Content uploaded by Reinhard Brandner
Author content
All content in this area was uploaded by Reinhard Brandner on Jan 27, 2014
Content may be subject to copyright.
1
In plane Shear Strength of Cross Laminated Timber
(CLT): Test Configuration, Quantification and
influencing Parameters
R BRANDNER 1) 2); T BOGENSPERGER 2); G SCHICKHOFER 1)
Graz University of Technology, Institute of Timber Engineering and Wood Technology 1)
Competence Centre holz.bau forschungs gmbh 2)
1 Abstract
Cross laminated timber (CLT) has become a well-known and widely applied two-dimensional,
engineered timber product worldwide. It constitutes a rigid composite of an odd number of orthogonal
and glued layers. Focusing on a single glued node loaded in plane in shear and composed of two
crossed board segments and the adhesive layer in-between, in principle three types of shear
mechanisms can be distinguished: mechanism I “net-shear” (shearing perpendicular to grain),
mechanism II “torsion” and mechanism III “gross-shear” (shearing parallel to grain). In fact, while
having generally accepted values for the resistance against mechanism II and good estimates for
mechanism III the resistance against “net-shear” (mechanism I) is still in discussion. In spite of
numerous investigations on nodes and on whole CLT elements in the past, a common sense
concerning the test procedure, the consideration and handling of distinct influencing parameters and
the quantification of the shear strength are open.
We focus on the in plane shear resistance of single nodes according to mechanism I. We (i) propose a
test configuration for reliable determination of the shear strength, (ii) determine the shear resistance in
case of shear loads perpendicular to grain, (iii) discuss influences of some parameters on the shear
strength of single nodes, and (iv) give a brief outlook concerning the resistance of CLT elements
against shear loads in plane.
2 Introduction
Cross laminated timber (CLT) constitutes a solid, laminar engineered timber product with high
resistances against loads in and out of plane. Common CLT is a rigid composite of an odd number of
orthogonal and face bonded layers. Each single layer consists of side-by-side aligned (finger jointed)
boards with or without edge bonding. In CLT without edge bonding gaps between the boards, more or
less regular in width, are evident. Common gap widths allowed by technical approvals for CLT are
2 (3) mm in the top and 4 (6) mm in the core layers (Brandner 2013).
We focus on the mechanical properties of CLT loaded in plane. In particular, the resistance in plane in
shear of CLT made of Norway spruce (Picea abies) is addressed. Three principle shear mechanisms
are distinguished: mechanism I “net-shear”, mechanism II “torsion” and mechanism III “gross-shear”
(see e.g. Bogensperger et al. 2007 & 2010, Blaß and Flaig 2012). Mechanism I “net-shear”
corresponds to shearing perpendicular to grain of the net cross sections in the controlling plane.
Mechanism III “gross-shear” is associated with shearing parallel to grain of the whole CLT element.
For clarification of these mechanisms, at first some simplifications for the mechanical treatment are
made according to Bogensperger et al. (2010).
2.1 Some general Comments on the Shear Mechanisms
Following Bogensperger et al. (2010) a representative volume element (RVE) is introduced, which is
in thickness equal to a CLT element and in width and depth equal to the width of one board plus the
half of the width of gaps between adjacent boards. Focusing on shear in a CLT element with constant
2
layer thicknesses (tl,i ≡ tl, i = 1, …, N) and an infinite number of layers N → ∞ (neglecting boundary
conditions) the RVE can be further simplified to a representative volume sub-element (RVSE). This
RVSE is in width and depth equal to the RVE but in thickness equal to tl, composed of both half
thicknesses of two orthogonal boards in one node and the face bonding in-between (see Fig. 1).
As consequence of N → ∞ a proportional shear force nxy,RVSE instead of the overall shear force nxy is
defined. The nominal shear stress τ0 is given as (see Bogensperger et al. 2010)
,RVSE
0
xy
l
n
at
τ=
⋅
, (1)
with a as width and depth, and tl as thickness of the RVSE, respectively. This more theoretical shear
stress corresponds to mechanism III “gross-shear” (see Fig. 1). Thereby a constant shear stress
distribution over the cross section is assumed, which may lead to shear failures parallel to grain in all
layers. Therefore, an intact edge bonding between the boards within one layer and the absence of
checks is required. Missing or insufficient connection between the boards at the edges disables the
transfer of shear stresses in that direction. For the resistance against mechanism III Blaß and Flaig
(2012) recommend a characteristic (5 %-quantile) shear strength of fv,gross,k = 3.5 N/mm². In view of
EN 338 and with kcr = 1.00 (factor which considers the influence of cracks on the shear strength) a
value of fv,gross,k = 4.0 N/mm² for CLT composed of boards of strength class C24 according to EN 338
(the common material used for CLT in Europe) is proposed (Flaig and Blaß 2013). In case of stress
relieves adaptation of fv,gross seems to be necessary.
Of course, in CLT composed of layers without edge bonding, shear force can only be transferred via
the cross sections of boards and via the gluing interface of the face bonding. Comparable conditions
are expected in edge bonded CLT exposed to common climate variations. Moisture induced stresses
caused by these climate variations lead to checks, which again restrict the possibilities for shear
transfer. Consequently, mechanism I and II become active and their verification mandatory in the
design process, even in cases of gap widths tgap → 0.
Fig. 1: (block left) RVE and RVSE of a CLT element; (block right) shear stresses in a RVSE: nominal shear stress τ0
(left), real shear stress τnet (middle; superposing left & right), torsional stress τtor on the gluing interface (right)
(Bogensperger et al. 2010; adapted)
Mechanism I considers the transfer of shear forces via the cross sections of boards within a RVSE.
Consequently, the shear stress is given as
net 0
2τ=⋅τ, (2)
with τnet as the shear stress dedicated to the net cross section (see Fig. 1). For calculation of stresses
caused by nxy in a real CLT element, considering e.g. boundary conditions caused by finite N and
variations in layer thickness, a procedure is provided e.g. in Bogensperger et al. (2010).
Shear strain in the RVSE, in case of insufficient or missing connection between the board edges,
causes also torsional strain in the surface bond layer. This may cause failure in the gluing interface,
which is dedicated to mechanism II “torsion” (see Fig. 1). Assuming polar torsion, the torsional shear
stresses τtor are given as
t
CLT
RVE
RVSE
t
l
RVE and RVSE of a CLT elem ent
Y
ZX
a = w
l
a = w
l
t
l
Y
ZX
a
a
t
l
Y
ZX
a
a
t
l
M
tor
nominal shear forces
idealised RVSE without checks
with edge bonded boards
shear forces
RVSE with checks or gaps, without edge bonding
half system!
2
with Mt
o
geometr
i
thickest
against
t
(2004)
a
rolling s
ring pat
t
commo
n
To conc
l
or gaps
d
shear”
a
against
quantifi
c
2.2
In gene
r
investig
a
et al. 20
or multi
p
2.2.1
I
Bosl (
2
1,200 x
tension
l
and co
m
b
ucklin
g
surface
b
was not
of τgross,
m
Later,
T
(2007)
m
dimensi
o
with ga
p
with tw
o
at all c
o
tests. T
h
compre
s
element
s
and at t
h
were no
t
b
oards
p
to 134.0
Andreol
l
point be
(ii) two
b
onding
)
forth fa
i
provide
d
dividing
account
or
as torsion
a
i
c paramete
r
layer gover
n
t
orsion, num
e
a
nd Jöbstl et
hear to zon
e
t
ern, and (i
i
n
sense to us
e
l
ude, in a re
a
d
ue to missi
n
a
nd mechani
torsion in
t
c
ation of the
Shear Me
r
al, investig
a
a
tions perfo
r
12), and (ii)
p
le RVSE; e
I
nvestigati
o
2
002) rep
o
1,200 x 85 (
5
l
oaded four-
h
m
pression. I
n
g
of single
b
b
onding can
observed. T
h
m
ean ≡ τ0,mean
≈
T
raetta et al.
m
ade tests
o
n 560 x 5
p
s between t
h
o
squared te
s
o
rners, was
h
e load on
s
sion. For a
s
were conti
n
h
e boarders
r
t
in shear b
u
p
er layer an
d
kN) can be
c
l
i et al. (201
2
nding as w
e
of four tes
t
)
, one dela
m
i
led in shea
r
d
fv,net = 12.7
the maxim
u
the real stre
a
l moment
a
r
ratio tl / a
.
n
s the desi
g
e
rous invest
i
al. (2004).
e
s exposed t
o
i
) surface a
r
e
ftor,k = 2.5
N
a
l CLT ele
m
n
g edge
b
on
d
sm II “torsi
o
t
he gluing
resistance a
g
chanism
I
a
tions on th
e
r
med on wh
o
investigatio
n
.g. Wallner
2
o
ns on CLT
o
rt on tes
t
5
x 17 mm)
m
h
inged steel
-
n
all four s
p
b
oards in t
h
be conclud
e
h
e mean ulti
≈
2.3 N/mm²
(2006) an
d
on three-la
y
60 x 120 (3
0
h
e boards o
f
s
t fields, eq
u
used for th
e
the steel f
r
continuous
l
n
uously bon
d
r
einforced b
y
u
t locally in
d
all five test
c
alculated.
2
) mention t
e
e
ll as (ii) di
a
t
s failed in
m
inated in t
h
r
perpendic
u
N/mm² as
m
u
m load by
ss distributi
o
tor
τ=
a
nd Ip as po
l
.
Thus, in
a
n (Bogensp
e
i
gations wer
e
A possible
r
o
shear is me
r
ea under to
N
/mm² as ch
a
m
ent at com
m
d
ing, in the
d
o
n”, need v
e
interface th
i
g
ainst shear
“net-she
a
e
shear stre
n
o
le CLT ele
m
n
s performe
d
2
004, Jöbstl
E
lements
t
s conduct
e
m
m³. The e
-
frame (see
F
p
ecimens w
i
h
e top-layer
s
e
d. A signifi
c
m
ate load
w
and τnet,mean
d
Bogenspe
r
y
er CLT el
e
0
+ 60 + 30
m
f
5 mm. A s
t
u
al in size a
n
e
three-poin
t
r
ame was
a
l
oad transfe
r
d
ed to the st
e
y
hardwood
l
compressio
n
s
a maximu
m
e
sts on thre
e
a
gonally in c
t
orsion (on
e
h
e lateral su
r
u
lar to grain
m
aximum v
a
the net cro
o
n in diago
n
3
tor
3
2
p
M
a
I
⋅=
l
ar moment
a
real CLT
e
rger et al.
2
e
made in t
h
r
edistributio
n
e
ntioned. De
s
o
rsion (see
e
aracteristic
(
m
on use the
o
d
esign proc
e
e
rification.
A
is contribut
i
according t
o
a
r”: State
-
n
gth of CL
T
m
ents (e.g. B
d
on single
n
et al. 2008,
H
e
d on fi
v
lements we
r
F
ig. 2, left).
i
th orthogo
n
s
as conseq
u
c
ant damag
e
w
as Fmax,mean
=
≈ 5.6 N/mm
²
r
ger et al.
e
ments o
f
m
m) mm³,
t
eel frame
n
d hinge
d
t bending
a
pplied in
r
the CLT
e
el frame
lamellas (se
e
n
. Neverthel
e
m
shear stre
s
e
and five la
y
c
ompression
e
with edge
r
faces (three
, allocable
t
a
lue of shear
ss section
m
n
al compress
i
0l
t
a
⋅τ ⋅
,
of inertia.
C
element wi
t
2
010). Con
c
h
e past, e.g.
B
n
of torsion
a
s
pite some i
n
e
.g. Jeitler 2
(
5 %-quantil
e
o
ccurrence o
f
e
ss both she
a
A
s there is
c
i
on concen
t
o
mechanis
m
-
of-the-Ar
t
T
in plane
b
osl 2002, B
o
n
odes (in di
m
H
irschmann
v
e-layer C
L
r
e freely pl
a
Consequent
l
n
al layers, t
h
u
ence of de
l
e
of the elem
=
325 kN, w
h
²
.
F
ig. 2: Tests
Andr
e
e
Fig. 3, left
e
ss, conside
r
s
s of at least
y
er CLT ele
m
by means o
f
bonding b
u
layer speci
m
t
o mechanis
m
stress. The
m
ultiplied b
y
i
on test, wh
i
C
onsequentl
y
t
h varying
l
erning the r
B
laß and Gö
r
a
l stresses fr
o
n
fluences of
004, Jöbstl
e
) torsional
s
f
checks du
e
r
mechanis
m
c
ommon se
n
r
ates on th
e
I.
t
y testing ca
n
o
gensperger
m
ension corr
e
2011).
L
T elemen
t
a
ced in a s
q
l
y, the CLT
h
e ultimate
l
aminated la
y
e
nts at zone
s
h
ich corresp
o
on CLT ele
m
e
olli et al. (201
2
)
. However,
r
ing the inn
e
τnet,mean ≈ 6.
0
m
ents loade
d
f
short steel
t
cracks, th
e
m
en with e
d
m
I (Fig. 2,
stress calcul
y
a correcti
o
i
ch is not pu
r
y
, τtor depen
d
l
ayer thickn
r
esistance o
f
r
lacher (200
2
om zones e
x
parameters
(
et al. 2004
)
s
hear strengt
h
e
to climate
v
m
s, mechani
s
n
se on the
r
e determin
a
a
n be classi
fi
et al. 2007,
A
e
sponding t
o
t
s with
d
q
uared, diag
was stresse
d
load was li
m
a
yers. An in
s
s
of load int
r
onds to she
a
m
ents: Bosl (
2
2
; right)
the observe
d
e
r shear fiel
d
0
N/mm² (F
m
d
(i) edgewis
angles. In t
e
e
other wit
h
d
ge bonding
)
right). Thi
s
l
ation was c
a
o
n factor, ta
k
u
re shear str
e
(3)
d
s on the
e
sses the
f
a RVSE
2
), Jeitler
x
posed to
(
i) annual
)
there is
h
.
v
ariations
sm
I “net-
r
esistance
a
tion and
fi
ed in (i)
A
ndreolli
a double
d
imension
onally in
d
in shear
m
ited by
s
ufficient
r
oduction
a
r stresses
2
002; left),
d
failures
d
of 5 x 5
m
ax = 78.1
e
in fou
r
-
e
st group
h
out edge
)
, and the
s
last test
a
rried out
k
ing into
e
ss but an
interacti
o
test pro
c
Sieder (
2
data is n
CUAP (
2
Approv
a
longitu
d
Howeve
perpend
i
mention
to the C
U
failed in
llel to g
r
mechan
i
the fail
u
verificat
this has
cross se
c
shear st
r
failure
a
11.5 N/
m
weighte
d
To sum
m
mechan
i
the one
fact, str
e
indicate
s
experie
n
element
s
In view
of the el
e
all thre
e
models,
wl = 15
0
2.2.2
I
Wallner
interfac
e
with a
(see Fig
failures
gluing
failures
horizon
t
In that
stresses
several
t
of 5.9 to
Based
o
determi
n
possible
tgap = 5
m
censore
d
o
n between
c
edure for C
L
2
013). Alth
o
n
ot presented
2
005) provi
d
a
ls (ETAs)
b
d
inal boards
r
, Jöbstl et
a
i
cular to gra
i
eight test s
e
U
AP proce
d
bending be
t
r
ain, whereb
y
i
sms are no
t
u
re schedul
e
t
ion. Accord
i
to be don
e
c
tion. The o
b
r
esses τnet,me
a
a
re in the ra
n
m
m², with
d
mean of 8.
4
m
arise: unt
i
i
sm I are mi
s
of Andreoll
i
e
ss levels of
s
that the re
s
n
ces made o
u
s
.
of the moti
v
ements “
b
o
a
e
, in princip
l
at least for
0
mm and tl
=
I
nvestigati
o
(2004) inv
e
e
, on three l
a
loading in
. 4, left). B
e
in rolling
interface,
parallel to
t
al board w
e
cases the
τv,net,mean a
t
t
est series ar
e
o
7.0 N/mm².
o
n Wallner
n
ing the bea
r
failure pla
n
m
m. As the
d
. Jöbstl et
a
shear and c
o
L
T columns
o
ugh a succe
s
.
d
es a test co
n
b
ased on th
e
enforces tr
a
a
l. (2008) an
d
i
n according
e
ries of thre
e
d
ure. None o
f
t
ween the lo
a
y
b
oth failur
t
conform t
o
e
d for she
a
i
ng to CUA
P
e
on the n
e
b
served mea
n
a
n at bendin
g
n
ge of 5.4 t
o
an overa
l
4
N/mm².
i
l now test
s
sing. The o
n
i
et al. (201
2
τnet at maxi
m
s
istance of C
u
tline also t
h
v
ation to est
a
a
rds” compo
s
l
e possible
s
representati
v
=
30 mm as
r
o
ns on Node
s
e
stigated rol
l
a
yer CLT of
N
compressio
n
e
side primar
y
shear at th
also she
a
grain in th
e
re observe
d
mean she
a
t
failures
o
e
in the rang
(2004) and
r
ing capacit
y
n
es of the
c
weaker of
a
l. (2008) d
i
o
mpression
i
with nodes
s
s
sful verific
a
n
figuration
fo
e
four-point
a
nsfer of sh
e
d
other repo
r
to mechani
s
and five la
y
f
these speci
a
ding points.
e
o
ar
P
e
t
n
g
o
l
l
Fig. 3
data regard
n
ly reported
2
). Other in
v
m
um test loa
d
LT against
s
h
e challenge
a
blish
b
earin
s
ing the syst
e
s
hear mech
a
v
e CLT dia
p
r
eference.
s
l
ing shear st
r
N
orway spr
u
n
y
e
ar
e
d
.
ar
of
e
Fig. 4:
CUAP (2
0
y
of single
n
c
ross sectio
n
both planes
i
d 20 tests
w
4
i
n the centr
a
s
tressed in
4
a
tion of the
p
fo
r determin
a
bending te
s
e
ar forces v
i
r
t that in al
m
sm
I, but rat
h
y
er CLT ele
m
i
mens failed
Some exce
p
: Test co
n
CUAP (
2
d
ing pure s
h
single test,
w
v
estigations
ds are in th
e
s
hear perpen
e
in generati
n
n
g models,
w
e
m CLT, it i
a
nisms. Bas
e
p
hragms, e.
g
t
rength and
s
u
ce. The set
u
Test confi
g
middle & r
i
0
05) an ad
a
n
odes (Jöbst
l
n
with wl x
s
determine
s
w
ith flat gra
i
a
l area of th
e
4
5° angle ca
n
p
roposed pr
o
a
tion of shea
r
s
t accordin
g
i
a the cross
m
ost all case
s
h
er
b
ending
f
m
ents (in tot
a
in shear per
p
p
tions failed
n
figurations C
L
2
006, right)
h
ear failures
w
hich failed
l
ead to fail
u
e
range of 6.
0
dicular to g
r
n
g failures
a
w
hich base o
n
s the aim to
e
d on them
i
g
. of 4 x 4 n
o
s
tiffness of
n
u
p was a sy
m
g
urations: Wa
l
i
ght)
a
pted test c
l
et al. 2008
;
tl = 200 x 1
0
the ultima
t
i
n
b
oard ma
t
e
panel (An
d
n
also
b
e fou
n
o
cedure by t
e
r
strength fo
r
to EN 408
.
layers and
t
s
not the int
e
f
ailures occ
u
a
l 90 speci
m
p
endicular t
o
in rolling sh
L
T elements:
T
in CLT el
e
in shear per
p
res others t
h
0
to 11.5 N/
m
r
ain is even
h
ccording to
n
strength a
n
d
efine relia
b
i
t is intende
o
des, five la
y
n
odes, in pa
r
m
metrical thr
e
l
lner (2004; l
e
o
nfiguration
;
Fig. 4). T
h
0
mm² at th
t
e load the
t
erial (Nor
w
d
reolli et al.
u
nd in Kreuz
i
e
sts is menti
r
European
T
.
A gap bet
w
t
he gluing i
n
e
nded failur
e
ur
. Jöbstl et
a
m
ens) tested
a
o
grain; nea
r
h
ear o
r
in sh
e
T
raetta et al. (
l
ements acc
o
p
endicular t
o
h
an in “ne
t
-
s
m
m² on ave
r
h
igher. Of c
o
mechanism
n
d stiffness
p
b
le strength
v
e
d to provid
e
a
yers and bo
a
a
rticular of t
h
e
e-point ben
e
ft), Jöbstl et
n
was deve
l
h
e setup pro
v
h
e vertical
g
test results
w
ay spruce)
w
2012). A
i
nger and
oned, the
T
echnical
w
een the
n
terfaces.
e
in shear
a
l. (2008)
a
ccording
r
ly 100 %
e
a
r
para-
2
006; left),
o
rding to
o
grain, is
s
hear”. In
r
age. This
o
urse, the
I in CLT
p
roperties
v
alues for
e
bearing
a
rds with
h
e gluing
ding tes
t
al. (2008;
oped for
v
ides two
g
aps with
are right
w
hich all
success
fu
fv,net,mean
=
fv,net,05 =
assumin
g
fv,net,mean,
M
In view
Bogens
p
quantify
Howeve
allow f
o
respect
t
differen
t
technica
single l
a
2013).
3
D
3.1
P
Based o
n
the fra
m
shear st
r
EN 408
forces
o
recomm
e
shear lo
a
data pro
c
gain the
shear an
shear is
expecte
d
Fig. 5:
Clarifyi
n
of three
for loa
d
(ρ12,mean
=
f
ully failed
=
12.8 N/m
m
11.1 N/mm²
g
a log
n
M
LE = 13.9
N
of current
p
erger et al.
2
y
the relevan
t
r
, the inter
a
o
r variation
o
t
o the thick
n
t
iation betw
e
a
l approvals
amellas are
D
evelop
m
P
rincipal
n
the succes
s
m
e of the M
a
r
ength of en
g
advanceme
n
o
f loading
e
ndations in
a
ds perpend
c
essing, e.g.
specimen
d
d compressi
o
somehow o
v
d
.
Configuration
a
tension or co
m
cross section (
r
n
g the possi
b
specimen e
a
d
ing in co
m
=
423 kg/m³
)
°
at the
m
², coeffici
e
. Applying
n
ormal di
s
N
/mm², CV[f
v
design pro
2
010) it is
m
t
resistance
r
a
ction relati
o
o
f test para
m
n
ess and wi
e
en flat and
of CLT av
a
(12) 20 to
4
m
ent an
d
Consider
a
s
fully prove
d
a
ster thesis o
g
ineered wo
o
n
ts of the s
e
and suppo
r
EN 408. In
icular to gr
a
by means o
d
irectly from
o
n perpendi
c
v
erestima
t
ed
.
a
nd geometric
p
m
pression (left
)
r
ight) (Hirsch
m
b
le influenc
e
a
ch were tes
t
m
pression
a
)
, respectiv
e
cross secti
e
nt of varia
t
maximum
s
tribution
v
,net,MLE] = 13
c
edures, w
h
m
andatory to
r
eliable. In
f
o
nship has n
m
eters in a r
a
dth of com
m
rif
t
grain
bo
ilable in Eu
r
4
0 (45) mm
d
Verific
a
a
tions
d
test setup
o
f Hirschma
n
o
d products
tup of Jöbs
t
r
t are in-li
n
contrast to
t
a
in is provi
d
f
MLE for r
i
full-size C
L
c
ular to grai
n
.
However,
b
p
arameters for
)
; test loaded i
n
m
ann 2011; ada
p
e
of loading
i
t
ed with cor
e
a
nd tension
e
ly. The h
y
5
ons. The
a
tion CV[fv,n
e
likelihood
e
with fv,net
.5 % and fv,n
e
h
ich verify
t
o
define a te
s
f
act, interact
i
n
ot been qu
a
a
nge at least
m
only used
o
ards. Base
d
u
rope the co
m
and (40) 1
a
tion of
a
o
f Jöbstl et
a
n
n (2011).
C
in EN 789
a
t
l et al. (20
0
n
e. The te
s
t
he setup of
d
ed. This all
o
i
ght censore
d
L
T elements
.
n
in the cros
b
ecause of t
h
testing shear p
e
n
compression
,
p
te
d
)
i
n tension or
e
layers wl
x
are 9.6 N
/
y
pothesis of
main stat
i
e
t] = 11.3 %
e
stimation
(
~ 2pLND,
e
t,05,MLE = 11
.
t
he shear r
e
s
t procedure
i
on of mech
a
a
ntified until
relevant fo
r
boards and
d
on a com
p
m
mon rang
e
00 to 240 (3
0
a
Test C
o
a
l. (2008) fu
r
C
onsidering
t
a
nd the shea
r
0
8) were ma
s
t specime
n
Jöbstl et al.
o
ws direct
u
d
data. Anot
h
.
A general
d
s layer. Con
h
e small ang
l
e
rpendicular to
,
including me
compressio
n
x
tl = 150 x 1
0
/
mm² (ρ12,m
e
equal med
i
i
stics gain
e
and the e
m
(
MLE) for
r
the ada
p
0 N/mm².
e
sistance o
n
for mechan
i
a
nis
m
I and
I
now. The
t
r
the practic
a
the annual
r
p
rehensive c
o
e
s in thickne
0
0) mm, re
s
o
nfigura
t
r
ther develo
p
h
e test conf
i
r
configurati
o
d
e, see Fig.
is rotated
(2008) only
u
se of test r
e
h
er advanta
g
d
isadvantag
e
s
equently, t
h
l
e of 14° onl
y
grain on singl
e
a
surement of
d
n
on the she
a
0
mm². The
e
an = 427 kg/
m
i
ans cannot
e
d from
t
m
pirical 5
%
right censo
r
p
ted statis
t
n
a single
n
i
s
m
I which
II cannot be
t
est proced
u
a
l use of CL
ring pattern
o
mparison
o
e
ss tl and wi
s
pectively (
t
ion
p
ments wer
e
fi
guration fo
r
o
n for solid
5. In brief:
d
14°, equ
a
one failure
e
sults witho
u
g
e is the pos
s
e
is the inte
r
h
e bearing c
a
l
y a small in
f
e
CLT nodes b
y
d
eformation a
n
a
r capacity t
w
mean shear
/
m³) and 9
.
b
e rejecte
d
ests are
%
-quantile
r
ed data,
t
ics are
n
ode (see
allows to
avoided.
re has to
T, e.g. in
,
e.g. the
o
f cu
r
rent
dth wl of
Brandner
e
made in
r
in plane
timber in
resultant
a
l to the
plane for
u
t further
s
ibility to
r
action of
a
pacity in
f
luence is
y
loading in
n
d fractured
w
o series
strengths
.
8 N/mm²
d
(Mann-
6
Whitney test; p = 0.7). For convenience in test preparation and execution, the main series were tested
in compression.
The aims of Hirschmann (2008) were (i) to investigate the applicability of the setup, (ii) to compare
the results with that gained from the setup of Jöbstl et al. (2008), and (iii) to analyse the influences of
selected geometric and material parameters on the shear perpendicular to grain resistance.
For clarification, the test series according to the setup of Jöbstl et al. (2008) are further given as “CIB”
and that according to Hirschmann (2011) as “EN”.
3.2 Material and Methods
The test material was Norway spruce (Picea abies) of nominal strength class C24 according to
EN 408. All material was classified according to the density. Thus, “matched samples” for series
“CIB” and “EN” were created. The material was conditioned at 20 °C and 65 % relative humidity to
reach an expected average moisture content of u = 12 %. Ten tests per series were executed. In the
reference test series “C” the core boards were flat grained (fgB) of wl x tl = 150 x 20 mm² and with
gaps of tgap = 5 mm. In all tests, top layers with 40 mm thickness were used. Following variations of
parameters were made (see also Tab. 1):
width wl: 150, 200 mm;
thickness tl: 10, 20, 30 mm;
annual ring orientation (AR): flat grain boards (fgB), rift grain boards (rgB) and heart boards
(hB);
gap width tgap: 1.5, 5.0, 25.0 mm.
As no rift grain boards were available, “pseudo rift grain boards” were produced by trimming out the
heart of heart boards and edge gluing of the residual parts.
The geometry of the test setups “CIB” and “EN” was planned to resist (i) compression at loading and
support, (ii) torsion in the gluing interface, and (iii) rolling shear in the gluing interface until failing in
shear perpendicular to grain in the net cross section of the core layer. A compilation can be found in
Hirschmann (2011). The test segments in the core were taken consecutively from 4 m long boards
with the aim to assure regions free of growth characteristics like knots, checks and reaction wood in
the expected failure zone of the specimen. Consequently, in tested series one to five specimens are
from the same board.
The tests were executed way controlled. The velocity was adapted to ensure an average time until
ultimate load of 300 ± 120 s.
3.3 Test Results
A summary of tested parameters and of main statistics is provided in Tab. 1. All executed tests in
series “CIB” and “EN” failed in the expected plane due to shear perpendicular to grain. Classification
according to density was successful comparing the series with equal parameters of “CIB” and “EN”.
However, series “G”, “H” and “I” of both setups show significant higher densities. For the test results
of “CIB” a MLE for right censored data, as in chapter 2.2.2, was executed.
Although mean and median shear strengths at equal parameter settings in series “CIB” are always
higher than in series “EN” (on average + 0.5 N/mm²), the hypothesis of equal medians cannot be
rejected in five of seven paired groups (Mann-Whitney test, p > 0.05), beside of series “C” and “I”.
The reasons for systematically higher shear strengths in “CIB” are seen in the load path. Whereas
setup “EN” provides resulting forces of loading and support in-line, in “CIB” the cross layer is
additionally stressed in bending. Furthermore, it can be assumed that the load path in “CIB” in
proportion to the shear stress leads to higher compression perpendicular to grain stresses. However,
due to the moment also an interaction of tension perpendicular to grain and shear is given. Following
the work of Spengler (1982) the higher compression stresses in “CIB” in comparison to “EN” are seen
as reason for the roughly 5 % higher shear strengths in “CIB”. A possible stiffening of the
compression zone attracts additional loads. It is concluded that both configurations provide
comparable test values. As the uncertainty in statistical inference in series “CIB” is higher and the load
7
path more complex, the test setup “EN” is preferred. Hirschmann (2011) also shows that in
comparison to “CIB” the setup “EN” allows testing of a wider range in examined parameters.
Although the material quality and parameter settings are comparable, the mean and dispersion of fv,net
in series “CIB_A” are significantly lower than in Jöbstl et al. (2008). In fact, in all series of
Hirschmann (2011) an unexpected low coefficient of variation is observed. One reason is caused by
the test preparation, whereby more than one specimen per series origin from the same board.
Considering the hierarchical material structure of timber, in case of a second order hierarchical model
with differentiation in variation within and between board properties, it is concluded that the results
are somehow biased. As the assignment of test specimen to former boards is possible, estimates for
coefficients of variation of fv,net within and between boards are 3.0 % and 3.6 %, respectively.
Following Källsner et al. (1997) an equicorrelation coefficient, as measure for the correlation of fv,net
within boards, can be estimated as ρequi ≈ 0.59. This equicorrelation is higher than found on average
for other strength properties (Brandner 2012). This is argued by the restriction of test material
regarding growth characteristics and by a strict classification in density. The coefficient of variation of
density CV[ρ12] is in the range of 2 % to 8 % (on average 4 %). However, the expected mean range is
6 % to 8 %; thus, the test material is very homogeneous. For material commonly used in timber
engineering a higher variation in shear strength than the herein observed range in “EN” test series of
CV[fv,net] = (5 to 10) % is expected. In view of the experiences reported in Jöbstl et al. (2008) a range
of CV[fv,net] = (12 to 15) % and a lower equicorrelation appears reasonable.
Tab. 1: Test parameters and main statistics of density and shear strength at 12 % moisture content according to
Hirschmann (2011); results partly adapted and reassessed
EN CIB
A B C D F G H I A B C F G H I
base p.
[-]
wl [mm] 200 150 200 150
tl [mm] 10 20 30 20 10 20
AR [-] 1
)
fgB rgB hB fgB fgB rgB hB fgB
t
g
a
p
[mm] 5.0 1.5 25.0 5.0 1.5 25.0
ρ12
[kg/m³]
quantity [-] à 10 à 10
mean 396 401 399 395 397 443 413 419 405 400 397 398 435 424 439
median 396 404 400 400 395 444 413 416 405 396 407 397 432 427 453
CV [%] 1.8 4.4 2.6 3.7 3.3 1.9 8.4 7.0 3.0 4.2 5.6 6.9 2.4 1.8 7.0
fv,net,12
[N/mm²]
min 10.0 10.2 8.4 6.4 6.3 8.2 8.5 7.1 – – – – – – –
mean 10.8 11.2 8.9 7.5 7.2 8.8 9.5 8.0
11.1 2
)
11.7 2
)
9.4 2
)
8.0 2
)
9.2 2
)
9.8 2
)
8.8 2
)
median 10.8 11.2 8.7 7.4 7.4 8.9 9.3 8.1
11.0 2
)
11.7 2
)
9.4 2
)
7.9 2
)
9.2 2
)
9.8 2
)
8.7 2
)
max 12.1 12.4 9.6 8.4 8.0 9.4 10.6 8.6 – – – – – – –
CV [%] 6.0 6.3 4.9 9.3 10.1 4.2 8.5 5.6 7.4
2
)
6.9 2
)
7.4 2
)
15.1 2
)
7.4
2
)
5.2
2
)
7.9 2
)
5 %-qu. 10.1 3
)
10.3 3
)
8.5
3
)
6.7
3
)
6.3
3
)
8.3 3
)
8.5 3
)
7.2 3
)
9.8 2
)
10.4 2
)
8.3 2
)
6.2 2
)
8.2
2
)
9.0
2
)
7.7 2
)
1) AR … annual ring orientation | fgB … flat grain boards | rgB … “pseudo” rift grain boards | hB … heart boards
2) statistics estimated by means of Maximum Likelihood Estimation (MLE) for right censored data, assuming f
v
,
net ~ 2pLND
3) empirical 5 %-quantile, gained from rank statistics
3.4 Shear Perpendicular to Grain: Load-Displacement and Failure Behaviour
Both setups, “EN” and “CIB”, show similar characteristic load-displacement behaviour, see Fig. 6
(left). The load-displacement curve can be divided in two main parts: the first part showing a roughly
linear course until the ultimate load Fmax is reached, and the second part a clear softening property,
where failure due to a new shear mechanism can be observed (see also Fig. 5, right).
In the first part, after some hardening until approximately 20 % of Fmax, a linear elastic material
behaviour within approximately 0.2 · Fmax to 0.8 · Fmax is given, followed by a regressive non-linear
relationship until Fmax. At this point, a combined failure of shear mechanisms I “net-shear” and II
“torsion” takes place, initiated by local exceeded resistance in opposite corners of the failure plane, at
the zones of interacting shear and tension perpendicular to grain. In the second part after the peak load,
softening is characterised by reaching a steady state at about 40 % to 50 % of Fmax, enabling large
deforma
early- a
n
This lea
d
(see als
o
Fig. 6:
The co
m
displace
m
implem
e
followi
n
mode I
a
has bee
n
paramet
e
To acco
u
must be
is show
n
(fgB), “
E
for mec
h
annual
r
numeric
a
also the
The seq
u
also be
enginee
r
under
t
lateral
s
simple
p
b
eams i
s
simplici
t
b
y a fi
x
length
f
where
a
acts an
d
Followi
n
t
It can b
e
b
ending
.
In that c
cantilev
e
a
tions. These
n
d latewood
.
d
s to a flexi
b
o
Jöbstl et al.
Typical load-d
i
cohesive elem
e
numerical resu
m
plexity of t
h
m
ent curve
e
nted in A
B
n
g the Dugd
a
a
nd shear sl
i
n
achieved.
T
e
rs of the nu
m
u
nt for both
implemente
d
n
in Fig. 6
(
E
N_G” (hB
)
h
anis
m
I (a
b
r
ing orienta
t
a
l study cle
a
non-linear l
o
u
ence in th
e
explained
b
r
ing model,
s
t
he circums
s
upport by
p
lanar mod
e
s
considere
d
t
y, this fixe
d
x
ed cantile
v
f
rom left s
u
a
n asymmet
r
d
shear for
c
n
g assumpti
o
T
is the tota
l
t
he elastic b
e
e
easily sho
w
.
After crac
k
ase a linear
i
e
r in bendin
g
deformatio
n
.
It follows
a
b
le composi
t
2008).
i
splacement be
e
nts in the nu
m
lts on characte
r
h
e failure be
h
in a satisf
y
B
AQUS. T
h
a
le-Barenbl
a
i
ding in mo
d
T
hus, chara
c
m
erical mo
d
shear mech
d
in both fai
l
(
right) toge
t
)
and “EN_
F
b
aqus I), an
d
t
ion was n
o
a
rly outline,
o
ad-displace
m
e
fracturing
p
b
y means
o
s
ee Fig. 7.
T
tance of
a
orthogonal
e
l of slende
r
d
in the she
a
d
-end beam
v
er beam
w
u
pport to t
h
r
ic boundar
y
c
e can be
o
ns are mad
e
l
shear force
e
haviour do
m
w
n that the
e
k
ing shear fo
r
i
ncreasing b
e
g
and tensio
n
n
s increase
s
a
successive
t
e of fixed-
e
h
avior exempl
a
m
erical model
r
istic (5 %-qua
n
h
aviour mot
i
y
ing manne
r
h
ese elemen
t
a
tt model of
d
e II and III
c
teristic (5
%
d
el, see Feic
h
anisms, me
c
l
ure regions,
h
er with th
e
F
” (rgB). N
u
d
(ii) cohesi
v
t considere
d
that at Fmax
m
ent behavi
o
p
rocess can
o
f a simple
T
hereby and
a
supposed
boards, a
r
fixed-end
a
r area. Fo
r
is replaced
w
ith half in
h
e middle,
y
condition
introduced.
e
:
to be trans
m
m
inates at b
e
e
lastic soluti
o
r
ces can onl
y
e
nding mo
m
n
.
8
s
hearing par
a
dissolution
e
nd beams,
a
ar
ily for series
(middle); co
m
n
tile) level (rig
h
ivated a nu
m
r
. Therefor
e
ts allow fr
a
elastic-plas
t
according t
o
%
-quantile) s
t
h
ter (2013).
c
hanism I “
n
,
see Fig. 6 (
m
e
average lo
u
merical res
u
v
e elements
d
in the nu
m
x
b
oth mech
a
our before
F
Fig. 7:
S
m
itted;
e
ginning (a);
o
n (a) trans
m
y
be transfer
r
m
ent develop
s
a
llel to grai
n
of the mate
r
a
ctive in ben
“EN_C”: singl
e
m
parison of av
e
h
t)
m
erical mod
e
e
, a FE-mo
d
a
cturing in
t
ic fracture
m
o
this theory
t
rength valu
e
n
e
t
-shear” a
n
m
iddle). Th
e
ad-displace
m
u
lts are sho
w
for mechan
i
m
erical mo
d
a
nis
m
s, I an
d
F
max, and the
s
S
imple enginee
r
after fractu
r
m
its shear fo
r
ed via the c
r
s
and conse
q
n
, mainly in
t
r
ial by separ
a
d
ing and te
n
e
and average
c
e
rage loa
d
-disp
e
l with the a
i
d
el with co
h
the observ
e
m
echanics.
S
occur after
e
s and fract
u
d II “torsio
n
e
outcome o
f
m
ent curves
w
n for (i) co
h
i
s
m
I and II
d
el. Howev
e
d
II, take pl
a
s
oftening af
t
r
ing model
e the loadin
g
r
ces as a pu
r
r
oss section
s
q
uences a su
c
t
he transitio
n
a
tion of ann
u
n
sion paralle
c
urves (left); p
l
p
lacement curv
e
i
m to mirror
h
esive ele
m
e
d failure
p
S
eparation i
n
a critical st
r
u
re energies
n
”, cohesive
f
the numeri
c
from series
o
hesive elem
(abaqus I
&
e
r, the resu
l
ace and co
n
t
erwards.
g
changes to
r
e shear fiel
d
s
of the cant
i
c
cessive fail
u
n
zone of
u
al rings.
l to grain
l
acement of
e
s with the
the load-
ents was
p
lanes by
n
fracture
r
ess value
are input
elements
c
al model
“EN_C”
ents only
&
II). The
ts of the
n
sequence
(b).
d
without
i
lever (b).
u
re of the
9
In brief: failure at Fmax due to shear forces perpendicular to grain is caused by exceeding the local
resistance of interacting mechanism I and II. The numerical model verifies this. Further a softening to
a steady state at approximately 40 % to 50 % of Fmax is given. A successive dissolution of the shear
fracture zone, by increasing shearing parallel to grain at the transition zone of early- and latewood and
separation of the annual rings, occurs. A flexible composite of fixed-end beams becomes active in
tension and bending. This is the cause for the high residual forces. A simple engineering model
demonstrated this sequence of fracturing. However, there is no doubt that the shear forces applied
perpendicular to grain lead to shearing parallel to grain. Consequently, the shear capacities and the
shear behaviour parallel to grain, in reference to a relatively small shear area and volume, indicate the
shear resistance perpendicular to grain.
3.5 Main influencing Parameters
In the following the investigated parameters (i) annual ring orientation, (ii) layer width, (iii) layer
thickness, and (iv) gap width are discussed individually regarding a possible influence on the shear
capacity perpendicular to grain. For statistical inference the Mann-Whitney test was used for testing
the hypothesis of pairwise equal medians. This was done although a symmetric distribution is not
realised in all series. Box-plots of all results of setup “EN” together with median values of setup
“CIB” are provided in Fig. 8.
Fig. 8: Box-plot of shear strength fv,net,12 of setup “EN” vs. parameter variations; median values of setup “CIB” included
3.5.1 Annual Ring Orientation
Investigating the influence of AR, the following series were tested: series “C” comprising flat grain
boards (fgB), series “F” with “pseudo” rift grain boards (rgB) and series “G” with heart boards (hB).
The parameters width wl = 150 mm, thickness tl = 20 mm and gap width tgap = 5 mm were kept
constant. The average densities of “fgb” and “rgB” are well comparable whereas both series “EN_G”
and “CIB_G” (hB) show significantly higher densities (mean difference 30 to 40 kg/m³). The results
are presented in Fig. 8.
As shear loads perpendicular to grain lead to failures in shear parallel to grain (see chapter 3.4) there is
evidence for influences caused by the parameter “annual ring orientation”. Keenan et al. (1985),
Denzler and Glos (2007), Dahl and Malo (2009) and Brandner et al. (2012) found significant higher
shear strength (on average 6 % to 40 %) in RL (radial-longitudinal) in comparison to TL direction
(tangential-longitudinal), Müller et al. (2004) not. In TL shearing occurs in the transition zone of
early- und latewood. In RL, shearing requires fracturing of early- and latewood. Consequently, a
higher resistance and a positive dependency of fv,RL on specimen’s global density are expected. Thus,
flat grain boards, in comparison to rift grain and heart boards, have commonly a higher resistance in
shear. The annual ring orientation in heart boards may comprise both, shearing in RL in the core and
in TL at the edges. In dependency of the width of the core lamella a resistance in-between flat and rift
grain boards is expected.
Statistical inference confirms the expectations regarding significant lower shear strengths in rift grain
boards in comparison to flat grain boards (p < 0.01). Also between series “rgB” and “hB” significant
differences in the medians are observed (p < 0.01). Some impact of the significant higher density in
shear strength f
v,net,12
[N/mm²]
12
11
10
9
8
7
6
10 20 30
thickness t
l
[mm]
150 200
width w
l
[mm]
1.5 5.0 25.0
gap t
gap
[mm]
fgB rgB hB
AR [–]
f
v,net,50,MLE
| setup „CIB“
f
v,net,05
| setup „EN“
„C“
„B“ „D“ „C“ „C“„B“ „A“ „H“ „I“ „F“ „G“
10
series “hB” cannot be excluded. As flat grain boards are commonly used in CLT production, relatively
high shear resistances can be realised. However, in both setups, “EN” and “CIB”, an interaction of
shear and compression perpendicular to grain occurs. Keenan (1973, 1974) observed that fv,RL is much
more influenced in case of interaction with σc,90 than fv,TL.
3.5.2 Layer Width
A comparison is made between wl = 150 mm (series “A”) and 200 mm (series “B”) wide boards. This
corresponds to a ratio of 1 : 1.33. The parameters tl = 10 mm, tgap = 5 mm and AR = “fgB” were kept
constant. The average densities of all series are in-line. Comparison shows that the hypothesis of equal
medians cannot be rejected (p > 0.05). This is also obvious considering the comparable ranges of
realisations in series “A” and “B”, see Fig. 8. However, as the range in commonly used board widths
(100 mm ≤ wl ≤ 240 mm) is much larger than tested some relevant influence on the shear resistance
cannot be excluded, in particular in wide boards were shearing at the edges more and more occurs in
TL, known to realise lower shear resistances (see chapter 3.5.1).
3.5.3 Layer Thickness
Test series “B” (tl = 10 mm), “C” (tl = 20 mm) and “D” (tl = 30 mm) were conducted for examining
the influence of layer thickness. The parameters wl = 150 mm, tgap = 5 mm and AR = “fgB” were kept
constant. The average densities of all series are in-line. The results are visualised in Fig. 8. In both test
setups “EN” and “CIB” and in all pairwise comparisons the hypothesis of equal medians was rejected
(p < 0.01). Two main reasons are identified: at first, the impact of size on shear strength parallel to
grain is well known and documented, e.g. in Brandner et al. (2012). They report on a regressive course
of shear strength with increasing shear area As. Secondly, load transfer from top layers to the core
layer via the gluing interfaces causes a locking effect. This locking effect, which restrains the shear
action, is at highest in the gluing interface and declines until the centre of the core lamella.
Fig. 9 contains a comparison of the size effect on shea
r
strength parallel to grain for construction timber, based on
a
literature survey and tests reported in Brandner et al. (2012),
some additional data sets for clear wood and the results
found for setup “EN”. The plot shows the shear strength
versus the shear area As. Deviating from the definition of As
in Brandner et al. (2012), for the herein presented test setup
and results As is defined by the cross section of the core
lamella, with As = wl · tl. Overall, good congruence is found.
The steeper regressive course in fv,net,mean vs. the shear area As
is dedicated to the locking effect. In view of the tendency to
standard lamella thicknesses tl = 20, 30, 40 mm an
extrapolation for 40 mm thick lamellas is required. Fig. 9: Size effect on mean shear strength
3.5.4 Gap Width
The influence of gap width on shear strength was analysed for tgap = 1.5 mm (series “H”), 5.0 mm
(series “C”) and 25.0 mm (series “I”). The parameters wl = 150 mm, tl = 20 mm and AR = “fgB” were
kept constant. The average densities of series “H” and “I” are well comparable whereas both series
“EN_C” and “CIB_C” (tgap = 5.0 mm) show significantly lower densities (mean differences of 15 to
20 kg/m³ in “EN” and 25 to 40 kg/m³ in “CIB”). Because of the dependency of fv,RL on the density in
softwood, an influence on fv,net cannot be excluded. The results are shown in Fig. 8.
As a quantitative correction of the differences in density is not available statistical inference is made
on observed pairwise median shear strengths. Significant differences in medians are found between
tgap = 1.5 mm and 5.0 mm and 5.0 mm and 25.0 mm in setup “CIB” (p < 0.05). High significant
differences (p < 0.01) are identified between tgap = 1.5 mm and 25.0 mm in both setups “EN” and
“CIB” and between tgap = 5.0 mm and 25.0 mm in “EN”. However, the hypothesis of equal medians
cannot be rejected comparing series with tgap = 1.5 mm and 5.0 mm in setup “EN” (p = 0.10).
shear strength f
v,12
[N/mm²]
10
1
100 1,000 10,000
shear area A
s
[mm²]
Brandner et al. (2012) | literature survey
Brandner et al. (2012) | test data
Bröker et al. (1987) | block shear; clear wood
Gaspar et al. (2008) | block shear; clear wood
Hirschmann (2011) | setup „EN“
11
In general, a decrease in the resistance with increasing gap width is expected. This is because of a
reduced influence of the locking effect as well as by increasing bending stresses in the gap. Thus, a
regressive course of shear strength versus gap width is expected.
4 Resistance in Shear Loads perpendicular to Grain: Proposal
In chapter 3 the resistance against shear loads perpendicular to grain was demonstrated and relevant
influencing parameters identified. The interaction of shear and compression perpendicular to grain,
which leads to some overestimation of the real shear resistance, was mentioned. However, at the
ultimate load interaction of shear mechanisms I “net-shear” and II “torsion” may counteract the shear-
compression interaction. In view of the material commonly used for CLT in Europe, flat grain boards
with cross section wl x tl = 150 x 30 mm² and a gap width of tgap = 5 mm (as upper boundary) are
defined as reference. Furthermore, a lognormal distribution (fv,net ~ 2pLND) and a coefficient of
variation CV[fv,net] = 15 % are assumed. Based on fv,net,12,mean = 7.5 N/mm² in series “EN_D” the
characteristic (5 %-quantile) shear strength is fv,net,05 = 5.8 N/mm². In case of lamellas with tl = 40 mm,
as the upper boundary of commonly used raw material, a value of fv,net,05 = 5.3 N/mm² is found by
extrapolating the power regression model, based on mean values of series “EN_B”, “EN_C” and
“EN_D”. However, these strength values are gained from examinations made on single nodes of a
three layer CLT element. The question remains if the verification of shear in plane, currently done on
single nodes and RVSEs, is representative for a whole CLT diaphragm. As demonstrated in chapter 2
this question cannot be answered yet, but an engineering judgement can be made.
In view of the bearing model for CLT in bending out of plane, we define a reference CLT diaphragm
of 4 x 4 nodes and of five layers, each composed of board material in reference dimension. Assuming
a shear load, homogeneously applied on the cross sections of this diaphragm, in total two times the
tested node in thickness direction are found to act in parallel. Due to allocated shear stresses, a failure
of the diaphragm in plane according to “net-shear” can only take place in cases where all nodes in x-
direction (direction of the top layers) fail. Again, a parallel system action, active in y-direction
(direction of the cross layers) of the diaphragm, can be identified. Of course, in a 4 x 4 element this
kind of shearing can occur on three planes, whereby the weakest plane governs the ultimate load. This
confirms to a serial system action, active in x-direction of the diaphragm. Considering the load-
displacement curve of shear perpendicular to grain, a non-linear behaviour, already before reaching the
ultimate load, and the ability to withstand large deformations on a moderate load level after softening
is found. Taking into account the remarkable possibility to transfer loads between the parallel active
nodes, there is evidence that the mean resistance of the diaphragm in shear will not be remarkable
different from the mean shear resistance of single nodes. However, because of the parallel system
action of 2 x 4 nodes a significant reduction in dispersion of fv,net is expected. On the one hand this
circumstance reduces the influence of serial system action between the shear planes, and on the other
hand it offers the possibility of rising fv,net,05.
Although a theoretical and practical verification is not available yet the current procedure of verifying
in plane shear resistance on single nodes (see e.g. Bogensperger et al. 2010) is judged as reliable and
proposed in the meantime until further progress is made. For simplicity a characteristic (5 %-quantile)
shear strength of fv,net = 5.5 N/mm² is proposed for all lamella thicknesses tl ≤ 40 mm.
5 Conclusions and Outlook
We presented a test configuration, which allows determining the resistance in shear perpendicular to
grain on single CLT nodes. Relevant parameters were investigated and their influence on shear
strength fv,net quantified. Thereby, the parameters (i) thickness of the core lamella tl, (ii) the annual ring
orientation AR, and (iii) the gap width tgap were found to affect the shear strength significantly.
Additional to testing, the load-displacement behaviour and in particular the failure process were
studied by means of a numerical and a simple engineering model. The interaction of both shear
mechanisms, mechanism I “net-shear” and II “torsion”, the fracturing in shear parallel to grain and
12
successive dissolution of the material was verified. Analogies to shear resistance parallel to grain of
structural timber were identified, in particular regarding the size effect.
Based on engineering judgement the shear resistance according to mechanism I was discussed for a
whole CLT diaphragm. In conclusion, a characteristic (5 %-quantile) value of fv,net,05 = 5.5 N/mm² for
common flat grain board material of Norway spruce with
tl ≤ 40 mm and tgap ≤ 5 mm, and the verification of shear in
p
lane on single nodes or RVSEs, including both, the
verification of mechanism I and II, is proposed.
Current investigations are made on a hardening property afte
r
softening and on the shear resistance at tgap = 0. Fig. 10
illustrates first results of flat and rift grain boards at tgap = 5
and 0 mm. Although and not to the full extend relevant for the
shear behaviour of a whole CLT diaphragm, a tremendous
ability to large deformations at a steady state on a relatively
high load level, followed by a hardening which exceeds
mostly the first, currently evaluated peak level, is observed.
Further tests and investigations on whole CLT elements are
scheduled.
Fig. 10: Single test results on specific
parameter settings
6 Acknowledgement
The work and progress made by Hirschmann (2011) and Feichter (2013), as part of their Master
theses, is thankfully acknowledged. Thanks also to Prof. Roberto Tomasi from University of Trento /
Italy for discussing several aspects concerning the tests published in Andreolli et al. (2012).
7 References
Andreolli M, Tomasi R, Polastri A (2012) Experimental investigation on in-plane behaviour of cross-
laminated-timber elements. Paper and Presentation, CIB-W18/45-12-4, Växjö, Sweden
Blaß H J, Görlacher R (2002) Zum Trag- und Verformungsverhalten von Brettsperrholz-Elementen bei
Beanspruchung in Plattenebene. Bauen mit Holz, 11:34–41; 12:30–34 (in German)
Blaß H J, Flaig M (2012) Stabförmige Bauteile aus Brettsperrholz. Karlsruher Berichte zum
Ingenieurholzbau, No. 24, Karslruher Institut für Technologie (KIT), Lehrstuhl für Ingenieurholzbau
und Baukonstruktionen, ISBN 978-3-86644-922-0 (in German)
Bogensperger T, Moosbrugger T, Schickhofer G (2007) New test configuration for CLT-wall-elements under
shear load. CIB-W18/40-21-2, Bled, Slovenia
Bogensperger T, Moosbrugger T, Silly G (2010) Verification of CLT-plates under loads in plane.
WCTE’2010, Riva del Garda, Italy
Bosl R (2002) Zum Nachweis des Trag- und Verformungsverhaltens von Wandscheiben aus
Brettlagenholz. Dissertation, Universität der Bundeswehr München, Fakultät für Bauingenieur- und
Vermessungswesen (in German)
Brandner R (2012) Stochastic system actions and effects in engineered timber products and structures.
Dissertation, Institute of Timber Engineering and Wood Technology, Graz University of Technology
Brandner R, Gatternig W, Schickhofer G (2012) Determination of Shear Strength of Structural and Glued
Laminated Timber. CIB-W18/45-12-2, Växjö, Sweden
Brandner R (2013) Production and Technology of Cross Laminated Timber (CLT): A state-of-the-art
Report. In: Harris R, Ringhofer A, Schickhofer (ed.) Focus Solid Timber Solutions – European
Conference on Cross Laminated Timber (CLT), 21st–22nd May 2013, Graz, Austria
Bröker F W, Meierhofer U A, Radović B (1987) Einfluss der Probengröße auf die Druckscherfestigkeiten
würfelförmiger Fichtenholzproben. Holz als Roh- und Werkstoff, 45:445–448 (in German)
CUAP Common Understanding of Assessment Procedure (2005) Solid wood slab element to be used as a
structural element in buildings. ETA request No 03.04/06, Österreichisches Institut für Bautechnik
(OIB), Vienna, Austria
global load F [kN]
50
45
40
35
30
25
20
15
10
5
0
0 5 10 15 20 25 30 35 40 45 50
global deformation [mm]
fgB, 10/150, t
gap
=5
fgB, 10/150, t
gap
=5
rgB, 10/150, t
gap
=5
rgB, 10/150, t
gap
=0
13
Dahl K B, Malo K A (2009) Linear shear properties of spruce softwood. Wood Science and Technology,
43:499–525
Denzler J K, Glos P (2007) Determination of shear strength values according to EN 408. Materials and
Structures, 40:79–86
EN 338:2009-10 Structural timber – Strength classes. European Standard, European Committee for
standardization
EN 408:2012-07 Timber structures – Structural timber and glued laminated timber – Determination of
some physical and mechanical properties. European Standard, European Committee for
standardization
EN 789:2004-10 Timber structures – Test methods – Determination of mechanical properties of wood
based panels. European Standard, European Committee for standardization
Feichter I (2013) Spannungs- und Traglastberechnungen an ausgewählten Problemen der Holz-
Massivbauweise in Brettsperrholz. Master thesis, Institute of Timber Engineering and Wood
Technology, Graz University of Technology (in German)
Flaig M, Blaß H J (2013) Shear strength and shear stiffness of CLT-beams loaded in plane. CIB-W18/46-
12-3, Vancouver, Canada
Gaspar F, Cruz H, Gomes A (2008) Evaluation of glued laminated timber structures – core extraction and
shear testing. World Conference on Timber Engineering (10th WCTE), Miyazaki, Japan
Hirschmann B (2011) Ein Beitrag zur Bestimmung der Scheibenschubfestigkeit von Brettsperrholz. Master
thesis, Institute of Timber Engineering and Wood Technology, Graz University of Technology (in
German)
Jeitler G (2004) Versuchstechnische Ermittlung der Verdrehungskenngrößen von orthogonal verklebten
Brettlamellen. Diploma thesis, Institute of Steel, Timber and Shell Structures, Graz University of
Technology (in German)
Jöbstl R A, Bogensperger T, Schickhofer G, Jeitler G (2004) Mechanical Behaviour of Two Orthogonally
Glued Boards. WCTE’2004, Lahti, Finland
Jöbstl R A, Bogensperger T, Schickhofer G (2008) In-plane shear strength of cross laminated timber. CIB-
W18/41-12-3, St. Andrews, Canada
Källsner B, Ditlevsen O, Salmela K (1997) Experimental verification of a weak zone model for timber in
bending. IUFRO S 5.02 – Timber Engineering, Copenhagen, Denmark
Keenan F J (1973) The shear strength of glued-laminated timber. University of Toronto, Department of Civil
Engineering
Keenan F J (1974) Shear strength of wood beams. Forest Products Journal, 24(9):63–70
Keenan F J, Kryla J, Kykong B (1985) Shear strength of spruce glued-laminated timber beams. Canadian
Journal of Civil Engineering, 12:661–672
Kreuzinger H, Sieder M (2013) Einfaches Prüfverfahren zur Bewertung der Schubfestigkeit von
Kreuzlagenholz / Brettsperrholz. Bautechnik, 90(5):314–316 (in German)
Müller K, Sretenovic A, Gindl W, Grabner M, Wimmer R, Teischinger A (2004) Effects on macro- and micro-
structural variability on the shear behaviour of softwood. IAWA Journal, 25(2):231–243
Spengler R (1982) Festigkeitsverhalten von Brettschichtholz unter zweiachsiger Beanspruchung, Teil 1 –
Ermittlung des Festigkeitsverhaltens von Brettelementen aus Fichte durch Versuche. Berichte zur
Zuverlässigkeitstheorie der Bauwerke, Heft 62, Laboratorium für den konstruktiven Ingenieurbau,
Technische Universität München (in German)
Traetta G, Bogensperger T, Moosbrugger T, Schickhofer G (2006) Verformungsverhalten von
Brettsperrholzplatten unter Schubbeanspruchung in der Ebene. In: Moosbrugger M, Schickhofer
G, Unterwieser H, Krenn H (ed.) Brettsperrholz – Ein Blick auf Forschung und Entwicklung. 5. Grazer
Holzbau-Fachtagung (5.GraHFT’06), 29th Sept. 2006, Graz, Austria (in German)
Wallner G (2004) Versuchstechnische Ermittlung der Verschiebungskenngrößen von orthogonal
verklebten Brettlamellen. Diploma thesis, Institute of Steel, Timber and Shell Structures, Graz
University of Technology (in German)
