Joint Workshop of COST Actions TU0601 and E55
September 21-22 2009, Ljubljana, Slovenia
Behaviour and modelling of timber structures with reference to
robustness
Sven Thelandersson
Lund University, Sweden
Daniel Honfi
Lund University, Sweden
Summary
A general description of the behaviour and modelling of timber structures in relation to
requirements for robustness of structural systems is given. Basic material characteristics for
wood and different types of wood products are described as well as the most common types
of joints. A brief review of expected failure modes for elements, joints and typical wood
based structural systems is included. Structural properties such as resistance, ductility and
redundancy are described for timber elements and typical timber structural systems.
The properties of timber and various wood based products are not ideal for achieving robust
systems due to the risk for brittle behaviour and the anisotropic nature of wood in certain
loading modes. Timber also exhibits large spatial variability which must be accounted for in
modelling and assessment of reliability. On the other hand, dowel type joints which are
frequently used in timber structures show high ductility when properly designed. The
relatively low self weight of timber structures implies low loads in post failure scenarios and
limited dynamic effects. Appropriate models are available to analyse structural timber
systems in post failure modes. Good results can often be achieved by elastic material
models for timber components and non-linear models for joints combined with geometrically
non-linear analysis.
Keywords
Timber, engineered wood products, joints, ductility, brittleness, anisotropy, shear walls,
robustness
Background / Introduction
According to the Memorandum of Understanding, Activity 5 of COST TU0601 deals with
structural behaviour models of relevance for robustness. This fact sheet gives an overview
of the behaviour of timber structures in relation to requirements for robustness of structural
systems. Structural properties such as resistance, ductility and redundancy are described for
timber elements and typical timber structural systems. Basic material characteristics and
different types of wood products are described as well as the most common types of joints. A
125
Behaviour and modelling of timber structures with reference to robustness
brief review of different failure modes for elements, joints and typical wood based structural
systems is included.
Problem statement / Key issues
In the context of robustness the main use of structural models is to describe the degree of
sensitivity of structural systems to extraordinary types of exposures often leading to local
failure. The requirements on the structural model should be that it must predict the mode of
failure and post-failure behaviour in a realistic manner. Key issues are ductility, energy
absorption, redundancy and geometrical non-linearity. When considering timber structures
as a class it is also important to include different types of engineering wood products apart
from sawn timber. Such products can have different properties of importance for the
structural response. Furthermore, a very important part of structural systems are its joints.
This is particularly important for timber structures, where a great variety of joints can be
found in practice.
Limitations
The discussion in this fact sheet mainly concerns issues of relevance for such timber
structural systems where robustness can be expected to be an issue due to the importance
of the built facility.
Basic material characteristics of wood
The material wood is biologically “produced” in the growing tree to meet the needs of the
tree itself. The wood cells are predominantly oriented in one direction, called the fibre
direction or grain direction. This is the strong direction, generally parallel with the longitudinal
axis of the stem. In the vicinity of branches, the fibres are redirected to form very efficient
“structural” joints between the stem and the branches, see Figure 1.
126
Behaviour and modelling of timber structures with reference to robustness
a
b
Figure 1: a) The fibre orientation around a knot creates a structurally efficient joint between
branch and stem; b) The fibre structure resembles flow lines around the knot
Both strength and stiffness in the fibre direction are very large in relation to the weight of the
material, especially in tension. Table 1 shows a comparison of strength/density ratios for
some structural materials.
Material Density Strength, Strength/Density
kg/m3 MPa 10-3 MPa⋅ m3 /kg
Structural steel 7800 400-1000 50-130
Aluminium 2700 100-300 40-110
Concrete, compression 2300 30- 120 13-50
Clear softwood, tension 400-600 40-200 100-300
Clear softwood, compression 400-600 30-90 70-150
Structural timber, tension 400-600 15-40 30-80
Table 1: Strength/density ratios for some structural materials
For clear wood this ratio is significantly higher than for other building materials. Also in the
case of structural timber, where natural defects reduce strength, the strength in relation to
weight is of the same order of magnitude as ordinary structural steel.
Structural timber
Sawn timber in structural dimensions is a non-homogeneous material, which contains growth
defects in the form of knots, zones with compression wood, oblique fibre orientation, etc.
Such growth characteristics, which once were created to serve the needs of the tree, will
127
Behaviour and modelling of timber structures with reference to robustness
usually reduce the strength significantly when the timber is cut and used for other purposes.
This is illustrated in Figure 2.
Clear wood Structural timber
Figure 2: The difference between clear wood and structural timber
Due to the large spatial and random variability in strength and stiffness, the mechanical
behaviour of timber can not be derived with any reliability from properties of clear wood. The
presence and character of knots and other defects vary from one timber board to another,
which means that the structural properties of sawn timber exhibit a significant variability.
Strength properties of structural timber are therefore usually determined by direct testing of
timber elements according to a standardised methodology, and strength is defined on the
element level rather than on the material level. Strength data associated with structural
timber therefore reflects moment, tension, compression and shear capacity of a timber
element, even if the data are expressed in stress units, obtained assuming that the theory of
elasticity is valid. The bending strength, for instance, is defined as the moment capacity of a
timber beam determined by testing, divided by the elastic section modulus. The influence of
defects is implicitly included in the strength values specified, and they can only be applied if
the stresses are determined by elastic theory.
Figure 3: Typical cumulative distributions for strength of structural timber (Spruce, Picea
Abies, grown in Scandinavia) in different loading modes. Hansson & Thelandersson (2001)
Consequently, different strength values are valid for different loading modes such as
bending, tension and compression. Figure 3 shows typical cumulative distribution functions
for strength of timber in tension, compression and bending. The coefficient of variation for
128
Behaviour and modelling of timber structures with reference to robustness
bending strength of structural timber is in the range 20-40 %, depending on species and
growth conditions as well as on the method and strategy used to grade the timber.
Anisotropy of wood
Wood is anisotropic, i.e. its physical properties depend on direction. The tension strength is
30-50 times smaller perpendicular to grain than in the fibre direction. The weakness
exhibited by wood in cross grain directions must always be carefully considered in design
and detailing of timber structures. Tension failure in this direction is also highly brittle.
Failures of timber structures observed in practice are often due to tension perpendicular to
grain, which commonly occurs e.g. in curved elements, in joints in timber structures, see
Figure 4 , and in discontinuities created by wholes and notches in timber members. The
effective load bearing capacity of timber members under tension perpendicular to grain
depends on the magnitude of the stressed volume since the probability of failure increases
with volume.
Our present understanding of fracture processes related to perpendicular to grain failure is
limited. In engineering design, empirical rules for detailing, such as minimum recommended
distance between fasteners as well as edge distances in mechanical joints, are commonly
employed to avoid the risk for splitting failure. The reliability of such design methods is
limited, however, especially when they are applied in large scale real world situations,
different from the laboratory tests they are derived from.
Wood has also low strength in compression perpendicular to grain, but here the behaviour is
very ductile, which may be utilised in design for robustness.
The strong anisotropy in wood is also valid for stiffness properties. The elastic modulus
perpendicular to grain is typically 50-80 times smaller than parallel to grain. When wood is
subjected to compression perpendicular to grain significant deformations will occur already
at low load levels.
Figure 4: Failure in tension perpendicular to grain in timber joint.
129
Behaviour and modelling of timber structures with reference to robustness
Engineered wood products
Engineered wood products (EWP) represent a broad class of materials intended for
structural use. They are typically produced from wood which has been processed to smaller
fractions by sawing, peeling, chipping, slicing or defibration. The wood constituents used for
EWP can be sawn laminations, veneers, strands, flakes or sawdust. The constituents are
bonded together to form panel products, timber-like elements in different sizes, or shaped
structural products. The bonding between constituents is most often made with adhesives
mixed with the constituents or sprayed on their surfaces, with application of heat and
pressure in the production process. One type of products is made by gluing solid sawn
timber or laminations into larger structural members. Some examples are shown in Figure 5.
Among advantages with EWP can be mentioned that the strength reducing effects of defects
present in solid wood will be more or less neutralised, depending on the type of product. The
growth defects of solid wood are either removed in the production process or distributed in
the finished product so that the strength is less affected and the variability of the product
becomes smaller. This means a more efficient utilisation of the material. As an example, the
probability density functions for glulam and structural timber are compared in Figure 6.
Although glulam is made from laminations of structural timber, the decisive strength, defined
as the 5th percentile, is significantly higher for glulam than for timber. The reason for this is
the load sharing between laminations in the glulam, which allows locally weak zones to
redistribute stress to adjacent stronger regions. Similar mechanisms affect the properties for
other types of EWP.
a b c d
Figure 5: Some examples of Engineered Wood Products.
a) LVL b) Composite I-beam c) Glulam d) Parallel strand lumber
130
Behaviour and modelling of timber structures with reference to robustness
Difference in 5th Difference in 5th
Frequency percentile percentile
Glulam
Timber
Strength
Figure 6: Typical probability density functions for structural timber and glulam.
Ductility/brittleness characteristics
One of the features related to robustness is the ability to redistribute stresses in structural
systems, which normally requires large deformation capacity. On a local level, timber has
generally low deformation capacity and often displays brittle type behaviour. This makes it
difficult to utilize redundancy as a mean to achieve robustness.
The response depends strongly on direction and type of loading. When subjected to
compression the response is rather ductile for loading both in the grain direction and
perpendicular to grain. However when loaded in tension, shear or bending the response is
brittle with very little deformation capacity. Fracture energy dissipation in tension and shear
is normally negligible in relation to the ductility required for robustness. The weakness in
tension perpendicular to grain makes it necessary to use some type of reinforcement if
ductility is desired. Reinforcement with fibre composites or metal screws has been shown to
be possible, but is presently little used in practice.
Non-linear material models for timber under compression are available and used for special
purposes. The most relevant, yet reasonably simple model for loading in tension or shear is
an elastic-brittle model. However, for investigations of brittle failure modes, advanced
models based on linear or non-linear fracture mechanics are being used, see e.g.
Gustafsson (2003). But the latter types of models are seldom relevant for problems where
large deformations are necessary.
Some types of engineered wood products have slightly more advantageous properties from
ductility point of view, but in general loading modes involving tension show limited ductility.
131
Behaviour and modelling of timber structures with reference to robustness
Joints in timber structures
Joints with dowel-type fasteners are the most common joints in timber structures. Dowel-
type fasteners include nails, staples, screws, drift pins, threaded rods and bolts. Figure 7
shows an opened nailed connection loaded in single shear after a test to determine the load-
carrying capacity. Different phenomena are observed:
• A relative displacement between the middle member and the side members occurred.
• The nails show plastic deformations due to bending.
• The timber close to the joint between the members is deformed plastically under the
action of the nails.
• The nail on the left is partly pulled out.
Since the load is transferred through compression in the wood the response is usually very
ductile. Another contribution to ductility comes from plastic yielding of the dowels. Joints
with other types of dowels such as bolts, screws or threaded rods show similar behaviour.
From these observations, three main parameters influencing the load-carrying behaviour of
joints with dowel-type fasteners like nails can be identified:
• The bending capacity of the dowel. The bending capacity is mainly influenced by the
dowel diameter and the yield strength of the dowel material. Plastic deformation capacity
is essential to provide bending capacity also after considerable deformation of the dowel.
• The embedding capacity of the timber or wood-based material. The embedding strength
primarily depends on the timber density, whereas the embedding capacity because of
the contact area also depends on the fastener diameter and penetration depth.
• The withdrawal strength of the dowel. Threaded fasteners provide higher withdrawal
capacities than smooth fasteners.
A simple limit state theory to predict the load capacity of timber connections with dowel-type
fasteners loaded perpendicular to the fastener axis was developed by Johansen (1949).
Johansen considered the dowel as a beam embedded in the timber. Apart from the joint's
geometry, the bearing capacity of the joint members and the bending capacity of the
fastener determine the failure mode and the load-carrying capacity of the joint.
132
Behaviour and modelling of timber structures with reference to robustness
Figure 7: Opened nailed connection after a test to determine the load-carrying capacity in
shear. Blass (2003)
More advanced models are available to predict of the load-slip behaviour of dowel type joint,
which is necessary from a robustness point of view. E.g. Foschi and Bonac (1977) used
non-linear finite-element analysis to derive load-slip characteristics for nailed connections,
properties required for a sophisticated non-linear analysis of structural systems in post
damage scenarios.
Although less common, adhesive joints are also used for timber structures. Commonly used
adhesive joints today have usually limited ductility. Energy absorbing and ductile adhesive
joints are being developed to some extent but are not ready for commercial applications.
Typical timber structural systems and behaviour related to robustness
As described above brittleness and material response of timber and other wood based
products are not ideal for achieving redundancy and load redistribution in structural timber
systems. However, there are some commonly used structural components which are
favourable for providing ductility and redundancy in structural systems. Widely used systems
with such properties are
• Light weight timber frame building systems
• Building systems based on solid planar wood panels
Timber frame buildings are built up by a skeleton of timber joists and studs covered with
panels fastened to the wood skeleton by metal fasteners. Two principal structural
components can be found in such systems:
• Shear walls (usually vertical)
• Diaphragms (usually horizontal)
133
Behaviour and modelling of timber structures with reference to robustness
The function of these components in a building is illustrated in Figure 8. The anatomy of a
typical shear wall is illustrated in Figure 9a and the behaviour under shear loading is
illustrated in Figure 9b.
Diaphragm
Shear
wall
Figure 8: Force transfer in multi-storey timber frame building under lateral loading.
a) b)
Figure 9a: Wood frame shear wall with plywood sheathing
Figure 9b: Shear load transfer in sheathed wall element
Shear walls as well as diaphragms designed in this way exhibit high ductility for in-plane
loading. This is achieved by the ductility in the metal connections between panels and
framing. Timber frame shear walls and diaphragms display large deformation capacity, when
134
Behaviour and modelling of timber structures with reference to robustness
properly designed. Such components are very effective in absorbing energy as illustrated in
Figure 10, showing the response under cyclic loading.
250 250
200 200
150 150
100 100
Design Load
50
Load (KN)
50 Design Load
Load (KN)
0
0
-50
-50
-100
-100
-150
-150
-200
-200
-250
-16 -12 -8 -4 0 4 8 12 16 -250
Displacement (mm) -16 -12 -8 -4 0 4 8 12 16
Displacement (mm)
a b
Figure 10: a) Typical response of shear wall with high ductility and energy absorbing
capacity. b) Typical response of shear wall with less ductility
Advanced methods for non-linear modelling of composite timber frame structures under
dynamic loading have been developed to a large extent, mainly in connection with research
related to earthquake resistance, see e.g. Karacabeily & Cecotti (2000). These analysis
methods can also be used to analyse and verify robustness performance.
Robust building structures can be designed on the basis of these types of timber frame wall
and floor components. Lightweight timber frame has also the advantage of low self weight
which reduces dynamic effects and required load bearing capacity in post failure situations.
Similar advantages can be achieved with solid wood planar panels which are now being
developed and used in buildings and bridges. Some of these products are shown in Figure
11. Modelling of robustness characteristics for structural systems built up from these types of
elements may be done by rather simple sub-models for the elements and non-linear models
for the joints connecting the elements.
135
Behaviour and modelling of timber structures with reference to robustness
Structural
joint
Cross laminated floor element Cross laminated wall element
Structural joint with
self tapping inclined
screws
Prestressed floor element
Figure 11: Typical solid planar wood elements
Robustness of long-span timber structures
Timber is often used for primary load-bearing elements in single storey long-span structures
for public buildings and arenas, where severe consequences can be expected if one or more
of the primary load bearing elements fail. A typical topology of such buildings is shown in
Figure 12. The system consists of main frames, secondary elements and bracing elements.
The main frame or columns and girders can be seen as key elements in the system and
should be designed with high safety against failure and under strict quality control. The main
frames may sometimes be designed with moment resisting joints between columns and
horizontal girders, but it is generally difficult to provide ductility in such timber frames due to
the inherent brittleness of the material. For this reason, scenarios, where one or more of
these key elements, fail should be considered at least for high consequence buildings. Two
alternative strategies may be applied:
• Isolation of collapsing sections
• Provision of alternate load paths
136
Behaviour and modelling of timber structures with reference to robustness
The first one is rather easy to provide by deliberately designing the secondary structural
system less strong and stiff or by using so called fuse elements in joints. But for large spans
and typical center distances large roof areas are still affected, leading to severe
consequences of failure of one main girder in buildings where many people may be present.
In that case it is desirable to design the secondary structural system and the bracing system
so that loss of capacity in the main girder or in one of the columns does not lead to collapse.
This may be achieved by appropriate joint design and utilization of e.g. catenary action, even
if it is often expensive and difficult for systems with timber as the main structural material.
Figure 12: Typical topology of long-span timber building
Summary and conclusions
The properties of timber and various wood based products are not ideal for achieving robust
systems due to the risk for brittle behaviour and the anisotropic nature of wood in certain
loading modes. Timber also exhibits large spatial variability which must be accounted for in
modelling and assessment of reliability. On the contrary, dowel type joints which are
frequently used in timber structures show high ductility when properly designed. The
relatively low self weight of timber structures implies low loads in post failure scenarios and
reduces dynamic effects. Appropriate models are available to analyse structural timber
systems in post failure modes. Good results can often be achieved by elastic material
models for timber components and non-linear models for joints combined with geometrically
non-linear analysis
137
Behaviour and modelling of timber structures with reference to robustness
References
Blass H.J. (2003) Joints with dowel-type fasteners. In “Timber Engineering” Ed.
Thelandersson, S., Larsen H.J., Wiley & Sons, Chichester
Foschi R.O., Bonac T. (1977). Load-slip characteristics for connections with common nails.
Wood Science, 9(3), 118-123.
Gustafsson P.J. (2003). Fracture perpendicular to grain- structural applications. In “Timber
Engineering” Ed. Thelandersson, S., Larsen H.J., Wiley & Sons, Chichester
Hansson, M., Thelandersson, S., 2003. Capacity of timber roof trusses considering statistical
system effects. Holz als Roh und Werkstoff, 61( 3), pp. 161-166.
Karacebeyli E., Cecotti A. (2000) Dynamic analysis of wood frame shear walls. Proceedings
12th World Conference on Earthquake Engineering, Auckland, New Zealand
138
