Engineering Structures 27 (2005) 145–154

www.elsevier.com/locate/engstruct

Optimum design of plane timber trusses considering joint flexibility

Simon Šilih, Miroslav Premrov, Stojan Kravanja∗
Faculty of Civil Engineering, University of Maribor, Smetanova 17, 2000 Maribor, Slovenia

Received 5 May 2004; received in revised form 20 September 2004; accepted 6 October 2004
Available online 26 November 2004

Abstract

The paper presents the optimization of metal-plate-connected plane timber trusses with special emphasis on joint flexibility. The
optimization was performed by the non-linear programming approach. Since various truss design parameters such as type of truss
configuration, span/depth ratio, number and type of diagonal and vertical members and type of joint connections simultaneously affect each
other, it is proposed that all of these parameters should be considered simultaneously in a single mathematical model. An optimization model
for cost optimization of timber trusses was thus developed. The economic objective function for minimizing the structure’s self-manufacturing
costs was defined, subjected to the design, stress and deflection (in)equality constraints. The finite element equations were as the equality
constraints defined for the calculation of the internal forces and the deflections of the structure. The stiffness matrix of the structure was
composed by considering fictitiously decreased cross-section areas of all the flexibly connected elements. Constraints for the dimensioning
of the timber members were determined in accordance with Eurocode 5 in order to satisfy the requirements of both the ultimate and the
serviceability limit states. The cross-section dimensions and the number of fasteners were defined as independent optimization variables. A
numerical example demonstrates the applicability of the optimization approach presented as well as the influence of the fasteners’ flexibility
on the optimal self-manufacturing costs.
© 2004 Elsevier Ltd. All rights reserved.
Keywords: Timber; Trusses; Fasteners; Flexibility; Optimization; Non-linear programming

1. Introduction large open areas with a few or no intermediate supports is

Timber construction is an important part of the
infrastructure in a number of areas around the world. Wood
has proved to be quite a resilient material, showing relatively
high ductility and low density. In addition, the flexibility
of mechanical fasteners provides a high damping capacity
between the connected timber elements. Well-built timber
structures maintain a good performance particularly under
the influence of wind and especially earthquake forces.
In the last few decades, the application of timber
trusses has frequently been noted in all aspects of building
construction. Timber trusses have become known for their
pleasing architectural appearance, lightweight design and
easy fabrication. The use of timber trusses to bridge over
still on the increase. These trusses are essentially lighter
than the analogous beam solutions. Many magnificent
space and plane timber trusses have been constructed all
over the world, covering public halls, stadiums, exhibition
centres and many other buildings. In this field, metal-plate-
connected timber trusses have been found to be favourable
structures for roof framings for spans greater than 20 m.
1.1. Main timber truss design parameters
In order to design a satisfactory and optimal timber truss
structure with the given load, span and boundary conditions,
some main design parameters need to be considered, on
which timber truss behaviour basically depends:

∗ Corresponding author. Tel.: +386 2 22 94 300; fax: +386 2 25 24 179. – type of truss configuration,
E-mail address: stojan.kravanja@uni-mb.si (S. Kravanja). – span/depth ratio,

0141-0296/$ - see front matter © 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engstruct.2004.10.001
146 S. Šilih et al. / Engineering Structures 27 (2005) 145–154
– number and type of intermediate members (diagonals and
verticals),
– type of joint connections.
The type of truss configuration is the first design
parameter, which has a significant influence on the
resistance, stiffness, weight and costs of the truss. The basic
configuration types of trusses are the flat, pitched or the
bowstring type. Many alternative types for different spans
can be found for example in Hoyle and Woeste [1] and
El-Sheikh [2]. Different inclinations of truss chords lead to
different values of axial forces and stresses in members. In
this sense, the pitched truss is for example the most efficient
type of configuration for the given centre-point imposed
loads. The flat truss is from the point of view of static
analysis not so pleasant a truss configuration, but it is often
selected for reasons such as easier serviceability of roofs, flat
ceilings.
The span/depth ratio (S/D) is the second main design
parameter. Truss stiffness increases as the S/D ratio
decreases. The decreasing of the S/D ratio (the increasing
of the truss depth) results in a larger intermediate distance
between the truss chords. The internal forces of the chord
thus decrease and, consequently, smaller chord members
may be used. A lower S/D ratio also results in a larger
inclination of the diagonal members with smaller internal
forces. The effective lengths of the compression diagonal
and vertical members become longer. The compression
elements consequently show slight compression resistance.
Finding an optimal S/D ratio is therefore a relatively
comprehensive task. The generally recommended S/D
ratios for various timber truss configurations are 1:6 to 1:8
for the bowstring type, 1:8 to 1:12 for the flat and 1:5 to
1:6 for the pitched configuration; see Hoyle and Woeste [1].
These ratios are not fixed and may be changed in accordance
with the use of better materials, connections or greater cross-
sections area of the chord members.
The number of intermediate members (diagonals and
verticals) represents the third main design factor. Both the
static determination and the force distribution in the truss
system significantly depend on this factor. The changing of
the number of intermediate members inevitably leads to a
number of consequences with direct bearing on the truss
costs; see El-Sheikh [2]. Basically, deeper trusses require
a lower number of intermediate members. Reducing the
number of diagonals increases the internal forces in chord
members and their effective length. The dimensions of the
compressed chord members must be significantly enlarged.
On the other hand, with the decrease in the number of
diagonals, the number of joints also decreases. This fact
is important for timber trusses, where the joints between
the chords and the connectors are not entirely as rigid as
they are in concrete or steel trusses. Undesirable slips in
joints lead to reduction in truss stiffness, which in this
way becomes smaller compared to the stiffness in trusses
with inflexible connections. Sometimes it is convenient that
timber trusses contain both the compressed timber and
tensile steel diagonals. As steel–timber connections are less
flexible than timber–timber connections, their use in timber
trusses may have an advantageous influence on the final
results.
The type of joint connections is the fourth main timber
truss design parameter. The flexibility of the joints in
timber trusses depends greatly on the selected/used type of
connections. The study presented in this paper is limited
to metal-plate-connected joints with metal fasteners (bolts,
dowels, nails etc.). Since the flexibility of bolts differs
from the flexibility of nails or dowels, the complete joint
flexibility primarily depends on the selected type of fastener.
While fastener dimensions (diameter, length) represent a
second factor, the number and disposition of the fasteners
in the connection constitute a third one. These factors are
presented in Section 2.
The main design parameters described have effects on
each other and may thus be treated together. Accordingly,
El-Sheikh [2] reported a numerical study of the S/D
ratio influence on the internal forces in truss members
considering different truss configurations and various
boundary conditions. A simple linear global analysis of
trusses was used. Since the study was performed for steel
trusses, it does not consider the influence of joint flexibility.
This paper, however, discusses timber trusses with respect
to the flexibility of the fasteners, which essentially increases
the comprehension of the task.
1.2. Design methods
When a high number of truss design parameters, designer
decisions and factors are involved in the analysis, the
designing of timber trusses can become a difficult and expen-
sive process. This has forced designers to find simpler and
cheaper alternative design methods, adequate at least for the
preliminary design state. Several approximate methods have
been developed in the recent past with different accuracies,
suitabilities and simplifications according to real truss condi-
tions; see El-Sheikh [3]. Approximate design methods which
additionally consider the flexibility of the joints in timber
trusses with respect to different diagonal members can for
example be found in Hoyle and Woeste [1], Stalnaker and
Harris [4], Steck [5] and Brüninghoff et al. [6].
The idea of the present study was to consider simultane-
ously all the design parameters and factors mentioned in the
Section 1.1 together, in a single mathematical truss model,
where structural optimization is performed rather than
classical analysis.
For more than four decades, trusses have not only been
successfully optimized but also very frequently used to
present, test and improve various optimization techniques.
Numerous research papers on this topic have been published
since the early 1960s, e.g. Schmidt [7]. While many papers
discuss the topology, shape and discrete sizing optimization
particularly of steel trusses, e.g. Lipson and Agrawai [8],
 S. Šilih et al. / Engineering Structures 27 (2005) 145–154 147

Prager [9], Šilih et al. [10], Kaveh and Kalatjari [11], and The compressed members are checked for compressive
the optimization of composite trusses, e.g. Kravanja and strength (2a) as well as for buckling criteria (2b, 2c):
Šilih [12], timber trusses have been quite neglected. fc,0,k
The paper presents the sizing optimization of metal- σc,0,d ≤ f c,0,d = kmod · (2a)
γm
plate-connected timber trusses considering the flexibility of
f c,0,k
the embedded fasteners. The optimization was performed σc,0,d ≤ kc,y · f c,0,d = kc,y · kmod · (2b)
by the non-linear programming (NLP) approach, where γm
all the design parameters mentioned in Section 1.1 were fc,0,k
σc,0,d ≤ kc,z · f c,0,d = kc,z · kmod · (2c)
simultaneously considered as (in)equality constraints. An γm
optimization model for the cost optimization of the timber where σc,0,d is the design compressive stress along the grain;
trusses was developed. An economic objective function f c,0,d is the design compressive strength along the grain,
was proposed to minimize the structure’s self-manufacturing depending on the characteristic compressive strength fc,0,k ;
costs, subjected to the design, stress and deflection kc,y and kc,z represent the coefficients which approximate
(in)equality constraints. The finite element equations were the buckling about the y and z axes in accordance with
as the equality constraints defined for the calculation of Eurocode 5 [13]. The coefficient of buckling about the y axis
the internal forces and the deflections of the structure. kc,y is determined as follows:
Constraints for the dimensioning of the timber members
1
were determined in accordance with Eurocode 5 [13] in kc,y =

order to satisfy the requirements of both the ultimate and ky + k 2y − λ2rel,y
the serviceability limit states; see Section 2. The NLP
k y = 0.5 · [1 + βc · (λrel,y − 0.3) + λ2rel,y ] (3)
optimization approach is discussed in Section 3. Since

special attention is given to the flexibility of the joints and 0.2 for solid timber
its influence on the final truss design, the latter is shown in βc =
0.1 for glued laminated timber.
Section 4.
The relative slenderness ratio λrel,y is

λy f c,0,k
2. Timber truss design criteria λrel,y = ·
π E 0,0.5
(4)
The design criteria for timber trusses are similar to ly
λy =
those for a timber beam and should be either stress- iy
controlled (ultimate limit state) or deflection-controlled ones
where λ y is the slenderness ratio corresponding to bending
(serviceability limit state).
about the y axis and l y is the buckling length of the
compressed element corresponding to bending about the y
2.1. Ultimate limit state (ULS) axis. The coefficient kc,z is obtained in a similar way.

2.1.1. Truss members 2.1.2. Fasteners

Chord and bracing (diagonal and vertical) tensile or
compressed members were checked for the ultimate limit
state, defined in Eurocode 5 [13]. Tension and compression
forces are limited by the choice of the timber quality and
its design strength. In the study, the contribution of the local
bending moments has been neglected.
The stress criteria for the tensile members with tensions
parallel to the grain σt,0,d are described by the following
expression:
f t,0,k
σt,0,d ≤ f t,0,d = kmod · , (1)
γm
where σt,0,d is the design tensile stress along the grain; f t,0,d
is the design tensile strength along the grain, depending
on the characteristic tensile strength f t,0,k ; kmod is the
modification factor, which takes into account the effect of
the duration of the load and the moisture content; and γm is
the partial safety factor for a material property. The values
for kmod are taken from Table 3.1 and the values for γm from
Table 2.2. of Eurocode 5 [13].
This study is limited to trusses, where all the timber
members are double shear connected with steel plates. Steel
plates of a thickness ts which is less than or equal to 0.5d
(d is the fastener diameter) are considered thin plates and
steel plates of a thickness greater than or equal to 1.0d
are considered thick plates. For intermediate values (the
thickness of steel plates lies between the values of 0.5d and
1.0d), a linear interpolation should be used.
Fasteners in all joint connections should satisfy the ULS
criteria in the standard form:
Fv,Rd ≥ N1d
Fv,Rk (5)
Fv,Rd = kmod ·
γm
where Fv,Rd is the design load carrying capacity per shear
plane per fastener, N1d is the ultimate force acting on a
fastener per shear plane and Fv,Rk is the characteristic load
carrying capacity per shear plane per fastener.
The values for Fv,Rk are taken from Eurocode 5 [13].
Fv,Rk for a steel plate of any thickness as the central member
148 S. Šilih et al. / Engineering Structures 27 (2005) 145–154

timber thickness or penetration depths (see Fig. 1(a), (b)),

M y,Rk denotes the characteristic fastener yield moment
and Fax,Rk is the characteristic withdrawal capacity of the
fastener.
The values for f h,1,k , f h,2,k and M y,Rk depend on the
type of fastener. They are taken from subsections 8.3, 8.5
and 8.6 of Eurocode 5 [13] for nailed, bolted and dowelled
connections, respectively. The minimum fastener spacing
and distances in the connection are defined according to
Table 8.2 for nails, Table 8.4 for bolts and Table 8.5 for
dowels.
(a) A steel plate as the central member.
2.2. Serviceability limit state (SLS)

Serviceability limit states for deflection require that the

maximal initial and final deflection must be calculated
within a specified range. The main distinction in the
calculation of the timber truss deflection, when compared
to steel or concrete truss deflection, is that an additional
deformability of the intermediate elements (diagonals and
verticals) to the upper and lower chord members has to be
considered. Mechanical fasteners in timber are not as rigid
as they are in concrete or steel, but they are flexible. Truss
(b) Steel plates as the outer members. deflection thus increases with the flexibility of the joints and
Fig. 1. Double shear connection with steel plates.
can be described as a sum:
f = f0 + f1 . (9)
of a double shear connection (Fig. 1(a)) is defined as follows: The deflection f0 represents the vertical displacement of the

 f ·t ·d (a) truss if the flexibility of the fasteners is not considered. If
 h,1,k 1


 f · t · d the influence of the local bending moments is neglected, it

  1
h,1,k

 4 · M y,Rk depends only on the axial forces Ni and the axial stiffness
 ·
 2+ −1 (E A/L)i of all i , i ∈ I , truss members in the form of
Fv,Rk = min f h,1,k · d · t12

 I

 Fax,Rk Ni · N i · L i

 + (b) f0 = , (10)

 4 E i,mean · Ai

 i=1
2.3 · M y,Rk · fh,1,k · d + Fax,Rk . (c)
4 where Ni is the axial force in the element i caused by the
(6) external load, while N i is the axial load caused by a virtual
unit force in the direction of the desired deformation; E i,mean
When thin steel plates (ts ≤ 0.5d) are designed as the outer
stands for the mean value of the modulus of elasticity. The
members of a double shear connection (Fig. 1(b)), Fv,Rk is
second term f 1 represents the deflection as a consequence
determined according to the following equation:
of the joint (fastener) flexibility of all m, m ∈ M,

 0.5 · f h,2,k · t2 · d (a) intermediate timber members (diagonals and verticals). It

Fv,Rk = min 1.15 · 2 · M y,Rk · f h,2,k · d can be accounted for in the form of

 + Fax,Rk . M 
(b) Nm · N m 1 1
4 f1 = · + , (11)
(7) K ser km,1 km,2
m=1

And, for thick steel plates (ts ≥ 1.0d) as the outer members where M represents the total number of intermediate timber
of a double shear connection (Fig. 1(b)), Fv,Rk is members, flexibly connected to the truss chords, km,1 and
km,2 are the numbers of fasteners at both ends of the
0.5 · f h,2,k · t2 · d (a)
m-th element considered and K ser denotes the fasteners’
Fv,Rk = min Fax,Rk
2.3 · M y,Rk · f h,2,k ·d+ (b) slip modulus, taken for different types of fasteners from
4
Table 7.1 of Eurocode 5 [13].
(8)
Since the intermediate members are flexibly connected,
where f h,1,k and fh,2,k are the characteristic embedment their stiffness decreases. In finite element analysis we
strengths in the timber members, t1 and t2 represent the consider the joint flexibility in such a way that cross-section
 S. Šilih et al. / Engineering Structures 27 (2005) 145–154
areas Am of all intermediate members are replaced by a
fictitiously decreased cross-section area A∗m :
Am
A∗m =  
E m,mean · Am
1+ Lm · 1
K ser ·km,1 + 1
K ser ·km,2
Eq. (12) for A∗m yields the exact solution for Eq. (9)
considering the flexibility of joints in the form of Eq. (11).
It should be noted that form A∗m can also be used for the
calculation of internal forces, but respecting the reduced
value for the stiffness of the fasteners. In this case,
considering Eq. (2.1) from Eurocode 5 [13], K u = 2/3 · K ser
should be considered instead of K ser :
Am
A∗m =  
E m,mean · Am
1+ Lm · 1
K u ·km,1 + 1
K u ·km,2
The total deflection defined in Eq. (9) should not exceed
the range of limiting values for beam deflections, depending
upon the level of deformation deemed to be acceptable. In
the following optimization procedure, the total deflection f
is limited with the recommended values given in Table 7.2
of Eurocode 5. The limiting value of the instantaneous
deflection for the simply supported beam is recommended
to be in a range from L/300 to L/500.
3. Optimization of timber trusses
3.1. The non-linear programming (NLP) optimization
approach
As the optimization problem of timber trusses is non-
linear, e.g. the objective function and (in)equality constraints
are non-linear, the non-linear programming optimization
(NLP) approach is used. The general NLP optimization
problem can be formulated as follows:
Min z = f (x)
subject to
h(x) = 0
g(x) ≤ 0
x ∈ X = {x | x ∈ Rn , xLO ≤ x ≤ xUP }
where x is a vector of continuous variables, defined within
the compact set X. The variables x are calculated between
their lower and upper bounds xLO and xUP . Functions
f (x), h(x) and g(x) are non-linear functions involved in
the objective function z, equality and inequality constraints,
respectively. All functions f (x), h(x) and g(x) must be
continuous and differentiable.
In the context of structural optimization, variables
include dimensions, cross-section characteristics, strains,
material characteristics, stresses, economic parameters.
Equality and inequality constraints and the bounds of the
variables represent a rigorous system of the design, loading,
stress, deflections and stability functions taken from the
149
structural analysis. The optimization of the structures may
include various objectives worthy of consideration. The
most popular criterion used today is the minimization of
(SLS). (12) mass. In this paper, an economic objective function is
proposed to minimize the structure’s self-manufacturing
costs. Hence, the trade-offs between different materials can
be appropriately accounted for.
3.2. The NLP optimization model TTO for timber trusses
The optimization model TTO (Timber Truss Optimiza-
tion) for the optimization of timber trusses was developed
according to the above NLP model formulation. GAMS
(General Algebraic Modelling System), Brooke et al. [14],
(ULS). (13) was used as the interface for mathematical modelling and
data inputs/outputs.
A general timber truss layout with its characteristic
elements (upper and lower chord, verticals, diagonals) is
presented in Fig. 2, where x i represents the local longitudinal
axis of element i , while yi and z i represent the principal
axes of the cross-section of the element i ; the axes X and
Y form the global coordinate system of the structure. Ai
and li stand for the cross-sectional area and the length of
member i , respectively. The cross-sections are considered to
be rectangular, where bi and h i represent the width and the
height of the cross-section of the truss member i .
An economic objective function is defined in the model
to minimize the structure’s self-manufacturing costs, subject
to design, stress and stability constraints, known from struc-
tural analysis. Internal forces are proposed to be deter-
mined by the finite element equations, while the dimen-
sioning is performed in accordance with Eurocode 5 [13].
The objective function is thus defined:
I M
min cost = ct · bi h i li + (c f m + c f l ) · 2 · km
i=1 m=1
J
+ (c f m + c f l ) · kj (14)
j =1
(NLP) where cost represents the self-manufacturing (material and
labour) costs of the structure; ct denotes the price of the
manufactured and embedded timber material per m3 ; the
sum of the products between widths bi , heights h i and
lengths li of i , i ∈ I , timber members represents the volume
of the truss in m3 (see Fig. 2); c f m is the material cost of one
fastener together with the adjoining steel plates, while c f l
denotes the manual labour costs for handling, assembling,
drilling and bolting, defined per fastener. Considering that
the required number of fasteners is equal for both ends of
each intermediate member, the total number of fasteners in
the m-th member is 2km , where km = km,1 = km,2 . The
last term of the objective function represents the sum of all
fasteners required in joints of the chord members. Variable
k j stands for the number of fasteners of the j -th joint, j ∈ J .
It is evaluated considering the resultant force on account of
the axial forces of all intermediate elements connected to
150 S. Šilih et al. / Engineering Structures 27 (2005) 145–154

Fig. 2. A plane timber truss.

Fig. 3. A plane timber truss with three different numbers of upper chord members.

joint j . Since the dimensions of steel plates depend directly
on the number of calculated fasteners, the costs of steel
plates are included in the values c f m and c f l .
The input data of the optimization model are the truss
geometry (coordinates of joints), the supporting and loading
conditions, the diameter of the fasteners considered, the
thickness of the metal plates, as well as the material
characteristics of all the components used (timber, fasteners,
plates).
The cross-section dimensions bi and h i of i , i ∈ I , truss
timber members and the number of fasteners km and k j are
defined as independent optimization variables.
The finite element equations for the calculation of internal
forces and deflections of the structure are defined as equality
constraints. The stiffness matrix of the structure is composed
considering fictitiously decreased cross-section areas of all
the intermediate timber elements (diagonals and verticals)
in accordance with Eqs. (12) and (13), where K u and
K ser are considered for the calculation of internal forces
and deflections, respectively. The ULS and SLS design
conditions, described in Sections 2.1 and 2.2, are defined as
inequality constraints.
4. Numerical example
4.1. Input data
The paper shows the numerical example taken from
Steck [5], where calculations for a truss with the maxi-
mal height of h = 300 cm and ten upper chord members
(n = 10) were made using the classical structural analy-
sis method. The paper goes on to demonstrate that structural
optimization of the timber truss has to be performed in order
to minimize the structure’s self-manufacturing costs. The
non-linear programming approach was thus used, where all
the design parameters, including joint flexibility, were simul-
taneously considered in the optimization process. The task
was to compare the optimal truss designs obtained between
the considered and the ignored joint flexibility. The material
properties and the loads for the optimization are assumed to
be the same as in Steck [5]. They are presented below.
4.1.1. Geometrical data
The span of the truss is L = 22.5 m, and it is assumed
to be constant during the calculation process. A study has
been done for six truss alternatives including two different
 S. Šilih et al. / Engineering Structures 27 (2005) 145–154 151

values of the maximal height, h = 300 cm—Truss A and

h = 200 cm—Truss B, and three chosen total numbers
n = 6, 10 and 14 of the upper (compressed) chord members;
see Fig. 3. Truss B was optimized twice: considering joint
flexibility (Truss B1) and ignoring it (Truss B2). The upper
chord inclination is assumed to be a constant (α = 10◦ ).
The top chord is laterally supported at a distance of 3.80 m
in all cases. All trusses are statically determinate. There Fig. 4. Minimum spacing of the fasteners and the definition of h lo
i .
is therefore no element stiffness influence on the internal
member force distribution. 4.1.5. ULS and SLS criteria
ULS: kmod = 0.9 for a short term load (snow), service
4.1.2. Timber characteristics class S2 (moisture content corresponding to a
A glued laminated timber GL32h (characteristic bending temperature of 20 ◦ C and the relative humidity of
strength f m,g,k = 32 N/mm2 , homogeneous glulam) was the surrounding air only exceeding 85% for a few
considered in the study according to the EN 1194 [15] weeks per year),
classification with the following characteristics: γm = 1.25 (glued laminated timber):
f c,0,k = 29 N/mm2 , ft,0,k = 22.5 N/mm2 , f t,0,k f t,0,k
f t,0,d = kmod · = 0.9 ·
E 0,mean = 13 700 N/mm2 , E 0,05 = 11 100 N/mm2 , γm 1.25
22.5
ρk = 430 kg/m3 , ρm 1 = 500 kg/m3 . = 0.9 · = 16.20 MPa (15a)
1.25
f c,0,k f c,0,k
4.1.3. The load f c,0,d = kmod · = 0.9 ·
A combination of permanent and variable load is γm 1.25
29
considered: = 0.9 · = 20.88 MPa. (15b)
1.25
permanent action: g = 2.0 kN/m Eqs. (1) and (2a–2c) are used to check the tensile,
(constant uniform load) and compressive and buckling resistance of the truss
variable action (snow): s = 5.0 kN/m elements, respectively.
SLS: As mentioned in Section 2.2 for instantaneous
(constant uniform load).
deflection for a beam on two supports with a
The self-weight of the truss members was automatically span L, the limiting value is recommended in
determined through the optimization process with respect Eurocode 5 [13] to be in a range from L/300 to
to the actual calculated element dimensions and added to L/500. In the present numerical example the value
the above permanent load. Both uniform loads are, in the L/300 = 75 mm is assumed.
calculation, approximated as nodal forces; see Fig. 3.
4.1.6. Optimization parameters
4.1.4. Fasteners The economic parameters considered for the objective
Dowels: M14 (diameter d = 14 mm), made of steel S 235; function Eq. (14) were costs ct = 900 EUR/m3 for
f u,k = 360 N/mm2 , the GL32h timber material, while c f l = 1.5 EUR and
K ser [N/mm] = 2 · ρm [kg/m3 ]1.5 · d [mm]/35 c f m = 1.0 EUR for one M14 S 235 dowel (including the
(Eurocode 5 [13], Table 7.1), corresponding part of the steel plates). The lower bounds
K ser = 2 · 5001.5 · 14/35 = 8944.27 N/mm (two on the cross-section dimensions bilo / h lo
i were taken as
shear planes). 12/12 cm in the cases of A and B1 (considering joint
Steel plates: thickness ts = 8 mm, steel S 235. The plates flexibility) and 6/6 cm in the case of B2 (ignoring joint
are placed as the central members of a double shear flexibility).
connection (Fig. 1(a)). Eqs. (6(a)–(c)) are thus It should be noted that the lower bounds on the cross-
used to calculate the characteristic load carrying section heights h lo
i of the elements also depend on the
capacity per shear plane per fastener Fv,Rk . The diameter of the dowels (Table 8.5, Eurocode 5 [13]).
design load carrying capacity of the fasteners is According to the arrangement of the fasteners considered
calculated using Eq. (5) by considering the values (two rows; see Fig. 4), the values of h lo
i amounted to 9d,
kmod = 0.9 for short term load and the safety factor where d is the dowel diameter.
γm = 1.3 for the connections.
4.2. Optimization

1 According to EN1194, the timber strength class C40 is used for lamels As soon as the above data were added to the general
in order to obtain timber GL 32h; ρm (C40) = 500 kg/m3 , EN 338 [16]. optimization model for timber trusses TTO, the NLP
152 S. Šilih et al. / Engineering Structures 27 (2005) 145–154
Table 1
Numerical results for h = 300 cm (Truss A)
n=6 n = 10
Total costs (EUR) 1865.35 2018.08
Timber mass (kg) 766.86 808.66
Chord dimensions b/ h (cm)
Lower chord 12/13 12/13
Upper chord 16/16 17/13
Total number of dowels 137 155
Max. deflection without joint
flexibility ( f 0 ) (mm) 31.05 32.80
Max. deflection considering
joint flexibility ( f = f 0 + f 1 ) (mm) 43.52 46.07
optimization was performed by the computer program
GAMS/CONOPT2, Drud [17]. The economic objective
function was defined to minimize the structure’s self-
manufacturing costs. While Truss A (h = 300 cm)
was optimized considering joint flexibility, Truss B (h =
200 cm) was optimized twice: considering the joint
flexibility (Truss B1) and ignoring it (Truss B2).
Since the NLP optimization handles continuous vari-
ables, the obtained variables of the final/optimal solution
take some real values between their defined lower and up-
per bounds. At this stage, the structure was fully exploited
considering either ultimate or serviceability limit state de-
sign conditions. As the aim of our research was to obtain a
structure of practical applicability, the final continuous op-
timal solution was reanalysed with the variables rounded to
their nearest upper integer values (i.e. 1 cm for cross-section
dimensions; 1 fastener).
It should be noted that during the optimization process,
two FEM analyses were performed simultaneously, one for
the calculation of the internal forces at the ultimate limit
state (using the slip modulus K u ) and one for calculation
of the deflections at the serviceability limit state (using the
slip modulus K ser ).
4.3. Results
The optimal results obtained for Truss A (h = 300 cm)
are shown in Table 1. As can be seen, the minimal self-
manufacturing costs and timber mass are obtained at the
lowest defined number of timber members n = 6. The
total number of fasteners increases with n, but not entirely
proportionally. The maximal deflection f 0 , in which joint
flexibility is ignored, is similar for all three cases of n, but
the deflection f 1 , in which joint flexibility is considered,
slightly increases with n. Consequently, the total deflection
f = f 0 + f 1 increases with the number of bracing members.
It is important to know that the deflection increment of
the joint flexibility should by no means be neglected; it
is in the range from 40% (n = 6) to 54% (n = 14)
according to f 0 . However, the height of the truss obtained
was too large (L/ h = 7.5) and the deflections were not
decisive ( f < 75 mm). Since the ULS criteria are competent
and the system is statically determinate, there is no joint
flexibility influence on the internal force distribution and,
n = 14 consequently, on stress conditions. Therefore, it is beneficial
to analyse further results for the case with h = 200 cm,
2239.66
L/ h = 11.25, from Table 2.
906.76
Table 2 shows the results and the minimal self-
12/13 manufacturing costs obtained for the truss of height h =
17/13 200 cm where joint flexibility is accounted for (Truss B1)
165 or is ignored (Truss B2). There are some considerable
differences between the results for truss B1 and those for
31.75
truss B2. In the case of truss B2, where joint flexibility is not
48.85 taken into account, the minimal timber mass and costs were
obtained in the case of ten upper chord members (n = 10).
It should be noted that the ULS criteria are still decisive.
Similarly to the Truss A case, the maximal deflection f 0 , in
which joint flexibility is ignored, is practically similar for all
three cases of n.
When joint flexibility is accounted for in the study (Truss
B1), the total truss deflection increases from 37% (n = 10)
to 48% (n = 14) and, consequently, the SLS instead of ULS
criteria become decisive. The optimal costs were obtained at
n = 6 compressed chord members, and the lowest mass was
calculated at n = 10 chord members.
In order to arrive at appropriate conclusions, it is
convenient to present the final results by comparing the
optimal costs and the masses obtained between different
truss alternatives; see Table 3. The influence of the height
of the truss on the final results is represented by the ratio
B1/A, which represents the ratio between the costs (masses)
obtained of the trusses B1 (h = 200 cm) and A (h =
300 cm) for different truss layouts (different numbers of
upper chord members n). The possible error caused by
neglect of the flexibility of the joints in the timber trusses is
exposed through the ratio B1/B2 which represents the ratio
of the costs (masses) obtained between truss B1 (considering
joint flexibility) and B2 (neglecting joint flexibility) for
different truss layouts.
Comparing Trusses B1 and B2, it is evident that the influ-
ence of fastener flexibility should not be neglected at all. The
total costs increase by up to 12.5%, and the timber masses
by up to 17%. The influence of flexibility increases with the
number of flexibly connected diagonal elements, represented
in this study by the number of the upper chord members n.
Additionally, it is convenient to compare the final results
for trusses with different heights (Trusses B1 and A).
Although Truss A is higher than Truss B1, the minimal
mass was obtained at Truss A. This occurred due to the
essentially smaller cross-section dimensions of the chord
members obtained at Truss A (Table 1) as compared to Truss
B (Table 2). Both the mass and cost ratios decrease with the
number of compressed chord members n.
Finally, in order to gain a general picture, all the
results obtained are summarized in Fig. 5. The results were
normalized with respect to the highest value obtained, to
which the value of 1.0 was assigned. The comparison of the
optimal costs obtained is presented in Fig. 5(a), while the
 S. Šilih et al. / Engineering Structures 27 (2005) 145–154 153

Table 2
Numerical results for h = 200 cm (Truss B: B1 and B2)
Truss B1 Truss B2
SLS joint flexibility criteria Without SLS joint flexibility criteria
n=6 n = 10 n = 14 n=6 n = 10 n = 14

Total costs (EUR) 2295.53 2320.12 2453.44 2172.16 2079.68 2181.57

Timber mass (kg) 922.24 900.07 926.91 885.92 783.16 792.54
Chord dimensions b/ h (cm)
Lower chord 17/15 17/16 17/15 17/15 17/14 17/14
Upper chord 18/18 21/13 21/13 18/18 21/13 21/13
Total number of dowels 174 192 210 162 186 204
Max. deflection without
joint flexibility ( f 0 ) (mm) 50.37 51.39 48.23 57.71 59.06 58.02
Max. deflection considering
joint flexibility (mm)
( f = f0 + f 1 ) 70.30a 70.71a 71.46a – – –
a It should be noted that the final displacements f are not equal to L/300 = 75 mm due to the rounding of cross-sections as described in Section 4.2. After
the first optimization phase (continuous optimization), the SLS conditions were active and f was equal to 75 mm in all cases considered.

Table 3
Comparison of timber mass and total costs

Costs ratio Timber mass ratio

B1/B2 B1/A B1/B2 B1/A

n=6 1.057 1.231 1.041 1.203

n = 10 1.116 1.150 1.150 1.113
n = 14 1.125 1.095 1.170 1.022

comparison of the masses is presented in Fig. 5(b). In both

cases, the highest value was obtained at truss B1 with 14
upper chord members (n = 14).

5. Conclusions
(a) Costs.
The paper presents the optimization of metal-plate-
connected plane timber trusses with respect to joint
flexibility. The optimization was performed by the non-
linear programming approach. Since various truss design
parameters such as the type of truss configuration, the
span/depth ratio, the number and type of diagonal and
vertical members as well as the type of joint connections
simultaneously affect each other, it was proposed that all
these parameters be simultaneously considered in a single
mathematical model.
The optimization model TTO (Timber Truss Optimiza-
tion) for cost optimization of timber trusses was thus
developed. The economic objective function was defined in
order to minimize the structure’s self-manufacturing costs,
subjected to the design, stress and deflection (in)equality (b) Timber mass.
constraints. The finite element equations were as the equality Fig. 5. Comparison of the results obtained.
constraints defined for the calculation of the internal forces
and the deflections of the structure. The undesirable slips
in the connections of timber trusses additionally resulted members were determined in accordance with Eurocode 5
in the reduction of truss stiffness. A stiffness matrix of the in order to satisfy the requirements of both the ultimate and
structure was therefore composed by considering fictitiously the serviceability limit states. The cross-section dimensions
decreased cross-section areas of all the flexibly connected and the number of fasteners are defined as independent opti-
elements. The constraints for dimensioning the timber mization variables.
154 S. Šilih et al. / Engineering Structures 27 (2005) 145–154
Not only are the results obtained optimal, but also the
optimization enables the conditions of either the ultimate
or the serviceability limit states to be fully exploited and
there is no reserve in the resistance of the structures. At this
point the comparison between results for trusses considering
joint flexibility and trusses ignoring it was performed for
different span/depth ratios and different numbers of internal
members.
The numerical example presented shows the applicability
of the optimization approach presented as well as the
influence of considering the flexibility of the fasteners on the
optimal self-manufacturing costs. On the basis of numerical
results, it is recommended that higher timber trusses be
designed with a lower span/depth ratio, with a smaller
number of diagonal and vertical elements and, consequently,
by using chord elements with smaller cross-sections. The
example demonstrates that the self-manufacturing costs can
increase by over 10% and the timber mass by over 15% if the
flexibility of joints is additionally considered in the design
process. However, there is less influence on the final results
if the ultimate limit state criteria are decisive.
References
[1] Hoyle RJ, Woeste FE. Wood technology in the design of structures.
Ames, Iowa: Iowa State University Press; 1989.
[2] El-Sheikh A. Optimum design of space trusses. Journal of the
International Association for Shell and Spatial Structures 1998;39:
159–68.
[3] El-Sheikh A. Approximate analysis of space trusses. International
Journal of Space Trusses 1996;11:321–30.
[4] Stalnaker J, Harris E. Structural design in wood. NY: Van Nostrand
Reinhold; 1989.
[5] Steck G. Fachwerbinde aus Brettschictholz un Vollholz.
In: Holzbauwerke: Bauteile Step 2, Konstruktionen, Details nach
Eurocode 5. Düsseldorf: Fachverlag Holz; 1995.
[6] Brüninghoff H et al. Eine Ausführliche Erläuterung zu DIN 1052,
Beuth–Kommentare (Berlin): Beuth Bauverlag; 1988 [Teil 1–3].
[7] Schmidt LA. Structural design by systematic synthesis. In: Proceed-
ings of 2nd conference on electronic computations. NY: ASCE; 1960.
p. 105–22.
[8] Lipson SL, Agrawai KM. Weight optimization of plane trusses. ASCE
Journal of the Structural Division 1974;100:865–79.
[9] Prager W. Optimization of structural design. Journal of Optimization
Theory and Applications 1970;6:1–21.
[10] Šilih S, Kravanja S, Bedenik BS. Shape optimization of plane
trusses. In: Hendriks MAN, Rots JG, editors. Finite elements in civil
engineering applications, Proceedings of the Third DIANA World
conference. 2002. p. 369–73.
[11] Kaveh A, Kalatjari V. Topology optimization of trusses using genetic
algorithm, force method and graph theory. International Journal for
Numerical Methods in Engineering 2003;58:771–91.
[12] Kravanja S, Šilih S. Optimization based comparison between
composite I beams and composite trusses. Journal of Constructional
Steel Research 2003;59:609–25.
[13] CEN/TC 250/SC5 N173. Eurocode 5: design of timber structures. Part
1-1, General rules and rules for buildings, Final draft prEN 1995-1-1.
Brussels: European Committee for Standardization; 2002.
[14] Brooke A, Kendrick D, Meeraus A. GAMS—A user’s guide.
Redwood City (CA): Scientific Press; 1988.
[15] EN 1194. Timber structures—glued laminated timber—strength
classes and determination of characteristic values. Brussels: European
Committee for Standardization; 1999.
[16] EN 338. Structural timber—strength classes. Brussels: European
Committee for Standardization; 2003.
[17] Drud AS. CONOPT—A large-scale GRG code. ORSA Journal on
Computing 1994;6:207–16.