Engineering Structures xxx (xxxx) xxx–xxx

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Cross laminated timber at in-plane beam loading – Prediction of shear

stresses in crossing areas
⁎
Henrik Danielsson , Erik Serrano
Div of Structural Mechanics, Lund University, P.O. Box 118, SE-221 00 Lund, Sweden

A R T I C LE I N FO A B S T R A C T

Keywords: Cross Laminated Timber (CLT) at in-plane beam loading conditions present a very complex stress state and many
CLT failure modes need to be considered in design. The work presented here aims at finding improvements of a
Beam specific analytical model for stress analysis and strength verification that has been suggested in literature and
In-plane loading which is also suggested as a basis for design equations for the next version of Eurocode 5. Although the model
Shear
has appealing properties it suffers from some drawbacks related to the assumed distributions of internal forces
Crossing area
which, based on comparison to finite element analysis, appear to be inaccurate. The main focus in this paper is
on model predictions regarding the distribution and magnitude of internal forces acting in the crossing areas
between longitudinal and transversal laminations. The proposed modified model assumptions regarding the
distribution of lamination shear forces, which in turn influence the forces acting in the crossing areas, are
suggested to be taken into account in design of CLT beams.

1. Introduction
Using Cross Laminated Timber (CLT) at in-plane beam loading
conditions is very relevant from a practical engineering point of view,
since the transversal layers have a reinforcing effect with respect to
stress perpendicular to the beam axis. The stress state is however very
complex and many failure modes and geometry parameters need to be
considered in design. A particularly challenging task is strength ver-
ification with respect to in-plane shear where three failure modes need
to be considered: gross shear failure (mode I), net shear failure (mode
II) and shear failure in the crossing areas between adjacent longitudinal
and transversal laminations (mode III).
Experimental tests on CLT beams are for example reported by
Joebstl et al. [1], Bejtka [2], Andreolli et al. [3], Flaig [4], Flaig & Blass
[5] and Danielsson et al. [6]. A comprehensive experimental in-
vestigation and design concepts of CLT diaphragms at shear loading are
further presented by Brandner et al. [7]. The stress state differs however
partly between in-plane loading of CLT diaphragms and CLT beams, for
example regarding the stresses relevant for shear failure mode III.
An analytical model for stress analysis and strength verification of
CLT beams has been presented by Flaig [4,8–10] and by Flaig & Blass
[5], including proposals for stress based failure criteria for relevant
failure modes. This model has also been used as a basis for design
equations in the ongoing revision work of Eurocode 5 (EC5). The shear
stresses acting in the crossing areas between longitudinal and trans-
versal laminations (relevant for shear failure mode III) can according to
Flaig and Flaig & Blass, with sufficient accuracy, be assumed to be
uniformly distributed in the beam width direction, irrespective of the
element lay-up. The torsional moments acting in the crossing areas are
furthermore assumed to be uniformly distributed over all crossing areas
in the beam height direction. Based on comparison to 3D FE-analyses as
presented in [11], both these assumptions seem to be inaccurate.
The aim of this paper is to give a brief review of the analytical model
presented by Flaig and Flaig & Blass and, in addition, also to present
improvements of that model. The improvements relate to the magni-
tude and distribution of internal forces and stresses relevant for de-
termination of load bearing capacity with respect to shear force. The
main focus is here placed on the magnitude and distribution, in both the
beam width and the height directions, of the forces and stresses acting
in the crossing areas and how these are influenced by the beam geo-
metry.
2. Analytical model
A brief review of the model presented by Flaig [4,8–10] and by Flaig
& Blass [5] for calculation of internal forces and stresses relating to
relevant failure modes for CLT beams is presented below. More detailed
reviews of the considered model are presented in [6,11], where also

⁎
Corresponding author.
E-mail address: henrik.danielsson@construction.lth.se (H. Danielsson).
URL: http://www.byggmek.lth.se/english (H. Danielsson).

https://doi.org/10.1016/j.engstruct.2018.03.018
Received 7 November 2017; Received in revised form 12 February 2018; Accepted 10 March 2018
0141-0296/ © 2018 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/).

Please cite this article as: Danielsson, H., Engineering Structures (2018), https://doi.org/10.1016/j.engstruct.2018.03.018
H. Danielsson, E. Serrano
Fig. 1. Illustration of beam model and definition of load and geometry parameters.
experimental tests and comparison of model predictions and test results
are presented.
The equations presented below are based on notation for geometry
and load parameters according to Fig. 1. They further relate to pris-
matic CLT beams without edge-bonding and composed of longitudinal
and transversal laminations having identical stiffness properties. Index i
refers to the position of the longitudinal laminations in the beam height
direction and index k refers to the position of the longitudinal and
transversal layers in the beam width direction.
Cross section forces and bending moments are considered on three
separate levels according to Fig. 1: (V ,N ,M ) refer to the forces and the
bending moment acting on the total cross section, (Vi ,Ni,Mi ) refer to the
sum of forces and bending moments acting in all k longitudinal lami-
nations for a certain i and (Vi,k,Ni,k,Mi,k ) refer to the forces and the
bending moment acting in an individual longitudinal lamination i,k .
2.1. Bending
The maximum normal stress in the longitudinal layers, due to
bending, is given by
M tnet,0 h2
σx = where Wnet =
Wnet 6 (1)
where M is the bending moment, tnet ,0 = ∑ t0,k is the sum of the widths
of the longitudinal layers, i.e. the longitudinal net cross section width,
and h is the beam height.
2.2. Shear mode I and mode II
Schematic illustrations of the shear stress distributions in the long-
itudinal and transversal layers are presented in Fig. 2. The illustrations
relate to beams having an integer number of longitudinal laminations,
m, of identical width, b0 , in the beam height direction. Section A-A
refers to a section through the centre of a transversal lamination while
section B-B refers to a section in-between two adjacent transversal
2
Engineering Structures xxx (xxxx) xxx–xxx
laminations. The shear stresses in the longitudinal and transversal
layers are here denoted τxy,0 and τxy,90 , respectively.
The maximum value of the gross shear stress (shear mode I) is given
by
3 V
τxy,gross =
2 tgross h (2)
where V is the shear force and tgross = ∑ t0,k + ∑ t90,k is the gross cross
section width. For verification with respect to gross shear failure, re-
levant for CLT elements with edge-bonding, characteristic shear
strength according to the strength class of the laminations according to
EN 338 and use of kcr = 1.0 is proposed in [9]. Brandner et al. [7]
suggest to use a characteristic shear strength of 3.5 MPa for CLT com-
posed of C24 laminations and to account for the influence of possible
cracks by disregarding half of the width of the outermost layer on each
side of the beam, when determining tgross and τxy,gross according to Eq.
(2).
The maximum values of the net shear stress in the longitudinal and
transversal layers (shear mode II) can according to [9], with sufficient
accuracy, be expressed as
3 V
τxy,0 = τxy,net ,0 =
2 tnet ,0 h (3)
3 V
τxy,90 = τxy,net ,90 =
2 tnet ,90 h (4)
where tnet ,0 = ∑ t0,k and tnet ,90 = ∑ t90,k refer to the net cross section
widths of the longitudinal and transversal layers, respectively. Eq. (3) is
based on assuming that the total shear force is uniformly distributed
over the m longitudinal laminations, i.e. Vi = V / m . The maximum va-
lues of the shear stress distributions τxy,90 and τxy,net ,90 are exactly equal
for beams with an even number of laminations, m, in the beam height
direction while they differ slightly for beams with an odd number m, see
Fig. 2. For net shear failure, a characteristic shear strength of 8.0 MPa is
suggested in [9]. Test results indicate an influence of the width t of the
H. Danielsson, E. Serrano Engineering Structures xxx (xxxx) xxx–xxx

t90 t90
A B A A B B
Wxy,gross
i=4 Wxy,gross b0
Wxy,net,90 Wxy,gross
y y
i=3 b0
x z
Wxy,90 = 0 h
i=2 Wxy,net,0 b0
V
Wxy,90 Wxy,0
Wxy,0
i=1 b0

A B A A B B
b90 t0 t0 t0
Long. layer Trans. layer Long. layer Trans. layer
t90 t90
A B A A B B
Wxy,gross
i=3 Wxy,gross b0
y Wxy,net,90 Wxy,gross y
x z
i=2 Wxy,90 = 0 b0 h
Wxy,net,0
V Wxy,0 Wxy,0
i=1 Wxy,90 b0

A B A A B B
b90 t0 t0 t0
Fig. 2. Illustrations of assumed shear stress distribution in longitudinal and transversal layers at sections A-A and B-B for beams with m = 4 and m = 3 longitudinal laminations in the
beam height direction, based on [4].

failing layer on the net shear strength, with increasing strength at de- 6V 1 1 1
τxz = ⎛ − ⎞
creasing failing layer width. Consequently, a design concept based on a b2 nCA ⎝ m2 m3 ⎠ (5)
reference net shear strength of 5.5 MPa, valid for layer widths t ≤ 40
q 1
mm, is proposed by Brandner et al. [7] together with specific strength τyz =
h nCA (6)
modifications based on the actual longitudinal and transversal layers
widths. 3V 1 1 1
τtor = ⎛ − ⎞
b2 nCA ⎝ m m3 ⎠ (7)

2.3. Shear mode III
In addition to the shear stress τxy , which is present in both long-
itudinal and transversal laminations, shear stresses τxz and τyz acting in
the crossing areas between the longitudinal and transversal laminations
are also present. Using the model as suggested by Flaig and Flaig &
Blass, these shear stresses can be decomposed into stresses due to three
basic in-plane loading situations: (a) shear stress parallel to the beam
axis τxz , (b) shear stress perpendicular to the beam axis τyz and (c)
torsional shear stress τtor .
Assumed shear stress distributions for a single crossing area are
shown in Fig. 3 and expressions for maximum stresses are given below.
Apart from the previously stated assumptions, these expressions are
based on the assumptions of equal width of the longitudinal and
transversal laminations, i.e. b0 = b90 = b, a beam composed of an in-
teger number of longitudinal laminations, m, in the beam height di-
rection and having nCA number of crossing areas in the beam width
direction. The maximum values of the shear stress components can ac-
cording to [9], with sufficient accuracy, be expressed as
The shear stress components τxz ,τyz and τtor represent for a specific
point either longitudinal shear, rolling shear or a combination of both.
A shear stress component giving pure longitudinal shear in the long-
itudinal lamination represents pure rolling shear in the transversal la-
minations, and vice versa. Thus, for failure in the crossing areas, and
based on experimental tests on crossing areas loaded in either uni-axial
shear, pure torsion, or a combination of both, failure criteria according
to
τtor τxz
+ ⩽ 1.0
fv,tor fR (8)
τtor τyz
+ ⩽ 1.0
fv,tor fR (9)
are proposed in [9]. Test results indicate a mean value of the torsional
strength of about fv,tor = 3.5 MPa and a mean value of the rolling shear
strength of about fR = 1.5 MPa. The corresponding characteristic values
are fv,tor ,k = 2.75 MPa and fR,k = 1.1 MPa. These strength values are in-
line with corresponding values suggested in the literature, see e.g. [7].
In relation to the failure criteria according to Eqs. (8) and (9) it should
however be noted that, from a theoretical point of view, the origin of

Fig. 3. Illustration of assumed shear stress distributions in the crossing areas.

3
H. Danielsson, E. Serrano Engineering Structures xxx (xxxx) xxx–xxx

2 the exact stress prediction according to the model assumptions. The

k = 1, k = 3 ratio t0,k / nCA,k is always constant for (symmetric) 3-layer elements, but
k=2
(t0,k / tnet,0 )·(nCA /nCA,k ) [-]

varies in general for 5- and 7-layer elements. The ratio t0,k / nCA,k is then
only constant for the specific case when the internal longitudinal layers
1.5 have twice the width of the external layers. This influence of the ele-
ment lay-up on the distribution of the shear stresses in the crossing
areas can according to [9] be disregarded and Eqs. (5)–(7) are stated as
1 being sufficiently accurate for element lay-ups used in practice.
Accounting for varying ratio t0,k / nCA,k between the k layers of
longitudinal laminations in the beam width direction, the maximum
shear stresses are given by
0.5
6V 1 t0,k 1 1
τxz,k = ⎛ − ⎞
b2 nCA,k tnet ,0 ⎝ m2 m3 ⎠ (10)

0 q 1 t0,k
τyz,k =
0 0.5 1 1.5 2 2.5 3 h nCA,k tnet ,0 (11)
t0,2 / t0,1 = t0,2 / t0,3 [-]
3V 1 t0,k 1 1
τtor ,k = ⎛ − ⎞
Fig. 4. Factor for shear stress prediction vs. relative width of longitudinal layers. b2 nCA,k tnet ,0 ⎝ m m3 ⎠ (12)
The influence of the relative widths of the longitudinal layers on the
the in-plane shear stress components acting in the crossing areas is ir- predicted stresses can be expressed by the factor (t0,k / tnet ,0 )·(nCA / nCA,k ) ,
relevant. The torsional shear strength fv,tor can thus only be understood i.e. by the ratios τk / τ according to Eqs. (10)–(12) and (5)–(7), respec-
as a fictitious strength parameter strongly related to the structural tively. The relationship between the relative longitudinal layer widths,
properties of the test conditions used to determine it. assuming a fixed net cross section width tnet,0 , and the predicted stress
The parallel to beam axis shear stress τxz increases with the distance for the respective crossing areas is illustrated in Fig. 4 for a 5-layer
from the neutral axis and Eq. (5) is consequently valid for the upper- element. The maximum stress increases by 33% for the case of equal
and lowermost crossing area in the beam height direction. The torsional width of all longitudinal layers and by 60% for the case of external
shear stress τtor , however, is by assumption equal for all crossing areas layers having twice the width of the internal layer, compared to the
in the beam height direction. The upper- and lowermost crossing areas reference case of t0,2/ t0,1 = t0,2/ t 0,3 = 2.0 .
are hence the ones relevant for strength verification according to Eq.
(8). 3.2. Stress distribution in beam height direction

3. Model improvements The assumption of equal torsional moments and torsional shear
stresses for all crossing areas in the beam height direction, mentioned
The analytical model for design of prismatic CLT beams, as pre- above, corresponds to assuming equal shear forces Vi for all longitudinal
sented by Flaig [9] and Flaig & Blass [5], has many appealing proper- laminations in the beam height direction, at sections B-B in Fig. 2. 3D
ties. The equations for prediction of maximum stress and failure criteria FE-analyses, as presented in [11,12], have however shown that this
are fairly simple, in spite the rather complex mechanical model. The assumption is inaccurate and that the torsional moments and shear
model does however suffer from some drawbacks, as pointed out in forces are significantly greater close to the beam centre-line compared
[6,11], relating to assumptions regarding the distribution of the in- to the upper and lower parts of the beam.
ternal forces and shear stresses acting in the crossing areas. Equal shear We here propose a more plausible assumption of the distribution of
stresses for all crossing areas in the beam width direction according to lamination shear forces Vi in the beam height direction resulting in a
Eqs. (5)–(7) is assumed, irrespective of the element lay-up. Comparison quite different, non-uniform, distribution of the crossing area torsional
to 3D FE-analyses as presented in [11] has shown that this assumption moments. Instead of assuming equal lamination shear forces, the total
is inaccurate and that the crossing area shear stresses instead are in- shear force is assumed to the divided between the individual lamina-
fluenced to a significant extent by the element lay-up in terms of the tions in accordance with the parabolic shear stress distribution found
respective relative widths t0,k / tnet ,0 of the longitudinal layers. The model from conventional engineering beam theory, τxy (y ) , considering a
is further based on the assumption of equal torsional moments for all homogeneous beam of width tnet,0 and height h.
crossing areas in the beam height direction. Also this assumption has The shear forces Vi can in general be expressed as
been shown to be inaccurate, since results of FE-analyses as presented Vi = αi V (13)
in [11] show uneven distributions over the beam height with sig-
nificantly higher values of torsional moments and shear stresses close to where αi are dimensionless weighting factors. We now define a di-
the beam centre-line compared to the upper- and lower parts. Sugges- mensionless parabolic function T (y ) as illustrated in Fig. 5 according to
tions for model improvements relating to these two issues are presented 3 1
T (y ) = (m2−4y 2 )
below. 2 m3 (14)
m /2
and hence such that T (−m /2) = T (m /2) = 0 and ∫ −m /2
T (y ) dy = 1.0 .
3.1. Stress distribution in beam width direction The weighting factors αi can then be determined as
2
∫−m/2+(i−1) T (y) dy = 6i−6i + m (6i−3)−2
m /2 + i
The assumption of equal shear stress for all crossing areas in the αi =
m3 (15)
beam width direction as stated in [9] is valid for certain types of ele-
ment lay-ups, but is not generally valid. The governing parameter is the and values of αi are given in Table 1 for 1 ⩽ m ⩽ 7 .
ratio t0,k / nCA,k , where t0,k is the individual longitudinal layer width and Derivation of the torsional moments Mtor ,i,k and the torsional shear
nCA,k is the number of crossing areas that the individual longitudinal stress τtor ,i,k can be carried out by equilibrium considerations of a part of
layer shares with adjacent transversal layers, i.e. nCA,k = 1 or 2. For a single longitudinal lamination. The shear forces Vi are then assumed
element lay-ups with constant ratio t0,k / nCA,k , Eqs. (5)–(7) should give to be divided between the k individual laminations in the beam width

4
H. Danielsson, E. Serrano Engineering Structures xxx (xxxx) xxx–xxx

Fig. 5. Illustration of parabolic function T (y ) and

shear force- and stress distribution.

Table 1 the crossing areas located closest to the beam centre-line, as was also
Weighting factors αi with i = 1,…,m for 1 ⩽ m ⩽ 7 . found from the FE-analyses presented in [11].
The maximum torsional shear stress according to Eq. (19) is always
m
greater than, or equal to, the maximum torsional shear stress according
1 2 3 4 5 6 7 to Eq. (12). Considering Eqs. (19) and (12), the difference in maximum
predicted torsional shear stress may be expressed by the ratio
7 – – – – – – 19/343
τtor ,i,k / τtor ,k = (max{αi}−1/ m3)/(1/ m−1/ m3) and is hence, for a given
6 – – – – – 4/54 49/343
5 – – – – 13/125 10/54 67/343
number m, governed by the weighting factors αi . Fig. 6 illustrates the
i 4 – – – 5/32 31/125 13/54 73/343 relationship between this ratio and the number of longitudinal lami-
3 – – 7/27 11/32 37/125 13/54 67/343 nations in the beam height direction, m. For m = 1, the torsional mo-
2 – 1/2 13/27 11/32 31/125 10/54 49/343 ment and hence the torsional stresses are zero. For m = 2, α1 = α2 = 1/2
1 1 1/2 7/27 5/32 13/125 4/54 19/343
and Eq. (19) gives equal torsional shear stress for the two crossing areas
in the beam height direction, as is also assumed in [9], and Eqs. (12)
direction based on their respective relative width t0,k / tnet ,0 according to and (19) predict the same maximum stress values. For odd numbers
m ⩾ 3, Eq. (19) gives a maximum torsional shear stress 1.50 times
t0,k greater than according to Eq. (12). The ratio between maximum stress
Vi,k = αi V
tnet ,0 (16) predictions for even numbers m is 1.40 for m = 4 and approaches 1.50
as m increases.
Consideration of rotational equilibrium of a part of a single long-
The distribution of the parallel to beam axis shear stress τxz over the
itudinal lamination i,k of length b90 in the x-direction, see Fig. 1, then
beam height, considering the influence of element lay-up as discussed
gives

1 t0,k ⎛ b
Mtor ,i,k = Vi b90 + ΔVi 90 −ΔMi⎞ 1.6
nCA,k tnet ,0 ⎝ 2 ⎠ (17)
(max{Di}-1/m3) / (1/m-1/m3) [-]

where ΔVi and ΔMi refer to the changes in shear force and bending 1.5
moment over the finite length b90 . For zero external load q and as-
suming a CLT element composed of an integer number of longitudinal 1.4
laminations, m, in the beam height direction, the torsional moment may
be expressed as 1.3
Vb90 t0,k 1
Mtor ,i,k = ⎛αi− ⎞
nCA,k tnet ,0 ⎝ m3 ⎠ (18) 1.2

For the case of equal width of all laminations, i.e. b0 = b90 = b , and 1.1
assuming a stress distribution according to Fig. 3 and corresponding to
assuming rigid body rotation over a shear compliant medium, the 1
maximum torsional shear stress at the middle points of the four sides of
crossing area i,k is then given by
0.9
3V 1 t0,k 1 100 101 102
τtor ,i,k = 2 ⎛αi− ⎞
b nCA,k tnet ,0 ⎝ m3 ⎠ (19) Number of longitudinal laminations m [-]
with αi according to Eq. (15). Eqs. (18) and (19) give distributions of Fig. 6. Ratio of torsional shear stress predictions τtor ,i,k according to Eq. (19) and τtor ,k
according to Eq. (12) vs. number of longitudinal laminations m.
the torsional moments and torsional stresses with maximum values for

5
H. Danielsson, E. Serrano Engineering Structures xxx (xxxx) xxx–xxx

3 4
Lamination number i 2
Wtor 3
Wxz 2
Wtor + Wxz 2

1
1 1
Normalized shear stress 0 0.25 0.5 0.75 1.0 0 0.25 0.5 0.75 1.0 0 0.25 0.5 0.75 1.0

5 6 7 8
6 7
5
4 6
5
4 5
3 4
3 4
3
2 3
2 2 2
1 1 1 1
0 0.25 0.5 0.75 1.0 0 0.25 0.5 0.75 1.0 0 0.25 0.5 0.75 1.0 0 0.25 0.5 0.75 1.0

Fig. 7. Distributions of maximum values for each individual crossing area of torsional shear stress τtor , parallel to beam axis shear stress τxz and the sum τtor + τxz in the beam height
direction for 2 ⩽ m ⩽ 8 . The torsional shear stress τtor is determined according to Eq. (19) and the parallel to beam axis shear stress τxz according to Eq. (20). Longitudinal lamination
number i is given on the vertical axis and normalized shear stress is given on the horizontal axis, normalized in such a way that max{τtor + τxz } = 1.0 .

in Section 3.1 and equal width of all laminations (i.e. b0 = b90 = b), is the choice used here, with fv,tor / fR = 2.33, the critical crossing areas are
given by either in the centre (m = even ) or next to centre (m = odd ).

12V 1 t0,k m+1

τxz,i,k = ai with ai = b ⎛ −i⎞
m3b3 nCA,k tnet ,0 ⎝ 2 ⎠ (20) 4. Discussion

where ai is the distance in the y-direction between the centre-line of the
gross cross section and the centre-line of the lamination i,k , see Fig. 1
and [6]. Maximum (absolute) stress values are found at the upper- and
lowermost crossing areas, i.e. for i = 1 and i = m respectively, with
|a1| = |am| = 0.5b (m−1) which applied in Eq. (20) yields Eq. (10).
The distributions over the beam height of the maximum values for
each individual crossing area of the torsional shear stress τtor , the par-
allel to beam axis shear stress τxz and the sum τtor + τxz are illustrated in
Fig. 7 for different beam geometries, considering the number of long-
itudinal laminations in the beam height direction, m. In addition, Fig. 8
shows the corresponding distributions normalized by their respective
strength values. Note that depending on the choice of the strength va-
lues fv,tor and fR , the position of the critical crossing area might vary. For
The maximum values of the torsional shear stress τtor according to
Eq. (19) and of the parallel to beam axis shear stress τxz according to Eq.
(20) in general occur at different crossing areas, see Fig. 7. Thus,
making use of these maximum values, the failure criterion of Eq. (8)
would be mechanically incorrect and would, assuming the failure cri-
terion to be valid, also yield conservative strength predictions.
Since the failure criterion in Eq. (8) is suggested to be used con-
sidering different shear strengths, fv,tor and fR , for the torsional shear
stress τtor and the parallel to beam axis shear stress τxz respectively, it is
not straightforward to determine which crossing area that is stressed
the most, see Fig. 8. An obvious solution, but for a practising engineer a
rather cumbersome one, is to consider both stress components and the
failure criterion for all m crossing areas in the beam height direction

3 4
2
Lamination number i

Wtor / fv,tor 3
Wxz / fR 2
Wtor / fv,tor + Wxz / fR 2

1
1 1
Normalized shear stress 0 0.25 0.5 0.75 1.0 0 0.25 0.5 0.75 1.0 0 0.25 0.5 0.75 1.0

5 6 7 8
6 7
5
4 6
5
4 5
3 4
3 4
3
2 3
2 2 2
1 1 1 1
0 0.25 0.5 0.75 1.0 0 0.25 0.5 0.75 1.0 0 0.25 0.5 0.75 1.0 0 0.25 0.5 0.75 1.0

Fig. 8. Distributions of maximum values for each individual crossing area of τtor / fv,tor ,τxz / fR and τtor / fv,tor + τxz / fR in the beam height direction for 2 ⩽ m ⩽ 8 , with
fv,tor / fR = 3.5/1.5 = 2.33 . The torsional shear stress τtor is determined according to Eq. (19) and the parallel to beam axis shear stress τxz according to Eq. (20). Longitudinal lamination
number i is given on the vertical axis and normalized shear stress is given on the horizontal axis, normalized in such a way that max{τtor / fv,tor + τxz / fR } = 1.0 .

6
H. Danielsson, E. Serrano
individually. Based on a fixed ratio of the considered strength values,
e.g. based on the suggested mean strengths as fv,tor / fR = 3.5/1.5 = 2.33,
it should however be possible to establish a modified failure criterion
which is sufficiently accurate and yet suitable in a practical engineering
context.
An interesting observation can be made from Fig. 6. The ratio be-
tween the maximum torsional shear stress based on the improved
model assumptions, Eq. (19), and the model as presented in [9], Eq. (7),
is in the range 1.40–1.50 for CLT beam elements composed of at least
three longitudinal laminations in the beam height direction, i.e. m ⩾ 3.
Taking this fairly constant ratio into account, and accounting also for
the influence of element lay-up in terms of the relative width of the
internal and external longitudinal layers, the maximum torsional shear
stress can be approximated as
3V 1 t0,k 3 1 1 improved model assumptions compared to the model presented in [9] is
τtor ,max ,k ≈ ⎛ − ⎞
b2 nCA,k tnet ,0 2 ⎝ m m3 ⎠ (21)
which for m ⩾ 3 gives no more than about 7% error compared to Eq.
(19). By this approximation, calculation of the weighting factors αi
according to Eq. (15) is not needed.
Although not specifically mentioned in Section 3, the improved
model assumptions regarding the distribution of the shear forces Vi in
the longitudinal laminations will also affect the maximum longitudinal
layer net shear stress τxy,0 according to Eq. (3). Considering the max-
imum lamination shear force to be max{Vi } = max{αi} V instead of as-
suming equal laminations shear forces Vi = V / m , gives an increase in
the predicted longitudinal layer net shear stress τxy,0 by a factor of
1.38–1.50 for m ⩾ 3. This may be relevant to consider for elements with
orientation such that the longitudinal layers are the internal layers and
the transversal layers are the external layers.
Equations, figures and tables presented in Section 3 relate to CLT
beam elements having equal lamination widths for the longitudinal and
transversal layers, i.e. b0 = b90 = b, and also having an integer number
of longitudinal laminations in the beam height direction, m. For beams
with geometries of more arbitrary character in terms on lamination
widths, as presumably encountered in practical engineering design si-
tuations, the internal force distributions and the stress distributions will
differ. The idealized case of equal laminations widths is however be-
lieved to yield results which are relevant also for the case of a more
arbitrary beam geometry.
5. Conclusions
Shear stresses acting in the crossing areas can according to [9], with
sufficient accuracy, be assumed as equal for all crossing areas in the
beam width direction irrespective of the element lay-up in terms of the
relative width of the internal and external longitudinal layers. Based on
comparison to 3D FE-analyses as presented in [11], this assumption
appears to be inaccurate. As illustrated in Fig. 4, by comparison of Eqs.
(5)–(7) and (10)–(12) respectively, the relative widths of the long-
itudinal layers have a significant influence on the predicted stresses.
The difference in maximum predicted stress is up to 60% for CLT ele-
ment lay-ups which are commonly used in practice. This relatively large
Engineering Structures xxx (xxxx) xxx–xxx
influence on the stress distribution should be considered in design of
CLT elements at in-plane beam loading conditions.
The assumption of equal torsional moments for all crossing areas in
the beam height direction, as used in the model presented in [9], has by
comparison to 3D FE-analyses also been shown to be inaccurate, see e.g.
[11,12]. A model improvement regarding the assumption of the dis-
tribution of the shear forces in the individual longitudinal laminations
Vi,k which consequently also affects the distribution of the torsional
moments in the crossing areas Mtor ,i,k has been presented here. The la-
mination shear forces, the torsional moments and hence also the tor-
sional shear stresses τtor ,i,k are, according to the improved model, not
uniformly distributed in the beam height direction and have maximum
values at the lamination/laminations closest to the beam centre-line.
The difference in predicted maximum torsional shear stress using the
40–50% considering elements with at least three longitudinal lamina-
tions in the beam height direction. The proposed model improvements
should hence also be taken into account in design of CLT elements at in-
plane beam loading conditions.
Acknowledgements
This work was financially supported by the Swedish Research
Council Formas through grant 2016-01090 for the project Cross
Laminated Timber Beams – Rational Structural Analysis. The support
from COST Action FP1402 is also gratefully acknowledged.
References
[1] Joebstl RA, Bogensperger T, Schickhofer G. In-plane shear strength of cross lami-
nated timber. In: Proc. CIB-W18, CIB-W18/41-12-3, St Andrews, Canada; 2008.
[2] Bejtka I. Cross (CLT) and diagonal (DLT) laminated timber as innovative material
for beam elements. Karlsruhe, Germany: KIT; 2011.
[3] Andreolli M, Tomasi R, Polastri A. Experimental investigation on in-plane behaviour
of cross-laminated timber elements. In: Proc. CIB-W18, CIB-W18/45-12-4, Växjö,
Sweden; 2012.
[4] Flaig M. Biegeträger aus Brettsperrholz bei Beanspruch in Plattebene [PhD thesis].
Karlsruhe, Germany: KIT; 2013.
[5] Flaig M, Blass HJ. Shear strength and shear stiffness of CLT-beams loaded in plane.
In: Proc. CIB-W18, CIB-W18/46-12-3, Vancouver, Canada; 2013.
[6] Danielsson H, Jelec M, Serrano E. Strength and stiffness of cross laminated timber at
in-plane beam loading. Report TVSM-7164. Sweden: Div of Structural Mechanics,
Lund University; 2017. Available for download at < www.byggmek.lth.se/english/
publications > .
[7] Brandner R, Dietsch P, Dröscher J, Schulte-Wrede M, Kreuzinger H, Sieder M. Cross
laminated timber (CLT) diaphragms under shear: test configuration, properties and
design. Constr Build Mater 2017;147:312–27.
[8] Flaig M. Design of CLT beams with rectangular holes or notches. In: Proc. INTER,
INTER/47-12-4, Bath, United Kingdom; 2014.
[9] Flaig M. In Plattenebene beanspruchte Biegeträger aus Brettsperrholz – Teil 1:
Effektive Festigkeits- und Steifigkeitskennwerte für die Schubbemessung.
Bautechnik 2015;92:741–9.
[10] Flaig M. In Plattenebene beanspruchte Biegeträger aus Brettsperrholz – Teil 2:
Brettsperrholzträger mit angeschnittenen Rändern, Durchbrüchen oder
Ausklinkungen. Bautechnik 2015;92:750–8.
[11] Danielsson H, Serrano E, Jelec M, Rajcic V. In-plane loaded CLT beams – tests and
analysis of element lay-up. In: Proc. INTER, INTER/50-12-2, Kyoto, Japan; 2017.
[12] Jelec M, Rajcic V, Danielsson H, Serrano E. Structural analysis of in-plane loaded
CLT beam with holes: FE-analysis and parameter studies. In: Proc. INTER, INTER/
49-12-2, Graz, Austria; 2016.