Você está na página 1de 9

402

World Conference on Timber Engineering


Auckland New Zealand
15 - 19 July 2012
INFLUENCE OF THE CONNECTION BEHAVIOUR ON THE
SEISMIC RESISTANCE OF MULTI-STOREY CROSSLAM
BUILDINGS


Iztok Sustersic
1
, Massimo Fragiacomo
2
, Bruno Dujic
3






ABSTRACT: The paper investigates the inIluence oI modelling diIIerent types oI connections in multi-storey cross-
laminated (crosslam`) buildings when perIorming diIIerent types oI analyses. More speciIically, linear modal response
spectrum analysis, nonlinear static (push-over) analysis assessed with a modiIied N2 method, and nonlinear time-
history analysis are considered herein. In the research done by other authors, so Iar only the behaviour oI certain types
oI crosslam building assemblies was investigated. In this paper it is shown how diIIerent connection details inIluence
the building`s global ductility and peak ground acceleration. Basic values oI overstrength Iactors Ior BMF 105 angular
brackets and 8 mm selI-tapping screws are also presented and the importance oI using the overstrength concept in
design is demonstrated on a case-study wall.
KEYWORDS: Cross-laminated timber, Multi-storey buildings, Seismic design


1 INTRODUCTION
123

Cross-laminated timber (also reIerred to as crosslam` or
Xlam`) is a wood-based product that has become
widespread over Europe in the last decade. Recently, it is
also Iinding its way in other continents, namely Canada,
the United States, and also Australasia. In addition to
low-rise residential housing, crosslam has a great
potential in multi-storey timber construction
particularly in earthquake-prone areas. UnIortunately,
research on seismic behaviour oI crosslam buildings is
hardly keeping up with the increased need oI saIer and
more demanding applications Ior the building market.
On the one hand, a lot oI experiments (both wall racking
and shake table tests) have been conducted over the past
years. But on the other hand, the number oI publications
systematically dealing with the inIluence oI boundary
conditions such a connection types, Iriction, vertical
loads, Iloor diaphragm inIluence, wall panes sizes, etc.
on the global seismic behaviour oI multi-storey crosslam
buildings has been Iairly scarce.
Most oI the research work done so Iar has been Iocused
on only one type oI crosslam wall assembly, where the
walls are composed Irom several narrower panels,
adjacently connected with a single or double step joint

1
Iztok Sustersic, CBD d.o.o. , Lopata 19G, 3000 Celje,
Slovenia. E-mail: iztok.sustersiccbd.si
2
Dr. Massimo Fragiacomo, University oI Sassari, Dept. oI
Architecture, Design and Urban Planning, Alghero, Italy.
Email: Iragiacomouniss.it
3
Dr. Bruno Dujic, CBD d.o.o., Lopata 19G, 3000 Celje,
Slovenia. E-mail: bruno.dujiccbd.si
|1|. The FE analysis oI such structures was also carried
out under the assumption oI a stiII Iloor diaphragm.
This paper Iocuses on the inIluence that diIIerent wall
connection properties have on the global seismic
response oI the multi-storey building. The input
parameters Ior the parametric study were:
connection types varying in strength;
inIluence oI Iriction between walls and inter-
story Iloor slabs; and
inIluence oI vertical load on walls.
The basic values oI overstrength Iactors Ior BMF 105
steel angular brackets (with ten 60 mm nails) and 160
mm long selI tapping screws are also presented. The
need oI using overstrength Iactors is demonstrated on a
case- study wall.

2 CASE STUDY BUILDING
2.1 GEOMETRY AND LOADS
A Iour-storey crosslam building (Figure 1) was modelled
in this study. The building has 140 mm thick 5-layer
crosslam walls along its perimeter. Inside the building
there are only two posts and a beam that support 140 mm
thick crosslam slabs running Irom wall A` to wall C`.
Wall A` is made Irom two separate panels, which are
connected only with a beam element pinned onto the
walls (Figure 1). The other walls are all made out oI
single panels. The wall panels are connected at the
bottom and at the top with BMF105 bracket connectors.
The spacing oI the brackets was determined according to
the base shear Iorce calculated with the lateral Iorce
method (explained more in detail later). The building
SESSI ON 41, LATERAL-LOAD RESI STI NG SYSTEMS 5
403
World Conference on Timber Engineering
Auckland New Zealand
15 - 19 July 2012
was modelled in SAP 2000 |2|. For
panels when perIorming linear elasti
static analysis, shell elements and
coeIIicients proposed by Blass and Fe
used. The Iloor diaphragms were mode
connections among adjacent vertic
schematized in diIIerent ways (explaine

Figure 1: FE model 3D view
Figure 2: Wall dimensions (in cm) and l
bracket connectors used for wall B
The Iollowing data was used Ior the
spectrum analysis oI the building accor
8 |4|: type 1 elastic response spectra an
1.0, T
B
0.15, T
C
0.40, T
D
2.00), b
1.0 and lower bound Iactor Ior the des
0.20. Ground acceleration was assum
with a building importance Iactor
permanent load oI the Iloor and rooI w
2.0 kN/m
2
, respectively. The imposed
and rooI was 2.5 kN/m
2
and 2.0 kN/m
2
,
selI-weight oI the outer walls was
building was assumed to be catego
according to EN 1991-1-1 |5|, so the
quasi-permanent load was 0.3 and the
Ior all Iloors except Ior the rooI w
assuming the rooI is accessible. The m
modelling timber
ic and nonlinear
d the reduction
ellmoser |3| were
elled as rigid. The
al panels were
ed in detail later).


ayout of BMF 105
e modal response
rding to Eurocode
d type A soil (S
behaviour Iactor q
sign spectrum
ed to be 0.25 g,
y
I
1.0. The
was 3.5 kN/m
2
and
load on the Iloor
respectively. The
1.2 kN/m
2
. The
ory oI use A
e value oI
2i
Ior
Iactor was 0.5
where it was 1.0
ass was modelled
as lumped in the centroids oI t
oI gyration oI Iloor in plan are
Table 1: Mass and mass radiu
mass
|t|
4th storey 19.4
3rd storey 31.5
2nd storey 31.5
1st storey 31.5


2.2 CONNECTION PROPE
An important design issue is h
oI the bracket connections. In
that stiIIness, strength and duc
and screwed connections are d
procedure proposed by Yasu
wood Iramed shear walls.
determined by the intersection

Figure 3: Yasumura & Kawai p
evaluation of strength, stiffness
The Iirst line is drawn through
curve corresponding to 10 a
F
max
. The second line is d
corresponding to 40 and 90
translated so that it becomes ta
curve. The intersection oI this
the yield load and the corresp
ultimate displacement correspo
decreasing part oI the load
strength F
u
is calculated so th
deIormation energies is achiev
plastic load-displacement curv
marked gray in Figure 3. For B
with ten 460 mm nails, a ten
and shear stiIIness oI 2767 k
The ultimate tensile strength w
the ultimate shear strength was

D
L
o
a
d
0.9 F
max
0.4 F
max
F
ma
K
y
0.1 F
max
F
y
d
y
the Iloors. Mass and radius
summarized in Table 1.
us of gyration for each floor
mass gyration radius
|t m
2
|
185.2
300.5
300.5
300.5
ERTIES
how to model the stiIIness
n this paper, it is suggested
ctility oI the steel brackets
determined according to the
umura and Kawai |6| Ior
The yielding load is
oI two lines (Figure 3).

procedure [6] for the
s and ductility
h the points on the loading
and 40 oI the peak load
drawn through the points
oI F
max
. The line is then
angent to the actual loading
s and the Iormer line gives
ponding displacement. The
onds to 80 oI F
max
on the
ding curve. The ultimate
hat the equivalence oI the
ved by assuming an elasto-
ve, the area under which is
BMF 105 angular brackets
sile stiIIness oI 6345 kN/m
kN/m were calculated |7|.
was calculated as 21 kN and
s 16.5 kN.
Displacement
0.8 F
max
ax,
d
Fmax
d
u
F
u
SESSI ON 41, LATERAL-LOAD RESI STI NG SYSTEMS 5
404
World Conference on Timber Engineering
Auckland New Zealand
15 - 19 July 2012
3 LINEAR ELASTIC ANALY
The use oI linear elastic analysis is gene
a Iorce based design (FBD). When us
seismic analysis, the nonlinear respons
implicitly considered with the use oI the
q, which is employed to transIorm the
spectrum into a design spectrum
reduced accelerations. On the basis
spectrum, seismic Iorces acting on t
calculated. According to Eurocode
method Ior calculating Iorces on a con
in plan and elevation is the lateral Io
same method may also be used Ior
structures irregular in plan but regula
these requirements are not met, the
spectrum analysis must be used. The ca
was initially designed according to
method, with the number oI brackets be
resist the shear Iorces and the overturni
ground Iloor.

3.1 LATERAL FORCE METHOD
The building main vibration period, bas
shear Iorces are calculated according t
behaviour Iactor q equal to 1 was used
compare the maximum peak ground acc
building can resist calculated with a no
analysis with the design peak ground a
g). The ratio oI the Iormer to the latter
a simpliIied basic value oI the behavi
The behaviour Iactor q was then calc
aIorementioned procedure Ior diII
connections and wall boundary conditio
T
0
= u,uS B
3
4
= u,S1
w = _0
k]
+_
LI

k
= 11S9 kN
F
b
= S
d
(T
0
) W = u S
2,5
q
W =
= u,2S 1
2,5
1
11S9 = 712 kN
The building connections were designed
base shear demand. To withstand a shea
kN, 44 brackets are needed in every dire
oI them were installed on each oI the pa
(Figure 2). The resistance to overturning
seismic action was conservatively assum
provided only by the walls running in th
seismic load and their corresponding bra
Under horizontal Iorces the walls are as
about the perpendicular wall on the peri
considering also the beneIicial contribut
vertical load assumed to be applied in th
oI each Iloor, the tensile strength oI the
Iound to be suIIicient. No reduction Ior
and tensile Iorce in the angle brackets w
design.

3.2 MODAL RESPONSE SPECTRU
When determining the elastic stiIInes
use in the modal response spectrum an
oI the shear and tensile stiIIness we
YSIS
erally regarded as
sing linear elastic
e oI a building is
e behaviour Iactor
e elastic response
characterized by
s oI the design
the structure are
8, the simplest
nstruction regular
orce method. The
certain types oI
ar in elevation. II
modal response
ase study building
the lateral Iorce
eing calculated to
ng moment at the
se shear and story
to Eurocode 8. A
in design so as to
celeration that the
onlinear pushover
acceleration (0.25
quantity provides
iour Iactor q |1|.
culated using the
Ierent types oI
ons.
(1)
(2)
(3)
d to resist the
ar Iorce oI 712
ection, hence 24
arallel walls
g moment due to
med to be
he direction oI the
acket connectors.
sumed to rotate
imeter. When
tion oI the
he centre oI mass
brackets was
coupled shear
was considered in
UM
s oI brackets Ior
nalysis, the values
ere calculated as
described earlier in section
nonlinearly modelled using e
elements (elements very rigid
stiIIness in tension) at the
underneath, and elastic links I
exact boundary conditions
compression and elastic behav
The wall models were then
symmetrically elastic (same
compression) links were
displacements (assumed as
caused by shear slip and rot
nonlinear and linear cases wer
horizontal load (Figure 4). Wi
elastic model, a modal respon
be carried out.


Figure 4: Wall calibration proc
The Iloor diaphragms were
connections among adjacen
schematized in diIIerent ways
3D model with rigid connectio
springs Ior the top and bottom
without any connection betwe
the same level (pseudo 3D mo
elastic springs between top/bot
(Iull 3D model with elastic con
Figure 5 compares natural pe
Iloor lateral displacements oI
diIIerent types oI models. The
very high Ior model i`, w
assumed as rigid. This results
is conservative. However iI th
hence even stiIIer, it could yie
due to natural periods being
spectrum outside the plateau r
the top displacements are on
actually expected. Although t
the general stability oI the buil
eIIects are signiIicant), it can
estimation.
From Figure 5 it can be also o
between considering and ign
between perpendicular walls
than 4 when long wall segm
under study. This result may
wall panels are used like in t
the SOFIE research project |1|

2.2. Each wall was then
elastic shell elements, gap
d in compression with no
interIace with the wall
Ior brackets to simulate the
oI (rigid) contact in
viour in tension and shear.
recalibrated so that only
stiIIness in tension and
used, and the target
a sum oI displacements
tation oI the wall) Ior the
re the same under the same
ith the so-determined Iully
se spectrum analysis could

cedure
modelled as rigid. The
nt vertical panels were
s: (i) with rigid links (Iull
ons); (ii) with linear-elastic
m connections oI walls, and
een perpendicular walls at
odel); and (iii) with linear-
ttom and same level panels
nnections).
eriods, base shears and top
I the building Ior the three
e stiIIness oI the building is
where all connections are
s in higher base shears and
he building was lower and
ld non-conservative results
in the range oI increasing
region. On the other hand,
nly 20 oI what could be
this does not compromise
ldings (unless second order
lead to underrated damage
observed that the diIIerence
oring vertical connections
at the same level is less
ments are used as in the case
be diIIerent when shorter
the buildings developed in
|.
SESSI ON 41, LATERAL-LOAD RESI STI NG SYSTEMS 5
405
World Conference on Timber Engineering
Auckland New Zealand
15 - 19 July 2012

Figure 5: Four storey case-study crosslam building
Comparison among different models (i, ii and iii) in terms
of vibration periods (T1, T2 and T3), base shear (Rx and
Ry) and top floor displacement (Ux and Uy).
4 NONLINEAR STATIC ANALYSIS
The nonlinear static analysis (NSA) procedure is more
complex than the FBD, however it allows the designer to
take into account the actual inelastic behaviour oI the
structure. Furthermore, it can be used Ior PerIormance-
Based Design (PBD), where the design is achieved Ior
diIIerent perIormance levels such as no damage, limited
structural damage, important structural damage without
collapse, etc. Each level is generally linked to the
structural displacement by deIining a damage index and
by assigning a limit value Ior every perIormance level.
The NSA procedures are generally based on the
evaluation oI the push-over curve, which represents the
response oI the structure under a lateral loading
distribution schematising the seismic action. A number
oI diIIerent methods have been proposed, including the
N2 method |8|, which has been adopted by the Eurocode
8 |4|.

4.1 N2 METHOD
The N2 method considers a perIormance point deIined in
terms oI both strength and displacement, where the
structural capacity is compared with the demand oI the
seismic ground motion. The base shear Iorce and the top
displacement oI a Multi-Degree-oI-Freedom (MDOF)
system are Iirst computed by means oI a non-linear
Push-Over Analysis and then converted respectively to
the spectral acceleration and displacement oI an
equivalent Single-Degree-OI-Freedom (SDOF) system.
The demand oI the seismic ground motion is represented
through the response spectrum in terms oI pseudo-
acceleration and displacement. Such an inelastic
spectrum depends upon the cyclic behaviour oI the
SDOF system and the characteristics oI the ground
motion (peak ground acceleration and shape), and can be
obtained Irom the elastic spectrum using suitable
reduction Iactors.

Figure 6: Relation among Rp, p and T [8]
The N2 method was Iound to provide the best
approximation among various NSA methods Ior SDOF
systems with diIIerent hysteretic models and Ior MDOF
systems |9|. However the N2 was originally not
developed Ior the design oI timber buildings with
speciIic hysteretic behaviour. ThereIore it must be noted
that the results derived in this study could be non-
conservative, because hysteresis loops with pinching,
slip and strength degradation (typical Ior connections in
timber structures) dissipate less energy than bilinear
plastic loops with the same ductility.
Nevertheless, it should be also pointed out that in the
analysed setups, the SDOF systems equivalent to the
multi-storey building have periods longer than T
c
which
is usually the value Irom where the reduction Iactor (R
u
)
and ductility Iactor (u) are considered to be the same
(Fig. 6), regardless the type oI hysteresis loop. The
0
0,1
0,2
0,3
0,4
0,3
0,6
0,7
11 12 13
[
s
]
er|ods
l
ll
lll
0
30
100
130
200
230
300
330
8x 8y
[
k
N
]
8ase shear
l
ll
lll
0
3
10
13
20
23
30
ux uy
[
m
m
]
1op d|sp|acement
l
ll
lll
SESSI ON 41, LATERAL-LOAD RESI STI NG SYSTEMS 5
406
World Conference on Timber Engineering
Auckland New Zealand
15 - 19 July 2012
results oI these analyses should thereIore be considered
as a preliminary study aimed to investigate the eIIect oI
diIIerent wall boundary conditions.

4.2 MODIFIED SDOF BILINEARISATION
PROCEDURE FOR THE N2 METHOD
The standard bilinearisation procedure oI the pushover
curve suggested within the N2 method assumes the
attainment oI the structure ultimate displacement when
the Iirst structural component (beam, column, wall)
reaches the near collapse (NC) state. In this paper,
however, the NC state was deIined as a global condition
on the entire Xlam building according to the Yasumura
& Kawai |6| procedure. ThereIore the NC state is
assumed to be attained at a top Iloor displacement such
that the base shear Iorce oI the structure drops by 20
Irom the peak value. The initial stiIIness oI the
bilinearised pushover curve is also deIined according to
the aIorementioned procedure instead oI using the
standard N2 method Iormula. It should be noted,
however, that the results in this case could be non-
conservative, as the storey shear Iorce in an individual
Iloor might drop below 80 oI the peak value beIore
the base shear does the same.

4.3 PARAMETRIC ANALYSIS
The Iollowing types oI connectors were analysed in the
parametric study:
1. BMF 105 angular brackets with twenty 460 mm nails
1st cycle backbone curve. Unlike all other cases where
the 3rd cycle backbone curve is used, the 1st cycle
backbone curve with an approximately 25 higher
strength was considered in this analysis.
2. BMF 105 angular brackets with twenty 460 mm nails
that exhibited ductile behaviour in experimental tests and
thereIore showed a desirable Iailure mode.
3. BMF 105 angular brackets with twenty 460 mm
nails, without the inIluence oI the vertical load. In this
case the inIluence oI vertical load on the panels,
calculated in accordance with the Eurocode 8 load
combination G
i
,
EI
Q
i
was neglected, as opposed to all
other cases.
All the aIorementioned cases have Iriction between the
wall elements and interstory plates neglected. Cases 4, 5
and 6 are the same as cases 1, 2 and 3 but consider the
Iriction. In the FEM program SAP2000 |2| the Iriction
isolator` elements were used Ior modelling the eIIects oI
Iriction. A Iriction coeIIicient oI 0.4 was used (both Ior
static or dynamic loading), meaning that 40 oI the
vertical load on walls was transIormed into horizontal
resistance. It should be noted, however, that in the
literature Iriction coeIIicient values between two pieces
oI timber range Irom 0.25 to 0.5. A 2D pushover
analysis oI wall B was then carried out Ior the six
diIIerent cases listed above.


Figure 7: Pushover curves of wall B for different types of
boundary conditions
Table 2: Ductility ratios, seismic demand and capacity of
the case-study building in terms of maximum acceleration
depending upon the wall connection boundary conditions
when the wall B is loaded in the Y direction
1 2 3
bracket upliIt ductility 3.50 4.12 4.12
bracket shear ductility 3.29 4.45 4.45
building ductility 2.39 2.36 1.93
target displacement |mm| 35 38 45
maximum displacement 83 81 70
max. peak ground
acceleration |g|
0.59 0.52 0.38
max.ground acc.
dcsIgn ground acc.

2.36 2.08 1.52
4 5 6
bracket upliIt ductility 3.50 4.12 4.12
bracket shear ductility 3.29 4.45 4.45
building ductility 2.59 2.86 3.13
target displacement |mm| 33 33 36
maximum displacement 80 74 66
max. peak ground
acceleration |g|
0.61 0.55 0.45
max.ground acc.
dcsIgn ground acc.

2.44 2.20 1.80

From Figure 7 it can be clearly seen that the stiIIness and
peak Iorce oI the analysed wall is markedly lower iI the
vertical load is neglected (case 3). Conversely, using the
Iirst cycle`s response backbone (case 1) results in an
increase oI the wall`s stiIIness and strength. However the
building`s global ductility does not change much iI using
either the 1
st
or the 3
rd
cycle`s (case 2) backbone curve.
The maximum allowed ground acceleration is higher Ior
the 1
st
cycle`s response, Iollowed by the 3
rd
and Iinally
the lowest Ior the case without vertical load applied. II
Iriction is considered in the analysis, the allowed ground
acceleration is generally higher Ior all cases. The
0
30
100
130
200
230
300
330
0.00 0.02 0.04 0.06 0.08 0.10
8
a
s
e

s
h
e
a
r

[
k
N
]
1op f|oor d|sp|acement [m]
WALL 8 - LIILC1 CI IkIC1ICN
1) 1 S1. C?CLL, nC l8lC1lCn 4) 1 S1. C?CLL, l8lC1lCn
2) 3 8u. C?LL, nC l8lC1lCn 3) 3 8u. C?CLL, l8lC1C1lCn
3) 3 8u. C?CLL, nC l8lC. nC C8Avl1? 6) 3 8u. C?CLL, l8lC1lCn, nC C8Avl1?
SESSI ON 41, LATERAL-LOAD RESI STI NG SYSTEMS 5
407
World Conference on Timber Engineering
Auckland New Zealand
15 - 19 July 2012
situation is, however, signiIicantly diIIerent Ior the
building`s global ductility Iactor. Namely, by neglecting
the vertical load results in a higher ductility Iactor.
However the allowed acceleration is still smaller (due to
a reduction in shear resistance).

5 NONLINEAR DYNAMIC (TIME-
HISTORY) ANALYSIS
The time-history analysis was carried out by selecting a
recorded earthquake ground motion (using the Rexel
soItware |10|) characterized by an elastic spectrum
similar to the standardised Eurocode 8 elastic spectrum.
The South Iceland earthquake (waveIorm ID is 4675,
earthquake ID is 1635) that occurred on the 17
th
June
2000 and had a magnitude oI 6.5 (see Figure 8) was
used, more speciIically: the X direction component,
scaled by 1.9, with peak ground acceleration oI 0.129 g
(non-scaled value) that was recorded in an area with
ground type A according to Eurocode 8 (rock) at 13 km
Irom the epicentre. Figure 9 compares the elastic spectra
(with 5 damping) oI the chosen recorded accelerogram
with the EC8 elastic spectrum, showing that there is
good correspondence between them.


Figure 8: New Zealand South Island accelerogram
(waveform ID 4675) as recorded in the X direction on a
solid rock base (ground type A according to Eurocode 8)

Figure 9: Elastic EC8 response spectrum (with 5 %
damping) and the chosen accelerogram (scaled by 1.9)
response spectrum
Nonlinear springs with hysteretic behaviour
characterized by pinching and strength degradation (type
pivot` in Sap 2000 |2|) were used Ior modelling the
behaviour oI timber connections (angular steel BMF 105
brackets) in shear (Figure 10) and tension (Figure 11).
The springs were calibrated on the results oI
experimental cyclic tests |11|.
Nonlinear dedicated Iriction Iinite elements (type
Iriction isolator` in Sap 2000) were used (Iriction
coeIIicient oI 0.4) Ior modelling Iriction between walls
and plates.
Reyleigh damping was used Ior the calculation oI the
damping matrix, which is computed as a linear
combination oI the mass and stiIIness matrixes with
coeIIicients a
0
and a
1
based on the assumption oI5
constant damping in the most important vibration modes.


Figure 10: Calibration of nonlinear spring element type
'pivot' of Sap 2000 to the experimental response in shear
of 2xBMF 105 angular brackets

Figure 11: Calibration of nonlinear spring element type
'pivot' of Sap 2000 to the experimental response in
tension of 2xBMF 105 angular brackets
-0.13
-0.1
-0.03
0
0.03
0.1
0.13
0 2 4 6 8 10
A
c
c
e
|
e
r
a
t
|
o
n

[
g
]
1|me [s]
1
8
1
c
1
D
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 1 2 3 4
S
a

[

]
1
0
0

g
]
1 [s]
LC8 elasLlc specLrum Chosen accelerogram's specLrum
-30
-40
-30
-20
-10
0
10
20
30
40
30
-0,03 -0,02 -0,01 0,00 0,01 0,02 0,03
S
h
e
a
r

f
o
r
c
e

[
k
N
]
S||p [m]
2 x 8MI 10S shear
LxL8lMLn1 lLM SlMuLA1lCn lLM 8ACk8CnL
-20
-10
0
10
20
30
40
30
60
-0,01 0,01 0,02 0,03
U
p
|
|
f
t

f
o
r
c
e

[
k
N
]
D|sp|acement [m]
2 x 8MI 10S up||ft
LxL8lMLn1 lLM SlMuLA1lCn lLM 8ACk8CnL
SESSI ON 41, LATERAL-LOAD RESI STI NG SYSTEMS 5
408
World Conference on Timber Engineering
Auckland New Zealand
15 - 19 July 2012

Figure 12: FEM model of Wall B. Diagonal trusses are
used instead of shell elements. Nonlinear link elements
used for modelling friction between plates, axial and
shear forces in steel angular brackets can be seen on
every floor.
To reduce the signiIicant computational demand a truss
schematization oI wall B was investigated as an
alternative to the use oI shell elements (Figure 12). In
this simpliIied FE model, the xlam wall panel is
schematized with a couple oI diagonal truss elements,
pin-connected at each storey, with an axial stiIIness
calculated so that the top deIlection oI the panel assumed
as Iully Iixed to its base is the same as the top deIlection
oI the panel modelled using shell elements and Iixed to
the base. Nonlinear link elements are used Ior modelling
Iriction between plates, axial and shear Iorces resisted by
angle brackets..



Figure 13: Comparison of wall B top floor displacements
when vertical load is considered or is neglected. Friction
was neglected in both cases.

Figure 14: Comparison of wall B top floor displacements
when friction between the wall and floor panels is
considered or is neglected. The vertical load was
considered in both cases.
In Figure 13, the top Iloor displacements are compared
when the eIIect oI vertical load was considered
(generally, a beneIicial eIIect) or neglected. Friction
between wall and Iloor panels was neglected in both
cases. The average maximum displacements are about
the same during the strongest cycles oI the earthquake
ground motion. However in the model without the
vertical load the displacement amplitudes reach higher
values in an earlier stage and the model continues to
sway with larger amplitudes even aIter the more
intensive part oI the ground motion is Iinished. The
gravity load, in Iact, causes a stabilising moment capable
to counteract the overturning moment due to the
horizontal seismic Iorces.
Figure 14 shows that Iriction seems to signiIicantly
inIluence the maximum top Iloor displacement. The
diIIerence is most obvious in the more intensive part oI
the ground motion, where the model with no Iriction
exhibits 3-times higher displacements on the top oI the
wall compared to the model with Iriction.
Based on these Iirst preliminary comparisonsvertical
load does not seem to noticeably aIIect the maximum
amplitudes oI the total displacements. However, since
the structure starts to experience higher displacements at
an earlier stage and damps out later the structural
damage would increase as the connecting elements go
through more high amplitude cycles, hence accumulating
more damage.
Friction seems to have a strong eIIect on the building
response. However it should be noted that the vertical
acceleration component was not taken into account in
this study. Vertical acceleration could result in the loss
oI contact between walls and plates, hence reducing
beneIicial eIIect oI Iriction. Nevertheless Iurther studies
with a higher number oI accelerograms should be carried
out beIore Iinal conclusions on this issue can be drawn.

6 OVERSTRENGTH FACTORS
6.1 BASICS
To ensure a structure can dissipate energy during a
seismic event, capacity-based design must be used so as
to avoid brittle Iailure mechanisms. This is achieved by
designing the structural elements which may Iail in a
brittle manner under an increased seismic demand E
d

given by

d od Rd d
E E ' (4)

where E
d
is the seismic demand on the structural element
calculated using the design spectrum reduced to allow
Ior the ductile behaviour oI the structure, y
od
is the
overdesign Iactor, given by

d
d
od
E
F
(5)
where F
d
is the design strength capacity oI the ductile
structural element, and y
Rd
is the overstrength Iactor,
given by

d
Rd
F
F
95 . 0
(6)
-60
-40
-20
0
20
40
60
0 2 4 6 8 10
D
|
s
p
|
a
c
e
m
e
n
t
[
m
m
]
1|me [s]
verLlcal nC verLlcal
-60
-40
-20
0
20
40
60
0 2 4 6 8 10
D
|
s
p
|
a
c
e
m
e
n
t
[
m
m
]
1|me [s]
lrlcLlon nC frlcLlon
SESSI ON 41, LATERAL-LOAD RESI STI NG SYSTEMS 5
409
World Conference on Timber Engineering
Auckland New Zealand
15 - 19 July 2012
where F
0.95
is the 95
th
percentile oI the actual peak
strength capacity oI the ductile structural element. Whilst
the overdesign Iactor depends on the rounding carried
out in the actual design and execution (i.e. installing
more brackets than necessary), the overstrength Iactor
depends on the type oI material and structural details.
The Eurocode 8 provides the values oI the overstrength
Iactors Ior steel and reinIorced concrete structures,
which are in the range 11.3, however there is no
provision Ior timber structures. In the New Zealand
timber standard |11|, a value oI 2 is suggested Ior the
overstrength Iactor.

6.2 EVALUATION OF THE OVERSTRENGTH
FACTORS
Dujic and Zarnic |12| carried out a small number oI tests
on cross-laminated wall connections with metal bracket
loaded in tension and shear and selI tapping screw
loaded in shear. The crosslam panel had three layers oI
boards with 30-34-30 mm thickness. Tests were
perIormed on BMF 105 angular brackets with ten 460
mm and 440 mm nails. Tests were also perIormed on
shear connections with 8 mm diameter selI tapping
screws.The yield point was computed according to the
Yasumura & Kawai procedure. The overstrength Iactors
were later calculated by the writers |7, 13|. In Table 3
only the overstrength Iactors oI selI tapping screws and
brackets with 60 mm nails are listed, as the 40 mm long
nails exhibited brittle plug shear Iailure not suitable Ior
seismic design. More details oI the calculation can be
Iound in |7|.

Table 3: Experimental results for BMF 105 angular
brackets with ten 60 mm long nails and Wuerth Assy II
8160/80 self tapping screws
No. oI spec.
tested
Overstrength
Iactor y
Rd

BMF 105 shear 3 1.26
BMF 105 upliIt 2 1.18
Wuerth 8160/80 5 1.63

The overstrength Iactors Ior angle brackets are 1.3 Ior
shear and 1.2 Ior upliIt. It must be noted, however, that
the number oI tested specimens on which the derived
Iactors are based is very small. More specimens should
be tested to draw Iinal conclusions on the overstrength
Iactors.

6.3 SINGLE WALL CASE-STUDY
When dealing with Iull length solid wall panels, the
chance oI a brittle Iailure oI the panel itselI is very low iI
a standard connection system is used. However, the
situation changes Ior walls with large openings, where
concentrations oI stresses in the corners may cause
timber to Iail brittle beIore the ductile mechanism in the
brackets or screws can take place. In such cases the
timber panels must be designed by taking into account
the overstrength Iactors oI the connections.
In the case-study discussed herein (a 240280 cm 3-
layer Xlam panel with a 144184 cm opening in the
middle), load is applied horizontally on the top oI the
cantilevered type wall.


Figure 15: Pushover curves for case "i" (full curve) and
case "ii" (dashed curve). Grey dotted line (20.4 kN)
denotes the base shear value where the Xlam timber
frame cracks.

Pushover curves (Figure 15) are presented Ior the wall
braced at the bottom with two BMF 105 brackets with
ten 460 mm nails Ior two cases: (i) brackets with
characteristic strength; and (ii) brackets with strength
exceeding the characteristic value by 30 (overstrength
oI 1.3). It is shown that the stresses in timber are
exceeded Ior the latter case, leading to crack Iormation
in the wall panel at the corner oI the opening. ThereIore
the timber panel should be redesigned Ior a higher load
corresponding to the overstrength oI the bracket
connection.

7 CONCLUSIONS
The paper investigates how diIIerent connection
properties can inIluence the building seismic response.
The elastic modal analysis has shown that rigid
connections between walls and plates result in very low
vibration periods, hence causing too low displacements.
However in the case-study building the prediction oI the
base shear Iorce is conservative, as the structure main
vibration period Iell in the plateau oI the design elastic
spectra.

The results Irom the nonlinear static analyses show that
iI the Iirst cycle hysteresis backbone is used in the
analysis to model the non-linear response oI angular
brackets, the building can withstand about 10 larger
peak ground acceleration. When including Iriction in the
model, the pushover curves are slightly steeper and the
allowable ground acceleration a bit higher.

The results Irom the nonlinear dynamic analyses show
that the vertical load does not noticeably aIIect the
maximum displacement amplitudes. However the
damage to the structure will be greater iI the vertical load
on walls is neglected. The connecting elements go
through more high amplitude cycles, hence accumulating
0
3
10
13
20
23
0 0.01 0.02 0.03 0.04
8
a
s
e

s
h
e
a
r

[
k
N
]
1op d|sp|acement [m]
SESSI ON 41, LATERAL-LOAD RESI STI NG SYSTEMS 5
410
World Conference on Timber Engineering
Auckland New Zealand
15 - 19 July 2012
more damage during an earthquake. Considering Iriction,
however, results in a much less intensive response oI the
structure, which may be non-conservative since the
vertical acceleration was not considered. Further
numerical analyses are needed to obtain Iinal
conclusions.
The overstrength Iactors oI angle brackets were
evaluated based on the results oI experimental cylic
tests. The value oI 1.3 under shear Iorce and 1.2 under
tensile Iorce were obtained, and should be used in
capacity base design oI crosslam buildings.

ACKNOWLEDGEMENT
The research support provided to the Iirst author by the
EU through the European Social Fund 'Investing in your
Iuture' is grateIully acknowledged.



REFERENCES
|1| Ceccotti A. New technologies Ior construction oI
medium-rise buildings in seismic regions: the
XLAM case. IABSE Structural Engineering
International, Special Edition on Tall Timber
Buildings 2008, 18(2):156-165.
|2| Computers & Structures Inc. SAP2000Integrated
Finite Element Analysis and Design oI Structures.
Computers & Structures Inc.: Berkeley, CA, 2000.
|3| Blass HJ, Fellmoser P. Design oI solid wood panels
with cross layers. 8th World ConIerence on Timber
Engineering, WCTE 2004, Lahti, Finland; 543-548.
|4| European Committee Ior Standardization (CEN).
Eurocode 8Design oI structures Ior earthquake
resistance. Part 1: General rules, seismic actions and
rules Ior buildings. EN 1998-1; Brussels, 2004.
|5| European Committee Ior Standardization (CEN).
Eurocode 1: Actions on structures Part 1-1:
General actions Densities, selI-weight, imposed
loads Ior buildings; 2002, Brussels, Belgium.
|6| Yasumura M, Kawai N. Evaluation oI wood Iramed
shear walls subjected to lateral load. Meeting 39 oI
the Working Commission W18-Timber Structures,
CIB; 1997, Vancouver (Canada), paper CIB-
W18/30-15-4.
|7| Fragiacomo M, Dujic B, Sustersic I. Elastic and
ductile design oI multi-storey crosslam masive
wooden buildings under seismic action. Engineering
Structures 2011; 33(11):3043-3053.
|8| FajIar P. A nonlinear analysis method Ior
perIormance-based seismic design. Earthquake
Spectra 2000; 16(3):573-592.
|9| Fragiacomo M, Amadio C, Rajgelj S. Evaluation oI
the structural response under seismic actions using
non-linear static methods. Earthquake Engineering
& Structural Dynamics 2006; 35(12):1511-1531.
|10| Iervolino I, Galasso C, Cosenza E. REXEL:
computer aided record selection Ior code-based
seismic structural analysis. Bulletin oI Earthquake
Engineering 2010; 8:339-362, DOI 10.1007/s10518-
009-9146-1
|11| New Zealand Standard. Timber structures standard,
NZS3603:1993. Published by Standards New
Zealand, Private Bag 2439, Wellington 6020, New
Zealand.
|12| Dujic B, Zarnic R. Report on evaluation oI racking
strength oI KLH system. University oI Ljubljana,
Faculty oI civil and geodetical engineering,
Slovenia, 2005.
|13| Sustersic I, Fragiacomo M, Dujic B. InIluence oI
connection properties on the ductility and seismic
resistance oI multi-storey cross-lam buildings. CIB-
W18. Alghero, Italy, 2011.

SESSI ON 41, LATERAL-LOAD RESI STI NG SYSTEMS 5

Você também pode gostar