Concentrated load introduction in CLT elements perpendicular to plane –

experimental and numerical investigations

Thomas Bogensperger1, Robert A. Joebstl2, Manfred Augustin3

ABSTRACT: In this contribution bending and shear tests of cross laminated timber (CLT) plates under concentrated
loads are presented. The so loaded structural members can fail either due to punching along a critical perimeter line in
the vicinity of the concentrated load or in bending. Two test configurations were developed and investigated by linear
elastic models. The obtained test results and observed failures as well as their correlation with the mechanical modeling
are shown in this paper. The established numerical model was a 3D solid model with different material behavior for all
acting stresses. The material behavior was implemented in a user subroutine for the FE program ABAQUS. By
comparison of measured and computed load displacement curves numerical models could be discussed regarding their
reliability and conclusions about missing input for an increasing accuracy of the model could be drawn.

KEYWORDS: CLT, concentrated load, point load, punching, rolling shear, FE-simulation, softening

1 INTRODUCTION 123 widely clear for internal section forces and moments
excepting twisting moments [3]. The Reissner-Mindlin
Cross laminated Timber (CLT) – also known as X-lam – plate theory corresponds to the Timoshenko beam theory
can be utilised as a structural element for loads in and/or but corresponding 2D plate-theories on a comparable
out of plane. Under uniformly distributed loads out of level of the modified γ-method or shear analogy method,
plane bending moment and shear forces are often e.g. implemented in FEM software for engineering
considered only in one direction like a beam. Several applications are still missing. Mechanical computation of
methods for the computation of internal section forces CLT for scientific purposes is carried out commonly
are available for these cases. For example the well with 3D solid FE models using non-linear material
known Timoshenko beam theory which takes into properties [1], [4].
account the shear weakness of layers under rolling shear Modern challenging architecture demands a two
stresses is one commonly used representative. Other dimensional load carrying behaviour of CLT plates in
methods are e. g. the modified γ-method and the shear special cases. Support by columns (e.g. Fig. 1) is often
analogy method. A recently worked out study [9] needed for such solutions.
showed that differences in the computation of stresses
determined by means of higher (plate) theories and
simple beam theories like Timoshenko's are negligible.
This paper deals with bending and shear failure of CLT
plates under concentrated loads out of plane [1], [6]. The
mechanical computation under concentrated loads is no
longer reliable with a one dimensional model.
Consequently the shear weak Reissner-Mindlin plate
theory is often used for engineering purposes. The
mechanical stiffness properties in this case are given in
[2]. Usually 5 stiffness values – two bending, one twist
and two shear stiffness values – are needed. Internal
section forces and moments of CLT elements under
concentrated loads show high values in a small zone
around the load introduction. Strength verifications are
Figure 1: example for CLT plates with column support.
1
Thomas Bogensperger, competence centre for timber
engineering and wood technology holz.bau forschungs gmbh, Several mechanical verifications have to be performed in
Inffeldgasse 24, 8010 Graz, Austria, [email protected]
2
Robert A. Joebstl, Haas Fertigbau Holzbauwerk GmbH & Co
such a case: One is the verification of compression
KG/Haas Holzprodukte GmbH, A-8263 Großwilfersdorf, perpendicular to plane (see Bogensperger et. al at CIB
Industriestraße 8, [email protected] meeting in 2011 [10]). In addition the verification of
3
Manfred Augustin, [email protected] bending and shear stresses in the CLT plate under the
high local loads (punching) are necessary (see 7-layered CLT section (CLT 7s*) led to a third cross
Bogensperger et. al at INTER meeting in 2015 [1]). The section (170 mm thickness) in order to cover a further
state of the art for bending and shear verification of CLT industrial manufactured CLT panel (producer 2). A span
panels is documented in [1], [7] and [8]. The bending to thickness ratio between 6.17 (CLT 5s) and 5.88 (CLT
verification is conducted with realistic load introduction 7s*) depending on the cross-section could be calculated
models where the finite size of the load introduction in which is very similar to the ratio of the tested specimen
opposite to single node load introduction avoids in [6] (Ratio there: 6.28).
numerical high gradients which can not develop under
Table 1: Stiffness values and lay-up of the tested CLT
realistic conditions. The verification of shear stresses in elements for configuration 1
case of concentrated loads should be carried out along a
perimeter line which is based on the findings in [6]. The
suggested location of this perimeter line is shown in Fig. E0 E90 G0 G90 # t1 t2 t3 t4 t5 t6 t7
2. N/mm²² mm

CLT 11600 ≈0 690 50 5 34 30 34 30 34 - -

5s
CLT 11600 ≈0 690 50 7 19 30 19 30 19 30 19
7s
CLT 11600 ≈0 690 50 7 20 30 20 30 20 30 20
7s*

An illustration of "CLT 5s" and "CLT 7s" and associated

shear stress distributions for a unit shear forces in
longitudinal (0°) and in transverse direction (90°)
(qx=qy=1 kN/cm) is given in Fig 3.

Figure 2: Controlling perimeter line based on the findings of

Mestek [6].

This controlling perimeter line was computed under an

assumed load spreading angle of 35° according to
Mestek [6] (Fig. 2).

2 EXPERIMENTAL TESTS
The dominance of bending versus shear verification
depends on the geometric properties of the CLT plate
(span to thickness) as well as on its lay-up. The internal
structure indicates the number, orientation and thickness
of a each single layer. Two different test setups were
developed at the competence centre holz.bau. forschungs
gmbh Graz/Austria in order to study both expected
failure modes (bending and shear) separately. Punching Figure 3: Shear stresses of CLT 5s respectively CLT 7s
shear failure was expected with configuration 1, while
bending failure was estimated for configuration 2. In addition CLT panels produced under laboratory
conditions were investigated with the same cross
2.1 Configuration 1 sections (CLT-5s and CLT-7s). The lamellas of these
(expected failure modes in shear) panels had to undergo a strong grading procedure. The
The tested CLT specimens of configuration 1 had lamellas used were graded in respect to the density of the
dimensions of 1.2 by 1.2 m. The load was introduced in boards within a range of 400 kg/m³ ± 20 kg/m³. Further
the centre of the plate with a manually driven hydraulic the annual ring pattern was graded. The allowance of
cylinder whereas the CLT plates were supported by knots was reduced in comparison to the industrially
several tension rods and steel profiles around the extent. produced elements. The idea behind this strict grading
The support of the CLT elements was provided at all 4 procedure was to achieve test results with a smaller
sides. The axes of the steel profiles (U100) were aligned coefficient of variation compared to the corresponding
to the grid of the used test field (50/50 cm). The layup of results of the industrially produced CLT panels. The
the investigated panels was chosen on basis of the complete test configuration is illustrated in Fig. 4.
availability based on industrial production (producer 1). Pictures of configuration 1 and some details can be seen
5- and 7-layered CLT panels with realistic thicknesses in Fig. 5.
(162 mm for 5-layered and 166 mm for 7-layered CLT The test specimens produced under laboratory conditions
panel) were investigated for use in engineering praxis An were denoted with “A” while the test specimens with
overview of the cross sections (CLT 5s and CLT 7s) can CLT elements from producer 1 are denoted with “B” and
be found in Tab. 1. A small change in the build up of the as series “C” resp. if they came from producer 2.
 For configuration 2 only CLT elements delivered by
producer 1 (denoted as "B") and producer 2 ("C") were
tested. The production of elements under laboratory
conditions was omitted [1][5].

Figure 4: Test configuration 1 (shear tests)

Figure 7: Overview test configuration 2

3 TEST RESULTS
3.1 RESULTS - CONFIGURATION 1
The observed load-displacement curves showed a
significant ductile behaviour [5]. Various initial failure
modes were detected in the different tests. A mixed
failure mode in bending and shear failure was observed
for configuration 1. Detailed values of the tests are given
in [1] and [5]. It is remarkable that the strength in rolling
shear due to punching forces could be determined with
fv,k,pu=kpu,v·fv,k,90=1.75·1.25=2.2 N/mm² when linear
elastic modelling is used which is a significant higher
strength value than it could be observed in standard tests
Figure 5: Overview and details of test configuration 1 [1]. The observed load displacement curves are
illustrated in Fig. 8.
2.2 Configuration 2
(expected failure mode in bending)
The test configuration 2 with expected failure mode in
bending under concentrated loads was designed with
significant higher length to thickness ratios. Restricted to
the grid of the load introduction points at the test field
the longer span a was chosen with l=4.0 m while the
shorter span was b=2.5 m. The mean length to thickness
ratios were between 19.5 and 18.6 depending on the used
cross section. A parameter study [5] showed that the 5- Figure 8: Load displacement curves - configuration 1
(a) 5-layered CLT; (b) 7-layered CLT
layered CLT elements behave worse in the global
bidirectional bending behaviour than the 7-layered ones.
Therefore tests with one directional bending behaviour 3.2 RESULTS - CONFIGURATION 2
with supports only along both shorter sides was proposed In configuration 2 in general bending failures occurred
for 5-layered CLT elements while a two directional (Fig 2). In addition secondary rolling shear failures
bending loading (with simple supports on all four sides) among other failure modes were observed. Numerical
of the CLT elements was implemented for the 7-layered predictions of strengths with bending strength
CLT element. These described test specimens are fm,CLT,k=aCLT·fm,k=28.8 N/mm² [7], [8] fitted quite well to
illustrated in Fig. 6. Two images of test configuration 2 the test observations in general. In case of bending
can be found in Fig. 7. failure an underestimation seems to exist when a bending
failure is forecast perpendicular to lamellas in the outer
layer (minor span direction). Furthermore an increase of
the local bending strength around the concentrated load
introduction due to the small volume with high stresses
can be expected. The available small number of tests
does not allow a reliable prediction of this increasement
in local bending. Therefore the bending correction factor
for local loads kpu,m is still assumed as 1.0 as a
Figure 6: Dimensions and type of supports of the test temporarily conservative approach. The local bending
specimens for test configuration 2 strength should be assumed as
fm,pu,k=kpu,m·fm,CLT,k =1.0·28.8. Two examples of the rolling shear modulus G90. The applied values in the
observed load displacement curves for configuration 2 analysis are summarised in Tab. 2.
are shown in Fig. 9.
Table 2: Elastic constants for the worked out numerical
model

E0=12000 N/mm² G0=690 N/mm²

E90=370 N/mm² G90=50 N/mm²

4.1.2 Mechanical model

The main aim of the simulation and computation on
basis of linear elastic material behaviour is to develop a
proper reliable test configuration in order to perform
Figure 9: Load displacement curves - configuration 2 relevant tests and make good predictions of the bearing
(a) 5-layered CLT; (b) 7-layered CLT
strength of the investigated CLT plates. Strength values
for CLT are available for 1D beam theory which also
In the analysis the observed test results were compared relates well to internal section forces of plate theories.
with the numerical ones: Whereas the numerical results Internal section forces and moments for CLT should be
and experimental data matched well and the differences determined with 2D structural shear weak plate theory
of the results can be explained for configuration 2 [8] a (Reissner-Mindlin) typically. Attention has to be paid to
good accordance with configuration 1 can only be found a realistic and accurate load introduction and support at
under some assumptions which are far too simple for a the boundaries. Modelling of load introduction was
good mechanical understanding [1]. In particular the solved with a rigid steel plate which is in contact to the
mechanical model should explain why the rolling shear CLT plate. The maximum bending moments can be
strength is remarkable higher under concentrated load found at the two symmetry-axis of the test configuration.
introduction than under distributed loads from test on 1D Dominating shear forces have to be considered at a
bending specimens. The improvement of rolling shear critical perimeter line with a distance to the outline of the
strength is in that case anyway in the range of 75%. In steel load introduction plate. Mestek recommends the
order to come to an increased understanding of this application of an load diffusion angle of 35º for
question numerical investigations were carried out. orthotropic CLT elements to set up this critical perimeter
line [6]. The bending and (rolling-) shear strength values
4 NUMERICAL INVESTIGATIONS to be considered in the vicinity of the concentrated load
are given in the previous section 3. Detailed results of
4.1 Linear elastic behaviour of the material
the elastic computations can be found in [5].
4.1.1 Elastic constants Based on the findings in [1] a shear strength value of
2.21 N/mm² could be computed (mean strength value:
When timber is loaded it should be considered as a linear 2.62 N/mm²; COV: 6%). Shear perpendicular to main
orthotropic material at the beginning. Best accordance orientation of CLT panels is dominating whereas a mean
can be achieved with a cylindrical coordinate system stress level of about 85% developed in the main
where the z-direction is parallel to the pith and r direction. The proposed strength value of 2.21 N/mm² is
respectively φ describes the cross section. A full based on a linear elastic stress computation and rolling
orthotropic material demands for 9 independent shear stresses have to be verified in the vicinity of
parameters: 3 moduli of elasticity in r, φ and z direction, concentrated load introduction. In one-dimensional
3 moduli of shear and 3 poison values. Values for these bending tests according to the present standards (e.g. EN
parameters can be found in literature, e.g. in [11]. In 16351) the characteristic rolling shear strength for CLT
general these values are based on experimental tests. plates are significantly lower (for spruce: fv,k,90=1.25
Alternatively the macroscopic stiffness values can be N/mm²). A mechanical explanation for the increased
predicted by a micromechanical model on the basis of rolling shear strength is necessary to understand and set
tissue-independent stiffness properties of hemicellulose, up regulations in a practical code. Consequently
cellulose, lignin, in a tissue-specific composition [12]. investigations applying a non-linear behaviour of the
Elastic constants used in the present numerical material have been worked out.
investigations had to undergo a further simplification
process. In the material model only differences between 4.2 Non-linear behaviour of the material
the parallel and perpendicular to grain are regarded. CLT
4.2.1 Post linear material behaviour
is build up with several layers and boards resp. which are
glued together. At the narrow side of the boards typically
General:
some gaps or cracks can be observed. Effects due to
Elastic stiffness values are given in Tab. 2. The material
poison numbers should be completely neglected
behaviour differs significant between several stress
therefore. With this only 4 relevant elastic constants
components. A relatively ductile behaviour can be
remain which are considered in the mechanical model.
expected under compression while the tress-strain
These 4 numbers are the modulus of elasticity parallel
behaviour is rather brittle under tension stresses. Shear
and perpendicular to grain, the shear modulus G0 and the
and rolling shear is neither ductile nor completely brittle be on the very conservative side. Tension strength
although rolling shear is less brittle than shear parallel to perpendicular to grain ft,90,mean shows a pronounced
grain.. The utilisation of a elastic plastic stress strain brittle behaviour with a dependence on the loaded
curve and/or softening in case of tension stresses has to volume. Results of mean values in [19] are shown in
be discussed in the given context. Generally quantile Tab. 4.
vales are needed for all kind of verifications. But mean
Table 4: Mean tension strength values [N/mm²], based on
values give the best agreement when load displacement the loaded volume
curves on basis of numerical results and test results are
compared. These values are not common available for
CLT elements. For simplification purposes a perfect
Volume [dm³] 0.95 2.25 4.95
elastic plastic behaviour in ductile case and simple linear
softening for brittle behaviour were chosen. ft90,mean [N/mm²] 2.11 1.58 1.26

Tension and compression in fibre direction:

A mean strength value of 1.50 N/mm² can be chosen as
Based on test results of 5 clients at the laboratory mean
an average value for the present computations. A tension
values for bending strength can be estimated. The results
softening effect follows the elastic stress strain relation
and the suggested mean value can be found in Tab. 3. As
after reaching the strength peak. A good estimation of
failures in CLT usually occur on the tension side these
fracture energy value of GFI = 300 J/m² for Mode I can
values are a good basis for the definition of tension
be found in [13] for the radial direction. A value for the
strength values under the specific load situation.
tangential direction is missing. As tension perpendicular
Rejecting the highest and lowest outlier-value for
to grain is a stress component of minor importance and
bending strength a mean value of 39 N/mm² was
tension softening is more challenging in computing than
introduced in the mechanical model. Due to volume
the consideration of a simple elastic plastic behaviour a
effect the strength value increased by 15% to meet the
value of 0.50 N/mm² together with a simple bi-linear
small area under high tension stresses in comparison to
elastic plastic behaviour is assumed for simplification
the standard test configuration in EN 408.
purpose. In addition the internal forces remain very low
The area under the linear softening curve corresponds to
which justifies this remarkable simplification.
the fracture energy in Mode I of spruce. Values for
Tests in [14] concerning compression perpendicular to
spruce can be found in Tab. 2 from [13]. GFI can be
grain deliver a mean value of 3.10 up to 3.80 N/mm²,
found with 1650 J/m² for tension in grain direction. In
depending on the thickness of the regarded lamellas. A
order to avoid numerical problems only particular zones
mean value of 3.5 N/mm² was chosen for the given
inherit this softening property whereas the other
computations in the numerical models.
elements remain elastic for the “smeared crack zone
modelling”.
Shear strength in fibre direction:
Table 3: Bending strength values [N/mm²] A mean strength value of 4.50 N/mm² can be found in
[15] for shear tests of glulam beams. Fracture energy can
be assumed for Mode II with a value from GFII = 300 to
Client fm,mean [N/mm²] f*m,mean without outliers 500 J/m² depending on the annual ring pattern [13]. The
1 - Series 1 44.2 44.2 consideration of the softening behaviour is the
mechanically correct assumption for this stress
1- Series 2 40.6 40.6 component. Only rolling shear failure has been observed
2 56.5 outlier, rejected so far in tests. A propagated shear strength together with
shear softening failure due to shear parallel to grain was
3 - Series 1 37.2 37.2 observed in the numerical model. It can be supposed that
3 - Series 2 39.4 39.4 a higher local shear strength can be applied in the
vicinity of high shear stresses. A reason for this can be
4 37.4 37.4 found in a volume effect which can be observed in
5 30.7 outlier, rejected experiments [20]. As shear in grain direction do not
provide the primary failure mode for CLT panels the
mean 40.8 39.8 shear softening was replaced by a simpler elastic plastic
stress strain relation. The maximum shear stress remains
An elastic plastic behaviour can be used for compression at 4.5 N/mm². This simplification can be justified for
parallel to grain. A simple bi-linear elastic plastic stress CLT panels because failure can be commonly identified
strain function can be assumed for simplification. The for rolling shear stresses.
plastic plateau can be chosen with 37 N/mm². This value
is based on compression tests of spruce parallel to grain Rolling shear strength:
in [14] as a mean value over all results. Rolling shear strength is the dominant shear component
for CLT under bending concerning its utilisation. A
Tension and compression perpendicular to grain realistic strength value should consider the stress
direction: interaction as it is described in [16] and [17]. Without
The tension strength perpendicular to grain is a rather considering stress interaction it is well known that the
sensitive and unreliable number and should be chosen to rolling shear strength is equal to the tension strength
perpendicular to grain. A hint for a reliable fracture Formulas for plain stress are derived in [16]. The
energy value for present Mode III can be found in [18] resulting stress strain interaction is shown in Fig. 10.
with the indication that this value GFIII is approximately Timber can be assumed as a natural fibre reinforced
“over twice as high as mode I in both the RL and TL composite and therefore the above mentioned
directions”. Values for Mode I can be found in [13] with conclusions can be transferred to it. A recent work at TU
GFI=300 J/m² for tension perpendicular to grain. No Graz [17] established by Kren dealt with stress
difference was made in [13] for the RL and TL direction. interaction in material with multi-plasticity planes. The
A value of 260 J/m² for GFI can be found in [18] and is starting point of this work was given by the idea to
listed in Tab. 5 for spruce based on a publication of improve the strength predictions for materials like
Reiterer et al. Concluding two different values were used timber. The embedded examples show various
in the worked out numerical investigations for GFIII: once applications of strength predictions of timber material.
500 J/m² and 800 J/m²; both laying in the cited range. The stress interaction between stresses perpendicular to
fibre direction and the shear stress is simplified to a bi-
Summary linear interaction relation figure [11]. The regarded shear
All material parameters which were used in the stresses are always in fibre direction. The fixed value n1
materially non-linear FE model are summarized in was chosen with 3.0 and n2 with 1.1 [17].
Tab. 5.
Table 5: Summary of the used strength values and the
considered post linear behaviour in the materially
non-linear FEM analysis

ft0,mean=39 N/mm² GFI=1650 J/m², tension softening

fc0,mean=37 N/mm² elastic-plastic
ft90,mean=1.5N/mm² GFI=300 J/m², tension softening
fc90,mean=3.5N/mm² elastic-plastic
fv0,mean=4.5 N/mm² GFII=300-500 J/m², stress softening
fv90,mean=1.5 N/mm² GFIII=500-800 J/m², stress softening

As elastic plastic behaviour was assumed for tension Figure 10: Plain stress interaction for fibre reinforced
materials acc. to Puck [16]
perpendicular to grain and shear in the grain direction for
simplification purposes. The modified material
parameters for these stresses can be found in Tab. 6.
Table 6: Summary of the used strength values and
considered post linear behaviour for the simplified
material behaviour

ft90,mean=0.5 N/mm² elastic-plastic (for simplification)

fv0,mean=4.5 N/mm² elastic-plastic (avoiding of shear failure)

4.2.2 Stress Interaction

The aim of stress interaction relations is given by the

idea to include positive and/or negative effects of Figure 11: Plain stress interaction between stresses perpen-
existing stresses on the regarding strength values. Thus dicular to grain and shear stresses acc. to Kren [17]
the individual strength values are no longer constants but
may vary with stress variation in the specific material The above mentioned simplifications cannot be adopted
point. Puck [16] studied various stress interaction directly for full 3D stress concentrations. Thus stress
formulas for fibre reinforced materials. Within them transformation (Fig. 12) with an additional unknown
stresses in fibre direction and stresses in the embedding rotation angle θ about the fibre direction has to be
matrix of the fibre reinforced composite develop. Puck performed. In order to determine the angle θ it can be
derived a plain stress interaction relation establishing assumed that failure happens in a plane where the stress
three different failure modes: utilization is maximized. This condition is usually be
- Mode A if tension stresses exist, evaluated by numerical methods [16]. Alternatively a set
- Mode B is dominant, if compression stresses are of various failure planes for different rotation angle θ
moderate and leads to the same result.
- Mode C is the governing failure mode in case of high Similar interaction relation as shown in Fig. 10 and
compression stresses perpendicular to fibre direction. Fig. 11 for normal stresses perpendicular to the fibre
direction and shear parallel to the fibre direction can be
obtained for normal stresses and rolling shear stress.
General analytical solutions cannot be derivated
anymore but solutions can be received by numerical
methods. The results and a comparison with the
interaction between normal stress and shear stress in
fibre direction are shown in fig. 13 on basis of equations
according to PUCK and in fig. 14 on those according to
Kren.

Figure 13: Stress interaction between stresses perpendicular

to grain and rolling shear acc. to equations of
PUCK [16]

Figure 12: Stress transformation with an additional unknown

rotation angle θ acc. to Kren [17]

The present problem of a concentrated load introduction

perpendicular to plane is dominated by the occurrence of
high bending stresses in fibre direction (σ11). The rolling
shear and normal stresses perpendicular to grain are
appear in the adjacent layers. These stresses are either
tension or compression stresses and may be denoted as Figure 14: Stress interaction between stresses perpendicular
σ22-stresses. The notation of rolling shear stresses may to grain and rolling shear acc. to equations of
be τ23. Additionally high compression stresses KREN [17]
perpendicular to grain due to the local load introduction
are acting in the third direction (σ33). The local load 4.3 FEM Model
introduction is highly predominated by the rolling shear
strength and the idea of handling this fact is that A full 3D solid FE-model was established in the FE-
especially the σ33 stresses are always compression program ABAQUS. The element type used was the
stresses and therefore an improvement of the rolling linear element C3D8R element with reduced integration.
shear strength could explain the high rolling shear The size of one FE element was approximately cubic
strength values which were observed in experimental test with a volume of about 1 cm³. The test configuration had
evaluation [1, 5, 6]. two symmetry-axis therefore only ¼ of the system was
modelled with FE. Only the test configuration with 5
layered CLT elements was investigated numerically. The
test configuration with 7 layers has to be treated
analogous. The established system with 45000 elements,
49419 nods and 148257 unknowns is depicted in Fig. 15.
As can be seen there a discrete volume is modelled with
tension softening behaviour for stresses in fibre
direction. This volume is shown red in Fig. 15. In
opposition to that shear softening of rolling shear is
modelled for all elements. The mean values of all
strength properties have to be introduced with values
described in the previous section of this paper. A
different material behaviour has to be taken into account
for different stresses and sign of stresses. This specific
material behaviour was implemented using the
parameters of Tab. 5 and Tab. 6 in different user
subroutine written in the FORTRAN programming
language.

Figure 16: Numerical results of the FE model without stress

interactions and comparison with experimental data

The shear softening behaviour on rolling shear at the

Figure 15: Numerical FE model for configuration 1
material point level delivers an apparent global elastic
plastic load displacement behaviour in shear. If the
The effects from the poison numbers were ignored in the rolling shear behaviour at the material point level is
implementation of the different subroutines for assumed to be elastic plastic, the global load
ABAQUS: displacement curve show a steady increasing path based
The simplest subroutine implementation can only on numeric results. Shear softening behaviour of shear
describe tension softening in fibre direction whereas all parallel to grain (V3) was rejected, because numerical
other stress strain relations behave perfectly elastic results show shear failure, where experimental data show
plastic behaviour (V1). The consideration of an shear definitely no failure due to shear in fibre direction.
softening effect for rolling shear replaces the elastic In a second step a comparison with stress interactions
plastic behaviour in the user subroutine version 2 (V2). included can be seen in Fig. 17. Both figures show that
In a next step the shear softening in the fibre direction the influence of pure stress interaction is relatively small
replaces the corresponding elastic-plastic behaviour in whereas different values for the fracture energy lead to
the user subroutine version 3 (V3). As mentioned this different results (V4).
model leads to dominant failure due to shear in fibre
direction. Solutions computed with subroutine V3 were
rejected for that reason. More investigations are
necessary if shear softening should be included in the
numerical model. The interaction relations were
implemented in the subroutine of version 2. As all
coupling terms between the stresses were neglected the
acceptable rolling shear stress on basis of the actual
stress components can be computed and interpreted as a
temporary valid rolling shear strength in the actual
iteration step. The difference between the curves in
Fig. 13 and Fig. 14 is that only one normal stress
perpendicular to grain was considered for interaction
curves in Fig 13 and Fig. 14 whereas all remaining 4 Figure 17: Numerical results of the FE model with stress inter-
stresses were considered in the user subroutine to actions and comparison with experimental data
compute the remaining allowable 5th stress component
(rolling shear stress). The influence of the normal A comparison between the numerical curves in Fig. 16
stresses in fibre direction σ11 to the stress interaction was and Fig. 17 shows almost identical curves. The
neglected. More effect of the stress interaction on the computed values show only a slight increase of 2% in
allowable rolling shear stress can be found in the model case of stress interactions. This difference is small
of KREN [17] than in the interaction equations of PUCK compared to the influence of the GFIII value which is
[16] (Fig. 13 and Fig. 14). The interaction equations of assumed on basis of recommendations found in [18].
the PhD work in [17] were used for the implementation
in the present user subroutine.

4.4 Results and comparison to the experimental data

5 Summary and conclusions
In a first step a comparison between the measured load The numerical results show good agreement with the
displacement behaviour of the tests and the numerical experimental data. Influence of stress interaction can be
values computed with the user subroutine V1 and V2 can neglected. Major influence in CLT structures in such
be seen in Fig. 16. As two values for the fracture energy loading situations can be seen in the post linear behavior
GFIII are considered (500 vs 800 J/m²) two results for V2 of the rolling shear. Here in the numerical computations
were computed. a linear shear stress softening was introduced. This linear
softening could be replaced by other functions like Verification Procedures for Cross Laminated
exponential one. The area under the softening curve Timber, Research Report, holz.bau forschungs
represents the fracture energy GFIII. Here research is still gmbh, Institute for Timber Engineering and Wood
missing to find a reliable value. A volume effect in shear Technology, Graz University of Technology.
failure leads to higher stress failure initiation values as [10] Bogensperger T, Augustin M, Schickhofer G.
the common well known shear strength values are in (2011): Properties of CLT-Panels Exposed to
general. This volume effect is well known. As shear Compression Perpendicular to their Plane, CIB
failure is not the dominant failure mode in the W18, 44th Meeting, Alghero, Italy
investigated structure, the softening behavior was [11] Neuhaus, F.-H., 1981: Elastizitätszahlen von
replaced by a simplier elastic-plastic behavior. Fichtenholz in Abhängigkeit von der
The calucated results are a supplement to the findings in Holzfeuchtigkeit (Elastic constants of wood
[1], where a relative high shear strength for concentrated depending on wood moisture). Tech. report,
loads was poroposed. Numerical results in figure 16 and Institut für konstruktiven Ingenieurbau, Ruhr-
17 show, that this high shear strength can be explaned Universität Bochum (in German).
with the relative mild shear stress softening behaviour [12] Hofstetter K., Hellmich C., Eberhardsteiner J.:
with relative high fracure energy GFIII. Development and experimental validation of a
continuum micromechanics model for the
ACKNOWLEDGEMENT elasticity of wood, European Journal of Mechanics
A/Solids 24 (2005) 1030-1053
The research work within the FFG COMET K-project [13] Mackenzie-Helnwein P., Müllner . H. W.,
“focus_sts” is financed by the Austrian Funding Agency Eberhardsteiner J., Mang H. A.: Analysis of
FFG and by contributions of local governments as well layered wooden shells using an orthotropic elasto-
as economical and scientific project partners. plastic model for multi-axial loading of clear
spruce wood, Comput. Methods Appl. Mech.
Engrg. 194 (2005) 2661-2685.
REFERENCES [14] Ruli A. Längs und quer zur Faserrichtung auf
[1] Bogensperger T. , Joebstl R. A.: Concentrated load Druck beanspruchtes Brettschichtholz, Diploma
introduction in CLT, Proceedings to the 2nd Thesis, TU Graz, 2004 (in German)
INTER meeting, Šibenik, 2015 [15] Schickhofer G.: Determination of Shear Strength
[2] Bogensperger T., Silly G.: Zweiachsige Values for GLT using Visual and Machine Graded
Lastabtragung von Brettsperrholzplatten, Ernst & Spruce Laminations, CIB W18, 34th Meeting,
Sohn publisher for architecture and technical Venice, Italy, 2001
science GmbH & Co. KG, Berlin. Bautechnik 91 [16] Puck A.: Festigkeitsanalyse von Faser-Matrix-
issue 10, 2014 (in German) Laminaten - Modelle für die Praxis, Carl Hanser
[3] Silly G.: Numerische Studien zur Drill- und Verlag München Wien, 1996, ISBN 3-446-18194-
Schubsteifigkeit von Brettsperrholz (BSP), 6 (in German)
diploma thesis, Graz University of Technology, [17] Kren B.: Plastisches Versagen strukturierter
2010 (in German) Materialien: Verallgemeinerte Modellbildung,
[4] Bogensperger T. , Joebstl R. A.: Concentrated load rechnerische Umsetzung & praktische Anwendung
introduction in CLT, presentation at EUROMECH auf Holz, Dissertation, TU Graz, 2014 (in
Colloquium, Dresden, 2015 German)
[5] Bogensperger T. , Wallner B., Augustin M.: [18] Smith I., Landis E., Gong M.: Fracture and
Durchstanzen von Brettsperrholzplatten, Research Fatigue in Wood, Wiley 2003, ISBN 0-471-48708-
Report to project focus_sts 2.2.3_2, holz.bau 2
forschungs gmbh Graz, 2015 (in German) [19] Stuefer A.: Einflussparameter auf die
[6] Mestek P.: Punktgestützte Flächentragwerke aus Querzugfestigkeit von BSH-Lamellen, Diploma
Brettsperrholz (BSP) – Schubbemessung unter Thesis, TU Graz, 2011 (in German)
Berücksichtigung von Schubverstärkungen, Ph.D., [20] Lackner H.: Festlegung einer Prüfkonfiguration
Munich University of Technology, 2011 (in für die Ermittlung der Schubfestigkeit von
German). Vollholz sowie Analyse des Größeneinflusses,
[7] Jöbstl R. A., Bogensperger T., Moosbrugger T., Diploma Thesis, TU Graz, 2011 (in German)
Schickhofer G.: A Contribution to the Design of
Cross Laminated Timber, CIB W18, 39th Meeting,
Florence, Italy, 2006
[8] Schickhofer G., Bogensperger T., Moosbrugger
T.: BSPhandbuch, Holz-Massivbauweise in
Brettsperrholz, Nachweise auf Basis des neuen
europäischen Normenkonzepts, publisher of Graz
University of Technology, 2011, ISBN 978-3-
85125-109-8 (in German)
[9] Bogensperger T., Silly G., Schickhofer G. (2012):
Comparison of Methods of Approximate