahall, Figure 3.31.
The last model including the stiffness of the joint is used to perform a buckling
analysis. The computation of such an analysis gives eigenvalues, which can help to
determine the load for which the buckling appears. The Figures 6.11 and 6.12 show
the first buckling modes of the trussed arch under load combinations LC1 and LC2.
Figure 6.11 First mode of buckling in load case 1
Figure 6.12 First mode of buckling in load case 2
The eigenvalue of the first mode is always above 4,0, which means that the buckling
should appear when the load is 4 times higher than the one given by the load
combination considered.
These results lead to say that the lower chord is safe regarding lateral buckling. The
structure does not need lateral bracing.

6.5 Partitioning of the arches
To transport the arch between the manufacture and the construction site, it has to be
cut in several parts. Each piece should fulfil the transportation requirements, which
are presented in chapter 2. This chapter introduces and compares two different
possibilities to partition the arch. The cuts are first done at the nodes, then both at the
nodes and in the middle of chord elements.
The models are computed in ABAQUS. The stiffness included in the models are the
same as before. The analysis is only carried out under the LC6.
In the first alternative, the cuts are only at the nodes. The arch is cut in four pieces.
The assembly will not require extra connections.
Figure 6.13 Location of the cuts
Table 6.12 Comparison of the forces
Original model Partitioned model

Difference
under LC 6 under LC 6
Max compressive
-4892kN -4582kN 6%
force
Max tensile force +1326kN +1271kN 4%
Table 6.13 Comparison of the moments

Difference
Max negative
-253kNm -269kNm 6%
moment
Max positive
+232kNm +201kNm 1%
moment

005:103
Original model
Partitioned model
Figure 6.14 Deformed shape
In the second alternative, the cuts are achieved both in the element and at the node.
The stiffness of the connections set in the middle of a chord element are defined by
the same values of the previous connections since they are also multiple steel plates
connections.
Figure 6.15 Location of the cuts.
The results computed by ABAQUS are presented in the tables below.
Table 6.14 Comparison of the forces

Difference
Max compressive
-4892kN -4711kN 4%
force
Max tensile force +1326kN +1266kN 4%

Table 6.15 Comparison of the moments

Difference
Max negative
-253kNm -286kNm 4%
moment
Max positive
+232kNm +192.5kNm 4%
moment
Original model
Partitioned model
Figure 6.16 Deformed shape

Although the difference between the original model and the partitioned ones is not so
relevant, it can be observed that the forces decrease. Since the arch is cut in pieces, the
chord cannot be anymore modelled as a continuous beam but in four smaller
continuous beams. As a consequence, the negative moments at the node are larger,
and the positive moments in the span are quite smaller.
The arch also deforms differently as it is shown in the Figure 6.15 and Figure 6.16.
The difference between the alternatives is not very significant. Thus, it seems
reasonable to assume that it is better to set the cuts at the node. In this way, no extra
connection is required in the chord, which is more economical.

005:103
7 Conclusion and recommendations
The aim of this thesis was to study a trussed arch for long span. A structure of 100m
span and 20m height was investigated with the finite element (FE) software
ABAQUS. Based on this configuration, some relevant parameters like boundary
conditions, number of diagonals and truss depth were modified in order to observe
their influence and determine the optimum static system. The evaluation between the
alternatives was performed under a triangular snow load distribution to accentuate the
difference.
The comparison of the different boundary conditions led to find the optimum system
regarding the force path, assuming lateral stabilization of the upper chord. The chosen
model consists in setting the abutments’ hinge at the lower chord and the crown’s
hinge in the upper chord (BC3). This model is also advantageous regarding buckling
problem in the lower chord.
Then, the number of diagonal elements was chosen regarding to local buckling
problem. The more diagonals there were, the less risk of buckling there was.
However, the issue also had to be considered from an economical point of view.
Therefore, the solution should be a compromise between all these considerations.
Hence, our recommendation is to take an angle of 40° between the chord and the
diagonal bars. This value is confirmed by several previous constructions.
The choice of the depth of the trussed arch was mainly done regarding to buckling
problems in the diagonal members. Transportation issues also had to be considered.
Hence, it is reasonable to select a depth of 3,5m
40º
3,5m
20m
100m
Figure 7.1 Optimum static system
Furthermore, several types of joints were assessed for the hinges and the truss
connections. Particular attention was paid on the truss connections. For long span
trussed arch, multiple steel plates connection is nowadays the only valuable joining
system. A preliminary design of this connection had to be accomplished in the early
stage because it often determines the minimum size of the cross-sections. The
connections were calculated so that their failure was ductile. Therefore, it resulted that
6 steel plates were suitable. Around 25 dowels were necessary to fasten the steel
plates in the chord and 15 dowels in the diagonals.

005:103 101
Figure 7.2 Multiple steel plates connections
Thereafter, the joints’ stiffness were estimated. The connections were modelled with
translationnal and rotational springs, with a stiffness depending on the number of
dowels. A complex FE model of the trussed arch including these stiffness was created
and worked out in ABAQUS. The results of this analysis showed significant changes
in the magnitude of the forces. Thus, it comes out that the design of the structure
assuming pinned connection is not on the safe side. However, as it is time consuming
to create a complex model, it can be suggested to increase the previous number of
dowels by 20%.
The cross-sections of the diagonals (230×645mm2) and of the chord (630×645mm2)
were checked according to the Eurocode 5. All the requirements regarding splitting
and buckling problems were fulfilled. It can be noticed that the cross-sections capacity
is not used more than 70%. Nevertheless, the optimisation is quite hard to achieve
since the connections determine the minimum size.
The risk of buckling in the lower chord was evaluated from a buckling analysis of the
structure. The computed critical load was very high. This result confirms that the
selected model doesn’t require special lateral bracing units on the lower chord.
Finally, the partitioning of the arch for the transportation between the manufacture
and the construction site was also approached. It appears that the cuts at the level of
the nodes are more appropriate. Furthermore, this solution is more economical since it
doesn’t require extra steel plates.

8 References
Eurocode 1: Basis of design and actions on structures (ENV 1991-2-1:1995).
Eurocode 1: Actions on the structures-General actions-Part 1-3:Snow actions (ENV
1991-2-3:1995).
Eurocode 1: Actions on the structures-General actions-Part 1-4:Wind actions (ENV
1991-2-4:1995).
Eurocode 5: Design of timber structures Part1-1: General-Common rules and rules
for buildings (EN 1995-1-1:2004).
Carling O., Svenskt Limträ Ab (2001): Limträhandbok. Svenskt limträ AB,

Stockholm.
Jorissen A. J. M. (1998): Double shear timber connections with dowel type fasteners.
Delft University Press, Delft.
Kinney J. S. (1957): Indeterminate structural analysis, Addison-Wesley Publishing
Company, Reading, Massachusetts, U.S.A.
Madsen B. (2000): Behaviour of timber connections. Timber Engineering Ltd.,
Vancouver.
Natterer J., Sandoz J.-L., Rey M., Fiaux M. (2000): Construction en bois : matériau,
technologie et dimensionnement Presses polytechniques et universitaires romandes,
Lausanne.
Other literature:
Johansson H. (2001): Systematic design of glulam trusses, Licentiate thesis, Luleå
University of Technology.
Thelandersson S., Larsen H. J. (2003): Timber engineering, Wiley, Chichester.

Appendix A: Comparison between ABAQUS results
and hand calculations
Calculation of the self weight of the superstructure:
G selfweight = G structure + G roof
G strucutre = G chord + G diagonal
G chord = ρ GL 32 C ⋅ g ⋅ V chord
V chord = 2 ⋅ (R ext ⋅α ⋅ 0 , 63 ⋅ 0 , 645 + R int α ⋅ 0 , 63 ⋅ 0 , 645 )
π
V chord = 2 ⋅ 0 , 63 ⋅ 0 , 645 ⋅ ⋅ 36 , 52 ⋅ (84 + 87 , 5 ) = 88 ,83 m 3
180
G chord = ρ GL 32 C ⋅ g ⋅ V chord = 400 ⋅ 9 ,81 ⋅ 88 ,83 = 348 , 56 kN
G diagonal = ρ GL 32 C ⋅ g ⋅ V diag
V diag = 26 ⋅ 0 , 645 ⋅ 0 , 270 ⋅ 5 , 47 = 24 ,8 m 3
G diagonal = ρ GL 32 C ⋅ g ⋅ V diag = 400 ⋅ 9 ,81 ⋅ 24 ,8 = 97 , 31 kN
G strucutre = G chord + G diagonal = 348 , 56 + 97 , 31 = 445

G roof = 0 ,8 ⋅ R ext ⋅ 2 ⋅ α = 1070 kN
G selfweight = G structure + Groof = 445,8 + 1070 = 1515,8kN
Calculation of the variable load:
The variable action, which is considered for LC1, is the uniform snow load.
Qsnow = 19,2 ⋅ l span = 19,2 ⋅ 100 = 1920kN
Load combination:
Qd = 1,35 ⋅ G se lfweight + 1,5 ⋅ Qsnow = 1,35 ⋅ 1515,8 + 1,5 ⋅ 1920 = 4926.3kN
As the system is symmetrical, it is reasonable to assume that the reaction forces are
equal to half of the loading:
4926,3
Rleft = Rright = = 2463kN
2
Q ⋅l
The horizontal thrust at the abutments is given by: H = d ,
8⋅ f
where f is the high of the arch.

4926,3 ⋅ 100
H= = 3078kN
8 ⋅ 20
The reaction forces given by ABAQUS are: Rleft=Rright= 2520kN and H=3211kN.
The model is also checked by making the equilibrium in the two first nodes of the
trussed arch.
Equilibrium at node 1:
Horizontal equilibrium:
FA ⋅ cos(75,88) + FB ⋅ cos(34,7 ) = 3079
FA Vertical equilibrium:
75,88 FB FA ⋅ sin (75,88) + FB ⋅ sin (34,7 ) = 2463
Results:
34,7 FA=414kN
FB= 3620kN
1
HA=3079kN
RA=2463kN
The second node is also checked.
In order to perform the equilibrium, an external force due to the snow load and the
self-weight of both the roof and the members of the upper chord is applied on the
node. It is assumed to be equal to:
Qd = 1,35 ⋅ (ρ gl 32 c ⋅ g ⋅ S chord + 0,8 ⋅ 12) + 1,5 ⋅ q snow = 1,35 ⋅ (0,4 ⋅ 9,81 ⋅ 0,63 ⋅ 0,645 + 0,8 ⋅ 12 ) + 1,5 ⋅ 19,2
Qd = 43,7kN / m
l 8,58
Fd = Qd ⋅ member = 43,7 ⋅ = 188kN
2 2

Equilibrium at node 2:
Horizontal equilibrium:
Fc ⋅ cos(31,86 ) = FD ⋅ cos(6,5) + 414 ⋅ sin (14,12 )
FC Vertical equilibrium:
FC ⋅ sin (31,86) + FD ⋅ sin (6,5) + 188 = 414 ⋅ cos(14,12)
188kN
31,86
Results:
2
6,5 FC=360kN
FD FD= 206kN
14,12
FA= 414kN

Appendix B: Number of dowels in the connections
Table A-1 Material and section properties of the trussed arch
Material properties Section properties
Timber Steel Diagonals Chord

2
ρGL32C = 400kg/m 3 Фdowel = 12mm 0,645 x 270mm 0,645 x 0,630mm2
fu,k = 510MPa t1=83mm, t2=85mm t1=83mm, t2=85mm
According to Eurocode 5, part 8.5.1.1,
M y , R ,k = 0,3. f u ,k .d 2, 6 = 97850 N .mm
f h , 0,k = 0,082.(1 − 0,01.d ).ρ = 28,8MPa
f h , 0,k
f h ,α ,k =
k 90 . sin α + cos 2 α
2
with k 90 = 1,35 + 0,15.d for softwood
Connection Steel plates – Diagonals

The load carrying capacity per dowel is calculated in
the way described in Section 5.4.2
Table A-2. Calculation of the load carrying capacity of the connection
Load carrying capacity per shear plane per dowel

F1/outer= 28749N
F1/outer= 14721N
F2= 13542N
F3= 13390 N
Load carrying capacity per steel plate per dowel
Outer part Inner part
F1/outer+F1/inner = 43469N
F1/outer+F3 = 42138N F1/inner+F1/inner =29441N
F1/inner+F2 = 28263N
F1/inner+F3 = 28111N F3+F3=26780N
F2+F3= 26932N
F3+F3= 26780N
Min= 26780N Min= 26780N
Characteristic load : R0,k = 160679N per dowel
Design load : R0,d = 98879N per dowel
The design load carrying capacity per dowel R0,d = 98879N/dowel for all diagonals
since the load is applied parallel to the grain.

005:103 107
Table A-3 Details of the calculations of the required number of dowels.
Stresses in the chords computed with
Combination
Abaqus MAX MAX
Element nb of
Load
STRESS FORCE dowel
LC1 LC2 LC3 LC4 LC5 LC6
(MPa) (kN) required
(MPa) (MPa) (MPa) (MPa) (MPa) (MPa)
1 1,21 4,55 1,64 1,91 5,17 5,37 5,37 6 935 9,5

12 -4,65 -6,82 -1,37 -2,02 -6,84 -7,24 -7,24 6 -1260 12,7
23 -2,17 -4,99 -1,87 -2,30 -5,60 -5,89 -5,89 6 -1025 10,4
34 1,94 5,37 0,91 1,16 5,48 5,66 5,66 6 985 10,0
55 -4,07 -5,80 -0,34 -0,81 -5,37 -5,65 -5,80 2 -1009 10,2
76 2,23 3,98 -0,02 0,20 3,54 3,68 3,98 2 693 7,0
Left
87 -3,66 -3,40 0,65 0,28 -2,50 -2,71 -3,66 1 -636 6,4

98 -2,76 -0,07 0,97 0,72 0,84 0,71 -2,76 1 -480 4,9
109 2,09 1,20 -0,50 -0,32 0,55 0,65 2,09 1 364 3,7
110 -0,64 2,49 0,45 0,39 2,77 2,77 2,77 6 482 4,9
111 -1,77 2,30 0,71 0,55 2,88 2,81 2,88 5 501 5,1
122 1,58 -1,32 -0,38 -0,25 -1,76 -1,70 -1,76 5 -306 3,1
123 7,50 -2,10 -0,32 -0,27 -2,35 -2,35 -2,35 5 -409 4,1
1 1,21 0,33 0,32 0,65 0,23 0,39 1,21 1 211 2,1
12 -4,65 -3,53 -0,39 -1,06 -2,98 -3,36 -4,65 1 -810 8,2
23 -2,17 -0,88 -0,38 -0,87 -0,67 -0,91 -2,17 1 -377 3,8
34 1,94 1,67 0,24 0,52 1,44 1,58 1,94 1 338 3,4
55 -4,07 -3,80 -0,20 -0,67 -3,28 -3,55 -4,07 1 -709 7,2
76 2,23 1,93 0,01 0,25 1,55 1,68 2,23 1 389 3,9
Right
87 -3,66 -3,41 0,06 -0,30 -2,84 -3,06 -3,66 1 -636 6,4

98 -2,76 -2,34 0,10 -0,13 -1,91 -2,06 -2,76 1 -480 4,9
109 2,09 1,50 -0,03 0,14 1,14 1,25 2,09 1 364 3,7
110 -0,64 -1,87 -0,24 -0,26 -1,96 -1,99 -1,99 6 -346 3,5
111 -1,77 -1,62 -0,06 -0,18 -1,45 -1,55 -1,77 1 -308 3,1
122 1,58 1,14 0,14 0,24 0,98 1,06 1,58 1 275 2,8
123 0,75 1,84 0,30 0,32 1,91 1,94 1,94 6 338 3,4

Connection Steel plates – upper chord / lower chord
To calculate the resulting force of the

two diagonal actions, the static
equilibrium of the steel plate is done. chord
The resulting force is applied at an
angle to the grain. As a result, fh,α,d has
to be used to calculate the number of
dowel required.
left diagonal right
diagonal
To simplify the calculation, the value of f α,h is calculated in each case. The minimum
of all the values is used to calculate the load capacity per dowel ( f α,h,d= 25,75MPa ).
Table A-4 Calculation of the load carrying capacity of the connection
Load carrying capacity per shear plane per dowel

F1/outer= 25651N
F1/outer= 13135N
F2= 12256N
F3= 112648N
Load carrying capacity per steel plate per dowel
Outer part Inner part
F1/outer+F1/inner = 38796N
F1/outer+F3 = 38299N F1/inner+F1/inner =26269N
F1/inner+F2 = 25390N
F1/inner+F3 = 25783N F3+F3=25296N
F2+F3= 24904N
F3+F3= 25296N
Min= 24904N Min= 25296N
Characteristic load : R0,k = 150992N per dowel
Design load : R0,d = 92918N per dowel

005:103 109
Upper chord
y Y
The angle between two diagonals is 97,62º.
The components of the resulting force are x X
calculated by:
X = Fleft _ diag . sin(48,81) − Fright _ diag . sin(48,81)

+ +
Y = Fleft _ diag . cos(48,81) + Fright _ diag . cos(48,81) 48,81
Fleft diag Fright diag

Forces in the
diagonals Resulting force (chords to the
f α,h
Load steel plate) required nb
node
magnitude (kN)
case of dowels
left right X Y Magnitude (MPa)
α (deg)
diagonal member (kN) (kN) (kN)
1 -1025 935 6 -1475 -356 1517 13,6 27,82 16,3
2 -1260 985 6 -1689 -464 1752 15,4 27,55 18,9
3 -1009 693 2 -1281 -387 1338 16,8 27,31 14,4
4 -636 364 1 -753 -256 795 18,8 26,96 8,6
5 -480 275 1 -568 -193 600 18,8 26,97 6,5
5 146 -306 5 340 23 341 3,8 28,78 3,7
6 501 -409 5 685 181 709 14,8 27,63 7,6
7 211 -377 1 443 43 445 5,5 28,68 4,8
8 338 -810 1 864 36 865 2,4 28,83 9,3
9 389 -709 1 826 77 830 5,3 28,70 8,9
10 364 -636 1 753 77 757 5,8 28,66 8,1
11 275 -480 1 568 58 571 5,8 28,66 6,1
12 131 -308 1 330 14 330 2,5 28,83 3,6
12 338 -269 6 457 124 473 15,2 27,58 5,1

005:103
Lower chord
The angle between two diagonals is 103,2º.
Fleft diag
+ Fright diag
The components of the resulting force are 51,63
calculated by: +
X = Fleft _ diag . sin(48,81) − Fright _ diag . sin(48,81) y Y

Y = − Fleft _ diag . cos(48,81) − Fright _ diag . cos(48,81)
x X
Forces in the
diagonals Resulting force (chords to the
f α,h
Load steel plate) required nb
node
magnitude (kN)
case of dowels
left right X Y Magnitude (MPa)
α (deg)
diagonal member (kN) (kN) (kN)
13 935 -1260 6 -1721 202 1733 6,7 28,60 18,6
14 985 -984 6 -1543 -1 1543 0,0 28,86 16,6
14 934 -1009 2 -1524 47 1525 1,7 28,85 16,4
15 693 -591 2 -1007 -63 1009 3,6 28,79 10,9
15 389 -636 1 -804 154 818 10,8 28,19 8,8
16 364 -480 1 -662 72 666 6,2 28,64 7,2
17 -306 501 5 632 -121 644 10,8 28,19 6,9
18 -409 482 5 699 -45 700 3,7 28,78 7,5
18 482 -409 6 -699 -45 701 3,7 28,78 7,5
19 -810 211 1 800 372 882 24,9 25,75 9,5
20 -709 338 1 821 230 853 15,7 27,50 9,2
21 -636 389 1 804 154 818 10,8 28,19 8,8
22 -480 364 1 662 72 666 6,2 28,64 7,2
23 -308 275 1 457 21 458 2,6 28,82 4,9

005:103 111
Appendix C: Checking of the steel plate
H
V
Figure C-1 equilibrium of the steel plate
• Resistance of the plate in the diagonal:

Bearing resistance
f u . d 0 .t
Fb , Rd = 2,5.α . ≥ Fb
γM
1260
Fb = = 90kN / dowel
14
e p 1 34,85 100 1
α = min 1 ; 1 − ;1 = min ; − ;1 = min{0,89;2,31;1} = 0,89
3.d 0 3.d 0 4 39 3.13 4
510.13.8
Fb , Rd = 2,5.0,89. = 94.79kN / dowel ≥ Fb
1,25
O.K
Resistance of the net section
Anet = 10272mm 2
f u . Anet
Ft , Rd = 0,9 ≥ Ft
γM
Ft = 934kN
510.10272
Ft , Rd = 0,9. = 3771kN ≥ Ft
1,25
O.K

005:103
Tensile failure of the plate
f y. A
Ft ,Rd = ≥ Ft
γM
984
Ft = = 164kN / plate
6
355.(238.8)
Ft ,Rd = = 614kN / plate ≥ Ft
1,1
O.K
• Resistance of the plate in the chord:
The two components of the resultant force acting on the steel plate in the chord have
to be checked.
Verification of the horizontal component H=1474kN.
Bearing resistance
f u . d 0 .t
Fb , Rd = 2,5.α . ≥ Fb
γM
1474
Fb = = 70kN / dowel
21
e p 1 50 100 1
α = min 1 ; 1 − ;1 = min ; − ;1 = min{1,28;2,31;1} = 0,89
3.d 0 3.d 0 4 39 3.13 4
510.13.8
Fb , Rd = 2,5.1. = 106kN / dowel ≥ Fb
1,25
O.K
Anet = 18254mm 2
f u . Anet
Ft , Rd = 0,9 ≥ Ft
γM
Ft = 1474kN
510.18254
Ft , Rd = 0,9. = 6702kN ≥ Ft
1,25
O.K

f y. A
Ft , Rd = ≥ Ft
γM
1474
Ft = = 245kN / plate
6
355.(428,3.8)
Ft , Rd = = 1105kN / plate ≥ Ft
1,1
O.K
Verification of the vertical component V=464kN.
Bearing resistance
f u .d 0 .t
Fb , Rd = 2,5.α . ≥ Fb
γM
464
Fb = = 22kN / dowel
21
e p 1 50 100 1
α = min 1 ; 1 − ;1 = min ; − ;1 = min{1,28;2,31;1} = 0,89
3.d 0 3.d 0 4 39 3.13 4
510.13.8
Fb , Rd = 2,5.1. = 106 kN / dowel ≥ Fb
1,25
O.K
Anet = 16896mm 2
f u . Anet
Ft , Rd = 0,9 ≥ Ft
γM
Ft = 464kN
510.16896
Ft , Rd = 0,9. = 6204,2kN ≥ Ft
1,25
O.K

005:103
f y. A
Ft ,Rd = ≥ Ft
γM
464
Ft = = 77,3kN / plate
6
355.(400.8)
Ft ,Rd = = 1032kN / plate ≥ Ft
1,1
O.K

Appendix D: Verification of the members
C
D
12 B
76
34
A
Table D-1 Verification of the chord members in the mid-span
Compressive Bending
Max force Max moment Combined
Stresses in stresses in
in N in N.m stress
MPa MPa
LC1 3,84.106 9,45 2,37.105 5,55 0,63 <1
LC2 3,92.106 9,63 2,66.105 6,23 0,66 <1
LC5 3,31.106 8,15 2,31.105 5,41 0,55 <1
LC6 3,65.106 8,97 2,53.105 5,93 0,61 <1
Combined bending and tension Member B
Tensile Bending
stresses in stresses in
in N in N.m stress
MPa MPa
LC2 1,09.106 2,67 1,58.105 3,70 0,35 <1
LC5 1,39.106 3,42 1,65.105 3,87 0,41 <1
LC6 1,33.106 3,26 1,66.105 3,89 0,40 <1
Combined bending and compression Member C
Compressive Bending
in N in N.m stress
MPa MPa
LC2 4,74.106 11,7 2,02.105 4,73 0,73 <1
LC5 4,69.106 11,5 1,82.105 4,27 0,72 <1
LC6 4,89.106 12,0 1,89.105 4,43 0,75 <1

005:103
Combined bending and compression Member D
Compressive Bending
in N in N.m stress
MPa MPa
LC5 4,41.106 10,9 2,25.105 5,27 0,70 <1
LC6 4,50.106 11,1 2,32.105 5,44 0,72 <1
Table D-2 Verification of the chord member at the level of the connection
Compressive Bending
in N in N.m stress
MPa MPa
LC1 3,84.106 10,3 1,71.105 4,37 0,52 <1
LC2 3,92.106 10,5 1,62.105 4,14 0,52 <1
LC5 3,31.106 8,90 1,35.105 3,45 0,40 <1
LC6 3,65.106 9,79 1,49.105 3,81 0,46 <1
Combined bending and tension Member B
Tensile Bending
in N in N.m stress
MPa MPa
LC2 1,09.106 2,92 7,10.104 1,82 0,28 <1
LC5 1,39.106 3,73 5,10.104 1,30 0,31 <1
LC6 1,33.106 3,56 5,70.104 1,46 0,30 <1
Combined bending and compression Member C
Compressive Bending
in N in N.m stress
MPa MPa
LC2 4,74.106 12,7 5,60.104 1,43 0,54 <1
LC5 4,69.106 12,6 4,50.104 1,15 0,52 <1
LC6 4,89.106 13,1 4,10.104 1,05 0,55 <1

Combined bending and compression Member D
Compressive Bending
in N in N.m stress
MPa MPa
LC5 4,41.106 11,9 7,30.104 1,87 0,50 <1
LC6 4,50.106 12,1 8,40.104 2,15 0,53 <1
Table D-3 Verification of the diagonal members in the mid-span
Combined bending and compression Member 12
Compressive Bending
in N in N.m stress
MPa MPa
LC1 9,02.105 5,18 0,00 0,00 0,28 <1
LC2 1,37.106 7,84 0,00 0,00 0,42 <1
LC5 1,40.106 8,04 0,00 0,00 0,44 <1
LC6 1,47.106 8,46 0,00 0,00 0,46 <1
Combined bending and tension Member 34
Tensile Bending
in N in N.m stress
MPa Pa
LC2 1,03.106 5,94 0,00 0,00 0,39 <1
LC5 1,39.106 7,97 0,00 0,00 0,52 <1
LC6 1,15.106 6,61 0,00 0,00 0,43 <1
Combined bending and tension Member76
Tensile Bending
in N in N.m stress
MPa MPa
LC1 4,36.105 2,50 0,00 0,00 0,13 <1

Table D-4 Verification of the diagonal members at the level of the connection
Combined bending and compression Member 12
Compressive Bending
in N in N.m stress
MPa MPa
LC1 9,02.105 5,65 1,60.104 2,23 0,20 <1
LC2 1,37.106 8,55 2,80.104 3,90 0,40 <1
LC5 1,40.106 8,77 3,00.104 4,18 0,43 <1
LC6 1,47.106 9,23 3,00.104 4,18 0,45 <1
Combined bending and tension Member 34
Tensile Bending
in N in N.m stress
MPa Pa
LC2 1,03.106 6,48 7,00.103 9,75.105 0,47 <1
LC5 1,39.106 8,70 6,00.103 8,36.105 0,61 <1
LC6 1,15.106 7,22 6,00.103 8,36.105 0,51 <1
Combined bending and tension Member76
Tensile Bending
in N in N.m stress
MPa Pa
LC1 4,36.105 2,73 1,60.103 2,23.105 0,19 <1

Untitled PDF

Uploaded by

Copyright:

Available Formats

Untitled PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Untitled PDF

Uploaded by

Copyright:

Available Formats

Timber trussed arch for long span

Master’s Thesis in the International Master’s Programme in Structural Engineering

Timber trussed arch for long span

Master’s Thesis in the International Master’s Programme in Structural Engineering

Department of Civil and Environmental Engineering

Master’s Thesis in the International Master’s Programme in Structural Engineering

©AUDREY FARREYRE & JEAN-BERNARD JOURNOT, 2005

Master’s Thesis 2005:103103

Department of Civil and Environmental Engineering

CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2005:103 I

CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2005:103 III

2 GLUED LAMINATED TIMBER- GLULAM 4

3 OPTIMUM STATIC SYSTEM 13

CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2005:103 V

4 PRELIMINARY STRUCTURAL ANALYSIS 47

VI CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2005:103

7 CONCLUSION AND RECOMMENDATIONS 101

CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2005:103 VII

Göteborg December 2005

CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2005:103 IX

Roman lower case letters

a1 Spacing of dowels within one row parallel to the grain

X CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2005:103

CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2005:103 XI

This subject was proposed by Roberto Crocetti, who is an engineer at Moelven

CHALMERS, Structural Engineering and Mechanics, Master’s Thesis 2005:103 1

Figure 1.2 Sketch of a trussed arch

2 CHALMERS, Structural Engineering and Mechanics, Master’s Thesis 2005:103

The main purpose of chapter 3 is to find an optimum static system. Different

The next step presented in chapter 4 consists in performing a preliminary analysis of

CHALMERS, Structural Engineering and Mechanics, Master’s Thesis 2005:103 3

2.1 Production process

4 CHALMERS, Structural Engineering and Mechanics, Master’s Thesis 2005:103

2.1.3 Standard shapes

CHALMERS, Structural Engineering and Mechanics, Master’s Thesis 2005:103 5

2.2 Material properties

Figure 2.4 Example of Glulam cross-sections

2.2.2 Composed glulam GL32C

6 CHALMERS, Structural Engineering and Mechanics, Master’s Thesis 2005:103

Figure 2.5 Norwegian forest

2.3.1 Hamar Olympiahall, Norway

Figure 2.6 Hamar olympiahall

CHALMERS, Structural Engineering and Mechanics, Master’s Thesis 2005:103 7

Table 2.1: Main information about the Hamar Olympic Hall

Name of the project Hamar Olympic Hall-Viking Ship

Architects NIELS TORP

Total height 37m

Max. span 96,4m

Depth of the truss 3,45m (from center line to center line)

Angle between the diagonals 40º

Dimension of the cross-section

Diagonal members 420x300mm2

Chord members 570x566mm2

Type of connections Multiple steel plates connections

Location of the hinges

Abutment Lower chord

Crown Lower chord

8 CHALMERS, Structural Engineering and Mechanics, Master’s Thesis 2005:103

Figure 2.8 Håkons Hall