See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/289240561

Seismic design of multi-storey cross laminated timber buildings according to

Eurocode 8

Article  in  Ingegneria Sismica · December 2013

CITATIONS READS

9 2,267

5 authors, including:

Maurizio Follesa Ioannis P. Christovasilis

Dedalegno, Firenze, Italy Aether Engineering
26 PUBLICATIONS   179 CITATIONS    25 PUBLICATIONS   301 CITATIONS   

SEE PROFILE SEE PROFILE

Massimo Fragiacomo Ario Ceccotti

Università degli Studi dell'Aquila Università Iuav di Venezia
218 PUBLICATIONS   4,067 CITATIONS    64 PUBLICATIONS   1,611 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

EISMIC BEHAVIOUR OF HISTORIC TIMBER-FRAME BUILDINGS IN THE ITALIAN DOLOMITES View project

All content following this page was uploaded by Ioannis P. Christovasilis on 28 April 2016.

The user has requested enhancement of the downloaded file.

 SEISMIC DESIGN OF MULTI-STOREY CROSS LAMINATED
TIMBER BUILDINGS ACCORDING TO EUROCODE 8
Maurizio Follesa, Department of Structural Engineering, University of Cagliari and
dedaLEGNO, via Masaccio 252, Firenze, Italy, [email protected]
Ioannis P. Christovasilis, dedaLEGNO, via Masaccio 252, Firenze, Italy,
[email protected]
Davide Vassallo, dedaLEGNO, via Masaccio 252, Firenze, Italy,
[email protected]
Massimo Fragiacomo, Department of Architecture, Design and Urban Planning,
University of Sassari, Palazzo del Pou Salit, Piazza Duomo 6, 07041 Alghero – Italy,
[email protected]
Ario Ceccotti, Director of the Trees and Timber Institute of the National Research
Council of Italy (CNR-IVALSA), Via Madonna del Piano 10, 50019 Sesto Fiorentino (FI) –
Italy, [email protected]

SUMMARY – Cross Laminated Timber (CLT) structures are nowadays increasingly used
worldwide and mostly in Europe where the system originated. However, in spite of this
diffusion which led to the construction of a great number of multi-storey buildings all over
Europe, still Eurocodes are almost completely missing provisions for CLT designers,
especially regarding the seismic design. Nevertheless, Eurocode 8 requires in most cases,
due to the regularity criteria being not fulfilled for most of the buildings, the use of the modal
response spectrum analysis method, i.e. the linear dynamic analysis. This method requires
the correct estimation of the lateral stiffness of the building in order to accurately calculate
the design seismic forces in the building, which may be significantly underestimated or
overestimated depending on the size of the building and the shape of the design spectrum.
This can be done by modelling each connection with different methods that are often based
on available test results, which are not easily accessible by a practicing engineer. This paper
provides a design approach for dynamic linear modelling of CLT structures using SAP 2000.
Equations are proposed based on available design codes and literature references, and used
to design a 3-storey case study building. Further provisions for the seismic design of CLT
buildings which are not included in Eurocode 8 are also given. Finally, the proposed design
model is also compared with the results of the shaking table tests conducted in 2006 in Japan
by CNR-IVALSA on a three-storey CLT building.

Key words: CLT structures, linear dynamic analysis, mechanical fasteners, shaking table test,
friction.

1. Introduction

Cross Laminated Timber (CLT) structural systems, in spite of their relatively new
diffusion, have been increasingly used over the last years for the construction of multi-storey
timber buildings in many parts of the world (North America, Japan and Australasia) and
mostly in Europe where the system originated about 20 years ago.
The reason for this success lies on the advantages over more traditional structural
systems, such as the increased speed of erection, thanks to the quick and dry construction
process, and the reduced environmental impact, in terms of CO2 emissions and energy
consumption. Moreover, other additional benefits such as the increased fire resistance, the
improved dimensional stability due to the cross-lamination process in the fabrication, and the
possibility to use low grade timber in the production, make the CLT system an excellent

1
alternative also to other traditional timber systems such as light-frame and glulam
construction for the erection of low to medium rise buildings.
However, the actual lack of design rules in the Eurocodes referred to the CLT system
requires some additional effort from the structural designer in order to find and explain the
design choices and the calculation methods adopted to fully meet the safety requirements
foreseen by the current building code. Especially the numerical modelling procedure of a
CLT building may result very difficult to practicing engineers, who cannot easily access test
results or specific literature references on the subject.

1.1. Structural system and seismic behaviour

Cross Laminated Timber (CLT) systems are structures in which walls and floors are
composed of cross laminated timber panels, i.e. panels made of at least three laminations
arranged alternatively at right angles and glued together. The laminations shall be made of
one or two layers of parallel timber boards.
Unlike vertical actions, which affect only a portion of the structure and some of the
structural elements, the seismic action is modelled via horizontal forces applied at the floor
level that are transferred to the ground involving all the structural members, as shown
schematically in Figure 1. Therefore, the continuity of the connections between the different
parts of the structure is particularly important and must be effective both in tension and
compression.

Figure 1. Seismic forces acting on a CLT building

With regard to the seismic response, a CLT building may be considered as a box-like
structure in which the walls and floors are made of solid wood panels with high in-plane
stiffness and strength, connected to each other by mechanical connections. The connections
of the ground floor walls to the foundation must have a dual function as shown in Figure 2, as
it must resist both overturning and sliding of the walls caused by the horizontal actions (wind
or earthquake) acting in the plane of the wall and in general on the entire building.
The connection of the walls to the foundation or to the supporting floor panels is
usually made with mechanical fasteners (hold-down anchors, steel brackets, anchoring bolts,
nails and screws) and should adequately restrain the wall against uplift and sliding. Uplift-
restrain connections are typically placed at wall ends and at opening ends, while sliding-
restrain connections are uniformly distributed along the wall length. Figure 3 illustrates
typical connections against uplift and sliding installed in a CLT building.

2
 The erection process follows a platform type of construction with walls of height
equal to the inter-storey height. Each wall assembly can be either made of a single CLT panel,
provided that the length does not exceed the maximum transportable length of typically 16 m,
or may be composed of a number of panels with a typical length of 2.5 m that renders the
transportation more convenient and economical. In this case, the connections between wall
panels are usually done with the interposition of a wood-based multi-layer panel that can be
inserted in grooves cut inside the wall or on one of its faces. Sometimes the same connection
is made with a half-lap joint. The connection is prescribed either with self-tapping screws
with a diameter varying from 6 to 10 mm or with nails of 3 mm diameter and spacing
depending on the seismic loads. Figure 4 illustrates some of the typical types of step joints
used in construction. Walls in perpendicular directions shall be connected with mechanical
fasteners (usually screws) to achieve a box-type behaviour.

Figure 2. Effects of the seismic forces acting on a wall and different function of the
connection elements

(a) (b)

Figure 3. (a) Hold-down placed at wall and at opening ends to prevent the uplifting of the
wall, and (b) angle bracket connection between the wall and the floor panel to prevent the
sliding of the wall

(a) (b) (c)

Figure 4. Three different step joints between vertical wall panels; (a) a cross-layer stripe
inserted into internal grooves, (b) a cross-layer stripe inserted in grooves on the internal side
of the wall, and (c) a half-lap joint

The experience gained from the research carried out so far has shown that buildings
with walls composed of panels with a maximum length of 2.5 m, if designed in full
compliance with the capacity design method, show a greater level of ductility than buildings
made of continuous walls and, therefore, a greater capacity to dissipate the energy induced by
the earthquake [Ceccotti et al., 2007].

3
 Horizontal diaphragms are made of CLT timber panels connected through horizontal
joints made with mechanical fasteners (screws or nails). The floor panels bear on the wall
panels and are connected to them usually with self tapping screws. The upper walls bear on
the floor panels, and are connected to the lower walls using metal connectors similar to those
used for the wall-foundation connection. Figure 5 illustrates the typical connections between
walls of a lower and an upper storey to the floor diaphragm.

Vertical joint

Horizontal joint
Wall panel
Floor panel Single piece wall

Horizontal joint
Tie-down

Hold-down
Steel brackets
Hold-down

Hold-down
Screws

Hold-down Base beam

Steel brackets

Figure 5. Walls and floors in platform type CLT buildings (after Follesa et al., 2011)

1.2. Seismic design and capacity-based design criteria of CLT buildings

The construction experience of CLT buildings and the research conducted over the
last years on the static and seismic behaviour has led to the derivation of some design rules
with regard to both the static and the seismic performance [Follesa et al., 2011].
By adopting the assumption of rigid horizontal diaphragms, the in-plane seismic shear
is divided among the walls underneath according to their stiffness taking into account an
accidental eccentricity in addition to any eccentricity between the centre of mass and the
centre of stiffness in the case of asymmetric distribution of stiffness.
To achieve the behaviour of a box-type structure it is important to ensure that local
failures which may compromise the box-type behaviour will not occur. The connections
devoted to the dissipative behaviour in a CLT building are:
 step joints between wall panels in case of walls composed of more than one
element;
 connections against sliding between walls and floor below, and between walls
and foundation; and
 anchoring connections against uplift placed at wall ends and at wall openings.
To ensure the development of cyclic yielding in the dissipative zones, all other
structural members and connections shall be designed with sufficient overstrength so as to
avoid anticipated brittle failure. This overstrength requirement applies especially to
connections shown in Figure 6 such as:
 connections between adjacent floor panels in order to limit at the greater
possible extent the relative slip and to assure a rigid in-plane behaviour;
 connections between floors and walls underneath thus ensuring that at each
storey there is a rigid floor to which the walls are well-connected; and

4
  connections between perpendicular walls, particularly at the building corners,
so that the stability of the walls themselves and of the structural box is always
assured.
Analyses carried out on the evaluation of the overstrength factor for the connections
mentioned above have led to the conclusion that a conservative value of 1.6 shall be used
[Jorissen and Fragiacomo 2011; Fragiacomo et al., 2011; Follesa et al., 2011].
The seismic resistance of shear walls should be higher at lower storeys and should
decrease at higher storeys proportionally to the decrease of the storey seismic shear, thus
leading to the simultaneous plasticization of the ductile connections in order to maximize the
energy dissipation of the building. To achieve the desired ductile behaviour, each single
connection should be detailed in order to avoid brittle failures. Special care should be used
when designing the dissipative connections to ensure the attainment of a ductile failure
mechanism characterized by the formation of one or two plastic hinges in the mechanical
fastener. Brittle failures such as splitting, shear plug, tear out and tensile fracture of wood in
the connection regions should be always avoided. Similarly, a brittle failure mechanism in the
weaker section of the steel plate should be prevented in connections with steel brackets or
hold-downs anchors connected to the wall panels with nails or screws.

1.3. Seismic force modification factor

Recent research work on the seismic behaviour of CLT buildings with pseudo-static
tests on individual structural elements (CLT wall and floor panels), pseudo-dynamic and
shake-table tests on full-scale CLT buildings, and numerical simulations [Ceccotti et al.,
2007; Ceccotti and Follesa, 2006] have shown that CLT buildings designed in accordance
with the aforementioned capacity design method and built with walls composed of panels
with a maximum length of 2.5 m can be designed with a value of the seismic behaviour factor
q of at least 3. In the current version of Section 8 of Eurocode 8 [EC8, 2004] there is no
provision for CLT buildings. However, for “Glued wall panels with glued diaphragms,
connected with nails and bolts”, which is the type of structure more similar as description to
the CLT system, the upper limit of the behaviour factor q is 2. Therefore, for compliance with
the current building codes, this is the value which is currently used when designing CLT
buildings.

Connection between
adjacent floor panels

Connection between
perpendicular walls

Connection between floors

and walls underneath

Figure 6. Connections to be designed with overstrength in order to fulfill the capacity

design criteria in Cross Laminated buildings (highlighted in red)

5
2. Numerical Modelling of CLT Buildings

2.1. General considerations

Before conducting the seismic design of a multi-storey CLT building, the designer
should follow the principles of conceptual design expressed in Section 4.2.1 of Eurocode 8,
which are given for all types of buildings. These principles are particularly important for CLT
structures, in order to achieve a good overall structural behaviour of the building, i.e.
structural simplicity, uniformity, symmetry and redundancy, bi-directional strength and
stiffness, diaphragmatic behaviour at storey level, and adequate foundation.
According to Eurocode 8 Part 1, depending on the fulfillment of the regularity criteria
in plan and elevation, there are three possible choices for the seismic analysis of a numerical
building model:
a) the lateral force method of analysis, i.e. the linear static analysis;
b) the modal response spectrum analysis, i.e. the linear dynamic analysis; and
c) non-linear methods, which could be either non-linear static (pushover)
analysis or non-linear time-history analysis.
Method (a) is usually followed for low-rise and simple buildings which meet the
regularity criteria in elevation. The horizontal forces may be calculated with an inverted
triangular distribution based on the mass and the height of each storey. The natural vibration
period may be calculated directly from the equation provided by the Eurocode 8 formula:

T1  CT  h 3 4 (1)

where CT may be assumed equal to 0.05, h is the height of the building in meters, and
T1 is measured in seconds. If the building also meets the regularity criteria in plan the
analysis may be performed using two planar models, one for each main horizontal direction.
Method (b) is used in cases where the building does not meet the criteria for regularity
in elevation, which is the most common situation. The correct estimation of the natural
vibration period of the building is in this case crucial, as it may lead to non-conservative
design if underestimated or overestimated, depending on the type and size of structure and on
the shape of the design spectrum [Sustersic and Dujic, 2012]. Moreover, it is important to
estimate accurately the displacements of the building for the verification at the SLS and ULS,
especially when the building is adjacent to other structures dynamically independent divided
by a seismic separation. Therefore, it is crucial to model correctly the stiffness of the shear
walls.
Both methods (a) and (b) are based on a linear elastic analysis, and the energy
dissipation capacity of the building is implicitly taken into account by dividing the forces
obtained from the linear elastic analysis by the behaviour factor q, which is associated with
the relevant type of structure and ductility class, as defined in Table 8.1 of Eurocode 8. It
should be mentioned however that no specific design rules and capacity design criteria are
given for CLT buildings within the timber part (Section 8) of Eurocode 8, and that the value
of the q factor given in Table 8.1 and currently adopted for CLT buildings is considered very
conservative [Follesa et al., 2011].
Method (c) is more complex than method (a) and (b) and requires the knowledge of
the non-linear monotonic (in case of static analysis) or cyclic (in case of time-history
analysis) behaviour of the structural elements which are detailed and designed for energy
dissipation, which in case of CLT buildings and of timber structures in general are
mechanical fasteners, e.g. self-tapping screws, annular ringed nails, angle brackets and hold-
down connectors. This is a more complex approach, which relies on the availability of
experimental data on the same type of connectors and fasteners used in the design. In
addition, a suitable finite element software should be used, which includes a proper non-
linear monotonic or hysteretic model capable of accurately simulating the actual non-linear
response of the mechanical fasteners.
6
 However, designers usually do not have access to experimental data that can be used
to calibrate the non-linear behaviour of the mechanical fasteners. Even when experimental
data is available, it may refer to connections with different types, number and diameter of
fasteners from those used in the actual design. Therefore, in most cases, the only possible
way to perform a reliable design of a multi-storey building is to (i) formulate a proper finite
element model of the CLT structure based on the available equations for the calculation of
the connection slip modulus given in Eurocode 5 and (ii) analyze it through a linear dynamic
analysis.

2.2. Description of the numerical model

The 3-dimensional numerical model of a CLT building proposed in this paper has
been implemented in the widespread software package for structural analysis SAP2000 [CSI,
2007], which is utilized to perform the modal response-spectrum or the static analysis and to
solve the associated equilibrium equations. A pre- and post-processing software specifically
developed by Tecnisoft (Modest, Ver. 8.1, 2013) was used to aid in the implementation.
Three types of elements are utilized, namely:
 4-noded, 24 DOFs, shell elements with membrane and bending capabilities for
the CLT wall panels with a typical mesh of 0.5x0.5 meters;
 2-noded, 6 DOFs, truss elements for the various mechanical connections
between wall panels; and
 2-noded, 12 DOFs, beam elements for the lintels connecting walls above
openings.
Figure 7 illustrates a typical schematization of a pair of CLT wall panels and the
connections at the base of the building as well as the connections with the upper floor walls.

Figure 7. Typical wall schematization

With this representation, the in-plane shear forces transmitted from the walls to the
walls underneath can be directly obtained from the axial forces of the horizontal truss
elements, while the uplifting vertical forces are obtained from the tensile forces in the vertical
truss elements. A rigid diaphragm constraint is used to constrain all nodes at the same level in
Figure 7. It should be noticed that a separate constraint is used to constraint the nodes at the
bottom of the wall and at the top of the wall underneath.
It is worth noting that the greater component of the inter-storey drift in a CLT
building is attributed to the deformation of the mechanical joints connecting the walls with
the floor diaphragms, which can be of the order of centimeters, while the shear and bending
deformation of the CLT wall panel itself remains within the order of millimeters.
The model described above is based on some simplified assumptions:

7
  floor diaphragms are assumed to be in-plane rigid while the out-of-plane
stiffness is not considered;
 the connection between perpendicular walls is assumed to be rigid;
 the connection between floors and supporting walls is assumed to be rigid; and
 hold-down connectors are not explicitly modelled.

2.2.1. Section and material properties for shell elements

Shell elements are defined with the same length, height and thickness as the
associated CLT wall panels. Orthotropic material properties are defined based on: (i) the
orthotropic properties of the wood class which the boards of the CLT walls are made of; and
(ii) the number of layers and their associated thickness and grain direction.
For the orthotropic material properties, nine values have to be defined, namely the 3
moduli of elasticity in the 3 orthogonal directions Ex, Ey, Ez, the 3 Poisson’s ratios νxy, νyz, νzx,
and the 3 shear moduli Gxy, Gyz, Gzx. The x axis is assumed along the length of the wall (in
the horizontal direction), the y axis in the direction perpendicular to the wall plane, and the z
axis along the height of the wall (in the vertical direction), as shown in Figure 8. Out of these
properties the most important ones are: (i) Ez that affects the vertical axial stiffness and the
bending stiffness; and (ii) Gzx that affects the in-plane shear stiffness of the wall.

Figure 8. Principal axes of a vertical wall element

Ez = E0,eq is the equivalent modulus of elasticity of CLT wall panels in the direction
parallel to the grain of the outer layers, calculated for a 5-layer CLT panel as suggested by
Blaß and Fellmoser [2004]:

  E  a3  a1 
E0,eq  1   1  90,T 
   E0 , L (2)
  E0 , L  a5 

where E0,L is the modulus of elasticity parallel to the grain of the longitudinal layers, E90,T is
the modulus of elasticity perpendicular to the grain of the transversal layers and a1, a3, a5 are
shown in Figure 9.
Ex = E90,eq is the equivalent modulus of elasticity of CLT wall panels in the direction
perpendicular to the grain of the outer layers, calculated for a 5-layer CLT panel as suggested
by Blaß and Fellmoser [2004]:

E  E  a3  a1 
E90,eq   90, L   1  90,L 
   E0 , L (3)
 E0,T  E0,T  a5 

8
 Figure 9. Definition of various thicknesses in a generic 5-layer CLT panel

where E0,T is the modulus of elasticity parallel to the grain of the transversal layers, E90,L is
the modulus of elasticity perpendicular to the grain of the longitudinal layers.
Furthermore, the equivalent modulus of elasticity to be used as Ey in the material
properties of the numerical model (see Figure 8) is calculated as:

1
 
E y  t tot    t i / E90, L ,i   t j / E90,T , j  (4)
 i 1, 3,....,n j 2 , 4 ,....,n 1 

where ttot is the total thickness of the CLT panel, ti and tj are the layer thicknesses in the
longitudinal and transversal directions, respectively, and n is the total number of layers.
The shear modulus G of the CLT panels is calculated as:

1
 
G  t tot    t i / GL ,i   t j / GT , j  (5)
 i 1,3,....,n j 2 , 4 ,...., n 1 

where GL and GT are the shear moduli of the longitudinal and transversal layers, respectively.

2.2.2 Section and material properties for horizontal and vertical truss elements

A pair of horizontal cross truss elements is used to connect each wall to the
foundation as well as to the wall of the upper floor where applicable. The section and
material properties of the trusses are computed based on the horizontal stiffness of the
connections used to transfer the shear force from the wall to the floor diaphragm below.
Typically, these connections consist of angle brackets or a pair of inclined screws at a certain
spacing. Connections between walls and upper diaphragms are considered rigid since they are
typically over-sized, namely designed for the overstrength of the dissipative connectors, to
satisfy capacity design requirements, but also because floor diaphragms are not explicitly
modelled but just schematized using a kinematic constraint of rigid floor (option
‘CONSTRAINT’ in SAP 2000).
The horizontal stiffness KH of the connection is computed based on the slip modulus
at serviceability limit state of each fastener Kcon used to connect the vertical metal plate of
each connector to the wall panel and assuming the metal plate to concrete connection as rigid
for the wall-foundation connection.
K con  2   m1.5  d / 23 (6)

In Equation (6), m and d signify the mean density of timber in [kg/m3] and the
fastener diameter in [mm], respectively, with the values of Kcon in [N/mm]. This equation is
given in Table 7.1 of Eurocode 5 [EC5, 2009] for the calculation of the slip modulus under
service load for dowels, bolts, screws and nails with pre-drilling used in steel to timber
connections. Since the nails are usually banged, a different formula for nail without pre-
drilling should be used. However, it was observed from the results of tests conducted on CLT
structures [Ceccotti et al., 2006] that Equation (6) provides a better match with the

9
experimental values. A possible reason for that is the effect of the cross layers, which
increases the stiffness of nails with respect to the calculated value for solid timber according
to the formula proposed for nails without pre-drilling. Thus, for n nails, the horizontal
stiffness of the wall-foundation connection is given by:

K H  n  K con (7)

For the wall to floor connection at upper storeys, the horizontal stiffness KH is
computed considering two horizontal springs in series: (i) the wall-metal plate connection,
with stiffness KH1; and (ii) the metal plate-floor connection, with stiffness KH2.

1
KH  (8)
1 1

K H1 K H 2

Both stiffness values are computed by multiplying the slip modulus of a single fastener by the
number of fasteners, as previously discussed. The axial stiffness EfAH of each horizontal truss
element is then calculated using Eq. (9):

K H  LW
E f AH  (9)
2

where Lw is the length of the wall section, which can be made of a single panel or more
adjacent panels connected with vertical step joints.
Similarly to the horizontal connection, the vertical step joints between wall panels are
simulated with a pair of vertical cross truss elements which are used to connect each wall
panel to the next one (see Figure 7). The section and material properties of the trusses are
computed based on the stiffness of the connections used to transfer the shear force from
adjacent wall panels. These connections are typically made of a cross-layered wood-based
panel (usually cross banded LVL panels or plywood) inserted in an internal groove or in a
groove cut on the internal side of the wall (case (a) and (b) of Figure 4, respectively) and the
connection is usually made of self-tapping screws with a diameter varying from 6 to 10 mm
or, sometimes, with 3 mm diameter nails and spacing depending on the seismic load.
The vertical stiffness KV of the connection is computed again based on the slip
modulus at serviceability limit state Kcon of each fastener of the panel to timber connection
given in Table 7.1 of Eurocode 5 [EC5, 2009] and in Equation (6) multiplied by the number
of fasteners. This value should then, be divided by 2, considering the double panel to timber
connection, and either multiplied by 2 for double shear connection like in the case of internal
groove step joint, or by 1 for groove on the internal side of the panel. Similarly to the
horizontal connections, the axial stiffness EfAV of each vertical truss element is then
calculated using Eq. (10):

KV  H W
E f AV  (10)
2

where Hw is the height of the wall and KV the stiffness of the step joint connection.
Vertical truss elements are also used to simulate the deformability of the floor
diaphragms along their thickness, namely perpendicular to their plane, as the walls bear on
the floor panels. Thus, the modulus of elasticity perpendicular to grain (E90 from Equation
(4)) is selected for the isotropic material properties. Since the shell elements are meshed with
a grid of 0.5 meters length, the cross sectional area of the vertical trusses is equal to 0.5
meters times the thickness of the wall above. At the foundation level, the modulus of
elasticity of concrete is used for the isotropic material properties. Forces in the vertical truss
10
elements of a wall are then utilized to calculate the tensile forces for the design of the hold-
downs, typically installed at each end of the wall to resist overturning moments from
horizontal seismic loads. It should be noted that although hold-down anchors play a major
role in the actual lateral stiffness and strength of the wall, they are not explicitly simulated in
the linear numerical model, mainly due to the nonlinear nature of their response that exhibits
markedly different stiffness in compression (where there is contact between wall and floor
panels) and in tension (where only the hold-downs resist).

3. Design of a three-storey case study building

The case study building considered in the design example is a 3-storey residential
CLT structure which was built and tested on the unidirectional 14,5m x 15m NIED Shaking
Table facility in Tsukuba, Japan in July 2006 by a joint group of Italian (CNR-IVALSA) and
Japanese (NIED) researchers within the SOFIE Project, which was a research project on the
seismic and fire behaviour of multi-storey CLT buildings funded by the Province of Trento,
Italy, and coordinated by CNR-IVALSA (Italian National Research Council – Trees and
Timber Institute) [Ceccotti and Follesa, 2006].

3.1. Geometric and structural configuration

The building has a square plan with dimensions of 7.0x7.0 m and a total height of
10.0 m with a double-pitched roof. The floor plan is symmetric in both directions and at each
floor there is a central wall parallel to the E-W direction with a central door opening of
2.25x2.40 m. The two outer walls at the first floor parallel to the E-W direction have two
openings; one of 2.25x2.20 m and one of 4.00x2.20 m. The outer walls parallel to the N-S
direction incorporate two window openings of 1.20x1.20 m. The net storey height is 2.95 m
for the first two floors and varies from 2.75 m at eaves to 3.80 m at the ridge for the third
floor. Plan and elevation drawings are displayed in Figure 10 while Figure 11 shows the
construction sequence of the building.
The walls are made of 5-layer (17-17-17-17-17)1 CLT panels 85 mm thick with the
primary orientation (orientation of the face grain) along the vertical direction. Each wall
assembly consists of three panels of about 2.3 m width. The floor panels, which are 1.35 m or
2.30 m wide, are oriented along the E-W direction and span the full length of the structure
with 6-layer (27-17-27-27-17-27) 142 mm thick CLT panels for the first two storeys and 5-
layer (17-17-17-17-17) 85 mm thick CLT panels for the roof. The CLT panels are considered
to be constructed of EN C24 timber class. The strength, stiffness and density properties are
listed in Table 1.

Table 1. Strength and density properties for EN C24 timber class

Characteristic strength in bending fm,k (MPa) 24.0
Characteristic strength in tension parallel to grain ft,0,k (MPa) 14.0
Characteristic strength in tension perpendicular to grain ft,90,k (MPa) 0.4
Characteristic strength in compression parallel to grain fc,0,k (MPa) 21.0
Characteristic strength in compression perpendicular to grain fc,90,k (MPa) 2.5
Characteristic strength in shear fv,0,k (MPa) 4.0
Mean modulus of elasticity parallel to grain E0,mean (MPa) 11000
Mean modulus of elasticity perpendicular to grain E90,mean (MPa) 370
Mean shear modulus Gmean (MPa) 690
5th percentile of modulus of elasticity parallel to grain E0,05 (MPa) 7400
Mean density ρmean (kg/m3) 420
3
Characteristic density ρk (kg/m ) 350

1
Values indicate the individual layer thicknesses in mm and bold fonts designate the layers parallel to
the primary orientation (face grain) of the CLT panel
11
 FIRST FLOOR PLAN 2nd - 3rd FLOOR PLAN ROOF PLAN

0,90
1,468

2,340

2,340

1,10
4,000

6,935

6,935
2,255

2,255

2,255

7,960
2,94
1,00
1,30

1,10
2,340

2,340
1,468

0,90
0,90 1,10 2,94 1,10 0,90 0,90 1,10 2,94 1,10 0,90
6,935 6,935 7,723

W
N
S
E
NORTH WALL ELEVATION SOUTH WALL ELEVATION EAST AND WEST WALL ELEVATION

7,960 7,960 7,723

ROOF THICKNESS 85 mm ROOF THICKNESS 85 mm

1,03
3,777

3,777

2,76
0,142

0,142

0,142
1,10
1,10 1,10

9,96
9,96

9,96

1,20
1,20

1,20
2,95

2,95

2,95
0,142

0,142

0,142
1,00
1,00

1,00

2,255 4,000 1,10

1,20

2,95
2,95

2,95

2,340 2,340 1,468 1,468

1,00
6,935 6,935 6,935

Figure 10. Plans and elevations of the case study building (dimensions in m)

Figure 11. Erection sequence of the case study building and direction of floor panels

3.2. Gravity loads and seismic weight

Gravity loads for the seismic combination were estimated based on the structural and
non-structural elements. The dead loads G of external and internal walls are 0.94 kPa and
0.79 kPa, respectively, while those of the floors and roof are 3.43 kPa and 1.28 kPa,
12
respectively, where the roof loads refer to the inclined area. The live loads Q for the floors
are 2.00 kPa for residential use and for the roof diaphragm no accidental load is considered
for the seismic combination. Based on these gravity loads, Table 2 lists the total dead and live
loads as well as the seismic weight of each floor of the building. The total seismic weight is
W = 692.45 kN.

Table 2. Total dead load, live loads and seismic weight for each level of the building
Level Dead load, G (kN) Live load, Q (kN) Seismic weight, G + 0.3Q (kN)
1 240.10 93.85 268.25
2 261.11 93.85 289.26
3 134.94 0.00 134.94
Sum 636.15 187.7 692.45

3.3 Design spectrum

The design response spectrum, illustrated in Figure 12, is defined from Section 3 of
Eurocode 8, based on a Type 1 elastic response spectrum with the following parameters:
 design ground acceleration g = 0.35 g;
 soil factor S = 1.2 for ground type B;
 lower limit of the period of constant spectral acceleration branch TB = 0.15 s;
 upper limit of the period of constant spectral acceleration branch TC = 0.50 s;
 value defining the beginning of the constant displacement response range of
the spectrum TD = 2.00 s;
 damping correction factor for 5% viscous damping η = 1; and
 behaviour factor q = 3.

Figure 12. Design response spectrum considered in the design example

3.4 Numerical model of the test building

Based on the description provided in the previous section, a 3-dimensional numerical

model has been developed for the 3-storey case study building. Floor and roof diaphragms
are considered rigid in their plane and schematized by means of master-slave constraints
applied to the reference nodes. Figure 13 shows the undeformed shape of the FE model and
Figure 14 shows the Wall IDs of the building. The properties of the shell elements
representing wall panels are computed according to the material properties shown in Table 1
and Equations (2-5) as:
13
 E x  4622 MPa
E y  370 MPa
(11)
E z  6748 MPa
G xz  G xy  G yz  690 MPa

Figure 13. Undeformed shape of the building FE model

Figure 14. Wall IDs of structure for first (left), second (middle) and third (right) floor

3.5. Preliminary analysis and design of the building

Before the final design of the building, a preliminary design is conducted using the
lateral force method of analysis. The natural vibration period of the structure is calculated
from the code equation shown in Equation (1), which in this case gives:

T1  CT  h 3 4  0.05  10 3 4  0.28 s (12)

The purpose of the preliminary design is to provide all the necessary information
needed to define the appropriate material and section properties of the elements used in the
final finite element model. The lateral stiffness of the connections of the wall panels to the

14
floor diaphragm below is assumed to be the same per linear meter for all wall assemblies of
the structure. A value of 20 MN/m/m is used, based on engineering judgement, to calculate
the required cross section of the pair of horizontal springs of each wall assembly according to
Equation (9). Similarly, vertical step joints between panels of the same wall assembly are
considered to provide the same generic stiffness per linear meter of 5 MN/m/m and the
required cross section of each pair of vertical springs is computed according to Equation (10).
The results of the preliminary analysis in terms of shear per unit length for each wall
and shear per unit height for each vertical step joint are presented in Table 3. The storey shear
forces are 206.2 kN, 165.0 kN and 78.8 kN for the first, second and third storey, respectively,
with a correction factor  = 0.85 applied according to Section 4.3.3.2.2 of EC8.

Table 3. Preliminary analysis and design results

Vert. Vert.
Shear Vert. joint
AB joint joint
Length Height per unit shear per No of
Wall ID stiffness screw screw
(m) (m) length unit height ABs
(kN/m) spacing stiffness
(kN/m) (kN/m)
(cm) (kN/m)
F1X-1-1 6.94 2.95 16.82 9.44 11 362260.8 10.0 16242.5
F1X-2-1 6.94 2.95 16.82 9.44 11 362260.8 10.0 16242.5
F1Y-1-1 1.47 2.95 20.50 N/A 3 98798.4 N/A N/A
F1Y-1-2 1.47 2.95 20.53 N/A 3 98798.4 N/A N/A
F1Y-2-1 2.34 2.95 16.92 N/A 4 131731.2 N/A N/A
F1Y-2-2 2.34 2.95 16.92 N/A 4 131731.2 N/A N/A
F1Y-3-1 2.29 2.95 17.55 N/A 4 131731.2 N/A N/A
F1Y-3-2 2.29 2.95 17.55 N/A 4 131731.2 N/A N/A
F2X-1-1 6.94 2.95 12.85 10.94 16 191609.6 7.5 53615.7
F2X-2-1 6.94 2.95 12.85 10.94 16 191609.6 7.5 53615.7
F2Y-1-1 6.94 2.95 9.97 8.76 12 143707.2 10.0 16242.5
F2Y-2-1 2.34 2.95 8.81 N/A 4 47902.4 N/A N/A
F2Y-2-2 2.34 2.95 8.81 N/A 4 47902.4 N/A N/A
F2Y-3-1 6.94 2.95 9.92 8.11 12 143707.2 10.0 16242.5
F3X-1-1 6.94 3.40 6.14 8.19 8 95804.8 10.0 18338.3
F3X-2-1 6.94 3.40 6.14 8.19 8 95804.8 10.0 18338.3
F3Y-1-1 6.94 2.75 4.78 9.47 8 95804.8 10.0 15194.6
F3Y-2-1 2.34 3.70 4.15 N/A 4 47902.4 N/A N/A
F3Y-2-2 2.34 3.70 4.15 N/A 4 47902.4 N/A N/A
F3Y-3-1 6.94 2.75 4.76 8.15 8 95804.8 10.0 15194.6

Two types of angle brackets (AB) are used in the design of the building, which are
manufactured by Rotho Blaas (2012): the WVS 90110 that is used for the shear-transferring
connections of the 1st storey, and the WB90 that is used for the connections of the upper
storeys. Table 4 summarizes the main properties of the angle brackets. The vertical step joints
between panels of the same wall assembly are constructed with a single plywood strip
27x150 mm inserted into a groove in the internal side of the wall panels, as shown in Figure
4b, and nails 2.8x80 that provide a design strength of 1.0 kN and a stiffness of 1047.9 kN/m
per connector. Table 3 lists the required number of angle brackets and the required screw
spacing in the vertical joints for each wall assembly as well as the corresponding stiffness for
each type of connection.

15
 Table 4. Design strength and stiffness of angle brackets
Angle Angle bracket to floor Angle bracket to wall Design strength Stiffness
bracket connection connection (kN) (kN/m)
WVS 1 12 steel rod class
11 4x60 anker nails 11.5 32932.8
90110 4.6
WB 90 8 4x60 anker nails 8 4x60 anker nails 5.8 11975.6

3.6 Final analysis and design of the building

Based on the results of the preliminary design, the numerical model is updated with
the correct stiffness contribution of the angle brackets and the vertical step joints to the
horizontal and vertical pair of springs, respectively. The building is analyzed with the modal
response spectrum analysis and Table 5 shows the fundamental periods and the mass
participation factors for each mode shape of the structure. The first mode shape, with a period
of 0.27 s that is similar to the 0.28 s computed with the code equation, is related to translation
along the N-S direction of the building. The second mode shape, with a period of 0.24 s, is
related to translation along the E-W direction of the building.

Table 5. Modal analysis results

Mass participation for Mass participation for Mass participation for
Period
Mode translation along N-S translation along E-W rotation along vertical
(sec)
direction (%) direction (%) direction (%)
1 0.27 83.91 0 0
2 0.24 0 84.27 1.35
3 0.16 0 0.83 84.76
4 0.09 12.59 0 0
5 0.08 0 12.59 0.03
6 0.06 2.88 0 0
7 0.06 0 0.11 10.99
8 0.05 0 1.61 0.09
9 0.04 0 0.02 1.94
Sum 99.38 99.43 99.16

The results of the response spectrum analysis in terms of shear per unit length, uplift
force for each wall, and shear per unit height for each vertical step joint are presented in
Table 6. The storey shear forces along the N-S direction are 205.6 kN, 165.0 kN and 80.4 kN
for the first, second and third storey, respectively, while along the E-W direction are 207.2
kN, 163.6 kN and 79.2 kN. These values are very similar to the values obtained from the
lateral force method of analysis.
Two types of hold-downs (HDs), the WHT 340 and the WHT 440, which are
manufactured by Rotho Blaas (2012), are used to restrain the building uplift. In addition, a
steel strap (SS) with dimensions of 100x1000x1.5 mm is used as a tie-down to restrain
external wall assemblies of the second and third storey against uplift. Table 7 summarizes the
main properties of the hold-downs and tie-downs. Self-tapping screws HBS 8x300 and
8x200 are prescribed for the connection of the floor and the roof panels, respectively, to the
supporting wall panels. The design strength per screw connector is 4.56 kN and 4.24 kN for
the 8x300 and 8x200, respectively. This type of connection is designed for a force equal to
the shear force of the supporting wall multiplied by an overstrength factor of 1.6. Self-
tapping screws HBS 8x200, with a design strength of 4.24 kN, are also prescribed for the
connection between perpendicular walls. This type of connection is designed for a force
equal to the maximum shear force per unit length between the two walls multiplied by the

16
height of the connection and further multiplied by an overstrength factor of 1.6. Table 6
shows the required number of ABs and the type of hold-downs/tie-downs for each wall
assembly, as well as the required spacing for the screws in the vertical step joints, the wall-to-
floor connections and the perpendicular wall connections.

Table 6. Final analysis and design results

Shear Vert. joint Vert. Joint Type of Wall-to- Perpendicular
No
per unit Uplift shear per screw hold- floor wall
Wall ID of
length (kN) unit height spacing down/tie connection connection
ABs
(kN/m) (kN/m) (cm) down spacing (cm) spacing (cm)
F1X-1-1 17.02 42.55 5.75 11 15.0 WHT440 15.0 10.0
F1X-2-1 17.02 42.55 5.75 11 15.0 WHT440 15.0 10.0
F1Y-1-1 22.52 34.84 N/A 3 N/A WHT340 10.0 10.0
F1Y-1-2 22.56 34.84 N/A 3 N/A WHT340 10.0 10.0
F1Y-2-1 17.52 34.01 N/A 4 N/A WHT340 10.0 10.0
F1Y-2-2 17.52 34.02 N/A 4 N/A WHT340 10.0 10.0
F1Y-3-1 15.90 46.61 N/A 4 N/A WHT440 10.0 10.0
F1Y-3-2 15.90 46.61 N/A 4 N/A WHT440 10.0 10.0
F2X-1-1 13.29 8.09 9.78 16 10.0 SS 20.0 15.0
F2X-2-1 13.29 8.09 9.78 16 10.0 SS 20.0 15.0
F2Y-1-1 10.41 24.25 5.67 13 15.0 SS 20.0 15.0
F2Y-2-1 8.42 4.65 N/A 4 N/A WHT340 20.0 20.0
F2Y-2-2 8.42 4.65 N/A 4 N/A WHT340 20.0 20.0
F2Y-3-1 9.29 14.65 4.69 12 15.0 SS 20.0 15.0
F3X-1-1 6.38 0.00 6.57 8 15.0 SS 20.0 20.0
F3X-2-1 6.38 0.00 6.57 8 15.0 SS 20.0 20.0
F3Y-1-1 4.83 0.00 8.19 8 10.0 SS 20.0 20.0
F3Y-2-1 5.22 7.97 N/A 4 N/A WHT340 20.0 20.0
F3Y-2-2 5.22 7.97 N/A 4 N/A WHT340 20.0 20.0
F3Y-3-1 4.18 0.00 6.39 8 15.0 SS 20.0 20.0

Table 7: Design strength and stiffness of hold-downs and tie-downs

Design
Type of hold- Connection to the floor for HDs Connection to the
strength
down/tie-down or to the wall below for SSs wall
(kN)
WHT 340 1 16 steel rod class 4.6 14 4x60 anker nails 36.5
WHT 440 1 16 steel rod class 4.6 18 4x60 anker nails 46.9
SS 100x1000x1.5 15 4x60 anker nails 15 4x60 anker nails 29.5

4. Validation of the Numerical Model

In order to validate the numerical modelling approach described in Section 2.2, the
experimental results from three low intensity shaking table tests of a three-storey full scale
CLT building [Ceccotti and Follesa, 2006] are compared with the numerical predictions from
linear dynamic analyses of a representative numerical model.
The test structure was tested with three different configurations, as shown in Figure
15, where the differences were the openings at the first floor along the W-E direction, the
direction of shaking. The third configuration is identical to the structure considered in the
case study building designed in the previous sections. The first configuration, considered in

17
this case for the validation of the numerical model, incorporated a 1.2 m long door opening at
the middle of the wall assemblies.
The ground motion records used for the shaking table tests are listed in Table 8. All
three configurations were tested under increasing amplitudes of Peak Ground Acceleration
(PGA) of 0.15g, 0.35g, 0.50g and higher. In order to validate the numerical model with a
linear analysis, the low intensity tests (with PGA equal to 0.15g) conducted on the initial
configuration, where the structure was not subjected to any strong shaking yet, were selected
as the most appropriate data for the appraisal of the numerical approach.

Figure 15. First (left), second (middle) and third (right) configuration of the test structure

Table 8. Details of the original ground motions used in the shaking table tests
Record
Country Date Station Component Duration (s) PGA (g)
Name
Kobe Japan 1995/01/16 JMA N-S 48.0 0.820
Imperial
El Centro California 1940/05/19 N-S 40.0 0.313
Valley
Nocera
Italy 1997/09/27 Nocera E-W 13.7 0.500
Umbra

4.1. Test structure and numerical model

The thickness and lay-up of wall and floor panels of the test structure were identical
to the case study building described in Section 3.1 and the orthotropic material properties
considered in the model are shown in Equation (11). The details of the shear-transferring
connections and the vertical step joints of the test building, which are those schematized in
the numerical model, are listed in Table 9, while Table 10 lists the corresponding stiffness
values of the connections according to the procedure described in Section 2.2 and presented
in Equations (6-10). Steel blocks were anchored on each floor diaphragm to augment the
gravity and seismic load and account for permanent and live loads in seismic combination as
prescribed by Eurocode 8. The dead and additional loads of the test structure are listed in
Table 11. Note that the dead loads are calculated for each storey by considering the floor
panels, half of the lower walls and half of the upper walls. The undeformed shape of the FE
model is shown in Figure 16.

Table 9. Details of the horizontal and vertical connections used in the test building
Vertical joint between wall
Storey Wall to floor horizontal connection for each floor
panels
6 BMF 90x48x3,0x116 for each wall assembly,
Self tapping screws 8x80
1 connected to the wall with 8 4x60 anker nails
spaced at 450 mm c/c
and to the foundation with one 16 anchor bolt
4 BMF 105 for each wall assembly, connected to
Self tapping screws 8x80
2 the wall and to the floor with 8 4x60 anker
spaced at 600 mm c/c
nails.
3 3 BMF 105 for each wall assembly, connected to Self tapping screws 8x80

18
 the wall and to the floor with 5 4x60 anker spaced at 900 mm c/c
nails.

Table 10. Total stiffness calculated for the horizontal and vertical springs
Total stiffness for angle bracket connections Total stiffness for vertical step
Storey
(kN/m) joints (kN/m)
K1=197597 (wall length 6.935 m - 6 brackets);
K=10479 (wall height 2.95 m - 7
1 K2=98798 (wall length 2.340 m - 3 brackets);
screws)
K3=98798 (wall length 2.868 m - 3 brackets)
K1=47902 (wall length 6.935 m - 4 brackets); K=7485 (wall height 2.95 m - 5
2
K2=23951 (wall length 2.34 m - 2 brackets) screws)
K1=22454 (wall length 6.935 m - 3 brackets);
K=4491 (wall height 2.95 m - 3
3 K2=14969 (wall length 2.340 m - 2 brackets);
screws)
K3=7485 (wall length 2.340 m - 1 bracket)

Table 11. List of dead and additional weights used in the test building
Storey Dead load (kN) Additional load (kN) Total load (kN)
1 69.3 150.0 219.3
2 69.3 150.0 219.3
3 47.4 0.0 47.4
Total 486.0

Figure 16. Undeformed shape of the FE model

4.2. Comparison with the test results

A modal linear analysis was carried out in SAP2000, with a modal damping of 5%
constant for all modes, to analyze the numerical model under the three recorded shake table
motions that were scaled to a peak acceleration of 0.15g. The comparison between the model
and the test results is made in terms of natural period, maximum base shear and maximum
displacements relative to the base measured at the centre of each storey. The experimental
results and the numerical predictions are presented in Table 12.
The results indicate that the numerical model predicts a more flexible response than
the actual test structure (natural vibration period of 0.195 s versus 0.166 s) and consequently
19
the maximum displacements are overpredicted from 20% to as much as 65%. Maximum
predicted base shear values are fairly well correlated to the experimental ones. This
difference in the lateral stiffness of the structure is attributed to the friction between
horizontal and vertical diaphragms, which is not accounted for in the numerical model but
can nevertheless contribute in reducing both the natural period of the building and the
measured displacements. A secondary effect that is not accounted for in the numerical model
is the contribution of the out-of-plane bending stiffness of the CLT floor panels in reducing
the rocking deformation of wall panels, although this effect can be more pronounced in the
direction of the floor panels, which in this case were spanning perpendicular to the seismic
direction.

Table 12. Comparison between model and experimental results

3rd storey
2nd storey 1st storey
Base shear (roof)
Period (s) displacement displacement
(kN) displacement
(mm) (mm)
(mm)
Kobe 0.15g
Model 0.195 77.3 2.52 1.77 0.74
Experimental 0.166 87.6 1.54 1.29 0.49
Difference (%) 17.47% -11.74% 64.01% 36.92% 52.53%
El Centro 0.15g
Model 0.195 138.4 5.37 3.64 1.40
Experimental 0.166 129.1 3.90 2.80 1.15
Difference (%) 17.47% 7.22% 37.78% 30.20% 21.41%
Nocera Umbra 0.15g
Model 0.195 193.90 7.17 4.94 1.93
Experimental 0.166 156.08 5.08 3.62 1.50
Difference (%) 17.47% 24.23% 41.20% 36.61% 28.26%

In order to quantify the effect of friction, a second numerical model is formulated that
incorporates an increased stiffness in the horizontal springs. The additional stiffness provided
by the friction was estimated first by selecting a coefficient of friction, which yields the
lateral strength at the base of the wall due to friction when multiplied by the gravity axial
load transferred by each wall. Then, a representative yield displacement is used to divide the
lateral strength and yield the stiffness due to friction. According to literature references
[Forest Products Laboratory, 2010], the coefficient of kinetic friction k depends on the
moisture content of wood and the roughness of the wood surface and of the second contact
surface, ranging from 0.3 to 0.5. In this example, a mean value of 0.4 has been assumed.
Consequently, considering the weight Wi at the base of each wall, the value of the friction
force Ff is calculated as:

F f   k  Wi (13)

The yield displacement is calculated by considering the yield displacement of the

weakest fastener of the shear-transferring connection, i.e. the 4 anker nails used in the
connection of the angle brackets. This may be evaluated, in the absence of test results, by
dividing the characteristic strength of the nail Rc,k, calculated according to EC5 with the
embedment strength of wood taken from the equations provided by Uibel and Blaß [2006 and
2007], by the slip modulus under service load of Equation (6), i.e.:

Rc ,k 2370
uy    0.79 mm (14)
2  1.5
m  d f / 23 2994

20
 Therefore, the stiffness due to friction is evaluated as:

Ff
Kf  (15)
uy

Adding the contribution of friction to the stiffness provided by the angle brackets, the
total horizontal stiffness of each wall assembly was recalculated and the results are shown in
Table 13.

Table 13. Total stiffness calculated for the horizontal and vertical springs taking into account
the friction contribution
Total stiffness for angle bracket connection Total stiffness for vertical step
Storey
(kN/m) joint (kN/m)
K1=197607 (wall length 6.935 m - 6 brackets);
K=10479 (wall height 2.95 m
1 K2=154189 (wall length 2.34 m - 3 brackets);
- 7 screws)
K3=124494 (wall length 2.868 m - 3 brackets)
K1=79252 (wall length 6.935 m - 4 brackets); K=7485 (wall height 2.95 m -
2
K2=55301 (wall length 2.34 m - 2 brackets) 5 screws)
K1=29771 (wall length 6.935 m - 3 brackets);
K=4491 (wall height 2.95 m -
3 K2=22286 (wall length 2.34 m - 2 brackets);
3 screws)
K3=14801 (wall length 2.34 m - 1 bracket)

Table 14 shows the comparison with the updated numerical model with the friction
contribution. The predicted natural period is still higher than the experimental one (0.182 s
versus 0.166 s) but apart from the displacement predictions for Kobe earthquake, a very good
correlation with the experimental results is obtained, thus leading to the conclusion that the
friction contribution should be taken into account in the calculation of the horizontal stiffness.
The modelling approach described in Section 2, where all the horizontal wall-to-floor
and vertical panel-to-panel connections are schematized with equivalent truss elements, may
require a long input procedure especially for large and tall buildings. For this reason,
sometimes simplified numerical models where all the joints between CLT elements are
considered as rigid are used by designers. In order to quantify the effect of modelling the
flexibility of the mechanical connections, a third numerical model was generated without any
connections among the shell elements. Figure 17 illustrates the comparison between the
experimental results and the numerical predictions from all three numerical models. As it can
be observed the differences from the experimental values of the third model are larger,
especially in terms of natural vibration periods and maximum displacements. This difference
in the natural period can lead to significant differences in the calculation of the design
seismic load, and confirms results of previous research [Fragiacomo et al. 2011].

Table 14. Comparison between model with friction contribution and experimental results
3rd storey
2nd storey 1st storey
Base shear (roof)
Period (s) displacement displacement
(kN) displacement
(mm) (mm)
(mm)
Kobe 0.15g
Model 0.182 82.3 2.54 1.75 0.76
Experimental 0.166 87.6 1.54 1.29 0.49
Difference (%) 9.64% -6.03% 65.32% 35.38% 56.65%
El Centro 0.15g
Model 0.182 123.3 4.05 2.74 1.15
Experimental 0.166 129.1 3.90 2.80 1.15
Difference (%) 9.64% -4.48% 3.91% -2.00% -0.27%
21
 Nocera Umbra 0.15g
Model 0.182 165.90 5.41 3.69 1.55
Experimental 0.166 156.08 5.08 3.62 1.50
Difference (%) 9.64% 6.29% 6.54% 2.04% 3.00%

80%

60%

40%

20%

0%

-20%

-40%

-60%

-80%

-100%
Kobe Kobe El Centro El Centro Nocera Nocera
Period Max. Base Max. Roof Max. Base Max. Roof Max. Base Max. Roof
Shear Displacemen Shear Displacemen Shear Displacemen

Rigid Connections -25.90% -16.19% -46.63% 16.05% -53.56% -4.28% -83.85%

Flexible Connections 17.47% -11.74% 64.01% 7.22% 37.78% 24.23% 41.20%
Flexible Connections + Friction 9.64% -6.03% 65.32% -4.48% 3.91% 6.29% 6.54%

Figure 17. Percentage difference between numerical models and experimental results

5. Conclusions

Design provisions for the seismic design and a numerical procedure for the linear
dynamic analysis of CLT buildings, which could be used by practicing engineers, are
proposed in this paper and explained with a design example of a three storey case study
building. The modelling procedure is based on code and literature references for the
calculation of the mechanical properties of CLT panels and mechanical connections.
The proposed numerical model is made of shell elements for the CLT wall panels and
truss elements for the mechanical connections between wall panels and between walls and
floors. The analysis is carried out using SAP 2000. Hold-down connections are not explicitly
modelled, but the uplifting vertical forces are obtained from the tensile forces of the vertical
truss elements which are used to simulate the deformability of the floor diaphragms along
their thickness.
The model has then been validated by comparing the analysis results of a sample
three storey building tested on a unidirectional shaking table in Japan in 2006 with the test
results. The comparison showed general good agreement for the natural vibration period and
base shear but a general overestimation of maximum displacements. Results in terms of
maximum displacements showed a much better correlation by taking into account the friction
contribution.
Finally a comparison was made with a third model without any connections among
the shell elements. The results in terms of differences from the experimental values of the
third model are large, especially in terms of natural vibration period, which can lead to
significant differences in the calculation of the design seismic load and maximum
displacements.
Although the model showed a good agreement with the test results, further
comparisons with other test results are needed in order to check and better calibrate the
modelling procedure, which is the following step of this work. Further research is also
needed to obtain prediction formulae for the evaluation of the slip modulus under service

22
 load of dowel type fasteners in CLT panels, in order to better estimate the correct linear
behaviour of CLT buildings via numerical models.

6. References

Blaß H.J., Fellmoser P., [2004]. “Design of solid wood panels with cross layers”, 8th World
Conference on Timber Engineering WCTE, Lahti, Finland, p. 543-8.
Ceccotti A., Follesa M., [2006]. “Seismic Behaviour of Multi-Storey X-Lam Buildings”,
COST E29 Workshop on Earthquake Engineering on Timber Structures, Coimbra, Portugal.
Ceccotti A., Follesa M., Lauriola M.P., Sandhaas C., [2006]. “SOFIE Project - Test results on
the lateral resistance of cross laminated wooden panels”, 1st European Conference on
Earthquake Engineering and Seismology Genevre, Switzerland.
Ceccotti A., Follesa M., Lauriola M.P., [2007]. “Which seismic behaviour factor for multi-
storey buildings made of cross-laminated timber panels?”, Proceedings of the 12th ANIDIS
Conference, Pisa, Italy, June 14th–17th (in Italian).
CSI, 2007. “SAP2000 V.11”, Computers and Structures Inc., Berkeley, CA, USA.
European Committee for Standardization (CEN), [2009] Eurocode 5— Design of timber
structures - Part 1-1: General - Common rules and rules for buildings, Brussels, Belgium.
European Committee for Standardization (CEN), [2004] Eurocode 8—Design of structures
for earthquake resistance, Part 1: General rules, seismic actions and rules for buildings,
Brussels, Belgium.
Follesa M., Fragiacomo M., Lauriola M.P., [2011]. “A proposal for revision of the current
timber part (Section 8) of Eurocode 8 Part 1”, 44th CIB W18, Alghero, Italy paper 44-15-1.
Fragiacomo M., Dujic B., Sustersic M., [2011]. “Elastic and ductile design of multi-storey
crosslam massive wooden buildings under seismic actions”, Engineering Structures, Special
Issue on Timber Structures, 33(11):3043-3053, doi:10.1016/j.engstruct.2011.05.20.
Forest Products Laboratory. [2010]. “Wood handbook—Wood as an engineering material”.
General Technical Report FPL-GTR-190. Madison, WI: U.S. Department of Agriculture,
Forest Service, Forest Products Laboratory.
Jorissen A., Fragiacomo M., [2011]. “General notes on ductility in timber structures”,
Engineering Structures, Special Issue on Timber Structures, 33(11):2987-2997.
Rotho Blaas S.r.l., [2012]. “Rothofixing: Timber Carpentry”, Cortaccia, BZ, Italy.
Sustersic I., Dujic B., [2012]. “Simplified cross-laminated timber wall modelling for linear
elastic seismic analysis”, 45th CIB W18, Vaxio, Sweden, paper 45-15-6.
Tecnisoft s.a.s., [2013]. “ModeSt – Version 8.1”, Prato, Italy.
Uibel T, Blaß H.J., [2006]. "Load Carrying Capacity of Joints with Dowel Type Fasteners in
Solid Wood Panels" 39th CIB W18, Firenze, Italy, paper 39-7-5
Uibel T, Blaß H.J., [2007]. "Edge joints with dowel type fasteners in cross laminated timber"
40th CIB W18, Bled, Slovenia, paper 40-7-2

23

View publication stats