EDITION 3:2022

Design of timber structures

Structural aspects of
timber construction
Vo ume

SWEDISH
WOOD
 Design of timber structures
Structural aspects of
timber construction Volume 1

Design of timber structures

Volume 3

EDITION 3:2022

Design of timber structures

Rules and formulas
according to Eurocode 5
Volume 2 J?

J SWEDISH
/If WOOD

: SWEDISH
8? WOOD

Design of timber structures Volumes 1-3 are adapted to Eurocode 5, Eurocode 0

and Eurocode 1.

• Volume 1: Structural aspects of timber construction

• Volume 2: Rules and formulas according to Eurocode 5
• Volume 3: Examples
Preface
This is the third revised edition of Design of timber structures Volume 1, Structural
aspects of timber construction published in 2015. Rules and standards change in
pace with the development of society, hence a publication of this type needs to
be reviewed regularly. Contrary to earlier English editions, the current version is
not adapted to the Swedish national choices in Eurocode 1995. Instead, the origi­
nal Eurocode 1995 is referred to, in order to simply adaptation to different
national parameters in the country using the book.

The book series Design of timber structures Volume 1-3 has been produced to make
it easier for structural designers to calculate timber structures and it is adapted
to Eurocodes. It is being used for higher education at universities and institutes.

The book series Design of timber structures Volume 1-3 includes Volume 2, Rules and
formulas according to Eurocode 5 as well as Volume 3, Examples. All three books are
available in English and Swedish. Since the books are available in both lan­
guages and due to the nuanced content, our goal is that they will play a role for
many users on different skill levels. Notice that the Swedish decimal holder “,”
was kept throughout all the text, instead of the English “.”. All photos are taken
in Sweden, unless otherwise indicated.

The authors of Volume 1 are active in educating engineers at universities and

institutes of higher education in Sweden, as Chalmers University of Technology,
Royal Institute of Technology, Linnaeus University in Vaxjo, Lulea University of
Technology and Lund University. Also RISE Research Institutes of Sweden has
contributed to the content.

Stockholm, April 2022

Patrice Godonou
Swedish Wood

Design of timber structures - Volume 1 3

 Design of timber structures - Volume 1

Contents

Introduction to design Design of timber joints 87

and design process 7 4.1 Dowel-types 88

1.1 General introduction to structural design 7 4.2 Shear capacity of single dowels 89

1.2 Life phases of a building project 7 4.3 Johansen theory - timber-timber joints
and panel-timber joints 91
1.3 Various requirements related to timber structures 12
4.4 Steel-to-timber joints 95
1.4 Eurocodes - general assumptions,
limit states and detailed design 15 4.5 Expressions for the resistance of
a single dowel 97
1.5 Concepts used for the limit state design
of timber and wood-based products 20 4.6 Tensile capacity of single dowels - rope effect 100

4.7 Combined loading 103

Structural properties 4.8 Joints depending solely on tensile capacity 103
of sawn timber and 4.9 Brittle failure modes and group effects
engineered wood products 23 in dowelled joints 104
2.1 Forestry and the production of sawn timber 23 4.10 Forces acting at an angle to the grain 107

2.2 Structure of timber 26 4.11 Punched metal plate fasteners 107

2.3 Physical properties of wood 31 4.12 Glued joints 109

2.4 Mechanical properties of wood and timber 33 4.13 Glued-in rods 111

2.5 Strength grading 42 4.14 Creating built-up column sections 112

2.6 Modified wood 46 4.15 Load distribution in joints 112

2.7 Engineered wood products 47 4.16 Stiffness of joints 114

2.8 End-user requirement on sawn timber 53

Composite timber elements 115

Design of structural timber elements 5.1 Structural elements

in ultimate limit state 54 with full composite action 116

3.1 Prismatic beam elements 54 5.2 Structural elements

with partial composite action 142
3.2 Design of slender members 62

3.3 Special timber elements 75

3.4 Portal frames 84

3.5 Arches 86

Design of timber structures - Volume 1 5

Design of timber structures - Volume 1

Horizontal stabilization 160

6.1 General considerations 160

6.2 The importance of bracing 163

6.3 Fundamental statics 164

6.4 Strength and stiffness requirements

for bracing systems 167
6.5 Typical bracing systems
for heavy timber structures 174

6.6 Special topics 176

6.7 Bracing of timber framed buildings 178

Design for serviceability (SLS) w

7.1 Performance requirements 183

7.2 Static deformation 185

7.3 Vibration 193

Timber building systems

for housing 203
8.1 Structural systems overview 205

8.2 Functional requirements and system solutions 211

8.3 Detailed design 217

Structural systems
for infrastructure 228
9.1 Timber bridges 228

9.2 Other exterior applications 236

9.3 Design for durability, wood protection 239

Symbols 242

Bibliography 246

Non-liability and copyright 253

Publications and websites from Swedish Wood 255

6 Design of timber structures - Volume 1

 1.1 General introduction
to structural design

Introduction to design
and design process
Robert Kliger

1.1 General introduction 1.1 General introduction to structural design 7

1.2 Life phases of a building project

to structural design 1.2.1 Conceptual design 8
7

1.2.2 Detailed design by verification 10

Structural design involves choosing the dimensions of load-bearing 1.2.3 Construction 11
members and modelling the load-bearing structure with regard to 1.2.4 Inspection and maintenance 11
1.2.5 Demolition 12
the requirements set for material resistance, performance and dura­
bility during the service life of the structure. 1.3 Various requirements related to
Structural design is based on verification. The aim of this verifica­ timber structures 12
tion is to show that the actual requirements are fulfilled for the cho­ 1.3.1 Stakeholders 12
sen material, dimensions and structural system. This verification can 1.3.2 Requirement categories 13
be achieved by calculation or testing, or by a combination of the two. 1.3.3 Requirements set by society - Eurocodes 14
1.3.4 Requirements set by customers
The design working life is the period for which it is assumed
and end users 14
a structure is going to be used for its intended purpose, with antici­
pated maintenance but without major repair being necessary. A design 1.4 Eurocodes - general assumptions,
working life of 50 years for buildings and 100 years for bridges is limit states and detailed design 15
often recommended. 1.4.1 General assumptions and verifications 15
1.4.2 Principles for limit state design 16
1.4.3 Verification based on the partial

1.2 Life phases of 1.5

safety factor method 17

Concepts used for the limit state design

a building project of timber and wood-based products 20
1.5.1 Load duration classes 21
1.5.2 Effect of moisture content
The life of a building project, irrespective of whether it is a house,
and service classes 21
industrial site or part of an infrastructure such as a bridge, can be
1.5.3 Partial factors for material properties
divided into different phases, see figure 1.1. These phases include and adjustments using various
design, construction, use of a building (working life) including main­ modification factors 22
tenance and repairs and, finally, at the end of working life of a struc­ 1.5.4 Difference in material response when the loads
ture, demolition. are applied in various directions in relation to
Conceptual design is the first stage of a building project. During the grain orientation of timber 22

this stage, needs are identified and examined, requirements for

potential solutions are defined, potential solutions are evaluated and
a suitable structural concept for further design is developed. Once
the concept has been chosen and well defined, the second stage of
the design process is reached by verification according to a required
design code. The second stage comprises preliminary and detailed
design. Forces loading the foundation, for example, are obtained

Figure 1.1 Phases of a project life, Niemeyer (2003)

Design of timber structures - Volume 1 7

1.2 Life phases of a building project

after the preliminary design, which makes it possible to design

the foundation at the same time as all the details in the main struc­
ture are designed. All possible design problems must be resolved
before the construction phase starts. Once the construction phase is
completed and a final inspection is made, a structure is ready for use
as it was intended. Inspection and maintenance are important phases
of the working life of all building projects. Different designs, build­
ing methods, materials and environmental exposure require different
amounts of maintenance during the working life of a structure. All
building components require some form of maintenance. The prema­
ture failure of some parts or a change in the intended use of a struc­
ture requires structural assessment and creates a need for repair or
strengthening.
New structural designs, assessments of remaining life, strengthening
and repair are common measures during the working life of a struc­
ture. At the end of the working life, a demolition phase is reached.
In this phase, it is important that materials used in the structure are
easy to dismantle, recycle and/or re-use.

1.2.1 Conceptual design

There are at least two different approaches describing conceptual
design; one is by Kroll et al. (2001) and one by Engstrom (2002).
The approach advocated by Kroll et al. is regarded as theoretical, but
it is very useful when it comes to understanding problem solving and
the creation of innovative solutions. The approach adopted by
Engstrom is more practical and a short summary is presented here.
A comparison of both methods is presented in figure 1.2. The concep­
tual design approach can be applied both at the beginning of a build­
ing process and during the working life of a structure in a process of
repair, strengthening and maintenance.

Kroll et al. (2001) Engstrom (2002)

• Needs identification • Intuitive phase
• Analysis of the task specification of goals
• Needs analysis
• Requirements, design criteria

• Generating ideas alternative concepts

• Design requirements
• Preliminary evaluation promising alternatives
• Technology identification
• Evaluation phase
• Weighting of design criteria
• Parameter analysis
• Risk analysis
• Parameter identification
• Final evaluation winning alternative
• Creative synthesis
• Preliminary sizing
• Evaluation
• Check of difficult parts
• Modifications final solution

Figure 1.2 Main issues applied in a conceptual design

8 Design of timber structures - Volume 1

 1.2 Life phases of a building project

Intuitive phase
This phase includes the definition of specific goals for the project,
understanding the client’s wishes and transforming them into a list
of requirements, understanding the design criteria as specified by
society via current design codes, generating simple concept solutions
for various ideas and the preliminary evaluation of each idea.
Understanding the client’s wishes and collecting all the necessary
information from possible stakeholders, which may affect the project,
is a crucial part of the intuitive phase. Literature studies, a search for
inconsistencies and interviews with users and other stakeholders
may provide the information necessary to define the aims and set
a list of priorities comprising different requirements. The design
requirements should be as clear as possible and should include issues
such as performance, quality parameters, reliability, safety, product
life span, aesthetics, ergonomics, economy and maintenance.
The main aim of the systematic search is to generate a wide range of
possible design concepts. A combination of intuitive and systematic
methods is probably the best way to generate new concepts. Creative
and innovative ideas are based on good theoretical and practical
knowledge. Various ideas should be represented by different struc­
tural materials or combinations of materials, a variety of structural
systems and production methods. Such ideas are created by a system­
atic search and are produced within the intuitive thinking process.
It is always advantageous to produce too many concept ideas rather
than too few. The main concern for a preliminary evaluation is to
reduce the number of concepts. This can be achieved by discussions
or by using preliminary evaluation matrices. It is important to have
a well-documented motivation of a solution that was made and per­
haps be able to backtrack to this step in the following process. Strandparken, Sundbyberg.

Evaluation phase and choice of the final concept

The evaluation phase includes ranking different concepts, perform­
ing a simplified risk analysis of each concept and making a final eval­
uation of concepts. Ranking can be done by weighing up objectives
and various requirements in order to make some kind of systematic
comparison of different concepts. A matrix showing the grades of
various requirements in relation to different concepts is produced.
The concept with the highest grade will obtain the highest ranking.
A risk analysis is an important part of the conceptual design. All pos­
sible risks should be listed for two or three concepts with the highest
ranking. These risks may include construction, transportation and
production risks, economic risks, accidental risks or weather risks,
for example. The result of the risk analysis should be included in
the final evaluation of some of the most promising concepts.

Preliminary sizing
The aim of preliminary sizing is to estimate the preliminary dimen­
sions of the load-bearing members in a building, such as trusses,
beams, columns, walls and floors. The thickness of the insulation
may be a decisive factor when it comes to the thickness of structural
members. The preliminary sizing often includes obtaining relevant
load combinations in order to estimate sectional forces acting on
a structural part which is being analysed.

Design of timber structures - Volume 1 9

1.2 Life phases of a building project

Estimations of section sizes are based on:

• experience and recommendations
• preliminary design forces (including interaction)
• preliminary design stresses
• preliminary deflections.

Rough estimations may be based on:

• rule of thumb
• previous experience
• simple analytical tools
• simple design software
• tables, diagrams and charts.

An example of a rule of thumb that applies for a timber beam with

a rectangular cross-section is that, if the ratio between depth and
width (h/b) > 6, full lateral support must be provided for the com­
pression edge of the beam and the beam must be supported at
the bearing points so that rotation is prevented.

1.2.2 Detailed design by verification

Outcomes and process
During the detailed design phase, all the key design decisions are
finalised. As a result of the detailed design, the following outcomes
are achieved:
• The structure is fully and unambiguously defined and validated.
• All major building subsystems are defined.
• All parts are fully engineered and co-ordinated.
• Costs and construction methods are established to a high level
of precision.
• Agreement is reached on tolerances between the companies
involved to ensure constructability, prefabrication and
transportation.
• Quality levels are established.
• Prescriptive specifications are completed.

The structural design process is regulated by the pertinent design

code, which applies to the geographical location where a structure is
going to be built. In different countries, the design codes may be
based on various principles, such as limit state design, reliability
design or allowable stress design, for example. In most European
countries, limit state design applies and all stages of this design pro­
cess are defined in a series of Eurocodes. The same design code must
be used when a structural design is produced for an entire structure.

Verification
Verification is the main part of structural design. The verification of
the load-bearing capacity of a structure or part of it is conducted in
the ultimate limit state, while the verification of whether it functions
is conducted in the serviceability limit state. To verify a structure in
the ultimate limit state, a designer needs to create structural and
load models by using design values for actions, material or product
properties and geometrical data. Load cases should be selected by
identifying load arrangements and possible deviations from assumed
directions and positions of actions and sets of deformations and
imperfections that should be considered simultaneously.

10 Design of timber structures - Volume 1

 1.2 Life phases of a building project

1.2.3 Construction
Process
The construction process usually involves the translation of designs
into the real production of a building. The design usually consists of
drawings and specifications, usually prepared by a design team
including surveyors, civil engineers, cost engineers (or quantity sur­
veyors), engineers for electrician, structural design, plumbing and
fire protection.
The increasing complexity of construction projects is creating
a need for design professionals trained in every phase of the project
life cycle and the development of an appreciation of the building as
an advanced technological system requiring the close integration of
many subsystems and their individual components, including
sustainability.
Building Information Modelling (BIM) is the process of generating
and managing building data during the life cycle of a building.
Typically, it uses three-dimensional, real-time, dynamic building
modelling software to increase productivity in building design and
construction. The process produces a Building Information Model
Glulam frame assembly for the Infanterimuseum, Karlstad.
(BIM), which includes building geometry, spatial relationships, geo­
graphical information as well as quantities andproperties of building
components.
A BIM is a way to handle information which is created during
the design and construction process but also during management,
that is all the way from conceptual design to construction and
through the service life. Modelling makes it possible to handle and
re-use the relevant information about the building.

1.2.4 Inspection and maintenance

Assessment
Structures are usually designed for a service life of 50 to 100 years.
Changes in load characteristics, deterioration with age, environmen­
tal effects and random actions may cause local or global damage to
the structure. Any structural system in service is subject to agerelated
deterioration, leading to potential concerns regarding maintenance,
health & safety, environmental and economic implications. The con­
dition assessment of aged structures, see for example Paik and
Melchers (2008), is an important part of engineering work, which
includes:
• inspection and maintenance of aged structures
• residual strength assessment of deteriorated structures
• reliability and risk assessment of aged structures
• models to predict structural deterioration.

The main steps in a condition assessment should be:

a) to record the damage, if any, and find the causes of distress
b) to assess the extent of distress and to estimate the residual strength
of structural components and the system including the foundation
c) to plan the rehabilitation and retrofitting/strengthening of a struc­
ture when needed.

Detailed visual inspection and Non Destructive Testing (NDT) play

an important role in condition assessments of existing structures. If
the structure is damaged by a natural disaster or by an accidental

Design of timber structures - Volume 1 11

1.3 Various requirements
related to timber structures

load, a great deal of expertise is required to interpret the field obser­

vationsand test results to make a proper assessment of the condition,
as well as analysing and evaluating safety.

Refurbishment and repair

The nature of refurbishment and repair is markedly different from
new construction work, as it is necessary to work within the con­
straints of a pre-determined situation. It is essential for those
involved in this type of work to acquire an intimate knowledge of
the structure that is going to be repaired. The documentation from
the original design, the definition of the alternative use and docu­
mentation from the assessment of the structure define the new situa­
tion and provide the basis for design of the new construction work.

1.2.5 Demolition
The demolition phase is the final stage in the service life of a struc­
ture. Today, there is still a real opportunity for improvement and for
reducing the environmental impact of the building sector when it
comes to re-using and recycling building materials. The built
environment makes an important contribution to the total national
environmental impact. Government policies are expected to play
an important role in improving the environmental performance of
the building sector. The energy and material use in construction and
during the service life of a structure represent a large amount of
the total energy and material consumption. By using buildings func­
tionally for as long as possible and by aiming during demolition for
the maximum re-use of materials and the adequate treatment of
the waste streams, a substantial environmental gain may be realised.
In this respect, timber structures may have a very large advantage in
comparison to other building materials. Demolition projects today
are more complicated if sustainability is incorporated into this stage.
The preparatory phase from policy to initiative must be well
described. Already during the conceptual design, decisions are made
which often have a major impact on the entire service life and sus­
tainability of a structure, including the demolition stage.

1.3 Various requirements

related to timber structures
1.3.1 Stakeholders
The most important parties are: a) society, through codes b) insurance
companies c) commissioners of buildings (representing building own­
ers or clients) d) contractors, builders or carpenters e) producers of
prefabricated elements f) suppliers of timber products, see table 1.1,
page 13. Society, insurance companies and building commissioners
set requirements for the components of the entire building, while
a building contractor or builder sets requirements for the timber
products. Society sets regulations to ensure a sufficiently low risk of
collapse during the time the building is in use. The serviceability or
functional requirements are often not regulated in detail.
The demands are general and it is up to the commissioners of build­
ings to formulate more precise requirements. The building commis­
sioner gives the contractor the job of constructing a building and

12 Design of timber structures - Volume 1

 1.3 Various requirements
related to timber structures

then expects a certain function from the building components, often

with no regard to the choice of material or design. A floor must fulfil
deflection and vibration standards, no matter how this is achieved,
for example. The contractor must satisfy both the commissioner’s
functional requirements and society’s safety regulations. The floor
joists must have a certain stiffness to meet the deflection require­
ments. The contractor imposes certain demands on the products.
The properties of the products should permit swift and efficient
production.

1.3.2 Requirement categories

The requirements set for timber or wood-based products can be
divided into the following categories: safety, function, desirable prop­
erties and irrelevant characteristics, see table 1.1. The safety require­
ments are set by society to prevent failure and damage which might
cause personal injury or heavy financial costs. Public perception does
not accept fatalities and injuries due to structural failure (in our
homes, during working time, during recreation activities and other
activities) during the design working life of a structure compared
with fatalities arising from other hazards and events. As a result,
structural safety is the principal requirement in all codes dealing
with structural design. The functional requirements are directly asso­
ciated with the user’s needs and expectations.
The desirable properties are those that give the product a greater
value for the user, apart from the basic needs. The functional
requirement limits are often diffuse, since they are often related to
comfort and human perception, which are difficult to quantify.
The irrelevant characteristics represent those properties that do not
affect the function of a product. For example, the building contractor
does not generally care about the colour, number or size of knots in
wall studs.

Table 1.1 Matrix of requirements from stakeholders with examples at different levels, Johansson et al. (1990)

Interested Requirements Requirements Desirable Irrelevant

parties based on safety based on function properties characteristics

Society, codes Safety of the structure. Function in general terms - -

and standards Safety from personal injury.
Design methods. Eurocodes

Insurance Load-bearing capacity Function in general terms. - -

companies during lifetime of Stiffness, durability, fire,
the structure. Design load-bearing capacity.
methods Comply with standards

Building Comply with the design Function in specific terms. Aesthetic considerations, Choice of material
commissioner standards Dimension, shape, durability, “warm” surfaces, shape,
User smell, load-bearing capacity, easy to paint, repaint
stiffness, vibrations and dismantle

Contractor, builder, Stiffness Dimension, shape, stiffness, Right length, tolerant Species, discoloration
carpenter smell, load-bearing capacity, to moisture, no and appearance of
moisture content, durability, impregnation. built-in timber
mould, rot, wane, weight, Right price
location of knots, ability to
nail, screw and glue

Design of timber structures - Volume 1 13

1.3 Various requirements
related to timber structures

1.3.3 Requirements set by society

- Eurocodes
General concepts
The fundamental requirements set by the European standard EN 1990
for the reliability of construction works include: structural safety,
robustness and serviceability.

Structural safety: A structure should be designed and executed in

such a way that it will, during its intended life with appropriate
degrees of reliability, and in an economic way, sustain all actions
likely to occur during execution and use. This deals with the safety of
people, the structure and its contents.

Robustness: A structure should be designed and executed in such

a way that it will not be damaged by events such as:
• explosions
• impact
• consequences of human errors

to an extent disproportionate to the original cause.

Serviceability: A structure should be designed and executed in such

a way that it will, during its intended life with appropriate degrees of
reliability, and in an economic way, remain fit for the use for which
it is required. This deals with the function, comfort and appearance
of the structure.

Durability
The durability of a structure or a part of it assumes to have a certain
length of working life based on the fact that appropriate maintenance
is given and that it remains fit for use during the design working life.
The structural design is suitably adapted to the surroundings of
the structure or its parts. Load-bearing parts should be designed in
such way that deterioration will not impair the durability and perfor­
mance of the structure exposed to the anticipated level of mainte­
nance. Guidance for design with regard to working life of a structure
is given in Thelandersson et al. (2012).

1.3.4 Requirements set by

customers and end users
The requirements set by customers and end users of timber products
differ for various products and uses. It is very important to distin­
guish between indoor and outdoor environmental conditions,
between load-bearing and non-load-bearing members and between
different construction methods.

Function, buildability, durability and maintenance

These requirements differ between building clients and users on one
hand and contractors, carpenters and builders on the other hand, see
table 1.1, page 13. For the first group, the types of building material
are not important, as long as they serve their function. Dimension,
shape, stiffness, vibration, load-bearing capacity, smell and durability
are the most important quality parameters to fulfil the function and
maintenance of the building, irrespective of the structural material.

14 Design of timber structures - Volume 1

 1.4 Eurocodes - general assumptions,
limit states and detailed design

For the second group, if timber is used as a structural material,

the shape of timber at the moisture content used during the con­
struction and the ability to nail, screw and glue, if necessary, are very
essential properties to facilitate more rational construction.
The moisture content in itself is not a problem for a building contrac­
tor. However, the effect of too high moisture content or too much
variation in moisture content can have a very negative impact on
the function, buildability, durability and maintenance of a structure.
The amount of knots in the timber is usually not a problem for
a building contractor. However, if the knots are situated in a place
where a piece of timber needs to be spliced or connected in
a mechanical type of connection, such as nails or screws, this may
create a problem for a carpenter.

1.4 Eurocodes - general

assumptions, limit states
and detailed design
1.4.1 General assumptions and verifications
The general assumptions for any structural design process are well
specified in EN 1990. Examples of these assumptions include:
• The choice of the structural system and the design of the structure
is made by appropriately qualified and experienced personnel.
• The execution is carried out by personnel with the appropriate
skills and experience.
• Adequate supervision and quality control is provided in design
offices and during the execution of the work, that is factories,
plants and on site. Vasaplan, Umea.

• The construction materials and products are used as specified in

EN 1990 to EN 1999, in the relevant execution standards, in refer­
ence material and product specifications, see figure 1.3.
• The structure is adequately maintained.
• The structure is used in accordance with the design assumptions.

Structural safety, serviceability

and durability

Actions on structures

De Design and detailing

Geotechnical and
seismic design

Figure 1.3 Links between Eurocodes

Design of timber structures - Volume 1 15

1.4 Eurocodes - general assumptions,
limit states and detailed design

Requirements EN 1990 establishes principles and requirements for safety, servicea­

Serviceability Safety bility and durability of structures, describes the basis of their design
Durability and verification and gives guidelines for related aspects of structural
reliability. EN 1990 is also applicable for the structural appraisal of
existing structures, developing the design of repairs and alterations
Design Situations or assessing changes of use. An acceptable degree of reliability for
Agents the majority of structures is expected to be obtained by design and
Actions execution according to Eurocodes 1 to 9, together with appropriate
Combinations of Actions quality assurance measures. The Eurocode for the design and detail­
Effects ing of timber structures is EN 1995.
Eurocodes are based on limit state design. A limit state defines con­
ditions beyond which the structure no longer satisfies the relevant
Ultimate Limit States Serviceability Limit States performance requirements. These conditions are classified into ulti­
mate and serviceability limit states. Ultimate limit states (ULS) relate
Figure 1.4 Requirements and terminology used to safety; states associated with collapse or with other forms of struc­
in the design procedure tural failure. Serviceability limit states (SLS) relate to those states in
which the structure, although standing, behaves in an unsatisfactory
fashion for example due to excessive deformation or vibration.
The verification or design procedure is illustrated in figure 1.4.
To verify a structure or part of it, a designer has to distinguish
between design situations, see table 1.2, agents, actions, combination
of actions and their effects, cf. figure 1.4. Examples of agents are
gravity, wind, snow, solar radiation, earthquakes and so on. Examples
Table 1.2 Design situations and their need for verification
of actions are load, pressure, temperature, ground acceleration and
Design situations Verifications so on. Combinations of actions are actions likely to occur simultane­
Persistent Normal use ULS, SLS ously. Finally, effects on the structure from combinations of actions
may be stress, internal force and moment, rotation, displacement
Transient Execution, temporary ULS, SLS
conditions, applicable and so on.
to the structure, e.g. When verification is performed by calculations, this often means
maintenance or repair using well-established design models which reflect the real mechani­
Accidental Normal use ULS cal response of the designed structure in the current limit state.
During execution ULS However, it is frequently not possible to imitate exactly the real
structure or real load magnitude or its variation. As a result, design
Seismic Normal use ULS, SLS
models are often simplified, depending on the level of accuracy to be
During execution ULS, SLS
achieved or the stage in the design process.

1.4.2 Principles for limit state design

The limit state design is based on a statistical approach, with
an assessment of the probability of reaching a given limit state, and
on establishing an acceptable maximum level of that probability for
design purposes. Consider the special case when the limit state of fai­
lure can be described in terms of load effect E and resistance R so that
failure will occur if E > R (the so-called fundamental case). The load
effect E usually corresponds to a maximum value during a reference
period T (often taken as one year for time variable loads). Both E and
R are random variables. The safety margin defined as Z = R - E is also
a random variable, so that the structure is regarded as safe if Z > 0. If
it is assumed that both R and E are normally distributed with mean
values p.R and p.E as well as standard deviations oR och oE, the probabi­
lity of failure Pf can be expressed as:

16 Design of timber structures - Volume 1

 1.4 Eurocodes - general assumptions,
limit states and detailed design

Table 1.3 Relationship between probability of failure Pf and safety index 0

p 0,00 1,28 2,32 3,09 3,72 4,27 4,75 5,20 5,60

Pf 0,5 10-1 10-2 10-3 10-4 10-5 10-6 10-7 10-8

1. = jp(z<o) = 0[O-azz]/o-z = 0(-/3) 1.1

where & is the standardised normal distribution function,

pZ = pR - pE and oZ2 = oR2 + oE2. p = p.Z/ oZ is the so called safety index.
The safety index p is often used as a measure of safety level and is
related to the same reference period T as is used for the load effect E.
The relationship between Pf and p is shown in table 1.3. When p
increases, the probability of failure decreases, that is the safety level
becomes higher.
Structural codes based on limit state design usually define a formal
safety level in terms of a minimum target safety index ptarget (or maxi­
mum permissible probability of failure). For Eurocode, the general
target safety index is set to 4,7 for a reference period of one year and
3,8 if a reference period of 50 years is used. In national applications,
many countries allow ptarget to be a function of the expected conse­
quences of failure, defined by reliability classes.
This means, for example, that lower values for ptarget may be used in
cases where the consequences of a failure with respect to human life
are regarded as small and vice versa.

1.4.3 Verification based on

the partial safety factor method
In normal engineering design, an evaluation based on safety index is
not practical. Instead the so called partial safety factor method is
used. Referring to the fundamental case in ultimate limit state intro­
duced in section 1.4.2, page 16, it has to be verified that the design
resistance Rd is larger than the design load effect Ed for the structural
element considered, that is:

d E.d
R.2> 1.2

This requirement has to be verified for each failure mode and a num­
ber of different load combinations depending on the design situa­
tion. The design value Ed of the load effect is determined on the basis
of permanent actions G, time-variable actions Q and accidental
actions A. The design value Gd of permanent actions is determined by:

Ga = rGGk 1.3

where yG is partial safety factor for permanent load G and Gk is

the characteristic value for permanent actions, typically defined as
the 50 % quantile or the mean value. Design values for variable
actions are defined by:

2d - Zq ' 2k 1.4

where yQ is partial safety factor of variable action Q and Qk is

the characteristic value of variable action Q, typically defined as
the 98 % quantile of the distribution for annual maxima of Q.

Design of timber structures - Volume 1 17

1.4 Eurokoder - generella antaganden,
granstillstand och detaljutformning

The design value Rd for resistance is mainly determined on the basis

of material parameters and dimensions. The design value fd of
a strength parameter f is determined from:

Td = 77— 1-5
/m

where ym is a partial safety factor for the material, fk is the character­

istic value for the material property f, typically the 5 % quantile and
n is a factor accounting for differences between insitu conditions in
the structure and the conditions in tests used to determine
the property.
The idea behind partial safety factors is that they should reflect
the uncertainty of each type of action. The partial safety factor yQ is
therefore usually larger than yG, due to the fact that variable loads
(for example snow, wind and imposed load) display greater uncer­
tainty than permanent loads (for example self weight). In addition,
the values of partial safety factors should be chosen so that
the desired level of reliability is achieved. For this purpose, the partial
safety factors are calibrated and optimised by calculating the reliabil­
ity indices fl obtained when the partial safety factor is used in design
for a set of simple and representative design cases. The objective of
this calibration procedure is to determine the partial safety factors so
that different design situations lead to uniform reliability levels fl for
different design situations as close as possible to the target reliability
level Btarget..
An important basis for the partial safety factor method is the con­
cept of characteristic value, which should be based on a clear statisti­
cal definition. It is obvious that the value of a partial safety factor for
both loads and material parameters depends on the definition of
the characteristic value used. For time-variable actions (for example
snow and wind), a typical definition of characteristic value Qk is that
the probability of exceeding Qk should be 2 percent per year. This is
the same as stating that Qk should have a return period of 50 years,
meaning that the load level Qk is exceeded in average once during
a 50-year period.
Permanent actions are more or less constant during the whole ser­
vice life of the structure. In contrast, variable actions change in time
and their characteristic value corresponds to rare points in time
when extreme levels occur. When more than one variable action con­
tributes to a certain load effect E, combination values of design
actions must be determined to account for the reduced probability
that extreme values of all variable actions will occur at the same
time. To obtain a design combination value for each load situation,
each action should be considered in turn as a leading action (with its
full value) and combined with the other actions considered with their
combination values, which are described by a reduction of the char­
acteristic value Qk by factors y0, y1 and ^2. The following load values
are introduced in EN 1990:
• The combination value (^0Qk): used for the verification of ULS and
for the characteristic combinations of irreversible SLS (conse­
quence of actions exceeding the specified service requirements will
remain when actions are removed).
• The frequent value (y1 Qk): used for the verification of ULS related
to accidental actions and for the verification of reversible SLS.
The frequent value is exceeded approximately 1 percent of
the time.

18 Design of timber structures - Volume 1

 1.4 Eurokoder - generella antaganden,
granstillstand och detaljutformning

• The quasi-permanent value (y2Qk): used for the assessment of

long-term effects of SLS, such as deflections or cracks, and for
the representation of variable actions in accidental combinations
of ULS. Corresponds to the time average of the variable action Q.

The factor y2 can also be regarded as a factor that converts loads with
short-term duration to an equivalent permanent action in order to
calculate the long-term deflection effected by creep.
EN 1990 defines load combination rules for the different design sit­
uations, shown in table 1.2, page 16. These rules define how perma­
nent loads and variable loads shall be combined to determine
the load effect. The following general format is, for example, valid
for design in persistent or transient design situations in the ultimate
state:

Ed = X 7g,j ■ Gk,j + /Q4 ■ <2k4 + X /qa • ^04 • Sk4 1.6

J>1 i>l

where indices j and i denote the jth permanent load component and
the ith variable load component respectively, and:

Qk,1 is the characteristic value for the leading variable load.

YQ1 is the partial safety factor associated with Qk1.

To verify a structure, it is necessary to consider different design situa­

tions, see table 1.2, page 16. The selected design situation should be
sufficiently severe and represent all the conditions that can reasona­
bly be foreseen as occurring during the execution and use of
the structure. For verifications in the ultimate limit state (ULS) or
serviceability limit state (SLS), there are a set of prerequisites for
calculations.

ULS deals with:

• rupture
• collapse
• loss of equilibrium
• transformation into a mechanism
• failure caused by fatigue.

SLS deals with:

• deformation
• vibration
• cracks
• damage in a negative way affecting use.

The procedure of limit state design can be described as:

• define the relevant limit states for which the structural behaviour
is to be checked
• for each limit state, determine the appropriate actions and
load combinations to be considered
• using appropriate structural models for design, and taking account
of the inevitable variability of parameters such as material proper­ Glulam hall under construction. Permanent cross bracing
ties and geometrical data at a gable wall transfers horizontal loading from stiff
• verify that none of the relevant limit states is exceeded. roof sheet to the ground

Design of timber structures - Volume 1 19

1.5 Concepts used for the limit state design
of timber and wood-based products

In ULS, the following types of limit state should be considered when

relevant:
• EQU: relating to the static equilibrium of a structure or any part of
it which is regarded as a rigid body.
• STR: relating to the internal failure or excessive deformation of
a structure or its member.
• GEO: relating to the failure or excessive deformation of the soil or
rock (ground).
• FAT: relating to the fatigue failure of a structure or structural
members.

1.5 Concepts used for

the limit state design of timber
and wood-based products
As a building material, timber differs from steel, reinforced concrete
or other composites in a number of ways. Wood is a biological and
natural material with highly variable properties. Wood is ortho­
tropic, it has different properties in different directions. It is further­
more hygroscopic, which means that the moisture content in
the material constantly changes with the relative humidity of the sur­
rounding environment. When a designer chooses a certain grade of
concrete or steel, the manufacturers of these materials produce them
according to well-established procedures to obtain a specific grade
with more or less controlled variability. When it comes to timber,
however, the designer chooses a grade which has been verified by
some type of non-destructive strength grading of sawn timber, see sec­
tion 2.5, page 42. This means sorting existing material, which was
formed during a long period of time and often in very different con­
ditions, based on statistical relationships.
Sawn timber is produced from a very large variety of logs sawn
from many different trees. These trees come often from forests with
very different silvicultural conditions and a very long rotation
period.
These prerequisites of strength-graded timber and the variable
characteristics of wood-based products are taken into account when
designing timber structures.
Some of the characteristics which are specific when it comes to var­
ious wood-based product design are:
• the duration of load for various types of load
• the moisture content in timber and service classes
• partial factors for material properties and adjustments using vari­
ous modification factors
• the difference in material response when the loads are applied in
various directions in relation to the grain orientation of timber.

20 Design of timber structures - Volume 1

 1.5 Concepts used for the limit state design
of timber and wood-based products

Table 1.4 Load duration classes

Load duration classes Accumulated duration Examples of loading

Permanent (P) > 10 years Self weight

Long-term (L) 6 months - 10 years Storage

Medium-term (M) 1 week - 6 months Imposed floor load

Snow load

Short-term (S) < 1 week Wind load

Snow load

Instantaneous (I) Wind gusts

Accidental load

Source: Table according to EN 1995-1-1:2004, 2.3.1.2.

1.5.1 Load duration classes

Timber experiences a significant loss of strength over a period of
time, see section 2.4.3, page 38. In order to take account of the loss of
strength, load duration classes were established to facilitate
the design procedure. The duration classes cover a range of durations
which may occur in practice and associated actions, see table 1.4.
The influence of load duration on the strength capacity of timber is
taken into account by assigning a factor kmod as a function of service
class defined in section 1.5.2 and one of the five load-duration classes
in table 1.4. The load-duration factor kmod is a reduction factor for
the characteristic strength of timber and wood-based products, vary­
ing between 0,2 and 1,1. Only for instantaneous load and dry wood,
this factor is > 1,0 and results in an increase in the characteristic
strength.

1.5.2 Effect of moisture content

and service classes
Moisture content and variations in moisture content play a very sig­
nificant role for all properties of timber and wood-based products.
Moisture content affects both strength and stiffness. To incorporate
this effect in design, three service classes have been defined in
EN 1995. They are:
• Service class 1 — when the average moisture content in most soft­
woods does not exceed 12 %
• Service class 2 — when the average moisture content in most soft­
woods does not exceed 20 %
Service class 3 — when the average moisture content in most soft­
woods exceeds 20 %.

1.5.3 Partial factors for material

properties and adjustments using
various modification factors
In order to take account of uncertainty in the resistance model used
for design, together with the unfavourable effects of geometrical
deviations of materials, characteristic strength values are divided by
the partial factor yM in order to obtain the design strength of
the material, see equation 1.5, page 18. The partial factor yM varies
between 1,3 and 1,2 for most wood-based material in the ULS, but,

Design of timber structures - Volume 1 21

1.5 Concepts used for the limit state design
of timber and wood-based products

for accidental combinations and all combinations in the SLS, the par­
tial factor yM is equal to 1. There are many different adjustments of
strength using modification factors. Commonly used factors in
Eurocode 5 are: kh,, kc,90,, kcrit,, kdis,, km,, ksys,, kv,, kvol . They correspond to:

• size effect (kh)

• an increase in compressive strength perpendicular to the grain
(kc,90)

• a reduction in strength due to the effects of lateral and torsional

instability (kcrit)
• an adjustment in the design tensile stress/strength perpendicular
to the grain in the apex zone of a double-tapered, curved or
pitched cambered beam (kdis)
• stress redistribution for beams loaded in flexure about both major
axes (km)
• member strength properties when several members or components
are equally spaced and connected by a continuous load distribu­
tion system (ksys)
• a reduction in shear strength due to a notch (kv)
• an adjustment in the design tensile strength perpendicular to
the grain in the apex zone of a double-tapered, curved or pitched
cambered beam which exceeds a certain reference volume (kvol).

To describe the effect of creep instantaneous deformations are multi­

plied with a factor which describes the additional deformation devel­
oping over time (kdef).
Further background and details about these factors are given in
later chapters of this book.

1.5.4 Difference in material response when

the loads are applied in various directions in
relation to the grain orientation of timber
Both mechanical and physical properties differ significantly in differ­
ent directions when it comes to fibre orientation, that is the radial,
longitudinal and tangential direction of sawn timber result in very
different properties. In design, it is very important to recognise
whether timber is loaded parallel or perpendicular to the grain. In
the first case, timber has very good strength, but, in the latter case,
the strength is typically 1/ 10 of that in the longitudinal direction
and often close to zero, see section 2.4, page 33.

22 Design of timber structures - Volume 1

 2.1 Forestry and the production
of sawn timber

Structural properties of
sawn timber and engineered
wood products
Marie Johansson

2.1 Forestry and the production 2.1 Forestry and the production
of sawn timber 23
of sawn timber 2.1.1
2.1.2
Europe 23
Swedish forestry 24
2.1.3 Production of sawn timber 25
In 2009, 3,95 billion hectares of land were classified as forest in
the world. Of these forests 25 percent were located in Europe, 21 per­ 2.2 Structure of timber 26
2.2.1 Material structure of wood 27
cent in South America, 16 percent in Africa, 15 percent in Asia,
2.2.2 Natural characteristics of wood 29
18 percent in North and Central America and 5 percent in Oceania.
Just under half of the wood use in the world is used as industrial 2.3 Physical properties of wood 31
wood (pulp, paper and wood products); the other half is used for fuel, 2.3.1 Wood and moisture 31
the main part of this in developing countries. The countries that pro­ 2.3.2 Shrinkage and swelling 31
2.3.3 Distortion of timber 32
duce the largest amounts of sawn timber are USA, Canada, Russia,
2.3.4 Density 33
Germany and Sweden. Canada, Russia and Sweden are also large
exporters of sawn timber. 2.4 Mechanical properties of
wood and timber 33
2.1.1 Europe 2.4.1
2.4.2
Strength and stiffness of wood 33
Strength and stiffness of structural timber 37
The forest sector is an important economic sector for Europe. 2.4.3 Influence of moisture 38
Studying Europe as a whole (including all of Russia) the forest sector 2.4.4 Influence of time 38
2.4.5 Influence of temperature 39
accounts for about 1 percent of the GDP and employs almost four
2.4.6 Influence of size 39
million people. Europe has 25 percent of the world’s forest resources
2.4.7 Long-term deformations 40
but only 17 percent of the total land area. The largest part, 81 per­
cent, of these forest resources is within the Russian Federation. This 2.5 Strength grading 42
means that 44,3 percent of the European land area is covered with 2.5.1 Relationship between strength, stiffness
forests. In all countries in Europe the area and the amount of timber and other parameters 43
2.5.2 Machine strength grading principles 44
in the forest is increasing every year. The differences within Europe
is, however, very large. In most of the available statistics, Europe is 2.6 Modified wood 46
divided into three regions, Western Europe, Eastern Europe (incl. 2.6.1 Processes 46
the Baltic States) and CIS countries (former Soviet republics). 2.6.2 Properties of modified wood 46
The forests in Western Europe are mostly privately owned (over 2.7 Engineered wood products 47
70 percent) while in Eastern Europe the share of privately owned for­ 2.7.1 Engineered wood products
ests is increasing from having been mainly state owned before 1990. based on sawn timber boards 48
In the CIS countries the forest is to almost 100 percent publicly owned. 2.7.2 Engineered wood products
The fragmented owner structure in Western Europe has made it com­ based on veneers 50
plex to manage the forests and in many countries strong private for­ 2.7.3 Engineered wood products
based on strands, chips or fibres 51
est owner associations are formed. Forest management are in most
2.7.4 Built up structures - I-beams 52
European countries highly regulated by the government, meaning
strict regulations regarding harvesting and re-planting. The European 2.8 End-user requirement on sawn timber 53
forests (excl. Russia) are to a large extent a mix of natural and planted
forests. The forests in Russia are still to a large extent natural, espe­
cially east of the Ural. Another important aspect of forest is the social
dimension. More than 90 percent of the European forests are open
for public access and the forest area for recreation is increasing.
The countries in Eastern Europe and the CIS countries before 1990
produced about half the sawn wood in Europe. Political changes

Design of timber structures - Volume 1 23

2.1 Forestry and the production
of sawn timber

Figure 2.1 Example of two softwood species. a) Spruce (Picea abies), b) pine (Pinus Sylvestris).

caused a drastic decrease in production after that. Today the produc­

tion in these countries is again rising and with a large production of
Engineered Wood Products. The demand for bioenergy from the for­
est has increased in Europe during the last ten years and this demand
is expected to increase as more countries try to replace fossil fuel with
renewable energy. The demand of renewable energy has increased
the amount of material taken from the forests; such as material from
thinnings, residues from harvesting, branches and stumps.
The European forest can be classified in boreal forest (The Nordic
countries and Northern Russia) while the middle of Europe and
the southern parts of Russia can be classified as temperate broadleaf
forest. In the boreal forests softwood species; spruce, pine and larch
are dominating, see figure 2.1. The softwoods are dominating also in
the rest of Europe but there are also more hardwoods to be found, as
for example beech, chestnut and alder. In production of sawn timber,
especially for the construction industry, the softwood species are
more commonly used.

2.1.2 Swedish forestry

On the Swedish level the forest industry is a very important industry
sector and stands for 9 — 12 percent of the total turnover, added value
and employment of the Swedish industry. Slightly over 12 percent of
the total export from Sweden is based in the forest industry. The total
export value in 2014 was 124 billion Swedish crowns, of this the pulp
and paper industry is responsible for approximately 3 / 4 while sawn
timber and engineered wood products is responsible for 1/ 4.

24 Design of timber structures - Volume 1

 2.1 Forestry and the production
of sawn timber

The Swedish forest area is in total 27,5 million hectares; of this

78 percent is productive forest while the remaining part is areas set
aside for conservation and unproductive areas. The Swedish forest is Other Packaging
to 50 percent privately owned while private and public companies 25 % 19 %
own 25 percent and 14 percent respectively. The growth in the Swedish
forests has been larger than the felling since 1950. The most common
Strips/mouldings 1 % Impregnated
species in the Swedish forests are spruce (40 percent) and pine 16 %
(39 percent). The remaining forest is birch (12 percent) and other Windows/doors 3 %

deciduous species. The birch is mostly used in the pulp and paper Glued laminated
products Buildings
industry. 4 % cFurniture
new
Spruce and pine are used both in sawmills and in the pulp and 12 %
6%
paper industry. In 2013, 76 million m3 sub (solid volume under bark) Flooring
logs was brought to the Swedish industry, of this approximately 6% Renovation and
enlargement
50 percent was sent to the sawmilling industry and the other 50 per­ 8%
cent to the pulp and paper industry. The sawmilling industry yielded
about 70 percent of this into sawn timber and engineered wood prod­ Figure 2.2 Estimated use of sawn timber in Sweden,
ucts while the remaining 30 percent was wood chips sent to the pulp Dackling (2002)

and paper industry. The total volume of sawn timber from the saw­
milling industry in 2013 was 17 million m3 of which 12 million m3
was exported. The largest export market for Swedish sawn timber in
2014 was Great Britain (2,7 million m3) with Egypt, Norway, Germany
and Denmark as next largest export markets (Skogsindustrierna 2014).
The use of Swedish sawn timber can be seen in figure 2.2, which
shows that a large part isused for packaging. Of the remaining sawn
material 55 percent is used directly in the construction industry.

2.1.3 Production of sawn timber

The following subchapter describes the production of sawn timber in
a Swedish (Nordic) context and is concentrated on the production of
structural timber from spruce (Picea abies). Forests in Sweden are dom­
inated by managed (planted) forests, where the rotation time is
approximately 80 years. Most forests are today planted and managed
through several thinnings during the growth. At the age of approxi­
mately 80 years the trees are harvested. After harvesting the logs are
taken to a sawmill or a pulp industry, and the branches are cut into
wood chips and taken away to be used as fuel wood. The ground is
then replanted with new spruce or pine plants.
The normal way of harvesting trees in Sweden is to make clear-cuts
in the forest of an even-aged stand using a harvester, see figure 2.3.

Figure 2.3 Harvester and logger in action in the forest

Design of timber structures - Volume 1 25

2.2 Structure of timber

The harvester has capability to measure the diameter of a log and

cross-cut the logs to lengths optimised for the market demand.
The logs are cut to pulp logs (smaller diameter) or saw logs. The logs
are then transported by a logger to the nearest forest motor road for
storing before transport by lorry to a sawmill or pulp industry. For
the sawmilling industry the maximum log length in Sweden is nor­
mally 5,5 meters with a diameter less than 40 cm.
At the sawmill the logs are normally graded by Biometria into dif­
ferent quality classes to decide the price of the logs for payment to
the forest owner. The logs are normally also graded into different
length and diameter classes. The logs are often debarked and run
through a metal detector before being sent into the saw line. In
the beginning of the saw line the logs are scanned to register at least
the outer shape of the logs and in more advanced scanners the inter­
nal properties are also recorded using x-ray. The result from the scan­
ner is used to optimize the sawing pattern. The logs are sawn to
boards of different sizes and sorted into batches of different dimen­
sions (and quality classes). The next step in the process is to stack
the boards before kiln drying. The kiln drying is made to reduce
the moisture content in the boards to levels between 12 % and 18 %
depending on the end use of the material. There are in principle two
kinds of kiln driers that can be used, a chamber kiln or a continuous
kiln. In the continuous kiln the stacked boards are brought into one
end of the kiln and slowly (5 — 7 days) transported through the kiln.
In the chamber kiln the stacked boards are placed within the cham­
ber of the kiln and the doors closed. The relative humidity and
the temperature in the kiln are possible to control and adapt so that
the drying process is performed to minimize the energy needed and
to optimize the quality of the sawn timber.
After kiln-drying the boards are un-stacked and run through
an adjustment mill where the boards are visually scanned and sorted
into different visual grades and unwanted parts of the boards are
sawn off. The boards are then often run through a planer to get their
Spruce forest final size. After planing, boards to be used as structural timber are
strength graded into different strength classes. (More about strength
classes and grading can be found in section 2.5). The sawn timber is
now ready to be packed, wrapped and shipped to its final destination.

2.2 Structure of timber

The wood material is built up by trees and the structure of the mate­
rial is optimised for the circumstances the tree is exposed to. The tree
stem is grown round to optimise the tree against wind loads, it grows
branches to carry up leaves (or needles) to take part in photosynthe­
sis, it stores chemicals in the centre of the tree to use as weapons in
case the tree is exposed to a threat and it forms reaction wood in
a leaning tree to optimize its internal stresses. The tree is also built
with internal pre-stressing (growth stresses) to minimize the risk of
crushing the wood at high compressive stresses. These are some
examples of how trees are optimized for their living. When man har­
vest the tree not all of these optimizations are beneficial for the use
of the sawn wood. To be able to use wood in the best possible way it
is necessary to know how wood is built up and the influence this
structure have on the behaviour of the material.

26 Design of timber structures - Volume 1

 2.2 Structure of timber

Figure 2.4 Basic structure of a cellulose molecule

2.2.1 Material structure of wood

Wood is a natural composite built up of mainly three elements;
50 percent carbon, 6 percent hydrogen and 44 percent oxygen in
the form of cellulose, hemicellulose and lignin. Nordic softwood spe­
cies have approximately the proportions 40 — 45 percent cellulose,
25 — 30 percent hemicellulose and 25 — 35 percent lignin. Cellulose
is a long organic molecule chain built up by glucose units, see fig­
ure 2.4. The number of glucose units can vary from a few up to
15 000 depending on the location of the cellulose molecule within
the cell wall. The cellulose molecules have a hydroxyl group ( — OH)
on the side which can form hydrogen bonds to the next cellulose
chain. Hemicellulose is the name for a group of shorter branched
molecule chains and lignin is the generic term for a group of three
dimensional polymers, Bodig and Jayne (1982).
The cellulose in wood are collected in crystalline strands enclosed
by hemicellulose and lignin. The strands are most often called
microfibrils. These strands make up the cell wall in the living tube­
shaped wood cell. The most common cell type in softwood is the trac­
heid which is tube-shaped with a size of approximately 2 — 4 mm in
length and 0,1 mm in width (diameter). The cell wall in wood con­
sists of four layers. On the outside there is a thin layer usually called
the primary wall where the strands are randomly oriented. Inside of
this is the proper cell wall consisting of three parts. The inner and
outer of these three layers (S1 and S3) have the strands oriented
around the tube shaped cell. The main purpose of these layers is to Spruce shoots
maintain the shape of the cell. Between these layers there is a thicker
layer (S2) with most of the strands oriented more or less parallel with
the length direction of the cell, see figure 2.5, page 28. This middle cell
wall layer (S2) represents about 85 % of the cell wall thickness and is
consequently the layer with the greatest influence on the properties
of the cell. The angle between the axial direction of the cell and
the micro fibrils in the S2-layer are often called the microfibril angle
(MFA). The cells are bound together by a layer called the middle
lamella consisting mainly of lignin; this acts as an adhesive layer
between the cells.
The wood tissue in softwood is made up from a large number of
these tracheids (90 % of the volume), often called fibre or grain.

Design of timber structures - Volume 1 27

2.2 Structure of timber

S 3
60 90 °
S 2
5 15 °
S 50 70 °
P Random

21

Figure 2.5 The structure of a wood cell. The figure shows the cell wall Figure 2.6 Principle of the cell wall growth during one
constructed of 4 layers around the central cell cavity. Typical microfibril angle in year, note the thin-walled earlywood cells and the thick walled
these layers is given in the figure, Hoffmeyer (1995). latewood cells. The number of cells in a real tree is higher.

The structure of hardwoods is often more complicated and also

includes other types of cells. The tracheids are oriented mainly in
the longitudinal direction of the tree. A growing tree forms new cells
through cell division in a layer between the wood and the bark;
the cambium. During the spring period the tree has a need of cells
that can transport water and nutrition up and down the tree.
The cells that are formed during this time is called earlywood cells
and have thin cell walls (2 pm) and large cell cavities to facilitate
transport of liquids in the tree stem. During the summer the need for
transport of water and nutrition is less and the tree also needs
greater strength to withstand external loads such as wind and snow.
During the summer and autumn the tree therefore forms cells with
thick cell walls (10 pm) and smaller cell cavity. These cells are nor­
mally called latewood, see figure 2.6. The proportion of latewood cells
has significant influence on the mechanical properties of the wood
material. The different cell types are often possible to distinguish by
their colour; the early wood is seen as a thick lighter coloured band
while the late wood is seen as a thinner band of darker wood.
The wood cells have small openings (pits) on the sides that connect
them to neighbouring cells. These openings allow for mineral solu­
Sapwood Heartwood
(light colour) (darker colour) tions and nutrition to be transported within the tree. There are vari­
ous types of pits occurring in different tree species. The type of pit is
the main factor governing moisture permeability in different species.
The only wood cells that are truly alive, that is have the capacity
for cell division, in a tree are the outermost cells close to the bark
(cambium). The cells deeper into the tree have the possibility to
transport liquids (sapwood). In the cells inside the sapwood zone,
the tree starts to store substances that protect the tree against biologi­
cal attacks that is extractive substances. The wood in this area of
the tree is called heartwood. In pine the heartwood can be distin­
guished by its darker colour, see figure 2.7. In spruce it is not possible
to see the difference between sapwood and heartwood based on
the colour. The centre of the tree is often referred to as the pith.

28 Design of timber structures - Volume 1

 2.2 Structure of timber

2.2.2 Natural characteristics of wood

There are a number of characteristics of wood that for engineering
use can be seen as defects, a few of them are described below.

2.2.2.1 Knots
All trees need leaves (hardwoods) or needles (softwoods) for transpira­
tion, respiration and the photosynthesis process. The area needed for
this process is so large that it is necessary for the tree to spread
the leaves and needles out onto branches. Each year a tree grows in
length at the top but also forms new branches at the sides. These
branches are each year gradually embedded in the stem wood
through the addition of more material on the outside of the trunk
and the branches; see figure 2.8. The material in the branch is
the same as in the main stem and the tissue systems of the two struc­
tures are interconnected. A longitudinal section through a knot
shows that the fibre orientation around the knot is distorted and that
the fibres pass in a sweep around the knot. The presence of knots has
a negative influence on most mechanical properties of sawn timber.
The fibres around the knot are no longer continuous when the wood
is sawn and thus affecting the strength, at least in the main fibre
direction. The size, shape, number and location of knots influence
the strength of the sawn timber. For structural use, sawn timber with
smaller and fewer knots are often graded into higher classes.

2.2.2.2 Spiral grain angle

The fibres of wood tend to grow in a spiral around the trunk of
the tree. The most accepted explanation, from the perspective of
the tree, is that it creates more flexibility in the stem against wind
loads. Spiral grain angle is defined as the angle between the fibre
direction and the direction of the pith. In the case of Norway spruce,
the spiral grain angle is zero close to the pith. A left-handed spiral
grain then develops and continues to increase during subsequent
growth, often reaching a maximum value of 3 — 4° within the first
ten annual growth rings. The grain angle then decreases towards
zero. Later in the life of the tree, the grain angle may change direc­
tion to become right-handed. The grain angle in this direction
increases slowly as long as the stem growth continues.

Figure 2.8 A knot seen from all four sides of a wood specimen 45 x 45 mm in square section

Design of timber structures - Volume 1 29

2.2 Structure of timber

Juvenile Mature 2.2.2.3 Juvenile wood

wood wood
Specific gravity Juvenile wood is formed during the first 5 — 20 years of the growth of
Cell length a tree. Juvenile wood is formed near the pith. It is characterised by
Strength a progressive increase in cell dimensions and changes in cell charac­
Cell wall thickness
teristics and the pattern of cell arrangement. According to most
Transverse shrinkage
Percent latewood assessments, juvenile wood have less desirable properties than
mature wood. The cells in the mature wood may be three to four
times longer than those found in juvenile wood. In addition to cell
Pith 5 - 20 rings Bark
length, there is a difference in cell structure, see figure 2.9. There are
fewer latewood cells in juvenile wood. In softwoods, the density is
typically 10 — 15 percent lower in the juvenile core and the strength
is normally 15 — 30 percent lower than that of mature wood. Within
the cell, the microfibril angle in the S2 part of the secondary wall is
characteristically larger in juvenile wood. Juvenile wood is reported
to shrink along the grain up to nine or ten times as much as mature
wood.

2.2.2.4 Reaction wood

Reaction wood is formed because that each part of a tree strives to
remain in its genetically predetermined, static equilibrium position.
Any displacement results in the formation of reaction wood so
arranged that the displaced organ is brought back to its original posi­
Figure 2.9 Juvenile to mature wood
transition of properties tion by the shortest possible route. In this theory, gravity has its place
since the displacement must be made in relation to the gravitational
field.
The reaction wood present in gymnosperms (softwoods) and angio­
sperms (hardwoods) is best referred to as compression wood and ten­
sion wood, respectively. Compression wood and tension wood are
names that have been used for a long time and they also indicate
the functions of these tissues in the living tree. Compression wood is
formed on the compressed side of a leaning softwood and tension
wood on the side subjected to tension. Tension wood is correspond­
ingly formed on the tensioned side of a leaning hardwood. As soft­
woods are commonly utilised in Sweden for structural purposes
the following text is focusing on compression wood. Compression
wood fibres are shorter and rounder than for normal wood. They
have distorted tips and thick cell walls. Intercellular spaces often
form between the rounded cells, which results in less connection
between the fibres. The S2 layer is much thicker in compression wood
cells than in normal cells but with a microfibril angle in the S2 layer
in the range of 30 — 50°. In severe compression wood, no S3 layer is
produced, but it is sometimes replaced by a warty layer. Needless to
say, every form, between severe compression wood and normal wood,
can also be present in wood.
In chemical terms, the same substances are present in both com­
pression wood and normal wood. The differences relate to
the amount of the substances. Compression wood contains
Reaction wood 30 — 40 percent more lignin and 20 — 25 percent less cellulose than
normal wood. The amount of hemicellulose is in general not
changed. Compression wood has higher specific gravity, a lower fibre
saturation point, reduced permeability, lower radial and tangential
shrinkage but substantially larger longitudinal shrinkage, higher
compressive strength but lower tensile strength, a lower modulus of
elasticity and it is more brittle in terms of fracture than normal
wood, Timell (1986).

30 Design of timber structures - Volume 1

 2.3 Physical properties of wood

2.3 Physical properties of wood

2.3.1 Wood and moisture
One of the external factors that have a large influence on the proper­
ties of wood and timber is water. The moisture content in wood is
always depending on the relative humidity in the surrounding air.
When water enters into a dry piece of timber the molecules are
firstly bound to the surface of the cellulose strands in the cell walls
(to the hydroxyl groups). When all the places in the cell walls are
filled the water molecules starts to fill up the cell cavity. The amount
of water in the wood is usually defined as moisture content (moisture
ratio) u, often expressed in % by weight, see equation 2.1:

m... - mArv
u = —----- *y ,100 2.1
Wdry

Here u is the moisture content (ratio) in %, mu is the mass of the wood

at moisture content u and mdry is the mass of the wood after drying in
103 °C for 24 hours. The moisture content level where the adsorption
and desorption of water is equal is called the equilibrium moisture
content (EMC). The relationship between relative humidity and
the equilibrium moisture content can be seen in figure 2.10.
One concept often used is fibre saturation point (FSP) which is
defined as the moisture content when the cell wall is filled with water
molecules but the cell cavities are still empty. For softwoods grown in
Europe the FSP is often between 27 % and 33 %. At higher moisture
content levels the cell cavities are starting to fill up with water.
The fibre saturation point is a cut-off point for many properties of
the wood material. For example swelling of the wood material occurs
when water molecules are bound in the cell wall, water in the cell
cavity do not influence the outer dimensions of a piece of wood.

2.3.2 Shrinkage and swelling

In the cell wall, the water is bonded to the surface of the micro
fibrils. When the moisture content is reduced, the microfibrils come
closer to each other, which make the wood shrink. Since most of
Equilibrium moisture content (%)

Figure 2.10 Relationship between equilibrium moisture content and the relative humidity

Design of timber structures - Volume 1 31

2.3 Physical properties of wood

Figure 2.11 Diagrammatic sketch of micro fibril arrangement and shrinkage in normal wood and in compression
or juvenile wood. Note the difference in micro fibril angles.
a) Shrinkage at normal micro fibril angle.
b) Shrinkage at large micro fibril angle (compression wood or juvenile wood).

the microfibrils are in the S2 layer, the micro fibril angle of this layer
will govern much of the behaviour during shrinkage and swelling.
Since the microfibrils are normally inclined at a small angle to
the longitudinal axis, almost all the shrinkage occurs in the trans­
verse directions. Juvenile wood and compression wood have microfi­
bril angles that are much larger than those of normal wood, which
results in far larger longitudinal movements, see figure 2.11. Typical
values for shrinkage strain in the different directions are for a change
in moisture content of 1 % is:
• Longitudinally 0,0001
• Radially 0,0015
• Tangentially 0,0030.

There is a factor of approximately 2 in difference between the shrink­

age in radial and tangential direction, this can be explained by
the difference in wood structure in these two directions. The shrink­
age in the longitudinal direction is very small but for large lengths
this can still be necessary to take into account. The presence of com­
pression wood or juvenile wood can have a large influence on, espe­
cially, the longitudinal shrinkage which can be a hundred times
larger in compression wood than in normal wood.

Twist 2.3.3 Distortion of timber

The varying shrinkage within a piece of wood can cause geometrical
changes of the wood cross-section that can make them difficult to use
(distortion), see figure 2.12. Distortion can be divided into four differ­
ent forms: twist, spring, cup and bow. Spring is the edgewise devia­
tion of a piece of timber from a straight line from end to end, while
bow is the flat-wise deviation. Twist is defined as lengthwise spiral
distortion. Cup is the deviation flat-wise from a straight line across
the width of the board. For studs used in the building industry, twist
is the distortion mode that causes the largest problems. Spring also
causes problems, while moderate bow can be accepted. Cup is not
normally a problem for structural timber, but it causes severe mate­
rial losses during planing.
Twist can be explained by the combined effect of the annual ring
orientation, spiral grain angle and shrinkage. The fact that
the shrinkage perpendicular to the actual fibre direction is large,
Bow combined with a spiral grain angle of some degrees as well as
the curvature of the annual ring causes each annual ring to twist and
Figure 2.12 Definition of distortion/warp modes consequently the whole board. The cup deformation is mainly caused

32 Design of timber structures - Volume 1

 2.4 Mechanical properties
of wood and timber

by the difference in shrinkage between the radial and tangential

direction in combination with the circular shape of the annual rings.
The two bending modes bow and spring are caused by both varia­
tions in longitudinal shrinkage within a single board and the release
of growth stresses in the tree/log when it is sawn.
The size of distortion of sawn timber can be influenced by different
sawing patterns and different drying regimes. A good sawing pattern,
for example sawing boards that do not include the pith, can reduce
the twist. The kiln-drying regime is also important, different drying
schedules, air flow and weights on top of the sawn timber can help to
minimize the distortion. It can also be beneficial to measure the spi­
ral grain angle and treat boards with large spiral grain angle differ­
ently, with for example heavier loads in the kiln.

2.3.4 Density
Density is a very important physical property of wood; it is correlated
to almost all mechanical properties. Density (p) is defined as:
m
P=V 2.2

where m is the mass (kg) and V the volume (m3). The density is mois­
ture dependent as both the mass and the volume is dependent on
the moisture content. Therefore the density for wood has to be
defined also in terms of moisture content. The most commonly used
definition in timber engineering is the density p12, that is based on
the mass and volume at 12 % moisture content. This is the moisture
content used in all standard tests for wood strength.
The density of the cell wall is about 1 500 kg/m3. This means that
density is a measure of the amount of pores in the wood material.
The normal density (p12) for softwood grown in the Nordic countries
varies between 300 and 600 kg/m3.

2.4 Mechanical properties

of wood and timber
It is usual to divide wood into small clear specimens (wood) and large
specimens that include natural characteristics (timber) such as; knots,
spiral grain angle, juvenile wood and reaction wood. Clear wood
specimens consist of only straight wood fibres without anomalies and
hence all properties are only dependent on the properties of
the wood fibre. For timber the effects of the natural characteristics
will be large and to a great extent decide the properties and behav­
iour of the specimen.

2.4.1 Strength and stiffness of wood

The structure of wood, tube shaped cell in a lignin matrix, results in
different properties in different directions, that is the material is ani­
sotropic. With a structure of tubes in a matrix it is easy to under­
stand that the compression strength is higher parallel to the fibres
than perpendicular to the fibres. Wood thus has different properties
in different directions and it is important to keep track of the loading
direction. For a complete picture of the properties of wood it is neces­
sary to define all the stresses in figure 2.13, page 34.

Design of timber structures - Volume 1 33

2.4 Mechanical properties
of wood and timber

Figure 2.13 shows the definition of the different stress in wood. For
timber the difference between R and T direction is often disregarded
and the directions are named g0 or g// and g90 or g± for the parallel and
perpendicular to the fibre direction respectively. To describe the “real”
behaviour of wood within the elastic range 12 constants are neces­
sary; for example the modulus of elasticity EL, ER, ET, the shear mod­
ules GLR, GLT, GRT and the Poisson ratios vLR, vRL, vLT, vTL, vRT, and vTR.
Normally, the Poisson ratios are assumed to be pair wise equal, which
make it possible to eliminate three. By disregarding the difference in
L - longitudinally tangential and radial direction, the number of variables can be
T - tangentially
reduced to six, often denoted E//, E±, G//, G±, v// and v where // denotes
R - radially
parallel to the fibre direction and ± perpendicular to the fibre direc­
tion. The modulus of elasticity in wood is often shortened to MOE.
Figure 2.13 Definition of normal- (a) and shear- ( t)
stresses in different directions in wood 2.4.1.1 Tension parallel to the fibre direction
Loading small clear specimens (that is without knots) it is possible to
investigate the effect of loading the wood material in different direc­
tions. A pure tension test in the fibre direction show that the stress­
strain relationship is almost linear up to failure, see figure 2.14. On
the material level two failure modes are possible (or a combination of
them); the middle lamella breaks and the fibres are being pulled out
of their matrix material or the fibre can break. The strength of wood
in tension parallel to the fibre direction (ft) is very high; the failure
stress is often of the magnitude of 100 MPa. The failure is, however,
often very brittle.

2.4.1.2 Tension perpendicular to the fibre direction

If the wood is loaded in tension perpendicular to the grain the forces
to pull apart the fibres or break the fibres are much lower. The stiff­
ness is also lower in this direction. Figure 2.15 shows that the strength
of wood loaded perpendicular (ft,90) to the fibre direction is very low,
Figure 2.14 a) Stress-strain relationship for clear wood
loaded in tension parallel to the fibre direction. Typical failure in most cases it is not possible to assume a higher strength than
in pure tension parallel to the fibre direction for b) thick-walled 0,5 MPa. It is also possible to see that the stiffness is much lower in
late wood and c) thin walled early wood. this direction.

2.4.1.3 Compression parallel to the fibre direction

In compression, the wood material will behave differently. In com­
pression parallel to the fibre direction the main stress will be parallel
with the axial direction of the fibres (tubes). Tubes loaded axially are
very stable and can withstand a high load level. When the load is too

zx o

a)

Figure 2.15 a) Stress-strain relationship for clear wood loaded in tension

perpendicular to the fibre direction. b) Typical failure in pure tension perpendicular to
the fibre direction.

34 Design of timber structures - Volume 1

 2.4 Mechanical properties
of wood and timber

Figure 2.16 a) Stress-strain relationship for clear wood

loaded in compression parallel to the fibre direction.
b) Typical failure in pure compression parallel to the fibre direction.

high some fibres will start to buckle and be driven into the other
fibres. When the buckling behaviour in the wood starts the possibil­
ity to take higher load will diminish and the behaviour can be classi­
fied as plasticising. Figure 2.16 shows the stress-strain relationship for
clear wood loaded parallel to the fibre direction. The compression
strength for wood in pure compression parallel to the grain (fc) is
around 80 MPa. Typical values for sc are 0,8 — 1,2 % with an ultimate
strain level su ~ 3sc.

2.4.1.4 Compression perpendicular to the fibre direction

In compression perpendicular to the fibre direction the tube shaped
wood cells will be crushed. Crushing a tube shaped structure from
the side requires low force and consequently both the stiffness and
the strength for this form of loading are low. It is however not possi­
ble to define a true failure in compression perpendicular to the grain.
When all the cells are crushed the stress level can increase again.
Figure 2.17 shows the stress-strain relationship for wood loaded per­
pendicular to the grain. The maximum stress level in compression
loading perpendicular to the grain (f) is 3 — 5 MPa. The failure in
compression perpendicular to the grain is then defined as the stress
level that gives 10 percent remaining deformations.
There is a difference in the amount of deformation for the same
load level depending on the orientation of the annual ring. The mod­
ulus of elasticity perpendicular to the grain is higher in the thick­
walled latewood than in the thin-walled earlywood. This is, however,
not taken into account when designing timber structures.

Figure 2.17 a) Stress-strain relationship for clear wood

loaded in compression perpendicular to the fibre direction.
b) Typical failure in pure compression perpendicular to the fibre direction.

Design of timber structures - Volume 1 35

2.4 Mechanical properties
of wood and timber

Figure 2.18 Definition of the angle a between the stress (force) and the fibre
direction

2.4.1.5 Loading in an angle to the grain direction

If the material is loaded in an angle between perpendicular to grain
and parallel to grain the strength is somewhere between the extreme
cases showed above. One equation for calculating the compression
strength depending on the angle (a), see figure 2.18, between the load
direction and the fibre direction was proposed by Hankinson (see for
example Dinwoodie 2000), see equation 2.3:

______ fo ' f90

r . 2_____ <• 2 2.2
/osin a + jgocos a

This equation has showed to work well also in tension. In graphic

form, see figure 2.19, it is possible to see that already a small deviation
from the parallel direction gives a considerable reduction in strength.

2.4.1.6 Shear strength

The shear strength of timber is highest in planes parallel to the fibre
direction, the strength varying between 5 and 8 MPa. The shear
strength perpendicular to the fibre direction tRT (rolling shear) is often
between 3 and 4 MPa (that is half the strength).
In timbers structures shear in the tRL and tTL direction are the most
common. The shear strength in the tTL direction is lower since
the failure line is going through the weaker early wood layer com­
pared with the tRL direction where the failure area has to go through
both the early and late wood. For the structural engineer it is how­
ever not possible to distinguish between these two and the lower of
the two values are used in the codes. For the rolling shear failure
the tube shaped fibres are rolling against each other instead of cut­
ting the cells in the longitudinal direction. Rolling shear is not so
common in normal square beams, but must be considered in for
Figure 2.19 Relationship between the failure strength
example glulaminated I-beams. The shear strength as well as the ten­
and the angle (a) between the fibre direction and force
direction according to Hankinson, Dinwoodie (2000) sion perpendicular to grain strength is highly influenced by drying
cracks in the material.

TTL

Rolling shear Shear longitudinal-tangential

Figure 2.20 Shear in different directions, Carling et al. (1992)

36 Design of timber structures - Volume 1

 2.4 Mechanical properties
of wood and timber

Figure 2.21 Longitudinal surface strains (sy) for maximum load 20 kN

2.4.2 Strength and stiffness

of structural timber
Larger pieces of wood (sawn timber) do not only contain straight
wood fibres but also anomalies such as knots, compression wood, spi­
ral grain and so on. It is therefore not possible to predict the mechan­
ical behaviour of sawn timber directly from the properties of clear
wood. For example the fibre deviations around a knot can give large
local stresses due to a loading in an angle to the grain.
The effect of a knot on the strain (and stress) field can be seen in
figure 2.21. This specimen, cross-section 45 x 45 mm including a knot,
see figure 2.8, page 29, was loaded in tension up to a level of 20 kN
(average stress 10,0 MPa). The strains were recorded with the meas­
urement system ARAMISTM manufactured by the company GOM dur­
ing loading. The system is based on evaluation of a random pattern
applied to the surface that deforms along with the material. Two CCD
cameras are placed at different angles in front of the specimen which
enables stereoscopic pictures to be taken of the specimen. On the basis
of the movement in the random pattern it is possible to evaluate 3D
displacement, rotations and strains.
The resulting strain shows that the strain distribution around a knot
is far from even. Areas far from the knot, with straighter grain, exhibit
less strain than areas close to the knot.
Strength values used for structural timber are therefore deter­
mined by testing with a standardised methodology. This means that
the strength values more reflect a strength value on the element level
than an actual strength value on the material level. The strength
value for bending is according to the European standard EN 384 and
EN 408 determined from a bending test performed with four-point
loading. The beam shall have a length of 19 times the beam height (H),
and be loaded with a clear span of 18H with a distance between
the inner loading points of 6H. The values for stiffness and strength
determined from a loading test to failure are defined using the elastic
section modulus. The determined strength values have to be recalcu­
lated to a height of 150 mm, which is the standard height for sawn
timber. This manner of testing results in strength values including
the effect of defects in the material (as well as possibly a small effect
of instability). The method of utilizing the elastic section modulus
when determining the strength also means that the strength values
can only be used in design models that are based on elastic theory.

Design of timber structures - Volume 1 37

2.4 Mechanical properties
of wood and timber

The strength values in tension and compression are determined in

the same manner with testing of a nominal dimension and determin­
ing a strength value based on this dimension.
The method of determining strength and stiffness values through
testing results in different strength values for different loading
modes such as bending, tension and compression.

2.4.3 Influence of moisture

Table 2.2 Approximate change in mechanical properties The mechanical properties of wood are affected by the moisture con­
of clear wood for a 1 % change in moisture content, tent. The lower the moisture content the higher the strength and
Hoffmeyer (2003)
stiffness. Often the strength and stiffness of timber are said to
Property Change (%) increase linearly with decreasing moisture content below the fibre
Compression strength 5 saturation point. Above the fibre saturation point there are no great
parallel to the fibre direction changes in strength and stiffness with increasing moisture content.
Compression strength 5 Tests have shown that the influence of moisture content is differ­
perpendicular to the fibre direction ent in different loading directions. For clear wood the change in
Bending strength 4 properties for 1 % change in moisture content can be seen in
parallel to the fibre direction table 2.2. This relationship is valid for moisture contents varying from
Tension strength 2,5
8 % to 20 %. Tests have shown that the influence of moisture content
parallel to the fibre direction on full size timber is not as pronounced as on small clear wood speci­
Tension strength 2
mens, especially in some loading directions. The tension strength is
perpendicular to the fibre direction almost independent of the moisture content for full size timber
Shear strength 3
while the compression strength is very much affected by moisture
parallel to the fibre direction content.
Modulus of elasticity
The influence of moisture content is normally taken into account
1,5
parallel to the fibre direction in design codes by reducing the strength values for timber used in
environments where high moisture content can occur. In Eurocode 5
all structural elements are assigned into different service classes
depending on the highest moisture content the element can be sub­
jected to during its lifetime, cf. section 1.5.2.

2.4.4 Influence of time

Another factor with a great influence on the mechanical properties
of wood and timber is time. Tests have shown that the bending
strength decreases with an increased loading time, that is duration of
load (DOL). This relationship seems to be valid in all loading modes
but is especially important in bending. The original curve for
strength versus time was established at the Forest Products Lab in
Madison, USA in the 1950’s based on small clear specimens and is
known as the Madison curve. The equation for the Madison curve is
nearly proportional to the logarithm of the time over which the load
is applied, see equation 2.4:

/ = 90,4-6,3 logf 2.4

where f is the stress level (percent) and t is the effective duration of

the load in hours. The curve shows that the strength after one year
has decreased to approximately 60 percent of the original short term
strength, see figure 2.22, page 39. The Madison curve shows a critical
stress level (50 percent) under which there is no duration of load
effects.
Further studies have shown that for full size timber this curve may
be different. Some research has indicated that the duration of load
effect is less for low-strength timber than for high-strength timber,
and that the effect is smaller than for clear wood, Hoffmeyer (2003).
Other studies indicate that the DOL-effect for structural timber is of

38 Design of timber structures - Volume 1

 2.4 Mechanical properties
of wood and timber

Loading time [h]

Figure 2.22 The relationship between relative bending strength and loading time
The strength at 5 minutes has been set to 100 percent.

the same order of magnitude as for clear wood. There is also an effect
of moisture content and temperature on the DOL curve. Higher mois­
ture content and varying moisture gives larger DOL-effect, the same
has been found for increasing temperatures.
To incorporate the effect of DOL in the design all loads are nor­
mally classified into classes depending on their duration. Depending
on the load duration class of the specified load, the strength is
reduced. The reduction is larger for long term loads than for short­
term loads. In Eurocode 5 the effect of duration of load and moisture
content is combined into one modification factor kmod, cf. section 1.5.1.

2.4.5 Influence of temperature

Both the strength and stiffness of wood is decreasing with increasing
temperature. The effect is, however, relatively small for normal tem­
peratures (-30 °C to +90 °C). At temperatures above 95 °C (or 65 °C for
long-term loads) there is a thermal degradation of the wood material.
At higher temperatures the cellulose chains are shortened and
the structure of the hemicellulose is changing. The effect of tempera­
ture is normally not taken into account in design codes.

2.4.6 Influence of size

Strength testing of timber has shown that there is a considerable
influence of the volume of the tested specimens. Larger specimens
break at a lower average tension stress level than smaller specimens.
This phenomenon is normally explained by the weakest link theory.
This theory says that “a chain subjected to tension is never stronger
than its weakest link”. For wood it is shown that the probability of
a large weakness occurring in the most loaded section is larger for
a large specimen than for a small specimen. Wood is a brittle mate­
rial and the Weibull theory has been used to explain the behaviour.
The Weibull theory assumes that the material is brittle and that
the defects are of random size and randomly distributed within
the specimen. If V1 and V2 are the volumes of two wood specimens
respectively and f1 and f2 are their respective strengths, equation 2.5
can be used to describe the relationship between these parameters.

Design of timber structures - Volume 1 39

2.4 Mechanical properties
of wood and timber

The factor k is the shape parameter of the Weibull distribution:

</lJ V2, 2.5

The volume is here the product of the width, depth and length, that
is b, h, and l of the specimens why the equation may be rewritten as
equation 2.6 with exponents associated with each dimension:

(4WkT
* ‘.iAr‘ .kt*' 2.6

In practice the width of a timber beam has a relatively small variation

and the effect of the width is generally ignored. The strength values
for bending strength as well as for tension strength are tested in
a standard configuration with a certain ratio between depth and
length. This makes it possible to reduce equation 2.6 to only the depth
but with a combined factor for the shape function taking both depth
and length into account.
The size (height) effect is often taken into account in design codes
by having different strength values for different sizes of the loaded
element. Eurocode 5 includes a modification factor kh that increases
the bending and tension strength for small dimensions, for sawn tim­
ber for a height less than 150 mm in bending and a width less than
150 mm in tension.
The volume effect is also very large in tension perpendicular to
the grain. Eurocode 5 therefore includes a modification factor kvol in
areas with high tension stresses perpendicular to the grain as for
example in curved elements or in the apex zone of double tapered
beams, see Chapter 3.

2.4.7 Long-term deformations

The loading time is affecting not only the strength but also the defor­
mation of wood. A piece of timber subjected to a constant load will
show increasing deformation with time, this effect is called creep, see
figure 2.23. When the load is removed most of the deformation will
be recovered, although there is almost always a small permanent
deformation remaining.
The creep curve in figure 2.23 shows the deformation for a relatively
low load/stress level. For low load levels the creep curve is convex.
Higher load/stress levels can lead to a rapidly increasing deformation
and lead to creep rupture, see section 2.4.4, page 38.
The deformation can be divided into three components; elastic,
delayed elastic and viscous deformation. The elastic deformation
appears directly after loading. After that the deformation will increase
slowly under a constant load. This increase in deformation is made
up of both the delayed elastic and the viscous deformation. The dif­
ference between these two components is visible first after unload­
ing. The delayed elastic deformation is reversible with time while
Figure 2.23 Creep curve, above the load-time curve and the viscous deformation is permanent.
below the deformation load-time curve There are several external factors that influence the magnitude of
the creep; temperature, load direction, stiffness, knots and moisture
content:
• Load direction — the amount of creep is different in different load
directions.
• Stiffness — tests have shown that material with higher modulus of
elasticity not only get less instantaneous deformation; the additional
creep deformation is also smaller for material with a higher modu­
lus of elasticity.
40 Design of timber structures - Volume 1
 2.4 Mechanical properties
of wood and timber

• Knots — timber beams with more knots (and lower MOE) exhibit
larger creep deformation under load.
• Moisture content — tests have shown that wood with higher mois­
ture content exhibit larger creep deformations than wood with
lower moisture content. Varying moisture content in combination
with loading have, however, been shown to give a significant
increase of creep deformations.
• Temperature — increased temperature leads to increased creep
deformations. This is especially noticeable at temperatures over
100 °C when the lignin is softening.

Increasing deformations for timber under varying climate conditions

are known as the mechano-sorptive effect. This effect was first
described by researchers from Australia during the 1950’s. This effect
is illustrated in figure 2.24 where the results from a study of creep in
small beech specimens loaded in bending are shown.
The influence of moisture content and load duration is important
when designing a timber structure. Engineered wood products, pro­
duced of wood and some kind of adhesive, may have different creep
properties than sawn timber. In these types of material the creep
behaviour may be governed by the properties of the adhesive. The creep
is larger for some engineered wood products than for sawn timber.
Creep is often taken into account in design, it is for example
important in the analysis of stresses in composite structures but
maybe most important in the calculation of long-term deformations
in the serviceability limit state. The increase in deformation due to
creep (ucreep) after an infinite time is often defined as a creep factor,
denoted (kdef) in Eurocode 5, times the instantaneous deformation
(uinst), see equation 2.7:

“creep = Winst 2.7

With this definition it is possible to calculate the final deformation,

see equation 2.8:

“fin = “inst + “creep = “inst (1 + *def ) 2.8

Figur 2.24 The deflection of small beech specimens loaded in bending to 1 / 8 and 3 / 8’s of
the short term strength in constant and varying climatic conditions, Hearmon and Paton (1964)

Design of timber structures - Volume 1 41

2 .5 Strength grading

With the help of the creep coefficient it is possible to estimate how

large the deformation of a structure will be after a long time with
a constant load. For normal sawn timber the creep factor kdef is
0,6 — 0,8 for normally dry conditions and as large as 2,0 for timber
subjected to high and varying moisture content.

2.5 Strength grading

The strength, stiffness and density of wood vary to a great extent. For
Norway spruce the variation in bending strength can be between 10
and 90 MPa. To be able to use this material in load-bearing structures
Frequency

it is necessary to have a better knowledge of the properties. The wood

material is produced by nature and it is therefore not possible to con­
trol the variation in properties by changing the manufacturing pro­
cess. For wood it is instead necessary to get an estimate of the proper­
ties and grade the material into different strength classes.
The characteristic strength value for all materials is normally
defined as the 5 %-fractile in the distribution of strength, see
Figure 2.25 Principle of the variation in strength for figure 2.25.
timber and the material split into three strength classes By grading the material into different strength classes it is possible to:
• improve the control of timber characteristics such as strength and
stiffness
• have a common classification within a market
• optimise the yield from the raw material
• optimise the use (good enough quality).

According to most standards the material is graded based on its bend­

ing strength. The mean modulus of elasticity in bending and density
also have to be checked so that it is within the limits of the grade. All
other parameters are estimated based on these values.
There are currently two types of grading: visual strength grading
and machine strength grading.

Visual strength grading

The visual strength grading technique relies on the assumed correla­
tion between visually detected defects and strength. In principle
there are grading rules establishing which type and size of defects
can be allowed in each strength class. The first visual strength grad­
ing rules were established in the US in the 1920’s and visual grading
rules were also established in several European countries during
the 1930’s. The grading rules are of course different for different
wood species, number of grades and so on. Traditionally visual
strength grading has been conducted by human inspectors looking at
each piece of timber as it passes them on a conveyor. This of course
gives problem in accuracy since the grader only has a couple of sec­
onds to assign the board to a certain grade. It also has the natural
drawback of only seeing the visual defects, defects within the mate­
rial is not possible to detect. The method only allows for the use of
simple combinations of defects. Today, there are scanning techniques
possible to use together with the visual strength grading rules that
improve the technique.

Machine strength grading

The machine strength grading technique relies on running the pieces
of timber through a machine that measures one or several parame­
ters non-destructively. These parameters are then used to predict

42 Design of timber structures - Volume 1

 2.5 Strength grading

the strength and possibly stiffness. The idea to use non-destructively

measurements as a basis for prediction of strength and stiffness was
presented in the 1950’s in both USA and Australia. The reason was
a wish to improve the accuracy of the grading. The most commonly
used parameters to measure non-destructively are modulus of elastic­
ity and density.

2.5.1 Relationship between strength,

stiffness and other parameters
The strength grading principles are built on measuring one or more
parameters nondestructively and use this measure to predict
the strength (and possibly stiffness and density). This means that to
use the grading principles it is necessary to have a good knowledge
about the relationship between these parameters and strength and
the influence of different kind of natural characteristics on these
relationships. For clear wood the following relationships have been
found between strength, density andmodulus of elasticity (MOE):
• Bending strength — modulus of elasticity R2 = 0,70 — 0,75
• Bending strength — density R2 = 0,60 — 0,65
• Modulus of elasticity — density R2 = 0,60 — 0,65

The properties in timber, however, vary to a large degree between dif­

ferent trees, logs and even within a tree or log. The natural character­
istics of wood also influence the parameters and their relationships,
and it is not only the size of the defects but also their location that
has an influence. In visual grading the occurrence of knots is the sin­
gle most important parameter for assigning boards into different
strength classes. Knots have also shown to be the most important fac­
tor in true strength tests. In strength tests of Norway spruce it has
been shown that approximately 90 percent of the failures start at
Nya Nibble farm, Ytterjarna.
the position of a knot. Spiral grain angle is the cause of failure in
some cases. Compression wood has a pronounced effect on the rela­
tionships between these parameters, it has high density and low MOE
but the strength is only marginally decreased.
There are a number of different non-destructively measurable
parameters that can be used in strength grading. Numerous tests have
been done to establish the relationship between different parameters
and the bending or tension strength of sawn timber. Studying a few
of these test series gives the following coefficient of determination
between bending strength and other parameters, see table 2.3.
The best predictor for strength is the stiffness or modulus of elas­
ticity. The coefficient of determination varies between 0,5 and 0,7

Table 2.3 Coefficient of determination between bending strength and a non-

destructively measured parameter based on a number of studies on sawn timber
(Johansson 2003)

Measured parameter R2
Knots 0,16 - 0,27

Annual ring width 0,20 - 0,44

Density 0,16 - 0,40

MOE, bending or tension 0,53 - 0,72

Knots and density 0,38

Knots and MOE 0,58 - 0,73

Design of timber structures - Volume 1 43

2.5 Strength grading

Figure 2.26 The relationship between the modulus of elasticity (edgewise) and bending
strength for 380 studs of the dimension 45 x 70 mm

between different studies; the difference is mainly due to different

measurement methods for the MOE. A typical result from a test of
bending strength and stiffness can be seen in figure 2.26.
The difference can also be attributed to different knot sizes in
the material in different studies. The modulus of elasticity includes
information about knots (lower MOE at the position of the knot), spi­
ral grain angle (lower MOE) and compression wood (lower MOE) which
is why it works better than the other parameters. It is of course also
possible to combine different parameters when grading timber.
Adding information about knots together with information about
the average density strengthens the prediction of strength. Adding
more information to the MOE does not strengthen the prediction to
any higher degree, probably because the MOE already includes this
information.

2.5.2 Machine strength grading principles

There are many techniques used in strength grading machines; static
flat-wise bending, MOE determined from vibration or measurement
of density by radiation are the most commonly used techniques.

2.5.2.1 Static flat-wise bending

Flat-wise bending of the timber is the principle that has been used
the longest and was the dominating technique among the grading
machines up until the early 21st century. The technique is based on
measuring the flat-wise MOE in three-point bending. The boards are
continuously feed flat-wise through a machine with three rollers and
the MOE is evaluated based on either 1) measurement of deformation
for a constant load or 2) measurement of the load necessary to create
a pre-set deformation, see figure 2.27, page 45. This gives a measure of
the variation of the MOE along the board. With a known MOE it is
possible to estimate the strength of the board and assign it into a spe­
cific strength classes.

44 Design of timber structures - Volume 1

 2.5 Strength grading

Figure 2.27 Principle of a static bending machine

2.5.2.2 MOE determined from vibration

The dynamic modulus of elasticity of a board can be determined by
introducing vibrations into it and evaluate its resonance frequency.
The principle is built on theory for vibration of prismatic beams and
the equation for determining the MOE from the first resonance fre­
quency for free-free support conditions in the axial direction can be
expressed as:

2912 29^2P 29

where EA1 is the modulus of elasticity, p is the density at the measure­

ment time, fA1 is the first resonance frequency in the axial direction
and l is the beam length. For other support conditions than the free-
free and for higher order frequencies the equation has to be modi­
fied. It is also possible to use transverse vibrations. For clear wood
specimens the dynamic MOE correlates well with the true MOE,
the same has been seen in several studies also for timber. The method,
however, has the drawback of measuring an “average” MOE of
the board while the static bending machines have, at least in theory,
the capacity to localise the position with the minimum MOE along
the board. Several studies have shown that for material with limited
knot size the techniques works as well as or better than the tech­
nique using static bending.

2.5.2.3 Other measurement techniques

With x-ray, y-rays or microwaves it is possible to measure the density
variation within a piece of timber. Several of the natural characteris­
tics of wood imply a density that is different from normal wood.
Knots and compression wood show for example a higher density and
juvenile wood has a lower density. Using enough accuracy in
the measurement procedure it is possible to characterise, for example
the size, position and shape of the knots within a piece of timber.
With known knot structure and threshold values for knot areas it is
possible to create a machine that can grade timber into different
strength classes.
There are also other types of techniques that can be used. One is
based on the wave velocity; this technique is based on measuring
the time it takes for a pulse, generated by a hammer, or ultrasonic
sound, to pass longitudinally through a board. The speed of the sound
wave through the material can be used as a basis to calculate

Design of timber structures - Volume 1 45

2.6 Modified wood

the average MOE of the board. It is of course also possible to combine

different types of techniques; for example combine a technique for
measuring MOE with a radiation technique to find the size and loca­
tion of knots. There are a number of grading machines on the mar­
ket; most of them are built on vibration techniques but almost as
many are built on flat-wise bending and x-ray. Some of these
machines also have the capability to use two of the techniques in
combination.

2.6 Modified wood

It is possible to change the properties of wood by modifying the cell
wall polymers and the matrix they are in. Modification of wood has
been the subject of years of research mainly focussed on increasing
dimensional stability and biological resistance. These modifications
have been studied to replace toxic wood preservation chemicals such
as copper, chrome and arsenic and creosote.
Over the past 15 years, a number of alternatives, more environ­
mentally friendly wood modifications have been reinvestigated, pat­
ented and commercialized. There are three basic technologies, some
with variations: acetylation, furfurylation and heat treatments. These
modifications have in common that they all increase dimensional
stability, reduce wood moisture content and reduce attack by wood
destroying fungi, see Hon and Shiraishi (1991) and Hill (2006).

2.6.1 Processes
Acetylation is a reaction between acetic anhydride and hydroxyl
groups that are located mainly on the hemicellulose and lignin poly­
mers located in the cell wall. The bonded acetyl bulks the cell wall
almost back to the green dimensions of the original wood. The reac­
tion is carried out in a vacuum/pressure cylinder after the wood is
dried to below 10 % moisture.

Furfurylation is a method where the wood is pressure impregnated

with monomeric furfuryl alcohol. The key reaction is a polymeriza­
tion of the alcohol bulking the cell wall with or without crosslinking.
The process leads to a permanent swelling of the wood. Some of
the added chemical can be located in the lumens.

Heat treatments are done by heating the wood to high temperatures

(190 °C to 220 °C) in the presence or absence of oxygen and some­
times heated in oil. The high temperature leads to a loss of mainly
hemicellulose polymers resulting in a loss of mechanical properties.
The degradation is more severe if the process is carried out in
oxygen.

2.6.2 Properties of modified wood

Acetylation results in a slight darkening of light woods and a mild
bleaching of dark woods. The best results are seen with an acetyl
weight gain of 18 percent to 20 percent. Dimensional stability is greatly
improved (80 percent +), decay resistance against brown-, white- and
soft-rot fungi is greatly increased (less than 2 percent in a standard
soil block test) and the equilibrium moisture content is greatly
reduced (reduced from 21 % to 5 %). The mechanical properties

46 Design of timber structures - Volume 1

 2.7 Engineered wood products

(modulus of rupture and elasticity) are unchanged over controls in

dry tests but wet strength and wet stiffness of acetylated wood are
much higher than controls. Acetylated wood is used in doors, windows,
decking, outdoor furniture and claddings.

Furfurylation changes the color of the wood depending on the level

of treatment. Wood treated to a weight gain of 30 percent is dark red
to brown and wood treated to 100 percent is dark brown. Mechanical
properties are improved, decay resistance is increased and the equi­
librium moisture content is reduced 30 to 60 % compared to untreated
wood. Furfurylated wood is often used in decking and cladding but
can also be seen in outdoor furniture.

Heat treatments result in major losses in mechanical properties

depending on the temperature, time and method of treatment.
Mechanical properties can be reduced more than 50 percent. Heat
treated wood is generally darker and can smell somewhat burned.
There is some increase in decay resistance, the equilibrium moisture
Scissor lift and boom lift during assembly of glulam
content is reduced (40 % to 50 %) and the wood is more brittle as
and CLT frame of a school building, Skelleftea.
compared to untreated wood. Heat-treated wood is used in non-struc-
tural applications such as decking and cladding.

2.7 Engineered wood products

Normal sawn timber can be found only up to certain dimensions,
due to the size of the trees and the industrial process. In Sweden
the maximum size of sawn timber is a depth of 245 mm and a length
of less than 5,5 meters. For larger dimensions it is necessary to use
some kind of Engineered Wood Product or EWP for short. EWPs are
made of wood in the form of sawn timber boards, veneers, particles
or fibres held together with, in most cases, some type of adhesive.
During the 1900’s a large number of different EWPs both in
the form of beams and in the forms of panels have been invented.
Figure 2.28 shows a timeline for the invention of different types of
EWPs.
The majority of these products were invented in North America
where lack of material from old growth timber has lead to an increased
use of material from smaller diameter trees, lower quality logs and
new tree species. In a controlled production process it is possible to
develop new value added products with good structural properties.

Glulam LVL I-joist PSL

Plywood Masonite Particle board MDF OSB X-Lam

Figure 2.28 Approximate year of development of different types of EWPs, beam type products above the time line
and panel type products below the time line. LVL - laminated veneer lumber, MDF - medium density fibreboard,
OSB - oriented strand board, PSL - parallel strand lumber, X-Lam - cross-laminated timber.

Design of timber structures - Volume 1 47

2.7 Engineered wood products

2.7.1 Engineered wood products

based on sawn timber boards
EWPs based on gluing together pieces of sawn timber can be produced
in the form of beams — glulam beams, or in the form of thick panels,
often glued with the sawn timber boards oriented in different direc­
tions in different layers, referred to as Cross-Laminated Timber — CLT.

2.7.1.1 Glued laminated timber - glulam

The oldest of the engineered wood products is glued laminated timber.
The first patent for a glue lamination process for sawn boards was
High strength awarded to Otto Hetzer in 1906. Glulam consists of several (at least
four) laminations (finger jointed sawn boards) bonded together with
adhesives. The boards are all oriented with their main fibre direction
in the axial direction of the finished glulam beam.
Normal straight glulam in Sweden consists of 45 mm thick lamina­
tions with a width of up to 215 mm. Wider beams can be produced
Low strength by gluing two or more glulam beams together. Glulam can either be
h

homogeneous or made to match the design stress levels in the beam

with higher strength laminations in the outer higher stressed regions
of the beam. This type of glulam is called combined glulam, see
figure 2.29.
Straight glulam is produced in a highly mechanised process.
The normal process is to start with the sawing and drying of the lam­
High strength inations. For the gluing process the laminations are dried to a moisture
content between 6 and 15 %. All laminations, have to have less than
5 % difference in moisture content, are thereafter strength graded.
Figure 2.29 The lay-up for a combined glulam beam The laminations are finger jointed together to create the correct length
of the laminations. The laminations are planed and the adhesive is
thereafter spread on the flat surfaces. In Sweden, the adhesive is
commonly a Melamine-Urea-Formaldehyde (MUF) adhesive, in some
special applications Phenol-Resorcinol-Formaldehyde (PRF) adhesive
are still used. The boards are then laid up in a press were the adhesive

Spruce logs Production of lamellas Kiln drying Strength grading

Finger jointing Planing Application of glue

Glueline pressure applied Planing of glulam element

Figure 2.30 The manufacturing process for straight glulam beams

48 Design of timber structures - Volume 1

 2.7 Engineered wood products

Difference in characteristic strength Difference in mean strength

Strength

Figure 2.31 Distribution function for the strength of glulam beams and structural timber

cures under a certain pressure and in a specified temperature.

The beams are finally planed and possibly a layer of wax to protect it
from moisture is applied. This mechanised production process also
includes quality controls in terms of regular checks of the strength of
the finger-joints and delamination tests of the glue line.
The beams can also be manufactured with a curvature, either
a small pre-camber to minimize deflection or as a curved beam.
The curved beams are normally made with thinner laminations and
forced into the desired curved shape before curing of the adhesive.
Tests have shown that in average glued laminated timber is not sig­
nificantly stronger than solid beams of the same size but the variabil­
ity in strength is lower, see figure 2.31. The low variability has often
been explained by smearing-out effects, i.e. cutting the solid wood
into smaller pieces and glue them together randomly causes the low-
strength defects to be more uniformly distributed and each defect has
a less significant effect.
The most common failure for a glulam beam subjected to bending
is a tensile failure parallel to the grain of the outer lamination, often
originating in a knot or a finger-joint. Another type of failure that can
occur is shear failure. Cracks can occur along the grain due to drying
effects and some researchers state that this will reduce the shear
capacity of the beams, but other researchers do not agree with this
conclusion. High tensile stresses perpendicular to the grain need to
be taken into account, especially in curved beams and beams with
holes and notches.

2.7.1.2 Cross-laminated timber - CLT

Cross-laminated timber (CLT) is a product that has won significant
market shares in the timber industry in the 2000’s. Cross-laminated
timber is made of sawn timber glued together in a layered structure
with the boards in each layer placed perpendicular to the layer above
and below; see figure 2.32. The exact lay-up of the cross-laminated
boards is dependent on the manufacturer. Cross-laminated timber
can be manufactured as a 3, 5, 7 or even more layer product with
varying thickness of the different layers. The use of an uneven num­
ber of layers implies that the outer layers always are oriented in Figure 2.32 Cross-laminated timber panel
the same direction, and better strength is obtained. with three layers

Design of timber structures - Volume 1 49

2.7 Engineered wood products

The size of CLT panels depends on the possibilities of the manufac­

turer but generally they are at least 60 mm thick and can be as large
as 500 mm thick, 3000 mm wide and with a length of up to 24 m.
The strength parameters for CLT are so far mainly set for each manu­
facturer after testing of the material by an independent third party.
CLT panels can be used as load-bearing vertical walls and as horizontal
floor diaphragms. CLT panels often arrive to the building site ready
for assembly with for example holes for doors and windows as well
as grooves for electrical fittings already prepared. The customization
of the elements is often done already at the factory with the help of
advanced CNC-machines.
There are also cross-laminated timber boards manufactured using
nails or screws instead of adhesive. The properties of this type of pan­
els vary somewhat from glued panels.

2.7.2 Engineered wood products

based on veneers
Another way of utilizing the raw material is to produce veneers from
the logs. Veneers are thin (2 — 4 mm) layers of wood. The logs are
debarked and steamed in hot water before being rotary peeled; see
figure 2.33. These thin veneers are then dried to the target moisture
content 12 — 15 % and often also strength graded before being used
in EWPs.
The veneers are then glued together using adhesives into different
formats of structural elements. The two most commonly used products
Figure 2.33 Veneer production from logs are laminated veneer lumber (LVL) and plywood.

2.7.2.1 Laminated veneer lumber - LVL

Laminated veneer lumber is made by gluing the wood veneer sheets
together to form thick (20 — 90 mm) structural panels with maximum
size 3 000 x 24 000 mm. For normal LVL all the layers are oriented
with the fibre direction in the same direction, normally in the long
direction of the finished product, see figure 2.34. After gluing and cur­
ing the thick panels are sawn into board dimensions. The process of
gluing the sheets together creates a structural element of higher reli­
ability and lower variability through defect elimination and distribu­
tion of defects, in the same way as for glulam.
In general LVL has high bending, tension and compression
strength, as well as high shear strength and a relatively high modu­
lus of elasticity. There is a considerable size effect in LVL that has to
be taken into account when calculating design values in bending for
LVL. Some grades of LVL are produced with some of the layers with

Figure 2.34 Laminated-veneer lumber - LVL, all veneers with the same fibre orientation

50 Design of timber structures - Volume 1

 2.7 Engineered wood products

the grain direction perpendicular to the fibre direction of the major­

ity of the layers. This production method allows for a production of
structural panels with a higher stiffness across the panel which can
be beneficial in some structural components, such as floor structures.

2.7.2.2 Plywood
Plywood was one of the first EWPs to be produced. It is produced in
much the same manner as LVL, but with the veneers laid up perpen­
dicular to each other. The number of veneers (plies) are always odd,
meaning that the fibre direction of the outer layers are always in
the same direction; see figure 2.35. Structural plywood is most com­
monly produced in sheets of the size 1 200 x 2 400 mm or 1 220 x
2 440 mm with a thickness of 12 — 24 mm. The outer veneers nor­
mally have the fibre direction oriented in the long direction of
the panel.
The structural properties of plywood are dependent on the number
and thickness of the veneers and load direction. It is important to
keep in mind the direction of the forces, bending perpendicular to
the plane of the panel and in-plane bending, see figure 2.36. The layers
with grain direction in the same direction as the normal stresses have
the highest stiffness and take the major part of the load. The contri­
bution to the load bearing of the layers with grain direction perpen­
dicular to the direction of the normal stresses is very small and can
in practice be disregarded.

2.7.3 Engineered wood products based on

strands, chips or fibres
EWPs can also be manufactured based on smaller wood elements
such as strands, chips or fibres. These small elements are glued
together in industrialised processes to form panels of different for­
mats. Due to the processing technique, these types of products have
in general a high density in the surface layers and a slightly lower
density in the middle. The strength of the product is also to a large
extent governed by the production technology, but also to a large
extent by the type and amount of adhesive used in the product.

2.7.3.1 Oriented strand board - OSB

The most commonly used panel for structural use is oriented strand
board — OSB. OSB is manufactured from thin wood strands (size
0,8 x 13 x 100 mm) sliced from mostly small-diameter logs. These
strands are bonded together using adhesive in a ratio approximately
95 percent wood and 5 percent adhesive under heat and pressure.

a) b)

Figure 2.36 Bending a) perpendicular to the plane and b) in-plane bending including distribution of normal stresses.

Design of timber structures - Volume 1 51

2.7 Engineered wood products

The strands are in the process oriented to form the best possible
product, this means that in the outer layers the strands are oriented
parallel to the long direction of the panel and in the inner layer with
the strands randomly oriented.
The most commonly used OSB product is in the form of panels with
a size of 1 200 x 2 400 mm with a thickness between 6 and 25 mm
used as sheathing material in walls or in floor structures. OSB can
also be manufactured in larger panels with width up to 3 m, length
up to 25 m and thickness up to 75 mm. These panels can be used as
structural elements in the same manner as CLT panels.

2.7.3.2 Chip, particle or fibre board

Chip, particle and fibre boards are all produced of wood chips of dif­
ferent sizes mixed with adhesives and hot pressed to panel form.
Chipboards are made of wood chips slightly smaller than the strands
in OSB, and are normally produced with 85 percent wood chips and
15 percent adhesive. Chipboards are uncommon in Sweden. Particle
boards are made from smaller wood particles; saw-dust combined with
adhesives. Particle boards are today mainly used in floor structures
but are often replaced with plywood or OSB.
Fibre boards can be produced using wood fibres and adhesive in
a process similar to the production of particle board or in a wet-process
using the lignin within the wood material as a binder. The later pro­
cess was developed by William H. Mason. The most commonly used
name for this type of product is masonite. Masonite is used as a struc­
tural element while other types of fibre boards are more commonly
used in joinery and furniture.

2.7.4 Built up structures - I-beams

Another type of EWP is I-beams, see also Chapter 5, page 115. These
beams are made up with flanges of sawn timber or LVL and with
webs of some kind of board material, plywood, OSB or masonite, see
figure 2.37. These beams are produced in automated production pro­
cesses under control of third party organisations. The webs are glued
to the flanges either in grooves in the flange or between two pieces of
the flange.
I-beams can be seen as built-up structures and designed as such.
The quality control for the manufacturing process is, however, very
difficult to fulfil without an industrialised production process. This
has lead to that most I-beams are produced in factories and that
characteristic strength values are given by the manufacturer.

52 Design of timber structures - Volume 1

 2.8 End-user requirement
on sawn timber

2.8 End-user requirement

on sawn timber
Different end use of the timber product means different requirements
on the sawn timber. A building consists of many parts to build a whole
structure; these parts are subjected to different requirements in terms
of geometrical tolerances and different types of loading for example.
This section presents some different types of commonly used struc­
tures and their requirements in terms of geometry, visual appearance
and mechanical properties.
A floor structure is a load-bearing horizontal structure with
the main purpose to divide to different storeys and to carry loads
from furniture and people transversally. A floor structure is often
produced with wooden beams (or EWP beams) with a spacing of
300 mm to 600 mm with a flooring panel above. Floor structures
have to fulfil three main design criteria; strength in terms of bending
and shear, deflection and vibrations. The design of a floor structure
according to the most commonly used codes, for example Eurocode 5,
shows that the criteria for deflection and vibration are the hardest to
meet. These are both dependent on the stiffness (modulus of elastic­
ity) of the material. For sawn timber used in a floor structure a high
modulus of elasticity is the most important material property. To
meet the requirements for a plane floor the distortions of the sawn
timber can not be too large. The most critical distortion mode for
a floor structure is spring as this will directly affect the shape of
the floor. Twist is also important as it is difficult to reach a good con­
nection between the flooring panel and the wooden beams when
the beam is heavily twisted.
Another commonly used structural element is the timber framed
wall with sawn timber studs. In this structure the sawn timber is Limnologen, Vaxjo.
used as vertical studs covered with a sheet material on both sides.
The main load on a wall is normal forces from above and a distrib­
uted load from the side caused by wind loads. The main design crite­
ria for a wall stud are compression parallel to grain and bending.
There is, however, also a criterion from building physics that requires
insulation. The requirement for insulation in the wall makes it nec­
essary to use larger beam dimensions than what is normally neces­
sary for the load bearing purpose. The requirement for strength of
a wall stud is therefore normally rather low. The requirement for
shape stability is still rather hard. The requirements for a straight
wall mean that there are strict requirements for twist of wall studs.
Other products like for example timber for roof trusses have high
requirements on tensile strength. There is also a requirement to limit
the occurrence of knots in the areas used for the connections in
a roof truss. Tensile strength is also important for material to be used
as laminations in a glulam beam. There are also geometrical require­
ments to be able to use the board in the glue lamination process. For
sawn timber delivered to a glulam factory there is also a strict crite­
rion for the moisture content.

Design of timber structures - Volume 1 53

3.1 Prismatic beam elements

Design of structural
timber elements
in ultimate limit state
Roberto Crocetti | Annika Martensson

3.1 Prismatic beam elements

3.1.1 Bending and shear 54
54
3.1 Prismatic beam elements
3.1.2 Axial forces 57
3.1.3 Bearing capacity 58 3.1.1 Bending and shear
3.1.4 Beams with holes and notches 60
Flexural members are those elements in a structure that are subjected
3.2 Design of slender members 62 to bending, and several types and forms of such members are used in
3.2.1 Column buckling 62
timber construction. Typical examples are solid section rectangular
3.2.2 Combined bending and axial compression 64
beams, floor joists, and purlins, see also figure 3.1.
3.2.3 Lateral buckling 65
3.2.4 Buckling of frames and arches 70 This section deals in detail with the general requirements that are
necessary for the design of flexural members made from straight solid
3.3 Special timber elements 75 timber or wood-based structural products of uniform cross-section in
3.3.1 Tapered beams 76
which the grain runs essentially parallel to the member axis.
3.3.2 Curved beams and
For beams design checks have to be done with respect to moment
pitched cambered beams 77
3.3.3 Design procedures 80 capacity, shear capacity, deflections and vibrations. In general, bend­
ing is the most critical criterion for medium-span beams, deflection
3.4 Portal frames 84 and vibration for long-span beams, and shear for heavily loaded
3.5 Arches 86 short-span beams.

b)

Figure 3.1 Examples of flexural members. a) External timber structure, b) floor joists.

54 Design of timber structures - Volume 1

 3.1 Prismatic beam elements

Figure 3.2 a) Cross-section of a rectangular beam.

b) Orientation of coordinate system in these compendiums.

The elastic theory of bending states that when, for example, a solid
rectangular member as shown in figure 3.2 is subjected to a bending
moment M about the y—y axis (the strong axis), the design stress at
any distance z from this axis will be:

M-z
3.1

where Iy is the second moment of area of the cross-section about

the y—y axis. The term Iy/ 0,5h is referred to as the section modulus
about the strong axis and is denoted by the symbol Wy. This means
that the moment capacity of a timber beam can be determined from:

MRd = /m,y,d'Wy 3.2

where fm,y,d is the design value of bending strength of the material.

The index R refers to resistance.
Similar expressions may be derived for the design bending capacity
about the z — z axis (the weak axis).
Lateral instability may affect a member subjected to bending when
the compression zone of the member is not fully restrained against
lateral movement and the relative slenderness for bending is high.
The design procedure for taking the effect of lateral instability into
account is given in section 3.2.2, page 64.
If a member is subjected to bending about both axes and the rela­
tive slenderness ratio for bending about the y—y axis is such that lat­
eral instability can be neglected, the design conditions to be met are
given by:

WyRd ^Rd
3.3

k+
^yRd ^zRd

where km is a modification factor that allows for redistribution of

stress in the element. The index E refers to forces and moments deter­
mined from the load effect.

Design of timber structures - Volume 1 55

3.1 Prismatic beam elements

Values for the factor km are given by Eurocode 5: For structural timber,
LVL and glued-laminated timber, km = 0,7 for rectangular sections,
and km = 1,0 for other cross-sections. For other wood-based structural
products, km = 1,0 for any cross-section.
When a beam is subjected to bending, shear stresses will also arise.
In accordance with elastic bending theory, shear stresses will be gen­
erated parallel to the longitudinal axis of the beam. The value of
the shear stress at any level in the cross-section of a beam, as derived
from elastic theory, is:

vs 3.4
I-b
where:
t is the shear stress at the required level.
V is the shear force at the position being considered.
S is the first moment of the area above the shear stress level
about the neutral axis.
I is the second moment of area of the cross-section
about the neutral axis.
b is the width of the cross-section at the shear stress level.

At any position along the beam the shear stress at the top and bot­
tom faces of the cross section will be zero and the maximum shear
stress will arise at the neutral axis position. For a rectangular section
of width b and depth h the maximum shear stress will occur at mid­
depth and will be:

3K
T =-------- 3.4 a
2b-h
This means that the shear capacity is given by:

V 3.5
V'Rd- 3

where:
A is the cross section area.
fv,d is the design shear strength.

The shear capacity may be reduced by cracks in the wooden member

caused by moisture induced stresses. This is considered in
Eurocode 5, which recommends that a reduced effective width bef
shall be used in equation 3.4 instead of the actual width b.

56 Design of timber structures - Volume 1

 3.1 Prismatic beam elements

3.1.2 Axial forces

This section covers the design of straight structural timber or glulam
in which the grain runs essentially parallel to the member lengths
and the members are subjected to the effects of axial compression or
tension.

3.1.2.1 Tension
Although the tensile strength, ft,0,k, of clear wood samples is greater
than the compression strength, fc,0,k, because tension failure occurs in
a brittle rather than in a ductile mode and also because of its sensi­
tivity to the effects of grain slope, knots and other defects, the tensile
strength of structural timber is generally less than the compression
strength. This is particularly the case for the lower strength classes.
Design with respect to tension is normally very simple and the main
consideration is the strength value of the wooden member. In some
cases it can be relevant to make considerations of the volume sub­
jected to the tensile stresses since it has been shown that the strength
is size dependent, see section 2.4.6, page 39.
In some cases the wooden members will be subjected to both bend­
ing and axial loading. In the case of combined bending and axial ten­
sion the design criteria are quite straightforward, that is failure will
not occur when the following relation is valid:

^Ed ^Ed <1

3.6
^Rd A\,Rd

where Nt,Ed is the tensile force due to external loads.

3.1.2.2 Compression
Examples of members subjected to compression forces are given in
figure 3.3 (columns, posts, stud members in walls or struts in trusses).
These are members that are subjected to a compressive action acting
parallel to the grain and along the centroidal x—x axis of the member.
When subjected to an axial load, as the slenderness of the member
increases there is a tendency for it to displace laterally and to eventu­
ally fail by buckling. Since the main focus of members subjected to
compression parallel to grain is on buckling stability, compression

Figure 3.3 Examples of members subjected to compression forces

a) Columns in a timber frame, b) studs in a multi-storey timber framed building, c) external columns.

Design of timber structures - Volume 1 57

3.1 Prismatic beam elements

will be discussed in total in section 3.2, page 62. The same applies for
members subjected to compression and bending, it will also be dis­
cussed in section 3.2.

3.1.3 Bearing capacity

In many structural systems based on timber or other wooden prod­
ucts, wood will be subjected to compression perpendicular to grain,
see figure 3.4. Since wood has low stiffness in this direction and exhib­
its large moisture induced movements, compression will lead to large
deformations. When this occurs in a connection between floor and
wall structure it may lead to undesired settlements in the buildings.
Although the main problem with compressive forces perpendicular
to grain usually is the deformation and not actual failure, most codes
treat this problem as an ultimate load situation, as a design with
regard to failure.
When timber is compressed perpendicular to the grain the wood
fibres withstand increased loading as they are squeezed together and
as they start to collapse the rate of load increase is reduced, see also
section 2.4.1.4, page 35. This behaviour continues until the fibres are
fully squashed and if the wood is strained beyond this stage the sus­
tained load will continue to rise. It is hard to define a clear failure in
this mode. The deformations may be very large but the wood will
still withstand the load. Therefore, the value of the compression
Figure 3.4 Timber frame structure
strength perpendicular to grain is often given as a limit value defined
Positions where significant deformations due to compressive
forces may occur are indicated by a circle. by a certain value of the strain. However, depending on the position
of the bearing area, the bearing length and the member depth, this
limit can be exceeded.
Examples of bearing effects (failures) are illustrated in figure 3.5.
The design check in the ultimate limit state can then be given by:

°c,90,d - ^c,90 ’ /c,90,d 3.7

Here oc90d is the design compressive stress perpendicular to the grain

and fc,90,d is the design compressive strength perpendicular to
the grain, while kc,90 is a factor that takes into account how the load­
ing is being applied. The value normally used for kc,90 is 1,0 but
higher values can be adopted if certain conditions apply, for instance

Figure 3.5 Bearing perpendicular to grain

58 Design of timber structures - Volume 1

 3.1 Prismatic beam elements

when a stud is situated on a rail, so that only a small part of the rail
is subjected to compression forces. Consider for example the case
shown in figure 3.5 d), page 58. If the rail rests on the whole bottom
surface, the stress will be less at that boundary than at the top
boundary, since the loaded area is larger there. It should also be noted
that the stressed area on the top also can be considered to be larger
than the cross section area of the stud, since the fibres in the rail in
the vicinity of the stud increase the resistance against the force.
The latter factor can be taken into account either in the factor kc,90 or
by dividing the force with a larger area than the cross section area of
the stud. This would mean that ac,90,d„„. can be determined either by
N/bh or by N/Aef where N is the design bearing force, b is the width of
the stud and h is the depth of the stud and Aef is the effective contact
area, which then is larger than bh.
It is in many cases important to estimate the deformation in
the zones subjected to compression since in most cases it is the defor­
mations and not the actual failure that will be of largest interest.
This is of larger importance in multi-storey buildings where
the deformations will add up from each floor.

Design of timber structures - Volume 1 59

3.1 Prismatic beam elements

a) b) c)

I
e)

Figure 3.6 Various types and causes of fracture perpendicular to grain

3.1.4 Beams with holes and notches

In structural design of timber structures it is important to avoid ten­
sile stresses perpendicular to grain since the strength in that direction
is very low, Gustafsson (1995, 2003). Tensile stresses can be induced
in several situations, see also figure 3.6:
• The geometrical shaping of the structural member can induce ten­
sion perpendicular to the grain when the structure is loaded. This
can be the case for curved members as well as for beams with
holes or notches, see figure 3.6 a), b) and c).
• Joints of different types can also lead to stresses perpendicular to
the grain, see figure 3.6 d), e) and f).
• Variations in moisture content can lead to undesired stresses
within an element, especially if abrupt changes in cross section
area occur.

Here, the problem with holes and notches will be discussed. Both
holes and notches may significantly reduce the load bearing capacity
of a beam and should preferably be avoided in design. Since tensile
forces are induced around a hole or a notch the failure will be brittle
and thereby sudden which means that there will be nearly no warning
before failure as there can be with other type of failures. In figure 3.7
the distribution of tensile stresses near the tip of a notch is presented
and it is obvious that the stresses can be larger than the tensile
strength. The theoretically calculated elastic stress will actually go

Figure 3.7 Stress at the tip of a notch according to linear theory and as
estimated in practice, full line shows theoretical stress and dotted line
estimated stress

60 Design of timber structures - Volume 1

 3.1 Prismatic beam elements

towards infinity. Shear stresses may also lead to failures in the vicinity
of a sudden change in cross section area.
As the very high stress is often concentrated in a very small region
it is difficult to try to determine the load bearing capacity of a beam
with a notch or a hole with a conventional stress criterion. Instead it is
necessary to rely on tests or on concepts based on fracture mechanics
instead of conventional stress criteria.
Fracture mechanics is a part of the science of the strength of mate­
rials. A solid body responds to extreme loading by undergoing large
deformation or fracture. Often the main interest for the engineer is
the magnitude of the load that causes fracture. In cases when there is
no or only minor stress concentrations, for example in the case of
a structural member in homogenous tension or bending, the calcula­
tion of the fracture load can be carried out by a conventional stress
criterion. On the other hand in the case of a very high stress concen­
tration some other approach is needed. Then different analyses can
be carried out applying theories of fracture mechanics. There are
a number of different alternative theories but none of them will be
described thoroughly here, only the basic ideas for the theory that
constitutes the equations given in Eurocode, namely linear fracture
mechanics especially analysis of the energy release rate. In this theory
the energy release in the fracture process region is the critical value,
that is when the energy release has reached a critical value Gc frac­
ture will occur. The critical value can be seen as a material parameter
that can be determined by tests. By using Gc the value of the fracture
load can be determined, Gustafsson (1995). Other parameters that are
necessary to know are geometrical characteristics for the notched
beam as well as the modulus of shear stiffness and the modulus of
elasticity parallel to grain since those affect the behaviour. In
the equations of Eurocode it has been assumed that t/(BGc) is propor­
tional to the shear strength, which has been shown to be the case
from tests.
In the design process the shear stress is controlled against the shear
strength in the following form:

where:
kv is a reduction factor that depends on the geometrical
conditions and material.
fv,d is the design value of the shear strength.
Vd is the design shear force.
b is the beam depth.
h is the beam height.
a is a non-dimensional parameter that is used to describe
the geometry of the notch.

For beams with a hole there is no strength equation given in Eurocode

but such relations can be found in other codes. The bases for the codes
differ and they don’t give the same results, Danielsson (2008). One
reasonable approach to deal with this is to avoid holes in timber
structures and when they are used to be observant of where they are
situated and use design equations with care. There are in general
some basic rules concerning for example hole placement, maximum
allowed hole size and minimum corner radius for rectangular holes
that need to be fulfilled.

Design of timber structures - Volume 1 61

3.2 Design of slender members

i iv

■^ \ a I tz \ a

Figure 3.8 Notations for regulations concerning hole geometry and placement

In table 3.1 and figure 3.8 examples of these are presented. It is gener­
ally recommended to place holes in the neutral axis of the beam.
Since there are large uncertainties in the few existing design rules,
the best thing at the moment is to assume that in areas with holes
practically no forces can be transferred by the timber but instead
reinforcement has to be used around the holes. This can for example
be done by reinforcing with rods or with plywood. In order to do so
there are some design advices given in literature that can be adopted
but also these should be used with great care.

Table 3.1 Examples on restrictions for geometry and placement of

unreinforced holes, Danielsson (2008)

Limtrahandbok DIN 1052:2004

lA > 0,5 h > 0,5 h

lv > h > h

lz > 1,5 h and > 300 mm > h and > 300 mm

hro > 0,35 h > 0,25 h

hru > 0,35 h > 0,25 h

a < 0,4 h < h

hd < 0,15 h < 0,4 h

r > 25 mm > 15 mm

3.2 Design of slender members

Timber members are often designed fairly slender. Therefore, it is
crucial to pay due attention during the design process, in order to
avoid problems related to stability.

3.2.1 Column buckling

As said previously there is a tendency for axial loaded members sub­
jected to compression to displace laterally which eventually can lead to
failure by buckling as shown in figure 3.9, page 63. The more slender
the member is the larger the risk for buckling. The slenderness ratio

62 Design of timber structures - Volume 1

 3.2 Design of slender members

is defined as the effective buckling length of the member, le, divided

by the radius of gyration, i, of the cross section:

i
where the radius of gyration about an axis i = ,iyA), I is the second
moment of area, and A is the cross-sectional area of the member.
Buckling will occur about the axis with the highest slenderness
ratio. The effective length le (or buckling length) of a compression
member is the distance along its length between adjacent points of
contra-flexure, where the bending moment in the member is zero.
Examples of values for the buckling lengths that can be used in prac­
tical design are given in figure 3.9. The values in figure 3.9 are slightly
higher than the theoretical values given by Euler theory because con­
sideration has to be taken to uncertainties in boundary conditions.
For an idealized perfectly straight column of length l having uniform
properties and pin jointed at both ends, the theoretical axial load at
which buckling will occur about the y — y or the z—z axes will be
the Euler buckling load for the respective axis. Expressing the Euler Figure 3.9 Effective buckling length le for different end
buckling load in terms of the effective length, it can be written as: conditions. l is actual column length.

3.9

The expression given above is a theoretical value of the load capacity

since no real column is perfectly straight and also other factors will
influence the behaviour of a column subjected to axial compression
such as:
• strength/stiffness — compressive strength and modulus of
elasticity of the material
• geometry of the member — cross-sectional sizes and length
• support conditions, which are accounted for by effective
buckling length
• geometric imperfections — deviations from nominal sizes, initial
curvature and inclination
• material variations and imperfections — density, effect of knots,
effect of compression, wood and moisture content.

This is usually described by introducing a reduction factor x that

depends on the slenderness ratio A and on the above mentioned factors
so that the compression capacity is given by:

NcR = %-fcA-A 3.10

where fc,d is the design value of the compression strength and A is

the cross section area.
For wooden members the influence of the above mentioned factors
have to be considered in the calculations. The reduction factor x has
been evaluated from simulations for a large number of columns,
Blass (1995). The columns were modelled by assigning them material
properties and geometric imperfections based on observations of real
columns. Material properties, geometry, initial curvature and devia­
tions from target values were chosen randomly for each simulated
column. Correlation between different properties was taken into
account. The calculation of the ultimate load for the simulated col­
umns was based on second order plastic analysis, using the plastic
deformation potential of timber subjected to compression. Based on

Design of timber structures - Volume 1 63

3.2 Design of slender members

Slenderness ratio of member, A

Figure 3.10 Typical graph showing the strength reduction in a structural

timber compression member as a function of the slenderness ratio

the simulations, buckling curves for different circumstances were

developed. In order to do so a measure called the relative slenderness
ratio A was defined:

3.11

where fc,0,k is the characteristic compressive strength of the timber

parallel to the grain and E0,05 is the fifth percentile of the elastic
modulus.
From the simulation results, curves have been derived in terms of
Arel,y (or Arel,z) from which buckling strength reduction factors, kc,y (or
kc,z), called instability factors, are obtained for structural timber and
glulam members subjected to axial compression, see also figure 3.10.
This gives the following form for compressive strength:

^c,Rd “ ' /c,0,d ' 3.12

N

3.2.2 Combined bending and

axial compression
For members subjected to compression and bending three cases can
be distinguished: the case when no instability effects are valid, a case
where buckling may occur and the case when both buckling and lateral
instability may occur. The last case will be discussed in section 3.2.4,
page 70. In the following the other two cases will be discussed.
For a member with a low slenderness ratio subjected to compres­
sion and bending, no buckling will occur and failure will be based on
the compressive strength of the member. Figure 3.11 shows interac­
tion diagrams for a member subjected to combined bending moment
and axial compression based on the application of elastic and plastic
theory. Under elastic theory, the failure condition will occur when
the combined compression stress in the member reaches the com­
pressive strength of the material. When plastic theory applies,
the material yields when it reaches the compressive strength allow­
and axial compression of a member about an axis
N and M is the applied load and moment and Nu and Mu is
ing the stress in the section to redistribute to some extent, which will
the ultimate load and moment, valid for either elastic (dashed enhance the strength. With the elastic theory approach, the sum of
line) or plastic (solid line) theory. the combined axial and bending stress to strength ratios cannot

64 Design of timber structures - Volume 1

 3.2 Design of slender members

exceed unity. With the plastic theory solution however, the axial
stress to axial strength ratio becomes a squared function enabling
the member to take a higher value of bending stress to bending
strength ratio for the same value of axial stress to strength ratio.
In the case where no strength reduction is occurring, that is when
the slenderness is low in the member, the strength benefits associ­
ated with the plastic behaviour of timber when subjected to compres­
sion stresses can be adopted. This is described in the following equa­
tion in the Eurocode 5 for the case with bending about both axes:

My^ | ^z,Ed [
3.13
,Rd Jid k ^c,O,Rd >

where km is a reduction factor that depends on the form of the cross

section, see section 3.1.1, page 54.
In the case when the slenderness value is high, that is when buck­
ling effects have to be taken into account, no benefit is taken of any
plastic behaviour in the member and the ultimate load is achieved
when the material reaches its failure strength in the extreme fibre.
This is in line with the elastic theory interaction approach shown in
figure 3.11, page 64. The equation to be used in design of members
subjected to both compression and bending is then given by Eurocode
in the following form for the case with bending around both axes:

My^A | ^z,Ed [ ^c,Ed

3.14
^y,Rd ^zjid ^c,y'^c,OJ<d

where kc is the reduction factor for buckling. The equation is valid

for buckling around the y — y axis.

3.2.3 Lateral buckling

3.2.3.1 Introduction
For an idealized, perfectly straight elastic beam, there will be no out-
of-plane deformations as long as the bending moment generated by
the applied external loading P is lower than the elastic buckling
moment Mcrit. When the bending moment generated by P exceeds
Mcrit, the beam buckles by deflecting laterally and twisting, as shown
in figure 3.12. If the compression zone of the beam is not braced to

Horisontal
displacement

Vertical
displacement

Rotation

Figure 3.12 Lateral-torsional buckling of simply supported beam

Design of timber structures - Volume 1 65

3.2 Design of slender members

Figure 3.13 Lateral buckling of a roof or floor system. a) Unbraced, b) braced.

prevent lateral movement, the member may buckle at a bending

stress that is less than the design value defined in section 3.1.1,
page 54.
In practice, however, as in axially loaded columns, beams are never
perfectly straight, nor perfectly homogeneous, and are often not
loaded in the plane that is assumed for design and analysis.
Rarely, moreover, does a timber beam exist with its compression
zone entirely free of all restraint. Most beams, in fact, are laterally
supported by such things as attached plywood or decking, nogging
pieces, purlins, or diagonal bracing.
It should be pointed out that it is necessary to examine not only
the individual beam for adequate bracing, but also the entire system.
Figure 3.13 a) shows a beam AB with a cross beam framing at
midlength; in such a case, buckling of the entire system is still possi­
ble unless the system is braced, as shown in figure 3.13 b). All too fre­
quently in design, the engineer encounters situations where the ade­
quateness of the bracing system against lateral torsional buckling is
uncertain. Typical examples are where the beam in question is not
braced at the compression zone but at or near the tension zone, or
purlins or light-gauge decking floor systems that rest on but are not
solidly attached to the beams.
However, it is better to assume no lateral support in doubtful situa­
tions. Alternatively, it may be possible in some cases to evaluate it as
an elastic restraint. Lateral support must not be ignored: many fail­
ures in timber structures are the result of inadequate bracing against
lateral instability. The engineer should also carefully consider
the construction stage when all the restraints which may eventually
act are not yet in place.

3.2.3.2 Theory of elastic stability of beams

A beam that is bent in the plane of greatest flexural stiffness may
buckle laterally at a certain critical value of the load. As long as
the load on such a beam is below the critical value, the beam will be
stable. As the load is increased, however, a condition is reached at
which a slightly deflected and twisted form of equilibrium becomes
possible. The lowest load at which this critical condition occurs rep­
resents the critical load for the beam.
To determine the critical load for the special case of uniform bend­
ing moment along the length of the beam, consider the beam offig­
ure 3.14 with applied end moments M, about an axle parallel to
the y-axis. The beam ends are free to rotate about the y-axis, but are
Figure 3.14 Lateral torsional buckling of laterally fixed against twisting (that is the beam ends are fixed against rotation
unsupported beam subjected to pure bending moment about z-axis).
about the strong axis

66 Design of timber structures - Volume 1

 3.2 Design of slender members

Figure 3.15 Lateral torsional buckling of laterally unsupported beam

Due to the external bending moment, the beam deflects both in

the vertical direction, that is parallel to z-axis, and in the transversal
direction, that is parallel to y-axis. These two deformations are inter­
dependent; when the beam deflects laterally, the applied moment
has a component which exerts a torque about the deflected longitudi­
nal axis which causes the beam to twist.
The displacement in the z-direction is denoted w and dw/dz = fl is
the angle of twist. From equilibrium of the buckled beam, the bending
and twisting moment at any cross section are according to figure 3.15:

dw
3-15
dx

3.16

Mz'~Mait-^ = McAt-0 3.17

dz

But by the theory of elasticity:

31 3 31 • I • 318 3.18
z z dx2
and:

3L.=G-K-30 3.19
dx

where:
E is the modulus of elasticity of the beam.
Iz is the moment of inertia about the weak axis.
Kv is the torsional stiffness factor.
G is the shear modulus.

Inserting equation 3.15 into equation 3.19 and then differentiating with
respect to x gives:

G-^v-^ = Mcrit-^
dx dx
3.20

Design of timber structures - Volume 1 67

3.2 Design of slender members

Inserting equation 3.18 into equation 3.17 and then substituting in

equation 3.20:

321

or:

d20 McA2-0 _ d20 ,2 „ „

—£- +----- ent—= Q =S> —^ + 0i-j8 = O 3.22
dx2 EIZGKV dx2

where:

02 =---- ------------ 3.23

Y EIZGKV

The solution of differential equation 3.22 is:

/?(x) = Acos(0x) + B-sin(0-x) 3.24

where A and B are arbitrary constants.

The boundary conditions are that the twisting angle is zero at

the supports, that is:

J8(x = 0) = 0 => A = 0

j8(x = ^) = O => .B sin(0-^) = 0 3.25

The nontrivial solution is therefore:

sin(0-^) = O => 326 = n-n 3.26

where n is any integer. For the fundamental buckling mode, n = 1.

Substituting equation 3.23 into equation 3.26 gives:

M7 = -^E-Iz-G-Kn 3.27

The corresponding bending stress is:

ffmcr,t=
m,cnt — v 6T
Wy = b-h2-t ^48h
z GKvy 3.28

3.2.3.3 The approach according to Eurocode 5

As common for buckling phenomena, Eurocode 5 proposes the use of
a reduction factor to be applied to the strength of the beam. In
the case of an unsupported beam, such a factor, kcrit is used to keep
the bending stresses in the compression zone of the beam lower than
the bending stresses which would induce lateral torsional buckling:

®m,d — ^crit ’ Jm,d 3.29

where:
m,dA
is the design bending stress.
fm,d is the design bending strength.
k crit is a factor which takes into account the reduced
bending strength due to lateral torsional buckling.

68 Design of timber structures - Volume 1

 3.2 Design of slender members

In Eurocode 5, kcrit is given as a function of Arelm, which is defined as

the relative slenderness for bending:

3.30

The critical bending stress for a simply supported straight beam of

constant section which is restrained against buckling at the supports
and loaded with a constant moment (see figure 3.14, page 66) is:

. _My,crit_ 6-7T
'0,05 ’ h ‘ ^0,05 ' Aor 3.31
m,crit W, bh2l
where the index 0,05 indicates 5th-percentile values of modulus of
elasticity and shear modulus (note that characteristic values are nor­
mally adopted when stability checks are performed).
For a rectangular section of width b, depth h, the moment of iner­
tia about the weak axis is Iz = (h • b3/ 12), and the torsional stiffness
factor for relatively slender beams can be taken to approximately
equal Itor ~ (b3 • h/ 3). Inserting these relationships in equation 3.31
the elastic critical bending stress can be written as:

7Cb2
'0,05 ’ ^0,05 3.32
= ITT
For softwood rectangular sections, the ratio of E0 05 / G0 05 is
approximately 16, and by applying this to equation 3.32, the critical
bending stress of a beam bent about its strong axis can be written as:

0,78-fe2 r
^”m,crit ~ h ^0,05 3.33

where:
E0,05 is the 5th-percentile value of the modulus of elasticity
parallel to the grain.
b is the width of the beam.
h is the depth of the beam.
l is the design span of the simply supported beam
between lateral supports at the ends of the beam.

Equation 3.33 is only valid for a uniform moment condition applied to

a simply supported beam where the beam ends are prevented from
moving laterally, are free to rotate in plan, and are fully restrained
against torsional rotation.
For situations with different end fixing conditions and where
moment is induced by other types of loading, as well as the cases
where load is applied at the compression (or tension) face rather than
the centroidal axis of the beam, the elastic critical moment can be
obtained by using the same expression but replacing the design span
l by what is termed the “effective length”, lef, of the beam:

0,78-fr2
C'm,crit 7 g ^0,05 3.34

The effective length is obtained by adjusting the design span to take

account of the effect of the moment distribution and end fixing con­
ditions. Values for commonly used cases in timber design are given
in table 3.2, page 70. The ratio between the effective length lef and
the span l is valid for a beam with torsionally restrained supports

Design of timber structures - Volume 1 69

3.2 Design of slender members

Table 3.2 Effective length as a ratio of the span and loaded at the centroidal axis. If the load is applied at the com­
pression edge of the beam, lef should be increased by 2 • h and should
Beam type Loading type lef /l
be decreased by 0,5 • h for a load at the tension edge of the beam,
Simply Constant moment 1,0 where h is the depth of the beam.
supported
Once the slenderness ratio for bending Arelm = fmk / gmcrit) 0,5 is known,
Uniformly distributed load 0,9 the reduction factor kcrit can be determined according to table 3.3.
Concentrated force at midspan 0,8 As can be seen from figure 3.16 and table 3.3, the value Arel,m = 0,75
Cantilever Uniformly distributed load 0,5 has been adopted as the limit below which the beam will be stiff
Concentrated force at 0,8
enough to prevent lateral torsional buckling. In other words, full
the free end bending strength can be achieved without risk for instability prob­
lems if Arel,m < 0,75. This has been shown also experimentally, see
Source: Table according to EN 1995-1-1:2004, 6.3.3.
Piazza et al. (2005).
When Arelm > 1,4, that is when fmk > 2 • gmcrit the beam is considered
k
Table 3.3 Values of crit as a function of A rel
to fail solely by elastic buckling and the elastic critical bending
Value of kcrit. Relative slenderness strength will become the design condition. Finally, for relative slen­
for bending, rel m A derness values between 0,75 and 1,4 the beam will fail in bending,
1 0,75
Arel,m < but after rather pronounced lateral (and vertical) deformations. Since
1 ,56 - 0,75 A rel,m 0,75 < A < 1,4
rel,m
there will be an interaction of bending in two directions (vertical and
lateral) and also torsion, the beam will not be able to achieve the full
) )2
1 / (*• A rel,m' A > 1,4
rel,m
design bending strength. Eurocode 5 adopts the approximation of
A graphical representation of kcrit plotted against the relative a linear relationship between these limits.
slenderness ratio for bending, A is shown in figure 3.16.

Figure 3.16 The value of kcrit for different values of the relative slenderness
ratio 1 rel,m

3.2.4 Buckling of frames and arches

In frames and arches, as in columns, the buckling resistance about
both the major axis and the minor axis must be verified by stability
checks.

3.2.4.1 Lateral behaviour with special emphasis

on out-of-plane buckling
It is obvious that a frame or an arch, which lies in one vertical plane,
must be prevented from toppling over sideways, see figure 3.17 a),
page 71. Two methods could be adopted to prevent this. One is fixed-
base connections, which is rather cumbersome to achieve and which
also requires — especially in case of large structures — a massive
foundation to prevent overturning. Another, more commonly used
method for achieving lateral stability is by bracing with members that

70 Design of timber structures - Volume 1

 3.2 Design of slender members

a)

c)

Figure 3.17 Considerations concerning the lateral behaviour of arches

a) Arch turning over sideways.
b) Lateral buckling due to high normal force.
c) The lateral stability can be achieved by lateral bracing of arches with other
elements, such as purlins from the roof structure. Cross bracing or some other
mechanism is needed to assure lateral stability.

are transversal to the frame or to the arch, as shown in figure 3.17 c).
The second major problem with respect to the behaviour of frames
and arches in the lateral direction is that of lateral buckling (or out-
of-plane buckling). Since timber elements can be fairly slender,
out-of-plane buckling of the type illustrated in figure 3.17 b) may
occur.
One solution to the out-of-plane buckling is to increase the stiff­
ness of the frame or the arch in the lateral direction by increased lat­
eral dimension. When braced according to figure 3.17 c), the frame or
the arch can become fairly slender. It should be noted that the same
system used to stabilize arches or frames from overturning laterally
also provides lateral bracing and prevents lateral buckling from
occurring.
In normal portal frames and arches, out-of-plane buckling is checked
in the same way as any other beam-column between lateral restraints
and torsional restraints provided by bracings. These bracings make
the effective lengths of each element easily identifiable. In the case of
a portal frame, for example, the out-of-plane buckling should be
checked in the unbraced zones as for a column with buckling length
according to figure 3.18.

z'
h

Figure 3.18 Buckling lengths for out-of plane buckling of a portal frame

Design of timber structures - Volume 1 71

3.2 Design of slender members

3.2.4.2 In-plane buckling

Buckling in the plane of the frame or of the arch is in general more
complicated than in normal beam-column elements. The common
method of analysis for frames and arches are:
• Simplified analysis
• Second order analysis

If simplified analysis is adopted, frames and arches can be verified

in the same manner as beam-columns, that is elements subjected to
simultaneous action of bending and compression. The calculation of
stresses due to external loading is based on linear elastic theory con­
sidering the equilibrium of the undeformed static system. Stresses
caused by geometrical and structural in-plane and lateral imperfec­
tions and induced in-plane deflections are taken into account by mul­
tiplying the compression and bending strength values by buckling
reduction factors kc and kcrit. The first buckling mode of a portal
Figure 3.19 In-plane buckling of a three-hinged frame (asymmetric buckling) has the shape shown in figure 3.19.
portal frame For the determination of the buckling load according to the simpli­
fied analysis, the buckling length le is required. In order to find a rea­
sonable value for such a length, let us first study a three-hinged arch.
For common rise to span ratios — that is h/1 within the range
0,14 — 0,3 — the lowest buckling mode for arches is unsymmetrical
h and has the shape shown in figure 3.20.
According to Timoshenko et al. (1963) the critical value of the inten­
sity of the load for a parabolic arch of uniform cross section subjected
to uniformly distributed load with one, two or three hinges, can be
Figure 3.20 Unsymmetrical buckling shape of an arch expressed by the following equation:
El
335 3 33 ■ 3 3’35

The numerical factor y4 is expressed graphically as a function of h / l

in figure 3.21, where the portion of the curves indicated by dotted
lines correspond to symmetrical forms of buckling. In these cases

h/I
Figure 3.21 4 values as a function of
y h /1
72 Design of timber structures - Volume 1
 3.2 Design of slender members

Figure 3.22 Forces acting on a half three-hinged arch

unsymmetrical buckling can still occur and in obtaining values of y4,

the curves for arches without central hinge (two-hinged arch) must
be used. For example, in the case of a three-hinged arch with h/l >
0,3 the y4 value is taken from the curve for a two-hinged arch.
Let us now derive the buckling length for a parabolic three-hinged
arch with constant cross section and uniformly distributed load. We
can assume a ratio f/l ~ 0,14 — 0,15 (which is a very common ratio
for timber arches). The arch length larch of a symmetrical parabolic
arch with span l and rise f can be calculated using the following
approximate formula:

3.36

The critical value of normal force at the crown of the arch:

?crit’^2 EI
Wcrit =
8f
----- = V34------ 3.37
can be compared with the critical load for an axially loaded column
with length s = larch / 2 (= 0,53 • l):

»t El El
0111
2
AL-i, = 7T • —-----
(/Ls)2
3 = 32 ■ —j852---------
-0,532^2
3--3 3.38

where ft is buckling length factor.

Imposing Hcrit = Ncrit, h/1 = 0,15 and y4 ~ 30 (see figure 3.22):

3 = 1,17 3.39

For practical design, the effective length for buckling is often

assumed to be:

^e = l,25-.S 3.40

Analogously, in the case of a portal frame the buckling length of

the “equivalent column” can be assumed to be:

= 1,25-a 3.41

with symbols shown in figure 3.23, page 74.

If second order analysis is adopted, the calculation of stresses is
based on geometric nonlinear theory considering equilibrium of
the deformed static system. When any frame or arch is loaded, it
deflects so that the shape under load is different from the unde­
formed shape, which is the shape in the analysis if this is ordinary

Design of timber structures - Volume 1 73

3.2 Design of slender members

Real Structure Equivalent Column Model

Buckling about ”y”

h
Figure 3.23 Buckling length for in-plane buckling of a portal frame
xjz x^ x^ x^ xL xL x^ xL

small deflection (first-order) analysis. The deflection causes the axial

loads in the members to act along different lines from those assumed
in the analysis, as shown in figure 3.24. The effects, commonly called
second-order effects or P-delta effects, can reduce the resistance of
the frame.
In second order analysis, design loads are gradually applied in
increments. Before an incremental load is applied on the structure,
displacements due to previously applied loads are calculated.
Incremental loads are applied at locations worked out after adding
displacements due to previously applied loads. Thus effect of all dis­
placements, including horizontal sway, due to previously applied
loads on member forces is considered in second order analysis. This
Figure 3.24 type of analysis is commonly performed by means of adequate finite
a) Asymmetric (or sway mode) deflection. element based computer programmes.
b) Symmetric deflection.

a)

b)

0,0025^
Figure 3.25 Examples of assumed
initial deviations in the geometry for
a) frames and arches, when
b) symmetrical load and
c) non-symmetrical load. c)

74 Design of timber structures - Volume 1

 3.3 Special timber elements

If geometrical imperfections are also modelled, member forces given

by second order analysis can be directly used for member design,
without reduction of the design strength due to buckling. Initial
deformation should be assumed to be affine with the deformation
figure under load and can be assumed to correspond to an initial
deformation which is found by applying an angle $ of inclination to
the structure or relevant parts, together with an initial sinusoidal
curvature between the nodes of the structure. Some examples are
shown in figure 3.25.

3.3 Special timber elements

By special timber elements is here meant timber components with a) =
a varying cross-sectional depth or curved members. Typical examples
of members with varying cross-sectional depth are pitched beams,
continuous beams with a deeper section over the intermediate sup­
ports, or frames. Typical examples of curved elements are pitched
cambered beams, curved beams and arches. Examples are shown in
figure 3.26. Special timber elements are normally made of glued lami­
nated timber, see section 2.7.1.1, page 48.
A peculiarity of special timber elements is that the distribution of
bending stresses is non-linear. In order to understand this non-linear Figure 3.26 a) Double tapered beam, b) curved beam,
behaviour, let us consider a curved beam element subjected to posi­ c) pitched cambered beam.

tive bending moment. The total displacements of the fibres in

the curved beam are proportional to the distances of the fibres from
the neutral surface. However, the strains of the fibres are not propor­
tional to these distances because the fibres are not of equal length.
Within the elastic limit the stress in any fibre of the beam is propor­
tional to the strain of the fibre and hence the elastic stresses in
the fibres of a curved beam are not proportional to the distances of
the fibres from the neutral surface. Hence, the neutral axis in
a curved beam does not pass through the centroid of the section.
The distribution of stress over the section and the relative position of
the neutral axis are shown in figure 3.27.
If the beam were straight, the stress would be zero at the centroidal
axis and would vary proportionally with the distance from the cen­
troidal axis as indicated by the dotted line in the figure.
The formulas which will be shown in the next paragraphs were
derived assuming the material to be isotropic. In reality, timber is
an anisotropic material. However, it has been shown experimentally
that the formulas derived for isotropic materials are also valid for
anisotropic materials such as timber, see Stalnaker et al. (1989).

Design of timber structures - Volume 1 75

3.3 Special timber elements

Figure 3.28 Bending stresses (am), shear stresses (t) and perpendicular to the grain tensile stresses
(at90) in double tapered beam (left) and single tapered beam (right)

3.3.1 Tapered beams

Structural timber elements, especially glulam components, are often
designed with a varying cross-sectional depth, for example symmetri­
cal double tapered beams, continuous beams with a deeper section
over the intermediate supports, or frames. As a rule, variations in
the cross-section are achieved by tapering the laminations along one
edge.
The distribution of bending stresses in tapered beams is non-linear.
Moreover, for such beams the highest shear stress is in general not at
the neutral axis. The location of maximum shear stress is, in fact,
closer to the tapered edge, see figure 3.28. Only at the support of a sim­
ply supported beam or at the free end of a cantilever beam the shear
stress is highest at the neutral axis. Note that the shear and bending
stresses shown in figure 3.28 are in the direction parallel to lamina­
tions, rather than parallel to the tapered side.
For small slopes of the tapered edge, say a < 7°, which covers
the major part of practical cases, the effect of tapering on the bending
stress is small. For design purposes, therefore, the maximum bending
stress can be calculated as for a beam with constant cross sectional
depth (that is am = M/W), both at the tapered edge and at the straight
edge. However, the bending strength fmd is reduced by a reduction
factor at the tapered edge to take into account that shear stress and
stress perpendicular to the grain are acting simultaneously with
bending stress at the tapered edge, see figure 3.29.
Knowing the stress component ama parallel to the tapered edge (fig­
ure 3.29, left), the stress acting on the planes parallel to the grain and

Figure 3.29 Stresses at the tapered edge of a beam. Left: bending stress parallel to the tapered edge
(principal stress). Right: bending stress a0 parallel to the grain.

76 Design of timber structures - Volume 1

 3.3 Special timber elements

Figure 3.30 Left: compression perpendicular to the grain occurs at the tapered edge.
Right: tension perpendicular to the grain occurs at the tapered edge. Nowadays the last mentioned
beam shape is not advised. It is from several aspects better and more common to use so called
fish beam, with curved laminations at the bottom surface and horizontal sawn at the top.

perpendicular to the grain respectively (figure 3.29, right) can be cal­

culated from the equation of statics:
2
(T0=Crma-COS a 3-42

. 2
(T90 =CTma-Sin (Z 3.43

3 3 3f44a •sma-cosa 3.44

Equations 3.42 - 3.44 show that — close to the tapered edge — there
are both stresses perpendicular to the grain (o90) and shear stresses
parallel to the grain (t). The magnitude of such stresses increases with
increasing slope of the tapered edge. If the tapered edge is located at
the compression side of the beam, compression perpendicular to grain
occurs. On the other hand, if the tapered edge is located at the tension
side of the beam, tension perpendicular to grain occurs. For this reason,
tapered edges with an angle exceeding 10° on the compression side or
2 — 3° on the tension side should be avoided (it should be remembered
that tension perpendicular to grain can cause brittle failure at very
low level of stress and should therefore be kept as low as possible).
Since both the moment and the depth vary along the axis of
the beam, maximum bending stress as a rule occurs not where
the moment is greatest but at a section nearer the supports, see fig­
ure 3.31. The position of this section can be determined analytically
from the condition:

rf(T(x) _ d
3.45
dx dx W(x),

For simply supported single tapered beams or symmetrical double

tapered beams with uniformly distributed loads, the critical section is
at the distance:

x----- 3.46

from the support, where h0 is the depth of the beam at the support
and hap is the maximum beam depth.

3.3.2 Curved beams and

pitched cambered beams
Amongst the most prominent advantages of glulam is the possibility
of designing curved structural members. During manufacture,
the individual laminations are bent to the desired form before
the glue has hardened. To avoid damaging of the laminations as they
are bent, the curvature must be limited.

Design of timber structures - Volume 1 77

3.3 Special timber elements

Figure 3.31 Left: bending moment tending to increase the curvature of the glulam member. Right: bending
moment tending to straighten the glulam member.

Thicker laminations cannot be bent as sharply as thinner ones.

Curvature must also be limited so that high residual bending stresses
are not present in the finished member. As an example, let us try to
estimate the bending stress in a lamination of a curved glulam beam
during manufacturing, when it is bent to a given radius of curvature.

Realistic values could be:

• lamination thickness: t = 33 mm
• radius of curvature rin = 8 m
• modulus of elasticity E = 13 000 MPa.

The theoretical bending stress which occurs in the lamination due to

bending is:
Et 13000-33
= 26,8 MPa 3.47
2'^ 2-8000

which is close to the bending strength of the lamination. However,

due to the creep characteristics of timber, the bending stresses (g) are
significantly relaxed during gluing, which is carried out with added
heat and moisture. These initial stresses can therefore often be
ignored in design. However, when the ratio of radius of curvature (rin)
to lamination thickness (t) is too small, the bending strength of
the beam should be reduced. According to Eurocode 5, such a reduc­
tion factor must be applied when rin /t < 240.
When bending moment is applied to a beam that is initially curved
in the plane of bending, radial stresses (as well as bending stresses)
occur. These radial stresses may be either tensile or compressive, see
figure 3.31.
When the applied moment tends to increase the curvature of
T90
dS the glulam member, the laminations are pressed more firmly
together, see to the left in figure 3.31. This means that compressive
radial stresses occur between laminations. On the other hand, when
the applied bending moment tends to straighten the glulam member,
the laminations try to move apart, see to the right in figure 3.31. This
means that tensile radial stresses occur between laminations. Tensile
radial stresses should be kept as low as possible since they may cause
cracking of the element.
This effect can easily be demonstrated by the reader by holding
together firmly several sheets of paper in an in initially curved shape.
Bending the bundle one way separates the pages while bending in
Figure 3.32 Simplified model for determining stresses
the other way compresses them together.
perpendicular to grain at the apex of a curved Figure 3.32 shows the apex zone of a curved beam subjected to con­
beam subjected to pure bending moment stant bending moment. Assuming, for simplification, a linear stress

78 Design of timber structures - Volume 1

 3.3 Special timber elements

distribution, it can easily been shown that the resulting tensile and
compressive forces, T and C respectively, give rise to a force T90 in
the radial direction:

d-&
3.48

Resultant forces T and C are equal to each other. For a beam cross sec­
tion with breadth b and depth h, the resultants can be easily derived:

, h
C = -T = — Om'b'2 3.49
2

By geometric considerations:

d^ = rd'& 3.50

Substituting equation 3.49 and equation 3.50 into equation 3.48:

, h dt
<Tm b - 3.51
2 m 2 r
Knowing the tension force perpendicular to the grain, the corre­
sponding tensile stress is:

T™ h i
a^=T^r~A-- =
bdl 4-r p
3.52

Equation 3.52 shows that the tension stress perpendicular to the grain
ot90 at the apex of a curved beam can be calculated approximately by
modifying the bending stress parallel to the grain (om = M/ W) with
a shape factor kp = h/ (4 • r). It is important to observe that increasing
the beam depth h and/or decreasing the radius of curvature r will
increase the magnitude of the stress perpendicular to the grain.
Various studies have shown that tension strength perpendicular to
the grain ft,90 is highly dependent on the stressed volume of the tim­
ber. The basic design value of tensile strength perpendicular to
the grain must therefore be modified, for example by multiplying it
by a modification factor kvol and kdis:

3.53

where:
V0 reference volume. In EC 5, V0 = 0,01 m3.
V stressed/curved volume determined with regard to
the geometry of the member.
kdis modification factor with regard to stress distribution in
the beam.

Values of kdis and V for beams loaded by uniformly distributed load

can be taken from table 3.4, page 80. V needs, however, not be taken
as more than 2/ 3 Vwhere Vb is the total volume of the beam.
In situations where the design tensile strength for stresses perpen­ Figure 3.33 Reinforcing of the apex zones of tapered,
dicular to the grain is exceeded, mechanical fastenings such as pitched cambered and curved beams by means of
glued-in rods or full-threaded screws may be used as reinforcement, mechanical fasteners
see figure 3.33.

Design of timber structures - Volume 1 79

3.3 Special timber elements

Table 3.4 Values of kdis and V according to Eurocode 5 for typical beam types
Beam type kdis V
Double tapered beam 1,4 Volume of the stressed part (1)

(1)

0,5hap 0,5hap

Curved beam 1,4 Volume of the curved part (1)

Pitched cambered beam 1,7 olume of the curved part (1)

h
ap

a /(1)
ap

rin
r = rin + 0,5 hap

Angles a and p in degrees, b relates to the width of the structural member.

3.3.3 Design procedures

3.3.3.1 Tapered beams
For slopes a < 10°, the design bending stresses amad and amiid
(see figure 3.34) may be taken as:
6Md
^"m,cc,d ^”m,0,d 3.54
Wxxo btf

at x = x0 at x = t/2

Figure 3.34 Single tapered beam showing critical cross sections for bending stresses

80 Design of timber structures - Volume 1

 3.3 Special timber elements

At the outermost fibre of the tapered edge, the stresses should satisfy
the following expression:

^m.a.d “ °m,O,d - ^ra,a ' fm,d 3.55

where:
°m,a,d OCh °m,0,d are the design bending stresses at an angle to grain
and at the straight edge, respectively.
Md is the design bending moment in the section x = x0.
x = x0 is the position of maximum bending stress
(x 0 = (h0 • l)/ (2 • h ) for simply supported beams
with uniformly distributed load).
Wx0 is the section modulus at the cross section x = x0.
fm,d is the design bending strength.
k m,a is a reduction factor that takes into account
the simultaneous action of bending stress,
shear stress and compression/tension stress at
the tapered edge.

The values of kmma for different slopes of the tapered edge are shown
in figure 3.35. The values are derived for the glulam class GL30c.

The bending stress at the apex must also be checked, even though it
seldom governs the design:

. . ^aP»d _ l. ^^aP.d
^m,d *D ~ t\'6 n 3.56
W
'r ap bh 2
^^ap
where:
k is a factor determined by finite element analysis
that takes into account the tapering of
laminations. Values of k for glulam GL30c
are given in figure 3.37, page 83.
Map,d is the design moment at the apex.
Wap is the section modulus of the beam at the apex.

k
Figure 3.35 Values of ma according to Eurocode 5 for different slopes of the tapered edge, glulam class GL30c

Design of timber structures - Volume 1 81

3.3 Special timber elements

The design tensile stress perpendicular to the grain due to bending

moment can be calculated as follows:

■t,90,d -k
“ Kp w " Kp
. 6^ap,d 3.57
FKap ^ap

where kp is a factor determined by finite element analysis, defined as

the ratio between perpendicular to grain stress and bending stress at
the apex. Values of kp for glulam GL30c are given in figure 3.38, page 83.
The design tensile strength perpendicular to the grain must then
be reduced to take into account the volume effect. According to
Eurocode 5, the following inequality must be satisfied:

^t.gO.d - ^dis ’ /t,9O,d 3.58

where:
kdis see table 3.4.
V see table 3.4.
ft,90,d is the tensile strength perpendicular to the grain.

3.3.3.2 Curved beams and pitched cambered beams

The most critical section of curved beams and pitched cambered
beams is normally the one at the apex, see figure 3.36.
The bending stress at the apex can be calculated as follows:

_ k ^ap.d _ > 6^ap,d

®m,d - tX’O ' - /V/> n 3.59
W
”ap bh 2
£,'»ap

where k is a factor determined by finite element analysis that takes

into account the geometry of the beam. Values of k for glulam GL30c
are given in figure 3.37, page 83.
The bending strength of curved laminations should be reduced for
taking into account the eigenstresses that occur when the lamina­
tions are bent during the manufacturing of the structural element.
This can be done by multiplying the basic value of bending strength
f m,dby a reduction factor kr :

^m,d = 3.'f^d 360

The value of kr decreases with decreasing ratio rin / t (see table 3.5,
page 83).

m,a

hap

z b z
7T zr

Figure 3.36 Bending stresses and tension stresses perpendicular to the grain for: curved beam (left) and pitched cambered beam (right)
82 Design of timber structures - Volume 1
 3.3 Special timber elements

Figure 3.37 Factor k according to Eurocode 5 for different radii of curvature, glulam class GL30c

The design tensile stress perpendicular to the grain due to bending

moment can be calculated as follows:

_ j ^ap,d z 6Afap,d
<\90,d"V w * "
p ^2 3.61
ylap ‘'"ap

where kp is a factor determined by finite element analysis, defined as Table 3.5 Reduction factor for bending strength r k
the ratio between perpendicular to grain stress and bending stress at r /t,
according to Eurocode 5 as a function of the ratio .n

the apex. Values of kp for glulam GL30c are given in figure 3.38. r t
where in = inner radius of curvature and = thickness of
the lamination
The design tensile strength perpendicular to the grain shall be
reduced then in the same manner as for tapered beams, see equa­ rin /t kr
tion 3.58, page 82. > 240 1

< 240 0,76 + 0,001- rin/t

k
p

Figure 3.38 Factor kp according to Eurocode 5 for different radii of curvature, glulam class GL30c
Design of timber structures - Volume 1 83
3.4 Portal frames

3.4 Portal frames

Frame structures of timber are normally executed in glulam.
The haunch can be made curved with continuous laminates, finger
jointed, jointed with steel dowels and slotted-in plates, or built-up,
see figure 3.39.
a)
The form of the frame should follow the funicular of the main load,
as far as functional and aesthetic considerations permit. The slope of
the roof should not be less than 14 — 15° due to, among other reasons,
the wish to reduce the deflection of the ridge.
Three-hinged portal frames are suitable for spans up to 30 — 40 m.
If spans are greater, the two halves of the frame will be too large to
transport in one piece. The two-hinged portal frame provides a stiffer
structure but generally means that the frame must be manufactured
and transported in three or more parts which are jointed with rigid
joints on the site. Joints can suitably be placed at positions in the struc­
ture with small moments. Rigid joints typically demand more compli­
cated workmanship than hinges and therefore command a higher
price. They are in addition often highly visible in an undesirable way.
Figure 3.39 Examples of three-pin portal frames The parts of the frame are on the other hand smaller than those in
a) Frame with curved haunches a corresponding three-hinged frame and therefore easier to transport.
b) frame with finger jointed haunches
c) built-up frame (knee braced frame).
Timber frames with one or no hinges are not usually employed for
load-bearing structures. The three-pin portal frame is the commonest
type. It is stable against horizontal forces in its own plane and stati­
cally determinate, which means that the moment distribution is not
affected by uneven foundation settlement, moisture movements or
by unforeseen deformations in joints and connections. Further,
the three-hinged frame is hinged into the foundations, which simpli­
fies their basic construction. In poor ground conditions the horizon­
tal reactions at the supports can be taken up by tension members
between the foundations (within or under the floor construction).
The load on the substrate is then principally vertical. In normal cases,
with roof slopes around 15°, the loading consists of self-weight and
snow plus possible concentrated loads which are critical. In case of
steep roof slopes, for example in churches or certain types of storage
buildings, loading combinations together with wind can be critical.
A preliminary, rough design can be carried out according to figure 3.40.
The haunches of portal frames can be either curved or finger
jointed. Curved haunches are usually made with a constant cross-sec­
tion. Radial tensile stresses (that is tensile stress perpendicular to
the grain) must be checked if load combinations give positive moments

hr = h/15 + Z/30

hf ~ 0,7 • hr

hn ~ 0,3 • hr (> 250 mm)

b ~ 0,15 • hr till 0,20 • hr

r> 8 m

Figure 3.40 Preliminary sizing of a glulam portal frame

84 Design of timber structures - Volume 1

 3.4 Portal frames

Figure 3.41 Finger jointed haunch with jointing piece

(inner edge in tension) at the haunch — usually in combinations

with wind load. Forces are calculated and checked in accordance
with the instructions in section 3.3.3.2, page 82. This means that
a glued glulam haunch can be treated like a curved beam.
Finger jointed haunches are usually designed with a jointing piece,
see figure 3.41. The angle between the force and the grain at the joints
should be limited, which is favourable for the loading capacity of
the haunch. Tension stresses at the tapered side of members should
be avoided. Therefore, for negative bending moments at the haunch
— which is normally the most common case in practice — the tapered
laminations should be placed at the underside of the frame, that is at
the compression side (as shown in figure 3.41). On the other hand, in
situations with predominance of positive bending moments at
the haunch, for example for portal frames with high roof slope and
high wind loads, it could be advantageous to place the tapered side at
the upper side of the frame.
The joints can be checked according to the following empirical
method:
• Bending moment and normal force perpendicular to the joints is
calculated at the centre of each joint;
• Effective cross-sectional area and effective section modulus at
the joints are calculated.

With symbols as in the figure 3.41, the following expressions apply:

bh
3.62
joint" cos/?

w.. = bh2
3.63
Jomt 6-cos2/?

where:
b is the width of the frame cross section.
t is the width of the tips in the finger joints.
f is the spacing of fingers centre to centre at base.

For most geometries of finger joints, it can be assumed that

(1 - t/f « 0,8.
The design value of the compressive strength f is determined
from the angle ft between longitudinal forces and the grain.

Design of timber structures - Volume 1 85

3.5 Arches

3.5 Arches
Timber arches are generally made of glulam, mainly due to the fact
that glulam can be produced in curved forms and with varying depth
without a great increase in price. As a rule, solid sections of constant
depth are used, but composite sections of I or box form also occur (see
Chapter 5, page 115), especially for large spans.
The form of the arch should be chosen so that the moments are as
small as possible. As a rule, this means that the arch follows
the thrust line (equilibrium polygon) of the dominating loading com­
bination. The influence of moments can however not be avoided
completely, as several load combinations must be taken into account,
each with its own thrust line. As a compromise a parabola is often
chosen. For functional reasons, for example to increase the headroom
near the supports, an elliptical or other arch form may sometimes be
preferable. The dividing line between frames and arches is fluid here.
The same result can be achieved by placing the arch on columns, see
figure 3.42, left. The horizontal support reactions caused by the arch
must in this case be taken care of by a tie rod between the springing
points of the arch. When the arch rests directly on the ground floor
slab, for example as in figure 3.43 right, the horizontal forces can be
taken up by the foundations if ground conditions permit, or by tie
rods under the floor or cast into it. To limit the size of the horizontal
reactions the rise of the arch should be equal or greater than 0,14 of
its span. For a parabola this corresponds to an angle at the base a of
30°.
For more parts, which are joined rigidly on the site, hinges and
rigid joints should be placed as in figure 3.43. The choice between
two- and three-hinged arches is made after similar considerations to
those for frames. Three-hinged arches are thus preferable over spans
of up to 60 — 70 m, while larger spans usually demand that the arch
is manufactured and transported in three parts.
f

Figure 3.42 Left: Arch with tie rod, on columns. Right: Arch springing from foundations.

Figure 3.43 Suitable placing of joints in arch structures. Left: Hinge, Right: Rigid joints.

86 Design of timber structures - Volume 1

 Design of timber joints

Design of timber joints

Helena Lidelow

Timber elements must be connected to each other to function as 4.1 Dowel-types 88

a system. Joints affect the structural behaviour of the elements for
4.2 Shear capacity of single dowels 89
example by providing moment stiff connections or hinges. Joints
4.2.1 Dowel action 89
between timber elements are also a strong contributor to the overall 4.2.2 Material parameters 89
economy in a building system, mostly due to the time spent in pro­
ducing joints. Joints in timber structures can be classified into one of 4.3 Johansen theory - timber-timber joints
the following groups: and panel-timber joints 91
4.3.1 Failure mode I 91
4.3.2 Failure mode II 93
• Traditional timber joints: were used in traditional timber con­
4.3.3 Failure mode III 94
struction when steel dowels or other steel details were unavailable. 4.3.4 Design situation 94
Dowels of stronger wood species such as oak could be used to sta­ 4.3.5 Double shear timber-to-timber joints 95
bilise the connection, which often was realised through cutting
out parts of the timber members and fitting them together as a jig­ 4.4 Steel-to-timber joints 95
4.4.1 Slotted-in steel plates 96
saw puzzle. Traditional joints are produced by hand and are
4.4.2 Double shear steel-to-timber joints 97
time-consuming to realise, therefore expensive. Furthermore,
the load transfer is limited, especially tension forces, which nar­ 4.5 Expressions for the resistance of
rows the scope to shortspan buildings and furniture. Restoration of a single dowel 97
older buildings is however an important area where knowledge of
4.6 Tensile capacity of single dowels
traditional joining techniques is valuable and there are numerous
- rope effect 100
large historical timber buildings preserved. 4.6.1 Eurocode 5 application 100
4.6.2 Determination of tensile capacity
• Dowelled joints: transfer forces through shear in mechanical fas­ of single fasteners 101
teners mounted at an angle to the force direction. This is the most
common fastener type for timber elements both in Sweden and in 4.7 Combined loading 103

the world. Dowel-type connections can be designed to be ductile, 4.8 Joints depending solely on
which ensures a safe structure. Dowel-type connections involve: tensile capacity 103
• Nails
• Screws 4.9 Brittle failure modes and group effects
• Dowels in dowelled joints 104
4.9.1 Group effect 104
• Nail plates (in combination with anchor nails/anchor screws) 4.9.2 Eurocode 5 application 104
and punched metal plate fasteners 4.9.3 Brittle failure modes in dowelled joints 105
• Bolts.
4.10 Forces acting at an angle to the grain 107

• Glued joints: can be used to connect structural elements, but also 4.11 Punched metal plate fasteners 107
to build up elements, for example glulam or LVL. Gluing requires
a controlled environment in production, since many glues are 4.12 Glued joints 109
4.12.1 Glue characterisation 109
affected by moisture content and temperature when setting. It is
4.12.2 Glue types 110
therefore not recommended to glue connections at a building site. 4.12.3 Strength 110
Glued connections often show a brittle behaviour, even though
the capacity might be high. 4.13 Glued-in rods 111

4.14 Creating built-up column sections 112

4.15 Load distribution in joints 112

4.15.1 Elastic load distribution 112
4.15.2 Plastic load distribution 113

4.16 Stiffness of joints 114

Design of timber structures - Volume 1 87

4.1 Dowel-types

4.1 Dowel-types
a) Nails are smooth or rugged cylinders with a head, see figure 4.1.
The diameter is smaller than 8 mm. Nails can be driven directly into
wood or the hole can be pre-drilled. In Eurocode 5 the terms smooth
b)
nails and other than smooth nails are being used. A smooth nail is
a nail with a smooth shank without profiling, it can for instance be
Figure 4.1 Examples of nails round, square or grooved nails. To other than smooth nails, all other
a) Round, smooth nail nail types with a shank with some kind of profiling are counted, it can
b) Grooved, smooth nail (here a brad).
for instance be annular ring shanked, round jagged or twisted nails.
Screws are threaded cylinders with a head that usually do not
require pre-drilling, see figure 4.2. If the diameter of the screw is maxi­
a)
mum 6 mm, it is theoretically treated as a nail.
Dowels are smooth (on rare occasions rugged) cylinders without
a head, see figure 4.3 respectively EN 14592 for dowel-type connectors.
According to Eurocode 5 dowels have a diameter 6 — 30 mm. All con­
b) nectors to be used in timber structures must be CE marked by
the supplier.
Nail plates are combined with anchor nails or anchor screws and
c) have a pre-drilled pattern for the holes designed to automatically ful­
fil the code requirements on spacing, see figure 4.4 a). Punched metal
Figure 4.2 Examples of screws plate fasteners (EN 14545) is a rationalisation of the nail plate with
a) Hexagon head wood screw requires pre-drilling nails, where ‘nails’ are punched and folded out from a steel plate, fig­
and is however often replaced by for instance
ure 4.4 b). The plate thickness is maximum 2 mm to enable folding of
b) wood screw with countersunk head or
c) double threaded wood screw. the nails. They are pressed into the wood under controlled conditions,
for example when manufacturing timber trusses.
Bolts are screws that usually have a hexagon head and that require
pre-drilling, see figure 4.5. The diameter is for construction purposes
commonly at least 6 mm. Bolts often have a smooth shank and
a threaded part nearby the nut.
Figure 4.3 Example of dowel The material in dowel-type fasteners is most often steel, which can
be treated to become rustproof or hardened. Steel is characterised by
its ultimate strength, fu and yield strength, fy. A common system for
classifying bolts is to mark the fastener with a combination of num­
bers for example 8.8 is translated to fu = 800 MPa and fy = 800 x 0,8
= 640 MPa.
Hereinafter a dowel referring to a dowel-type connector, such as
nail, screw, steel dowel and bolt.

Figure 4.5 Bolt, often to be completed with washer and nut

88 Design of timber structures - Volume 1

 4.2 Shear capacity of single dowels

4.2 Shear capacity of

single dowels
4.2.1 Dowel action
The fastener is mounted in an angle to the force direction, most often
perpendicular to the same. When loaded, the dowel will press
against the surrounding timber members, creating embedding pres­
sure against the dowel, see figure 4.6.
The dowel will act as a beam with a distributed load from
the embedding pressure. If the dowel is stocky it will not bend, but if
it is more slender it will deform in bending ultimately creating one
or more plastic hinges in the dowel. Once deformed, the shear action
can be complemented with a tensile action in the dowel. The tensile
action can be increased by using bolts with a head and/or a nut to
achieve anchorage, by mounting screws in an angle to the force
direction to carry loads on the threads or by using screws or other
dowels with a rough surface to increase withdrawal strength.

4.2.2 Material parameters

The load carrying capacity of a dowel-type connection in shear is
determined by three parameters; the embedding strength of the tim­
ber fh, the dowel strength represented by the yield moment My and
the anchorage capacity enabling tensile action in the dowel Fax.

4.2.2.1 Embedding strength

The embedding strength is the pressure the wood around the dowel
can sustain. It is determined through a test made according to
EN 383, see figure 4.7 a), where a stocky dowel is pressed against
a wooden hole. The embedding strength is determined as the maxi­
mum force applied divided by the projected area of the dowel,
A = d x t and is measured in MPa. A typical test result is shown in Figure 4.6 Stocky and slender dowel
figure 4.7 b).

Figure 4.7 Embedding strength. a) Test set-up, b) typical test result.

Design of timber structures - Volume 1 89

4.2 Shear capacity of single dowels

The embedding strength itself is affected by several parameters:

• The density of timber: a higher density gives a higher embedding
strength.
• The fastener diameter d: small diameters yields higher embed­
ding strength than larger diameters. The hole can be allowed to
exceed the fastener diameter by 2 mm without any effect on
the embedding strength, Blass (2003).
a
• The angle between grain and load direction: the highest
embedding strength is obtained in compression parallel to
the grain and the lowest perpendicular to the grain.
• The friction between dowel and timber: dowels with a rough sur­
face obtain higher embedding strength than those with a smooth
surface.
• The moisture content in wood: as for all other strength properties,
the moisture content affects the embedding strength negatively if
it is high and positively if it is low.
• Any reinforcement of timber in tension perpendicular to
the grain: the failure in embedding is initiated by a crack along
the grain caused by the dowel expanding the hole when pressing
against the wood. It is thus the strength in tension perpendicular
to the grain that limits the embedding strength. Any reinforce­
ment in tension perpendicular to the grain will therefore increase
the embedding strength.
• If the hole is pre-drilled or not: if the hole is pre-drilled, most of
the load parallel to the grain will be carried through compression
parallel to the grain, while holes that are not pre-drilled will give
Figure 4.8 Load situation in a pre-drilled and not pre­ a load situation where the embedding strength is a mixture
drilled hole. The fibres are cut off upon predrilling while they between compression parallel and perpendicular to the grain, see
are bent when the hole is not pre-drilled. figure 4.8.

Empirical expressions for the determination of characteristic embed­

ding strength for softwood capture characteristic density and the fas­
tener diameter:

/hOk = 0,082pkJ”°’3 4.1

40>k=0,082(1-0,01d)pk 4.2

where fh,0,k is the characteristic embedding strength under loading

parallel to the grain. Note that the density pk should be inserted in
kg/m3 and the diameter d in mm in the equations 4.1 - 4.2.
For nails the embedding strength does not vary to any larger extent
with the angle between load and grain direction. For bolts the effect
is larger and a reduction is made using Hankinson’s formula if
loading takes place at an angle a to the grain:

/h,a,k = 4 . 43 * ----- 2—
*90 sin ct + cos a
[N/mm2]
L J
4.3
*90 = 1,35 + 0,015^

where fhak is the embedding strength under load directed in an angle

a to the grain direction and d is the diameter of the fastener in mm.
k90 is here given for softwood. The embedding strength fh,0,k is calcu­
lated according to equation 4.2. For other materials than softwood,
such as particleboard, plywood and fibreboard, refer to Volume 2,
Chapter 10, alternatively to Eurocode 5, sections 8.3 and 8.5.

90 Design of timber structures - Volume 1

 4.3 Johansen theory - timber-timber
joints and panel-timber joints

4.2.2.2 Yield moment

The yield moment is referred to as the plastic moment in steel design.
It is the moment needed to produce a plastic hinge in a dowel and is
determined through a test for nails according to EN 408, see figure 4.9,
or by calculating the plastic moment for larger dowel diameters.
The characteristic value of the yield moment My,Rk of a dowel-type
fastener is dependent on the fastener diameter d in mm and the mate­ Bent to an angle of 45°
rial of the dowel represented by fu the ultimate steel strength in N/mm2.
Empirical expressions for the determination of yield moment are Figure 4.9 Determination of yield moment for nails
given for nails and larger diameter dowels:

MvRk = -^-180 d2’6

[Nmm] for round nails 4.4
600

MvRk = — 270d2'6 [Nmm] for square and grooved nails 4.5

y,Rk 600

MyM=0,3/urf2’6 [Nmm] for all fasteners with d > 8 mm 4.6

For all other dowel-type fasteners testing must be performed accord­

ing to EN 14592, to determine My,Rk, alternatively acquired from
the supplier.

4.3 Johansen theory

- timber-timber joints and
panel-timber joints
When a dowel joint is loaded in shear it can fail in a number of failure
modes depending on the relation between the embedding strength,
the yield moment of the dowel and the thickness of the timber mem­
bers. The possible failure modes are displayed in figure 4.10, page 92.
The thicknesses are labelled 1 for the thinner member and 2 for
the thicker member (2 for the inner and 1 for the outer member in
joints with two shear planes). If the timber members are of different
species, a relation between their embedding strengths is defined as:

= 4.7
Jh,1,k

This situation can also occur if timber members of the same species
are mounted at an angle to each other implying that the embedding
strength may be different in the two members.
The first theory to establish the load-carrying capacity for a dow-
elled joint was put forward by Johansen (1949). He identified three
possible failure modes coupled to the number of plastic hinges form­
ing in the dowel; failure mode I has no plastic hinges, failure mode II
has one plastic hinge forming in the dowel and failure mode III has
two plastic hinges, see figure 4.10, page 92.

4.3.1 Failure mode I

Referring to figure 4.10, page 92, failure modes marked with I are
mode I failures. The resistance of failure modes a (I), b (I), g (I) and h (I)
are calculated similarly to the embedding strength described in sec­
tion 4.2.2.1 and shown in figure 4.7, page 89.

Design of timber structures - Volume 1 91

 4.3 Johansen theory - timber-timber
joints and panel-timber joints

t1 t2

For failure mode I the following relation is valid, see figure 4.11:

/vJJk “ /h4X^ 4.8

To determine Fv,Rk for the second timber member, index 1 is replaced

by index 2. The dowel remains straight and the timber fails due to
the embedding pressure. This failure mode is common in bolted con­
nections, but occurs very seldom in nailed joints.
There is also a possibility that the dowel can remain straight but
rotate. This occurs only in joints with a single shear plane (failure
mode c (I)). To calculate the resistance of such a joint a moment equi­
librium is set up.

Figure 4.11 Embedding failure Figure 4.12 Mode I failure

in mode I by rotation of the dowel

92 Design of timber structures - Volume 1

 4.3 Johansen theory - timber-timber
joints and panel-timber joints

Once again starting from the embedding failure:

fvftk = /h,l,k^1 = = ^/h,l,k^2

*1 = Pb2

Equilibrium at the interface gives:

A, 2
A/lft = /hlkd\ — + a1(b1+ — - a. f b, + =w
ien j n,i»K I 2 \ 2/ 1I 2 2 -a?
</D,14C I —

( 'I ( b2
4 I °2 ~
bright = /h2* I = Pfhi 02 ~ ”2" I

Equalling and replacing b2 = b 1 /p leads to:

2 P
Expressing a1 = (t1 - b 1) / 2 and a2 = (12 - b2)/ 2
and substituting a second order equation remains:

The solution for b1 is entered in the expression for Fv,Rk:

F^ = fb^db1 = ^4k
V,KK ■'QJa -1 1+ 4.9

In a situation where p = 1 and the timber members have the same

thicknesses, the equation is simplified to:

^=/h^(>/2-l) 4.10

In conclusion, rotation of the dowel is the worst case for failure

mode I when the timber members have the same thickness.

4.3.2 Failure mode II

If the thickness of one of the timber members is large enough, the dowel
will bend inside the timber, forming a plastic hinge, see figure 4.13.
Failure mode II refers to modes d, e, and j in figure 4.10, page 92.
Recognizing that the moment has its maximum in the plastic hinge,
the shear force must equal zero at the plastic hinge:

^v,Rk = /h,l,k^l = /h,2,k^2 = /^/h,ljc^2

Z>i = pb2

Moment equilibrium at the hinge:

( h 1 A A O '
^v.Rk 2 + f■'DJ.JC
J L1,Z,K —
y»KK “~fh2kd bltd(b, 1/1 bZ2 +——— 1 ‘’11,141
\1 +a.) 11 ^11+^2Z+ ~^ry
t1 t2

Substituting fh2k = pfh1k, b2 = b 1 /p and a1 = (11— b 1) / 2 leads to a second

order equation for b1. Inserting the solution into Fv,Rk in the same way Figure 4.13 Failure mode II
as for failure mode I gives:

4.11
v>Rk 2+p /h,l,k“^1

Design of timber structures - Volume 1 93

 4.3 Johansen theory - timber-timber
joints and panel-timber joints

If P = 1 (fh,1,k = fh,2,k = fu<), the expression is simplified to:

P /h,k^ 3MvRk
21 +----- ^-1
"v,Rk g

If instead the timber member on the left-hand side is thicker,

the expression for the resistance will end up in:

l2J82(l + j8) + ^1 + 2^y’Rk

^v,Rk 4.12
1 + 2/? f^dt2

F 3M RL
_ fhX^2 2/1 +----- ^-1
V fhXdt2

When P = 1, it is seen that equations 4.11, page 93, and 4.12 describe
the same resistance.

4.3.3 Failure mode III

The most favourable failure mode in terms of ductility is failure
mode III, where the dowel is bent in both timber members (mode f
and k). To derive the equation for the resistance a moment equilib­
rium is set up for the inclined part of the dowel, given that the shear
force in the dowel is zero in both of the plastic hinges:

^yjlk + ^yjik = /hAJc^ll ^2 + ( b i u

i2=^L^z,1= BZ

P NWv+0

- fh^'d-bi -yjZMyjikfh^d 4.13

t1 t2
Using P = 1, the equation is reduced to:

Figure 4.14 Failure mode III Fyflk ~ yZMy-gkfh^d

4.3.4 Design situation

All resistances for all failure modes must be established, whereafter
the lowest value for the particular geometry is taken as the resist­
ance. As an example, the resistances for a nailed joint with grooved
nail with a diameter of 4 mm and a timber member thickness on
the headside of 7d = 28 mm is presented in figure 4.15, page 95, as
a function of the thickness of the second member.
As the thickness of the timber member on the point-side increases,
the failure mode changes from mode b (I) to c (I) and finally mode
e (II). To reach mode f(III), the timber member on the headside addi­
tionally needs to be made thicker. A full presentation of resistances
for failure modes in single shear is assembled in table 4.1. This set-up
is fully consistent with Eurocode 5. Unlike the theoretical derivation

94 Design of timber structures - Volume 1

 4.4 Steel-to-timber joints

Resistance single shear wood-wood connection with grooved nails, d = 4 mm andt1 =28 mm

4 -------- Mode a (I)

-------- Mode b (I)
3,5
-------- Mode c (I)
-------- Mode d (II)
3
-------- Mode e (II)
-------- Mode f (III)
2,5
-------- Lowest

1,5

0,5

0
0 10 20 30 40 50 60
Thickness t2 [mm]

Figure 4.15 Resistances in different failure modes for timber-to-timber single shear joint

descibed above and which is based on Blass (2003), the formulas in

table 4.1, page 97, in some cases have been adjusted for correspond­
ence with test results.

4.3.5 Double shear timber-to-timber joints

In double shear joints (modes g — k in figure 4.10, page 92) rotation of
the dowel is restrained due to symmetry in the loading situation.
Otherwise, the same approach to determine the resistance can be
used as for single shear joints and failure modes I, II and III can all
occur. The formulas for the resistances are presented in table 4.2,
page 98. This set-up is fully consistent with Eurocode 5. Unlike
the theoretical derivation descibed above and which is based on Blass
(2003), the formulas in table 4.2, page 98, in some cases have been
adjusted for correspondence with test results.

4.4 Steel-to-timber joints

Using a steel plate as one of the members in a timber joint is very
common. Theoretically, this has the consequence that the forming of
the plastic hinge is always located at the timber-steel interface. This
increases the capacity of the joint compared to timber-to-timber con­
nections. There is a condition on the thickness of the steel plate tsteel
to enable the forming of a plastic hinge:

tsteel > d ^ fixed support

tsteel < 0,5d ^ pinned support

If the thickness of the steel plate, tsteel is larger than or equal to

the diameter of the dowel, it is stiff enough to provide a fixed sup­
port for the forming of a plastic hinge. If the thickness is less than or
equal to half the diameter of the dowel, the support from the steel

Design of timber structures - Volume 1 95

4.4 Steel-to-timber joints

Failure modes and resistances, annular ring shanked nails 60 x 4

Mode I
Mode II (d)
Mode III

Fvk,Rk

Figure 4.16 Resistances and failure modes for single shear steel-to-timber joint with fixed support of the nail from the steel plate

plate can be regarded as pinned and the dowel will merely rotate in
the hole without the formation of a plastic hinge. The resistance of
the joint is higher if a plastic hinge forms. Interpolation between
resistance formulas is permitted if needed. A full presentation of for­
mulas for the resistance is found in tables 4.3 - 4.4, page 98 -99.
An example of failure modes and resistances is given in figure 4.16.

4.4.1 Slotted-in steel plates

Steel plates are very efficient in carrying loads in joints, but they are
not always a good solution when there is a risk for fire since the steel
plates early on in the fire becomes very hot and loose their strength.
Steel plates in structures subject to fire must therefore be protected
e.g. by fire proof paint or coverage with wood. A solution is therefore
to hide the steel plates inside the timber member from the begin­
ning, which is often also preferred from the point of view of appear­
ance. This is done by making slots into the timber member and
inserting the steel plates. Holes for dowels are pre-drilled both in
wood and steel and the dowels areinserted in the holes to complete
the connection, see figure 4.17.
The resistance of a joint with slotted-in steel plates is derived in
the same manner as described for timber-to-timber joints. The only
difference is the automatic location of the plastic hinges, which will
be placed at the steel-timber interface. Furthermore, the condition
for the thickness of the steel plate need not be fulfilled. A plastic
hinge will form regardless of the thickness of the steel plate due to
symmetry in the loading situation. However, the steel plate must be
made thick enough to withstand the embedding pressure from
the dowel. The resistances for joints with slotted-in steel plates are
Figure 4.17 Joint with slotted-in steel plates given in table 4.5, page 99.

96 Design of timber structures - Volume 1

 4.5 Expressions for the resistance
of a single dowel

4.4.2 Double shear steel-to-timber joints

In double shear connections it is common to replace both the outer
members with steel plates, see figure 4.18. This can be done in several
ways. One possibility is to have nails that end within the timber
Figure 4.18 Double shear steel-to-timber joint
member, which makes the joint a single shear steel-to-timber joint. It
is also possible to have the dowel protrude all the way through
the timber member and both steel plates, see figure 4.18. This is then
a double shear steel-to-timber joint, for which the resistances are pre­
sented in tables 4.6-4.7, page 99-100.
The production of a double shear steel-to-timber joint is most easily
accomplished if both steel and timber can be pre-drilled in the same Figure 4.19 Special fastener designed for steel-to-
operation. Special fasteners have been developed to accomplish this, timber joints enabling drilling the hole and mounting
both for slotted-in steel plates and multiple shear steel-to-timber the fastener in one working moment
joints, see figure 4.19.

4.5 Expressions for

the resistance of a single dowel
Tables 4.1 - 4.7 express the resistance per fastener per shear plane. If
for example two shear planes are present, the calculated value must
be multiplied by 2. f> = fh2k/fh1k. The presentation can also be found
in Eurocode 5, Chapter 8.

Table 4.1 Single shear timber-to-timber joint

d
ff

^v,Rk - /h,2,k^2^

V1Rk
= 1.05
2+0 V
Al + 0) + + A- A
4/?(2AuA 4
/
d

F = i,05 Al^H202(1 + 0) + 4^1 + 2^)^_ A^gk

VlRk 1 + 20 V H V fhlkdt% H 4

Design of timber structures - Volume 1 97

 Table 4 .3 Single shear steel-to-tim ber joints, where tstee| > d (thick steel plate)

98 Design of timber structures - Volume 1

o
 4.5 Expressions for the resistance
of a single dowel

Table 4.4 Single shear steel-to-timber joints, where tteel < 0,5d (thin steel plate)

5= =

DL

4
- 0,4/^ijtZ

l
4
<k
_________ p
----1

4
* ^=14572^4^+

1
4
'
Table 4.5 Slotted-in steel plates

d
r
^v,Rk =
■4-J ~ A----------------------- ______ L

+J-

p - f tA Io j. 4^y,Rk _ 1 . ^ax,Rk
* 'v.Rk Jh,l,kh« J2 + , ,.2 1 1 .
/hjjc"'! J 4 ■4-J

+J-

d
r ■

[
* ^v,Rk-2>37A?y-Rk^i-1’k^'1 4
/
4
i

Design of timber structures - Volume 1 99

4.6 Tensile capacity of
single dowels - rope effect

4.6 Tensile capacity of

single dowels - rope effect
In all failure modes in shear where the dowel is bent (marked with *
in tables 4.1 - 4.7), some part of the load uptake also occurs in tension.
Depending on the surface and end anchorage of the dowel, the part
carried in tension can be larger or smaller. The surface of the dowel
can have a higher anchorage resistance Fax due to:
• twisted dowels
• annular rings
• threading (the dowel is then a screw or a bolt).

The anchorage of the dowel can be enhanced by:

• washers and nuts on the headside
• washers and nuts on the pointside.

The contribution of tension to the shear capacity of a single dowel

can be substantial. Kuipers and Van Der Put (1982) showed that
threading can increase the resistance of a joint by as much as 2,6
times the shear capacity calculated through tables 4.1 - 4.7, omitting
Table 4.8 Maximum contribution from rope effect in the second term in the right part of the formulas. The effect of tensile
relation to the shear capacity of a single dowel-type action can be determined either by empirical formulas or by testing.
fastener Since the empirical expressions are derived for a multitude of cases,
Fastener type Percentage
testing is suggested if a particular joint is to be used repeatedly.
Round nails 15 %

Square and grooved nails 25 %

4.6.1 Eurocode 5 application
Other nails 50 % In Eurocode 5, the rope effect is taken into account by adding
Screws 100 %
the term F/ 4 to the expression for the shear capacity of a single
dowel according to section 4.4. The contribution from the rope effect
Bolts 25 %
is limited to given percentages of the shear capacity (Tables 4.1 - 4.7)
Dowels 0%
as presented in table 4.8.

100 Design of timber structures - Volume 1

 4.6 Tensile capacity of
single dowels - rope effect

4.6.2 Determination of tensile capacity

of single fasteners

4.6.2.1 Nails
For nails, the capacity in tension is dependent on the surface rough­
ness along the nail and the anchorage capacity of the nail head. These
can be characterised through fax and fhead respectively, which can be
determined through tests according to EN 1382, EN 1383 and/or
EN 14358. If these strengths are not known, empirical expressions
exist to establish them for smooth nails (in other cases the nail is
tested according to EN 14592):

,/'xk =20-10 4.14

/head,k - 70-10 pk 4.15

Expressions for calculating the withdrawal capacity Fax,Rk for the nail,
simply based on the surface area of the nail or the head, are then
given as:
■/axi^pen
^axjlk = min- (other than smooth nails) 4.16
■/headjc^h

^ixjc^pen
T^ax.Rk = min (smooth nails) 4.17
fax.k.dt "*■ -/headjc^h

where tpen is the penetration depth and dh the diameter of the nail
head. (For annular ring shanked nails tpen is the penetration depth for
the threaded part). Conditions apply to the thickness of the members,
see Eurocode 5, 8.3.2(7). Permanent axial loads are not allowed for
smooth nails. In figure 4.20, the addition of the withdrawal capacity
to the shear capacity is shown for grooved nails.

Contribution from withdrawal capacity, grooved nails 60 x 4

Fv,Rk
Fax,Rk
Fax,Rk /4

Figure 4.20 Withdrawal and shear capacity of grooved nails. Fv,Rk is the capacity in shear
and FaxRk the capacity in tension. The total capacity is the sum FvRk + FaxRk / 4. The joint is of
the same type as in figure 4.16, page 96.

Design of timber structures - Volume 1 101

4.6 Tensile capacity of
single dowels - rope effect

4.6.2.2 Bolts
For bolts, the withdrawal capacity is dependent on the anchorage
capacity of washers and nuts and the tensile capacity of the bolt itself.
When determining the anchorage capacity of the washer, the charac­
teristic compression strength perpendicular to grain for wood can be
taken as three times the characteristic base value due to concentrated
loading:

^'ax.washer.Rk “ ^/c,90,kAvasher 4.18

If an entire steel plate with thickness tsteel is used instead of washers,

Awasher should be replaced with a circular area having the diameter:

D = min- 12i
4J
steel

where d is the bolt diameter.

4.6.2.3 Screws
The withdrawal capacity of screws is dependent on the threading and
can reach very high values. The withdrawal strength can be determi­
ned as:
4.19 4 0,52^’5^’1pk0'8 4 and /rf [mm], pk [kg/m3] 4.19

where lef is the length of the threaded part embedded in wood.

The characteristic withdrawal capacity of a connection with screws is
established as:
_ rTf^dl^
4.20
ax,a,Rk i)2cos2a + sin2a

where a is the angle between the screw axis and the grain (a > 30°)
and kd is min(d/ 8;1) with d in mm, and n is the number of screws act­
ing together in the connection. The outer thread diameter d must be
6 < d < 12 mm and 0,6 < d 1 /d < 0,75 where d1 is the inner thread
diameter.
The contribution from the withdrawal capacity for bolts (with
washer and nut) and screws is shown in figure 4.21.

Fvk,Rk
Fax,Rk (bolt)
Fax,Rk (bolt)/4
Fax,Rk (screw)
Fax,Rk (screw)/4

Timber thickness [mm]

Figure 4.21 Shear and withdrawal capacity for bolts and screws. The absolute value of the withdrawal capacity is not
equal for bolts and for screws. The total capacity of the bolt respectively the screw in shear is taken as FvRk + FaxRk / 4.

102 Design of timber structures - Volume 1

 4.7 Combined loading
4.8 Joints depending solely on tensile capacity

4.7 Combined loading

If a joint is loaded with simultaneous shear load and axial load,
the combination of these loads must be taken into account in design.
The following interaction relations can then be used:

2ax,Ed ,1--------
--------- 2v,Ed Si1 .. x
/(smooth.. nails) .4.21
^ax,Rd ^v,Rd

rax,Ed
(other types of nails and screws) 4.22
^^ax»Rd y <2v3d>

4.8 Joints depending solely on

tensile capacity
The tensile capacity of screws is so large that it can be beneficial to
mount them in such a manner that they are loaded only in tension
or compression. This can be accomplished by using the analogy with
a truss, where shear forces are carried through a force couple in ten­
sion and compression. Examples of applications are shown in
figure 4.22.
The capacity of the screws is determined only with the use of
the tensile capacity described in section 4.6, page 100.

Tensile screw

Compressive screw

Figure 4.22 Screws loaded solely in tension or compression, Blass (2003)

Design of timber structures - Volume 1 103

4.9 Brittle failure modes and
group effects in dowelled joints

4.9 Brittle failure modes and

group effects in dowelled joints
A joint is considered to exist if at least two fasteners are present in
a connection. Oftentimes, several fasteners placed in a group are
needed to carry the load in the connection. The fastener spacing
needs to be kept at certain distances to prevent splitting between
the fasteners. Rules for this are given in most codes, see Volume 2,
Chapter 10.

4.9.1 Group effect

C When a number of fasteners are loaded, they seldom reach their
load-carrying capacity at the same time. This is due to local varia­
0,25 -
tions in timber strength, hole sizes, misalignment of holes and
0,20 - an uneven load transfer between the connecting members. Cramer
(1968) and Lantos (1969) described what happens if a row of fasteners
0,15 - is loaded in tension and found that the first and last fasteners receive

b)
0,10 -

0,05 -

Ihil
1st 2nd
Fastener position
nth
the highest load level, that is they will fail first, see figure 4.23.
The group effect is largest in joints with stout dowels, behaving
nearly linear elastic, see further section 4.12, page 109, on glued joints
where the uneven stress distribution causes the same type of problem.
The total capacity of such a joint is lower than the sum of the capaci­
ties of the individual dowels. In joints with slender dowels, for exam­
ple nails, the plasticity of the fasteners enables redistribution of loads
Figure 4.23 Load distribution in bolted joint,
so that the total joint capacity is close to or equal to the sum of
Lantos (1969). a) Force distribution in side member, the individual fasteners’ capacity, Blass (1990).
b) force distribution along the row of fasteners.

4.9.2 Eurocode 5 application

For dowel-type fasteners placed in a row parallel to grain a group
effect is applied:

«ef = nkrf (screws d< 6 mm, nails and staples) 4.23

n
(screws d > 6 mm, bolts and dowels) 4.24

where a1 is the distance between fasteners parallel to the grain.

The group effect is strongly dependent on the distance a1 between
fasteners placed in a row parallel to the grain. For nails, the value of
kef = 1 if a1 > 14d,, that is there is no group effect. For nails with
the common spacing 10d, kef = 0,85 and for dense spacing, a1 = 7d,
kef = 0,7, Eurocode 5: section 8.3.1.1(8), see figure 4.24, page 105.
Reduction of the load-bearing capacity can be avoided by shifting
the nails ± d in the direction of the fibres.
For bolts, the effect is also dependent on the distance between
the fasteners and for the minimum spacing of 5d the effect is signifi­
cant, see figure 4.25, page 105. If the spacing is larger than 13d, no
group effect is expected.
The group effect shows a stronger dependency on spacing for nails
than for bolts.

104 Design of timber structures - Volume 1

 4.9 Brittle failure modes and
group effects in dowelled joints

nef, (7d)
nef, (10d)
nef, (14d)

Figure 4.24 Efficient number of nails in a row dependent on spacing

nef, bolt (5d)

nef, bolt (10d)
nef, bolt (13d)

Figure 4.25 Efficient number of bolts in a row dependent on spacing

4.9.3 Brittle failure modes

in dowelled joints
Failure according to Johansen theory occurs for single dowels placed
at appropriate spacing to prevent splitting before bending of
the dowel or embedding failure takes place. In many joints, large
spacing is not possible to use since this would require large dimen­
sions of timber members, ending up in uneconomical structures.
However, when fasteners are closely spaced the risk for splitting
increases. This can induce brittle failure in the joint before embed­
ding failure occurs. Brittle failure in joints should be avoided and
the capacity for the brittle failure modes must be checked along with

Design of timber structures - Volume 1 105

4.9 Brittle failure modes and
group effects in dowelled joints

Figure 4.26 Brittle failure modes in mechanical timber joints

the capacity of the single fastener and the group of fasteners. Brittle
failures can be of different types; row shear, block shear, plug shear
and splitting, see figure 4.26.
If the joint consists of several rows of fasteners block/plug shear or
row shear failure may occur. For nailed joints, row shear never occurs
due to the small diameter of the fasteners. Furthermore, block shear
is uncommon since nails seldom protrude through the timber mem­
ber. Nailed joints should therefore be checked only for plug shear
failure. Dowelled joints should be checked for block shear failure
according to Eurocode 5, but row shear failure is left out. Row shear
failure commonly shows a higher capacity than block shear, since
Figure 4.27 Faces along which splitting occurs in plug splitting occurs along more faces compared to block shear failure.
shear failure. In block shear failure no bottom face forms
The faces can be designated as in figure 4.27.

4.9.3.1 Block shear failure

The resistance in block shear consists of two possible contributions;
the resistance in tension of the end face and the resistance in shear
of the side faces. The tensile and shear capacities cannot be added
since they are associated with different stiffnesses. The resistance is
taken as the maximum value of the tensile and the shear capacities
of the block:

^bsjtd = max- (dowel-type connectors) 4.25

The area Anet,t is the net area at the end of the plug and the areas Anet,v
is the net area of the sides of the block. The factors 1,5 and 0,7 are
empirically derived factors to accommodate for two different phe­
nomena. In the case of the tensile capacity, the tensile strength can
be higher locally at a joint, since the probability of having defects is
lower in a specified area than in a full size member, Larsen (2003).
The shear capacity is reduced due to a volume effect that affects
the shear strength when the loaded area is large, Johnsson (2004).
The areas should be reduced with the size of the fastener holes, as
indicated by the subscript “net” in equation 4.25. The method is
described in Eurocode 5, Annex A.

4.9.3.2 Plug shear failure

Equation 4.25 can be used to establish both the block shear and
the plug shear capacity. What differs is the calculation of the area
Anet,v, which for plug shear failure can be taken as the sum of the net
bottom shear area and the net shear area along the sides of the plug.

106 Design of timber structures - Volume 1

 4.10 Forces acting at an angle to the grain
4.11 Punched metal plate fasteners

Figure 4.28 Joints loaded at an angle to the grain

4.10 Forces acting at

an angle to the grain
Wood is weak in tension perpendicular to the grain. Producing joints
such as those in figure 4.28 is not recommendable. A better solution is to
transfer the forces in compression as in figure 4.29.
If it is impossible to transfer forces in compression it is important to
place the fasteners as far from the loaded edge as possible to avoid split­
ting of the timber member. More or less sophisticated models exist to
predict the capacity of a joint loaded in an angle to the grain, some of
them based on linear fracture mechanics, Gustafsson et al. (2003). In
Eurocode 5, section 8.1.4, the phenomenon is taken into account by
satisfying:

^v,Ed - ^90,Rd 4.26

where Fv,Ed = max{Fv,Ed,1; Fv,Ed,2}, that is the largest of the two shear forces
in the beam on each side of the connection as shown in figure 4.28.
The design resistance is determined on the basis of the characteristic
resistance F90,Rk determined as:

^90,Rk - 146 (with b, he and h taken in mm and with F90,Rk in N) 4.27

1-^
h
where h, he and b are defined in figure 4.28. Figure 4.29 Load transfer in compression

4.11 Punched metal plate

fasteners
Punched metal plate fasteners are an integration of nails and nail plates.
They are produced by punching nails directly from a steel plate, see fig­
ure 4.4 b), page 88. Due to limitations in the punching process, the thick­
ness of the original steel plate cannot exceed 2 mm, which has the con­
sequence that the ‘nails’ have low capacity. The length of the ‘nails’ is
commonly 8 — 15 mm, Nielsen (2003). The most common usage of
punched metal plate fasteners is in the production of timber trusses, see
figure 4.30. In Sweden the application is prominently for the housing
industry, with common spans of 9 — 15 m of these trusses, but they can
reach towards 40 m span. Punched metal plate fasteners are installed
through pressing them into wood by means of a hydraulic press.
Therefore it is not recommendable to use punched metal plate fasteners
for joints on the building site.

Design of timber structures - Volume 1 107

4.11 Punched metal plate fasteners

Due to the production process, the punched metal plate fastener will
show an orthotropic behaviour. The strength of the fastener is depend­
ent on the angle to the load direction and for a full characterisation
a) of a punched metal plate fastener no less than 11 strength parameters
Cantilever (recessed support) need to be determined; shear, tensile, compressive and anchorage
strengths in perpendicular directions and three constants which
describe the anchorage behaviour, EN 1075 and EN 14545.
When designing a joint with punched metal plate fasteners,
the tensile, compression and shear strength must be checked along
with the risk for buckling of the plate at unsupported sections of
b)
the assembly. The procedure is described in detail in Eurocode 5, sec­
tion 8.8, but is seldom performed by hand calculations. Well developed
Punched metal software exists for the concurrent design of punched metal plate fas­
plate fastener teners and trusses, taking into account the effect of movements in
the joints, which affects the force distribution in the truss.
Punched metal plate fasteners can also be used to reinforce timber
c)
itself, but also the area around joints, Nielsen (2003). In that case, it
is not the punched metal plate fastener in itself that is interesting to
design, but rather the effect it creates by preventing failure perpen­
dicular to the grain in timber. Examples of usage are, see figure 4.31:
• increasing the compression strength perpendicular to grain by pro­
viding a secondary load path in the connection and stiffening
the wood
• increasing the stiffness in bending by reinforcing the compressive
side of the timber member
• increasing the tensile capacity perpendicular to grain in the joint
area to prevent brittle failures such as block shear or splitting and
also provide high embedding strength at the most heavily stressed
section in a dowelled joint.

The usage of punched metal plate fasteners to strengthen dowelled

joints can double the loadcarrying capacity, Blass (2003).
Figure 4.30 Trusses produced with punched metal plate
fasteners, Nielsen (2003)
a) Duopith (standard) or W with double cantilever
b) Duopitch with french heel and double cantilever
c) Scissor
d) Monopitch
e) Attic.

Reinforcement Reinforcement

c)

Figure 4.31 Strengthening of timber structures using punched metal plate

fasteners, Nielsen (2003)

108 Design of timber structures - Volume 1

 4.12 Glued joints

4.12 Glued joints

Glued joints are frequently used in timber engineering, especially
when producing elements. They are not frequently used as a joining
technique between members due to:
• a lack of precise design rules
• the need for controlled environmental conditions when gluing,
making only factory production a feasible option.

Glued timber joints can be categorised according to figure 4.32.

The production of glued joints must take place under controlled
conditions with a quality control to ensure consistent quality of
the end result.

4 .12.1 Glue characterisation

The strength of glued joints is difficult to describe analytically. If
the glue is ductile, the shear strength describes its behaviour well. If
it is brittle, the fracture energy, Gf, of the glue is the best descriptor.
To characterise the brittleness, a brittleness ratio has been defined as
fV/Gf. A full characterisation from experiments of the shear strength
fv, the tensile strength ft, the fracture energies in shear GfT and in ten­
sion Gfo are available only for a few glues, see table 4.9.

Structural adhesive joints

Pure wood-wood joints (lumber, glulam, LVL) Hybrid wood joints

End joints (knees, reinforcements) Continuous • Joints with glued in rods:

• Scarf joints (and special form - Threaded steel rods
• Lap joints work truss) joints - Rebars
• I-beams with - FRP rods
- End joints
- Reinforcements with plywood and LVL straps continuous webs • Joints with glued in
• Finger joints • Glulam steel and FRP plates
- Glulam lamellas (large) lumber cross sections
• Large finger joints:
- Straight beams
- Knee joints:
- with equal materials
- with corner pieces from plywood, LVL, etc.
• Knee and truss joints with gussets from plywood, LVL, etc.
• End joints with axially loaded hard wood dowels

Figure 4.32 Classification of different types of structural adhesive joints, Aicher (2003)

Table 4.9 Characteristics of adhesives and comparison with wood, Aicher (2003)

fv GT
f fv2 / GfT f Gf. f2 / Gf.
Glue [N/mm2] [N/mm] [N/mm3] [N/mm2] [N/mm] [N/mm3]

Resorcinol/Phenol 8,4 0,7 99 6,4 0,36 114

PVAC 7,4 2,1 26 — — —

Polyurethane 2,8 0,6 14 1,0 0,23 4

Wood (Picea Abies) 8,9 1,0 83 2,9 0,3 28

Design of timber structures - Volume 1 109

4.12 Glued joints

One of the problems is that pure tension or shear is seldom present

in a real joint, making it impossible to separate strength components
even in testing.
From table 4.9, page 109, one can conclude that the glue formerly
often used for making glulam, resorcinol/phenol is more brittle than
wood itself, while polyurethane is less brittle than wood.

4.12.2 Glue types

Resorcinol/phenol has two components and creates a reddish brown
glue line with good mechanical properties also in fire. The glue can
set at room temperature, but an elevated temperature (+ 40 °C)
reduces the curing time. The glue line is resistant to moisture
change. Resorcinol is expensive and a blend with 50 — 75 % phenol is
therefore used. Up until a few years ago, resorcinol/phenol was
the most common glue when producing glulam in Sweden. Currently
Melamine-Urea-Formaldehyde (MUF) is used for glulam production,
since this glue has a whitish glue line. It has the same structural
properties as resorcinol/phenol. The curing time is an hour in room
temperature, but only a few minutes if using a high frequency oven
(microwave technology) during setting.
Polyurethane is also made up of two components, but is weaker
than resorcinol although the demand for pressure during curing is
not so high. Polyurethane is preferable to epoxy due to the risk for
allergic reaction. Epoxy is on the other hand known to bond almost
any material and can be used for gluing on site.

4.12.3 Strength
The strength of a lap joint was first described by Volkersen (1938)
who arrived at the conclusion that the distribution of shear stress t(x)
in the joint can be described as:

p cosh(px/^) (1-a) sinh(pxZ^)

2 sinh(p/2) (1 + a) cosh(p/2)
4.28
Mi
^2^2

Refer to figure 4.33 for notations. The coordinate x increases from left
to right along the length l.

b)

Figure 4.33 Lap joint

110 Design of timber structures - Volume 1

 4.13 Glued-in rods

The most important result from the Volkersen theory is that the stress
distribution in the joint is non-uniform. The strength of the joint
cannot be determined by calculating tm, which is the average shear
stress value. There will be peak stresses at the ends of the glue line
where failure will be initiated. The same phenomenon, but less
prominent, is behind the group effect in dowels placed in a row.
A deeper study of the shear stress t(x) will reveal that above a certain
length, there is no change in peak stress or in other words, lengthen­
ing the connection will not increase its strength. Volkersen theory
can be generalised to non-linear fracture mechanics, Gustafsson
(1987). By studying the limit cases for non-linearity, fully plastic bond
line and fully brittle bond line, Gustafsson (1987) arrived at upper
and lower bounds for establishing the capacity of a joint:

4.29
Pmm ~ ^/^(l +

If a fully brittle bond line can be assumed, the capacity is not influ­
enced by the length of the bond line. However, the range for the brit­
tleness ratio where non-linear fracture mechanics is valid is depend­
ent on the bond length through:

0,1(1 + a)<co< 10 (1 + a)
^2/v2 4.30
(D =—:i2—
Z1E1GfT

For smaller a, plastic conditions apply, while for larger a, linear frac­
ture mechanics should be used. Using l = 400 mm, t = 45 mm, and
E1 = 13 000 MPa and fv and GfT for a resorcinolphenol glue according
to table 4.9, page 109, a = 27. For a = 1 (symmetrical lap joint), the limit
value for a = 20, which means that linear fracture mechanics can be
applied. Changing glue to polyurethane makes a = 3,8, which means
that non-linear fracture mechanics must be used.

4.13 Glued-in rods

A special application of a gluing is glued-in rods, figure 4.34. This is
an appealing solution since the joint is hidden inside the timber
member and the connection is strong and stiff. Theoretically, there
are no general design rules, but substantial testing have established
good knowledge in the field, Gustafsson and Serrano (2001) and
Tlustochowicz et al (2010).

a)

Figure 4.34 Failure modes for glued-in rods, Tlustochowicz et al. (2010)

Design of timber structures - Volume 1 111

4.14 Creating built-up column sections
4.15 Load distribution in joints

4.14 Creating built-up

column sections
When designing timber columns and other members of a timber
structure, it is not unusual to need other cross sections than rectan­
gular ones (crosses and boxes are common). These can be built-up
from several timber members and mechanically jointed as described
in Eurocode 5, Annex C. Another option is to order this kind of structure
directly from the glulam manufacturer, who then uses glue to fabricate
the column, which leads to full composite action in the cross section.
To obtain enough pressure during production, 1 — 2 nails/dm2 is
needed and the nails should have sufficient anchorage length and
preferably be ringed to further increase the anchorage. Alternatively
screws can be used. Sufficient pressure can also be created through
a distributed weight on the glued member.

4.15 Load distribution in joints

4.15.1 Elastic load distribution
In a joint, the individual fasteners will receive different loads
depending on the global load applied to the joint. In tension and
compression, it is a matter of calculating the load components and
applying them to the individual fastener, but for moment resisting
joints the actual geometry of the joint will produce different loads in
different fasteners depending on their position in the joint. Assume
a timber joint loaded with a vertical eccentric load with lever arm e
with respect to the centre of gravity of the joint. The elastic load dis­
tribution can then be determined as shown in figure 4.35:

1. Translate the eccentric forces to the centre of gravity of

the fastener group.

2. The vertical force on each fastener is Fyi = F/n, where n is

the number of fasteners.
3. From the moment M = F • e the load vector Fmi results. Assuming
a linear force-slip relation, Fmi = K8 = Kari, where K is the slip
modulus of the connector, 8 is the displacement due to rotation in
the joint and a is the rotation angle, see bottom figure.
4. The external moment and the sum of the internal moments
must be equal: M = F • e = Sri Fmi = Sri Kari = Ka Sri2

5. The term Sri2 is a characteristic of the joint and is called the polar
moment of inertia: Ip = Sri2 = S(xi2 + yi2)
6. The load component Fmi can now be established through:

F^-Kar,-^
2p
Figure 4.35 Load distribution in a moment resisting joint Myx Mx{
—- t
Fmxi■ =----tF ’ • = myi
•'p yp

112 Design of timber structures - Volume 1

 4.15 Load distribution in joints

7. Vector summation gives the load effect on one fastener

(including F= Fx/n in the general case when the external load F
has a component Fx in the x-direction):

= ^xi + ^nxi )2 + (^yi + ^myi

8. Fi should be compared to the capacity of one fastener.

4.15.2 Plastic load distribution

As dowelled joints show a plastic behaviour, there is a possibility to
increase the total resistance of the joint, by allowing the outermost
dowels to plasticize and thereafter redistribute their load to the dow­
els closer to the centre of rotation, see figure 4.35, page 112. Assuming
that the plastic behaviour in figure 4.7 b), page 89, can be idealized to
elasto-plastic behaviour, there is a possibility to establish a lower
bound and an upper bound for the plastic resistance of the joint.
Note that the principle of superposition does not apply when assum­
ing plastic behaviour. There is a limit on the displacement for one
dowel, since the assumption of elastoplastic behaviour is limited by
the possible displacement of the dowel. A reasonable limit could be
that the maximum allowable displacement would be umax < 4 uy
where uy is the elastic displacement. The practical consequence is
then that dowels closer to the rotational centre than 0,25rmax (rmax
being the maximum distance from the centre of rotation to the out­
ermost dowel) are not be taken into account when establishing
the plastic resistance of a joint.

4.15.2.1 Upper bound for plastic resistance

An upper bound F + is established through the principle of virtual
work, (Larsen and Riberholt, 1999). A centre of rotation is assumed
somewhere along a line perpendicular to the load. Often this line
passes nearby the elastic centre of gravity, see figure 4.35, page 112.
Assuming that all dowels have equal yield load Fy, and reusing figure
4.35, page 112, the internal and external works Wi and We are:

^i=XariFy
i=l 4.31
We = aeF+

Equaling Wi and We leads to:

n
4.32

To find the optimal value for F +, an iteration procedure is needed,

where the centre of rotation is assumed and F + calculated. F + will
have a minimum value when the exact plastic capacity is obtained.
The optimum position of the centre of rotation is usually close to
the elastic centre of gravity but on the side opposite the external load.

Design of timber structures - Volume 1 113

4.16 Stiffness of joints

4.15.2.2 Lower bound for plastic resistance

The lower bound for the plastic resistance is established through
dividing the fasteners into two groups; those that resist the moment
and those that resist the normal force. The selection is made under
the condition that equilibrium with the imposed load must be pre­
served. If the joint is loaded with a force acting through the centre of
•2
gravity, the lower and upper bound will coincide as the direction of
Fy
J. all reaction forces will be the same for all fasteners. The method is
exemplified in figure 4.36.
Fy 6
In figure 4.36, the lower bound F - for the resistance is established
through:
,3

Vertical equilibrium: F - 4 • Fy = 0
Moment equilibrium: F • e - Fy • r = 0

y where Fy is the yield load for the fasteners in this case assumed equal
Figure 4.36 Example of lower bound establishment for all of them. The lowest value of F found from these two equilib­
of plastic load distribution in joint, Larsen and Riberholt rium relations constitutes the lower bound solution found from
(1999) the assumed force distribution, that is:

F = min- 4.33

4.16 Stiffness of joints

The slip modulus Ku of connections in the ultimate limit state should
be taken as two thirds of the slip modulus in the serviceability limit
state, Eurocode 5, section 2.2.2:
2
4.3U 4 —K^roCl 4.34
The slip modulus in serviceability limit state, Kser, in N/mm can be
established through, Eurocode 5, section 7.1:

P^/23 (for dowels, bolts, screws and pre-drilled nails)

45
p^d°’g/3Q (for non pre-drilled nails)

where pm is the mean density in kg/m3 of wood and d the outer diam­
eter in mm of the fastener. In joints between wood and steel and
between wood and concrete, Kser established using the wood density
can be multiplied by 2.

114 Design of timber structures - Volume 1

 Composite timber elements

Composite timber elements

Bert Norlin

By composite timber elements here is meant I-beams, hollow section 5.1 Structural elements
beams and stressed skin panels (SSP). As examples, the first two are with full composite action 116
typical beams having cross-sections as shown in figure 5.2, page 116, 5.1.1 Glued thin webbed beams 116
while the last one is a kind of building block covering a large surface 5.1.2 Glued thin flanged elements (SSP) 133
as in figure 5.14, page 133.
5.2 Structural elements
Composite timber elements are usually, but not necessarily, made
with partial composite action 142
of more than one material. Common wood based materials used are 5.2.1 Built-up bending elements 143
structural timber, glued laminated timber, laminated veneer lumber 5.2.2 Summary of equations 156
(LVL), plywood, fibreboard, particleboard, oriented strand board (OSB), 5.2.3 Failure criteria for built-up beams 157
etc. Other materials such as steel, aluminium, concrete, plastics, and
even glass are possible to use. But here we restrict ourselves to wood
based materials.
Basically these elements carry the load through bending action,
especially in horizontal structures like floors, roofs, bridge decks, etc.
Of course axial forces must be considered for structures affected by
such forces, like in wall elements and columns.
The common property of all these composite elements is that
the section is divided into wide outer flange parts with an intermedi­
ate thin web part. Naturally, the flanges of an SSP-element are very
wide and very thin, but the web panels are still quite thin in relation
to the flanges. The same is also true for an ordinary I-beam, even
though the flanges may have a small width the web panel will have
a much smaller thickness. When these kinds of elements are subjected
to bending, most of the bending moment is carried by the flanges
while a minor part is carried by the web. The shear force on the other
hand is mainly carried by the web panel or panels. The only exception
is T-shaped sections for which no real distinction between a flange
and web part can be made.
Consider now a small cut-out from the compression flange of
an ordinary I-beam as shown in figure 5.1 a). The axial stresses change
slightly when moving from one end to the other end of the cut-out.
The only thing that can cause this change is the shear stress affecting
the web panel. All the shear stress must be transferred into the flange
in order to build up the axial stresses or vice versa. Now if the web-
to-flange connection is sensitive to shear deformation a small slip
will result between the two materials, as shown in figure 5.1 b). If this
slip is negligible we have so called “full composite action” (fca) in

a a + da Slip

a)

Figure 5.1 A small cut-out of the web-to-flange connection of an I-beam

a) No slip or negligible slip, b) non-negligible slip.

Design of timber structures - Volume 1 115

5.1 Structural elements
with full composite action

Figure 5.2 Typical cross-sections of thin webbed beams, I-sections as well as

rectangular hollow sections (box-sections)

which case the beam’s stiffness and stress magnitudes can be based
on ordinary beam bending theory as explained in section 5.1. If
the slip is non-negligible the applied load will be counteracted differ­
ently such that with increasing slip the flanges will carry less of
the applied bending moment through axial compression and tension
of the flanges and more through individual bending of each flange and
web panel. This will result in a decreasing transfer of shear stresses
over the interface. In addition to this redistribution from axial flange
forces to flange bending the stiffness will decrease leading to increased
deflection. This state of non-negligible slip is referred to as “partial
composite action” (pca) and is explained in section 5.2, page 142.
Typically, full composite action is usually assumed for glued con­
nections having very small glue line thicknesses. Partial composite
action must be assumed if the glue is replaced by mechanical fasten­
ers and/or the interface have a thickness such that its shear deforma­
tion is not very small. Actually, the shear deformation of the web
panel itself can be treated as a slip and can be accounted for through
the theory explained in section 5.2, page 142.

5.1 Structural elements

with full composite action
I-beams and hollow section beams are treated in section 5.1.1 while
stressed skin panels are dealt with in section 5.1.2, page 133. But bear
in mind that most design issues are similar for these two types of
composite elements.

5.1.1 Glued thin webbed beams

Naturum, Takern. Examples of thin webbed beams are shown in figure 5.2. Two basic
shapes can be distinguished, that is open I-sections and closed hollow
sections (box-sections). The flanges are usually made of structural
timber, glued laminated timber or LVL, while web panels are made of
hard fibreboard (hardboard), OSB, plywood or particleboard. In most
commercial products sawn timber flanges in combination with hard­
board, OSB or sometimes plywood are used.
Figure 5.3, page 117, shows a side view of two typical I-beams (or
hollow section beams), one having small web slenderness and one
having large web slenderness. The slenderness limit between these
two categories is approximately hw /bw = 35.

116 Design of timber structures - Volume 1

 5.1 Structural elements
with full composite action

Figure 5.3 I-beam or hollow section beam having: a) small web slenderness or b) large web slenderness.

Possible failure positions are indicated symbolically and given a spe­

cific number in figure 5.3. The type of failures indicated by the num­
bers are:

1. Compression failure of the compression flange, which in case of

insufficient lateral bracing may involve the influence of lateral
torsional buckling.
2. Tension failure of the tension flange.
3. Shear failure of the web panel.
4. Shear buckling failure of the web panel.
5. Shear failure of the flange-to-web interface, that is the glued con­
nection between them.
6. Shear failure of open web panel joints and web panels with
holes.
7. Buckling failure due to axial loading, not shown in figure 5.3.
8. Patch loading failure, that is concentrated loading acting perpen­
dicular to the flanges may cause failure in the neighbouring web
panel region, either by a pure compression failure of the web
panel in combination with some local flange bending (stocky
web panels) or by a local buckling failure of the web panel that is
also influenced by some flange bending (slender web panels).
9. Compression failure of the outer flange surfaces caused by con­
centrated forces acting perpendicular to the grain, which is
checked using the rules described in section 3.1.3, page 58.
10. Overall web buckling failure, which is quite rare for normal tim­
ber I-beams. Another type of buckling that may add to this buck­
ling mode is flange induced web buckling due to increasing
beam curvature.
11. Local web buckling caused by bending moments and/or axial
force. This is more or less obscure for timber I-beams.
12. In the serviceability limit state one must check deflections, pri­
marily caused by bending but sometimes significantly influenced
by shear deformations in the web. Not shown in figure 5.3.
13. The influence of vibrations and resonance frequencies in floors
and bridge decks may also be important issues in the serviceabil­
ity limit state, not shown in figure 5.3.

Design of timber structures - Volume 1 117

5.1 Structural elements
with full composite action

Aspects 10 and 11, out of this extensive list of failure modes and
requirements to be checked, are of little or no importance for beams
of practical dimensions, and not treated in this book. The remaining
aspects must at least be thought of whenever you attempt to design
a beam of this kind. Usually some of them can be ruled out for a par­
ticular problem. Detailed design rules and recommendations are given
in the following sections regarding most of the identified failure
mechanisms and serviceability requirements.

5.1.1.1 Transformed or fictitious cross-sections

Here a transformed cross-section merely means a fictitious cross-sec­
tion where the width of all parts except one has been changed such
that the same modulus of elasticity can be assumed for the entire
cross-section. The width is measured parallel to the relevant axis of
bending. If a cross-section is made of two or more materials having
different modulus of elasticity a transformed cross-section will sim­
plify equations and actual design calculations. Using a transformed
cross-section is by no means necessary from a theoretical point of
view, but it brings practical simplifications. Figure 5.4 shows
the transformation from a real crosssection into a fictitious
cross-section.
For a cross-section made up of 3 pieces as shown in figure 5.4
the fictitious widths are determined as:
EE E
\fic 5 51^’ 5h 5 5j1^ 5 5-’ \fic = 5’1

if part 2 is taken as reference. A proper value of the modulus of elas­

ticity must be used for each material making up the cross-section. For
wood based materials this value will in the long run be affected by
creep, as discussed in section 2.4.7, page 40. The influence of creep is
two folded for beams with a composite cross-section. First, the deflec­
tion and deformation of the beam increase with time. Second,
the internal distribution of stresses between members will change
with time. The outcome is that members that are more prone to
creep will be unloaded, while members less sensitive to creep will
carry more of the load with increasing time. Yet another problem is
that two different moduli are given in material property tables for
timber materials, for example in Eurocode 5. One value corresponds
to a lower percentile value (usually 5 %) and one corresponds to
the mean value of the modulus. Best practice is here to always use
the mean values of the modulus for each material whenever a fictitious

a)

Figure 5.4 Going for a fictitious or transformed cross-section

a) The real physical cross-section,
b) fictitious cross-section drawn to scale, if part 2 is taken as reference with
E1 = 2, E2 och E3 = E2/2.

118 Design of timber structures - Volume 1

 5.1 Structural elements
with full composite action

cross-section is needed. The simple reason for this is that we must try
to predict the most likely distribution of stresses and avoid stress dis­
tributions affected by large scatters in material data. Note, however,
that this reasoning applies to finding the best stress distribution
affecting a composite cross-section and may not be entirely correct
for instability problems, which is discussed in connection to equa­
tion 5.4 below.
Eurocode 5 uses the concepts of instantaneous and final conditions
(“inst” and “fin”) to handle the creep problem. In a more general case
we have n parts where i = 1, 2...... n; and if one of the i-values is asso­
ciated with the reference material, r, the fictitious width of the ith
part is:

r,mean
„ ^.mean (1 + ^2 V)
where "i,ULS,fm “ F T. ~ V 5.2
r,mean + V^2 ^i,def /
E. (1 + ^.Hrf)
i,mean \ r.def /
iUi,SLS,fin “ T T. 7 V
r,mcan + ^j;def /

Here ULS and SLS refer to the ultimate limit state and the serviceabil­
ity limit state, respectively. ki,def is the creep factor of the ith member
as if subjected to permanent loading and specified in Eurocode 5 for
different materials. y2 is the reduction factor for quasipermanent
loading and should be determined for the load causing the largest
stress. If that is a permanent action, a value of 1 should be used.
When and how to use the different a.-values in equation 5.2 is
explained as:

«.._„.
i,ULS,inst
= V i,SLS,inst» is used in the ULS or SLS under instantaneous condi-
tions, that is as if all design loads are applied during a period
shorter than a year or so. This value should, in the ULS, be
used to check the resistance of those members that are most
prone to creep.

Ai,ULS,fin is used in the ULS under final conditions, that is towards

the end of a design life of some 50 years or so. This value
should, in the ULS, be used to check the resistance of those
members that are less sensitive to creep. The technique takes
creep into account in an approximate way. Especially large
variable loads act for short periods of time, under which not
much creep will take place.

A i,SLS,fin is used in the SLS under final conditions to check deflections

and deformations after a long time like 50 years. This value
will give an estimation of creep deformations towards
the safe side, as only the creep factor for permanent load kdef
is used.

In practical design calculations the centre of gravity (GC), second area

moment of inertia and bending stiffness are calculated for
the fictitious cross-section, and in case of bending about the y-axis
Askims square.
usually denoted as zfic, Iy,fic and Er Iy,fic. Note that zfic is measured from
an arbitrarily chosen point of reference and that GC is the neutral

Design of timber structures - Volume 1 119

5.1 Structural elements
with full composite action

axis in case of pure bending, that is no external axial loading. If you

follow the above recommendations for p. you will end up with as
much as three different values of all relevant cross-sectional
properties. As an example the second area moment Iy,fic will take on
three different values: Iy,ULS,inst = Iy,SLS,inst, Iy,ULS,fin and Iy,SLS,fin.
Observe that when stresses are calculated for the fictitious section,
such that they are assumed to be uniformly distributed along
the width bi,fic for a given z-coordinate, they are also fictitious and
must be smeared out over the real width using the appropriate value
of p.i from equation 5.2 in the expression:

^ai^i^fic where 53i^n 5.3

For global stability problems such as column buckling and lateral tor­
sional buckling of composite beams, the recommendation to always
use the mean value of the elastic and shear moduli will be in conflict
to the definition of slenderness parameters (see sections 3.2.1, page 62,
and 3.2.2, page 64), which are generally defined as:

For column buckling For lateral torsional buckling

A:,rel “
Axial resistance
Critical axial load
^crit,rel ------- 5----------------
Bending resistance
Critical bending moment
5.4

The conflict is that the slenderness parameter should use the most
probable ratio, which is achieved if both the numerator and denomi­
nator are based on either characteristic or mean values of
the involved material parameters. Only characteristic material prop­
erties are available for the numerator (the resistance is based on fm,k,
ft,k, fc,k, etc.), while both characteristic and mean values are available
for the denominator (the critical load depends on E, G, etc. and not
on any strength parameters). It is therefore recommended that E- and
G-values used in stiffness expressions are taken as (lower) characteris­
tic values, while the fictitious section is still based on mean values of
the moduli. Typical stiffness values involved in finding critical loads
are Er Iy,fic, Er Iz,fic, Er Iw,fic and Gr It,fic for y-axis bending, z-axis bending,
warping and torsion, respectively.
The difference is usually smaller than 5 percent in the final design
resistance of a typical composite cross section, if comparing results
obtained using either characteristic or mean values of the elastic
moduli involved in determining the dimensions of the fictitious
cross-section. If global instability is a part of the analysis the differ­
ence usually stays within 10 percent, if mixing characteristic and
mean values when determining the slenderness parameter according
to equation 5.4. Hence, we conclude that a principal mistake regarding
the used moduli will have a small impact on the final result, if com­
pared to the large scatter of material data and the magnitude of par­
tial coefficients used. The same also holds if mixing up the i-values p
given in equation 5.2, page 119, which in general will have a small
influence on the final result. One can really ask if all the extra work
p
associated with these different i-values are worth the effort in practi­
cal design situations.

5.1.1.2 Bending resistance

Here a typical I-section is taken as an example, other sections (like
box-sections) are assessed in the same principal way. In figure 5.5,
page 121, an I-section having flanges of equal dimension but different
strength grades is illustrated. The stress distribution for pure bending

120 Design of timber structures - Volume 1

 5.1 Structural elements
with full composite action

^ aw,c,max,d

„
aw,t,max,d

Figure 5.5 Stress distributions on an I-section (or box-section) used to determine its bending moment resistance My,Rd
Note that the “real” stress distributions indicated is valid if Ew < Ef,c < Ef,t.

is shown. It is assumed that the area moment of inertia Iy,fic is deter­

mined according to equation 5.2, page 119, using the timber compres­
sion flange as reference and that the tension flange has a much better
strength grade than the compression flange. Note that this kind of
asymmetry should be avoided in practical design, because the beam
may accidentally be placed upside down.
In figure 5.5 six dots indicate positions on the cross-section where
the bending stress may become critically large in comparison to
the strength of the relevant material. The absolute value of these
stresses are determined using Navier’s equation for the fictitious
cross-section, but also multiplying by the appropriate E i /Er-ratio as in
equation 5.3. The stress for any of the dots is:

= £dot My,Ed | |
udot,d £ t pdotl 5.5
r y,fic

where zdot is the distance from the neutral axis to the dot and Er is
the modulus of elasticity of the reference material (in this case
the timber of the compression flange). It is now tempting to use equa­
tion 3.6 in section 3.1.2, page 57, to verify the resistance of the flanges.
This is because the total stress affecting a timber flange can be split
into two parts: one pure compressive or tensile stress at the centre of
each flange and one pure bending stress, taken as the difference
between the outermost edge stress and the centre stress. Finally, sub­
stituting these stresses into the appropriate interaction formula will
verify the resistance. It has however turned out that such an approach
is on the safe side compared to experimental results and also a bit
cumbersome. The recommended approach is much simpler. Usually
the flange depth is much smaller than the beam depth, in which case
their resistance is almost entirely controlled by the compression or
tensile strength of the flange material. Hence a reasonable design cri­
terion would be to avoid stresses at the flange centres that are greater
than the material strength at these points. But this approach is too
optimistic if the flange depth becomes large in relation to the beam
depth. This obstacle is overcome by realising that the bending
strength is greater than the compression strength and much greater
than the tensile strength for most timber materials. So the problem is
solved by adding the additional requirement that the stress at

Design of timber structures - Volume 1 121

5.1 Structural elements
with full composite action

the outermost fibres of the flanges must be smaller than the bending
strength. For the dots in figure 5.5, page 121, the following six verifica­
tions must be made:

Compression flange: O'f.c.d 7c,o,d ^f’.c.rnax.d - .4,d ~ -4w,d

5.6
Tensionflange: Cf^d A0,d CT£t4nax,d <-An,d CTw.t,m«,d Aw,d

where all g-stresses are calculated using equation 5.5; fc0d and fmd are
the compressive and bending strength of the timber in the compression
flange, respectively; ft,0,d and fm,d are the tensile and bending strength
of the timber in the tension flange, respectively; fc,w,d and ft,w,d are
the compressive and tensile strength of the web, respectively.
Using fc,w,d and ft,w,d will slightly underestimate the resistance of
the web panel/panels. The three main reasons for this are:
1) The strength values have been determined based on testing of
fairly wide and uniformly stressed specimens. But in a beam just
a small volume close to the edge is stressed to this level. 2) A lamina­
tion effect in the flange, where the timber will allow for some stress
redistribution from the web into the flange before final web failure.
3) Web material of wood based sheeting has a tendency to break away
the surface material if subjected to compression. This is more or less
prevented for a surface that is glued to the timber. 4) The shear defor­
mation of the web panel actually leads to a small increase of the axial
stresses in the web, which slightly counteracts the favourable effects
of remarks 1 to 3. By considering all these effects it should still be
possible to increase these particular strength values by 10 percent,
without hesitation. Such an increase is presently not promoted by
the Eurocode.
Note that the verification criterion (equation 5.6) does not cover
the risk of lateral torsional buckling. The background to lateral tor­
sional buckling of ordinary rectangular timber beams are treated in
section 3.2.3, page 65. Here some additional comments regarding thin
webbed composite beams will be given. One very simple and easy to
use method is to consider the compression flange to be an ordinary
column that can buckle sideways without having to drag the rest of
the cross-section with it. The column is cut out between two points at
which the compression flange is sufficiently braced laterally. Pinned
boundary conditions are assumed at these two points, that is
the flange is treated as an Euler 2 column. Verification of the resist­
ance is given by:
<7,..,
t,c,d 5.kf n.
c J c,0,d 5.7

where kc is the reduction factor for flexural column buckling as

described in section 3.2.1, page 62. The bad thing with this model is that
it grossly underestimates the bending resistance of the beam, especially
for beams with large torsional stiffness G It,fic, like box-sections. Yet
another problem is that equation 5.7 is designed to work for one mas­
sive section of solid timber, but a flange usually has some web panels
with different material properties attached to it. This is, however,
a small and negligible problem in relation to neglecting the rest of
the cross-section. A better approach to find the “true” resistance with
regard to lateral torsional buckling is to rewrite equation 5.7 as:

■^y,Ed - ^y,LT,Rd = ^crit ^y.Rd 5.8

where My,LT,Rd is the moment resistance due to lateral torsional buckling,

kcrit is the reduction factor covering up for lateral torsional buckling

122 Design of timber structures - Volume 1

 5.1 Structural elements
with full composite action

and My,Rd is the design resistance not influenced by instability and

taken from equation 5.6, page 122. The reduction factor kcrit is obtained
following the sequence: critical value ^ slenderness parameter ^
reduction factor, that is:

My,cnt ent

The designation A refer in the following sections to Acritrel unless other­

wise stated.
In determining A the bending moment resistance Rk is needed,
which is the characteristic resistance without taking lateral instabil­
ity into account. This value is easily obtained using equation 5.6,
page 122, with characteristic values rather than design values. Next
step is to find the critical bending moment My,crit for the actual support
and loading conditions at hand. Note that this is a theoretical value
obtained for a beam without any imperfections whatsoever. The basic
idea is that the critical value captures the influence of geometry,
boundary conditions and how the load is entering into the beam.
Normally, My,crit is calculated based on some elementary load case
from the literature. Cross-sectional properties needed are typically E
Iz,fic, E Iw,fic and G It,fic for z-axis bending, warping and torsion, respec­
tively. They should be determined as discussed in section 5.1.1.1,
page 118, that is Iz,fic, Iw,fic and It,fic are calculated based on mean values
of elastic and shear moduli, while E and G in front of them should
reflect the characteristic value. It is not straightforward to find My,crit ,
but not unrealistically problematic. Guidance is given in many hand­
books. Next, the slenderness parameter for lateral torsional buckling
is obtained as:

~^y,Rk
A< 5.9
Vy My.cntv dim)'

The value of My,crit should be taken as close as possible to the design

section xdim where the beam is subjected to a combination of stiff axis
bending, weak axis bending and torsion. Sometimes it is obvious
where xdim is located, sometimes one has to make a qualified estima­
tion. It is of course always on the safe side to choose xdim such that
My,crit has its maximum, but that may not always be the same position
as where the beam section will fail.
Finally, kcrit can be read from a suitable design curve of an instabil­
ity chart. Presently, only one design curve exists for lateral torsional
buckling of timber beams, and that curve is actually valid for sec­
tions of rectangular structural timber and glulam. No specific curve
exists for more complicated sections. In figure 5.6, page 124, the one
and only curve for lateral stability (as given in section 3.2.3.3, page 68)
is brought together with the two curves used for ordinary column
buckling of structural timber and glulam (as given in section 3.2.1,
page 62). A fourth extremely important curve 1/A2 is also depicted.
This curve exactly represents the theoretical critical load of a beam
or column without any imperfections, that is critical bending
moment in case of a beam and critical axial load in case of column
buckling. You may easily verify this by rewriting 1/A2 using equa­
tion 5.9 and then substituting into equation 5.8, page 122, which
results in the critical bending moment. It works the same for flexural
column buckling, except that the bending moment is replaced by
axial force. In either case it is important to realise that the critical
curve 1/A2 represents an absolute upper limit of the resistance, no
postbuckling resistance is possible.

Design of timber structures - Volume 1 123

5.1 Structural elements
with full composite action

Critical load
kcrit
kc glulam
kc structural timber

Figure 5.6 Buckling curves for lateral torsional buckling and flexural column buckling

The whole idea behind the chart offigure 5.6 is that the influence of
geometry, support and loading arrangements enters via A, while
the buckling curves only corrects the resistance with respect to all
relevant imperfections, where the most important ones for wood
based materials are non-linear material and bow imperfections.
The critical load is determined under the assumption of a linear elas­
tic material that can sustain infinitely large stress. In reality kcrit or kc
can never exceed unity. We see that the two kc-curves give a slightly
greater reduction than the kcrit-curve. It can also be shown that
a curve for lateral torsional buckling must give smaller reductions
than for flexural column buckling. Since no specific curve exist for
lateral torsional buckling of composite I- and box-sections it is here
proposed that the present kcrit-curve can be used also for composite
beams, especially since most composite beams have smaller bow
imperfections than timber beams with rectangular cross-sections.
And if one for some reason feel uncertain about the applicability of
the kcrit-curve it should be safe to use the upper kc-curve for flexural
buckling of glulam.
The procedure to find the resistance with regard to lateral torsional
buckling is summarized as follows: The characteristic short term
properties should be used when calculating the slenderness accord­
ing to equation 5.9, page 123, since the slenderness only reflects geom­
etry, boundary conditions and load arrangement. But after determin­
ing the reduction factor kcrit as a function of A the reduction should be
applied in equation 5.8, page 122, where My,Rd should be calculated
based on design values modified for duration of load with k mod
according to section 2.4.4, also see section 3.2 in Volume 2. It is recom­
mended to determine the slenderness parameter A under instantane­
ous (short term) conditions only, since a determination under final
conditions (long term) would have a small and insignificant impact
on the ultimate resistance. But note that in finding My,Rd itself both
“inst” and “fin” conditions may have to be considered.

5.1.1.3 Shear resistance of web panels

without open web joints
Two kinds of failures are possible for a uniform web panel without
any open joints, either a pure panel shear failure or an out of plane

124 Design of timber structures - Volume 1

 5.1 Structural elements
with full composite action

bending failure caused by shear buckling. For web panels with open
joints the reader is referred to Thielgard and Larsen (1978).
Approximately, we may assume panel shear failure if hw /b w < 35 and
shear buckling failure if 35 < hw/bw < 70. Other kinds of buckling
phenomena may limit the resistance for values above 70. But web
panels having a slenderness above this limit are rarely used in practi­
cal applications. Actually, a slenderness hw /bw above 70 should be
avoided unless the designer is experienced enough to handle all
instability scenarios that may occur. Note that the limit 35 is sup­
posed to be valid for all board materials, but values adjusted to better
match the properties of individual board materials are given in
table 5.2, page 139.
Let us start with a pure panel shear failure under linear elastic con­
ditions. The shear stresses counteracting the shear force Vz,Ed are then
distributed over the cross-section as shown in figure 5.7.
The simple failure criterion is that tvEd < fvmind. The shear stress can
be obtained using the well-known shear formula as given by:

^z,Ed ^y,fic(Z) V V
_ z,Ed z,Ed
Tv,Ed ~ £ 5.10
Iy,fic
. bw

where the left expression gives the shear stress parallel to the x- or
z-axis at any point z in the web panel. Sy,fic (z) is the static moment of
all area on either side of the z-coordinate, which is measured from
the centre of the fictitious cross-section. The only problem with this
formula, aside from its complexity, is that it overestimates the magni­
tude of the shear stress by some 5 to 10 percent for ordinary I-beams.
The reason is that it is derived under the assumption that no defor­
mations are caused by the shear stresses and that only axial strains
caused by the bending moment contribute. In reality shear stresses
will cause shear straining, which in turn will reduce the shear stress
below the value given by the left part in equation 5.10. If we instead
assume all shear stresses to be uniformly distributed over
the cross-sectional area of the web panel, Aw measured from flange
centre to flange centre, a much simpler relation is obtained. We have
in this simpler failure criterion neglected the slightly parabolic shape
of the stress distribution in the web panel. But this is justified by
neglecting the shear contribution from both outer flange parts. And
it is also justified by the fact that the actual shear stresses are smaller,
as described above. The simplified version of equation 5.10 is recom­
mended for practical design, as in Eurocode 5. Note thatfvmind is
the minimum value of the panel shear strength taken parallel or per­
pendicular to the beam axis. The shear strength of plywood varies

Tv(z)
Tv,Ed
y

Figure 5.7 A piece of an I-beam subjected to shear force

Design of timber structures - Volume 1 125

5.1 Structural elements
with full composite action

considerably with the angle between its fibre orientation and the lon­
gitudinal beam axis. The shear strength of plywood is more than twice
as large as the tabulated value for an angle of 45°. In Eurocode 5
fv,min,d is denoted fv,0,d and no distinction is made between panel shear
stresses acting parallel and perpendicular to the beam axis.
Figure 5.8 shows two side views and one end view of typical I-beams
for which shear buckling is limiting the resistance. The buckling pat­
tern is clearly indicated. Note that the web panel must under all cir­
cumstances be furnished with transverse web stiffeners at the sup­
ports, whenever shear buckling is limiting the resistance. But more
stiffeners are rarely needed, unless large concentrated forces are
applied to the beam in which case additional stiffeners may be needed
at those locations. Note that web stiffeners may be needed at the sup­
ports even if shear buckling is no problem. This is because of crushing
or local buckling of the web panel due to concentrated loading, that
is patch loading may limit the resistance. The best practice to avoid
patch loading failure where concentrated force are applied, like at
supports, is to always use web stiffeners at these locations. How to
check the patch load resistance if no stiffeners are used is not explic­
itly covered in this book.
By rewriting the right part of equation 5.10, page 125, and introduc­
ing a reduction factor xv covering the shear buckling phenomenon
we have:

^z,Ed - ^z,Rd “ %v /v,min,d ~ %v \^w + ) A,min,d 5.11

Since shear buckling is an instability problem we may proceed in

the same way as for lateral torsional buckling in the previous section,
that is follow the chain: critical value ^ slenderness parameter ^
reduction factor, that is using tc ^ kv ^ xv, where tcr is the critical
shear buckling stress that will cause buckling of a web panel without
imperfections, that is a linear elastic and perfectly flat plate having
no residual stresses. The value of tcr catches the influence of the plate’s
geometry, boundary conditions and the shape of the in-plane stress
field acting across the plate. The critical shear buckling stress should
be determined based on characteristic values of the web panel’s elas­
tic properties, like E and G. Formulas and accompanying values of
the buckling coefficient kt used to calculate tc may be taken from
some relevant literature, for instance Lekhnitsky (1968). However,
two cases that cover most practical applications are given here. In
both cases the plate is rectangular, simply supported around all 4
edges and subjected to a uniform field of shear stress as shown in fig­
ure 5.9, page 127. These conditions are similar to those experienced by
a cut-out part from any of the web panels indicated in figure 5.8.

bf
Flange

Stiffener
Web

Cross-section
h /bw 2 70
Figure 5.8 I-beams for which shear buckling is limiting the resistance, that is 35 < w

126 Design of timber structures - Volume 1

 5.1 Structural elements
with full composite action

k
Figure 5.9 Buckling coefficient T for a simply supported plate subjected to uniform shear stress
around all four edges. a) Isotropic plate, b) orthotropic plate.

For isotropic plates, see figure 5.9 a), the buckling coefficient and its
corresponding shear buckling stress can be determined as:

^=5,34 + 4,0 5.12

where v is the Poisson’s ratio, which for hardboard can be set to 0,20
and to zero for particleboard if no better values are available. Note
that wood based board materials can be viewed as isotropic if
the fibre orientations are random within the plane of the plate (like
for particleboard and hardboard, but not for plywood and OSB which
have a clear grain direction). The buckling coefficient is simply a par­
abolic approximation of the theoretically “true” result shown in fig­
ure 5.9 a). For orthotropic plates it is a bit more complicated but
the buckling coefficient can be taken from figure 5.9 b), where param­
eters fl1 and fl2 together with the shear buckling stress are given by:

and figure 5.9 b) => k and 5.13

In equation 5.13 Dx is the plate bending stiffness about the x-axis, Dz is

the plate bending stiffness about the z-axis and Dxz is the torsional
stiffness of the web panel. The critical buckling stresses predicted by
equations 5.12 and 5.13 are in general greater because of the torsional
restraints provided by the flanges and the stiffeners. For an isotropic
web panel having all four edges fully built in (infinite rotational stiff­
ness of flanges and stiffeners) the critical stress is 60 to 70 percent
greater than calculated by equation 5.12, but that is in practice not to
achieve.

The slenderness parameter is defined as:

5.14

Design of timber structures - Volume 1 127

5.1 Structural elements
with full composite action

Critical load
Design curve

Figure 5.10 Design shear buckling curve 0,8/2v , compared to the critical curve 1/2v2

where fv,min,k is the characteristic minimum panel shear strength

taken parallel to the x- or z-axis. Finally the reduction factor xv,
accounting for all imperfections, is obtained using a proper design
curve. In Hoglund (1990) it is shown that:

Zv = min 5.16

is a simple and suitable curve, which compared to experimental

results and advanced numerical simulations give results slightly on
the safe side. The design curve is shown in figure 5.10 together with
the critical curve for a plate without imperfections. A comparison of
the two curves indicates large post buckling resistance for Av > 1,25
and no influence of shear buckling for Av < 0,8, for Av < 1,25 the resist­
ance is smaller than the critical value. Note that the post buckling
resistance can only be utilised if the web panel is equipped with
transverse web stiffeners at all supports, but more stiffeners are gen­
erally not needed. However, if more and closely spaced stiffeners are
added an increase in resistance is possible, because of a larger critical
stress and contribution from flange bending (vierendeel beam effect),
as described later in this section, see figure 5.11, page 130.
The reduction factor xv can for many commonly used designs be
further simplified, as to obtain a lower limit value of the resistance.
Let us check an I-beam having an 8 mm thick hardboard web of
strength quality HB.HLA2 according to EN 13986. The characteristic
panel shear strength, modulus of elasticity and shear modulus are
fv,min,k = 16 MPa, E = 3 680 MPa and G = 1 520 MPa, respectively.
Poisson’s ratio is, from linear elasticity, estimated to v = E/ (2 G) — 1
= 0,21. Substituting equations 5.12, page 127, and 5.14, page 137, into
equation 5.15 and at the same time introducing numerical values for
fv,min,k, E and v gives equation 5.16, page 129. But before substitution,
the critical stress according to equation 5.12, page 127, is increased by
25 percent to account for some torsional rigidity coming from
the flanges and the buckling coefficient kT is set to 5,34.

128 Design of timber structures - Volume 1

 5.1 Structural elements
with full composite action

For a long plate (like Iw > 2 hw) kT is rapidly approaching 5,34

according to figure 5.9 a), page 127, hence 5,34 is a safe assumption for
an ordinary I-beam having transverse stiffeners only at the supports:

7C2E
*3
1,25 'i
^-0,8
h
ZV=^O,8
•Ar,minjc v,mm,k
5.16

b
0,8 = 29,1—
hw

If repeating this process for different panel materials, isotropic as well

as orthotropic, the number in place of 29,1 will end up between 28
to 73 and in the majority of cases between 30 and 40. More detailed
values of this coefficient are given in the two rightmost columns of
table 5.2, page 139. In Eurocode 5 a value of 35 has been chosen to be
applicable to all wood based panel materials. The only difference
being that for orthotropic materials equation 5.13, page 127, is used for
the critical load and the buckling coefficient kT is taken as 3,81, which
is the smallest value in figure 5.9 b), page 127. Finally, replacing 29,1 in
equation 5.16 by 35 and substituting into equation 5.11, page 126, gives:

WFz,Rd = M 1+/ 5.17

v,min,d
k w?

in which 35 bw can be thought of as an effective depth of the web

panel with regard to shear buckling. Note also that from equation 5.16
it immediately follows that shear buckling is no problem if hw /b w <
35, that is if 29,1 is replaced by 35. Note also that avoiding slender
web panels for which hw /bw > 70 means that xv is always greater than
0,5, alternatively Av is always smaller than 1,6. It is possible to use
more slender web panels, but then local buckling from axial stresses
(bending stresses) and web buckling caused by transverse compres­
sive stresses from distributed loads and beam curvature must be thor­
oughly checked. Concentrated transverse loads are no problem
because web stiffeners must always be used at such locations if
the web panel is so slender that shear buckling is possible, that is
hw /bw > 35.
Beams having closely spaced transverse stiffeners have greater
resistance than predicted by equation 5.17. The reasons are two: first,
the resistance should be based on equation 5.11, page 126, including
the increased shear buckling resistance gained by the additional stiff­
eners, and second, the flanges will now contribute more to the shear
resistance because of local bending between adjacent stiffeners as
shown in figure 5.11, page 130.
A greater portion of the shear force is carried by the flanges after
buckling of the web panel. It is as if the shear modulus of the web
panel gets smaller due to the growing shear buckles.
Stiffeners and flanges constitute an ordinary framework with consid­
erable stiffness, greater stiffness with decreasing spacing of stiffeners.

Design of timber structures - Volume 1 129

5.1 Structural elements
with full composite action

Point of
contraflexure

Vf
Vf/2

b) w /2
Figure 5.11 Local flange bending between transverse web stiffeners contributes to the shear resistance of
the web, if shear buckling is limiting the resistance

The web panel connects flanges and stiffeners to fairly stiff frame
corners. The maximum shear force that can be carried by the flanges
alone, can be estimated from the bending resistance of a flange. Half
the shear force is acting at the point of contraflexure, where
the bending moment in the flange is zero, see figure 5.11 b). At any of
the 4 points indicated we have:

V
Kf I w
Mf _ 2 2 2 .
v <— f bf—hfL
-£ 5.18
W{ b{hf f 3 7f’md t
6

The bending moments can as a rule not reach the full resistance of
the flange, because the shear deformation of the web panel will not
be large enough. It is, therefore, reasonable not to use more than
a little less than half of 2/ 3, say 0,3. The total shear resistance is now
the sum of equations 5.11, page 126, and 5.18. But the panel shear
strength according to the right part of equation 5.10, page 125, cannot
be exceeded. Hence we get:

%v ^z,max,Rd + ^’^/f,md
V
z,Rd = mm< w 5.19

V
z,max, Rd v,min,d

5.1.1.4 Shear strength of flange-to-web connections

For a composite beam it is important that the shear stresses in
the web panel can be transferred into the flanges without any failure
of the connecting glue line. Only glued connections are considered in
this section, mechanical connections are treated in section 5.2, page 142.
Shear stresses to be transferred across sections 1 — 1 and 2 — 2 in fig­
ure 5.12 a), page 131, can be estimated by the traditional shear for­
mula, which if expressed as a failure criterion takes the form:

tv,Ed = Th
<f
J gl>nun,d if
gl 5.20
y,fic "gl "gl

where fgl,min,d usually is the planar (rolling) shear strength of the web
panel if not for some odd reason the shear strength of the flange
material in the longitudinal direction is smaller. Equation 5.20 is valid
if the shear stresses are assumed to be uniformly distributed along
the depth of each glue line hgl. The parameter ngl is the number of
glued surfaces per web panel and flange, that is 1 for a box-section
and 2 for an I-section. In this case all the glue lines are of the same

130 Design of timber structures - Volume 1

 5.1 Structural elements
with full composite action

Tv,2,Ed

dx
GC

Section A-A
a)

Figure 5.12 Shear stresses in glued flange-to-web connections. a) Entire cross-section, b) sections 1 -1 and 2-2 cut open
indicating distribution and direction of shear stresses, c) as in (b), but for the bottom flange having a deep glue line.

depth, which is not a requirement. The parameter ASyific is the first

area moment of the flange part belonging to each glue line i, i = 1 or
2 taken with respect to GC. It should be calculated for a part of
the flange extending from the glue line to a point of zero shear
stress, usually a free edge or a line of symmetry. In the upper part of
figure 5.12 a) is shown how the shear stresses are distributed along
the flange width, that is stresses on any sectional cut parallel to
the z-axis. Note also that equation 5.20, page 130, gives the “real” stress
and not a fictitious stress, because the stress is uniformly smeared out
along the real glue line hgl as depicted in figure 5.12 b). The assumption
of uniformly distributed shear stresses works fine as long as hgl <
4bw / ngl, which was shown by Foschi (1970). For greater glue line
depths there is a non-negligible concentration of the shear stresses
towards the re-entrant corners where the flange meets the web panel.
The concentration is more severe for a double sided glue line (I-section)
compared to a single sided glue line (box-section). For hgl > 4b w /ngl
Foschi (1970) has suggested a modified version of equation 5.20, which
reads:
, A0,8
_ ^z,Ed A^y.i.fic "gi \i
r if
4bw
h >— 5.21
V,Ed ” 1 fi h 1 4b gl,min,d 11 "gl5*
y,fic gl "gl

A correction is possible to utilise for plywood-webbed beams having

the grain direction of the face veneers parallel to the flanges. The rea­
son is that the veneer which is sensitive to planar (rolling) shear is
located one veneer thickness away from the glue line surface where
the shear stress is smaller than at the surface. The shear stress obtained
from equations 5.20, page 130, or 5.21 can be reduced as shown in fig­
ure 5.13, page 132 , by assuming a linear decline of the shear stress,
which results in:

pt-h
n sv
"gi_____
Sing =1 single sided glue line
Tv,red,Ed - T’ where 5.22
b Mg! = 2 double sided glue line

"gl

Design of timber structures - Volume 1 131

5.1 Structural elements
with full composite action

Figure 5.13 A possible reduction of shear stresses for plywood. a) One glued and one free surface, b) two glued surfaces.

Note that the reduction given by equation 5.22, page 131, might be
a bit too optimistic for a double sided glue line if the number of
veneers is smaller than 6.

5.1.1.5 Bending and shear deflections

The shear deflection is not negligible for full composite beams having
webs of relatively slender wood based sheet materials, and must be
added to the usual bending deflection. In addition to this the different
flange and web materials have different creep properties, making it
necessary to distinguish between “instantaneous” and “final” deflec­
tion. Instantaneous means that no creep has yet occurred and final
means full creep deformations accumulated during a period of about
50 years. This means that two different second area moments have to
be calculated for the transformed cross-section using “inst” and “fin”
values of the elastic moduli in equation 5.2, page 119. The proper elastic
modulus to use for each case and each material is given in table 5.4,
page 158. Finally, the deflections should be calculated as described by
the principles given by equations 5.85 and 5.86, page 159. These
expressions are, however, indexed as to make sense for beams sub­
jected to partial composite action. But by changing index “fca” to
“bending”, “slip” to “shear” and erasing factor “2” from the last three
terms of equation 5.86 they are ready to use for full composite beams
as well. Make sure to use kdef for the reference material used in equa­
tion 5.2 in all the bending terms and kdef for the web material in all
the shear terms of equation 5.86.
Once E0,mean Iy,fic,inst and E0,mean Iy,fic,fin are known the bending deflection
is easily calculated using traditional elementary cases for beam bend­
ing found in the literature. E0,mean is the mean modulus of elasticity of
the reference material taken directly from material standards, see for
instance section 3.4 in Volume 2. Unfortunately, formulas for shear
deflections are seldom given in such literature. The shear deflection
can, for statically determinate beams, be found by direct integration
of the shear angle caused by the shear force. It is assumed that all
shear deformation is concentrated to the web panels and that
the entire shear force is carried by them. The differential equation to
integrate then becomes:

A (' K(x)
^^shearW q A 5.23

where Gw is the mean shear modulus of the web directly from mate­
rial standards, Aw is the total web area taken between flange centres

132 Design of timber structures - Volume 1

 5.1 Structural elements
with full composite action

and V(x) is how the shear force varies along the beam. Direct integra­
tion of equation 5.23, page 132, and determination of integration con­
stants with regard to the boundary conditions result in a function for
the shear deflection. For a simply supported beam subjected to uniform
loading q, that is the most common support and loading conditions
in the world, the maximum shear deflection at midspan becomes:

qL1
23
W,shear,max 5.24

5.1.1.6 Advantages and disadvantages

with I- and box-sections
As with all structural solutions there are advantages and disadvan­
tages, some are of a statical nature and some of a practical or eco­
nomical nature. Some advantages of I-sections in contrast to hollow
sections are:
• easy to put web stiffeners in place at end and intermediate supports
to counteract the influence of patch loading and shear buckling
(actually this holds for all kinds of web buckling)
• if necessary, easy to reinforce open web joints after manufacturing
• slightly easier to reinforce or to arrange flange joints if needed
• easier to make holes in the web panel for installations.

Some advantages of hollow sections in contrast to I-sections are:

• much better torsional stiffness and bending stiffness about the minor
axis, which is good for avoiding lateral torsional buckling
• easier to put thermal or sound insulation in place and generally
more suitable in production
• if the web panels are made of hardboard it is somewhat more diffi­
cult to nail using an ordinary hammer, that is whenever the nail
must first penetrate the hardboard.

5.1.2 Glued thin flanged elements (SSP)

Examples of SSP-elements are shown in figure 5.14. Two basic shapes
can be distinguished, open T-sections and closed hollow sections
(H-sections). The flanges are usually made of particleboard, plywood,
OSB or hardboard; but tension flanges made of steel sheeting are not
uncommon. The web parts are usually made of sawn timber, glued i' ' 1 1 ' 1 1 "i

laminated timber or LVL; but web parts that are themselves compos­
ite beams are not unusual.
Figure 5.15, page 134, shows one typical SSP-element, an H-section
having two thin flanges and webs of rectangular timber pieces.
Some relevant failure locations are indicated symbolically and
given a specific number in figure 5.15. The indicated failures refer to Figure 5.14 Typical cross-sections of stressed skin panels
a section with stocky web parts that are not sensitive to any kind of (SSP), H- as well as T-sections are shown
buckling. If, however, the web parts are made of thin sheeting
materials the relevant failure modes from figure 5.3, page 117, and sec­
tion 5.1.1, page 116, should also be checked. The most relevant failure
modes of elements with stocky webs are:

1. Compression failure of the compression flange, which includes

the effect of local buckling, shear lag and if necessary flange
curling.
2. Tension failure of the tension flange, including the influence of
shear lag and possibly flange curling.
3. Bending failure of the web parts.

Design of timber structures - Volume 1 133

5.1 Structural elements
with full composite action

Figure 5.15 Double flanged SSP-element. a) Cross-section, b) elevation, c) top view.

4. Shear failure of the web parts. The resistance is verified as for struc­
tural timber, for which design rules are given in section 3.1.1, page 54.
If the webs are made of sheet material, the rules in section 5.1.1.3,
page 124, apply.
5. Shear failure of flange-to-web interface, that is the glued connection
between them. The resistance is basically checked as for I-girders,
which is described in section 5.1.1.4, page 130. Note, that if the inter­
face itself is strong enough, there is still a risk for a pure panel shear
failure of the flange material on either side of the web members
both for the tension and compression flanges. The exact points for
this failure mode are not indicated with numbers in figure 5.15, but
the position is almost as for point 5 but in the flanges themselves.
6. Failure of web parts with holes. No detailed design rules are given
in this book, see for example instructions from the manufacturers.
7. Flexural buckling failure caused by axial loading, not shown in fig­
ure 5.15. Flexural buckling is treated as for ordinary columns in sec­
tion 3.2.1, page 62.
8. Compression failure of the outer flange surfaces or the web parts in
contact with the flanges caused by concentrated forces acting per­
pendicular to the grain, which is basically checked using the rules
given in section 3.1.3, page 58. But the contact area between flange
and web can be slightly increased if the sheeting material has good
compression properties in its thickness direction.
9. The influence of flange curling, that is transverse bending of both
the compression and tension flanges due to the element curvature
caused by the external bending moment. Of course some flange
curling may also be induced by transverse loading to a flange in
the regions between the webs. The basic influence of both flange
curling and transverse loading is reduced bending stiffness as
the distance between the flanges decreases. As a consequence,
the axial loading of the flanges as well as the bending deflection of
the element will increase.
10. In the serviceability limit state one have to check deflections pri­
marily caused by bending. Shear deformations can at least for
stocky webs be neglected, but should be added for slender web pan­
els. The influence of vibrations and resonance frequencies in floors
and bridge decks may be of profound importance. More information
about these topics is given in Chapter 7, page 183.

Detailed design rules and recommendations are given in the following

sections regarding the identified failure mechanisms under 1, 2, 3 and
9, also the deflection part of 10 is briefly discussed.

134 Design of timber structures - Volume 1

 5.1 Structural elements
with full composite action

5.1.2.1 Bending resistance

The bending resistance may at the same time be influenced by shear
lag, local flange buckling and flange curling. For a long element with
flanges having a low width-to-thickness ratio none of them may have
any substantial influence on the resistance, while the opposite is true
for short elements with slender flanges. The common denominator of
these influences are that they are all accounted for by changing
the physical cross-sectional dimensions of the SSP-element into
an effective cross-section; for which the resistance check is carried
out in exactly the same way whether or not the resistance is affected
by shear lag, local buckling and flange curling. The shape and basic
dimensions of the effective cross-section for H- and T-shaped ele­
ments are given in figure 5.16.
How the effective cross-section is obtained will be shown later.
First the resistance checks to be made are given, if the dimensions of
the effective cross-section are known. The resistance is simply veri­
fied by checking the stress levels at the locations of the dots in fig­
ure 5.16, that is in almost the same way as for I- or box-sections. But
before making these checks the effective cross-section is for conveni­
ence usually changed into a “transformed” section, to account for dif­
ferent elastic moduli and creep properties of the web and flange
materials, as described in section 5.1.1.1, page 118. The axial stress at
each dot in figure 5.16 is obtained by equation 5.5, page 121, and then
compared to the appropriate design strengths as:

°f,c,d - /c,0,d OT -4,90,d (Compression flange)

- Z,o,d or ftenoji (Tension flange) 5.25

(Ww,max,d h-'rn.d(
(-bef (Web edge)
^.eue^e,

The last inequality of equation 5.25 applies to webs of structural tim­

ber having a rectangular shape. Comparing the maximum edge stress
to the bending strength is only relevant if the stress distribution over
the web is sufficiently close to a pure bending distribution, which is
usually the case for H-shaped SSP-elements. If, however, the stress
distribution can be split into one tension or compression part plus
one bending part the checking of the web should be made through
the interaction criteria described in Chapter 3, page 54, for tension and
compression. This is usually the case for T-shaped SSP-elements.

bf bf bf bf

bf/2 bf/2

Compression
NA NA

Tension

Figure 5.16 Effective cross-sections of stressed skin panels (SSP), H- as well as T-sections

Design of timber structures - Volume 1 135

5.1 Structural elements
with full composite action

beff /2 /2
beff If the webs are made of wood based sheeting, then the last inequality
of equation 5.25, page 135, is replaced by a verification of the edge
stresses being smaller than the tensile and compressive strengths of
the board material, as appropriate.
If glued thin webbed beams are used as webs in an SSP-element
the last inequality of equation 5.25 is replaced by equation 5.6, page 122.
Now, the methods of finding the effective cross-section with regard
to the influence of shear lag, local buckling and flange curling will
be discussed. We start by discussing shear lag in combination with
local buckling, whereafter flange curling is treated separately.
Both shear lag and local buckling cause a non-uniform stress distri­
bution along the flange width. The maximum stress appears around
the web panels, while the minimum stress appears between them.
/2 Shear lag may influence both compression and tension flanges, while
/2
beff beff
local buckling is only relevant for compression flanges.
The flanges of any beam are subjected to in-plane shear stresses,
that is acting in a plane “perpendicular” to the webs. It is these shear
stresses that gradually build up the axial stresses in the flange, but at
the same time they will cause some small shear deformation of
the flange. This shear deformation will consume some of the axial
straining needed to build up full axial stress in the flange material
Buckle between the webs, resulting in a non-uniform stress distribution as
shown in figure 5.17 a). From a designers point of view this fairly com­
plicated stress distribution is handled through the concept of effective
b) width. The real flange width is replaced by an effective width for
which gmean b f = gmax b eff is supposed to be valid.
Figure 5.17 Non-uniform stress distributions, in a wide
To find the best effective width in each particular case is difficult
compression flange between two webs, caused by:
a) shear lag only,
because it is influenced by many factors, out of which the most
b) local buckling only. important are: shear properties of the flange material, loading config­
uration, boundary conditions, ratio of flange width to span length
and local buckling. In Eurocode 5 the effective width is simply given
as a constant fraction of the span regardless of this complexity.
The span is the distance between points of contraflexure or zero
bending moment. In figure 5.18 a comparison between the simple
Eurocode expression and theoretically more correct relations are
given for a flange made of OSB loaded either perpendicular or parallel

osb ± q
osb q
Z'

osb ± P
osb Z' P
EC5: 0,15Z
beff
bf

Figure 5.18 The influence of shear lag for an OSB flange in a simply supported SSP-element
subjected to two different loading conditions

136 Design of timber structures - Volume 1

 5.1 Structural elements
with full composite action

to its main grain direction, here denoted by ± or //, respectively.

The SSP-element is simply supported and loaded by either a uniformly
distributed load acting along the entire span or a concentrated load
at midspan, labelled q or P. It is clear that the Eurocode expression of
0,15 l gives a good effective width only if the ratio of flange width to
span is about 0,16 for a q-load and 0,20 for a P-load. For longer spans
the effective width is overestimated by as much as 20 percent, while
the effective width is underestimated for shorter spans. Flanges of
hardboard, particleboard and plywood show similar results, but ply­
wood shows larger scatter and deviations between the Eurocode cur-
veand theoretical results. The theoretical curves were obtained by
FE-analysis under linear elastic conditions.
More accurate predictions of the effective width are obtained by
approximating the theoretical curves with straight lines on the inter­
val of practical interest 0,05 < b f /l < 0,30. The function expression for
such fitted results is given by:

h (h
f 5.26
°f J

where the coefficients ft0 and ft1 are given in table 5.1 together with
the traditional Eurocode expressions.
Buckling of compression flanges gives a non-uniform distribution
of axial stresses which is similar in shape to the distribution caused
by shear lag, as indicated in figure 5.17 b), page 136. Shear lag reduces
the influence of local buckling because it leads to smaller compres­
sive stresses in the part of the flange that is sensitive to buckling.
The other way around local buckling will reduce the influence of
shear lag. But we start to discuss local flange buckling as a stand-alone
problem. As an instability problem it can be accounted for using
the same principal
approach as outlined for shear buckling in section 5.1.1.3, page 124,
resulting in an effective flange width (b eff = b f xc). This process may be
summarised as: critical compressive stress ^ slenderness parameter
^ reduction factor, that is using gcr ^ Xc ^ xc, where index c indicates
compression. The slenderness parameter is defined in the same way
as equation 5.14, page 127 and the design curve is chosen as in equa­
tion 5.15, page 128, but index v is replaced by c to indicate compression.

Table 5.1 Effective width due to shear lag in wide compression or tension flanges
Widths according to both Eurocode 5 and equation 5.26 are given.

Board material Eurocode 5 Type of loading 2) 3) Coefficient p 0 Coefficient p 1

bfff in (5.26) in (5.26)

OSB 0,15 l q 1,03 0,72

p 0,97 1,05

Plywood, 3± 1) 0,10 l q 1,05 1,15

p 0,93 1,35
Plywood 0,10 l q 1,06 1,80

p 0,83 1,60

Particleboard 0,20 l q 1,03 0,61

or fibreboard
p 0,98 0,90

1) 3± denotes a three veneer plywood with the face veneers oriented perpendicular to the axial stresses.

2) q is a uniformly distributed load on a simply supported beam.

3) P is one concentrated load at midspan on a simply supported beam.

Design of timber structures - Volume 1 137

5.1 Structural elements
with full composite action

This results in a reduction factor expressed as:

^-0,8
%c = 5.27

where ocr is the critical buckling stress for a long plate simply sup­
ported along all four edges and subjected to uniform compression in
its longitudinal direction. The factor 1,25 increases the buckling
stress by 25 % to approximately account for some of the rotational
stiffness provided by stocky web members (total rotational constraint
along the webs would give about 60 %). If, however, the web members
are thin and slender 1,25 should be replaced by 1,0, as thin board
members provide little rotational stiffness. of,c is the compressive
stress in the flange at the load level where the bending resistance of
the entire SSP-element is exhausted, under the assumption that no
local buckling affects the resistance of the flange. It is always on
the safe side to set of,c = fc,0,d or fc,90,d, as appropriate. But if the bending
resistance is governed by the tension flange or webs, rather than
the compression flange, a proportionately reduced value can be used.
Also, if shear lag is affecting the compression flange the above value
of of,c can be reduced by the factor beff,shear lag/bf to take the beneficial
effect of shear lag into account.
Note that it is not self-evident that that the design curve should be
0,8 /Ac, but it must under all circumstances have a similar shape.
A comparison to the more complicated curves used for steel in
Eurocode 3, observing that those curves also take the influence of
residual stresses into account, indicates that 0,9 /Ac might be a better
choice. But choosing 0,8 is slightly on the safe side, especially for
slender flanges, and can also be said to implicitly account for some
small influence of flange curling. This simple formula for the design
curve makes it possible to express the effective width as a constant x
flange thickness, as we shall see next.
The critical stress ocr is calculated based on linear elastic theory and
a lower limit, on the safe side, is obtained by assuming the flange to
be a long plate simply supported along all four edges. But the value
can easily be increased or decreased if other boundary conditions are
more relevant, for example increased if the flange is equipped with
closely spaced transverse stiffeners. Compression flanges in SSP-
elements are either made of isotropic or orthotropic sheet materials
and the critical stress is conveniently expressed as:

Isotropic plate

5.28

„ _k Orthotropic plate
cr hfb?

where Dx and Dy is the bending stiffness of the flange about its x- and
y-axis, respectively. Ex and Ey are the elastic moduli for bending about
the two main axes of the sheet material as if it would be homogeneous
over its thickness, that is the moduli usually given in various stand­
ards and test reports for timber based sheet materials. The buckling
coefficient kaiso = 4,0 applies to long isotropic plates. For orthotropic
plates the buckling coefficient kaorth depends on two parameters:

138 Design of timber structures - Volume 1

 5.1 Structural elements
with full composite action

D
and j82 =— 5.29

where G is the shear modulus. But, for long plates (l > 2bf) that are
only loaded by uniform compression in their longitudinal direction,
the value of ft1 does not matter and a fairly good approximation of
the buckling coefficient is:

JL
°’nHh
orttl
= 2,0 + 2,0 P2
& = 2,0 + JeTe 5.30
V x ;
By using kaiso = 4,0 and kGoiih from equation 5.30 in equation 5.28,
page 138, and substituting the obtained critical stresses into equa­
tion 5.27, page 138, the reduction factor for the effective width is
a constant times hf/b f and the effective width itself is just a constant
x hf. Obviously the constant is a unique number for every sheet
material on the market. In table 5.2 the results of such calculations
are summarised for some common sheet materials together with
the effective widths recommended in Eurocode 5. It appears that
the precision of the Eurocode 5 constant is rather poor and too much
on the unsafe side for OSB, particleboard and fibreboard. The reason
for this difference is that the design curve is taken as 1,0 /Ac rather
than 0,8/Ac for the Eurocode 5 constant, which is definitely too opti­
mistic if Ac < 1,30 and the unfavourable effect of flange curling is not
considered; or if the favourable effect of shear lag is not implicitly
taken into account.
Following Eurocode 5, the effective width should be taken as
the smallest width obtained either from the Eurocode column of
table 5.1, page 137, or table 5.2 for compression flanges and only from
table 5.1 for tension flanges. In addition, the influence of flange
curling is not even mentioned. The main advantage of this approach
is its simplicity, but its precision might be very poor. It is pointless to
perform complicated and tedious analyses of SSP-elements with

Table 5.2 Effective width due to buckling of compression flanges according to Eurocode 5 compared to values obtained
by equations 5.30 5.28 5.27
, and applied to 139 different board products for which material properties are available.
Effective width = const x hf. The two rightmost columns show the effective width (const x hw) in case of shear buckling,
as described in section 5.1.1.3. Minima and maxima obtained for different products in each category are given.

Board material 1) 2) 3) Eurocode 5 Compression 4) Shear

const Min const Max const Min const Max const
OSB, // 25 19 21 32 36

OSB, ± 25 21 23 40 45

Plywood, 3, // 20 12 15 22 27

Plywood, 3, ± 25 17 22 44 61

Plywood, // 20 17 24 35 63

Plywood, ± 25 19 32 44 73

Particleboard 30 23 26 35 40

Fibreboard 30 20 28 28 44

1) // The flange major axis of bending oriented transversely to the SSP-element, for example face veneers of plywood oriented parallel to the webs.
2) ± The flange major axis of bending oriented longitudinally along the SSP-element, for example face veneers of plywood oriented

perpendicular to the webs.

3) 3 Denotes plywood made of only 3 veneers.
4) It is assumed that the compression flange governs the resistance such that afc = f.0 d or f,90 d, as appropriate.

Design of timber structures - Volume 1 139

5.1 Structural elements
with full composite action

regard to the creep properties and moduli of elasticity of the materials

making up the cross-section if the effective width can be off target by
as much as 50 %, while the influence of creep gives a variation of
usually no more than 10 % in the calculated bending resistance.
Instead, the bending resistance can be more precisely targeted by
following the recipe:
1. Calculate, with regard to shear lag, the effective width of both
the tension and compression flanges through equation 5.26,
page 137, using coefficients from table 5.1, page 137.
2. Reduce, if relevant, the effective width of the compression flange
further with regard to buckling according to equation 5.27, page 138.
In equation 5.27: use equation 5.28, page 138, to find the critical
stress based on buckling coefficients koiso = 4,0 and/or kGoiih from
equation 5.30, page 139, and set o£c = (f^d or f;,90,d) x bef-she^ lag/bp as
appropriate. If after this reduction the resistance is still governed
by the tension flange, of,c can be reduced even further to reflect
the “real” stress level in the compression flange. But note that
such an extra reduction is iterative and usually quite impractical
for ordinary hand calculations, especially since the effect on
the final result is quite small.
3. Calculate the position of the neutral axis and the second area
moment of inertia of the effective and transformed cross-section.
Determine the influence of flange curling according to the proce­
dure given below. Move each flange for which flange curling must
be considered towards the neutral axis by a distance estimated as
a fraction of umax according to equation 5.31. Recalculate the posi­
tion of the neutral axis and the second area moment of inertia
with regard to the new flange positions.
4. Check the bending resistance according to equation 5.25, page 135.

Flange curling is a problem for thin and wide flanges and affects both
tension and compression flanges. The axial bending stresses in
the flanges will always have a very small component acting towards
the neutral axis of the SSP-element, which is a consequence of
the bending curvature. It is equivalent to a distributed load applied
to the flange surface, causing transverse bending of the flange. This
results in a reduced distance between the flanges, which in turn
reduces the second area moment and increase the axial flange stresses.
The maximum transverse bending deflection of the flange in between
two neighbouring web panels can be estimated from elementary beam
bending theory applied to a flange strip supported on a number of
webs. Following the approach taken in StBk-N5 (1980) the maximum
flange deflection can be estimated to:

5.31

where of,mean is the average axial stress in the flange calculated as (of,c,d
or o) x beff/ bf, that is the average stress with respect to shear lag and
buckling. Ef is the modulus of elasticity with respect to flange bending
about an axis parallel to the webs. Set Ef to Emean under instantaneous
conditions in service class 1, 0,8 Emean in service class 2 and 0,4 Emean for
plywood that can be used in service class 3. Set Ef to Emean/(1 + ^2 k f)
under final conditions with ^2 determined for the load causing
the largest stress. zf is the absolute value of the distance from the neu­
tral axis to the centre of the flange. K is a modification factor taking
the support conditions of the flange into account and is given below.

140 Design of timber structures - Volume 1

 5.1 Structural elements
with full composite action

If umax is greater than 5 % of the distance between flange centres it is

recommended that flange curling is considered by moving the flange
a fictitious distance, that is a fraction of umax, towards the neutral
axis. The recommended value of K and its corresponding fictitious
distance is:
K = 0,15 if the flange is simply supported on two neighbouring
webs. If needed move the flange 0,64 umax.

K = 0,065 if the flange is continues over intermediate webs and

supported by at least three equally spaced webs.
If needed move the flange 0,58 umax.

Note that the above recipe for finding the moment resistance is appli­
cable to SSP-elements for which the transverse load enters into
the element right on top of the webs. No external transverse loading
can be applied directly to the flanges in between the webs, which
would cause severe transverse bending of them. In such cases, like in
floors and outer roof surfaces, the flange must first be designed such
that it can carry this transverse bending from distributed and/or con­
centrated forces. After that the recipe can be applied, but the flange
slenderness is then usually so small that only shear lag needs
attention.

5.1.2.2 Bending and shear deflections

The bending deflection is determined exactly as for full composite
I-beams as discussed in section 5.1.1.5, page 132. The shear deflection is
usually quite small and is negligible for elements having rather
stocky webs of structural timber, glulam, etc. If, however, the webs
are made of thin sheet materials the shear deflection should be calcu­
lated in the same way as described for I-beams in section 5.1.1.5. Note
that effective flange widths need only be determined with respect to
shear lag when calculating deflections in the SLS. Flange buckling
and flange curling need not be considered, because the stress levels
are much smaller in the SLS than in the ULS, making the effect of
these phenomena negligible, at least for normal SSP-elements.

5.1.2.3 Advantages and disadvantages

with T- and H-shaped SSP-elements
Here some statical as well as practical advantages and disadvantages
of SSP-elements are discussed.
The most difficult problem to solve for SSP-elements is the continu­
ity of a thin tension flange. Wood based sheeting are usually available
in lengths much smaller than the element length and must therefore
be jointed in regions transferring large tensile stresses, which is pos­
sible but complicated and expensive. One solution is to use LVL or
steel sheeting for the tension flange and thereby avoiding unwanted
joints.

Some advantages of T-sections compared to H-sections are:

• Easy to add extra local reinforcement to secure load transfer of
large concentrated forces that may otherwise cause local damage.
• Simpler to connect the elements to each other and to the supports.
• Easier to install and hide installations within the elements, when
carried out on site.
• Combining two similar elements with limited or no contact in
between them may result in very good sound insulation proper­
ties, which may also be good for fire protection.

Design of timber structures - Volume 1 141

5.2 Structural elements with
partial composite action

Two advantages of H-sections compared to T-sections are:

• Better bending resistance and bending stiffness, which makes it
possible to span greater distances without using much more
material.
• A higher degree of prefabrication is possible, avoiding costly com­
pletions on site. Execution quality of insulation and installation
works will be better due to more suitable factory conditions
regarding tools, climate and control.

5.2 Structural elements with

partial composite action
In timber engineering built-up cross-sections are quite common.
The parts are often connected by glue, in which case the slip between
the parts may not be much of a trouble. But slipping between mem­
ber parts must be considered when mechanical connectors are used
instead of glue or when the quality of the glue-line cannot be relied
upon. In figure 5.19 are shown some different beam structures having
built-up cross-sections for which partial composite action may play
a substantial role.
Suppose that we have a beam or SSP-element subjected to bending
and that it is made of two or more individual parts connected to each
other by some mechanical connectors, soft and thick glue interfaces
or some rather thick interlayer that is only capable of transferring
shear stresses but no direct stress (such as the core of a sandwich ele­
ment). By partial composite action is then meant that the shear
deformation between separate parts is non-negligible. This shear
deformation is hereafter also referred to as slip deformation or sim­
ply slip. The principal behaviour of an element having only two parts
connected by one shear transferring interlayer is depicted in
figure 5.20.
The response is characterised by some slip of the intermediate
layer, which may be categorised into three different cases. First, we
Figure 5.19 Typical composite beam elements for which have full composite action (FCA) if the shear connection is infinitely
partial composite action is relevant stiff. Second, we have no composite action (NCA) if the shear con­
nection has no stiffness at all, that is the two layers act independently
of each other and no shear forces are transferred along the joint line.

Full composite action

FCA
Strain

No composite action Partial composite action

NCA PCA
Strain Strain

Figure 5.20 Three degrees of composite action, for equal midspan deflections

142 Design of timber structures - Volume 1

 5.2 Structural elements with
partial composite action

Third, we have partial composite action (PCA) if the shear connec­

tion has a finite stiffness, that is the “true” response that must be in
between the first two limiting cases. The first two cases are well cov­
ered by traditional beam bending theory and will not be treated any
further in this section. Here, we shall focus on the third case charac­
terised by partial composite action.
The partial composite action is governed by either of the differen­
tial equations 5.50 or 5.51, page 146-147. Equation 5.50 was first derived
by Stussi (1947), but he had no term to account for the temperature
difference between the two layers. Equation 5.51 was derived by
Granholm (1949), also neglecting the influence of temperature differ­
ences. Newmark (1951), derived equation 5.50 independently of
the first two authors. After these pioneers many others have followed
publishing minor or major modifications to the basic equations given
here. Such modifications include the influence of an external axial
load, creep, more than two layers, non-linearities or de-bonding.
Also, many contributions deal with different solution procedures like
symbolic closed form solutions, finite difference methods and finite
element methods. Only the basic equations are needed to solve
the problem treated in this book. The other mentioned references
will be left without any further treatment.
As a curiosity it is worth to mention that researchers working in
this field may be divided into two main categories. The first category
is those working with composite beams connected by mechanical
connectors like nails, screws, studs and also glue. The second group is
concerned with all kinds of sandwich beams where the layers are
connected by some kind of foam or honeycomb, for example
expanded polystyrene (EPS). But the basic equations are the same no
matter which group the researcher belongs to. Despite this the two
groups almost never reference each other’s work. The authors refer­
enced here belong to the first group.

5.2.1 Built-up bending elements

We will start by deriving the governing equations for a two layered
composite beam subjected to bending only. We will then go on to
show certain simplifications of the analysis, making it possible to
obtain approximate solutions with relative ease. Then the equations
are modified such that three layered double symmetric sections can
be properly analysed. The equations of a single symmetric three lay­
ered beam with unequal parts are not treated here, but are given in
Oduyemi and Wright (1991).

5.2.1.1 Basic assumptions

Consider, as in figure 5.21, page 144, an infinitesimal segment cut out
along the longitudinal axis of any element. The differential equations
5.50 and 5.51, page 146 - 147, govern the response of this small piece,
when subjected to external loading. These equations are derived
using the following assumptions and limitations:
a) All material remains linear elastic.
b) Small displacement theory is assumed.
c) The two layers must have equal deflections, i.e. they must not
separate and the model does not consider any tensile or compres­
sive stresses in the thickness direction. Furthermore, both layers
are assumed to have equal radius of curvature, which is justified
by the fact that the thickness of the element is much less than
the radius of curvature.

Design of timber structures - Volume 1 143

5.2 Structural elements with
partial composite action

d) No shear deformation within each layer is considered, only the shear

deformation of the core is accounted for. Also, any influence of straining
perpendicular to the longitudinal axis is neglected. This implies that
the assumption in normal beam bending theory, that plane cross-sections
remain plain, holds for each layer.
e) The core is only assumed to have shear stiffness and its only purpose is to
act as a shear connection between the two layers.
f) The element may also be subjected to a uniform strain difference between
the two layers, for instance caused by temperature differences between
the two layers.
g) In addition to the strain difference the only allowed load is a bending
moment caused by some external load acting in the z-direction. External
axial loads are not accounted for.
h) The model can only account for deflections within the xz-plane and
the cross-section must, therefore, be symmetric about the z-axis.
A

Core

Original
configuration

Displaced configuration

x
Figure 5.21 The beam cross-section and a small cut-out d along the beam axis
In the figure the following notations are used (other symbols are explained as they are introduced):

A1 Cross-sectional area of layer 1 I2 Second moment of inertia, for the plane

Cross-sectional area of layer 2 cross-sectional area of layer 2
A2
b Width of the core material M1 = M1 (x) Internal bending moment in layer 1
E Modulus of elasticity, which is assumed to be equal M2 = M2 (x) Internal bending moment in layer 2
for both layers. If equality does not hold the method of M = M (x) External bending moment affecting
transformed cross-sections can be used the entire cross-section
Y = Y (x) Shear angle of the core N1 = N1 (x) Internal axial force in layer 1
G Shear modulus of the core material N2 = N2 (x) Internal axial force in layer 2
h core Core thickness q = q (x) Transverse loading
ht Distance between the centres of layer 1 and 2 u = u (x) Shear deformation of the core
H = H (x) Shear flow transmitted, by the core, u1 = u1(x) Displacement of the centre of layer 1
between the two layers u2 = u2 (x) Displacement of the centre of layer 2
I1 Second moment of inertia, for the plane w = w (x) Deflection
cross-sectional area of layer 1
x, y, z Co-ordinate axes

144 Design of timber structures - Volume 1

 5.2 Structural elements with
partial composite action

The directions of the arrows in figure 5.21, page 144, indicate positive
forces, moments and displacements. The prime symbol, ', indicates
differentiation of the above functions with respect to x, two prime
symbols indicate that the differentiation is performed twice, etc.

5.2.1.2 Derivation of the governing differential equations

for a two layered beam
From figure 5.21, page 144, it is clear that the deformation of the core is:

M = «2 - M1 + Atw' 5.32

and after differentiation:

u' = u2-u{ + htw" = e2-e1 + htw‘ 5.33

where:
du.
£. = —1
1 dx

du.
E =--- —
2 dx

are the simple engineering strains along layer centres 1 and 2, respec­
tively. We have also assumed both layers to have the same deflection
w(x) and approximately the same radius of curvature. The strains
may, by using traditional bar theory and positive tensile strains, be
expressed as:

£. =---- L + St. 5.34

1 E\ u

N2
£.2 =-- — + £tt2, 5.35

where st1 and st2 are uniform strains caused by temperature changes
in layers 1 and 2, respectively. By substituting equations 5.34 and 5.35
into equation 5.33 we obtain:

5 ' = htw-------L +---- —6Et.~Et. 5.36

‘ E^ ‘2 u

Now, since both st1 and st2 are constant along the beam the difference
between them is replaced by one constant:

£t = et2 “ £tl 5.37

Force equilibrium for the cross-section shown in figure 5.21, page 144,
requires that:
N3 = -TVj 5.38

because there is no external axial load affecting the beam.

Substitution of equation 5.37 and 5.38 into equation 5.36 yields:

m' = Aw"-
fl 1
------ +-------- ^+£t 5.39
‘ 1^4 ea^ 1 ‘

Design of timber structures - Volume 1 145

5.2 Structural elements with
partial composite action

Let now the material properties of the core material enter into this
derivation. When the shear flow H affects the infinitesimal cut-out in
figure 5.21, page 144, it is displaced an angle that may be written as:

Shear stress
y =----------------- H
=— 5.40
Shear modulus G b

and by using equation 5.40 the resulting displacement is:

H -N'
5.41
Gb
where the shear stiffness of the core, kcore, has been introduced:
* ""
core i
“core

In equation 5.41 it is also used that the force equilibrium of layer 1, of

the infinitesimal cut-out in figure 5.21, page 144, requires that H = -N'1.
Differentiating equation 5.41 once and equating the result to equa­
tion 5.39, page 145, gives:

1
kcore
= Atw"- A + £i 5.43

Equating the external moment M and the internal moments in

the cross-section, taken about the centre of layer 2, yields:

M4 M,+M,-.X,h. 5.44

The assumption that both layers have equal radius of curvature

gives:

,.__A£ m2
5.45
£A ei2

and equation 5.44 may by using equation 5.45 be rewritten as:

5.46
EIy + EI2 EIr + EI2

Substituting equation 5.46 into equation 5.43 and rearranging gives:

1 l 1 t A2 ' k h
^core
N. = —5-^r M - k et 5.47
1 E A A A+A> 1 E (Zj + ^J core ‘

Introducing the cross-sectional constants:

q _
kcore h
5.48
■E

2 k ( 1 1 A2
£02 = _c»e _L + _L +---- 1--- 5.49
e A A A+A,

makes it possible to write equation 5.47 in its simplest form:

1 1 = CM-kcore ett
N'-o^N. 5.50

146 Design of timber structures - Volume 1

 5.2 Structural elements with
partial composite action

This is a second order linear differential equation with constant coef­

ficients that may be solved easily once the boundary conditions of
the problem are known. In equation 5.50 the unknown function rep­
resents the variation of the axial force along layer 1. Sometimes it
may, however, be more convenient to express the governing differen­
tial equation in terms of the deflection function. This may be
achieved by solving for N1 in equation 5.46, page 146, and substituting
the result and its derivatives into equation 5.43, page 146. The result,
after introducing two new crosssectional constants, is:

5.51

where:

5.52

5.53

When explicitly solving a problem it is often slightly simpler to work

with equation 5.50, page 146, than equation 5.51. But sometimes it is
convenient to express the boundary conditions in terms of known
deflections and sometimes you just want to know the deflections. So
if one solves for N1 in equation 5.50 it is necessary to have an expres­
sion that gives the deflection w once N1 is known. This link between
the deflection function and the axial force function may be derived
by solving for N1 in equation 5.50 and substituting the result into equa­
tion 5.46, page 146. This first step leads to:

f n" C k e
M+h 5.54
jo2 co2 co2

After rearranging terms and introducing the constant defined by

equation 5.52, we have:

5.55

Where the factor in front of M may be rewritten as:

5.56

-1 A! + A2

-1 1-1
E J +1 + ^fca
1 2 4 + ^2

Design of timber structures - Volume 1 147

5.2 Structural elements with
partial composite action

After a number of algebraic manipulations it is clear that the denom­

inator at the final stage in equation 5.56 is equal to the bending stiff­
ness for the beam in case of full composite action, where it may be
easily verified that:

5.57

is the second area moment of inertia for the cross-section when full
composite action prevails. Using the result from equation 5.56 in equa­
tion 5.55, page 147, gives:
„ -M C
w =------------------ 5.58
EI{ica core

Finally, by integrating equation 5.58 twice, the link between the deflec­
tion function and the axial force function may be expressed as:

W=Wfca N,1 + k„£

core tt—
2 + D3x + D4 5.59
^core®2

where wfca is the deflection in case of full composite action, that is no

shear deformation of the core material occurs. Note, also that
the term wfca is computed with regard to the moment distribution
only and that solutions to such elementary cases often exist in the lit­
erature. Coefficients D3 and D4 are integration constants that must be
determined using appropriate boundary conditions. For many practi­
cal cases both coefficients may often be zero. It follows directly from
equation 5.59 that D4 = 0 when w = wfca = N1 = 0 for the same x-coordi-
nate and that D3 = 0 when w' = w'fca = N1' = 0 for some other
x-coordinate.
To compute stresses within each layer it is necessary to know M1
and M2. Expressions for these quantities can be obtained by combin­
ing equations 5.44, page 146, and 5.45, page 146, which yields:

M.------ 5.60
1 I'+I^ 1 t7

M,-—(M + jV.A) 5.61

2 T■'1+T \
+ 12
1

5.2.1.3 Solution procedure

In order to find a function N1(x) satisfying equation 5.50 it is necessary
to have an expression for the moment distribution M(x) along
the beam in question. Here it will be demonstrated how to find
a solution for a beam subjected to transverse loading causing a bend­
ing moment distribution that can be expressed as a polynomial of
order four. Of course higher orders are possible, but are quite rare in
practical applications. Note that moment distributions expressed as
polynomials are by far the most common type of loading assumed in
classic beam bending theory. Besides from this, the beam may have
quite general boundary conditions. The moment distribution along
the beam may be written as:
4
M(x) = ^ajX! 5.62

i=0

148 Design of timber structures - Volume 1

 5.2 Structural elements with
partial composite action

where the ai’s are known coefficients for statically determinate

beams and for statically indeterminate beams one or more of them
must be expressed in terms of some indeterminate quantity, for
example an unknown support reaction or bending moment at a built
in support. Note that the beam may have to be divided into more
than one interval along the x-axis, that is the function representing
the moment distribution may in general be different for each inter­
val depending on the actual loading and boundary conditions. Here
just one interval having a continuous moment distribution is
assumed. Substituting equation 5.62 into 5.50, page 146, gives:

( k E
N'.'~(D
1 2N.=C
1 n —0016 1 + a.x + a~x2 + o,x3 + a.x4
a0/^1234 5.63

This second order differential equation is linear and has constant

coefficients and may be solved using standard mathematical proce­
dures given in most schoolbooks on elementary differential equa­
tions. The total solution is the sum of the homogeneous solution and
one particular solution:

5.64

The homogeneous solution is:

.V111(x) = Z>1cosh(fflx) + Z>2sinh((ox) 5.65

where the constants D1 and D2 must be determined with regard to

the boundary conditions of the problem. The particular solution
must be a fourth order polynomial because the right part of equa­
tion 5.63 is such a polynomial. By substituting a polynomial trial
solution and its second order derivative expressed as:

M,P(x)=Xz’i%! 311(1 Mj(x)=Xz’(z’-1)z,ixI 2 5.66

’ i=0 ’ i=2

into equation 5.63 and identifying terms a linear equations system in

the unknown bi’s are set up. A symbolic solution to that system
expresses the unknown bi’s of the particular solution in the known
ai’s, which together with equation 5.65 are substituted back into equa­
tion 5.64 yielding:

Nr (x) = Dr cosh ((Ox) + D2 sinh((0x) -

^core£t c 2
+ xax + x2
C
ao
c I ®2> ‘2
5.67
co2 6xy 12x2 24
+ 3
+
»2> o4

Furthermore, it is usually necessary to have an expression for

the deflection function, which is obtained by substituting equa­
tion 5.67 into equation 5.59 resulting in:

Design of timber structures - Volume 1 149

5.2 Structural elements with
partial composite action

+ DyX + —

D. coshl cox) + IX sinh( (ox) + k w £.

1 \ / 2 \ / core t

c 5.68
^core®2

Note that the principal of superposition holds because linear elastic

material is assumed, that is our differential equation is linear.
Therefore, solutions to elementary cases may be superimposed in
exactly the same way as in classic beam bending theory. If you have
a library of solutions to some elementary cases they can be combined
in numerous ways to find solutions to new elementary cases, rather
than explicitly solving equations 5.50 and/or 5.51, page 146 -147, possi­
bly with the aid of equations 5.67, page 149, and/or 5.68.
One consequence that immediately follows is that it would not
have been necessary to carry all terms including st through equa­
tions 5.50 to 5.68. If in equation 5.50 simply setting M = 0 as if no
external bending moment is present, a new elementary case only
dealing with st is formed. Its solution can then be added to any other
case as needed. But now the easiest way to find the solution to this
case is to recognise that setting M = 0 is equivalent to setting all ai’s
to 0 in equations 5.67 and 5.68, after which the coefficients D1, D2, D3
and D4 are determined with respect to actual boundary conditions.
It will now be demonstrated how to find the complete solution to
one elementary case of practical interest. The case of a simply sup­
ported beam subjected to a concentrated load P acting at mid-span is
chosen and it is also assumed that st = 0. The total span is l and
the moment distribution has a discontinuous derivative at mid-span,
at 1/2. It is therefore necessary to divide the beam into two parts, one
for 0 < x < 1/2 and one for 1/2 < x < l. But because of symmetry it
suffices to find a solution for one of the intervals, preferably 0 < x <
1/2. The bending moment is expressed as M(x) = P/ 2 • x. By compar­
ing to equation 5.62, page 148, we see that a1 = P/2, while a0, a2, a3 and
a4 = 0. The coefficients D1, D2, D3 and D4 of equations 5.67 and 5.68 may
now be determined using some adequate boundary conditions.
The axial force in each layer as well as the deflection must be zero at
x = 0, which may be expressed as N1(0) = 0 and w(0) = 0. These two
conditions give without any calculations that D1 = D4 = 0 in equations
5.67 and 5.68. At mid-span the axial force N1 and the deflection w
have their maximum and their derivatives must therefore be zero,
N'1(1/2) = 0 and w'(1/2) = 0. From equation 5.59, page 148, it follows
that D3 = 0. Finally, equation 5.67 gives:

(o£ CP
= Z>2®cosh ft)2 2 ° d2 =
ft)l
2 G)3 cosh 5.69
~T
The total solution is obtained by substituting the results of equa­
tion 5.69 back into equations 5.67 and 5.68, and after some rearranging
of terms and developing of the deflection function (wfca for full com­
posite action) we arrive at:

150 Design of timber structures - Volume 1

 5.2 Structural elements with
partial composite action

sinh(fi)x)
-(OX
, f tot'
cosh —
I 2 ,

lx
5.70
8ELtea 2

sinh(<ax
c2 ------- 7---- x— U'-A
J 1
< cosh I — 2 J ,

One can of course find solutions to other problems, being loaded dif­
ferently and having other boundary conditions, by using the same
principal approach. A summarising recipe may look like:

a) Divide if necessary the entire beam into a number of intervals for

which the moment distribution is continuous, that is has a con­
tinuous first order derivative or in other words can be written as
one single function expression. Note also that all cross-sectional
constants must not change along each chosen interval.
b) For each interval identified under point (a), express the moment
distribution as a polynomial and identify the ai’s in equations 5.62,
page 148, 5.67, page 149, and 5.68, page 150. Note also that if
the beam is statically indeterminate one or more of the ai’s have
to be expressed in terms of all the statically indeterminate quanti­
ties, such as a reaction force or bending moment.
c) Figure out relevant boundary conditions such that equations 5.67
and 5.68 plus their first and second order derivatives can be used
to find expressions for the coefficients D1, D2, D3 and D4.
Remember that D3 and D4 are often zero due to the structure of
equation 5.59, page 148. One equation is needed for each Di and for
each statically indeterminate quantity. Boundary conditions
securing continuity of equations 5.67, 5.68 and their derivatives
across interval borders are useful if the beam is subdivided into
several intervals. Note also that all Di’s are in general different on
each side of an interval border.
d) Set up all chosen boundary conditions as a system of linear alge­
braic equations and solve it for the unknown Di’s and if present
the statically indeterminate quantities. The solution can be either
symbolic or numeric. A symbolic solution is more versatile and
can be reused over and over again for different input data like
dimensions and stiffness properties, while a numeric solution is
only valid for exactly the same input data as those used for set­
ting up the equation system. Some useful boundary conditions
are: N1 = 0 at a free end whether or not the beam is supported or
free to deflect at this end; N'1 = 0 at a position where no slip can
occur; w = 0 at a pinned support; w' = 0 at a built in end.

The complete symbolic solutions to a large number of practically use­

ful elementary cases are given in Elsander (1999). Here, the solution
to just one more case is given, which is a simply supported beam sub­
jected to uniform loading q.

Design of timber structures - Volume 1 151

5.2 Structural elements with
partial composite action

Shear flow along the interface

eee fca
------ PCA
••• NCA

x-coordinate / m

Figure 5.22 Partial composite action in comparison to the upper and lower limits

The axial force in part 1 is:

/ z X 2
M (x)= “4" Jsmh(t»x) - cosh (tax) + “(x2 -j +1 5.71

It is also possible to use analogies in order to solve the differential

equation 5.50, page 146, because other physical phenomena might be
controlled by a differential equation of exactly the same mathemati­
cal form. For instance, the equation controlling mixed torsion (that is
Saint Venant torsion in combination with warping torsion) has
an equivalent mathematical form. Also, the equation controlling
deflection of an elastic bar subjected to transverse loading in combi­
nation with axial tension is equivalent in shape.
As already mentioned, the principle of superposition holds because
small displacements and linear elastic material are assumed.
Therefore, solutions to elementary cases may be superimposed in
exactly the same way as in classic beam bending theory.
By using equation 5.70, page 151, some graphics presenting the basic
difference in results between full composite action (fca), partial com­
posite action (pca) and no composite action (nca) have been derived
and are presented in figure 5.22. Parameters used are: l = 2,40 m,
P = 5,00 kN, C = 6,49 X 10-9 mm-3, rn = 9,50 x 10-4 mm-1, E = 11,0 GPa
and Ifca = 3,90 x 107 mm4. These values are representative for
a T-section made of two joists of C24-timber having dimensions
45 X 45 mm (turned edgeways) and 45 X 120 mm (turned flatways).
These joists are nailed together using one row of 120 X 3,8 mm
smooth grooved nails with a spacing of s = 60 mm. Instantaneous
conditions have been assumed.

5.2.1.4 Relations between the slip modulus of mechanical

connectors and shear stiffness of the interlayer core
The governing equations have been derived assuming the presence of
an interlayer core material characterised by a shear stiffness as given
by equation 5.42, page 146. But all derived equations are also valid for
mechanically connected beams like nailed timber beams if the spacing
between the discrete connectors is small enough. Here, small enough
means that the stiffness of the connectors can be assumed to be
smeared out along the beam axis as if the connection is continuous.

152 Design of timber structures - Volume 1

 5.2 Structural elements with
partial composite action

For mechanically connected beams a different shear stiffness must be

used, so that equation 5.42, page 146, must be replaced by the shear
stiffness of the mechanical connectors, which may be computed as:

, kn
k =— 5.72
s
where k is the slip modulus of a single connector (for instance Kser or
Ku for a timber structure designed according to Eurocode 5), s is
the constant spacing between the connectors in the longitudinal
direction and n is the number of connectors placed on a line perpen­
dicular to the longitudinal direction. Suppose that two (or more) dif­
ferent types of connections connect the two layers, for example a shear
flexible core of thickness hcore and mechanical connectors with slip
modulus k and spacing s. One must then distinguish between parallel
or series arrangement of the different connections. In the parallel
case both types of connectors connect the two layers independently
of each other, for example mechanical connectors that go right
through a core material. In the series case the shear flow is trans­
ferred from one of the layers into the first type of connection then
into the next, etc., until the second layer is reached. An example is
a shear flexible core that is connected to each layer by mechanical
connectors that do not fully penetrate the core material, but have
the same properties at each side of the core. For the discussed exam­
ples the stiffness in the parallel case may be written as:

, kn Gb
k<°
-~
* +r~ 5.73
core
and for the series case it may be expressed as:
1
5.74
2 S ^core
kn Gb
Equations 5.73 and 5.74 are merely examples and must be properly
adjusted in each particular case. For more complicated interlayer
connections, as in latticed girders and girders with packs or gussets
(Vierendeel type), expressions for kcore are for instance given in Larsen
and Riberholt (2005).
In connection to equations 5.40 to 5.43, page 146, is explained that
the shear flow along the interface is the first derivative of the axial
force in layer 1, i.e. H = -N'1. The shear force acting on an individual
connector is therefore:
£
57x) = -N!(x)— or 57x) = ku(x) 5.75
n
and from equation 5.41, page 146, the slip along the interface is
expressed as:

m(x) = 5.766^ or «(x) = -#'(x)-^- = 57

)
* 5.76
k_ . kn k

5.2.1.5 Approximate solutions of

the governing differential equations
Equations 5.50 or 5.51, page 146 -147, may be solved approximately. In
many practical applications the approximate solution is good
enough, and indeed it forms the basis for a very practical method of

Design of timber structures - Volume 1 153

5.2 Structural elements with
partial composite action

finding the flexural buckling resistance of built-up columns, see

Eurocode 5 or Larsen and Riberholt (2005).
In short the approximate solution is based on the exact solution to
an elementary case in which a simply supported beam is loaded by
a sinusoidally distributed load, that is half a sinewave. The load and
bending moment is expressed as:

f \ • | 7£X I
?w=^osm — ; M" = -q gives 5.77

where q0 is the maximum amplitude of the distributed load. By sub­

stituting the bending moment into equation 5.50, page 146, and set-
ting st = 0 we obtain:
t_ fix'
N"-(O2N1 = Cq0 sin 5.78
n k
It can now be easily verified that D1 and D2 in the homogeneous solu­
tion must be zero, meaning that the particular solution is the entire
solution. By attempting a sinusoidally shaped particular solution
the total solution becomes:

Tlx
sm 5.79
~T

By substituting equation 5.79 into equation 5.59, page 148, noting that
D3 = D4 = st = 0 and topping off with some algebraic manipulations
the deflection function is found to be:

4 1 +------ -
?o^4
w(x) = /dmfSin 5.80
^4^4a 1+P Y
+ l at J

During the algebraic rearranging the second area moment in case of

full composite action Ifca and the constant CM from equation 5.53,
page 147, have been introduced. From equation 5.80 it is clear that ydmf
represents a displacement magnification factor, which multiplied by
the fca-displacement gives the “correct” displacement accounting for
interlayer slip. Another way of understanding equation 5.80 is to see:

EIf = EIf —!— 5.81

ef fca v
• dmf
as an effective bending stiffness of the cross-section, which is con­
stant along the entire beam. Note that this is the only elementary
case for which the bending stiffness can be viewed as constant along
the beam. For all other cases the hyperbolic terms from the homoge­
neous solution makes such a simple interpretation impossible. But
we can expect that equations 5.79 to 5.81 give a very good approxi­
mate solution in cases where a beam is supported and loaded such
that the external bending moment distribution is similar to half
a sine-wave. The method for designing built-up beams and columns,
characterised by partial composite action, in Eurocode 5 depends
entirely on this simplified solution. In Eurocode 5 the effective bend­
ing stiffness in equation 5.81 is expressed differently, but is really
the same thing.

154 Design of timber structures - Volume 1

 5.2 Structural elements with
partial composite action

5.2.1.6 Three layered double symmetric beams

The characteristics of a three layered beam are shown in figure 5.23.
Note that even the properties and spacing of the fasteners connecting
parts 1 and 2 must be the same as for those connecting parts 2 and 3.
Also, the strain difference st between the layers are left out, because
it is rare to have a symmetric influence from this kind of effect.
It is for a double symmetric three layered beam actually possible to
use most of the previously defined equations. By going through
the derivation phase as explained by equations 5.32 to 5.51, it can be
shown that the differential equations 5.50 and 5.51, page 146 -147,
does not change. It is just some of the cross-sectional constants
together with a few of the equations that need to be modified.
The cross-sectional constants become:

C_kcore *t
valid for DS3L

valid for DS3L

£ 2(2/1 + 72)J
5.82
C , 1 ,
valid for DS3L
je(271 + Z2)
Q —
kcore
valid for DS3L
M EAX

where each equation is labelled with DS3L, which means “double

symmetric 3-layered”. This labelling is just used to avoid mixing up
with the two layered case. The equations that need adjustment are
the bending moment in each part (Equations 5.60 and 5.61, page 148),
which change to:

A1, I1
Undeformed configuration

h1

h2

h1 = h3

A3 = A1
I3 = I1

E3 = E1

N1= - N3

Figure 5.23 A piece of a three layered double symmetric beam

Design of timber structures - Volume 1 155

5.2 Structural elements with
partial composite action

M = Mi. =---- —---- (jW + .V./l.) valid for DS3L

1 3 2/j + z/ 1 t;
I 5.83

Md va----- v----- 1 va + va A ) valid for DS3L

2 2^ + 7/ 1

and the deflection given by equation 5.59 changes into:

2C
W=Wfca------------ -jVj vDfXD3^ valid for DS3L 5.84
Aore®

The more general single symmetric three layered case is more com­
plicated. A detailed derivation will result in a system of differential
equations, where each equation is similar to the one given in equa­
tion 5.50, page 146. In general one equation is needed for each inter­
face between the parts, that is two equations for a 3 layered beam,
three equations for a 4 layered beam, etc. Equations for the 3 layered
case can for instance be found in Oduyemi and Wright (1991).

5.2.2 Summary of equations

A summary is made in table 5.3 of the most relevant equations to be
used for 2 and 3 layered beams, including formulas for connector
force and interface slip. Note that the 3 layered beam must be double
symmetric. Note also that the st-term has been left out.

Table 5.3 Summary of equations for 2 layered and 3 layered double symmetric beams. No st-term is present.

Explanations Equations common for 2 and 3 layered beams

Differential equation in N1 #1"-®2#1 = CAf

Differential equation in w w"" - cd2 w"-cdO-M-M"]

Mb \ M /
V
Connector force
F(x) = -N'Ax)— or F(x) = k • u(x)
n

Interface slip
m(x) = ~~rL— or M(x)= “-^1 (x) A— = —1—
£core k

Two layers Three layers, double symmetric

Relation between c 2C
N1 and w w-wfca , 2Ni + D3x + D4 W-Wfca , 2-V1+Z>3X + D4
Acre®
Constants k A k A
Q— core Q— core
’E ( A + A 2E | 2r,+r,
(1 f
n _ kcore
—+ ±+_?L' g,2 = Arae 1
U + 2 (2Z1+A)J
*,2 1
E ■^2 A+A j E

c i 1
G+A, 2Z1 + A)

k r i _
k
con
(J _ core
M E IV M EA
"A,
Bending moment in
M.------ M + N,ht) M. = M,-------—(M + N. h\
parts 1 and 2 1 W 1 3 2^ + 1^ 1

M,----- (M+N,ht M.------ —(M + N.h,}

2 A+A 2 2^ + Z/ 1

156 Design of timber structures - Volume 1

 5.2 Structural elements with
partial composite action

Equations 5.72 to 5.74, page 153, give possible relations between kcore
and k for mechanically connected parts or layers. Equations 5.70,
page 151, 5.71, page 152, and equations 5.79 to 5.81, page 154, give solu­
tions to 3 different elementary cases of practical interest.

5.2.3 Failure criteria for built-up beams

5.2.3.1 Recommendations regarding
the stiffness properties to be used
For ordinary beams designed under the assumption of full composite
action we need only to account for the difference in elastic modulus
and creep properties of the parts making up the cross-section, as
explained in section 5.1.1.1, page 118. It is recommended to perform
the analysis for partial composite action in the same way, that is
using a transformed cross-section such that all parts can be assumed
to have the same modulus of elasticity. It may also be necessary to
make relevant checks at “instantaneous” and “final” conditions in
order to cope with the creep effects.
There is one additional complication for partial composite beams
due to the creep properties of the shear interface. Mechanical con­
nections in timber materials are definitely sensitive to creep. Usually,
the connection is more prone to creep than the timber itself, mainly
due to relatively high stress levels in the small timber volumes sur­
rounding the connector. In Eurocode 5 it is recommended to use
a creep factor kdef for the slip modulus of the connector that is twice
as large as for the timber itself, that is to use 2 kdef. Creep usually
plays a substantial role even if a thick shear interface layer (core) is
used without any mechanical connectors, because such materials are
often sensitive to creep.
Here, we will concentrate on mechanical connectors in built-up
timber beams. Their slip modulus k is in Eurocode 5 characterised by
either Kser or Ku = 2/ 3 x Kser. Kser is the gradient of a straight line going
through a point on the load-slip relationship corresponding to
approximately 40 percent of the connector’s ultimate resistance,
while Ku corresponds to the 70 percent level.
How will the internal forces and stresses change in the composite
beam if the slip modulus increases or decreases? The answers are that
a decreasing slip modulus will give:
• A smaller axial force N1 but the bending moments M1 and M2 will
increase such that the net outcome is greater stress at the outer­
most edge of each layer.
• A smaller connector force F, i.e. the shear flow H = -N'1 is
decreasing and more of the load is carried by individual bending
of the layers. We will have no composite action at all for a slip
modulus equal to zero.
• A greater deflection.

An increasing slip modulus will give the opposite effects to those listed
above. Luckily a large change in the slip modulus will produce a much
smaller change in member forces, member stresses and deflection.
Reducing k to half its value or doubling it will change these values in
the order of 10 %, under typical conditions. But note that the connec­
tor force will typically change by the same order as k.
From the above description it is recommended that Ku is used when
checking resistances in the ultimate limit state (ULS). This holds even
if the weakest part of the beam is the connectors, because then their
slip modulus will be even smaller than Ku for the most loaded (and

Design of timber structures - Volume 1 157

5.2 Structural elements with
partial composite action

failing) connectors, hence subjected to a smaller load than actually

calculated. Therefore, Ku is still a good choice, also noting that many
connectors along the beam are not fully used. Of course, if one really
wants to be on the safe side, Kser might be used when the connectors
are the weakest part of the beam. If the connectors are much
stronger than the timber, then a value closer to Kser may be beneficial
when checking the timber resistance.
In the ULS it may still be necessary to distinguish between “instan­
taneous” and “final” conditions due to the creep properties of the tim­
ber itself. Here the key recommendation is to check the members
most prone to creep at instantaneous conditions, while members less
sensitive to creep are checked at final conditions.
In the serviceability limit state (SLS) it is recommended that Kser is
used when determining all the cross-sectional constants.
Following the recommendations given in Eurocode 5 as close as
possible, the slip modulus and the modulus of elasticity may for
the instantaneous and final conditions be taken from table 5.4.
Note, that the relations of table 5.4 should be used to calculate all
cross-sectional constants of a transformed cross-section when deter­
mining the distribution of internal forces between members of differ­
ent materials having different creep properties. After that the final
SLS deformation is estimated using equation 5.86, page 159, the method
outlined here will slightly overestimate the final creep deformation
in that all the cross-sectional constants are determined using the creep
factor for permanent loading. In the ULS y2 shall be determined for
the load causing the greatest stress, which effectively is a way of cal­
culating a kind of weighted mean value of the cross-sectional con­
stants that will give reasonable estimations of internal forces and
stresses.
Equations 5.59, page 148, and 5.84, page 156, are made up of two
terms: the first is called wfca as it is the deflection in case of full
composite action, while the second is called wslip as it is the additional
deflection caused by the connector slip.
In wfca the modulus of elasticity E is controlling the deflection,
while the slip modulus k (kcore) is controlling the deflection in wslip. Of
course these moduli are also hidden within Ifca, C and rn, but are then
purely relational and determine the distribution of internal forces

Table 5.4 Recommended values for the slip and elastic modulus of composite timber beams
for which mechanical connectors are used and partial composite action prevails

Instantaneous conditions, inst Final conditions, fin

Serviceability limit state, SLS P

■^SLSjfaist
F
— ^(^mean p _ ^0,mean *
*SLS,fc-i ,
det
^SLS.inst = ^scr K *
k — *
SLS’fc 1 + 2L,
det

Ultimate limit state, ULS F

FULS,inst — ^(Xmean p _ ^0,mean
ULS’fc_ uV2kiet

^ULS.inst = *
u
k________
1 + 2^

* A note on determination of final deformations. Cross-sectional constants, that is Ifca, C, rn, CMb and CM of transformed cross-sections,
should be based on ESLS,fin and kSLS,fin for the instantaneous condition before the effect of creep is added. But all moduli beyond these
fictitious cross-sectional constants should keep their instantaneous values ESLS,inst = E0,mean and kSLS,inst = Kser, because they will at a later
stage be multiplied by either (1 + kdef) or (1 + 2 kdef) when going from instantaneous to final deformation as given by equation 5.86.

158 Design of timber structures - Volume 1

 5.2 Structural elements with
partial composite action

between the members. Suppose now that we have permanent loading

Q
denoted G and variable loading denoted 1 for the leading variable
Q
load and i for the other variable loads, i is 2, 3, etc. Following
the recommendations in Eurocode 5 to use the characteristic
combination in Eurocode 0 for determining SLS-displacements,
the instantaneous deflection may formally be expressed as:

Wfca,SLS,inst + Wslip,SLS,inst+

w
vinst = (Qi) +
Wfca,SLS,inst (^1) + ^sUp.SLS.inst 5.85

X ( Wfca,SLS,inst (^O,/ Qi) + Wslip,SLS,inst (Vofii

and the final deflection as:

Wfca,SLS,fin ( ^ ) (1 + ^dcf ) + Wsl ip.SLS.fin ( ^ ) (1 + 2 £dcf ) +

Wfca,SLS,fin (fil )(^ + V^2,l ^dcf ) + Wslip,SLS,fin (fil )(^ + ^2,1 ^dcf ) +
5.86
X(Wfca,SLS,fin(fii)(v<O,i + ^2.i^def )+ Wslip^LS,fin (ei)(ni+2nifcdrf))

y
where 2 is a reduction factor giving the quasi permanent load value,
y
while 0 gives the combination value of the load.
In equations 5.85 and 5.86, make sure to use E0,mean in all wfca and Kser
in all wslip. Note also that shear deflection caused by shearing of each
timber member is neglected, which is justified by the fact that its
contribution usually is small compared to wslip.

5.2.3.2 Failure criteria in the ULS

Any failure criterion must be adjusted to meet the requirements of
each particular material used as members and connectors. Stresses in
each member must in general be checked, remembering that shear
lag must be considered in very wide flanges and in wide compression
flanges local buckling as well. The effective width can be determined
using the approach given in section 5.1.2.1, page 135. The integrity of
the shear interface/interfaces must also be checked, that is shear
stresses in a continuous core material or connector forces if discrete
mechanical fasteners are used. Here, the failure criteria of a beam
made of timber members joined by mechanical connectors will be
discussed.
If the flanges are made of structural timber, glulam, LVL or alike,
then the interaction formulas in section 3.1.3 (or if the bending part in
a member is small the criteria in equation 5.6, page 122, or 5.25,
page 135) should be checked for each part. If lateral torsional instabil­
ity is also judged to be a problem then one may proceed as outlined
in section 5.1.1.2, page 120, but making proper adjustments due to
the influence of partial composite action.
If the connectors are made of nails, staples, screws, dowels or bolts,
then the failure modes according to Chapter 4, page 87, must be
checked. Note, however, that bolts (and maybe also dowels) are quite
unsuitable to use as connectors in this kind of composite beam,
because the large initial slip caused by the clearance between bolt
and hole will give large deflections before any shear force can be
transmitted.

Design of timber structures - Volume 1 159

6.1 General considerations

Horizontal stabilization
Roberto Crocetti

6.1 General considerations

6.2 The importance of bracing
160

163
6.1 General considerations
6.3 Fundamental statics 164 In design of structural systems, the way lateral stability is achieved is
of fundamental importance. For the sake of simplicity let us
6.4 Strength and stiffness requirements approach the problem by studying a planar structure. All structures
for bracing systems 167 undergo some changes in shape under load. In a stable structure,
6.4.1 Perfectly straight column 167
the deformations induced by the load are typically small. Moreover,
6.4.2 Column with initial imperfections 169
6.4.3 Beam, truss or column systems 171
in a stable structure, internal forces are generated by the action of
the load; these forces tend to restore the structure to its original
6.5 Typical bracing systems shape after the load has been removed. In an unstable structure,
for heavy timber structures 174 the deformations induced by a load are typically large and tend to
6.6 Special topics 176
continue to increase as long as the load is applied. By contrast,
6.6.1 Forces arising due to an unstable structure does not generate internal forces that tend to
the slope of the roof 176 restore the structure to its original configuration. It is the fundamen­
6.6.2 Stability of the upper chord of tal responsibility of the structural engineer to assure that a proposed
low-arch or low-truss bridges 177 structure does indeed form a stable configuration. Stability is a cru­
6.6.3 Bracing of continuous beams cial issue in the design of structures that are assemblies of discrete
at intermediary supports 177
elements. For example, the post-and-beam structure illustrated in fig­
6.7 Bracing of timber framed buildings 178 ure 6.1 a) is apparently stable. However, any horizontal force tends to
6.7.1 Floor and roof diaphragms 178 cause deformations of the type indicated in figure 6.1 b).
6.7.2 The in-plane resistance of shear walls Clearly, the structure has no capacity to resist horizontal loads, nor
under horizontal loading 179 has it any mechanism that tends to restore it to its initial shape after
horizontal load is removed. The large changes in angle that occur
between members characterise an unstable structure that is begin­
ning to collapse. Consequently, this particular pattern of members is
referred to as a collapse mechanism.
There are just a few fundamental ways of converting a self-stand­
ing structure of the general type shown in figure 6.1 from an unstable
to a stable configuration. These methods are illustrated in figure 6.2,
page 161.

Figure 6.1 a) Column-and-beam assembly,

b) instability under horizontal loads,
c) instability in a wall-and-plate assembly.

160 Design of timber structures - Volume 1

 6.1 General considerations

Figure 6.2 a) Diagonal bracing, b) shear wall, c) rigid joints.

The first is to add a diagonal member to the structure, see figure 6.2 a). Table 6.1
In such a way, the structure cannot undergo the “parallelogram- Diameter Horizontal stiffness
ming” indicated in figure 6.1 a), page 160, without a very large elonga­ k F/
= d
tion of the diagonal member — which will not occur if the diagonal
No diagonal k = k0
is adequately sized. Another method used to assure stability is
d = 10 mm k » 7 • k0
through shear walls, that is rigid planar surface elements that resist
changes in shape of the type illustrated, see figure 6.2 b). A third d = 20 mm k » 25 • k0

method used to achieve stability is by stopping the angular changes d = 30 mm k = 50 • k0

that occur between members and that are associated with collapse,
see figure 6.2 c). In this case the connections between members are
such that the angle of intersection between members remains con­
stant under any loading. Such connections are often referred to as
rigid connections (or rigid joints).
There are, of course, variants on the basic methods of assuring sta­
bility. Still, most structures composed of discrete elements rely on one

6m
or the other of these basic approaches for stability. More than one
approach can be used in a structure, for example a structure having
both rigid joints and diagonals. This is normally done when the stiff­
ness of the structure needs to be improved and, in such a case,
the redundancy is obviously increased. As an example, consider
the following glulam frame, with clamped connection at the base of
the columns and pinned connections between beam and columns, see
figure 6.3. The system is stable without any diagonal bracing. However, Figure 6.3 Influence of the diagonal bracing
the introduction of a steel diagonal dramatically increases the hori­ on the stiffness of the frame
zontal stiffness of the frame, even when relatively small diameters of The column are bent around their strong axes.

the steel bar are used.

In the case of a real building (which by nature is of spatial charac­
ter) the way it resists lateral forces influences not only the design of
vertical elements, but the horizontal spanning system as well.
The basic issues associated with the effect of lateral forces due to
wind or earthquakes on a spatial structure are illustrated in figure 6.4.
Lateral forces not only cause structures to deform horizontally. They
can also cause twisting or torsional deformations. If no adequate
resistance to such mechanisms — for example shear walls, diagonal
bracings or rigid frames — are present to resist lateral forces, unac­
ceptable displacements and possibly complete collapse can occur.

Figure 6.4 a) Basic structure without lateral stability devices,

b) major lateral displacements (“racking”) due to horizontally-acting wind or earthquake loadings,
c) torsional displacements.

Design of timber structures - Volume 1 161

6.1 General considerations

Reality Codes

Figure 6.5 Wind pressure and suction, reality versus building codes. In some building codes the model for
wind pressure and suction may be more sophisticated with slightly different coefficients.

Wind forces on the side of simple buildings of the type shown in fig­
ure 6.5, for example, would be picked up by surface members, which
in turn transfer them to secondary frame elements. Earthquake forces
have a slightly different character than wind forces, as they are nor­
mally concentrated at high-mass areas, such as roofs or floors, but for
the purpose of this discussion, we assume that the effects of wind or
earthquakes can be represented by a series of forces acting laterally
on the primary structure.
It should be emphasised that a primary role played by a roof or
a roof plane is, besides resisting vertical loads, to transfer lateral
forces to side shear walls, cross braces, or frames. Figure 6.6, page 163,
illustrates two methods for picking up forces acting on transverse
wall and transferring them to side shear planes. The figure also illus­
trates how forces on the longitudinal face are transferred to side
reverse shear planes.
The use of an edge beam approach illustrated in figure 6.6 a) demands
that the edge beam be sized for both vertical and horizontal loads.
This approach is often used for relatively small buildings. As building
dimensions increase, a larger part of the roof plane is typically uti­
lised for stiffness. An entire bay, for example, might be cross braced
as a way of providing sufficient in-plane stiffness for a large building,
see figure 6.6 b). In these cases, it is essential that the whole system be
organised such that the rigid horizontal plane (at the roof) both receive
externally induced horizontal forces and transmit them to side shear
planes (at the walls). Unless there are two separate systems for trans­
mitting horizontal forces, for example purlins and struts, the floor or
roof beams (purlins) must be designed to carry both bending due to
vertical loads and compression due to horizontal loads.

162 Design of timber structures - Volume 1

 6.2 The importance of bracing

Purlin

Loading on gables
a) Forces from secondary frames are resisted by an edge beam b) Forces are transferred through roof members to the roof
with high lateral strength and stiffness, and carried directly to diaphragm which transfers loads to side diaphragms (note that
side shear walls or diaphragms (typically small structures only). in this case purlins must be designed to carry both compressive
forces and bending from vertical load).

Rigid roof diaphragm

in-plane truss or rigid frame)

Loading on longitudinal faces

c) Forces from secondary framing are picked up by rigid roof d) Horizontal and vertical rigid planes or diaphragms can be
plane or diaphragm and transmitted to rigid side planes placed anywhere as long as loads can be transmitted to them.

Figure 6.6 Force transfer to rigid wall or roof-planes. The graphic convention of a grey tone with dotted “X”
is used herein to depict a rigid plane, regardless of whether it is a wall, truss, or rigid frame.

6.2 The importance of bracing

Bracing effectively resists the racking of the roof, especially under lat­
eral loads from a direction other than parallel or perpendicular to
the building. In other words, an effective bracing provides the roof
with beneficial torsional stiffness in the roof plane. Bracing is also
often needed during the construction stage. Whilst for minor timber
structures sufficient bracing can be achieved by means of wood panel
sheathing, for heavy timber structures bracing consists, generally, of
either:
• steel rods capable of acting in tension only
• timber struts, which can take both tension and compression
• profiled steel shear diaphragms.

When steel rods are used, they are usually arranged in crossing pairs,
so that if racking puts one of the elements into compression (which it
is incapable of resisting), its complement will resist racking by tension.
Bracing may also be necessary to prevent lateral buckling of
the entire roof plane and to keep the filler beams (purlins or separate
struts) in place so that they can give the proper lateral support to
the compression zone of the beam under in-plane bending, see fig­
ure 6.7, page 164.

Design of timber structures - Volume 1 163

6.3 Fundamental statics

Buckling length: 0

Figure 6.7 Roof structures with main beams braced in different ways:
a) Unbraced. Buckling length of the main beams lef = l.
b) Braced with crossing pairs of steel rods. Buckling length of the main beams lef = l/4.
c) Braced with profiled metal sheathing. Buckling length of the main beams lef = 0.

The specific arrangement of the bracing in the roof plane depends on

different aspects, such as:
• the position of the lateral force resisting elements
(wall bracing systems)
• the required stiffness of the roof structure
• the magnitude of the force flow.

The wall bracing system which resists lateral forces may be arranged
in a symmetrical or asymmetrical manner. For the symmetrical case
they will only carry direct forces as caused by translation. For
the asymmetrical case, additional forces due to rotation (torsion) are
generated, see figure 6.10 b), page 165.

6.3 Fundamental statics

H/4 H/4 H/4 H/4 As a rule, for closely spaced, parallel, supporting, vertical structures
(that is the wall bracing systems), the roof or floor diaphragm can be
considered as rigid, see figure 6.8. Therefore, lateral load distribution
is known, being just a function of the tributary fagade area. However,
in the case of a statically indeterminate system, the load distribution
is not obvious if rigid frames or shear walls are placed rather far
apart. In such a case, in fact, the roof or floor diaphragm becomes
more flexible, see figure 6.9, and therefore, lateral load distribution
depends on the ratio between the rigidity of the roof (or floor) dia­
H
phragm and the rigidity of the supporting structure. Though there is
H/3 H/6 H/6 H/3 no such thing as infinitely rigid or flexible diaphragm, the concepts
are still useful for developing some understanding about lateral force
flow. Normally, concrete slabs and some types of steel deck are con­
sidered be rigid diaphragms which are assumed to distribute the hori­
zontal forces to the supporting vertical structures in proportion to
their relative rigidity R.
For a building with rigid roof diaphragm and four equally stiff
shear walls (or equivalent frames) each wall carries the same force,
H see figure 6.8 a). In the case b), the interior walls only provide one-half
Figure 6.8 Rigid roof diaphragm
the rigidity of the exterior walls. In this case, the total horizontal
a) Four walls with equal rigidity R, action H must be redistributed in proportion to the wall rigidities.
b) walls with different rigidities. The total rigidity is:
D
X^ = 2(Z? + y) = 3J? 6.1

164 Design of timber structures - Volume 1

 6.3 Fundamental statics

Hence, each of the external walls carries:

6.2

while each of the interior walls resists:

V = H-^ = ^ 6.3
1 3-R 6

Roof with wood based sheathing, or with some type of steel deck, are
generally considered to be flexible diaphragms. They are normally so
much more flexible in comparison to the lateral supporting system
that the supporting structures are considered to be rigid. The behav­
iour of the flexible diaphragm may be visualised as simply supported,
discontinuous beams sitting on unyielding supports. Thus, the loads
can approximately be distributed as based on tributary area. For
the case in figure 6.9 a), note that the smaller interior walls carry dou­
ble the force as the longer exterior walls, because the tributary areas
are twice as large as the ones for the exterior walls; the stiffness of
the walls is not considered since they are assumed infinitely rigid.
The behaviour of semi-flexible or semi-rigid diaphragms, on
the other hand, is highly indeterminate. They are assumed to distrib­
ute the loads like a continuous beam on flexible supports.
If the lateral force-resisting structures are arranged in an asymmet­
rical fashion, then in addition to translation (that is direct shear),
rotation (that is torsion) is generated. The magnitude of the generated Figure 6.9 a) Flexible roof diaphragm,
rotation is also influenced by the building proportions, see figure 6.10. b) semi-rigid roof diaphragm.
The solution for the distribution of the lateral forces to the asym­
metrically arranged shear walls can be approached by distinguishing
between determinate and indeterminate systems as based on the fol­
lowing simplified assumptions:
• Consider the roof as a rigid diaphragm. Flexible diaphragms are
considered incapable of transmitting rotational forces.
• Neglect the bending stiffness of the walls about their weak axis.
• Consider intersecting walls as separate units. There is no shear
transfer at the corners.
• Consider the shear flow along the wall to be constant.
• Neglect the effect of columns in resisting lateral forces.

Hence, according to these assumptions, as based on statics, a mini­

mum of three shear walls or other types of lateral force-resisting pla­
nar systems is necessary to stabilize a building against force action
from any direction; the arrangement of the walls shall be neither
parallel nor concurrent according to the laws of statics. When more

a)

Figure 6.10 Effect of building proportions on asymmetrically braced structure

a) Square b = a, b) rectangle b = 4 • a, problematic arrangement, c) rectangle b = 4 • a, preferred arrangement.

Design of timber structures - Volume 1 165

6.3 Fundamental statics

than three shear walls are present, the structure is statically indeter­
minate and the force distributions can be found by means of either
approximate methods or computer-based approaches. As an example,
let us study the force distributions in the building shown in figure 6.11.
The building has a channel-shaped stabilising system and the hori­
zontal loads act mainly along two orthogonal directions, that is x and
y. Let us first consider the equilibrium for loads acting along y-direc-
tion, see figure 6.12.
The total horizontal force “Y” acting perpendicular to the long
fagade of the building must be resisted by the forces P1, P2 and P3. It is
worth to remember that the wall assembly is not considered as a single
channel-shaped unit but as three individual, independent elements.
Moment equilibrium about the intersection of the unknown forces P1
Figure 6.11 Illustration of a building stabilised by and P3 yields the unknown wall force P2:
a channel-shaped bracing system
xn 2
^M = 0 => y-2 a-P2-3 a = 0 => P^ = ~ Y 6-4

Equilibrium in x-direction results in:

v 2
SX = 0 => p1-p2 = o => p1 = p2 = - y 6.5

Equilibrium in the y-direction results in:

^Fy = o => p3-y = o => p3 =y 6.6

For the case where a lateral force resultants “X” acts perpendicular to
the short fagade, see figure 6.13, the force P3 along the core wall is
zero.

In fact:

X^y=0 => P3=0 6.7

Rotational equilibrium force about force P1 yields:
xn 2
68m = 0 => X-2-a — P2-3-ti = 0 => /2= —-X 6.8

Equilibrium in x-direction results in:

^Fx=0 => P1 + P2 = X =► 6.9

a

Figure 6.12 Resisting mechanism for horizontal loading Figure 6.13 Resisting mechanism for horizontal loading
perpendicular to the long fagade perpendicular to the short fagade

166 Design of timber structures - Volume 1

 6.4 Strength and stiffness
requirements for bracing systems

6.4 Strength and stiffness

requirements for
bracing systems
The function of lateral bracing is to provide lateral support to a mem­ Fbr /2
ber in order to prevent it from moving laterally at the bracing posi­
tion. Normally, the same bracing elements that are used to prevent
lateral movements due to external transversal loading (for example
z
wind or earthquake) serve also to increase the buckling strength of
the primary members, such as beams and columns.
An adequate bracing system requires both strength and stiffness.
A simple brace design formulation such as designing the brace for
1 — 2 % of the member compressive force addresses only the strength
z
criterion. If brace connections are too flexible, they can have
a very detrimental effect on the stiffness of the main structure.
Fbr /2

6.4.1 Perfectly straight column

P
Let us study an axially loaded column braced at mid height by means
of a bracing element represented in figure 6.14 by a spring with stiff­ Figure 6.14 Model of a initially straight column braced
ness C. In order to simplify the problem, we assume that the column at mid-height with applied perturbation
is hinged at the mid span. In other words, the bending stiffness of
the column is ignored (the more slender the column, the more realis­
tic is the assumption). Let us now imagine to apply a very small
transversal displacement at the column mid-height and let us study
the equilibrium of the system.
The equilibrium of the system can be studied taking moments
about the hinge at mid-height:
P C1 A
y\M = 0 => PA—— a = 0 => PA------—a 6.10
^2 2
where Fbr is the bracing force. Note that the term on the left hand
side of equation is the external moment Munst, which tends to unstabi­
lize the system, while the term on the right hand side is the stabiliz­
ing moment Mst which occurs due to the presence of the spring at
mid-height. Three types of situations are possible, namely:

Ca „ 2P
Unstable Manst > Mst => P>----- =} C <----- 6.11
2 a

,, .„ _ Ca „ 2P
Stable *
Afu^cM => P < — => C >----- 6.12
2 a

Ca 2P
Critical na = M.sit => P =----- => C =-----
M„unsi 6.13
2 a
If the equilibrium is stable, it means that when an external perturba­
tion is applied, and when the source of this perturbation is removed
the system will return back to its original position, that is the perfectly
straight position. On the other hand, if we imagine to gradually
increase P, and for each small increment of P we give a small pertur­
bation at the mid-height of the column that produces a displacement
A, a condition will be reached in which the column — after the given
perturbation — will not return to its original position. The load cor­
responding to this situation is referred to as the critical load.

Design of timber structures - Volume 1 167

6.4 Strength and stiffness
requirements for bracing systems

However, the maximum axial load that the column can resist cannot
be larger than the buckling load of the column, which in this case is:
, E l
P<PE = n2- — 6.14
a
PE is the Euler load for a strut with the length a under compression, E
is the modulus of elasticity parallel to grain and I is the second
moment of area about the z-axis, according to figure 6.14, page 167.
The corresponding buckling shape, when the stiffness of the spring
C > CE is large enough for achieving either a stable or a critical equilibrium,
is shown in figure 6.15.
The spring stiffness CE which is necessary in order to achieve
the buckling shape shown in figure 6.15 is therefore:
„ 2-Pp „ 2 El
CE------ - = 2-n2—3- 6.15
P a a
Figure 6.15 Buckling mode for spring stiffness C> CE Figure 6.16 shows the relationship between the load carrying capacity
of the column and the stiffness of the bracing element, for an ideal
column. When C < CE, there are two equilibrium situations, one stable
PE
with no transversal displacements and one critical with a small trans­
versal displacement. This critical situation is achieved when the rela­
tionship between P and C is that given by equation 6.13, page 166, (in
figure 6.16 this relationship is described by the line AB). When C > CE
there will also be two possible equilibrium situations, again, one stable
PE with no transversal displacement and one critical when P = PE, with
the formation of a buckling shape with two half-waves, see figure 6.15.
The above model — though slightly more complicated — can also
be used to determine the critical load on a column with several inter­
mediate braces each with the same stiffness C. Figure 6.17 shows
a column with three intermediate equidistant braces. Increasing
the brace stiffness, the buckling mode of the column switches from
PE a single half-sine wave to two, three and finally four half-sine waves,
4 which occurs when C approaches the value (3,4 • PE ) / a, Yura (1996).
If the number of braces is further increased, the brace stiffness nec­
CE essary to form a series of half-sine waves between all the brace points
Figure 6.16 Influence of the brace stiffness on
is slightly larger. For practical cases, when the number of intermedi­
the loading carrying capacity of an ideal column ate braces is n > 4, the required stiffness should be CE > (4,0 • PE)/ a,
Note that PE and CE are defined in equation 6.14 and Yura (1996).
equation 6.15 respectively.

Figure 6.17 Load-stiffness relationship for a column with three intermediate

supports. For the case with three intermediate supports, the full bracing stiffness
requirement is achieved when CE = (3,4 • PE)/a, where PE = n2 EI/a2.

168 Design of timber structures - Volume 1

 6.4 Strength and stiffness
requirements for bracing systems

6.4.2 Column with initial imperfections P

It is worth observing that for perfectly straight columns with full

bracing (that is C > CE) there is no force in the brace even at buckling
because there is no displacement at the brace point. However, tests
have shown that brace forces do develop when a column is subjected
to axial loading. The model shown above, with initial straight col­
umn, does not give any information about the deformations of
the bracing and the corresponding forces which are generated in it.
In order to estimate deformation and corresponding bracing force,
Winter (1958) proposed a rigid link model with fictitious hinges and
with an initial out-of-straightness, A0, see figure 6.18.
As a load P is applied there will be additional deflection A which
can be determined by taking moments about point n:
P C A
^M = 0 =5 P-(Ao + A)-^-a = O =5 P-(A0 + A) = ^--a 6.16

Observing that AT = A0 + A, equation 6.16 can be rewritten in the fol­

lowing manner:
Figure 6.18 Model of column braced
PA C-(21T-21O) 6.17
at mid-height an initial out-of-straightness

T 2

Let us first study the system when the stiffness of the spring is C = CE,
that is stiffness such to make the column to buckle according to fig­
ure 6.15. We impose:

2-P■tif
C = CE = 6.18
a
Substituting equation 6.18 into equation 6.17 gives:

P-A-y - PE -(^t ^o) 6.19

For C = CE and A0 ^ 0, the heavy solid line in figure 6.19 shows the rela­
tionship between P and AT/A0 given by equation 6.19. For P = 0, AT = A0.
When P increases and approaches the buckling load PE = n2 • E • I/ a2,
the total deflection AT becomes very large (theoretically AT tends to
infinity). For example, when the applied load P is about 90 — 95 % of
PE, the total deformation AT becomes as large as 10 — 20 times A0.

Figure 6.19 Relationship between Pand AT for a braced column with initial out-of-straightness

Design of timber structures - Volume 1 169

6.4 Strength and stiffness
requirements for bracing systems

In general this is not acceptable for a real structure. Consider for

example a glulam column, with length h = 6 m and initial out-of­
straightness A 0 = h/ 500 = 12 mm, braced at mid-height. If the chosen
brace stiffness should be only C = CE, when a compressive load of
the magnitude of 0,95 • PE is applied on the top of the column,
the deformation at mid-height could be as large as 20 x 12 = 240 mm.
On the other hand, if a brace stiffness twice the value of the ideal
stiffness is adopted, that is C = 2 • CE, Equation 6.19, page 169, becomes:

------ 2- 1--------- — valid for P < PF 6.20

k ^T/A) >

In this case, the deflections are much smaller than for the case with C
= CE, as shown in figure 6.19. For example, when the load just reaches
the buckling load (P = PE), then AT = 2 • A0, not infinity as in the case
with C = CE. For C = 3 • CE and P = PE, AT = 1,5 • A0.
It is evident that the larger the brace stiffness, the smaller AT. Let
us now estimate the forces that occur in the brace when brace stiff­
ness is varying (note that C is derived from equation 6.17, page 169):

2P 4) + ^ 2P
■A-------- A. 6.21
a A J a
We study first the system when the stiffness of the spring is C = CE. In
that case (see also equation 6.19, page 169):

A 2-P _A_ 6.22

1~P/PE a 1-P/PE

which shows that the brace force is directly related to the magnitude
of the initial imperfection. If a member is fairly straight, the brace
forces will be small. Conversely, members with large initial out-of­
straightness will require stronger braces. Equation 6.22 can be rewrit­
ten in the following manner:

P_ = 1___ L_ LA 6.23
Pe Fbr/P «
A plot of equation 6.23 for an initial imperfection A0 = (2 • a) / 500 is
shown in figure 6.20. If the brace stiffness is just equal to the ideal
stiffness, then the brace force becomes very large as the buckling
load is approached; this occurs because AT becomes very large, as
shown in figure 6.20. For example, at P = 0,84 • PE, equation 6.22 gives

Fbr/P

P F
Figure 6.20 Relationship between and br for a braced column with initial
A
out-of-straightness equal to 0/(2 • a) = 1/500

170 Design of timber structures - Volume 1

 6.4 Strength and stiffness
requirements for bracing systems

a brace force of 5 % of PE. It should be emphasised that equation 6.22,

page 170, is only valid for Fbr /P > (2 • A0 / a), otherwise equilibrium is
not possible.
Therefore, it can be concluded that a brace system will in general
not be satisfactory if its stiffness is just equal to the theoretical ideal
required stiffness CE, because the brace forces will get too large. If
the brace stiffness is ‘over designed’, for example by C = 2 • CE or
C = 3 • CE as shown in figure 6.20, page 170, then the brace forces will
be more reasonable.

N
6.4.3 Beam, truss or column systems
Where a bracing system is required to provide lateral stability to
a series of compression or bending members (for example columns, N

trusses or beams), this is effectively achieved by providing lateral

stiffness using truss or diaphragm action within the plane of Figure 6.21 Typical bracing of beam system for
the bracing structure, see figure 6.21. preventing lateral buckling
For the general case of a series of similar compression members
that require to be braced at positions along their length, it can be
assumed that each compression member under load will have an ini­
tial out-of-straightness A0, as shown in figure 6.22.
A single member subjected to an axial force N can alternatively be
analysed by replacing the effect of the initial deviation from straight­
ness on the deflection behaviour, by the effect of an equivalent lat­
eral load qh1 acting on the member when it is in an initially straight
condition — such that the maximum bending moment in each case
will be the same, see figure 6.23:

KT A 8-N-A0
N ’ Ao - <7hr g => tfhi - ^2 6.24

Where n members are to be braced and each member is subjected to

a compression force N, assuming that all members contribute to
the loading to be taken by the bracing system, the total compression
load will be n • N:

8-jV-J0
?h =w' 6.25
£2

Sine-shaped
N N N deformed profile

Bracing system

Figure 6.22 Model for the determination of the brace forces: Figure 6.23 Equivalent lateral load acting on the bracing system
bracing system for a series of compression or bending members

Design of timber structures - Volume 1 171

6.4 Strength and stiffness
requirements for bracing systems

For timber members, the initial out-of-straightness defined as A0 /l is

normally assumed to be 1/ 500 and 1/ 300 for glulam and structural
timber respectively. Substituting these values into equation 6.25 yields:
TV
- for glulam structures 6.26
h 60-I
and:
TV
f for structural timber structures 6.27
h 40-^

For a member subjected to bending, such as that shown in figure 6.21,

page 170, the action of the bending moment can conservatively be
assumed to be M ~ 2/3 • N • h, where h is the depth of the beam.
It is important to observe that if the bracing is inadequately stiff,
the maximum load carrying capacity of the beam cannot be
achieved. This is shown in figure 6.24. The test results also show that
lateral deflections of the bracing beam (and thus the lateral
deflection of the timber beam) dramatically increase when the load
approaches a certain value, which depends on the lateral bending
stiffness of the bracing beam (UPE-type steel beams were used as
bracing in the experiment). As an example, when the bending stress
in the tested timber beam is 20 MPa, the corresponding bending
moment is 7,5 kNm and the vertical load is 8,2 kN/m. The lateral load
on the bracing system which is generated by the out-of-plane deflection
of the timber beam is approximately:
• 0,70 kN/m (out-of-plane deflection ~ 25 mm) for the less stiff
bracing beam (UPE 50)
• 0,35 kN/m (out-of-plane deflection ~ 6 mm) for the stiffer
bracing beam (UPE 80).

These results show that the brace force increases with decreasing lat­
eral stiffness of the bracing system.
Experimental results also show that in case of double pitched
beams with upper side inclination of 5°, the brace force is 10 % to
30 % higher than in the case of similar timber beams with parallel
upper and lower sides, Holzbau Handbuch (2000).

Timber beam bending stress cm at mid-span [MPa]

B = 25 mm, H = 300 mm
—o— Bracing beam: UPE 50, I = 9,12 x 106 mm4
—•— Bracing beam: UPE 80, I = 19,4 x 106 mm4
G = 500 MPa
E = 11 000 MPa

Figure 6.24 Test setup for determining brace force, and test results
The struts are connected to the top edge of the tested timber beam (Holzbau Handbuch, 2000).

172 Design of timber structures - Volume 1

 6.4 Strength and stiffness
requirements for bracing systems

According to Eurocode 5, the bracing system shall — in addition to

other horizontal loads, for example wind or earthquake — be designed
for a horizontal, uniformly distributed load:

9h n'30h-Ar 6.28

where:
M maximum moment in the beam.
h depth of beam.
l span of the beam [m].
n total number of laterally braced beams.
kcrit factor that takes into account the risk for lateral buckling
in an unbraced beam (see section 3.2.3.3, page 68).

The reduction factor k in Eurocode 5 intends to consider that greater

care in workmanship can be expected in large span structures. Since
it is not representative under normal conditions, it has here been
omitted.
The laterally bracing structure marked with (1) in figure 6.25 must
be designed for the load qh according to equation 6.28 in addition to
the external load from wind qw shown in the figure.
Fixing of purlins or roofing sheets in the roof beam should be
designed for a force of:

*2 “ ”b 6.29
n
where:
a distance between purlins or, in sheets direct on
the roof beams, distance between screws.
nb number of roof beams braced by the force transferred
via the purlin or roof sheeting.

The bracing structure shall be stiff enough to limit the deflection due
to the design load qh to l / 700 and due to the total load, including for
example wind load, to l / 500.

] <— qw

A-A

qw
n beams

Figure 6.25 Lateral bracing of roof beams. 1) Laterally bracing structure, 2) roof beams.

Design of timber structures - Volume 1 173

6.5 Typical bracing systems
for heavy timber structures

6.5 Typical bracing systems

for heavy timber structures
To be able to take up horizontal loads, heavy timber structures are
normally stabilised in accordance with one of the following
alternatives:
• The diaphragm action of the roof which transfers the horizontal
load to wall diaphragms, normally placed in the gable walls.
• Wind trusses in the roof which transfer the horizontal load from
columns to braced columns, normally placed in the gable walls as
in figure 6.26.
• One or both of the columns is rigidly fixed to the foundation as in
figure 6.27 a).
• One or both of the columns is rigidly fixed to the beam as in fig­
Figure 6.26 Bracing of heavy timber structure using ure 6.27 b), forming a three- or two-pin frame respectively.
wind trusses and wall bracing • The structure is complemented by diagonals as in figure 6.27 c). For
functional reasons this is usually only possible in gables.
• The structure is rigidly connected to a wall diaphragm in its own
plane.

a) The diaphragm action in flat profiled sheet metal roofs can be used
to stabilise glulam timber structures under the following conditions:
Figure 6.27 Examples of structures which are stable in
• The sheets forming the roof can be jointed so that shear forces can
their own plane
be transferred.
• The roof diaphragm is stiffened along its edges by beams designed
and fixed to take up axial forces. Edge beams at right angles to
the profiling can be of timber, while edge beams parallel with
the profiling are most easily made in steel.

The roof diaphragm can be designed according to the same principles

as for steel structures.
Frames and arches are stable in their own plane, as are column
and beam systems with columns rigidly fixed in the foundations.
Systems with hinged columns must, on the other hand, be stabilised
by diaphragm action in the roof or by wind trusses in the plane of
the roof. The principle of wind bracing with trusses along the side of
the building is shown in figure 6.28.

Figure 6.28 Wind bracing

of frame and arch structures

174 Design of timber structures - Volume 1

 6.5 Typical bracing systems
for heavy timber structures

The roof beams (primary beams) are often used as bottom and top
chords in the truss, with some of the purlins as posts (or compression
struts). The purlins, if they are used as posts for the wind truss, can
not be designed as Gerber beams. Only diagonals are added as stiffen­
ers, often crosses of steel rods or sometimes timber diagonals, which
are stiff in compression. The joints between the various components
forming part of the wind truss must be designed with regard to
the forces which arise and to eccentricity. Truss posts and chords are
often in different planes.
The wind truss is often designed so that it can be used for stabilisa­
tion during the erection of the structure. In smaller buildings (say
with a longitudinal dimension of about 30 — 40 m), only one wind
truss in the roof can be sufficient. The purlins or the roof diaphragm
must then be able to transfer both compression and tension forces
from the other end. In longer buildings it can be suitable, on the other
hand (not least with regard to stability during erection) to arrange
two or more wind trusses in different bays within the building.
The wind truss can with advantage be placed in the second bay from
the end. The design of joints is then not affected by the end wall
structure (gable), which usually differs from that in the rest of
the building. Other types of bracing arrangements for achieving dia­
phragm action in the roof are shown in figure 6.29.
Since roof diaphragms or wind trusses are often used as lateral
bracing of roof beams, they must — in addition to other horizontal
loads — be designed for a horizontal, uniformly distributed load q,
derived according to the method shown in section 6.4.3, page 171.
The structure in hall-type buildings with large spans (> 15 m) contain­
ing spaces where large numbers of people are present at the same time,
for example sports or exhibition halls and department stores, shall be
designed in such a way that the risk of progressive collapse as a result
of an accidental load is acceptably small.
It is important to observe that heavy timber structures must be
able to transmit loads from the roof level onto the foundations.
Moreover, stability must be ensured against all types of mechanisms
that can occur, including lateral buckling. figure 6.30 illustrates this
concept. If a bay of a structure is to be stabilized against lateral loads,
the first thing to do is to prevent columns to move laterally which
may be achieved, for example, by introducing diagonal members as
shown in figure 6.30 a). However, although the columns are now

Figure 6.30 Example of a proper lateral bracing of a bay in a larger timber building

Design of timber structures - Volume 1 175

6.6 Special topics

braced, loads from the roof plane cannot be transmitted to the foun­
dation through diagonal members. Therefore, the beams need — at
their support points — a bracing system which permits the trans­
mission of loads from the roof to the underlying structure, for exam­
ple as shown in figure 6.30 b), page 175. Although the system may look
stable now, the beams are still prone to lateral buckling, which may
be prevented for example by creating a horizontal truss at the roof
plane, where the beams work as compression and tension chords, see
figure 6.30 c), page 175.
An accidental load can for example be a car which runs into a col­
umn, or a gas explosion within the building and as a rule it is not
reasonable to design individual structural components to withstand
such loads. Instead, the structural system should be designed so that
a primary part of the building, such as a main beam or column, may
collapse without the building as a whole collapsing. In loading com­
binations with accidental loads, other loads are as a rule assumed to
act with their normal (frequent) values and not with their extreme
values (characteristic values).
Normally, it is sufficient to estimate the horizontal stability of
the remainder of the building if one main beam fails. In buildings
containing spaces for large numbers of people it must also be shown
that the damaged bay can be spanned by an alternative load-bearing
system, for example by utilising the line- and membrane effect of
purlins and roofing. This can however also be a disadvantage in cases
where a primary girder fails due to a design or construction error
which is similar for all girders in the building. In that case adjacent
girders may also fail in a progressive manner following the failure of
one bay. Several such cases have been experienced during recent
years; see Sorensen et al. (2011). The risk of progressive failure is then
higher if the bays are structurally connected to each other in the lon­
gitudinal direction of the building. In such situations a better strategy
is to decouple the system to ensure that failure in one bay will not
q spread to adjacent bays.

6.6 Special topics

In this section some relevant topics related to bracing systems will be
briefly discussed, namely:
• forces that arise due to the slope of the roof
• stability of the upper chord of low-arch or low-truss bridges
• bracing of continuous beams at intermediary supports.

Downwards force 6.6.1 Forces arising due to

the slope of the roof
In design it should be observed that the wind truss is seldom plane.
Upwards force At nodes where the members of the truss change direction, for exam­
ple at the ridge of a double-pitched roof, force components across
Figure 6.31 Upward and downward forces that generate the roof arise which must be taken care of by the primary structure,
due to the slope of the roof see figure 6.31.

176 Design of timber structures - Volume 1

 6.6 Special topics

6.6.2 Stability of the upper chord of

low-arch or low-truss bridges
In absence of upper chord bracing of a low-arch or low-truss bridge,
the lateral buckling of the top chord is only resisted by the elastic
reaction of the vertical and/or diagonal members of the truss or arch.
In that case it is difficult to verify sufficient bracing of the compressed
part of the truss or arch. The stability of the upper chord or of the arch
can be increased by increasing the rigidity of the lateral supports, for
example as shown in figure 6.32. The upper chord of a truss or an arch
can then be considered as a bar with hinged ends compressed by forces
distributed along its length and elastically supported at intermediate
points as shown in figure 6.32. At the supports there are usually
frames or bracing members of considerable rigidity, so that the ends Arch
of the chord can be considered as immovable in the lateral direction. N
For a constant cross section of the chord and a constant compres­
sive force, the minimum rigidity, at which the supports begin to
behave as though they were absolutely rigid is found in section 6.4.1, Figure 6.32 The stability of a low-arch bridge can
page 167. be increased by increasing the stiffness of the lateral
supports (that is by increasing the spring stiffness C in
the model on the right).
6.6.3 Bracing of continuous beams
at intermediary supports Main beam

In buildings, for example of hall type, the primary beams are often
made continuous, that is with the beam supported by more than two
columns. As a rule, the horizontal frame bracing, for example wind
truss or roof slab, is at the level of the top of the beam. Therefore, in
the zone around the intermediate support, the beam — which is sub­
jected to compression at the underside — must be braced at the under­
side in order to prevent it from buckling sideways (lateral torsional
buckling). The bracing of the underside of the beam can be performed
for example as shown in figure 6.33.
The bracing should preferably be such that it can only take ten­
sion. This is to avoid possible torsion of the beam due to uneven load
distribution which could occur, for example, when strong wind takes
place after snowing. In such a case big masses of snow could collect
in one span of the beam leaving the adjacent span more or less Figure 6.33 Bracing of the underside of a continuous
unloaded. Thus, if the bracing diagonals shown in figure 6.33 were beam at intermediate supports
able to take both compression and tension, the beam would twist on
the support with possible risk for collapse of the roof or part of it.
The horizontal force acting on the bracing can be calculated by
the following formula, Limtrahandbok Del 2 (2016):

6.30
70 h
where:
M the moment in the beam at support.
h the depth of the beam.

Similar bracing arrangement as that shown in figure 6.33 may be

needed also for simply supported beams or arches in circumstances
where negative bending moment can occur. An example could be
long truss beams that support a light-weight roof. In such a case
the lower chord of the truss may need to be braced in order to avoid
lateral buckling due to wind suction.

Design of timber structures - Volume 1 177

6.7 Bracing of timber framed buildings

6.7 Bracing of timber framed

buildings
The behaviour of a roof or a floor under lateral loading depends on
its structure. Whilst trussed roof framing and concrete slabs are rela­
tively rigid, skin constructions (for example made of wood based pan­
els) are normally rather flexible. In flexible skin-framed structure
the floor or roof sheathing or decking acts as a diaphragm, when
properly fastened with shear connectors to the purlins and to
the perimeter members so that shear can be transferred. A typical
flexible diaphragm is shown in figure 6.34.
V

Diaphragm

Edge beam

Main beam

Figure 6.34 Diaphragm action of the roof (flexible diaphragm, for example
wood based panel sheathing)

6.7.1 Floor and roof diaphragms

A diaphragm can be visualised as the thin web of a large roof “plate
girder” (or I-girder), primarily resisting shear, while the boundary
members — that is edge beams or walls perpendicular to the load
direction — act as the plate girder flanges by carrying the moment in
axial action, see figure 6.35, page 179.
The maximum moment is resisted by the longitudinal edge mem­
bers in compression and tension:

mdA =------------
g 6.31

Hence, the boundary members must carry the following axial com­
pression and tension forces, respectively:

„ „ M w-f2
T = C--------------- 6.32
b 8 b
All of the bending is assumed to be taken by the edge beams and con­
sequently they must be continuous or detailed to be able to transfer
the tensile or compression loading to adjacent sections.

178 Design of timber structures - Volume 1

 6.7 Bracing of timber framed buildings

T
M
v

c R V

c
c Plate girder
model analogy
v

v
I

Figure 6.35 “Plate girder” analogy for a roof structure: model and assumptions

The roof skin transfers the shear to the vertical braced frames (shear
walls, diagonal bracing or rigid frames). The maximum shear the roof
web must carry is:

6.33

All of the shear must be taken by the panel material. The shear stress is
higher closer to the edges of the diaphragm. The shear flow, v, (N/mm)
which the diaphragm and its connections must be designed for is:
V
v=—
b
This means that the force in individual fasteners connecting
the sheathing to the frame must be designed for a force Fv = v • s,
where s is the spacing between the fasteners. Similarly the sheeting
panel must be designed for an in-plane shear stress t = v/ t, where t is
the thickness of the panel.
h

Obviously, the mathematical interpretation of the diaphragm behav­

iour under lateral loads is an oversimplification, since the degree of
flexibility or rigidity is highly indeterminate.

6.7.2 The in-plane resistance of

shear walls under horizontal loading
Timber frame shear walls can be sheathed on one or both faces with
the sheathing securely fixed to both studs and rails, enabling the wall
to act as a rigid diaphragm.
The sheathing panels may be made of wooden based materials Figure 6.36 A typical shear wall unit built up of
such as plywood, OSB (oriented strand boards) or hardboard with a timber frame and a sheet
dimension b x h = 1 200 x 2 400 mm, see figure 6.36.

Design of timber structures - Volume 1

6.7 Bracing of timber framed buildings

Figure 6.37 Wall diaphragm resisting racking loads: a) Unloaded wall diaphragm, b) racking resistance, c) overturning.

The thickness of the panels is normally 12 mm or more. Gypsum

boards or various types of cement based boards can also be used as
sheathing panels. The frame members (studs and rails) are normally
made of strength graded timber. Typical cross section adopted for
such members is 45 x 120 mm2 and the fixing between sheathing and
framing is normally made by means of nails or screws.
The sheathing in shear walls is usually relatively stiff with regard to
in-plane deformations and the shear stiffness of the walls depends
mainly on the stiffness of the sheathing-to-framing joints. When used
in low buildings the stabilizing capacity is normally high compared
to the actual horizontal loading.
The sheathing-to-framing joints are normally also decisive for
the racking resistance when the lateral external force is applied as
shown in figure 6.37 b). Horizontal sliding of the wall is resisted by
anchorages such as anchor bolts that connect the bottom rail to
the substrate. Overturning about the bottom corners may also occur
(particularly in walls with large height/width ratios) if appropriate
holding-down fixings are not provided, as shown in figure 6.37 c). In
most timber-framed buildings, beams and floors are designed as sim­
ply supported elements on pin-jointed walls and the lateral strength
and stability of the structure is provided by the use of such vertical
diaphragms or shear walls. In timber design, the in-plane lateral
resistance of a wall diaphragm is referred to as the racking resistance
of the wall.
Different models can be used to evaluate the racking resistance of
a shear wall, most of them based on the shear capacity of an individ­
ual fastener, for example a nail connector. Herein, only one model is
described, namely the “plastic model”, which gives a lower bound
solution, Kallsner et al. (2009). This method can only be applied if
the leading stud — that is the stud which is subjected to upwards ver­
tical reaction force — is fully anchored against uplift (normally, hold­
downs at the end of the wall are used to prevent uplift of the leading
stud). In such a case, the shear wall can be regarded as fully anchored
and all the sheathing-to-framing connectors can be assumed to reach
their maximum shear capacity.
Let us study first a single wall unit. A uniform shear flow is assumed
along the edge of a panel, in compliance with equilibrium condition
of the wall, see figure 6.38, page 181.
The racking capacity of the wall unit Hd can be determined from
equilibrium of horizontal forces at the top rail:

HA = f\,Ab 6.35

180 Design of timber structures - Volume 1

 6.7 Bracing of timber framed buildings

Figure 6.38 Forces in the plastic shear wall model

Hd Hd_;
2 2

where fvd is the shear flow strength (N/mm) at the sheathing-to-frame

joint and b is the width of the panel. In reality, the shear between
sheathing and frame is not continuous but punctual, since it is trans­
mitted by the connectors — which are placed with a centre-to-centre
spacing s (typically s = 100 mm at the perimeter of the wall).
According to the plastic model all individual connectors carry
the full design load Fv,d. For equal spacing s between the connectors
we get f'v,d = Fv,d /s, so that:

„ b
Hd=-Fvd 6.36
s

The hold-down force of the panel can also be calculated as:

Vd = vd-/i = --Fv>d 6.37

s
where h is the height of the shear wall.
For walls made up of several panels (or wall units), for example
according to figure 6.37, page 180, the racking capacity is the sum of
the racking capacity of each single wall unit:

H3=£Hdj 6.38
i

where Hd,i is the resistance of wall unit i.

In design, when openings are present in the shear wall, the capac­
ity of a wall unit including an opening is normally disregarded.
Moreover, the wall must be anchored on each side of the opening.
This means that a building from a static point of view will consist of
a number of continuous vertical strips of shear walls that act
separately.
The plastic model described above rests on the assumption that
the connectors between panels and framing behave in an ideal plas­
tic manner. For most wood based panels connected to the framing
with nail and screw connectors this is a reasonable assumption.

Design of timber structures - Volume 1 181

6.7 Bracing of timber framed buildings

For other panel products such as gypsum boards or cement based

panel products the behaviour of the connections may be more brittle
and application of models based on plasticity cannot be guaranteed.
Models based on elastic behaviour of the connectors are also availa­
ble; see for example Carling (1992). Utilizing an elastic model means
that the racking resistance will be determined by high forces in a few
connectors near the corners, that is a more conservative design than
that obtained from the plastic model.

Sport Center, Ostersund Arena, Ostersund.

182 Design of timber structures - Volume 1

 7.1 Performance requirements

Design for serviceability (SLS)

Annika Martensson

Unlike strength considerations it is very rare that serviceability con­ 7.1 Performance requirements 183
siderations, such as functionality and appearance as well as comfort
7.2 Static deformation 185
of the users, alone can lead to the collapse of a component or struc­
7.2.1 Deformation 185
ture but they are nevertheless very important in design for a number
7.2.2 Long term and climate effects 187
of reasons, for example: 7.2.3 Load combinations 189
• to keep the visual appearance and functional requirements of 7.2.4 Limitation of deformations 190
the component or structure within acceptable limits 7.2.5 Methods to avoid deformation problems 192
• to prevent damage to brittle finishes
• to prevent undue deflection of roof structures so that for instance 7.3 Vibration 193
7.3.1 Dynamic loads 194
rainwater ponds can cause leakage through the structure
7.3.2 Response to dynamic loads 195
• to limit the effects of creep 7.3.3 Design approaches to limit
• to provide sufficient stiffness in the construction so that vibrations vibrations in timber floors 198
do not lead to problems, for example with comfort. 7.3.4 Avoidance of vibration problems 201

Some of the notations used in this chapter coincide with the ones
adopted in Eurocode but the discussion is general and describes
the phenomenon and the design methods that can be used in service­
ability state analysis.

7.1 Performance requirements

Deflections and movements in a building must be limited to avoid
damage and other undesirable effects in service. A building must
have sufficient strength to carry the maximum expected loads with
an adequate factor of safety. Structures designed in the ultimate limit
state may often be relatively slender, which means that they may be
prone to large deformations. Therefore, the serviceability criteria
used in design are of great importance, especially for lightweight
structures like timber.
Serviceability requirements will in many cases be decisive in
the design process. This means that it is important that the magni­
tude of the deformations can be predicted with adequate accuracy
and that requirements and limits are correctly formulated. This is
often a problem in the design process today since the knowledge is
lacking with regard to both requirements and prediction methods.
Timber is a material exhibiting relatively large deflections when
subjected to sustained load. Since wood is an anisotropic material,
different loading modes will lead to different types of response. Its
sensitivity to environmental changes, for instance moisture induced
movements, has also a significant effect on the deformations in struc­
tural timber systems. Both long term deformation (creep) and envi­
ronmental effects must be considered when serviceability behaviour
of timber structures is studied.
In principle, serviceability requirements should be set by the client
in agreement with the builder and not by codes. The codes give, how­
ever, general principles for loads, material parameters and calculation
methods which may be used for design in serviceability limit states.

Design of timber structures - Volume 1 183

7.1 Performance requirements

Excessive deflection of structural components can be manifest in

damage to structural as well as non-structural elements, in detrimen­
tal effects on use and in various effects perceived as disagreeable by
the persons occupying the building.
Deformation of a structural element seldom leads to failure of that
single element, with the exception of deformations connected to
the stabilising system, see Chapter 6, page 160. If a number of elements
exhibit rather large deformations or if the deformation in one element
leads to subsequent movements in other elements, this may change
the primary structural system or cause instability of the whole struc­
ture. Such system effects can also lead to unacceptable effects from
the serviceability point of view. It is therefore of interest not only to
investigate the behaviour of each element, but to consider the whole
system, including the connections, in a deformation analysis. Such
an analysis is, however, comparatively complex and is therefore only
undertaken in very special cases.
Deformations may be unfavourable for the use of the building in
many ways. Examples are insufficient slope on roofs and problems
with opening doors and windows. In special buildings such as gym­
nasiums or facilities with high requirements of surface planeness for
example due to sensitive equipment, deflections may also lead to
troubles.
Serviceability limits should also be applied to buildings and their
components to avoid nonstructural elements being introduced into
the load path. If there is insufficient separation between deflecting
primary structural elements and non-structural components, the load
may be transferred through the non-structural element. Such
non-structural elements may include walls, windows, ceilings, floor
and roof coverings, facades, lifts etc. The resulting damage may
include cracking and buckling of some elements of moderate stiff­
ness and integrity, or tearing and folding of more flexible items such
as coverings and linings. Whereas these forms of failure in general
do not affect the stability of the structure, they may render the build­
ing unserviceable. Limitations in order to prevent damage to
non-structural elements should in many cases be determined on
basis of variable loads, with respect to the higher values that these
may attain during shorter periods.
Excessive deflections do not look attractive to the public and may
lead to a feeling of insecurity. These observations are subjective, and
although the deflections will not endanger the structure or shorten
its lifetime, it is appropriate to limit deflection to ensure that confi­
dence is maintained. Such limits apply especially to long term deflec­
tions and not so much to shorter but recoverable deflections that
may occur due to high load peaks.
It is evident from the previous discussions that limitation of defor­
mation is relevant in many different situations and for various more
or less important reasons. In practical design it is convenient to
define two principally different categories of deformation control:
• to avoid permanent damage (irreversible effects)
• to ensure a good appearance and general utility (reversible
effects).

The designer may then refer each specific situation to one of these
categories.

184 Design of timber structures - Volume 1

 7.2 Static deformation

Problems with regard to serviceability include also vibrations of dif­

ferent origin. Vibrations due to foot-fall and/or machines of different
types lead to human discomfort but may also lead to problems with
sensitive equipment as well as structural behaviour. The requirements
in those cases are often even more complex than with static deflection.
The same discussion as with the static deformation is, however, valid
in many cases since it is a matter of usage of the structure that is
decisive for the acceptable vibrations.

7.2 Static deformation

For most structures the loads consist of permanent loads Gi and varia­
ble loads Qi. For timber structures where the variable loads often
dominate the deflection will fluctuate to a great extent during Deflection

the lifetime of the structure.

Figure 7.1 illustrates in principle the deflection behaviour of
a beam loaded with permanent load, G, and snow load, Q. The total
deflection can be subdivided into one part w1 due to permanent load
immediately after loading and one part w2 which is variable during
the lifetime of the structure, Martensson (2003). The variable part w2
consists of a reversible portion w2,inst which is present only during
those periods when the variable load is present, and a continuously
increasing portion creep, which for all practical purposes may be
considered as irreversible. Load peaks with short duration, such as
those illustrated in figure 7.1, occur both for snow load and imposed
1 2 3 4 5 6 46 47 48 49 50 Year
(live) loads in the most common types of buildings.
Figure 7.1 Time variation in principle for deflection
(lower figure) of a beam with permanent and variable
7.2.1 Deformation loads according to the upper figure. Curve A shows
the creep deflection if the beam is loaded with
In figure 7.2 a schematic figure of the deformation for a timber beam
the characteristic loads Gk + Qk during the whole period.
is given with the different deflection parts. It can be assumed that
the beam may have a pre-camber (which can be the case for a glulam
beam), wc. During short periods, there will be high load peaks, lead­
ing to a large deflection winst. This deflection may occur any time dur­
ing the structure’s service life.
One part of the load will be sustained for a longer time period and
will give rise to a gradual increase in the deflection, wcreep. In order to
determine the total deflection after long time, wnet,fin, the sum of wc,
winst and wcreep can be calculated.
In the calculations of the deformations the value on the modulus
of elasticity and in some cases the shear modulus is used. It is most
reasonable to use the mean values of these moduli in the serviceabil­
ity state.
When a member is subjected to bending, in addition to deformation
due to the effect of the bending moment it will also deform due to

net, fin

Figure 7.2 Definitions of deflection

Design of timber structures - Volume 1 185

7.2 Static deformation

the effect of the shear forces, and the significance of the shear defor­
mation will primarily be a function of the ratio of the modulus of
elasticity E of the member to its shear modulus G and the depth to span
ratio, Porteous and Kermani (2007). Consider, for example, a simply
supported rectangular beam of depth h and design span l carrying
a point load at mid-span. The ratio of the instantaneous deflection at
mid-span caused by the shear forces, winst,v, to the instantaneous
deflection at mid-span caused by the bending moment, winst,m, will be:

Winst,v J ; £ f
7.1
Wmstjn ’ G UJ

With softwood for structural purposes, E/G is approximately 16 and

for practical beam design, h/ l will range between 0,1 and 0,05 result­
ing in a shear deformation between 5 and 20 percent of the flexural
value. This shows that in some cases it can be of interest to take these
deformations into account when designing timber structures.
One special case of deformation is settlement of structures. This is
often treated as an ultimate limit state design problem by checking
the compressive stresses in contact areas where the timber is com­
pressed perpendicular to grain, see figure 7.3, Martensson (2003). It is,
however, in practice often a serviceability problem since the settle­
ments of the joints can be a problem rather than actual failure of
the joint. In a multi-storey wood frame building, high magnitude
forces can result in the lower floors and at the foundation level.
The compressive forces perpendicular to grain lead to deformations
which may become relatively large especially in comparison with
deformations parallel to grain.
In order to estimate the magnitude of the compression of a joint it
is important to observe that the whole joint design has to be consid­
Figure 7.3 Platform frame construction
ered. See for instance a joint like the one in figure 7.4. If the bearing
Circles indicate details with high compressive stress length covers the whole surface of the ground plate, that is le = 0, it
perpendicular to grain. would be reasonable to assume that the deformation A would be
equal to:

A-^
7.2
E,

where d is the ground plate depth, E± is the elastic modulus perpen­

dicular to grain and the stress is given by:

7.3

where b is the ground plate width, l is the length of the loaded surface
and P is the applied load. Experiments have shown that in the case
where the load does not cover the entire surface the deformation is
less than that defined by equations 7.2 and 7.3.
The shorter the bearing length, the less deformation is obtained for
maintained stress level, which can be seen as an increase in stiffness
with decreasing bearing length. Another factor that affects the behav­
iour is the influence of different ground plate depths, d. Directly under
the surface of the loading plate (or stud) the stress will be equal to
the value given by equation 7.3. If it is assumed that the ground plate
is resting on the whole bottom surface, the stress would be less at
that boundary than at the top boundary, since the loaded area is
Figure 7.4 Typical layout of a bearing joint larger there. This would mean that the deformation over the ground

186 Design of timber structures - Volume 1

 7.2 Static deformation

plate depth is not as large as a calculation based on equations 7.2 and

7.3, page 186, would indicate. This can be considered in estimations of
the deformation by introducing an effective compressed area Aeff =
pbl instead of bl in equation 7.3, with p > 1,0 for all cases where le ^ 0.
There exists different formulations for p, but in all cases they depend
on le and on d. Tentatively a value on p of 1,5 — 2 can be used for
cases where the stud is not close to a plate end. In the latter case
a lower value on p should be used, Martensson (2003).

7.2.2 Long term and climate effects

Creep behaviour in timber and wood-related products, see for exam­ Figure 7.5 Schematic description of creep effects in
ple figure 7.5, is a function of several factors, and there exists complex a timber beam for different moisture contents

models that can be used to calculate deformations. The higher

the moisture content the larger is the creep, that is the increase in
deflection with time.
In practice there is a problem also with varying humidity which
also influences the behaviour, see figure 7.6. Variations in moisture
content, even at relatively low moisture levels, lead to larger increases
than at constant high humidity. Therefore it is important to try to
estimate the moisture content levels and variations in the surround­
ing during the lifetime of the structure.
In addition to the effects of pure creep under constant load and
creep under varying humidity, often called mechano-sorptive creep,
there is the effect of varying load levels during the structures lifetime.
All these effects have to be taken into account in the design process.
In principle, timber subjected to compression perpendicular to
grain exhibits the same behaviour in terms of deformations and time
effects as a beam subjected to bending, but with some additional
effects. The deformation in this case is further increased by wood
shrinkage perpendicular to grain which is far more severe than
shrinkage parallel to the grain, which is normally negligible.
The creep effects are also more severe than for a beam subjected to
bending. Figure 7.7, page 188, shows examples of deformation in
a joint with the layout shown in figure 7.4, page 186.

-------- LVL
-------- Glulam
-------- Structural timber
-------- I-beam

Figure 7.6 Relative creep in outdoors sheltered environment

(Relative creep = beam deflection at time t / initial deflection) Stress level 2 MPa.
Ranta-Maunus, Kortesmaa (2000).

Design of timber structures - Volume 1 187

7.2 Static deformation

Figure 7.7 Deformation of a joint where timber is subjected to constant compression perpendicular to grain. Same stress level 2,6 MPa in all cases.
The joint had been conditioned to 60 % RH before loading while it was exposed to RH = 30 % during the whole loading period, Martensson (2003).

sp = one stud and one plate,

spp = one stud and two plates,
sppp = one stud and three plates,
Figure 7.7 shows that there is a significant increase in deformation
ssp = two studs and one plate,
sssp = three studs and one plate,
during the two week test period. The pure shrinkage (60 % RH to
ssssp = four studs and one plate. 30 % RH) was about 1/ 3 of the additional deformation increase after
the initial deformation, that is the major part of the increase in
deformation was due to creep. Figure 7.7 also shows that there is
a non-linear increase in deformation with increasing depth d, as
mentioned in section 7.2.1, page 185.
The complex models of the behaviour of wood subjected to varying
load as well as varying climate are not feasible for design in practice
and instead more simplistic models can be used. In the following
a simplified method to consider creep effects is described:
For a member subjected to a constant load over the lifetime,
the instantaneous deflection (winst) and the creep deflection (wcreep) are
related as follows:

^creep _ ^def^inst 7.4

where kdef is a deformation factor whose value is dependent on

the loaded material as well as its moisture content and variations in
moisture content.
For structures or members complying with the above conditions
the final deformation, wfin, can then be written as:
Wfrn = Wrnst + WCrcCp = ^(1 + ^f) 7.5

The final deformation under permanent and variable loading will

then be as follows:

• For permanent actions, G, where the load is constant in time:

Wfin,G = W>nst,G + Wcrcep,G = +*

drf) 7.6

• For variable actions, Qi:

Wfin,Q, = W,nst,Ql + WcreepjQi = W^q/1 + 7.7

The value of kdef is decided based on climate conditions and type of

material, see also section 2.4.7, page 40. A number of experimental
studies have been made in order to estimate values for this factor and
examples of values can be found in handbooks and codes. Different
service classes are usually defined in order to classify the environ­
mental conditions that the structural elements are subjected to.
The value of kdef will increase with increasing humidity in the sur­
rounding climate. It is then important to remember that not only

188 Design of timber structures - Volume 1

 7.2 Static deformation

the maximum or average values of the relative humidity are of inter­

est but also variations in humidity. If it is reasonable to assume that
large temporal variations in humidity will occur, it could be an idea
to use a more severe service class than if only the maximum humid­
ity value is used.
Values of kdef are based on results from creep tests under constant
load, and the uncertainties in the figures are relatively large. Since
creep effects also depend on the time period that the load will be sus­
tained, the factor y2 is introduced in equation 7.7, page 188, to describe
this effect (y2 = 1 for constant load in time). For a time variable load
Q(t) with characteristic value Qk the reduced load y2 • Qk can be seen
as the time average of the load Q(t). In Eurocode, this is referred to as
the quasi-permanent value, and values for the factor y2 are specified
for different types of variable loads.
In the case where a structure consists of elements/materials with
different creep behaviour special considerations have to be taken. In
these situations, the creep behaviour will affect the stiffness and
stress distribution and a linear relationship between the instantane­
ous and the creep deflection described earlier will not apply even for
low stress levels. This is described in Chapter 5, page 115.
For the final deformation analysis the loading will be the same as
that used for the instantaneous deformation and the creep effect on
displacement behaviour is achieved by using reduced stiffness
properties:
p Timber storage, Norrkoping.
p _ mean 7.8
^mean^n ,
1 + 'Mef

where Emean,fin is the final mean value of the modulus of elasticity,

Emean is the mean value of the modulus of elasticity and kdef is
the deformation factor for timber and wood-based products.

7.2.3 Load combinations

When calculating the deformation according to the previous sections
one has to define the relevant loads to be used. Three different load
combinations can be defined in order to define relevant design situa­
tions. The load combinations for serviceability limit states defined in
EN 1990 are given below for the general case with several variable
loads Qk,i.

Characteristic combination:

ZGkj+ek,i+X^oAj 7.9
ya ;>i

where yQ is the combination value of the variable load. This com­

bination gives a high value on the load and should normally be used
in cases where excessive deformations may lead to irreversible effects
or permanent damage.

Frequent combination:

£Gkj+nAi+5>2A,i 7.10

JU i>l

where y1 1 Qk1 is the frequent value of the load Qk1 and y2i Qki is
the quasi-permanent value of the variable load Qk,i. This is the combi­
nation to be used for the assessment of effects that are reversible,

Design of timber structures - Volume 1 189

7.2 Static deformation

that is they occur with a certain frequency but the effects will be
reduced when the load decreases again.

Quasi-permanent combination:

7-11
j>l i>l

The value y2iQki is the quasi-permanent value of the variable load

Qk,i. This is the combination to be used for the assessment of long­
term (creep) effects.
If equations 7.5 - 7.7 and 7.9 - 7.11 are adopted the final deformation
wfin under permanent and variable loading will be as follows, when
the serviceability design is made on the basis of the characteristic
load combination, equation 7.9, page 189:

For permanent actions, G:

Wfin,G = + Wcreep,G = ^inst.oO + *drf) 7-12

For the leading variable action, Q1:

Wfin,Q1 = Wtast,Q1 + = Winst>Qi (1 + ) 7-13

For accompanying variable actions, Qi:

Wfin.Q, = Winst;Qi + ^cnxp.Q, = ^insl.Q, (’/'o.i + F2^drf) 7’14

The final result for permanent load and n variable actions will be:
n
Wfln = WfinjG + ^fin.Q, 7-15
i=2

It is important to notice that this load combination will give high val­
ues for the total deformation since the deformation from the variable
load is calculated based on the characteristic value. This is reasonable
if the control of deflection concerns permanent damage, since
the effect of the high loads occurring for short time periods are then
of interest.
In the case of control with respect to appearance and utility it can
be more suitable to calculate the deflection based on the frequent
load combination equation 7.10, page 189, or the quasi-permanent load
combination equation 7.11. The formulas 7.13 and 7.14 for calculating
the final deformation from variable loads will then be modified.
Using for instance the frequent load combination as a basis we get:

For the leading variable action, Q1:

^fm.Q! =Winst,Q1 + W'crcep,Q1 = W'inst,Q1 7 ^1,1+ 7-16

For accompanying variable actions, Qi:

Wfin,Q, = Winst>Qi + WcreepQi = WinstQi (</2.i + l^def) 7’17

7.2.4 Limitation of deformations

Any deflection limit which is set can be specified due to functional
reasons or purely for visual reasons. For example, it has been found
by experience that beam deflections not exceeding l / 300 are usually

190 Design of timber structures - Volume 1

 7.2 Static deformation

acceptable. This is often used as a value when designing in

the serviceability state. It is, however, important to consider why
the control is done and on what type of structure and for which
loads. If for instance the beam is cambered to off-set dead load deflec­
tion, the deflection limit applies to movement under imposed load­
ing only. For beams spanning a short distance (say less than 2 m) it
seems rather unnecessary to be too pedantic about observing this
kind of deflection limit, Ozelton and Baird (2006), Porteous and
Kermani (2007).
In the design of beams for storage and for roof structures, it is usu­
ally acceptable to permit deflections in the order of l /200 — l / 150.

The following points should be considered in the decision of

the deflection limit:
• the span
• the type of structure and the usage
• the risk of damage to the ceiling or covering material
• aesthetic requirements
• the number of times and length of time when maximum
deflection is likely to occur
• roof drainage Connection detail in the glulam frame of a bandy hall, Nassjo.

• the effect on such items as partitions over or under the position of

deflection.

For the different situations it is important to define the problems

that may arise due to deflections and whether they may lead to prob­
lems with regard to appearance or damage. Some examples are given
here on combinations of limit values, load combinations and calcu­
lated deflection values:
• A non-bearing wall below the deflected beam that may be dam­
aged if the deflection of the beam is too large. Determine
the required value on free space between the beam and the wall
and use that as a limit. The design value of the deflection can be
calculated with the characteristic load combination in equation 7.9,
page 189, and formulas 7.12 - 7.15, page 190. This can be seen as
an example of control with respect to permanent damage.
• A floor in a living room without any sensitive materials and any
risk for damages to adjacent structures can be checked with regard
to long term loads only. Large values of deformations for a shorter
time period will not lead to any problems. In this case it can be
enough to calculate the deformations with a quasi-permanent load
combination, that is:
n
Wfm,qp — ^inst,G ^creep,G ^creep,Q, —
j=l
n
^’mst.G (1 + *def 7 + 7.18
i-1

• In some cases it can be more relevant to control the long term

deflections only, that is the extra deflection that occurs with time,
omitting the instantaneous parts in the control. This can be
the case when the beam is cambered so that the instantaneous
effect of the permanent load is taken care of. And when the con­
trol is made with respect to appearance and utility and not
damage.

Design of timber structures - Volume 1 191

7.2 Static deformation

• For cases where it is of interest to control deformations connected

to bearing stress these should be based on a serviceability load
combination. Since the elastic deformations have been shown to
be only a minor part of the deformation in most cases, this means
that short term higher load levels will not affect the magnitude of
the deformation significantly. The deformations depend to a great
extent on the long term behaviour and it is therefore reasonable to
base the calculation on a quasi-permanent load combination. It is
then reasonable to use higher values on the factor kdef than for
deflection controls since the creep perpendicular to grain is higher
than along the grain. In this case it can also be important to esti­
mate the free shrinkage in the compressed part.

7.2.5 Methods to avoid

deformation problems
In practice it is impossible to avoid deformations when constructing
with traditional building materials, as for example timber. But with
better prediction tools it ought to be possible to avoid deformations
of such a magnitude that they may cause problems. In addition to be
able to predict the magnitude of the deformations during the service
life it is of interest to find design solutions that lead to smaller defor­
mations. For beams it is important to reduce the humidity variations.
This means that one has to be observant on the climate conditions in
the building where the timber is to be used. One especially important
aspect of the climate is if the timber structure is situated in a struc­
ture where the climate is different on different sides of the structure,
as in the case with a climate separation structure. Different climate
on the different sides may increase the problems with deformations.
For joints subjected to compression it is important to avoid or
reduce compression perpendicular to grain since this is the most sen­
sitive direction both with regard to load and to effects of moisture
----- Stud
variations. In figure 7.8 two examples of joints are presented where
Sole plate
only a minor part of the timber is subjected to compression perpen­
Top plate -— dicular to grain. There may also be problems if different type of
Stud------ Floor joist
materials is used in the building. If, for instance, a beam is situated
on a concrete wall in one end and on a timber structure in the other,
Edge support beam
this often leads to inclination of the beam. In this case one may
introduce a wooden piece between the concrete and the beam so that
deformations of the same magnitude will occur in both ends.
Another method to avoid the negative effect of moisture variations
in timber structures and thereby decreasing the deformations is to
Floor joist
use some sort of coating that makes it more difficult for the moisture
to move into the timber. This is not practical for most cases since it is
laborious and uneconomical.
It is also of major importance that the engineer really focuses on
the question: what is the acceptable serviceability level? In many cases
Figure 7.8 Examples of joints that can be used the designer uses very simple rules of thumbs for the limit values,
in order to reduce the problems with compression which in some cases are adequate but in some cases are too crude.
perpendicular to grain

192 Design of timber structures - Volume 1

 7.3 Vibration

7.3 Vibration
Serviceability requires consideration of the comfort of the user, and
into this comes dynamic effects such as vibration when persons walk
across a floor. This is a complex topic since it relates to the mass of
the floor structure and the actual arrangement of structural members
in the floor providing lateral distribution of applied dynamic load
and damping of the generated vibration.
Users of buildings and other engineered structures sense low-fre­
quency motion in three ways:
• acceleration causes forces on the body that are felt by the balance
organs
• visual cues (for example movement of objects resting on or hang­
ing from the structure relative to the observer).
• audio cues (for example creaking or rattling created by motion of
the structure).

Surveys for domestic buildings indicate that audio cues are very
important to perceptions of how well floors perform, Smith (2003).
Human induced vibrations in structures are almost always a prob­
lem of serviceability, in that they are a source of annoyance to
the users. In some instances the person experiencing the motion is
also the cause of it, while in other instances it is activities of others
that cause annoyance. Thus, the activity of the person experiencing
the vibration is important. When a person walks across a floor he or
she will tolerate much larger amplitude vibrations than when sitting
quietly resting, reading or writing. Categorisation of human percep­
tion, and tolerance, needs to reflect both the activity being under­
taken and the relationship between the source and the sensor. In this
respect, the following definitions are used (Smith, 2003):
• Springiness of a floor is associated with the sensation of self-gen­
erated floor deflection and vibration from a single footstep during
the time of contact between foot and floor surface.
• Vibrational disturbances caused by foot-fall on a floor are charac­
terised by perception of floor vibration induced by other persons
than the one that is disturbed.

Springiness is a problem usually only associated with lightweight

floors or those that are flexible under concentrated load. Such floors
are common in light-frame timber construction, and in other types of
buildings having timber joisted floors. In terms of the response of
a floor system, springiness encompasses static flexibility and impul­
sive velocity response, while vibrational disturbance encompasses
impulsive velocity response and stationary vibration response. It is
mainly the latter one that theoretical models can describe.
It is well established that humans are most sensitive to vibrations
in the frequency range 4 to 8 Hz, which is the range of vibration for
human internal organs, and thus researchers agree that structural
frequencies in that range are undesirable. Human sensitivity to
low-frequency accelerations is highly variable and depends also on
situation — are they working, sitting down, standing up, resting etc.

Design of timber structures - Volume 1 193

7.3 Vibration

7.3.1 Dynamic loads

7.3.1.1 Human induced loads

According to the various load sources and applicable countermeas­
ures, structures affected by human induced vibrations can be
grouped as:
• residential buildings
• office buildings
• industrial buildings
• pedestrian structures (footbridges, walkways in shopping malls)
• gymnasia and sports halls
• dance and concert halls.

There are many variations of rhythmic body movements leading to

a large variety of dynamic loads. Activities generating synchronised
rhythmic movements such as those due to several or more people
dancing or exercising are especially problematic. Several people act­
ing synchronously for 20 seconds or more can lead to approximately
periodic loads that produce almost steady state structural vibration.
Forces from human motion depend upon many factors including
the characteristics of the person(s) involved, the activity being under­
taken (for example walking, running, jumping), the number of peo­
ple, whether activities of different people are coordinated, and
the characteristics of the floor surface. Annoying vibration of timber
floors is commonly associated with walking and running forces.
Vertical forces due to an individual foot impact have been measured
and characterised, figure 7.9. As can be seen, there are two peaks in
a force-time history with the first corresponding to heel strike contact
and the second to toe-lift off contact. Flexibility of both footwear and
floor coverings influences forces. The peak force is much higher, but
the duration much shorter, for running than for walking.
The response of short or medium span floors is a mixture of low-fre­
quency forced vibration and highfrequency resonant vibration.

194 Design of timber structures - Volume 1

 7.3 Vibration

7.3.1.2 Mechanical equipment

Powered mechanical equipment usually exerts reciprocating or har­
monic loads on floors (for example industrial machinery, washing
machines, spin dryers). Depending upon the nature of the equipment
and the orientation of the axis about which moving parts spin,
induced motion can be horizontal, vertical or a mixture of the two.
Ideally, mechanical equipment should be isolated from floors used by
humans or premises where sensitive equipment is housed. If that
cannot be done, attention should be directed to avoiding resonance.
Except for some reciprocating machinery, resonance is usually auto­
matically avoided because the excitation frequency is much higher
than frequencies of the most energetic structural modes. Audible
effects (noise) associated with operation of mechanical equipment
heightens human perception of and sensitivity to motion, and should
therefore be minimised.

7.3.2 Response to dynamic loads

From a structural standpoint, a timber floor can be treated as
a two-dimensional thin plate structure reinforced with a series of
beams. Typically, this two-dimensional system is simplified as
a one-dimensional beam structure for design under specified live and
dead loads as in the previous sections. The situation for design against
annoying floor vibrations is a far more complex issue. Performance is
then influenced by stiffness, mass, damping and the socalled two-way
action of the floor structure. In many cases, the static stiffness prop­
erties of timber floor systems are adequate to ensure satisfactory
vibration performance. In some instances however, floor systems
designed merely to meet traditional deflection criteria under uniformly
distributed loads have been found to exhibit vibration problems.
New construction practices have had a profound impact on
the vibration characteristics of some timber floors. Amongst these is
the use of prefabricated engineered wood joists, concrete toppings
and floating floors. The availability of engineered wood joists and
trusses has led to more long-span and continuous multi-span floor
systems, while the use of a concrete topping has dramatically altered
the mass characteristics of timber floors.
There exist a number of different design methods with regard to
vibration problems and a number of studies have been made to
improve the methods, Hu et al. (2001); Brunskog et al. (2006).
Significant progress has been made towards achieving a better under­
standing of the vibration parameters influencing human perception
of vibrations, development of design approaches for control of floor
vibration, modeling floor responses to static and dynamic loads, and
understanding the influence of components and construction details
on vibration performance.

7.3.2.1 Factors affecting human response

to floor vibrations
Numerous efforts have been made to identify factors affecting human
response to vibrations. Both laboratory studies and large-scale studies
on floors in situ have been performed. In both cases subjective assess­
ment by persons together with performance tests are the evaluation
factors. In the context of residential construction, floor vibration ser­
viceability relates to human response to vibration effects excited by

Design of timber structures - Volume 1 195

7.3 Vibration

human activities. The major cause of the excitation is the motion of

the occupants on the floors. Essentially, this type of excitation con­
sists of a train of impacts, each causing a transient vibration response
characterized by its decay with time. Depending on the intervals
between successive impacts and damping of the vibration, adjacent
transient vibration responses may interact with each other. Ohlsson
(1984) suggested that this interaction can be significant and should
be accounted for in any design method. This differed from other
researchers who considered response from a single footfall force.
Research has shown that continuous excitation caused by walking
action is critical for floor systems with fundamental natural frequen­
cies lower than 9 Hz. This is due to resonance effects caused by walk­
ing motions in assembly occupancies. However, fundamental natural
frequencies of timber floors are rarely below 9 Hz, so resonance rarely
occurs in residential occupancies. Analysis of transient response to
a single footstep force may therefore be adequate in evaluating vibra­
tion properties of timber floors.
When considering human response to transient vibration in a floor
system, it is parameters such as frequency components, magnitude of
response and damping of the vibration that are the most important
factors.

7.3.2.2 Frequency components

Floors are usually divided into low-frequency floors and high-fre­
quency floors. Low frequency floors have a fundamental frequency
below about 8 Hz, and high-frequency floors have a fundamental fre­
quency above about 8 Hz. Low-frequency floors are generally heavy
structures such as concrete floors. This classification of floors into
low-frequency and high-frequency ones has its origin in the different
responses of the floor types to human walking. For lowfrequency
floors the low frequency parts of human walking (the continuous
parts) are the most important because they cause a resonant response
of the floor. This means that a person staying still may feel this reso­
nance vibration. A high-frequency floor is more responsive to
the impulsive parts of human walking. In this case a person standing
still might feel the impacts caused by another person walking by, and
the walking person might get a feeling of springiness.
For a floor system excited by a footstep force, the transient vibra­
tion response is of two types, see figure 7.10, page 197. The initial
forced vibration due to the heel strike force component of the foot­
step lasts for typically 30 — 100 ms. This is followed by free vibration
that decays with time. The free-vibration response contains the fre­
quency components coinciding with the natural frequencies of
the system. The frequency range of vibration modes excited depends
on the duration of the footstep force. The shorter the duration,
the broader the frequency range. Only the first few modes can be
excited effectively in a typical timber floor by the footstep force from
a normal walking person.
Most researchers agree that, for a given level of response and
damping, the lower the frequency, the longer the vibration will per­
sist above the threshold of perception, that is the low frequencies are
of major concern, Hu et al. (2001). However, Ohlsson (1984) has
reported that, in addition to the fundamental natural frequency,
the spacing between the first few adjacent natural frequencies can
also contribute to human discomfort.

196 Design of timber structures - Volume 1

 7.3 Vibration

Figure 7.10 A time history response of timber floor caused by a footfall impact, depicting the forced and
free vibration components; duration of the forced vibration section matches that of the impact duration

The fundamental natural frequency f1 of a two-way structural system,

such as a timber floor, is governed primarily by the system stiffness
in the along-joist direction EIx, unit mass m and span l. Spacing of
two adjacent natural frequencies is controlled by the ratio between
the cross-joist direction stiffness EIy and EIx. The spacing increases
with increasing EIy. EIx is mainly governed by joist stiffness. Many
construction details, such as floor decking thickness and material,
attachment of floor decking to joists, joist spacing, and ceiling boards
contribute to the effective EIy.

7.3.2.3 Vibration magnitude

The transient response of floors to an impact can be represented by
a time history of displacement, velocity or acceleration. Researchers
have used quantities such as peak value of the response, or root­
mean-square (rms) value of the entire response history, as a measure
of human response to vibration. The peak value is extracted from
the initial part of the response (forced vibration) due to an impact.
However, the peak velocity used by Ohlsson (1991) is from a pure free
vibration response excited by unit impulse. The rms value is deter­
mined from the entire response, including the initial forced and free
vibration parts of the transient. Although both peak and rms values
are related to the characteristics of the impact and the floor properties,
the rms value includes the effect of the duration of the transient.

Design of timber structures - Volume 1 197

7.3 Vibration

7.3.2.4 Damping
Damping is a property which influences vibration amplitudes under
forced vibration and the rate of decay of vibration amplitudes under
free vibration. Damping relates to the dissipation of energy, or to
the conversion of the mechanical energy associated with a vibration
to a form that is unavailable to the vibration. Increased damping
results in rapid decay of a free vibration.
Many floor researchers have attempted to use damping to quantify
the decay of the transient vibration induced by footstep impact, and
hence correlate human response to damping. Despite these attempts,
the problem of how to make use of damping information is largely
unresolved because of the limited knowledge on quantification and
measurements of damping in floor structures having a multiplicity of
vibration modes. Damping values measured by researchers are global
values arising from various sources. Material damping and friction
between components are thought to be the major sources of damping
found in structural systems.
In timber floors, material damping depends on the materials used
for construction. The construction details such as the manner that
components are attached, and the boundary conditions at supports,
contribute to the frictional damping. The material damping is usu­
ally small. The major contributor to damping appears to be due to
friction, a factor that has been found in for example cases where
the presence of partitions on floors has provided exceedingly high
damping.
End conditions of bending members can greatly affect damping. It
has been shown that for timber beams with ideally simply supported
ends, the damping ratio associated with the fundamental mode is
about 1 percent, which corresponds approximately to the material
damping. When the same members had ‘clamped’ ends the damping
ratio increased to about 8 percent. For timber joisted floors effective
Karolinska, Solna.
damping ratios are in the range of 1 to 3 percent depending upon
the details of the floor and the mode being considered. Application of
imposed masses (objects) on the surface of a floor can greatly increase
the damping, especially if the system is lightweight or small.
Because of inherent low self-weight, timber floors often do not
exhibit a significant amount of inertial damping. Exceptions can
occur with large systems, especially if they have thick concrete top­
ping or support large amounts of imposed mass.

7.3.3 Design approaches to limit

vibrations in timber floors
The fundamental research in understanding the factors affecting
human response to floor vibrations has paved the way for develop­
ment of design approaches to prevent vibrations. Some examples are
presented here.

7.3.3.1 Limiting point load deflection

It has been shown that one possible method to predict human
response to floor vibration is to determine the static deflection under
a point load. A common method to do this is to determine the static
deflection under a 1 kN point load at the centre of a single beam in
the simplest model or under a two-way floor system. The deflection
limits used in these rather simple models differ depending on regula­
tion or handbook, but also on span length and aim with the design.

198 Design of timber structures - Volume 1

 7.3 Vibration

This approach can be seen as a method to model the effect of a step

action. Here one method to do this is presented: The static load simu­
lating the foot force effect is 1 kN applied at the centre of the floor
and the deflection of the floor at this point, a, must be no greater
than a certain limit value. The deflection of a point load is given by:

Pt3
a =------ 7.19
48£7

where in this case P = 1 kN. If the most simple approach is used equa­
tion 7.17 gives the value of the deflection with the flexural rigidity EI
for one single joist. In most cases however, this will overestimate
the deflection since the joist is only one element in a two-way
load-bearing system. In order to take this into account, the rigidity in
both directions of the floor can be used in order to calculate
the value of the deflection:

P£3
a=K 7.20
48EI
where k is a load distribution factor that can be calculated according
to the following:

-4,7y32+ 2,9)8+ 0,4 when 0</J<0,3

K= 7.21
0,8 + 0,2/3 when 0,3<j8<1,0
with:

Wt(^
P= m 7.22
W
where (EI) is the flexural rigidity of floor in the stiffer direction, that
is along-joist (Nm12/m) and (EI)B is the flexural rigidity of floor in
the direction perpendicular to the stiffer direction, that is across-joist
(Nm2/m), s is the spacing between the joists and l is the span of
the joists, Ohlsson (1984).
Note that other methods exist to perform this calculation and that
various limit values for a are also discussed.

7.3.3.2 Limiting point-load deflection

and peak-velocity due to unit impulse
The use of static response parameters such as deflection, while pro­
viding some control, does not always produce satisfactory perfor­
mance. Researchers are aware of this limitation, and recent research
has focused more on studying dynamic parameters. Among the first
to propose dynamic-based parameters for design was Ohlsson (1991).
A laboratory study was conducted involving the use of dynamic test­
ing equipment and limited subjective evaluation along with theoreti­
cal reasoning. Ohlsson concluded that to account adequately for
the important factors that affect human response to floor vibration,
two parameters should be checked for lightweight floors having nat­
ural frequencies above 8 Hz:

1. Static deflection limit under 1 kN load at floor centre (1,5 mm)

2. Peak velocity v due to “unit impulse of 1Ns” < 100[f1z - 1] (m/s) / Ns
where f1 is the fundamental natural frequency and Z is the damp­
ing ratio for f1.

Design of timber structures - Volume 1 199

7.3 Vibration

The first criterion is similar to that presented in section 7.3.3.1, page 198,
and Ohlsson (1991) stated that this is a control of the low-frequency
components (< 8 Hz) that are semi-static in nature. The second crite­
rion is required to limit the magnitude of the transient response due
to the heel impact of a footstep. As discussed in previous sections,
the impact excites higher frequency components in a timber floor
(8 — 40 Hz) and its response is governed by stiffness, mass and damp­
ing of the floor. Believing that contributions to the total motion from
vibration modes with natural frequencies higher than 40 Hz are
insignificant, Ohlsson (1991) proposed the following equation for cal­
culating the peak velocity due to unit impulse for a rectangular floor
system simply supported on all four sides, valid for f < 40 Hz:

4(0,4 + 0,6n40)
V = —i--------------- — 7.23
mB£ + 200
where n40 represents the number of eigenmodes with eigenfrequencies
Car park with glulam and CLT frame, Skelleftea.
lower then 40 Hz and is given by:
-1025
7.24

where B is the width of floor (m), l the span (m); m the mass per unit
area (kg/m2), (EI) is the flexural rigidity of floor in the stiffer direction,
that is along-joist (Nm2/m) and (EI)B the flexural rigidity of floor in
the direction perpendicular to the stiffer direction, that is across-joist
(Nm2/m).
Ohlsson (1991) also provides an equation for calculating the funda­
mental natural frequency of a floor:

,_2L 7.25
71 2£2 N m

Since the introduction of this method it has been used quite exten­
sively and has in many cases shown satisfying results, that is floors
designed according to this method have shown a satisfying
behaviour.
In the second criteria above the damping ratio Z has to be determined
which is a difficult matter. Ohlsson (1991) states that the value on Z
can be 1 percent but also states that higher values can be relevant.

7.3.3.3 Classification of floor vibrations

The methods described previously work well in some cases, but are
less adequate in other cases. The design methods give usually only
a limit value for floors and it is often unclear for the designers what
the limit value actually means. How much better is the floor actually
if the limit value is lowered with 50 percent? In order to get a better
description of the actual behaviour of a floor structure, Talja and
Toratti (2006) have introduced a method for classification of different
floors based on the response to vibrational excitations. They per­
formed subjective tests and measurements on timber-, steel- and con­
crete floors both in laboratory as well as in buildings just after con­
struction. Their research has resulted in a method for the assessment
of vibration serviceability of floors.
For high-frequency floors (f1 > 8 Hz) they have found that point
load deflection is the best indicator when designing with respect to

200 Design of timber structures - Volume 1

 7.3 Vibration

Table 7.1 Vibration classes in office and residential buildings (Talja and Toratti 2006)

A Special class for vibrations inside one apartment.

Normal class for vibrations transferred from another apartment.
The vibration is usually imperceptible.

B Higher class for vibrations inside one apartment.

Lower class for vibrations transferred from another apartment.
The vibration may be perceptible but usually it is not annoying (inside one apartment).

C Normal class for vibrations inside one apartment.

The vibration is often perceptible and some people may feel it annoying (inside one apartment).

D Lower class for vibrations inside one apartment.

For example attics and holiday cottages.
The vibration is perceptible and most people feel it annoying (inside one apartment).

E Class without restrictions.

Table 7.2 Tentative acceptance limits for vibration classes (Talja and Toratti 2006)

Class a
rms [m/s2]
Peak vertical velocity
3 [mm]
Global displacement of load bearing
member due to 1 kN point load
A < 0,03 < 0,12

B < 0,05 < 0,25

C < 0,075 < 0,5

D < 0,12 < 1,0

E > 0,12 > 1,0

floor vibration. For low-frequency floors (f1 < 8 Hz) the recommended
parameter to use is acceleration, because of its good correlation with
subjective ratings.
Talja and Toratti (2006) propose a classification of floors into five
classes, see table 7.1. This classification presumes human walking to
be the designing load case and it is valid for floors in residential and
office buildings. It is based on the sensory perception of a sitting per­
son and the human perception of vibrating objects.
This method has the advantage that it gives a tool where the designer
can be more aware of the actual floor performance and on the target
performance to be achieved. Talja and Toratti (2006) also gives tenta­
tive limit values on deflections and frequencies for the different
classes, see table 7.2.
Other approaches to modelling and prediction of the dynamic
response can be found in literature but for all of them it can be said
that there are uncertainties and that it is often difficult to find any
one model that can cope with all situations. All the design approaches
are semi-empirical in nature and provide satisfactory solutions for
the particular category of floors on which they are based. None
appears to work entirely satisfactorily when applied to other catego­
ries of floors. In many cases it can be said that the best knowledge of
the behaviour of a floor is still obtained by testing it.

7.3.4 Avoidance of vibration problems

Since the problems with vibrations are complex in nature it is also
difficult to give one recommendation that solves the problems. One
practical strategy is to try to design structural systems that have rela­
tively high natural frequencies, a method referred to as tuning
the frequency. The method means that the lowest and most energetic

Design of timber structures - Volume 1 201

7.3 Vibration

structural frequencies of the structures are higher than the excitation

frequency. The objective is to avoid coincidence between loading and
response frequencies.
Addition of extra material (increasing the stiffness) or reduction of
the span are methods that are used. Since there has been efforts made
to decrease the amount of material and increase span lengths due to
other aims in the design process, it is not surprising that reports of
vibration problems in structures have increased lately. It is normally
easier to increase the strength to weight ratio than to increase the stiff­
ness to mass ratio of construction materials.
Other means of avoiding “problem” floors are obviously available.
A potentially effective approach is to increase the damping (for exam­
ple add artificial damping or tuned mass dampers), but this is nor­
mally a complex and/or expensive solution. Attention to details such
as provision of adequate blocking or cross-bracing (see figure 7.11) and
to be observant of detailing and connections between different parts
of the systems are also important factors.
A special type of floors is floating floors and raised floors, Talja and
Toratti (2006). These types of superstructures are increasingly used
Flat roof made of glulam and CLT. because of impact sound insulation requirements and because of
the flexibility of mounting the installations. The vibration or move­
ments of objects, as clinking of glassware or leaf movement of plants,
are typical for these type of floors and it has been shown that such
effects are highly dependant on the flexural stiffness of the top sur­
face board. Local deflections are caused by soft floating floors and
this needs to be limited to avoid such disturbances. On the positive
side, floating floors with sufficient bending stiffness of the top layer
may effectively distribute point loads to various floor joists and thus
increasing the floor vibration performance.
Local deflections are most often difficult to predict based on engi­
neering calculations and it is recommended that these are determined
based on tests. The bending stiffness of the floating floor top layer
has a major impact on the local deflection.
It is also important to be observant to transmission of vibrations
between different rooms via the floor joists. With continuous beams,
vibration can be transferred from one apartment (or room) to another.
Even if no problems are experienced in the room where the vibrations
emanates the neighbors can experience nuisance from them. And it
has been shown that vibrations coming from a neighboring apart­
ment are more irritating than if the vibration source is in the same
apartment.

Figure 7.11 a) Plywood sub-flooring on timber joists. To the left with wood cross-bracing between joists and to the right with
blocking between joists. b) Plywood sub-flooring with concrete topping on wood I-joists (metal cross-bracing between joists),
Smith (2003).

202 Design of timber structures - Volume 1

 Timber building systems for housing

Timber building systems

for housing
Helena Lidelow

Timber buildings systems can be used to construct a multitude of 8.1 Structural systems overview 205
buildings. Single-family homes have a relatively simple structure, 8.1.1 Surface elements 205
while greater challenges meet the material when facing multi-storey 8.1.2 Prefabricated Box Units 207
or long-span buildings. Multi-storey timber construction in 8.1.3 Post and beam systems 208
the Nordic countries started around 1995 with the traditional on-site 8.2 Functional requirements and
construction utilizing the 2" x 4" (45 x 95 mm) platform system. system solutions 211
Relatively soon, prefabricated systems emerged. In the Nordic coun­ 8.2.1 Fire 211
tries, prefabricated systems dominate, while systems with a larger 8.2.2 Acoustics 213
degree of craftsmanship are used in central Europe. Timber building 8.2.3 Horizontal stability 214
systems are common in the Nordic countries and the German speak­ 8.2.4 Critical moisture level 216
ing countries in central Europe; Germany, Switzerland and Austria
8.3 Detailed design 217
and to some extent Italy. Also, timber building systems exist in
8.3.1 Wall elements 217
the United Kingdom, while timber is less used in countries around 8.3.2 Floor elements 220
the Mediterranean. Globally, USA use timber extensively for housing, 8.3.3 Wall-floor connections 222
the most common technology being a simple 2" x 4" frame stabilised 8.3.4 Services 224
by sheathing. New Zeeland and Australia also apply timber structures 8.3.5 Balconies 226
in housing. LVL is a common framing material both in New Zeeland, 8.3.6 Staircases and elevators 226
Australia and USA. Finally, Japan has a long tradition using timber in
structures. Oftentimes, the great ductility in a timber frame ensures
good racking resistance under earthquake loading.

Geometrically, building systems belong to one of the following

categories:
• Panel systems: are based on planar building elements, for example
walls or floors. These systems utilise either light frames or solid
wood elements as their basic technology. The panels can be full
size or partitioned in widths of 1 200 mm for ease of transporta­
tion. Floor elements are not suitable for long-span structures,
the limitation in span length is about 8 — 10 m. Roof elements
however, can have longer spans.

• Modular systems (volumetric modules): utilize the light frame sys­

tem mainly, but examples exist where solid wood elements apply.
The main principle is that the entire volumetric box consisting of
walls, floor and ceiling, as well as inner claddings and all services
are assembled in a factory and delivered to the building site for
erection. Modular systems are not suitable for long-span structures
due to transportation limitations.

• Post and beam systems: are sometimes named big frame structures
and are very common for industrial and commercial buildings.
These systems are also used for buildings or parts of buildings,
where the load-carrying structure is visible. The post and beam sys­
tem is a system based on a structural grid of beams and columns,
typically with pinned connections. The structure is instead stabi­
lised through diagonal bracing or shear walls. Post and beam sys­
tems are suitable for long-span structures and are used for arenas.

Design of timber structures - Volume 1 203

Timber building systems for housing

Building systems can also be classified according to their connection

with the industry sector as:

• Open systems: the specifications and building elements can be

combined and purchased by anyone. Suppliers manufacture their
elements based on an open standard of dimensions, which all sup­
pliers adhere to.

• Closed systems: the specifications and building elements are used

by the supplier only who makes his profit from the efficiency in
building with the system. Closed systems are suitable for integra­
tion of different functions in a company, for example design and
production.

A supply chain is a group of companies, which produce a product by

letting it pass through their production processes in sequence.
The building system is transferred to the customer or the next party
in the supply chain at the decoupling point. The earlier the decou­
pling point is placed in the supply chain, the more open the building
system. In Sweden there are currently no real open building systems
for modular systems or cross-laminated timber (CLT). There is, how­
ever, an agreement on dimensions prescribed in EN ISO 8560 and
most actors in the building element industry follow the standard and
produce elements in sizes of 900 mm, 1 200 mm or 2 400 mm (corre­
sponding to 9M, 12M and 24M). The dimensional standard also sim­
plifies both design and prefabrication, since main dimensions are
already negotiated and only special cases need to be addressed.
EN ISO 8560 is not used for arenas and other long-span structures.
Handbooks (Limtrahandbok Del 1-4, 2016) exist, presenting solutions
and details for designing glulam structures. With the help of
the handbook, a glulam structure can be realised regardless of sup­
plier input and glulam beam-and-post systems may therefore be con­
sidered open systems. In Finland and Denmark the development of
open building systems has lead to a component supplier market
where several suppliers compete with similar products. The Open
Timber Construction System (2003) was published in Finland to pro­
vide prerequisites for an open market with timber elements and it
describes detailed solutions for element lay-up and connections for
any supplier to use, produce, and sell. A strong governmental body is
needed to provide basis for open building systems, if the market is
left to regulate itself, closed building systems often result.
Technical design of the frame is always the core part of the con­
tract with the supplier of a building system, which also can include
the erection of the structure. Thus, the delivery includes much more
than a building material and the supplier has to create an organisa­
tion for this, in-house or through partnerships. Integrating large
parts of the supply chain in one company is beneficial since the pos­
sibility to optimise processes becomes larger if several processes are
controlled. The risk with integration is financial since control often
leads to investments in automation, which in turn builds up a fac­
tory that needs to be fed with a continuous flow of products to pro­
duce. The benefits seem to be larger since the majority of building
systems are closed or proprietary systems. This is true for most of
the building systems in Sweden and the majority of timber building
systems in Germany, Switzerland, and Austria. Patent rights for new
solutions is important in closed building systems.

204 Design of timber structures - Volume 1

 8.1 Structural systems overview

8.1 Structural systems overview

8.1.1 Surface elements
8.1.1.1 Light frame system
The light frame building system using joists and studs placed at even
distances with the voids filled with insulation and covered by sheath­
ing to stabilise the element was born in USA in the 1830’s. The rapid
Joists
expansion in USA was created by a large demand for housing and
combined with machine sawing and industrially produced nails,
the light timber frame became a success that has spread all around
the world and is still used extensively. The platform framing tech­
nique is still the most common in USA when constructing multi-fam­
ily houses, see figure 8.1. Lower-floor
In USA, light frames are commonly produced on site, but in top plate
the Nordic countries they are prefabricated in a factory. The light
frame technology is the backbone of single-family housing produc­ Ground plate
tion in the Nordic countries, where the approach to order a stock
house, have it produced and erected on site by the same company (a
closed building system) has been the norm ever since the 1950’s, see
figure 8.2.
With the introduction of glulam and LVL, the timber joists in a light Figure 8.1 Platform framing, McGraw-Hill (2011)
frame can be replaced with a stronger material, which enables longer
spans. Being so light, the frame is prone to vibrate and deflect, which
limits floor spans to 8 — 10 m. Another common problem with light
timber frame is their low mass, which can lead to poor acoustic per­
formance, see further section 8.2, page 211. The height limit using light
frame technology is currently 6 — 7 storeys. The compression strength
perpendicular to the grain for the ground plate limits higher build­
ings, unless the structure is strengthened by special methods.

Figure 8.2 The load-bearing structure of a detached single-family house

Design of timber structures - Volume 1 205

8.1 Structural systems overview

Figure 8.3 a) Structural system with cross-laminated elements. b) X-Lam, Massivtrahandboken (2006).

8.1.1.2 Solid wood systems

Solid wood systems are usually based on planar cross-laminated (CLT
— cross-laminated timber) elements, see figure 8.3, see also section 2.7.1.2,
page 49. The cross-lamination is established through gluing planed
and strength graded timber together, much in the same manner as
when producing glulam. It is possible to use small dimension timber
in the production of solid wood panels, which makes them attractive
for sawmills. The prefabricated elements are delivered to the build­
ing site, where the assembly can be done by the manufacturer them­
selves or another contractor, see figure 8.3.
Solid wood elements, such as cross-laminated timber, are most
common in Germany, Austria and Switzerland where they are also
used in single-family houses to obtain specific architectural expres­
sions. In countries with warmer climates, the wood element can be
applied without extra thermal insulation. This is not likely in
the Nordic countries, where the CLT frame often is completed with
a light frame with insulation in the voids.
De massiva traelementen kan aven byggas upp av brador eller
plankor som forbinds mekaniskt med aluminiumspik i stallet for
med lim. Dartill kan bradorna eller plankorna ha frasta spar pa ena
sidan sa att de fardiga vaggelementen har innesluten luft som okar
deras isolerande egenskaper sa att ingen ytterligare isolering kravs.

8.1.1.3 Engineered Wood Products

Engineered Wood Products (EWP) are mainly used as a substitute for
sawn lumber in light frame timber buildings. There are many differ­
ent types of EWP, see section 2.7, page 47, but the main types used in
building systems are I-joists with web of wood based panel, trusses,
and LVL with parallel or cross-wise glued veneer layers.
An example from Sweden is Masonite Flexible Building System
with I-joists built up by an OSB web or chipboard and flanges of
structural timber, see figure 8.4 a), page 207, where joints are strength­
ened with thick (20 — 40 mm) wood based panel. Also I-joists built up
by a hardboard web and flanges of LVL are present on the market.

206 Design of timber structures - Volume 1

 8.1 Structural systems overview

The intention is to use the elements for industrialised timber con­

struction with a high degree of prefabrication, but at the same time
this type of building system is relatively open and offers the possibil­
ity to be applied along with other types of construction methods,
such as building on-site, panel elements or beam and post structures.
Trusses of structural timber are effective as floor joists, see fig­
ure 8.4 b). The voids created by the truss are suitable for running ser­
vices. The truss might be too high for many applications, and the many
joints may infer large flexibility in the truss, leading to excessive
deflections for larger spans. The design must be performed carefully
to balance the benefits and drawbacks of a truss joist.
To address the problem with vibrations and poor acoustic perfor­
mance in light-weight floors, they can be completed with a concrete
topping, creating a composite floor element, see figure 8.5. Also con­
crete topping without interaction or fillings with for instance gravel
exist.
Composite timber-concrete floors are successfully used in Austria
and Switzerland with several variations, see for example Natterer et
al. (1996), and several designs and types of fasteners have been tested
to reach desired performance, see section 8.3.2, page 220.

8.1.2 Prefabricated Box Units

Figure 8.4 a) I-joists with web of wood based panel.
8.1.2.1 Light frame modular systems b) Floor truss joists.
An extension of the light frame building system is combining panel
elements into volumetric modules at a factory off-site, see figure 8.6.
Walls, floors and ceilings are produced using a light frame structure,
Concrete slab Shear connector Timber/glulam beam
using the same methods as when construction a single-family house.
The floors and ceilings are hung onto the walls constructing a volume.
Apart from the structural system, also claddings, services, and fur­
nishing (for example kitchen furniture and equipment) are mounted.
Thereafter, the modules are covered to be weather protected and
transported to the building-site for assembly.
To assemble the modules on top of each other, a helping device is
used in the form of a little dub (male-female connection). Between
the modules, polyurethane tape or Xylodyne is used for acoustic pur­
poses. Since this is a very expensive product, it is applied only on top Figure 8.5 Composite timber-concrete floor,
of the studs, not continuously along the rim. The floors hang from Lukaszewska (2009)

Figure 8.6 Volumetric element or module made with a light frame

Design of timber structures - Volume 1 207

8.1 Structural systems overview

the walls in modular systems since the outer walls of the module are
used as vertical load-carrying components.
Utilising this kind of system gives the opportunity to control
the whole building process from architectural and structural design
to completion. The architectural design of the buildings built with
modules needs to be carried out with knowledge of the building sys­
tem and the limitations and possibilities in the production system.
Modular systems are suitable for student housing, hotels and housing
for the elderly, but is also adaptable to condominiums or tenancies.
Each module can comprise one or a few rooms, or even the whole
apartment in the case of student apartments. The size of the modules
is determined by manufacturing, handling and transportation
limitations.

Figure 8.7 Assembly of modular elements 8.1.2.2 CLT modular systems

This type of system is based on modules of CLT (or semi-solid with air
cavities). The modules are entirely built and equipped in the factory
and transported to the building site, see figure 8.7. Using X-Lam
instead of light frame systems opens up the possibility for stiffer sta­
bilising walls.

8.1.3 Post and beam systems

Post and beam structures have always been very common for timber
structures. Through larger cross-sections of members and conse­
quently larger span lengths, the post and beam systems experienced
a renaissance with the development of engineered wood products
such as glued laminated timber and laminated veneer lumber.

8.1.3.1 Short-span structures

The post and beam structures can have various structural design and
many types of buildings fall into this category, see for example fig­
ure 8.8 a), based on a post and beam timber structure with composite
timber-concrete floors, also used in the first timber parking house in
Sweden, see figure 8.8 b).
A post and beam system for multi-storey timber buildings named
“Tra8” in timber was developed in Sweden by the glulam manufac­
turer Moelven Toreboda, see figure 8.9 a), and is built up by glulam
and Kerto (LVL). The system is based on rectangular division in plane,
with a maximum span length of 8 metres, Moelven Toreboda. This
gives at present a possibility to use the system for both residential
and non-residential structures such as offices. The company has
Figure 8.8 a) Block of houses, Uppfinnaren in Vaxjo.
an intention to increase the span length in the future, but this limi­
b) Car park Ekorren, Skelleftea. tation is related to the floor element height, which can be substantial
for timber structures.

8.1.3.2 Long-span structures

For long-span structures, glulam presents a very economical alterna­
tive. The main competitor is steel structures. The maximum span
length is often set to be 80 — 100 m, but much longer built examples
exist. The most economical structures built are however ice hockey
arenas with a span of 36 m. A recent trend in Sweden has been to
also build arenas for bandy, which need to have a much longer span,
roughly 70 m. Examples of structures for long-span structures are
shown in figure 8.10, page 210.

208 Design of timber structures - Volume 1

 8.1 Structural systems overview

Figure 8.9 a) The post and beam system "Tra8", Askims torg.
b) Hockey hall, Ekvallen.
c) Bandy hall, Nassjo.

Such long-span structures are not used for housing, which means
that most of the functional requirements in place for housing are not
viable for this type of structures. Acoustics is seldom a problem;
moisture safety and thermal insulation are currently poorly
addressed. Load-carrying capacity and fire are the most prominent
requirements to address for the structure.
To reach an economical structure, optimisation of the load-carry­
ing capacity is therefore a must, which includes efficient design of

Design of timber structures - Volume 1 209

8.1 Structural systems overview

10°

Single pitched beam Two or three-hinged arch

a = 3 - 10°

Double pitched beam

a

Double pitched beam with curved underside Three-hinged portal frame with curved haunches

a >14

Three-hinged portal frame with finger-jointed haunches

Three-hinged tied trussed rafter Three-hinged portal frame with knee bracing

Figure 8.10 Structural systems in glulam, Limtrahandbok (2016)

210 Design of timber structures - Volume 1

 8.2 Functional requirements
and system solutions

joints, see Chapter 4, page 87. Compared to short-span structures

intended for housing and offices, long-span structures are relatively
easy to optimise.
Apart from the main load-carrying structure (the primary system)
a secondary structure of beams might be needed between the pri­
mary beams. The secondary structure is made of beams out of glu­
lam, structural timber or LVL and the purpose of it is to support
the sheathing covering the building, see figure 8.11. Optimisation of
the building system must embrace both the primary and the second­
ary structure. Since the main functional requirement is load-carrying
capacity, the main indicator for efficiency is the volume of glulam
spent. Roughly 30 percent of the glulam volume is spent on the sec­
ondary structure.
Figure 8.11 Primary, secondary structure, and
roof sheathing

8.2 Functional requirements

and system solutions
All structural solutions for building structures need to fulfil the func­
tional requirements for its users. For timber structures the most diffi­
cult and economically decisive ones are:
• fire
• acoustics
• horizontal stability
• critical moisture level.

8.2.1 Fire
The European codes for fire safety in timber buildings have been
revised and handbooks published, Fire Safety in Timber Buildings
(2010). Buildings should limit the risk related to fire, limit the spread
of fire and fire gases inside the building, and allow the inhabitants to
escape from the building or to be evacuated or saved. The safety of
firemen is also taken into consideration.
The building fire documentation that is mandatory for all build­
ings should include: fire safety classes of used materials and ele­
ments, division into fire cells, evacuation plan, ventilation during
fire, and a description of used technical solutions, including any
additional installation like sprinklers or fire detection systems. It is
recommended to integrate fire safety design in the very early stages
of building design.
Structural elements, materials, claddings and surface finishes,
depending on their function are assigned one of the following prop­
erties, EN 1995-1-2:
• R — load-bearing capacity
• RE — load-bearing capacity and integrity
• REI — load-bearing capacity, integrity and insulation
• E — integrity
• EI — integrity and insulation.

Requirements related to these attributes are specified in terms of

time in minutes, under which a specific function shall be preserved
from the start of a fully developed fire. As an example for buildings
up to four storeys high, the fire safety class needs to be REI 60, while
for higher buildings the requirement is REI 90, where the number
indicates the time in minutes. It should be noted that the resistance

Design of timber structures - Volume 1 211

8.2 Functional requirements
and system solutions

time is related to an internationally standardised fire with specified

temperature as a function of time. The actual fire resistance time in
an actual fire in a building may be different from that observed in
relation to the standardised fire.

8.2.1.1 Timber
The wood material itself burns and is destroyed in a fire. However, it
burns in a controlled manner and quite slowly. A protective layer of
coal is created on the surface when a wooden piece is subjected to
fire. If the wooden piece is thick, the core of the wood will remain
intact for a good deal of time. The charring rate is of the order
0,7 mm/min. In the case of large cross sections (for example glulam),
the slow charring rate makes it possible to evacuate the building
before collapse. The glue used for glulam can sustain fire well.
Design for fire when looking at large cross sections (as those used in
long-span structures) is made by first calculating a design load using
the load combination for fire in EN 1990. Then, the reduction of
the cross section is calculated corresponding to the required fire
resistance time (for example R 60), using the design charring rate
specified in EN 1995-1-2. If the structure can sustain the load with
Ulls hus, Uppsala.
the remaining cross section, the fire resistance is sufficient.
When considering light-weight structures, the cross section of
wood is much smaller (typically 45 mm on the short side). This
dimension is too small to sustain a fire during for example 60 min­
utes and therefore the structure needs to be completed with some
type of cladding to increase the fire resistance.

8.2.1.2 Sheathing
In light frame timber walls, the fire resistance depends in decreasing
importance upon the following parameters, Fire Safety in Timber
Buildings (2010):
• the sheathing firstly exposed to fire
• the sheathing secondly exposed to fire
• cavity insulation protecting the sides of the timber member
• type of insulation in cavity, stone wool is generally better than
glass wool.

A design model is presented in Fire Safety in Timber Buildings (2010),

where the idea is to calculate the remaining cross section after expo­
sure to fire and make sure that this remaining cross section can sus­
tain the imposed load. The resistance time for the sheathing is
deducted first and the fire exposure is assumed to start once all
boards have “failed”. The method allows for calculation of the fire
resistance of newly developed structures, without the need for per­
forming fire tests.
The most common measure to fulfil the fire safety requirements is
to apply sheathing of gypsum boards. Some Nordic manufacturers
produce special fire resisting gypsum boards that have a fire resist­
ance of approximately 40 minutes. Combined with one layer of com­
mon gypsum board they create a fire safety assuring class R 60. It is
important that the edge joints of boards in both layers do not coin­
cide. Boards of wood-based materials usually have a low fire resist­
ance, for example plywood and OSB have a fire resistance shorter
than 10 minutes. Note that the boards often play several roles, not
only do they function as fire protection, they also stabilise the build­
ing and provide acoustic insulation.

212 Design of timber structures - Volume 1

 8.2 Functional requirements
and system solutions

8.2.1.3 Sprinklers
Sprinklers are a type of active fire protection. They are used to a vary­
ing extent in different countries; some prescribe them in buildings
higher than for example 4 storeys while some do not. There are basi­
cally two types of sprinkler systems: industrial and residential.
Residential sprinklers are an important novelty since most fires take
place in residential buildings. The residential sprinklers are quite
easy to assemble in the building and using them gives a basis for risk
assessment that may lead to a technical exchange in fire safety
design, see Ostman et al. (2002). Some examples of successful applica­
tions of technical exchange when using sprinkler systems are:

• combustible facade in more than 2 storeys may be used

• lower requirements on protection against fire spread through win­
dows in the same building
• reduced demand on the outer facade layer in residential buildings
• increased distance to evacuation path is allowed.

One of the well-known weaknesses of timber structures is their sensi­

tivity to moisture (valid also for insulation materials of mineral
wool). The sprinkler system could be a cause of damage in case of
a false fire alarm and subsequent activation of the sprinkler system.
In case of fire, residential sprinklers bring comparatively the same
amount of water inside the building as in the case of a manual extin­
guishment. The industrial sprinklers on the other hand, supply much
larger amounts of water and the water damage is more severe. In
case of an actual fire it is difficult to verify what damage, caused by
water or by fire is worse. It is claimed that application of residential
sprinklers saves lives; however, still they are optional for 3 — 4 storey
buildings in the Northern European countries. Residential sprinklers
are connected to the ordinary cold water installations as they operate
on much smaller amounts of water than industrial sprinklers.
The heads are installed in rooms where the fire statistically often
starts.

8.2.1.4 Material modifications

Engineered wood products offer the possibility of material modifica­
tions. By adding different chemicals or using specific adhesives
increased fire resistance can be achieved through impregnation. This
option is not so commonly used due to high cost and uncertainties of
the long-term effect of the treatment.

8.2.2 Acoustics
Acoustic regulations for housing, educational buildings and some
other building categories now exist in most countries in Europe, but
findings from comparative studies show that extent and strictness as
well as descriptors vary considerably across Europe. The acoustic per­
formance areas dealt with are e.g. airborne and impact sound insula­
tion, reverberation time, traffic noise, service equipment noise.
Comparing countries, there is in general no consistency of contents,
structure or enforcement of acoustic regulations, see Rasmussen
(2018).
Timber structures are treated as light due to their relatively low
mass in comparison to concrete and steel structures. An overview of
the state of the art regarding acoustics in timber frame buildings is

Design of timber structures - Volume 1 213

8.2 Functional requirements
and system solutions

given in Forssen et al. (2008) and a handbook has been published,

Ljunggren (2011). The solid wood technique (for example X-Lam) is
definitely heavier than light frame systems, but the acoustic perfor­
mance of these systems is still fairly poor. For solid wood floors and
light-weight structures, the major problems are the flanking trans­
mission and the impact sound insulation at low frequencies
(50 — 125 Hz). Currently, well performing solutions however exist to
solve both problems. Flanking transmission is solved by using special
vibration insulation between building parts. Examples are Sylomere,
Xylomer, Xylodyn or Stepisol, which are product brands for a damp­
ing material with specified stiffness. Impact sound insulation is
dependent on the mass and the stiffness of the floor. Wooden floors
must be completed with heavier sheathing materials such as gypsum
board in several layers to increase mass and increase energy absorp­
tion. A very common solution is also to create two separate floor
sub-systems, that is a load-carrying floor slab with a suspended ceil­
ing underneath. These two units are separate and thus sound insula­
tion is improved, see for example figures 8.20, page 220, and 8.22,
page 221. It is important that the fastening of the suspended ceiling is
very flexible, to limit transfer of vibrations from the floor to the ceil­
ing. It is also convenient to place ducts in the space between the floor
and the ceiling.
Building systems with volumetric modules have the double layer
solutions of floor and ceiling integrated in the system itself. It is sel­
dom possible to make acoustic design calculations for floor solutions
in advance, new solutions need to be verified through laboratory
tests and sometimes also through measurements on site. An interna­
tional project, Silent Timber Build, is however in progress (2016),
with the aim to present calculation tools that will be able to better
predict the acoustic performance.
In the case of solid plane floor elements it is common to apply
a floating floor above the load bearing structure to suppress
the flanking transmission. The surface weight of this floating floor is
enhanced by a layer of heavy material, for example sand or concrete.
It has been shown that the airborne sound transmission as well as
the impact sound can be sufficiently decreased using this solution.
Another possibility is to combine a solid wood floor with a suspended
ceiling and walls made from frames clad with plasterboards, as plas­
terboard structures are in general much less prone to flanking trans­
mission than solid wood, such as a X-Lam structure. Further recom­
mendations regarding reduction of the flanking transmission can be
found from Martinsons and in Massivtrahandboken (2006). Solutions
present on the Swedish market that meet the requirements for
the acoustic class B are between 300 mm (with floating floor of con­
crete) up to 600 mm thick for span lengths up to about 8 m. For
larger span lengths, up to about 12 m the thickness of the floor struc­
ture can reach 700 mm.

8.2.3 Horizontal stability

Since timber structures are categorized as light structures, one of
the key structural issues is the stabilisation against horizontal loads.
In the Nordic region, wind load is the main cause for horizontal
loads, while several areas in the world need to address earthquakes
as the main source of horizontal loading, see Tavoussi et al. (2008).
Movements and deformations caused by these loads can be highly
inconvenient or even dangerous for the inhabitants. The problem

214 Design of timber structures - Volume 1

 8.2 Functional requirements
and system solutions

becomes more severe with an increasing number of floors.

Conventionally, in low-rise buildings, the following two main
approaches are utilised:
• diagonal bracing
• shear wall action.

Residential low-rise buildings are often stabilised through shear wall

action, where the sheathing (wood based panels or gypsum board) is
nailed or screwed to the timber frame in the walls to assure sufficient
resistance against horizontal loads. Often, this method is the most
effective and economically advantageous, since it assures relatively
ductile performance and no expensive materials or connectors are
used. In commercial buildings, where the number of windows are
generally less and the aesthetic issues can be of less importance, diag­
onal bracing can be utilised, either being made of timber members or
steel chords. In some cases, albeit unusual, stiffening in one direction
can be realised through moment-resisting connections, see also
Chapter 6.
When dealing with taller structures it is necessary to assure a good
inhabitant comfort also in the higher storeys against possible hori­
zontal deformations in the structural system. For these structures
the above methods may not be sufficient. One possible solution is to
connect the building, along its full height from foundation to roof
with a steel rod anchored to the ground.
The shear forces in tall buildings are transferred from the roof,
walls and floor diaphragms into the foundation via a system of con­
nections. The transfer of loads can require special solutions; for
example, cross-laminated timber panels (X-Lam) can be used instead
of the more common to timber frame nailed or screwed sheathing
boards, to limit excessive horizontal deformation. The difficulty is
that solutions favourable for the stiffness and strength of the build­
ing may be in conflict with the acoustic performance, since increased
stiffness is negative for the sound insulation. Design of stabilising
systems using boards as sheathing is described in Chapter 6 and
EN 1995-1-1. Important factors to ensure good horizontal stabilisa­
tion are, see figure 8.12:
• The sheathing board should be fastened to the timber frame along
all edges. The sheathing board cannot be allowed to end before
Figure 8.12 Force flow in stabilising wall section
reaching the top or bottom plate of the frame.
• The number of and distance between fasteners is decisive for
the horizontal resistance.
• Wall sections with openings cannot be part of the stabilising sys­
tem, unless plastic calculation is applied, see Girhammar and
Kallsner (2010).
• The leading and trailing stud must be sufficiently connected to
the foundation (which could consist of a floor, a wall or the foun­
dation) to avoid uplift.
• It is advisable to spread the stabilisation forces to as many walls as
possible to avoid heavy up-lift forces.

Another alternative can be to use the staircase or elevator shafts, often

constructed in concrete, as the stabilising mechanism for the whole
building. This solution is common for medium-rise buildings in many
European countries. In Sweden this solution is often avoided, since it
is argued that the presence of a concrete shaft inside a timber build­
ing may make it visually, economically and also environmentally

Design of timber structures - Volume 1 215

8.2 Functional requirements
and system solutions

unattractive. Also, such a solution introduces problems due to differ­

ent settlements of timber and concrete, which implies a need to
adjust the elevator stops over the years in service. It is though a feasi­
ble solution structurally, which has been shown by Nordlund (2010).
In systems based on modules, the basic shear wall action needs to
be utilised, as modules are mainly manufactured as light frame sys­
tems built up by sheathing boards on a timber frame. The main prob­
lem for this system is that the joints between the modules should be
robust enough and at the same time accessible during assembly, to
provide satisfactory transfer of shear forces. In addition, a reliable
anchorage to the foundation must be provided to prevent the uplift
of light structures, see Girhammar and Kallsner (2010) and Vessby
(2008).

8.2.4 Critical moisture level

One of the important issues related to using wood as a structural
material is its sensitivity to moisture and, generally speaking,
weather exposure, especially during the assembly phase. It is impor­
tant to emphasise that not only timber is sensitive to moisture, but
also other materials used in construction such as gypsum boards or
insulation materials. In the light of these requirements, the timber
construction companies often use some sort of weather protection
during assembly, especially in cases when assembly is not completed
within one day, see for example figure 8.13.
For smaller houses the weather protection issue leads to
an increased degree of prefabrication. Buildings (for example single
family) can be entirely manufactured off-site and then transported to
the destination and the assembly can be completed in % — 1 full
working day. In this case, additional conditions for a successful mois­
ture protection of the structure are a dry concrete plate and protec­
tion of the unassembled elements.
However, this method does not apply for non-residential buildings
or multi-family houses. For these buildings the assembly time as well
as the exposure to weather is longer, so the principal protection
against weather should be an integral part of the building system
design and assembly process. For multi-storey buildings a very high
degree of prefabrication, for example volumetric module elements,

Figure 8.13 Roof used as weather protection during erection of the structure

216 Design of timber structures - Volume 1

 8.3 Detailed design

can be a solution to avoid moisture influence. Apart from using

a construction method assuring assembly of the roof in an early
stage, other methods are utilized.
The most advanced method of protecting construction works from
weather influence is to cover the building site with suitable tents
(light bearing structures with fixed tarp), see figure 8.14.
There are tents covering the whole intended structure from
the start, with openable roofs or tents protecting only the uppermost
storey where the construction work is ongoing, successively moving
up as the work progresses. A large advantage of this kind of weather
protection is that it enables using a travelling crane inside it, see fig­
ure 8.14 b). However, this method also has some limitations, for
example ventilation problems during warmer periods, logistic com­
plications and higher cost in comparison to other methods, Larsson
and Soderlind (2006). Also, special foundations must be made for
the columns carrying the tent upwards. Examples of the most com­
mon weather protection systems can be found in Axelson et al.
(2004).
The general evaluation of applying weather protection systems is
positive, since they provide a considerably improved working envi­
ronment, which in turn leads to shortened production time, higher
level of safety at work and improved final quality, see Martinsons
(2010) and Larsson and Soderlind (2006). The main criticism regarding
these systems is their high cost, as well as there are different opin­
ions about their real necessity to fulfill the moisture requirements.
As an example in case of the project Inre Hamnen in Sundsvall in
Sweden, see Boverket (2010), the cost of the weather protection was
approximately 4,6 percent of the production cost in the project. This
seems to be the argument for their limited use. However, develop­
ment to decrease the cost of weather protection should proceed since
they are especially advantageous for the construction process during Figure 8.14 a) Movable tent over the whole building.
the autumn-winter season, see Larsson and Soderlind (2006). Alvsbacka Strand, Skelleftea.
b) Movable tent from the inside with integrated
travelling crane.

8.3 Detailed design

The presentation below gives some examples of detail solutions,
mainly for multi-storey buildings, but several are also applicable for
small buildings (for example semidetached houses). Others can be
found via direct contact with suppliers. The fundamentals regarding
details and technology for building systems with solid wood elements
can be found in CLT handbooks, KL-trahandbok (2017). Generally,
there are just a few patented solutions within timber construction,
but at the same time the technical information is usually well pro­
tected by the companies.

8.3.1 Wall elements

Wall elements in wood are generally of two basic types; light-frame
elements or frame elements with solid wood panels. Light-frame ele­
ments are built up with vertical solid studs or so called lightweight
studs (I-joists), with regular spacing 450 or 600 mm, see figure 8.15 a),
page 218. To ensure good thermal insulation in external walls a cross­
wise layer 45 mm thick often is added. This can be done either on
the inside or on the outside of the vertical studs. The external clad­
ding can be made of wood, steel sheets or cement based materials, but
should be ventilated to enable leaking or condensed water to dry out.

Design of timber structures - Volume 1 217

8.3 Detailed design

22+22 Wood cladding

22 Ventilation space between horizontal battens
Wind protection
45 Thermal insulation between horizontal studs s600
195 Thermal insulation between vertical studs s600
Vapour barrier
45 Thermal insulation between horizontal studs s600
15+13 Two layers of Gypsum boards

a)

25 Wood cladding
34 Ventilation space between vertical battens
Wind protection
195 Thermal insulation between vertical studs s600
83 CLT
Vapour barrier
45 Thermal insulation between horizontal studs s600
15 Gypsum board (one or two)

Figure 8.15 Example of a) external wall with light-frame structure,

b) external wall with solid wood frame.

In the example shown in figure 8.15 a) 22 mm vertical boards with

overlapping vertical battens are used, see also horizontal section in
figure 8.18, page 219. The vapour barrier is preferably placed between
the inner cross-wise layer, if present, and the load-bearing stud layer
to ensure that the barrier is not penetrated by fasteners to the sheath­
ing boards and when installing fasteners to hang for example cup­
boards and decorations on the wall. Two layers of plasterboard or
other dense materials may be needed to provide good sound insula­
tion and fire protection.
VA wall element built from a solid wood panel can be almost any­
thing from a single plate to more complicated units consisting of sev­
eral timber panels. Usually panels with thickness 72 — 95 mm have
sufficient load bearing capacity for multi-storey buildings, see fig­
ure 8.15 b). X-Lam-panels are commonly used and can be joined either
by screwing or gluing. The solid wood element provides excellent
stiffness to the structure and possibility to transfer large horizontal
wind forces. However, in external walls, thermal insulation is gener­
ally not sufficient and the solid wood wall needs to be complemented
with a light-frame stud layer on the outside. The example in figure
8.15 b) also has an insulation layer on the inside. The vapour barrier
is applied on the inner side of the inner insulation layer as in figure
8.15 a).

218 Design of timber structures - Volume 1

 8.3 Detailed design

1. 45 x 95 Vertical studs s600

2. 45 x 95 Sole plate
3. 15 + 13 Gypsum board
4. Sound and fire insulation
5. Sound sealant
6. Sound sealant
7. Holders for insulation
8. Sound and fire insulation

15 Gypsum board
94 CLT
2 x 30 Sound insulation board
94 CLT
15 Gypsum board

b)

Figure 8.16 a) Apartment separating wall in light timber frame.

b) Apartment separating wall with cross-laminated timber,
thickness approximately 280 mm, (KLH).

Two examples of apartment separating wood-frame walls are presented

in figure 8.16. The example in figure 8.16 a) is typical for timber frame
buildings with a double wall with air space between the two sides to
provide good sound reduction. Double gypsum linings are used to meet
requirements of fire resistance. The wall studs are braced against
buckling in the weak direction by steel ties to ensure stability in fire Figure 8.17 Horizontal section showing corner
after the gypsum boards are lost. connection of external wall with cross-laminated timber

The example in figure 8.16 b) with double CLT panels with sound
insulation boards at the centre provides good sound reduction for air­
borne sound (61 dB).
Apartment separating walls in timber tend to become quite thick,
which reduces the area possible to sublet for the client. Thick walls
can decrease the return on investments, what also arises for modular
buildings, where the wall structure is double in the building system.
Outer corners in walls also needs to be connected, see figures 8.17
and 8.18:
External walls are thicker than internal walls to provide good insu­
lation. The thickness of external walls increases with increasing
demands on energy efficiency. External walls are subject to several
functional requirements; thermal insulation, moisture barrier, sound
insulation from sounds in the outer environment, horizontal, and
vertical load-carrying capacity. External walls of solid wood panels
need to be completed with insulation material, thus implying that
a regular light-weight wall must be built outside the actual load-car­
rying structure, see figure 8.17. Today there are also systems where
the facade is hung up in the primary structure without extra studs,
what also reduces thermal bridges. Light frame walls are on the other Figure 8.18 Horizontal section of external wall
hand not quite as strong as solid wood walls. The connection between connection in a light frame system. The wall cross section is
light frame walls in the outer corner can be realised as in figure 8.18. the same as in figure 8.15 a), completed with corner details.

Design of timber structures - Volume 1 219

8.3 Detailed design

8.3.2 Floor elements

One of the toughest challenges to handle in the design of timber
buildings is to find a good solution for floor elements. The different
types of load-bearing timber elements which can be used in a floor
are described in section 8.1, page 205. The challenge is to limit
the total depth of the floor for given requirements of span length.
From a structural point of view this is governed by serviceability
a) I =1 requirements concerning vibrations and deflection, see Chapter 7,
page 183.
Wooden floor The design is also strongly dependent on the requirements con­
3 Expanded polyethylene
30 Gipcrete
cerning fire resistance, sound performance and integration with ser­
7- 8 Rubber layer vices. For floors in single-family houses or within the same apart­
21 OSB panel ment, simple floor designs can be found only considering structural
400 Insulation between timber truss beams serviceability requirements. The real challenge is for floors separating
25 Resilient channel
13+15 Gypsum board
apartments, where the floor must be designed to meet strict fire
resistance and sound insulation requirements, see section 8.2, page 211.
An example of floor design used in a timber frame building system
is shown in figure 8.19. This type of solution is typical for timber
frame systems used for residential construction in North America
and UK, although this design is adapted for higher sound perfor­
mance. The main truss joists are quite deep and was chosen to allow
installation of services horizontally. The ceiling consists of a double
gypsum layer suspended in metal sheet profiles (resilient channels)
specially designed for reduction of sound transfer through the floor,
see figure 8.19 b).
A light-weight floor of the type shown in figure 8.19 is quite com­
plex in design with many layers and is not robust to handle during
construction. It is quite sensitive to errors which may lead to reduced
sound performance, for example if installations unintentionally cre­
ates contact between the suspended ceiling and the structure above.
For this reason floors in light-weight systems are often designed with
complete separation between the load-bearing structure and the ceil­
Figure 8.19 a) Typical floor design in timber frame ing below. The ceiling needs a load-bearing structure separate from
building system. b) Photo of the bottom part of the floor
the floor itself and cannot be connected to the floor. An example of
with electric installations.
this approach is shown in figure 8.20. As described in section 8.2.2,
page 213, separation between top and bottom parts of the floor is

22 Particle board
45 Wood stud (45 x 120)
12 Particle board
245 Wood joist +
120 Insulation
145 Wood joist
28 Wood batten
15 Fire resistant gypsum board

Figure 8.20 Apartment separating floor design

220 Design of timber structures - Volume 1

 8.3 Detailed design

a) b)

Figure 8.21 Illustration of a) cross-laminated element, b) cassette element and c) composite floor design.

an advantage automatically obtained in prefabricated volumetric

type construction leading to good acoustic performance.
Floor elements utilising solid wood elements can be divided into
three main types, see figure 8.21:
• Plane elements: nailed or dowelled plates, stress-laminated plates,
cross-laminated plates.
• Cassettes: plates with empty spaces, for example H-beams (made
out of thinner beams of structural timber, LVL or glulam as webs)
and solid wood or wood based panels as flanges.
• Composite floors with timber and concrete (different designs are
possible).

Which solution is used is basically dependent on the acoustic

requirements the floor needs to meet and the span needed. The plane
floor elements usually must be accompanied by either a floating floor
or a suspended ceiling due to the sound insulation requirements (for
floors between apartments, in particular), see for example figure 8.22.
Composite floors, where timber is combined with another material,
present many possibilities for good technical performance, but are
somewhat expensive to produce, see figure 8.23, page 222. The idea is
to have timber in tension and concrete in compression, while the fas­
teners are loaded mainly in shear. The stiffness of the connections
completely governs the performance of the floor, see Chapter 5,
page 115. Composite timber-concrete structures can also be used to
500

lining 28 x 70 si 200 mounted on site is cut after mounting the suspended ceiling to the walls

Figure 8.22 Example of a floor cassette that can be used for span lengths up to 6,5 m
The deck is made of a 3-layered cross laminated plate, supported by glulam beams with bottom flanges with a suspended ceiling.

Design of timber structures - Volume 1 221

8.3 Detailed design

restore existing timber floors by casting a concrete plate on top of

already existing joists. Floor spans can reach up to 12 m with this
technique and it can also be combined with pre-stressing to reach
even longer spans.

Figure 8.23 Different types of connectors in timber-concrete floors, Lukaszewska (2009)

8.3.3 Wall-floor connections

There are two main principles of placing floor elements; they can be
either laid on top of the load-bearing walls, see figure 8.24 a), page 223,
or hung on the walls, see figure 8.24 b), page 223. Placing it on top of
the walls can lead to large forces in compression perpendicular to
the grain and difficulties to achieve air tightness. Hanging floors can

222 Design of timber structures - Volume 1

 8.3 Detailed design

result in a weaker joint with a risk for increased deflection (note that
the connectors are not visible in figure 8.24 b). It can also hang only on
the flanges, which often is the case for cassette floors, see figure 8.22,
page 221.
In many cases the structural details are not commonly available.
Solutions can be found only as examples of existing structures. In
the wall-floor connection, many functional requirements must be
met. Firstly, the vertical load bearing must be granted. In wood,
the compression stresses perpendicular to the grain must be moni­
tored for the vertical load. Secondly, the horizontal resistance must
be sufficient; any connections must be able to transfer horizontal
forces between stories. If the floor is placed on top of the wall as in
figure 8.24 a), the horizontal load must pass the floor element.

1. Horizontal stud
1 2. Wind protection
3. Sole plate
4. Edge beam
6 5. Top plate
2 6. Thermal insulation between vertical studs
7
7. Gypsum board
8 8. Vapour barrier

a)

1. Wind protection
2. Sole plate
3. Top plate
4. Thermal insulation between vertical studs
5. Vapour barrier
1
6. Gypsum boards
7. Edge beam
2

J
3

4
1

b)

Figure 8.24 a) Example light-frame structure with floor element placed on top of the walls.
b) Example light-frame structure with floor element hanging from the walls.

Design of timber structures - Volume 1 223

8.3 Detailed design

In contradiction to the requirements on force transfer, good acoustic

performance is achieved when having loose connections between ele­
ments. Oftentimes, a solution is to have some kind of vibration
2 x 13 Gypsum board damping material between the elements. This weakens the structural
PE-foil
stiffness of the connection though. In the wall-floor connection of
45 x 195 Studs
Coach screw an external wall, the vapour barrier must be continuous when pass­
195 Insulation
10 Plywood ing the connection. Finally, the connection itself must have sufficient
9 External gypsum fire resistance, for example connectors of steel must be covered and
Facade
wooden pieces with small dimensions should be covered by plaster­
board or equivalent. An example of a detailed solution of a wall-floor
connection can be observed in figure 8.25. In this two-storey project
stress laminated floors were used. This kind of floor elements exhibits
Suspended ceiling
worse acoustic performance (especially regarding step sound insula­
tion) than cassette floors, which can be acceptable for non-residential
buildings.
When hanging floor elements on the wall a fastener is needed.
A special hanger has been developed in the example in figure 8.26.
To avoid problems with acoustic performance and vibrations being
Figure 8.25 Solid floor element supported on
a timber frame wall, Massivtrahandboken (2006)
transferred in the wallfloor connection some vibrational damping
needs to be applied. In figure 8.26 the vibrational damping is applied
inside the hanger. In figure 8.24 b), page 223, it is instead accomplished
by inserting damping between the sole plate of the upper wall and
the top plate of the lower wall. A special example is shown in fig­
ure 8.27 where a patented roller support between modules is used.

8.3.4 Services
Timber structures are definitely advantageous compared to concrete
and steel structures regarding the services, primarily due to easiness
of making openings on site. On the other hand it is always more
complicated to integrate the services in prefabricated building com­
ponents than during on-site construction. Clients always demand
a documentation of the installation system when a building system is
used. Generally, in prefabricated building systems, it is favourable to
Figure 8.26 Hanger, MFB Handbok (2010) lead the services vertically to reduce the number of passages between
fire cells, see Lennartsson (2009). In multi-storey residential buildings

------- 4

------- 5
1. Multilayer cross-laminated timber
2. 145 Thermal insulation
3. Wind protection Element joint
4. 34 x 45 Vertical battens
producing a ventilation space 6
5. Wood cladding
6. Steel support reducing flanking
sound transmission
7. 2 x 13 Gypsum board

Figure 8.27 a) Floor-wall connection. b) Roller support.

224 Design of timber structures - Volume 1

 8.3 Detailed design

| | Dwelling

| | Shaft

| | Service central

• Canalisation with
connection point room

Figure 8.28 a) Single and b) multiple shafts for services, Lennartsson (2009).

it is more advantageous to have one shaft for each staircase so a large

number of horizontal pipes can be avoided, instead of one shaft for
the whole building accompanied by a large number of horizontal
installations, figure 8.28.
With the new energy standards, which demand in practice heat
exchange and reuse of prewarmed supplied air to all the rooms, it is
seldom possible to gather all the services in one shaft, especially for
larger apartments, figure 8.29. Supply to the apartments includes
water, electricity and low current. Generally, these are small dimen­
sion installations. The only exceptions are ventilation ducts that are
space-demanding. The small dimension installations are usually
placed in the shafts close to or inside the staircase, which makes
them accessible for service work.
The evacuation of air from apartments requires larger dimensions
(normally 100 mm), see figure 8.30, page 226. For these ducts the main­
tenance demand is much lower and thus the installation can be inte­
grated in the wall structure (with 120 mm studs) or in separate shafts.

a)

Figure 8.29 a) Installation shaft. b) Service layout.

Design of timber structures - Volume 1 225

8.3 Detailed design

Regarding the horizontal installations within the apartments, there

are two applied principles:
• placed in the service space created by the suspended ceiling
• integrated in the building part.

For a system based on panel elements (for example CLT), the service
layer can be placed either on top in a floating floor or under the solid
wood panel in the suspended ceiling. The choice of solution is strictly
connected to the step sound insulation. If the ceiling is not included
in the floor element, the services can be placed in this space before
finishing, figure 8.30. If the ceiling is integrated in the floor element,
services are installed at the factory. With cassette floor elements,
the height of the floor does not need to be increased because of ser­
vices since the spaces between joists can be used for this purpose,
figure 8.31.

8.3.5 Balconies
a) An example of a solution of a 1 200 mm x 1 200 mm balcony for
a multi-storey timber building is presented in figure 8.32, page 227.
The solution consists of a CLT plate placed in a steel frame anchored
in the CLT structure, see Martinsons. However, many different system
solutions for timber balconies have been developed the last years, see
for example Pousette et al. (2011).

8.3.6 Staircases and elevators

The walls of the staircase or the elevator shaft can have a stabilising
function in the building. Solid wood elements (for example CLT) fit
this purpose very well due to good lateral stiffness of the material.
On the other hand, the two most important functions of these build­
ing parts are communication and evacuation. Therefore the struc­
tures in the staircase shaft have to meet the fire safety requirements.
Several different combinations of structural materials can be used in
the stairs. Wooden staircases are in general relatively cost effective,
b) but they tend to result in bad acoustic performance. Staircases in
Figure 8.30 a) Ventilation ducts placed in steel are most suitable for complicated shapes, but on the other hand
a special shaft. b) Services placed under they are the most expensive ones and they might result in an indus­
a floor element before finishing. trial appearance, unless they are combined with wood or tiles.
The design of stairs is usually made in collaboration with the stair
supplier, who has profound knowledge in the field and uses special­
ised design software.

Figure 8.31 Placement of tubes between joists in a cassette floor

226 Design of timber structures - Volume 1

 8.3 Detailed design

1. CLT
2. Sealing layer
3. Pressure-treated battens
4. Pressure-treated decking
5. Z-profile of two welded steel L-profiles
6. Drip plate
7. Steel L-profile
8. Facade ribs
9. Lintel
10. Fixing with going-through rod
11. Drip plate
12. Tension steel tie

b-b

Figure 8.32 Solution for small balcony

Design of timber structures - Volume 1 227

9.1 Timber bridges

Structural systems
for infrastructure
Anna Pousette

9.1 Timber bridges

9.1.1
228
Types of timber bridges 228 9.1 Timber bridges
9.1.2 Loads on bridges 230
9.1.3 Beam superstructures 231 Wood is a suitable material for bridges since it is a light material
9.1.4 Deck superstructures 232 with high load capacity in relation to its own weight. Timber is used
9.1.5 Details 235 as structural material both for pedestrian bridges and road bridges.
9.2 Other exterior applications 236 For road bridges the requirements include sufficient carrying capac­
9.2.1 Noise barriers 236 ity for vehicular traffic with large concentrated wheel loads and this
9.2.2 Vehicle barriers 238 results in different methods, measures and types of bridges compared
to pedestrian bridges. Timber bridges have proved to be strong enough
9.3 Design for durability, wood protection 239
to carry full traffic loads for small and moderate spans.
9.3.1 Inspections and maintenance 241
Bridges are built to pass obstacles and facilitate the movement of
people and materials. Timber bridges have undergone a renaissance
over the last twenty years. New techniques for timber structures have
made it possible and competitive to build new timber bridges.
The important development steps for wooden bridges are new struc­
tures, new materials and new connections, as described by Taylor et
al. (1992). Durable and economic bridges for wide spans or high loads
can be built with these new concepts together with structural protec­
tion. The technical development includes for example more efficient
use of materials such as glued laminated wood, new types of connec­
tions, and better knowledge of structural behaviour of materials and
components. Timber bridges in the old times were often made of tim­
ber beams with a wearing surface of wooden planks. A breakthrough
for the modern timber bridges came with the stress-laminated bridge
deck that improved the possibilities to build durable wooden bridges,
described in Ritter (1990). Because of the limited movements of
the deck, they could also be used with a durable asphalt layer on
the surface.
The selection of a timber bridge is often based on costs, aesthetics,
environmental concerns, construction time and weight. There are
many ways to design a timber bridge and there are many different
bridge types. It is important for the planning and construction of
timber bridges to consider a general protection concept, which
includes design, choice of materials and structural detailing. With
a good design, timber bridges can be competitive both in terms of
construction and maintenance. With good architectural design tim­
ber bridges can also be attractive.

9.1.1 Types of timber bridges

Which type of bridge that is most suitable in each case depends on
the prerequisites for spans, free height and type of traffic. The bridge
should also fit into the landscape and it should be attractive both for
those who travel on the road and for those who see it from
the surroundings.

228 Design of timber structures - Volume 1

 9.1 Timber bridges

The planning of a new timber bridge should take into account:

• topography at the bridge location
• type of crossing, a valley, a rail, river or other obstacles
• connections to surrounding roads
• nearby buildings or landscape
• ground conditions on site
• standards and regulations.

There are many types of timber bridges, see table 9.1. The bridge deck
typically consists of beams or a deck plate. For longer spans the bridge
deck can be complemented by other structural parts. Different bridge
types such as trusses, arches, cable stayed bridges and suspension
bridges can be used for both road bridges and pedestrian bridges, see
TraGuiden (www.traguiden.se).

Table 9.1 Bridge types, static systems

Bridge type Term Suitable spans1)

Beams 5 - 30 m

& &
£
Kings post 10 - 50 m

__ £
&
Strut frame 20 - 40 m

— Truss 10 - 40 m
oo
>n---------

Arch 20 - 70 m

—
Cable-stayed 20 - 130 m

( J &
Suspension 20 - 130 m

77 77 *7

Design of timber structures - Volume 1 229

9.1 Timber bridges

Kings post truss bridges: have vertical rods that transfer loads from
the bridge deck to diagonal members at the sides of the bridge deck.
Kings post bridges can have large spans with a relatively small height
of the bridge deck.

Strut frames: are old building systems that have diagonal compres­
sion struts under the bridge deck. The construction below the deck
enhances the load-bearing capacity of the bridge deck or beam and
transfers the compressive forces to the abutments or walls.

Trussed beams and trusses: can span longer with less material than
homogeneous beam sections, but require larger structural depth.
These bridges are built of glulam and are sometimes combined with
tie rods of steel. They have many joints, which are very important for
the strength, stiffness and durability of the structure. The trusses can
be placed under the bridge deck or at the sides of the bridge deck.
When possible, trusses below the bridge deck are favourable because
several trusses can then be used, the secondary system will be sim­
pler and the trusses will be protected against weather. When trusses
are placed at the sides of a road bridge the depth of the truss should
be big enough for a horizontal wind bracing system to be installed
above the traffic lanes. The truss chords can also be designed with
curved elements.
Trusses at the sides of the deck are used for pedestrian bridges with
a width of 2 — 3 m and trusses with depth of about 1 — 1,5 m can act
as bridge railing together with complementary parts. Experience
shows that protection against moisture is important, but usually for
aesthetic reasons the cladding does not cover the entire trusses. Yet
for truss bridges, all exposed upper surfaces of the lower chord and
the lower diagonal joints should be covered using metal plate
cladding.

Arches: are usually made of glulam. The lamellas are bent during
the gluing process and they can be performed with different radii.
Due to limitations and efficiency of production and transportation
the glulam arches are often made as 3-hinge arches. The arches can
be placed below or above the bridge deck. Arches above the deck are
placed at the sides and are combined with crossbeams under the deck
and steel hangers. The joints are critical to the load-bearing capacity
and durability, and the arches should be protected with cover plates
and cladding. The arches can be treated with chemical impregnation
or be untreated if they are well covered.

Suspension bridges and cable-stayed bridges: consist of a wooden

bridge deck which is suspended from flexible steel cables or tie bars
from pylons or towers. They are suitable for long clear spans. Most
timber suspension bridges have been built for pedestrians.

9.1.2 Loads on bridges

Bridge design includes verification in serviceability and ultimate
limit states for combinations of actions with design values of perma­
nent, variable, and accidental actions as well as actions during execu­
tion. Loads on timber bridges should be calculated according to
the relevant parts of Eurocode EN 1991; densities, self weight, snow
loads, wind actions, thermal actions, actions during execution, acci­
dental actions, traffic loads on bridges. Self weight is a permanent

230 Design of timber structures - Volume 1

 9.1 Timber bridges

action. According to the material standards, EN 338 respectively

EN 14080, the self weight of wood is based on the strength class, for
example mean value 3,5 — 4,8 kN/m3 for C14 — C40, respectively
3,7 — 4,9 kN/m3 for GL20 — GL32. The self weight of asphalt and
asphalt concrete is about 24 — 25 kN/m3. Bridges with roofs must be
calculated for snow load on the roof. Ski and snowmobile bridges
should also be designed for snow loads. Other bridges are assumed to
have the snow removed and are not designed for snow loads.
Traffic loads on bridges correspond to the impact of traffic on
the bridge in vertical and horizontal direction due to the weight of
vehicles, braking and acceleration forces, centrifugal and transverse
forces, loads on footways and cycle tracks, crowd loading, etc. They
are free, variable loads and should be considered as short-term loads.
For road bridges there are in accordance with Eurocode EN 1991-2
several load models with values corresponding to regular traffic with
cars, busses and trucks. The load models contain large concentrated
Figure 9.1 Example of traffic load on road bridge
loads (wheel loads) acting on a surface of 0,4 m x 0,4 m and a uni­
(distributed load and wheel loads)
formly distributed load foreach lane. Load Model 1 is shown in
figure 9.1.
There are two types of traffic loads for pedestrian bridges according
to Eurocode EN 1991-2. They are uniformly distributed load corre­ 3 000
sponding to people and crowds, and concentrated loads (wheel loads)
of a service vehicle, shown in figure 9.2. These loads should not be
combined.
There are static and also aesthetic demands on railings and moreo­
ver, the railings are subject to weathering. Wooden pedestrian
bridges usually have timber railings. Railings on road bridges must
withstand the required impact forces of vehicles. This is evaluated by
full-scale crash tests according to standards. Most often steel railings
Figure 9.2 Example of traffic load on pedestrian
are used on wooden road bridges. The connection of the rail posts to bridge (wheel loads of service vehicle)
the timber deck is very important and needs to be verified.
Vibrations caused by pedestrians are important when designing
timber structures, due to the low weight of the wood material. Both
vertical and horizontal vibrations can cause a feeling of unsafety for
pedestrians standing or walking on the structure. Verifications con­
cerning vibration for pedestrian bridges due to pedestrian traffic
should be done and a verification of the comfort criteria should be
performed. The comfort criterion is defined in terms of maximum
acceptable acceleration of any part of the deck. Damping ratios for
timber bridges can be taken according to Eurocode EN 1995-2.
Fatigue is rarely decisive for timber structures, and is usually not
checked for pedestrian bridges.

9.1.3 Beam superstructures

For beam bridges the beams constitute the main structure. For other
bridge types beams can be combined with additional structures such
as arches or struts, see table 9.1, page 229. The beam bridges are often
constructed with glued laminated beams located below the bridge
deck and with transverse or longitudinal plank decks, as described in
Pousette (2008). The traditional beam bridge typically consists of
the main beams, cross beams, transversal bracing, transverse floor
beams and a plank deck, see figures 9.3 and 9.4, page 232. This type of
bridge is suitable for pedestrian bridges, but also for small road
bridges. The glulam beams are usually manufactured with a pre-cam­
ber so that the bridge is slightly curved. Beam bridges have a simple
static system and they are easy to manufacture and build. The beams

Design of timber structures - Volume 1 231

9.1 Timber bridges

must be transversely braced to provide lateral strength and rigidity

and a horizontal truss between the main beams can be utilised to
care of the wind loads across the bridge. The bridge decks are nor­
mally open plank decks, but also flatwise glued laminated beams or
other woodbased panels can be used together with a waterproofing
layer under the wearing surface.
Simply supported beams are easy to design. Longer bridges over
several supports can be continuous, which is favourable in design.
Bridges over several supports can also be designed with Gerber joints
in a multi-span bridge system. Gerber beam means that the exterior
span beams are cantilevered and support a drop-in beam in the inte­
rior span. Bridge superstructures can be prefabricated into parts or
complete units that are transported to the building site and lifted
into place.
The beam depth depends on the span, the bridge width, the num­
ber of beams and the beam widths. A pedestrian and bicycle bridge
can have two beams for widths up to about 2,5 m. Three beams can
handle widths up to about 4 m. Service vehicles usually require a bit
deeper beams than the surface load alone. For a simply supported
beam with a span of 10 m the required beam depth is about
600 — 800 mm, and for a 20 m span approximately 1 200 — 1 400 mm.
Concentrated vertical loads should be considered at a reference
plane in the middle of the bridge deck (floor beams). For concen­
trated loads an effective load area with respect to the middle plane
should be assumed, see figure 9.5. This gives the number of floor
beams that are loaded with the wheel load.
If structural timber members are exposed to abrasion by traffic,
Figure 9.4 Beam bridge with four glulam beams the depth used in the design should be the minimum depth allowed
before replacement.

9.1.4 Deck superstructures

Stress-laminated deck superstructures consist of timber lamina­
tions of planks, or at larger spans of glued laminated beams, that are
placed edgewise and compressed transversely with pre-stressing high-
strength steel bars, see figure 9.6. The bars are placed in pre-drilled
holes through the laminations and are stressed in tension using
a hydraulic jack. The tension force compresses the laminations so
that the deck behaves like one large, solid plate of wood, as described
Figure 9.5 Beam bridge by Pousette et al. (2002). This leads to a stable and stiff deck which is
suitable for very high wheel loads.
A stress-laminated deck functions as both main structure and
bridge deck. The deck is stiff in its transverse direction, thus nor­
Pre-stressing bar
mally no special wind bracing is needed to withstand e.g. wind loads
across the bridge. A deck plate can also be combined with additional
structures such as arches or struts, see table 9.1, page 229.
Deck superstructures are suitable for road bridges, but also for
pedestrian and bicycle bridges. The bridges are usually provided with
a water-tight layer protecting the wood against moisture from above
Figure 9.6 Stress-laminated deck and an asphalt pavement, functioning as wearing protection, see fig­
ure 9.7, page 233. Together with a cladding on the edges this gives
the bearing structure a good weather protection. The bridges have
either steel or wood railing.
A stress-laminated deck is easy to build and has a good load distri­
bution of the wheel loads. Load transfer between the laminations is
developed by compression and friction. The friction is created by

232 Design of timber structures - Volume 1

 9.1 Timber bridges

Railing

Protective layer
and asphalt pavement

Cover plate

Cladding
Timber deck plate

Figure 9.7 Stress-laminated bridge Pre-stressing bar

the transverse compression applied to the deck. The deck acts as

a continuous slab without transverse or longitudinal joints that can
adversely affect wearing surface performance. Since the load transfer
between the laminations is developed from friction, butt joints are
permitted within certain limitations and thus all laminations do not
have to be continuous. There are few limitations in production of
length and width of a stress-laminated bridge. It is also possible to
manufacture curved bridge decks according to specific needs
and/or locations.
The analysis of laminated timber deck plates should be based upon
one of the following (according to Eurocode EN 1995-2): orthotropic
plate theory, modelling deck plate by a grid or simplified method. In
an advanced analysis, for deck plates made of softwood laminations,
the relationships for the system properties (elasticity and shear mod­
ulus in different directions) are given. The Poisson ratio may be taken
as zero.
Concentrated vertical loads should be considered at a reference
plane in the middle of the deck plate and an effective load area
should be assumed, see figure 9.8. The width bw is the wheel width and
bw,middle is the width at the middle plane calculated from the load dis­
persion angles ft1 and ft2 for different materials. Dispersion angle ft of
concentrated loads for various materials can be found in table 5.2 in
Eurocode EN 1995-2. Figure 9.8 Stress-laminated bridge
In a simplified analysis the deck plate may be replaced by one or
several beams in the direction of the laminations with the effective
width bef calculated as:

^ef “ .middle + a 9.1

where:
bw,middle should be calculated according to figure 9.8.
a is given in the standard, for example 0,3 m for
a stress-laminated deck plate. This beam is loaded
with distributed load and wheel loads and the required
deck thickness can be calculated.

Design of timber structures - Volume 1 233

9.1 Timber bridges

When wheel loads are placed on the deck, the transverse bending
produces a tendency of gaps forming between the laminations on
the lower side of the deck and the transverse shear creates a tendency
of vertical slip of the laminations, see figure 9.9. A sufficient level of
compressive pre-stress between the laminations prevents both actions.
For transverse bending, the compressive stress directly counteracts
the tension on the lower side of the deck. Vertical slip is prevented by
Transverse bending
the friction between the laminations resulting from the compressive
pre-stress. Maintaining an adequate level of pre-stress is the most

■
important issue regarding a stress-laminated deck plate. If
the pre-stressing forces are reduced over time due to shrinkage or
creep, it is always possible to re-stress them. The loss of pre-stressing
force is considerable during the first period after construction, but
after re-stressing a number of times during the service life the force
seems to stabilize as the moisture content also stabilizes. However,
Transverse shear (rolling shear) the pre-stressing force will always vary significantly with temperature.
The long-term pre-stressing forces shall be high enough to ensure
Figure 9.9 Actions of wheel loads that no inter-laminar slip occurs. The maximum allowed shear force
on stress-laminated decks
caused by vertical and horizontal actions depends on the coefficient
of friction, the minimum long-term residual compressive stress due
to prestressing, and the thickness of the plate. The coefficient of fric­
tion is determined based on wood species, roughness of contact sur­
face, treatment of the timber, moisture content, and residual stress
level between laminations. In areas subjected to concentrated loads,
the minimum long-term residual compressive stress due to pre-stress­
ing between laminations should not be below 0,35 N/mm2.
According to Eurocode EN 1995-2 the long-term residual pre-stress­
ing can normally be assumed to be greater than 0,35 N/mm2, provided
that the initial pre-stress is at least 1,0 N/mm2, the moisture content
of the laminations at the time of pre-stressing is not more than 16 %,
and the variation of in-service moisture content in the deck plate is
limited by an adequate protection layer.
The resulting pre-stressing forces should act centrally on the tim­
ber cross-section. The compressive stress perpendicular to the grain
during pre-stressing in the contact area of the anchorage plate shall
be verified. As the pre-stressing forces at the edge can be too high for
the softwood of the deck, hardwood parts are often used for distribu­
tion of the pre-stressing.
Stress-laminated bridges are common for both pedestrian bridges
and road bridges. There are three kinds of stress-laminated structures:
deck plates, T-beam bridges and box-beam bridges. Deck plates are
built as either simply supported or continuous over several supports.
T- and box-beam bridges are usually simply supported with joints at
each support. T- and box-beam bridges are used for longer spans. This
design becomes more complex, see figures 9.10 and 9.11. Transfer of
shear forces between web and flanges as well as deformations for
asymmetrical loads are significant in the design of T-beams and
box-beams.
T-beam bridges and box-beam bridges become somewhat deeper
than a deck plate with the same span, but they are built with less
material. Suitable spans and an approximate indication of required
depths are shown in table 9.2, page 235, for simply supported bridges
with a deck width of about 3 m. The depth depends on the load,
the paving, the wood material, and the width. For T- and box-beam
bridges it also depends on the number of webs and their width and
material.
There are also other types of decks manufactured of planks or glu­
Figure 9.11 Stress-laminated box-beam bridge lam beams placed side by side and nailed or glued together, or of

234 Design of timber structures - Volume 1

 9.1 Timber bridges

Table 9.2 Stress-laminated deck plates, width 3 m.

Deck type Bridge type Suitable span (m) Depth (m)

Stress-laminated deck plate Road bridge 5 - 25 0,3 - 1,2

Stress-laminated deck plate Pedestrian bridge 5 - 20 0,2 - 0,6

Stress-laminated T-beam bridge Pedestrian bridge 20 - 30 0,8 - 1,51)

Stress-laminated box-beam bridge Pedestrian bridge 20 - 30 0,6 - 1,21)

1) Depends also on the number and type of webs.

Fastener
cross-laminated timber (X-Lam) or laminated veneer lumber (LVL).
Concrete
X-Lam consists of at least three layers of crosswise glued boards (most
commonly of softwood), but also five or seven layers are used, see Non-structural
also section 2.7, page 47. The X-Lam boards can be combined with intermediate layer
glued laminated beams to form a T-section. The connection between
Wood
the bridge deck and the beams is important to transmit the shear
forces.
The idea of timber-concrete composites is to combine the high ten­
sile strength of timber with the high compression resistance of con­ Figure 9.12 Timber-concrete composite deck, with
crete. The first applications were in structural systems for new build­ intermediate layer
ings and in renovation. The shear forces can be transmitted by
mechanical connectors or by direct contact between timber and con­
crete (grooved connection). The stress distribution depends on
the rigidity of the connectors.
Timber-concrete connections in composite beams can be designed
as laterally loaded dowel-type fasteners. Where there is an intermedi­
ate non-structural layer between the timber and the concrete (for
example for formwork), see figure 9.12, the strength and stiffness
parameters should be determined by applying a special analysis or by
tests.
For grooved connections, the shear force should be taken through
direct contact pressure between the wood and the concrete cast in
the groove. Sufficient resistance of the concrete and timber parts of
the connection should be verified. The concrete and timber parts must
be held together to avoid separation and the connection should be
designed for a small tensile force between the timber and the concrete.

9.1.5 Details Figure 9.13 Wooden railing on pedestrian bridge

The design of connections in timber bridges is important for

the load-bearing capacity and durability. Connections such as nails,
bolts and timber connectors can be used, but some connections have
too low capacity and give greater slip than desirable. For rigid con­
nections with large forces, for example in truss bridges, embedded
steel plates in sawn slots in the members to be connected can be used
together with dowels. Steel is also used in other components of timber
bridges. Steel members can be used to transfer moments, for example
rigid frames for lateral stability. For cross beams, steel beams are often
used due to the low shear capacity of wood which would require large
cross-sections. Also, for tensile members as vertical suspension bars,
steel is the preferred material.
Railings are usually made of wood or steel, see figures 9.13 and 9.14.
They must comply with load and safety requirements. There are also
requirements for railing height and dimensions of the various com­
ponents of a railing. Especially for pedestrian bridges there are rec­
ommendations for space between the bars and clear distance between
parts. This means that to meet the requirements, the railing has to be
designed so that it can not be climbed. In addition, the aesthetic

Design of timber structures - Volume 1 235

9.2 Other exterior applications

design is of great importance for railings. Especially for pedestrian

bridges the railings should be aesthetically pleasing, and there are
many alternative designs to choose from.
Support attachments for timber bridges must be designed to resist
the vertical and lateral forces transmitted from the superstructure to
the substructure. Bearings for beam bridges can often be simple with
a beam attached to the substructure. Stress-laminated deck bridges are
most frequently placed on rubber strips and connected with steel con­
nections, see figure 9.15. The movements of wooden bridges are usually
relatively small. The transition between road and bridge must be tight,
so that water cannot flow down to the abutment under the bridge.
The abutment should be designed so that water will drain away.

9.2 Other exterior applications

Except for timber bridges there are a number of other applications of
exterior wooden structures in the infrastructure, for example noise
barriers, vehicle barriers, utility poles, lamp posts, bike shelters, bus
shelters and street furniture. Some of these are described below.

9.2.1 Noise barriers

Noise is disturbing and unwanted sound. A noise barrier is an exterior
structure designed to reduce the spreading of noise from roads, rail­
ways, or industrial sources of disturbing sound. Traffic noise from
roads consists of sound from tires, vehicle engines, transmission sys­
tems etc. At speeds above 50 km/h, tyre noise is the dominating source
of noise and it varies by vehicle type and speed. Other methods of
reducing the noise include improving automobile aerodynamics and
tyre design, and choosing low-noise paving material. Despite this work
on noise reduction at the source and also the use of noise reducing
windows and buildings, many people who live along and near main
highways are exposed to high noise levels.
Noise barriers can be efficient means for noise reduction. Location,
topography, cost and aesthetics are of additional importance in
the final choice of a noise barrier. Normally, the benefits of noise
reduction far outweigh aesthetic impacts for residents protected from
unwanted sound. These benefits include lessened sleep disturbance,
improved ability to enjoy outdoor life, reduced speech interference
and reduced stress, as noise can give rise to physiological changes
and various health problems. Disadvantages of noise barriers include
aesthetic impacts for motorists and neighbours (particularly if attrac­
tive views are blocked), costs for design, construction, and mainte­
nance and also necessity to design special solutions for drainage that
the barrier might interrupt.
When sound waves spread from a sound source the noise level
decreases with increasing distance from the source. Sound can be
reflected, refracted or absorbed. The sound is reflected when the sound
wave hits a hard surface. When a barrier is placed in a sound field
the noise decreases behind the barrier, see figure 9.16, page 237.
The decrease is mainly due to the height of the noise barrier. Higher
barriers than 3 — 4 meters are not being built, mostly for practical
reasons. The noise shadow behind the barrier is not complete, since
some of the sound will be spread over the edge. The dampening
effect of the barrier is greatest immediately behind it, and decreases
as the distance to the barrier increases.

236 Design of timber structures - Volume 1

 9.2 Other exterior applications

Figure 9.16 Noise barrier

Noise that is reflected by noise barriers raises the noise level on

the opposite side of the road. Sound absorbing materials such as min­
eral wool on the barrier can reduce the effects of this reflected sound.
It is used especially in urban areas. A noise barrier must be completely
airtight. Cracks or gaps in the barrier reduce the sound insulation.
The barrier should also fit tightly to the ground. A few centimetres
gap between the barrier and the ground can remove half the noise
reduction or more. The openings for the passage of road users should
be designed with overlap.
To be efficient the noise barrier must also have the ability to isolate
the airborne sound. Noise insulation for a barrier should be at least
20 — 25 dB. This is achieved through good woodwork and a noise bar­
rier with double panels.
Loads on noise barriers are wind load, dead load, snow load, snow
clearance effects, and impact of vehicles. Wind loads on free-standing
walls should according to Eurocode EN 1991-1-4 be specified for
the zones A, B, C and D, see figure 9.17. A wall with a return corner
will get a reduced value of the pressure coefficient in zones A-C. In
zone D the pressure coefficient is always 1,2.
Noise barriers made of wood can be constructed as walls. The noise
barrier is built up with posts at 2 — 5 m distance and between the posts
there is a single or double panel. Wood in contact with the ground
should be resistant or pressure-treated wood. The posts of the noise
barrier must be anchored to the ground. Ground and foundation
work stands for a large share of the cost for noise barriers. Therefore,
the distance between the foundation points should be as large as pos­
sible. There is however a need for finding a reasonable combination
of number of support points and robustness of the intermediate pan­
els. The greater the distance between the posts, the longer and
thereby more robust the panel elements have to be to stand wind
loads. The foundation can be made with concrete blocks or piles.
Frost damages of the foundation should be prevented, due to the risk
of damages on posts and connections.

Figure 9.17 Wind load on noise barrier (highest load in zone A near the end)

Design of timber structures - Volume 1 237

9.2 Other exterior applications

Due to exposure to splashing of water from the ground, the lower

part of the barrier close to the support points run the biggest risk of
damages. Vegetation adjacent to the barrier increases the risk of dam­
ages. The panel should preferably be at least at a distance of 300 mm
from the ground, or at least 100 mm if there is an effective splash
guard. The wall must fit tightly to the ground and the connections
should be able to take up displacements in the ground.
Material selection for noise barriers is important to make them fit
in with the surroundings and with existing buildings. Variation of
materials along the barrier may help to create experience of variety
along a long and monotonous wall. Also solutions based on glass
may reduce the effect of a dominant wall, see figure 9.18, and vegeta­
tion along the noise barriers may be included as part of the design.
Noise barriers along a road should be built with two fronts, one
towards the road and one towards residents. The road front should be
varied, but given a whole and large scale so that the drivers are not
distracted. Drivers move fast and do not experience the details of
the design. Detailed design is however important regarding the resi­
dent front where people move slowly and will get close the barrier.

9.2.2 Vehicle barriers

Barrier design depends on the intended use of an area. Dimensions
and materials are chosen depending on the requirements and
the surrounding environment. Some barriers keep vehicles within
defined spaces such as a road, or specific parking areas.

238 Design of timber structures - Volume 1

 9.3 Design for durability, wood protection

Vehicle barriers can be divided into different types such as parapets,

bollards, fences, and wooden guard-rails.
Barriers and parapets in car parking areas should be designed to
resist the required horizontal loads to withstand the impact of a vehi­
cle. The horizontal load is normal to and uniformly distributed over
the length of the barrier. The force is calculated based the mass and
velocity of the vehicle and can be assumed to act at bumper height.
Wooden guardrails are series of low posts tied together by wooden
rails. Guardrails block and control vehicular access, and are used to
line roads. They should be designed to resist the required horizontal
loads of a vehicle. A fence is a structure functioning as a boundary or
barrier, usually made of posts, boards, wire or rails. The purpose is to
keep people in or out of an area, and fences are not designed to with­
stand vehicular impact.
A bollard is a post with no stringer or rail, commonly 0,5 to 1,2 m
tall, and used one by one or grouped. The purpose is to block vehicle
entry and they are used to define boundaries e.g. around a parking
lot, outside buildings, or at a pedestrian entrance. When bollards
define a pedestrian entrance, they should be at least 0,9 m apart to
allow pedestrian and wheelchair passage. They can be more closely
spaced when they define a boundary. Using bollards provided with
lightning, creates lit up zones revealing details in the immediate sur­
roundings and can provide guiding lights for the safety of pedestrians
and motorists.

9.3 Design for durability,

wood protection
Serviceability, working life and durability is often included in
the requirements for exterior wood applications. Durability is
the property of service during a given time without losing defined
functions. Economically reasonable working life is the time during
which the performance will be maintained at a level compatible with
the requirements when all relevant aspects are taken into account.
Considered aspects comprise costs of design, construction and use, as
well as inspections, maintenance and environmental aspects. They
also include consequences of failure and costs for repair, partial
renewal and disposal.
The durability of outdoor timber structures is affected by precipita­
tion, wind and solar radiation. The effects of direct weathering of
structural timber members can be reduced by structural protection,
or by using timber with sufficient natural durability, or timber treated
with preservatives against biological attacks. Exterior structures of
wood should be designed for optimal durability when selecting for
example bridge type, appropriate design of details, dimensions, joints,
and treatments including coatings, which may influence the durabil­
ity. Decay fungi are the main threat, and the most important issue is
to keep the moisture content of the timber below critical levels.
The growth of fungi in wood (Picea abies) depends on the humidity
and if the moisture content is below 20 % no growth will occur.
The interaction between materials, design, construction and mainte­
nance is important. Materials and treatments should be selected based
on requirements of durability and service life, including the risk of
decay fungi, and the ability to inspect, maintain and repair.

Design of timber structures - Volume 1 239

9.3 Design for durability, wood protection

Most critical in the design of details is the influence of moisture on

timber-based materials, and this can be minimized mainly by:
• Minimizing moisture penetration by efficient water shelter, appro­
priate cladding, end grain protection, avoiding exposure of end
joints, using fastener types minimizing the risk of cracks, ensuring
appropriate distance between timber parts and the ground, appro­
priate surface coatings or other treatments.
• Ensuring rapid drying if the material takes up excessive moisture,
using diffusion open treatments, avoiding narrow gaps in joints,
avoiding vegetation close to the exposed structure.

Covering of the structural elements will protect the structure from

rain, for example a roof on a bridge or a cladding on an arch, as
shown by Kleppe (2010). Where a partial or complete covering of
the main structural elements is not feasible, durability can be
improved through one or more of the following measures:
• Limiting standing water on timber surfaces through appropriate
inclination of surfaces.
• Limiting openings, slots, etc., where water may accumulate or
infiltrate.
• Limiting direct absorption of water (for example capillary absorp­
tion from concrete foundation) through use of appropriate
Pedestrian and bike cable-stayed bridge, Skelleftea.
barriers.
• Limiting fissures and delaminations, especially at locations where
the end grain would be exposed, by appropriate sealing and/or
cover plates.
• Limiting swelling and shrinking movements by ensuring an appro­
priate initial moisture content and by reducing in-service moisture
changes through adequate surface protection.
• Selecting a geometry for the structure that ensures natural ventila­
tion of all timber parts.

The risk of increased moisture content close to the ground can be

reduced by:
• Covering of the ground with coarse gravel or similar to limit
vegetation.
• Using an increased distance between the timber parts and
the ground level.

Resistance to corrosion is important. An example of especially corro­

sive conditions is a timber bridge where corrosive de-icing cannot be
excluded. The risk of stress corrosion should be taken into account.
The effect of chemical treatment of timber, or timber with high acidic
content, on the corrosion protection of fasteners should be taken into
account. For steel parts and fasteners, galvanization is sufficient in
most cases but stainless steel is also used.
Treated timber can provide protection against rot and/or insects.
Pressure-treated wood is typically used when there is a risk of
long-lasting moisture impact, when safety measures are important,
structural elements are difficult to replace or when the wood is located
close to the ground. Pressure-treated wood shall be manufactured
according to European standards.
Some countries such as Norway prescribe 100 years of service life
for bridges. In Norway the wood in bridges is often impregnated
using chemicals to provide sufficiently long durability. Two methods
are used, either copper salts or creosote or a combination of both.

240 Design of timber structures - Volume 1

 9.3 Design for durability, wood protection

A double impregnation using salt and creosote is common, as

described by Ekeberg (2010). Each lamella is pressure-treated with
salt solution before planing and gluing. Then the entire structural
member with holes etc. is given a pressure treatment with creosote
oil. This double impregnation has been applied for arches and trusses.
Stress-laminated deck plates protected by a water-tight membrane are
usually only impregnated with creosote oil.
Other countries have prohibited the usage of chemical treatments
containing creosote due to it being considered more or less harmful
to nature. Therefore the design is very important as well as the design
of details and the maintenance. Sweden has requirements concerning
wooden bridges with technical service life of 40 and 80 years respec­
tively. The difference between these cases is that extensive sheltering
is required and a maintenance plan has to be prepared for technical
service life of 80 years. Impregnation with copper salts is permitted
for exposed members, but bridge decks and other structural compo­
nents that are protected by sheltering can be made of softwood with­
out any chemical treatment or by other wood species with acceptable
natural durability.

9.3.1 Inspections and maintenance

Wood protection and maintenance should be considered during plan­
ning since the reason for decay is usually poor detailing and/or
neglected maintenance. Regular maintenance is needed to ensure
a long life of outdoor structures of any material, timber as well as
other materials. This includes all measures to ensure proper perfor­
mance of the structure to prevent damage from weathering and
decaying organisms. For bridges, it is usually required that they should
be inspected regularly to provide a basis for assessing their condition
and function, and the planning of future actions. Timber bridges are
inspected in a similar way compared to other bridges, but the meth­
ods and measurements are adapted to the material performance char­
acteristics of wood, as described in Pousette et al. (2006).
Thorough cleaning is recommended once a year to remove dirt like
soil and/or leaves from the stucture. Openings should be kept free for
efficient drainage. Coated surfaces can be washed with clean water
and the appearance and function of components of the structure
should be checked. It is important to note the signs of decay in wood
components, observable as for example very soft spots etc.
Maintenance painting should be carried out after cleaning and/or
sanding. Usually the heavily exposed surfaces such as the top of
handrails need more frequent maintenance painting than vertical
surfaces. If there is an excessive thickness developed or cracking and
flaking occurs after several painting interventions, the structure
requires a total renewal of the paint layer.

Design of timber structures - Volume 1 241

Symbols

Symbols
Symbols used in EN 1995-1-1.

Symbol Explanation Fv,Rd Design load-carrying capacity per shear plane

per fastener; Design racking load capacity
Latin upper case letters
Fv,Rk Characteristic load-carrying capacity per shear
A Cross-sectional area
plane per fastener
Aef Effective area of the total contact surface
Fv,w,Ed
M Design shear force acting on web
between a punched metal plate fastener and
the timber; effective total contact surface Fx,Ed Design value of a force in x-direction
perpendicular to the grain
Fy,Ed Design value of a force in y-direction
Af Cross-sectional area of flange
Fx,Rd Design value of plate capacity in x-direction
„
Anet,t Net cross-sectional area perpendicular to Design value of plate capacity in y-direction
Fy,Rd
the grain
Fx,Rk Characteristic plate capacity in x-direction
t
Anet,v Net shear area parallel to the grain
Fy,Rk Characteristic plate capacity in y-direction
C Spring stiffness
G0,05 Fifth percentile value of shear modulus
E0,05 Fifth percentile value of modulus of elasticity
Gd Design value of shear modulus
Ed Design value of modulus of elasticity
Gmean Mean value of shear modulus
Emean Mean value of modulus of elasticity
H Overall rise of a truss
Emeanfin
e Final mean value of modulus of elasticity
If Second moment of area of flange
F Force
1tor Torsional moment of inertia
FA,Ed Design force acting on a punched metal plate
Iz Second moment of area about the weak axis
fastener at the centroid of the effective area
Kser Slip modulus
FA,min,d Minimum design force acting on a punched
metal plate fastener at the centroid of the «
Kser,fin Final slip modulus
effective area
Ku Instantaneous slip modulus for ultimate limit
Fax,Ed
M Design axial force on fastener states

Fax,Rd Design value of axial withdrawal capacity of Lnet,t Net width of the cross-section perpendicular
the fastener to the grain

Fax,Rk Characteristic axial withdrawal capacity of the L net,v

, Net length of the fracture area in shear
fastener Design moment acting on a punched metal
MA,Ed
Fc Compressive force plate fastener

Fd Design force M ap,d Design moment at apex zone

Fd,ser Design force at the serviceability limit state Md Design moment

Ff,Rd Design load-carrying capacity per fastener in Rl,

M y,Rk Characteristic yield moment of fastener
wall diaphragm N Axial force
F,c,Ed Design compressive reaction force at end of Design splitting capacity
R90,d
shear wall
R90,k Characteristic splitting capacity
^.t,Ed Design tensile reaction force at end of shear
Rax,d Design load-carrying capacity of an axially loa­
wall
ded connection
Fi,vert,Ed Vertical load on wall
Rax,k„ Characteristic load-carrying capacity of an axi­
F,v,Rd Design racking resistance of panel i or wall i ally loaded connection

Fa Lateral load Rax,a,k, Characteristic load-carrying capacity at an

Design force from a design moment angle a to grain
FM,Ed
Tensile force Rd Design value of a load-carrying capacity
Ft
Characteristic tensile capacity of a connection Ref,k Effective characteristic load-carrying capacity
Ft,Rk
of a connection
Fv,0,Rk Characteristic load-carrying capacity of
Riv,d Design racking capacity of a wall
a connector along the grain
Rk Characteristic load-carrying capacity
Fv,Ed Design shear force per shear plane of fastener;
Horizontal design effect on wall diaphragm Rsp,k Characteristic splitting capacity

Source: EN 1995-1-1:2004, 1.6

242 Design of timber structures - Volume 1

 Symbols

Rto,k, Characteristic load-carrying capacity of fc,90,k Characteristic compressive strength

a toothed plate connector perpendicular to grain

Rv,drl Design racking capacity of a wall diaphragm ft Design tensile strength of flange

V Shear force; volume fu Characteristic embedment strength

Vu, Vl Shear forces in upper and lower part of beam fhead,k Characteristic pull-through parameter for nails
with a hole
f, Fundamental frequency
wy Section modulus about axis y
fm,k, Characteristic bending strength
Xd Design value of a strength property
fm,y,d„ Design bending strength about the principal
Xk Characteristic value of a strength property y-axis

Latin lower case letters fm,z,d„ Design bending strength about the principal
z-axis
a Distance
fm,a,d„ Design bending strength at an angle a to
a Spacing, parallel to grain, of fasteners within
the grain
one row
f:.0.d Design tensile strength along the grain
Minimum end distance to the centre of gravity
a1,CG
of the screw in each timber member f:.0.k Characteristic tensile strength along the grain

ad Spacing, perpendicular to grain, between rows ft,90,d Design tensile strength perpendicular to
of fasteners the grain

a2,CG Minimum edge distance to the centre of f:.w.d Design tensile strength of the web
gravity of the screw in each timber member
f• Characteristic tensile strength of bolts
a3,c Distance between fastener and unloaded end
fv,0,d Design panel shear strength
a3,t Distance between fastener and loaded end
fv,ax,a,k, Characteristic withdrawal strength at an angle
a4,c Distance between fastener and unloaded edge a to grain

a4,t Distance between fastener and loaded edge fv,ax,90,k Characteristic withdrawal strength
perpendicular to grain
abow Maximum bow of truss member
fv,d„ Design shear strength
abow,perm Maximum permitted bow of truss member
h Depth; height of wall
adev Maximum deviation of truss
hap Depth of the apex zone
adev,perm Maximum permitted deviation of truss
hd Hole depth
b Width
he Embedment depth; loaded edge distance
bi Width of panel i or wall /
hefr Effective depth
bnet Clear distance between studs
hf,c Depth of compression flange
bw Web width
hf,t Depth of tension flange
d Diameter; outer thread diameter
hrl Distance from lower edge of hole to bottom of
d, Diameter of centre hole of connector; inner
member
thread diameter
hru Distance from upper edge of hole to top of
dc Connector diameter
member
def Effective diameter
hw Web depth
dh Head diameter of connector
i Notch inclination
fh,i,k Characteristic embedment strength of timber
kc,y, kc,z Instability factor
member i
kcr Crack factor for shear resistance
fa,0,0 Characteristic anchorage capacity per unit area
for a = 0° och p = 0° kcrit. Factor used for lateral buckling

fa,90,90 Characteristic anchorage capacity per unit area kd Dimension factor for panel
for a = 90° och p = 90°
kdef Deformation factor
fa.^ Characteristic anchorage strength
k.dis Factor taking into account the distribution of
fax,kt Characteristic withdrawal parameter for nails stresses in an apex zone

f,0.d Design compressive strength along the grain kf,1, kf,2, kf,3 Modification factors for bracing resistance

f,w.d Design compressive strength of web kh Depth factor

f,f,c,d„ Design compressive strength of flange ki,q Uniformly distributed load factor

Source: EN 1995-1-1:2004, 1.6

Design of timber structures - Volume 1 243

Symbols

km Factor considering re-distribution of bending uinst,Q,i Instantaneous deformation for accompanying

stresses in a cross-section variable actions Qi

kmodd Modification factor for duration of load and wc Precamber

moisture content
wcreep Creep deflection
kn Sheathing material factor
w.fin Final deflection
kr Reduction factor
winst* Instantaneous deflection
kR,red Reduction factor for load-carrying capacity
wnet,fin Net final deflection
ks Fastener spacing factor; modification factor for
v Unit impulse velocity response
spring stiffness
Greek lower case letters
ks,red Reduction factor for spacing
a Angle between the x-direction and the force
h
kshape Factor depending on the shape of
for a punched metal plate; Angle between
the cross-section the direction of the load and the loaded edge
k System strength factor (or end)
sys

kv Reduction factor for notched beams p Angle between the grain direction and the
force for a punched metal plate
kvol> Volume factor
Pc Straightness factor
ky eller kz Instability factor
Y Angle between the x-direction and the timber
l a,min Minimum anchorage length for a glued-in rod
connection line for a punched metal plate
l Span; contact length Partial factor for material properties, also
YM
Distance from a hole to the centre of accounting for model uncertainties and
lA
the member support dimensional variations

Effective length; Effective length of A Slenderness ratio corresponding to bending

lef y
distribution about the y-axis

Distance from a hole to the end of the member Az Slenderness ratio corresponding to bending
lv
about the z-axis
lZ Spacing between holes
,
Arel,y Relative slenderness ratio corresponding to
m Mass per unit area bending about the y-axis

n40 Number of frequencies below 40 Hz ,

Arel,z Relative slenderness ratio corresponding to
bending about the z-axis
nef Effective number of fasteners
pk Characteristic density
Pd Distributed load
Pm Mean density
q Equivalent uniformly distributed load
Oc,0,d Design compressive stress along the grain
r Radius of curvature
O„c,a,da Design compressive stress at an angle a to
s Spacing
the grain
s0 Basic fastener spacing
f A
Of,c,d Mean design compressive stress of flange
r-n Inner radius
Off,c,max,dA Design compressive stress of extreme fibres of
t Thickness flange
^pen Penetration depth Mean design tensile stress of flange
Of,t,d
ucreep Creep deformation Of,t,max,d
ft A Design tensile stress of extreme fibres of
flange
Ufin Final deformation
Om,crit Critical bending stress
ufin,G Final deformation for a permanent action G
Om,y,d. Design bending stress about the principal
Ufin,Q,1 Final deformation for the leading variable
y-axis
action Q1
Om,z,d. Design bending stress about the principal
Ufin,Q,i Final deformation for accompanying variable
z-axis
actions Qi
Om,a,d. Design bending stress at an angle a to
uinst Instantaneous deformation
the grain
uinst,G Instantaneous deformation for a permanent
ON Axial stress
action G
Ot,0,d Design tensile stress along the grain
uinst,Q,1 Instantaneous deformation for the leading
variable action Q1 Ot,90,d Design tensile stress perpendicular to the grain

Source: EN 1995-1-1:2004, 1.6

244 Design of timber structures - Volume 1

 Symbols

0w,c,d Design compressive stress of web

°W,t,d Design tensile stress of web

Td Design shear stress

TF,d Design anchorage stress from axial force

TM,d Design anchorage stress from moment

Ttor,d Design shear stress from torsion

V0 Factor for combination value of a variable

action

V1 Factor for frequent value of a variable action

V2 Factor for quasi-permanent value of a variable

action

z Modal damping ratio

Source: EN 1995-1-1:2004, 1.6

Design of timber structures - Volume 1 245

Bibliography

Bibliography
Chapter 1 - Introduction to design and design process
Engstrom, B.: Conceptual Design — Methodology. Course literature on structural analysis and design.
Department of Structural Engineering. Chalmers University of Technology, Gothenburg, Sweden, 2002.

Johansson, G., Perstorper, M., Kliger, R.: Kvalitetskrav pa byggnadsvirke. Kvalitetskrav pa byggnadsvirke.
(In Swedish). BFR R105:1990, Stockholm, 1990.
Kroll, E., Condoor, S., Jansson, D.: Innovative Conceptual Design — Theory and Application of Parameter
Analysis. Cambridge University Press. Cambridge, UK, 2001. ISBN 0521778484.

Niemeyer, S.: Conceptual Design in Building Industry. Master’s Thesis 03:6. Department of Structural
Engineering. Chalmers University of Technology, Gothenburg, 2003.
Paik, J.K., Melchers, R.E.: Condition assessment of aged structures. Woodhead Publishing Limited, 2008.
ISBN 1 84569 334 5.

Thelandersson, S., Isaksson, T., Fruwald, E., Toratti, T.,Viitanen, H., Grull, G., Jermer, J., Suttie, E.:
Bestandighet for utomhustra ovan mark. Guide for utformning och materialval. Report TVBK-3060.
Faculty of Engineering, Lund University, 2012.

Chapter 2 - Structural properties of sawn timber

and engineered wood products
Anon: State of the World’s Forests 2007. Food and Agriculture Organization of the United Nations,
Rome, 2007. ISBN 978-92-5-105586-1.
Anon: State of the World’s Forests 2007. Food and Agriculture Organization of the United Nations,
Rome, 2009. ISBN 978-92-5-106057-5.

Anon (2005a): European Forest Sector Outlook Study — 1960 — 2000 — 2020 Main report. United Nations
publications, 2005. ISSN 1020 2269.
Anon (2005b): Global Forest Resources Assessment 2005. FAO Forestry Paper 147. Food and Agriculture
Organization of the United Nations, Rome, 2005.

Bodig, J., Jayne, B.: Mechanics of wood and wood composites. Van Nostrand Reinhold company Inc.
New York, 1982. ISBN 0-442-00822-8.
Dackling, M.: Den svenska travarumarknaden. Forbrukning inom delsektorer 1995 — 2000. (The Swedish
market for sawn timber. Use within different industry segments, in Swedish). Institutionen for industriella
produktionssystem, avdelningen for Skog- och trateknik, Vaxjo universitet, Vaxjo, 2002.

Dinwoodie, J.M.: Timber: Its nature and behaviour. Second edition. E & FN Spon, London, 2000.
ISBN 0-419-23580-9.
Hankinson, R.L.: Investigation of crushing strength of spruce at varying angles of grain,
Air Force Information Circular No. 259, U. S. Air Service, 1921.

Hearmon, R.F.S. Paton J. M.: Moisture content changes and creep of wood.
Forest products journal 14 (8) pp. 357 — 359, 1964.
Hill, C.A.S.: Wood Modification: Chemical, Thermal and Other Processes. John Wiley&Sons,
Chichester, England, 239 pp, 2006. ISBN 0-470-02172-1.

Hoffmeyer, P.: Wood as a building material (A4), in Timber Engineering STEP 1, Blass, H.J.,
Aune P., Chooet B.S., Editors. Almere, Centrum Hout, the Netherlands, 1995.
Hoffmeyer, P.: Strength under long-term loading, in Timber Engineering, S. Thelandersson and H.J. Larsen,
Editors. Wiley: Chichester. ss.131-152, 2003. ISBN 0-470-84469-8.

246 Design of timber structures - Volume 1

 Bibliography

Hon, D.N.S., Shiraishi, N. (ed).: Chemical modification of wood, Handbook on Wood and Cellulosic Materials,
Marcel Dekker, Inc., New York, NY, Chapter 15, 703 — 756, 1991.

Johansson, C.-J.: Grading of Timber with Respect to Mechanical Properties, in Timber Engineering,
S. Thelandersson, Larsen, H. J., Editors. Wiley: Chichester. ss. 23 — 45, 2003. ISBN 0-470-84469-8.
Skogsindustrierna, The Swedish Forest Industries: Facts and Figures.

Timell, T.E.: Compression wood in gymnosperms. New York; Springer-Vlg, Cop, 1986. ISBN 0-387-15715-8.

Chapter 3 - Design of structural timber elements

in ultimate limit state
Anon: Timber Engineering STEP 1 and STEP 2. H.J. Blass et al. Centrum Hout, the Netherlands, 1995.
Belluzzi, O.: Scienza delle Costruzioni, vol. IV (Structural Mechanics, vol IV), in Italian, Zanichelli,
Bologna, Italy, 1966.

Blass, H.J.: Columns. In “Timber Engineering STEP 1”. Ed. Blass et. al., Centrum Hout, the Netherlands, 1995.
Carling, O.: Dimensionering av trakonstruktioner (Design of Timber Structures), in Swedish,
AB Svensk Byggtjanst & Tratek, Stockholm, Sweden, 1992.

Danielsson, H.: The strength of glulam beams with holes. A Probabilistic Fracture Mechanics Method and
Experimental Tests. Division of Structural Mechanics. Report ISRN LUTVDG/TVSM--09/3069--SE (1 — 124), 2008.
Gustafsson, P.J.: Notched beams and holes in glulam beams. In “Timber Engineering STEP 1”.
Ed. Blass et. al., Centrum Hout, the Netherlands, 1995.

Gustafsson, PJ.: Fracture perpendicular to grain — Structural applications. In “Timber Engineering”.

Ed. Thelandersson S., Larsen H.J., Wiley & Sons, Chichester, UK. 3.40, 2003.
Gross, H., Crocetti, R.: Limtrahandbok Del 1 — 3, in Swedish, Svenskt Tra, Stockholm, 2016.

Martensson, A.: Short and long term deformations of timber structures. In “Timber Engineering”.
Ed. Thelandersson S., Larsen H.J., Wiley & Sons, Chichester, 2003.
Piazza, M., Tomasi, R., Modena, R.: Strutture in Legno (Timber Structures), in Italian,
Hoepli, Milan, Italy, 2005.

Porteous, J., Kermani, J.: Structural Timber Design to Eurocode 5, Blackwell Science Ltd, Oxford, UK, 2007.

Stalnaker, J.J., Harris, E.C.: Structural Design in Wood. Van Nostrand Reinhold, New York, 1989.

Timoshenko, S., Gere, J.M.: Theory of Elastic Stability (2nd edition). McGraw Hill International Book Company,
Singapore, 1963.
Thelandersson, S., Larsen, Timber Engineering. John Wiley & Sons Ltd. West Sussex, UK. 3.41, 2003.

Chapter 4 - Design of timber joints

Aicher, S.: Structural Adhesive Joints including Glued-in Bolts. Timber Engineering.
Larsen H.J. and Thelandersson S. Eds., Wiley & Sons, 2003. ISBN 0-470-84469-8

Blass, H.-J.: Load distribution in nailed joints. In proceedings from the International Council for
Research Innovation in Building and Construction, Working Commission CIB-W18, paper 23-7-2,
Lisbon, Portugal, September 1990.

Blass, H-J.: Joints with Dowel-Type Fasteners. Timber Engineering. Larsen H.J., Thelandersson, S. Eds.,
Wiley & Sons, 2003. ISBN 0-470-84469-8.

Cramer, C.O.: Load Distribution in Multiple-Bolt Tension Joints. Journal of the Structural Division
94(ST5):1101-1117, 1968.

Design of timber structures - Volume 1 247

Bibliography

Gustafsson, P-J: Analysis of Generalised Volkersen-joints in terms of Non-linear Fracture Mechanics.

Mechanical Behaviour of Adhesive Joints, Pluralis:323 — 338, 1987.

Gustafsson, P.-J., Serrano, E.: Glued-In Rods for Timbers Structures (GIROD): Final Report for WP1 —
Development of a calculation model, Lund University, 2001.

Johansen, K.W.: Theory of Timber Connections. International Association of Bridge

and Structural Engineering 9:249 — 262, 1949.

Johnsson, H.: Plug Shear Failure in Timber Connections — Avoiding Brittle and Promoting Ductile Failures.
Doctoral thesis 2004:03, Lulea University of Technology, 2004.

Kuipers, J., Van Der Put, T.A.C.M.: Betrachtungen zum Bruchmechanismus von Nagelverbindugen.
In Ingenieurholzbau in Forschung und Praxis, Eds Ehlbeck J. and Steck G., Bruderverlag Karslruhe, 1982.
ISBN 3-87104-049-5.

Lantos, G.: Load Distribution in a Row of Fasteners subjected to Lateral load.

WoodScience 1(3):129-139, 1969.

Larsen, H.J. och Riberholt, H.: Traekonstruktioner (in Danish). SBI-anvisning 194,
Statens Byggeforskningsinstitut, 1999.

Larsen, H.J.: Structures Based on Solid Timber Products. Timber Engineering.

Larsen, H.J., Thelandersson S. Red., Wiley & Sons, 2003. ISBN 0-470-84469-8.

Nielsen, J.: Trusses and Joints with Punched Metal Fasteners. Timber Engineering.
Larsen H.J., Thelandersson S. Red., Wiley & Sons, 2003. ISBN 0-470-84469-8.

Tlustochowicz, G., Serrano, E., Steiger, R.: State-of-the-art review on timber connections
with glued-in steel rods. Materials and Structures 44(5):997-1020, 2011.

Chapter 5 - Composite timber elements

Elsander: Elementarfall for tvaskikts samverkansbalkar med linjart elastiskt beteende.
Royal Institute of Technology, div. of Steel Structures, Master of Science Thesis 125, 1999. (In Swedish)

Foschi: Rolling shear failure of plywood in structural components. Forest Products Laboratory, Madison.
Information report VP-x-67, 1970.
Granholm, H.: Om sammansatta balkar och pelare med sarskild hansyn till spikade trakonstruktioner.
(On composite beams and columns with particular regard to nailed timber structures). Transactions,
Chalmers University of Technology, Rep. No. 88, 1949. (In Swedish)

Hoglund: Design of timber beams with webs of plywood or structural board.

Royal Institute of Technology, div. of Steel Structures, Report, 1990.
Larsen, Riberholt: Trakonstruktioner, Beregning. Statens Byggeforskningsinstitut. SBI-anvisning 210,
H0rsholm, 2005. (In Danish).

Lekhnitsky: Anisotropic plates. Gordon & Breach Science Publishers, New York, 1968.
Newmark, N.M., Siess, C.P, Viest, I.M.: Tests and analysis of composite beams with incomplete interaction.
Proceedings, Society for Experimental Stress Analysis 9(1): 75-92, 1951.

Oduyemi, T.O.S., Wright, H.D.: Partial interaction analysis of double skin composite beams.
Journal of Constructional Steel Research 19(4): 253-283, 1991.
StBK-N5: Tunnplatsnorm. AB Svensk Byggtjanst, Stockholm, 1980. (In Swedish)

Stussi, F.: Zusammengesetzte vollvandtrager. (Built-up girders). International Association for Bridge
and Structural Engineering, IABSE, Vol. 8, pp. 249-269. (In German)
Thielgard, Larsen: Limede I-bj^lker af tra. Aalborg University, inst. for Bygningsteknik.
Report 7804, 1978. (In Danish)

248 Design of timber structures - Volume 1

 Bibliography

Chapter 6 - Horizontal stabilization

Anon: Timber Engineering STEP 1 and STEP 2, Blass, H.J., et al., Centrum Hout, the Netherlands, 1995.

Belluzzi, O.: Scienza delle Costruzioni, vol. IV (Structural Mechanics, vol IV), in Italian,
Zanichelli, Bologna, Italy, 1966.
Carling, O.: Dimensionering av trakonstruktioner (Design of Timber Structures), in Swedish,
AB Svensk Byggtjanst & Tratek, Stockholm, Sweden, 1992.

Caironi, M., Bonera, L.: Il legno lamellare, il calcolo (Glued laminated timber, design), in Italian,
Habitat Legno editor, Edolo (BS), Italy, 1989.

Gross, H., Crocetti, R.: Limtrahandbok Del 1 — 3, in Swedish, Svenskt Tra, Stockholm, 2016.
Holzbau Handbuch: Reihe 2 Teile 2 12 folge 1, Verbande und Abstutzungen — Grundlagen, Regelnachweise
(Connections and bracing — basics, design guide), in German, Informationdienst Holz, Deutschen
Gesellschaft fur Holzforschung e. V., Munich, Germany, 2000.

Kallsner, B., Girhammar, U.A.: Horisontalstabilisering av traregelstommar — Plastisk dimensionering av

vaggar med trabaserade skivor (Horizontal stabilisation of timber buildings — Plastic design of wood-based
light-frame shear walls), in Swedish. SP Wood Technology — Technical Research Institute of Sweden,
Stockholm, Sweden, 2009.

Piazza, M., Tomasi, R., Modena, R.: Strutture in Legno (Timber Structures), in Italian,
Hoepli, Milan, Italy, 2005.

Porteous, J., Kermani, J.: Structural Timber Design to Eurocode 5, Blackwell Science Ltd, Oxford, UK, 2007.
Schodek, D.L.: Structures, 2nd edition, Prentice-Hall, New Jersey, 1992.

Schueller, W.: Horizontal-Span Building Structures, John Wiley & Sons, New York, 1983.
Stalnaker, J.J., Harris, E.C.: Structural Design in Wood. Van Nostrand Reinhold, New York, 1989.

Sorensen, J.D., Dietsch, P., Kirkegaard, P.H., Kohler, J.: Design for Robustness of Timber Structures.
COST Action E55 “Modelling of the Performance of Timber Structures”. European Science Foundation,
Brussels, 2011. ISBN 978-3-8322-9949-1.
Timoshenko, S., Gere, J.M.: Theory of Elastic Stability (2nd edition).
McGraw Hill International Book Company, Singapore, 1963.

Thelandersson, S., Larsen, H.J.: Timber Engineering. John Wiley & Sons Ltd. West Sussex, UK, 2003.
Vessby, J.: Analysis of shear walls for multi-storey timber buildings. PhD thesis, Linnaeus University,
Vaxjo, Sweden, 2011.

Winter, G.: Lateral bracing of columns and beams. Proc. ASCE 1958,84 (ST2), 1561-1-1561-22, 1958.

Yura, J.: Winter’s bracing approach revisited. Engineering Structures, Vol. 18, No. 10, pp. 821 — 825,
Elsevier Science Ltd, UK, 1996.

Chapter 7 - Design for serviceability (SLS)

Brunskog, J., Johansson, A.-C. Bahtijaragic,Z.: Svikt och vibrationer i bjalklag,
in Swedish, Bygg & teknik 3, 2006.

Hu, L.J., Chui, Y. H., Onysko, D.,: Vibration serviceability of timber floors in residential construction,
Prog. Struct. Enging. Mater, 3:228 — 237, 2001.
Martensson, A.: Short and Long-term Deformations of Timber Structures, Timber Engineering,
Eds: Larsen H.J., Thelandersson S., Wiley & Sons, 2003.

Ohlsson, S.: Svikt, svangningar och styvhet hos bjalklag — dimensioneringsmetoder,

in Swedish, Byggforskningsradet, 1984.

Design of timber structures - Volume 1 249

Bibliography

Ohlsson, S.: Serviceability criteria — especially floor vibration criteria.

1991 International timber engineering conference. TRADA. Vol. 1, 58 — 65, 1991.

Ozelton, E.C., Baird, J.A.: Timber Designers’ Manual; Blackwell Science Ltd, 2006.
Porteous J., Kermani, A.: Structural Timber Design, Blackwell Science Ltd, 2007.

Ranta-Maunus, A., Kortesmaa, M.: Creep of timber during eight years in natural environments.
WCTE2000 Conference. Whistler, CA, 6 p, 2000.
Smith, I.: Vibrations of Timber Floors: Serviceability Aspects. Timber Engineering,
Eds: Thelandersson, S., Larsen, H.J., John Wiley & Sons Ltd. West Sussex, UK, 2003.

Talja, A., Toratti, T.: Classification of Human Induced Floor Vibrations.

Journal of Building Acoustics, Vol. 13, No. 3, 211—221, 2006.

Chapter 8 - Timber building systems for housing

Axelson, K., Larsson, B., Sandberg, S., Soderlind, L.: Weather protected production.
Possibilities and experiences (in Swedish). Report FoU-Vast 0404, Sveriges Byggindustrier,
Gothenburg, Sweden, 2004.
Boverket: Ett utvecklings- och informationsprojekt for trahusbyggande i massivtra.
Internet: Sundvalls inre hamn, Boverket, 2006. ISBN 91-7147-963-5.

Fire Safety in Timber Buildings — Technical Guideline for Europe.

SP report 2010:19, SP Technical Research Institute of Sweden, Stockholm, Sweden, 2010.

Forssen, J., Kropp, W., Brunskog, J., Ljunggren, S., Bard, D., Sandberg, G., Ljunggren, F., Agren, A., Hallstrom,
O., Dybro, H., Saint-Gobain, I., Larsson, K., Tillberg, K. Jarnero, K., Sjokvist, L-G., Ostman, B., Hagberg, K.,
Blomkvist, A., Olsson, A., Ekstrand C-G., Johansson, M.: Acoustics in wooden buildings. State of the art 2008.
Vinnova project 2007-01653. SP Rapport 2008:16, SP Hallbar Samhallsbyggnad, Stockholm, 2008.

Girhammar, U.A., Kallsner, B.: Horisontalstabilisering av traregelstommar. Plastisk dimensionering (in

Swedish). Handbook, SP Technical Research Institute of Sweden, Stockholm, Sweden, 2010.

Gross, H., Crocetti, R.: Limtrahandbok Del 1 — 3, in Swedish, Svenskt Tra, Stockholm, 2016.

Gyproc’s handbook, in Swedish, www.gyproc.se

KLH: www.klh.at
Lennartsson, M.: Modularity in Industrialised Timber Housing. A Lean Approach to Develop
Building Service Systems. Lulea University of Technology, Licentiate thesis, 2009.

Lindbacks: www.lindbacks.se
Ljunggren, S.: Ljudisolering i trahus — en handbok for konstruktorer, in Swedish.
SP Swedish Technical Research Institute, Stockholm, 2011.

Lukaszewska, E.: Development of Prefabricated Timber-Concrete Floors.

Doctoral thesis, Lulea University of Technology, 2009.
Martinsons: www.martinsons.se.

MFB Handbok, in Swedish, www.mfbmiks.se

Moelven Toreboda, www.moelven.com/se

Natterer, J., Hamm, J., Favre, P.: Composite wood-concrete floors for multi-storey buildings.
Proc. Of the International Wood Engineering Conference, 431 — 435, 1996.
Nordlund, P.: Horisontalstabilisering av hoga trahus via trapphuset (in Swedish).
Master thesis, Lulea University of Technology, 2010.

250 Design of timber structures - Volume 1

 Bibliography

Pousette, A., Sandberg, K., Ekstedt, J.: Trabalkonger. SP Rapport 2011:82,

SP Sveriges Tekniska Forskningsinstitut, 2011. ISBN: 9789187017155.

Setra Tralyftet, www.setragroup.com

Solid wood handbook, www.martinsons.se, 2010

Svenskt Tra, www.svenskttra.se

Tavoussi, K., Winter, W., Pixner, T.: Development of earthquake bracing system for multi-storey buildings
using slender shear wall elements in cross-laminated timber (CLT), World Conference on Timber Engineering,
Miyazaki, Japan, 2008.

The Open Timber Construction System: The Open Timber Construction System
— Prefabricated Element Structures. WoodFocus, Finland, 2003.

TraGuiden, The Swedish Forestry Industries Federation (Skogsindustrierna), www.traguiden.se, (In Swedish).

Tyrens & Derome: Traplattformen — en plattform for industriellt trabyggande. CBBT, 2011.

Vessby, J.: Shear walls for multi-storey timber buildings. Licentiate thesis, School of Technology and Design,
Vaxjo University, Sweden, 2008.
Ostman, B., Arvidson, M., Nystedt, F.: Boendesprinkler raddar liv — Erfarenheter och
brandskyddsprojektering med nya mojligheter (in Swedish), SP Rapport, SP Hallbar Samhallsbyggnad, 2002.

Chapter 9 - Structural systems for infrastructure

Ekeberg, P.K.: Technical concepts for long span timber bridges. Proceedings,
International Conference, Timber bridges, ICTB2010, Lillehammer, Norway, 12 — 15 september 2010.

Kleppe, O.: Durability of Norwegian timber bridges. Proceedings, International Conference,

Timber bridges, ICTB2010, Lillehammer, Norway, 12 — 15 september 2010.
Pousette, A.: Trabroar — konstruktion och dimensionering. SP Hallbar Samhallsbyggnad.
SP RAPP 2008:50, ISSN 0284-5172, 2008. ISBN 978-91-85829-73-6. (In Swedish)

Pousette, A., Jacobsson, P., Gustafsson, M., Horttanainen, J., Dahl, K.:
Stress Laminated Bridge Decks, Part II. Tratek — Swedish Institute for Wood Technology Research,
Nordic Timber Bridge Project, Nordic Wood, 2002.
Pousette, A., Fjellstrom, P.-A.: Inspection of timber bridges, Proceedings of 7th
International conference on short and medium span bridges, Montreal, Canada, 2006.

Ritter, M.A.: Timber bridges: design, construction, inspection, and maintenance.

Forest Service, Washington, U.S., 1990.
Taylor, R.J., Keenan, F.J.: Wood highway bridges, Canadian Wood Council, Ottawa, Ontario, Canada, 1992.

TraGuiden, The Swedish Forestry Industries Federation (Skogsindustrierna), www.traguiden.se, (In Swedish).

References for standards

EN 338:2016 Structural timber — Strength classes.
EN 383:2007 Timber structures — Test methods — Determination of embedding strength
and foundation values for dowel type fasteners.

EN 384:2010 Structural timber — Determination of characteristic values of mechanical

properties and density.
EN 408:2010/A1:2012 Timber structures — Structural timber and glued laminated timber
— Determination of some physical and mechanical properties.

Design of timber structures - Volume 1 251

Bibliography

EN 409:2009 Timber structures — Test methods — Determination of the yield moment of

dowel type fasteners.

EN 1075:2014 Timber structures — Test methods — Joints made with punched metal plate fasteners.
EN 1382:2000 Timber structures — Test methods — Withdrawal capacity of timber fasteners.

EN 1383:2000 Timber structures — Test methods — Pull through resistance of timber fasteners.
EN 1990:2002/A1:2005/AC:2009/AC:2010 Eurocode 0: Basis of Structural Design.

EN 1991-1-1:2002/AC:2009 Eurocode 1: Actions on structures — Part 1 — 1:

General actions — Densities, self-weight, imposed loads for buildings.

EN 1991-1-2:2002/AC:2009/AC:2013/AC2:2013 Eurocode 1:
Actions on structures — Part 1 — 2: General actions — Actions on structures exposed to fire.

EN 1991-1-3:2003/AC:2009 Eurocode 1:
Actions on structures — Part 1 — 3: General actions — Snow loads.

EN 1991-1-4:2005/AC:2009/AC:2010/A1:2010 Eurocode 1:
Actions on structures — Part 1 — 4: General actions — Wind actions.
EN 1991-1-5:2003/AC:2009 Eurocode 1:
Actions on structures — Part 1 — 5: General actions — Thermal actions.

EN 1991-1-6:2005/AC:2008/AC:2013 Eurocode 1:
Actions on structures — Part 1 — 6: General actions — Actions during execution.
EN 1991-1-7:2006/AC:2010/A1:2014 Eurocode 1:
Actions on structures — Part 1 — 7: General actions — Accidental actions.

EN 1991-2:2003/AC:2010 Eurocode 1:
Actions on structures — Part 2: Traffic loads on bridges.
EN 1993-1-5:2006/AC:2009 Eurocode 3:
Design of steel structures — Part 1 — 5: Plated structural elements.

EN 1995-1-1:2004/AC:2006/A1:2008/A2:2014 Eurocode 5:
Design of timber structures — Part 1 — 1: General — Common rules and rules for buildings.
EN 1995-2:2004 Eurocode 5: Design of timber structures — Part 2: Bridges.

EN 14080:2013 Timber structures — Glued laminated timber and glued solid timber
— Requirements.
EN 14358:2006 Timber structures — Calculation of characteristic 5-percentile values and
acceptance criteria for a sample.

EN 14545:2008 Timber structures — Connectors — Requirements.

EN 14592:2008+A1:2012 Timber structures — Dowel-type fasteners — Requirements.

EN ISO 8560:1986 (1999) Technical drawings — Construction drawings — Representation of modular sizes,
lines and grids.

252 Design of timber structures - Volume 1

 Non-liability
Copyright

Non-liability
By using the content of Design of timber structures - Volume 1, you agree to the terms of use below.

The authors and reviewers have been responsible for interpreting building regulations, research
papers, industrial documents and so on, with a view to conveying current design practices.
The material presented is only intended to provide guidance; the final design responsibility lies with
the structural engineer or designer.

All information is provided in its present state and without any form of guarantee, to the extent
permitted by prevailing legislation. Although the publisher has taken all reasonable steps to try and
provide reliable information in Design of timber structures - Volume 1, the publisher does not guarantee
that the content is free from errors, mistakes and/or omissions, or that the content is up to date and
relevant to the user’s needs.

The publisher, Swedish Forest Industries Federation, also provides no guarantees for any outcomes
arising from use of the information contained in Design of timber structures - Volume 1. All use of
information in Design of timber structures - Volume 1 is the user’s own responsibility and is done at
the user’s risk.

Swedish Forest Industries Federation does not accept liability for any damage or injury that may arise
due to the content of Design of timber structures - Volume 1.

Copyright
Swedish Forest Industries Federation owns the rights to the content
Design of timber structures - Volume 1.

The content is protected under copyright law. Misuse will be subject to a penalty.
Copying of the content is prohibited.

Design of timber structures - Volume 1 253

 Chapter 6
Design of timber structures Roberto Crocetti — Faculty of Engineering,
- Volume 1 LTH, Lund University — Division of Building Science,
© Swedish Forest Industries Federation, 2022 Lund, Sweden
Third UK edition
Chapter 7
Annika Martensson — Faculty of Engineering,
Publisher LTH, Lund University — Division of Building Science,
Swedish Forest Industries Federation Lund, Sweden
Swedish Wood
P.O Box 55525 Chapter 9
SE-102 04 STOCKHOLM Anna Pousette — RISE Research Institutes of Sweden
Sweden
Tel: +46 8762 72 60 Fact checkers
Anders Gustafsson, RISE Research Institutes of Sweden
Marie Johansson, RISE Research Institutes of Sweden

Language reviewer
Patrice Godonou — Swedish Wood
Project managers
Johan Frobel — Swedish Wood Illustrations
Patrice Godonou — Swedish Wood Vendela Martinac — Thelander Arkitektur & Design AB
Cornelia Thelander — Thelander Arkitektur & Design AB
Editor
Patrice Godonou — Swedish Wood Photo credits
Tommy Andersson, page 182
Authors Johan Ardefors, page 208 (top)
Chapter 1 Per Bergkvist, page 19
Robert Kliger — Chalmers University of Technology, Anders Bjornfot, page 217 (top)
Structural Engineering, Goteborg, Sweden Patrick Degerman, page 47, 200, 208 (middle and lower)
Bertil Enquist, page 29
Chapter 2 Ake E:son Lindman, page 4, 9, 15, 43, 53, 57 (right), 116, 198
Marie Johansson — LNU, Linneus University, Gremo, page 25 (right)
Department of Building Technology, Vaxjo, Sweden Anders Gustafsson, page 217 (lower)
Thomas Harrysson, page 212
Chapter 3 Soren Hakanlind, page 209 (top and lower right), 119,
Roberto Crocetti and Annika Martensson — 189, 191
Faculty of Engineering, LTH, Lund University, Marie Johansson, page 24
Division of Building Science, Lund, Sweden Kerstin Jonsson, page 26, 27
Martinson Group AB, page 240
Chapter 4 and 8 Rottne Industri AB, page 25 (left)
Helena Lidelow — LTU, Lulea University of Technology, Setra Travaror AB, page 11
Department of Civil, Environmental and Natural Stora Enso, page 202
Resources Engineering, Lulea, Sweden Gosta Wendelius, page 209 (lower left)

Chapter 5 Graphic production

Bert Norlin — KTH, Royal Institute of Technology, Origoform
Department of Civil and Architectural Engineering,
Stockholm, Sweden ISBN 978-91-985212-5-2

254 Design of timber structures - Volume 1

Publications and websites from Swedish Wood
Publications
Order via www.swedishwood.com/publications.

Design of timber Design of timber Design of timber

structures, Volume 1 structures, Volume 2 structures, Volume 3
Structural aspects of Rules and formulas Examples.
timber construction. according to Eurocode 5. 64 pages. A4 format.
316 pages. A4 format. 64 pages. A4 format.

Commercial Grading Joinery Handbook Tra! Magazine

of Timber Everything for small Quarterly magazine show­
The recently updated basic carpenters and students, casing the best in timber
manual for grading timber. from design to machine architecture and interior
81 pages. A5 format. knowledge and surface design from Sweden and
treatments. beyond.
120 pages. A4 format.
 SWEDISH
ffl® WOOD
Swedish Wood’s aim is to increase the size and value of the market for P.O Box 55525
Swedish wood and wood products in construction, interior design and pack­ SE-102 04 Stockholm
aging. Through inspiration, information and education, we promote wood as
a competitive, renewable, versatile and natural material. Swedish Wood also Sweden
lobbies on behalf of its members on key industry andtrade issues. Tel: +46 8 762 72 60
Swedish Wood represents the Swedish sawmill industry and is part of
the Swedish Forest Industries Federation. In addition, Swedish Wood repre­
sents the Swedish glulam and packaging industries, and collaborates closely
with Swedish builders’ merchants and wholesalers of wood products. ISBN 978-91-985212-5-2
2022-04

© Swedish Forest Industries Federation, 2022.