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

net/publication/304355276

Evaluation of the Mechanical Properties of Cross Laminated Timber with

Elementary Beam Theories

Article  in  Construction and Building Materials · September 2016

DOI: 10.1016/j.conbuildmat.2016.06.082

CITATIONS READS

51 5,392

5 authors, including:

Ioannis P. Christovasilis Michele Brunetti

Aether Engineering Italian National Research Council
25 PUBLICATIONS   430 CITATIONS    68 PUBLICATIONS   648 CITATIONS   

SEE PROFILE SEE PROFILE

Maurizio Follesa Michela Nocetti

Dedalegno, Firenze, Italy Italian National Research Council
32 PUBLICATIONS   448 CITATIONS    61 PUBLICATIONS   638 CITATIONS   

SEE PROFILE SEE PROFILE

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

Forest Therapy View project

Wood Production and Promotion View project

All content following this page was uploaded by Ioannis P. Christovasilis on 01 August 2016.

The user has requested enhancement of the downloaded file.

Paper Title: Evaluation of the Mechanical Properties of
Cross Laminated Timber with Elementary Beam Theories

Authors: Ioannis P. Christovasilis, Michele Brunetti,

Maurizio Follesa, Michela Nocetti, Davide Vassallo

Journal: Construction and Building Materials

DOI: 10.1016/j.conbuildmat.2016.06.082

Accepted for publication on 15-06-2016

Authors’ pre-publication version

Free access link until 12 of August, 2016, for interested readers by

clicking the following link
http://authors.elsevier.com/a/1TG1h3O1E14sbS

 Evaluation of the Mechanical Properties of
Cross Laminated Timber with Elementary
Beam Theories

Christovasilis, I.P.a
a Aether Engineering s.a.s., Via Quintino Sella 6/A, 50136, Florence, Italy
Corresponding author, Email: ipc@aethereng.com

Brunetti, M.b
b CNR-IVALSA, Istituto per la Valorizzazione del Legno e delle Specie Arboree,
via Madonna del Piano 10, I-50019 Sesto Fiorentino (FI), Italy
Email: brunetti@ivalsa.cnr.it

Follesa, M.c
c dedaLEGNO, Via Masaccio 252, 50132, Florence, Italy
Email: follesa@dedalegno.com

Nocetti, M.d
d CNR-IVALSA, Istituto per la Valorizzazione del Legno e delle Specie Arboree,
via Madonna del Piano 10, I-50019 Sesto Fiorentino (FI), Italy
Email: nocetti@ivalsa.cnr.it

Vassallo, D.e
e dedaLEGNO, Via Masaccio 252, 50132, Florence, Italy
Email: vassallo@dedalegno.com

Abstract
This paper presents a study on the assessment of the mechanical properties of Cross Laminated

Timber (CLT) panels based on four-point bending tests. The most recent or well-established analytical

theories have been implemented to estimate stiffness and strength properties under loads

perpendicular or parallel to the principal plane of CLT panels from laboratory tests. The main

objectives were to evaluate each proposed theory in predicting the associated deformation and failure

mechanisms and to assess the reliability of the estimated properties with respect to the expected

values and in terms of consistency among specimens with different layer configurations. The results

2
indicate that the bending response is on average well represented in the implemented theories for

the two cases of loading and in terms of both elastic and strength properties. For loads perpendicular

to plane the characteristic rolling shear strength appears to have a significant variability among the

different layups for all three applied methods, while for loads in plane the consideration of a combined

rolling and torsional shear failure criterion provides more consistent results with respect to a less

rigorous approach.

keywords: cross laminated timber, mechanical properties, bending test, parameter identification,

rolling shear

3
1 Introduction
Cross laminated timber (CLT) is a relatively new engineered wood product that is steadily increasing

its share in the building market for a number of advantages over traditional wood products and over

the main construction materials such as reinforced concrete and steel. In fact, Brandner [1] reports a

15-20% annual increase in CLT production capacity with an estimate of 500,000 m3 overall production

volume for 2012. CLT panels are constructed from layers of structural boards that are glued

orthogonally to each other as shown in Figure 1, forming bidirectional elements that can be used as

wall diaphragms, floor plates or even beam and column elements.

Figure 1: Construction detail of CLT panels

The mechanical behavior of CLT panels is complex, mainly due to the orthogonality in the grain

direction of successive layers and the inherent anisotropy of timber. In general, the structural

response of CLT panels has been systematically studied for the case of loads perpendicular to plane

(vertical or wind forces for floor or wall panels respectively) or for loads in plane (forces on vertically

oriented CLT beams or shear and vertical forces on wall panels).

Most of the testing and certification procedures of CLT panels in Europe, nowadays, are based on four-

point bending tests of strip-shaped CLT specimens, according to EN 408 [2] or based on the recently

published EN 16351 [3], with uniform boundary supports and applied forces along the width. The

mechanical properties that are of primary interest when conducting four-point bending tests of CLT

panels are the mean values of the section bending and shear stiffness (elastic properties) and the

4
characteristic values of bending and shear strength properties. These properties have then to be used

with an applicable analytical theory to yield representative material properties of the boards of the

CLT panels such as the mean modulus of elasticity parallel to grain and the characteristic values of

bending strength and shear strength.

This study aims to present the most recent or well-established analytical theories, as applied within a

numerical framework that implements the Euler-Bernoulli and the exact Timoshenko beam element

[4], for the assessment of the mechanical properties of CLT panels from laboratory tests. The main

objectives are to investigate the level of accuracy of each proposed theory in predicting the associated

deformation and failure mechanisms of a CLT panel and to evaluate the reliability of the estimated

properties with respect to the expected values and in terms of consistency among specimens with

different layer configurations.

1.1 Loads perpendicular to plane (out of plane)

The mechanical response of CLT panels under loads perpendicular to plane is, nowadays, fairly well

understood as it has been the primary field of study for the qualification of CLT elements as structural

components of floor systems. Although CLT panels are bi-dimensional elements that can resist

bending around both orthogonal in plane axes, it is common practice to analyze and design them as

beam elements when used as floor or roof plates because they typically have a maximum width of 2.5

meters, for ease of transportation, and the connection of two adjacent panels does not transfer

bending moments.

Various theories have been proposed, mainly in Germany and Austria, for the analysis of CLT panels

under loads perpendicular to plane and the majority focuses on the calculation of the effective

bending and shear stiffness based on the layer characteristics (thickness and grain direction). In

chronological order, according to the references provided by Thiel and Schickhofer [5] for publications

in the German language, Kreuzinger in 1999 presented the Shear Analogy method and Blass and

Gorlacher in 2003 suggested the Modified Gamma method, which stems from the Mechanically

5
Jointed Beams Theory or Gamma method, available in Annex B of Part 1-1 of Eurocode 5 [6]. Finally,

Schickhofer et al. proposed the Timoshenko beam theory in 2009 as a simplified alternative.

The Modified Gamma method is probably the most common approach in Europe. It can be

implemented with the Euler-Bernoulli beam element as no shear deformations are considered, but

accounts indirectly for them by calculating the effective bending stiffness based on the efficiency of

the connection between the longitudinal layers (those with grain parallel to the beam length). This

connection is provided by the shear stiffness of the internal transverse layers (those perpendicularly

oriented to the beam length). The method is applicable to a 3- or 5-layer CLT panel but can be

extended to 7- and 9-layer panels, as well. The effective bending stiffness EIeff,Gamma is calculated as:

EIeff ,Gamma   Ei  bi  hi3 / 12     γi  Ei  bi  hi  zi2  (1)

i i

where Ei is equal to the modulus of elasticity parallel to grain E0 for longitudinal layers while for

transverse layers Ei is considered in this method equal to zero; bi is the width, hi the thickness and zi

the distance from the centroid of the ith layer to the centroid of the cross section; γi is the connection

efficiency factor that is nonzero only for longitudinal layers and equal to unity for the middle layer.

The latter is computed as:

1
 π 2  Ei  bi  hi 
γi   1  2  (2)

 L eff   G 90 ,j  b j  / h j


where Leff is the effective length of the beam; j refers to the transverse layer connecting the ith layer

with the central layer and G90 is the shear modulus in the plain perpendicular to grain or rolling shear

modulus. As it can be observed from Eq. (2), the connection efficiency factor depends on the effective

length of the beam, that is the length of the beam between the two zero-moment points (inflection

points). Figure 2 illustrates qualitatively the normal and shear stress diagrams in a CLT section for

various values of the connection efficiency factor γ. As γ approaches unity, the longitudinal layers work

as part of the whole cross section, while, as it approaches zero, they work as independent layers under

bending.

6
(a) (b) (c) (d)

Figure 2: Normal (left) and shear (right) stress diagrams in a 5-layer CLT section (coloured hatched layers
represent longitudinal boards) (a) for a connection efficiency factor (γ) equal to (b) 1.0, (c) 0.5, and (d) 0.1

As for the Timoshenko beam theory, the bending stiffness is calculated as shown in Eq. (1) considering,

though, γi equal to unity for all layers and Ei equal to E90 for transverse layers, where E90 is the elastic

modulus perpendicular to grain. Thus, the stress profiles are similar to those shown in Figure 2a. The

effective shear stiffness GAeff,Timo is calculated as:

GAeff ,Timo  κ  GA  κ   Gi  bi  hi  (3)

i

where Gi is equal to the shear modulus in planes parallel to grain G0 for longitudinal layers and to the

rolling shear modulus G90 for transverse layers; κ is the shear correction factor that, for most common

layer configurations, lies between 0.20 and 0.25 [5] and is computed as:

EIeff ,Timo2
κ (4)
S2  z,E  z  
GA   dz
z
Gz  b z 
where S is the first moment of inertia of the integrated section. This approach is simpler to implement

and more general than the Modified Gamma method, since it can be applied in case of any number of

layers and it is applicable for multi-span beams, as the bending stiffness does not depend on the

effective length of the beam.

Perhaps the most accurate method, but also the most computationally intensive, is the Shear Analogy

method, which is based on the combination of two beams constrained to have the same vertical

deflection at any point along their length. The first beam (A) accounts for the bending stiffness of each

layer independently, which is equal to the first part of Eq. (1) and Ei equal to E90 for transverse layers.

The bending stiffness of the second beam (B) is equal to the second part of Eq. (1) with the same

7
modification on Ei and considering γi equal to 1 for all layers, thus the sum of the two bending

stiffnesses is equal to the one of the Timoshenko beam theory. Shear deformations are neglected for

beam A while for beam B the shear stiffness is calculated as:

1
2
 h1 n 1
 h  hn 
GA eff ,B ,SA  α CEN    i   (5)
 2  G1  b 1 i 2  G i  b i  2  Gn  b n 
where αCEN is the distance between the centroids of the lower and the upper layer. This approach is

as general as the Timoshenko beam theory, but can simulate more complex phenomena because the

two beams interact with each other through an exchange of a distributed load perpendicular to their

length even when the external force is zero. For interested readers, more information on the specifics

of the Shear Analogy method applied to simple examples can be found in the Canadian CLT Handbook

[7].

Other analytical studies include the composite theory or Kappa method [8] and the work on the design

and system effect of CLT panels [9] that has been extended in [10] and [11] on the development of a

bearing model with a reference cross section, similar to what has been done for Glued Laminated

Timber (GLT) products in EN 1994 [12]. Finally, a laminated plate theory has been proposed in [13] for

accurate results in deformation as well as in stress components, for thick CLT plates under distributed

and concentrated forces. Many of the studies mentioned above include also experimental results

mainly from bending tests of CLT specimens. Alternative methods have been also explored as found

in [14], where a solid automated procedure is presented for the determination of elastic properties of

CLT panels based on vibration tests and modal analysis, and in [15] on the bending stiffness of CLT

panels under loads perpendicular to plane derived from bending tests or by modal analysis. The latter

study showed that specimen strips wide at least two boards in longitudinal layers are appropriate in

evaluating the stiffness properties of a rectangular panel.

1.2 Loads in plane

Studies on the mechanical behavior of CLT panels under loads in plane have focused on the shear

transfer mechanism and more specifically on the evaluation of an equivalent shear modulus with

8
reference to the gross cross section, and on the identification of deformation mechanisms and failure

criteria for shear strength analysis and design. The flexural response is dominated by the layers

running along the length of the beam. A typical rectangular cross section with height equal to the

height of the prototype and width equal to the sum of the thickness of each longitudinal layer t0 can

be considered for the bending stiffness and the moment induced stress components. The shear-

transfer mechanism is characterized by different stress profiles in the longitudinal and transverse

layers since the boards of each layer are not bonded to each other along the narrow edges. Even in

the case of bonded edges, cracks typically appear with time in planes parallel to the edges, so that

longitudinal (horizontal) layers transfer shear forces through the vertical cross sections and transverse

(vertical) layers transfer shear forces through the horizontal cross sections. Thus, there is an additional

torsional moment transferred in the crossing areas of orthogonally bonded lamellae and its magnitude

is linearly related to the thickness and stress value of the two layers. The glued connections are

typically very effective and it is rare that failure of a CLT panel will depend on the actual bond strength

of the connection.

Bogensperger et al. [16] proposed a verification method, named Bogensperger theory in this paper,

for CLT panels under loads in plane extending upon the investigations of Joebstl et al. [17]. This study

considers two deformation and failure mechanisms under shear stresses of each layer and under

torsional shear between layers. The equivalent shear modulus is expressed as a combination of two

springs in series as:

2 p2
  tm  
Geq,B  G0  1  6  p1     (6)
  αLB  
 
where G0 is the shear modulus in the planes parallel to grain, tm is the mean thickness of all layers, αLB

is the width of the lumber boards, typically 150 mm, and (p1, p2) are parameters with initial theoretical

values of (1,0) and can be calibrated according to test or finite element data on various configurations.

Flaig and Blass [18] presented a detailed study, named Flaig and Blass theory in this paper, of the

shear stiffness and strength of CLT beams loaded in plane. Three deformation mechanisms are

9
considered; one in shear of the lamellae and two in shear and torsional shear of the crossing areas.

The equivalent shear modulus is expressed as:

1 1
 1 1   1  K α 2 n m2  
Geq,F       1 /  SL LB  CA  2   (7)
 G0 
 G 0 Geff ,CA    5 t gross m  1 
where KSL is the slip modulus of crossing areas, nCA is the number of glue lines between longitudinal

and transverse layers within the element thickness, m is the number of longitudinal lamellae within

the beam height and tgross is the thickness of the panel.

Vessby et al. [19] investigated the in plane response of CLT panels and verified the high strength and

stiffness provided by glued connection with respect to mechanical ones. Andreolli et al. [20]

conducted an experimental investigation on the behavior of CLT panels under in plane loads and

reported the stress components for the Bogensperger theory. Brandner et al. [21] extended previous

work on CLT panels under in plane loads and talked about test configuration and influencing

parameters.

2 Materials and methods

2.1 Specimens and testing program
The specimens tested were CLT panels made of spruce (Picea abies) boards, strength graded visually

as C24 according to DIN 4074-1 [22]. The boards were finger jointed and face glued with a

polyurethane adhesive. Three different layups were considered as shown in Table 1: (i) 3-layer panels

made of 19 x 150 mm boards, (ii) 3-layer panels made of 44 x 120 mm boards, and (ii) 7-layer panels

made of 44 x 120 mm boards. The specimens for each layup were CLT panels cut in strips so that

layer’s parallel to the test span consisted of at least two laminations, in accordance with EN 16351 [3].

The testing apparatus consisted of a Zwick-Roell testing machine with a load cell accuracy of 1% and

the deformations were measured by means of inductive displacement transducers HBM with an

accuracy of 0.1%. The tests performed were four-point bending tests, both to derive bending stiffness

(Modulus of Elasticity – MoE tests) and bending strength (Modulus of Rupture – MoR tests) in the

10
moment-critical configuration and to derive the shear strength (MoR test) in the shear-critical

configuration. The moment-critical configuration was defined according to EN 408 [2], with a length

span, based on the vertical centerlines of the two supports, of 18 times the cross section height of the

specimen, symmetrically divided in three parts, so that the distance between the external support and

the nearest load application point was 6 times the height, as shown in Figure 3. The local deformations

were measured in the neutral axis on both sides of the specimen and the mean of the two measures

were used to calculate the local stiffness that is primarily related to a bending deformation

mechanism. In the same test setup, the total deformations were measured in the central point on the

tension edge and used to calculate the global stiffness that is related to both bending and shear

deformation mechanisms. The load was then applied until failure and the maximum load was

recorded. The shear-critical configuration was adapted to maximize the shear stress in the part

between the support and the load application point: the length span was 9 times the cross section

height of the specimen, symmetrically divided in three parts. The load was applied until failure and

the maximum load was recorded. Table 2 lists the characteristics of the specimens tested in the two

different configurations for loads applied both perpendicular and parallel to plane of the panel.

Table 1: Investigated panel layups

Layer thicknesses * Lamination width
Layup
(mm) (mm)
57-3layer 19-19-19 150
132-3layer 44-44-44 120
298-7layer 44-44-44-34-44-44-44 120
* Bold numbers indicate layers with grain direction parallel to the longitudinal axis

Figure 3: Moment-critical test configuration according to EN 408. h = specimen height; a = distance between
external support and load application points; d = distance between load application points; Lsp = test span;
s = reference distance for local deformation measurement.

11
Table 2: Specimen number and dimensions for each test configuration. N = number of specimens; h = height; b
= base; l = length of the specimen; a = distance between external support and load application point; Lsp = test
span. See also Figure 3
h b l a Lsp
Layup Configuration N
(mm) (mm) (mm) (mm) (mm)
Moment-critical / out of plane 7 57 300 1200 342 1026
57-3layer

Shear-critical / out of plane 7 57 300 650 171 513

Moment-critical / in plane 7 300 57 6000 1800 5400
Shear-critical / in plane 7 300 57 3000 400 2700
Moment-critical / out of plane 7 130 250 2500 780 2340
132-3layer

Shear-critical / out of plane 7 130 250 1500 396 1188

Moment-critical / in plane 7 250 130 5000 1500 4500
Shear-critical / in plane 7 250 130 2500 500 2250
Moment-critical / out of plane 7 300 300 6000 1800 5400
298-7layer

Shear-critical / out of plane 7 300 300 3200 900 2700

Moment-critical / in plane 7 300 300 6000 1800 5400
Shear-critical / in plane 7 300 300 3000 900 2700

2.2 Beam theories and numerical modelling

Three theories (the Modified Gamma method, the Timoshenko beam theory and the Shear Analogy

method) were considered for the case of loads perpendicular to plane, while the Bogensperger and

the Flaig and Blass theories were considered for the case of loads in plane. A numerical framework

was implemented in a computer program on the MATLAB platform [23] in order to calculate the

response of the CLT panels under external loads in terms of displacements and section forces (moment

and shear). Each specimen was modelled with three beam elements in series representing the three

parts of the beam; the two external parts which are under constant shear and the middle part which

is under constant moment. The exact Timoshenko beam element [4] was used in this numerical

implementation to include shear deformations where applicable. The 4x4 stiffness matrix K is equal

to:

 12 6  Le 12 6 Le 
 2
2  L e  (1  6 / α) 
2
4  L e  (1  3 / α) 6  L e α  GA eff  L e2 / EIeff
K  α  EIeff / (β  L e3 )   , (8)
 12 6  L e  β  α  12
 
sym 4  L e2  (1  3 / α)

where Le is the length of the element, EIeff is the effective bending stiffness of the cross section and

GAeff is the effective shear stiffness of the cross section. This element is super-convergent and does

12
not exhibit shear-locking, thus, no meshing of the beam elements is required. When shear

deformations were neglected, the stiffness matrix was reduced to the well-known 4x4 stiffness matrix

of the Euler-Bernoulli beam element by calculating its limit for GAeff approaching infinity.

Each beam theory was implemented by assigning to the stiffness matrices the values for the effective

bending and shear stiffness of the cross section, as shown in Section 1.1. For the implementation of

the Shear Analogy method, each of the three parts of the beam was modelled with a combination of

two beam elements in parallel. The two beams in parallel were then meshed with a length equal to a

quarter of the height of the respective cross section and the vertical degrees of freedom of the nodes

lying on the same geometrical points were restrained to have the same displacements.

Regarding the estimation of the elastic properties, single- and dual-parameter identification

procedures were performed using the nonlinear optimization toolbox available in MATLAB. For each

case, an objective function was defined based as the difference between measured and predicted

displacements. The optimum parameters that minimize the objective function were then calculated

based on a given convergence criterion. For the single-parameter identification, three estimates of

the modulus of elasticity parallel to grain were calculated for each test and for each theory; the first

one based on the global stiffness measurement, the second one based on the local stiffness

measurement, and the third one based on both measurements. For the latter case, equal weights for

the percentage differences between predicted and measured values were defined in the objective

function. For the dual-parameter identification, an additional elastic property, as explained later,

other than the modulus of elasticity was selected and the optimum pair of elastic properties was

computed based on the pair (global and local) of stiffness measurements.

The mechanical properties calculated from the experimental results were evaluated with the use of

verification indices. These indices were computed by dividing each property value, such as the mean

modulus of elasticity or the characteristic bending and shear strength, yielded from the application of

a given theory on the experimental results, with the corresponding expected value based on the

13
material properties of the boards constituting the CLT panels. Indices greater than unity indicate that

the estimated property values verify the expected property, while those below unity fail to verify it.

For elastic properties, expected values were given as mean estimates while for strength properties

they were expressed as 5th-percentile estimates. Normal and shear strength verification indices are

presented for both test configurations for comparison and verification of the controlling failure

mechanism (in bending or in shear).

Finally, all samples of data or numerical estimates were considered in this study to follow a lognormal

distribution. So, the median value, called and labeled as “Mean” in the text, represents the geometric

mean and the coefficient of variation (CoV) is the dimensionless standard deviation of the natural

logarithms of the sample values. Characteristic 5th-percentile values (Char.) are calculated according

to EN 14358 [24] with Kγ = 2.25 for seven specimens and a minimum CoV of 0.05.

2.2.1 Loads perpendicular to plane

The evaluation of elastic and strength properties from the experimental results requires first the

determination of redundant elastic properties, which are in order of significance: the rolling shear

modulus G90, the shear modulus in planes parallel to grain G0 and the modulus of elasticity

perpendicular to grain E90. Since the base material was strength graded C24, expected values of the

elastic properties can be found in EN 338 [25] (E0 = 11000 MPa, E90 = E0/30 = 370 MPa, and G0 = E0/16

= 690 MPa). The rolling shear modulus is perhaps the property that has the highest variability because

of the influence of the orientation of the growth rings. Fellmoser and Blass [26] studied the values of

rolling shear modulus reporting a value between 40 and 80 MPa. Ehrhart et al. [27] presented a

comprehensive study in this area identifying sawing pattern and board geometry as the main

parameters that affect the rolling shear properties. For spruce boards, a mean rolling shear modulus

of 100 MPa is recommended for boards with a width-to-thickness ratio (αLB / tLB) greater than four,

with a linear relationship for ratios less than four given by (30 + 17.5*(αLB / tLB)) MPa. Similarly, for the

characteristic rolling shear strength a value of 1.4 MPa is proposed for a width-to-thickness ratio

greater than four, with a linear relationship for ratios less than four given by (0.2 + 0.3*(αLB / tLB)) MPa.

14
Expected mean values for elastic and strength properties have been recently proposed by Schickhofer

et al. [11] for classes specifically defined for CLT panels, based on tensile properties (T-classes) of the

lamellae, with a 5-layer bending reference cross section of 600 mm in width by 150 mm in depth. Since

the experimental study presented herein was part of the certification process of CLT panels of a

specific CLT producer, it was preferred to retain the actual strength class profile of the material tested

(C24), according to the production practices of the manufacturer.

In this study, the redundant elastic properties were selected to be linearly related to E0 (E90 = E0/30,

G0 = E0/16, G90 = G0/10) as this assumption predicts stress profiles that are dependent only on the

layup of the cross section. The first two proportionalities were derived directly from EN 338 [25] while

the third was based on common engineering practice. For dual-parameter identification the rolling

shear modulus G90 was selected as the second parameter, with G0 = 10·G90 for the redundant shear

modulus in planes parallel to grain, with an upper limit of G90 = E0/30.

The stiffness verification indices ρPERP,E were calculated as the ratio of the mean modulus of elasticity

of each sample over the expected value of 11000 MPa. The normal strength verification indices ρPERP,σ

were computed by dividing the characteristic stress obtained for each layup σm,k by the characteristic

bending strength fm,k = 24.0 MPa. The former was adjusted for the size effect with a factor of kh as

proposed in [10] with tgross given in mm.

σm,k σm,k
ρPERP,σ   0.1 (9)
k h  fm,k  150 
   fm,k
 t gross 
The shear strength verification indices were computed for each case as the maximum of two

verifications obtained for shear in planes parallel to grain with characteristic strength fv,k = 4.0 MPa

[22] and for rolling shear with characteristic strength fvR,k = 0.8 MPa, considered twice the

characteristic strength in tension perpendicular to grain [22].

 τ τ 
ρPERP,τ  max  v,k , vR,k  (10)
 fv,k fvR,k 

15
where τv,k and τvR,k are the characteristic shear stresses for planes parallel and perpendicular to grain,

respectively.

2.2.2 Loads in plane

The basic redundant elastic properties needed for the case of loads in plane are related to the shear

stiffness in planes parallel to grain and the torsional stiffness among the layers in the direction

perpendicular to the CLT panel. For the shear modulus a value of G0 = E0/16 was considered as for the

case of loads perpendicular to plane. The method of Bogensperger utilizes the calibrated parameters

(p1, p2) shown in Eq. (6) to calculate the torsional shear stiffness. In this study, the values of (p1, p2) =

(0.5345, 0.7947) were used for 3-layer panels and the values of (p1, p2) = (0.4253, 0.7941) for 5-layer

and 7-layer panels. For the method of Flaig and Blass a slip modulus of the crossing areas KSL = 5

N/mm2/mm [16] was considered and the ratio E0/KSL was kept constant. A dual-parameter

identification procedure was performed for the Flaig and Blass theory considering the slip modulus of

the crossing areas as a second parameter.

The shear stress resultants for the Bogensperger theory were computed as shown below considering

the x direction along the direction of the beam and the z direction along the vertical direction of the

cross section, as shown in Figure 3:

 3 / 22V 
 τ xz  
 H   tRVSE 
  (11)
 τ  3 / 2  3  V  max  tRVSE  
 tor αLB H   tRVSE 
 

where V is the shear force, H is the height of the cross section and tRVSE represents the thicknesses of

the Representative Volume Sub-Elements (RVSE) as defined in [16]. Note that a multiplication factor

of 3/2 was used to account for the maximum shear stress with respect to the average shear stress,

much like in the case of a rectangular beam.

The shear stress resultants for the Flaig and Blass theory were calculated as:

16
  3/ 2V 
 τ xz,gross  
H   tgross 
 
 3/ 2V 
 τ xz,net  
 H  min   t0 ,  t90  
 
 6V  1 1  
 τ yx       (12)
 αLB 2  nCA  m2 m3  
 qz V  max  t90  / 2 
 τ yz   
 m  αLB  t90  L sup  m  αLB 
 
 τ  3 V   1  1  
tor  
 αLB 2  nCA  m m3  

where t0 and t90 represent the thicknesses of the layers parallel and perpendicular to the direction of

the beam and Lsup is the length along the beam of the support through which the force is applied and

was taken equal to 60 mm. This theory considers additional shear stress resultants in the xy and yz

planes, the first as part of the resisting shear mechanism of the panel and the second as part of the

load transfer mechanism of external vertical loads through the layers oriented in that direction.

The stiffness verification indices ρPAR,E were calculated similarly to the case of loads perpendicular to

plane as the ratio of the mean modulus of elasticity of each sample over the expected value of 11000

MPa. The normal strength verification indices ρPAR,σ were computed by dividing the characteristic

stress obtained for each layup by the characteristic bending strength of fm,k = 24.0 MPa, which was not

adjusted for any size effect. The shear strength verification indices were defined as the maximum of

two verifications for the Bogensperger theory, ρPAR,τ,B, and the maximum of four verifications for the

Flaig and Blass theory, ρPAR,τ,F. For the Bogensperger theory these indices were calculated as:

 τ τ 
ρPAR,τ,B  max (1): xz ,(2): tor  (13)
 fxz,k ftor,k 

where fxz,k = 5.0 MPa and ftor,k = 2.5 MPa. For the Flaig and Blass theory the verifications were

calculated as:

 τ τ τ τ τ τ 
ρPAR,τ,F  max (1): xz,gross ,(2): xz,net ,(3): yx  tor ,(4): yz  tor  (14)
 fxz,gross,k fxz,net,k fvR,k ftor,k fvR,k ftor,k 

17
where fxz,gross,k = 4.0 MPa, fxz,net,k = 10.3 MPa, ftor,k = 2.75 MPa and fvR,k = 1.1 MPa.

3 Results and discussion for loads perpendicular to plane

3.1 Elastic properties
Table 3 presents the experimental results for each layup in a normalized concept, considering the

exact Timoshenko beam theory as given in Appendix A. The first column shows the effective bending

stiffness calculated from the local stiffness measurement as shown in Eq. (A.1), normalized by the

bending stiffness of the gross cross section considered as a single isotropic material with E = 11000

MPa. The second column shows the effective shear stiffness calculated from both measurements as

shown in Eq. (A.2), normalized by the shear stiffness of the gross cross section considered as a single

isotropic material with G = 690 MPa. Finally, the third column shows the effective bending stiffness

calculated from the global stiffness measurement neglecting shear deformations as shown in Eq. (A.3),

normalized by the bending stiffness of the gross cross section.

The effective bending stiffness based on the local measurement is, as expected, higher than the one

based on the global measurement with the infinite shear rigidity assumption, which leads to the

lowest estimate for the effective bending stiffness. There is an exception for test No. 7 which indeed

leads theoretically to negative effective shear stiffness in order to equate the predictions based on

the two measurements. The effective bending stiffness based on the global measurement has lower

CoV, thus, higher reliability than the one based on the local measurement, while the CoV is

significantly increased for the shear stiffness corresponding to equal global and local bending

flexibility, demonstrating that this way of calculating the shear stiffness is not very reliable.

3.1.1 Single-parameter identification

Figure 4 illustrates the probability density functions of the modulus of elasticity parallel to grain

calculated for the three layups separately, as well as for all layups together, with the thee theories

implemented in the present study. The Modified Gamma method returns higher values when the

identification is based on the local stiffness with respect to the global stiffness. On the contrary, the

18
Timoshenko beam theory returns lower values when based on the local stiffness. The Shear Analogy

method returns lower values when based on the local stiffness only for the 57-3layer panels, while for

the other two layups the mean values are very close to the mean values from the global stiffness

estimates, but the CoV is higher.

Table 3: Effective elastic properties expressed as percentage of the stiffness of the gross cross section: results of
MoE tests for loads perpendicular to plane in the moment-critical configuration. EIeff,lo and EIeff,glob = local and
global effective bending stiffness; GAeff = effective shear stiffness; CoV = coefficient of variation; Char. =
characteristic value
Layup 57-3layer 132-3layer 298-7layer
GAeff EIeff,glob GAeff EIeff,glob GAeff EIeff,glob
Test EIeff,loc EIeff,loc EIeff,loc
(EIeff,loc) (1/GAeff=0) (EIeff,loc) (1/GAeff=0) (EIeff,loc) (1/GAeff=0)
No. [%] [%] [%]
[%] [%] [%] [%] [%] [%]
1 101.8 18.9 81.5 90.8 36.2 81.1 91.7 28.9 80.1
2 81.6 145.1 79.6 89.0 37.0 79.8 90.8 27.9 79.1
3 92.3 27.4 79.9 83.9 49.2 77.6 86.6 30.0 76.5
4 83.5 78.0 79.6 96.4 71.4 90.6 88.7 30.9 78.5
5 80.9 65.5 76.6 105.4 21.8 85.7 102.7 18.2 81.7
6 87.5 59.2 81.9 96.2 25.0 81.4 75.6 42.5 69.9
7 65.1 -52.7* 69.0 100.0 26.3 84.6 97.8 17.2 77.7
Mean
84.0 53.2 78.2 94.3 35.2 82.9 90.2 26.8 77.6
[%]
CoV [-
0.14 0.74 0.06 0.08 0.42 0.05 0.10 0.32 0.05
]
Char.
61.3 10.1 68.8 79.2 13.7 73.8 72.2 13.1 69.1
[%]
*
Value excluded from Mean and CoV calculations

Yet, the most striking observation is that the probability density functions based on the global stiffness

assessment are almost identical for all three theories within a given layup, for all three layups. The

CoV of the obtained values for each layup is that shown in Table 3 for the effective bending stiffness

based on the global deformation, but it is remarkable that all three theories predict very similar mean

values.

19
(a) (b)

(c) (d)

Figure 4: Probability density functions of the modulus of elasticity parallel to grain from MoE tests for loads
perpendicular to plane for (a) the 57-3layer group, (b) the 132-3layer group, (c) the 298-7layer group, and (d) all
layups. kGlob = estimate based on global modulus; kLoc = estimate based on local modulus.

The stiffness verification indices ρPERP,E are shown in Table 4 where the values less than unity are

highlighted in gray. It can be seen that the estimates from the Timoshenko beam theory are not

verified when the local deformation measurement is taken into account. Among the three theories it

is clear that the Shear Analogy method returns very similar values for the three layups and it is the

most accurate approach for the best estimate of the elastic properties.

Table 4: Verification indices ρPERP,E for the MoE tests for loads perpendicular to plane in the moment-critical
configuration. kGlob = estimate based on global modulus; kLoc = estimate based on local modulus.
Theory Modified Gamma Timoshenko Shear Analogy
kGlob + kGlob + kGlob +
Layup kGlob kLoc kGlob kLoc kGlob kLoc
kLoc kLoc kLoc
57-3layer 1.01 1.09 1.04 1.02 0.87 0.93 1.02 0.96 0.99
130-3layer 1.08 1.23 1.14 1.09 0.98 1.03 1.09 1.09 1.09
298-7layer 0.99 1.16 1.06 1.01 0.96 0.98 1.01 1.02 1.01
All 1.03 1.15 1.08 1.04 0.94 0.98 1.04 1.02 1.03

20
In order to understand the discrepancies between local and global estimates for the first two theories,

the bending moments at each common point of the two beams implemented in the Shear Analogy

method can be used to calculate a connection efficiency factor γ, similarly to the Modified Gamma

method but for each cross section of the beam, as:

EIA MB  x 
γ x    (15)
EIB MA  x 

Figure 5 illustrates the calculated connection efficiency factor for the three layups based on Eq. (15).

The connection efficiency factor as predicted by the Modified Gamma method is also plotted with

values of 0.79, 0.78 and 0.83 for the 57-3layer, 132-3layer and 298-7layer, respectively.

It can be observed that the minimum values are obtained at the supports and they seem to converge

at a value of 0.5. In the middle part of the beam, which is under pure bending in terms of section

forces and is monitored with the local deformation measurement, the connection factor is parabolic

with maximum values at the center around 0.95 and minimum values at the sides (at a distance of

2.5h from the center) around 0.65. So, it becomes clear that a constant value of the connection

efficiency factor as assumed in the Modified Gamma method and the Timoshenko beam theory is an

approximation that can differentiate the estimated values with respect to the Shear Analogy method.

For example, the Timoshenko beam theory considers the maximum bending stiffness capacity with γ

equal to unity, which naturally leads to lower estimates of the modulus of elasticity for the same

deformation measurement. Similarly, the γ factors considered in the Modified Gamma method are

relatively low estimates for the middle part as γ ranges above 0.9 for 50% of its length and this leads

to higher estimates of the modulus of elasticity.

21
Figure 5: Connection efficiency factor (γ) based on the Shear Analogy method for the moment-critical test
configuration. Lsp = test span; h = specimen height.

3.1.2 Dual-parameter identification

Table 5 and Table 6 list the results for the bending modulus of elasticity and the rolling shear modulus,

respectively, for the single- and dual-parameter identification based on both global and local

measurements. The results indicate that the mean shear moduli evaluated from the dual-parameter

identification are greater than the single-parameter estimates but the CoV is also significantly higher,

especially for the Timoshenko beam theory and the Shear Analogy method. This is a direct

consequence of the experimental data as was shown in Table 3 and discussed therein. The elastic

moduli are in general lower by about 5% of the single-parameter estimates and the verification indices

for the Timoshenko beam theory as shown in Table 5 are not verified. Evaluating the mean shear

moduli listed in Table 6 with the values proposed by Ehrhart et al. [27], it can be stated that the

recommended values in [27] are in the right direction since the first layup, with a width-to-thickness

ratio greater than four exhibits a significantly increased value respect to the one considered in the

single-parameter identification, while this is not the case for the other two layups with width-to-

thickness ratios of 2.7. This is true mostly for the Shear Analogy method, partially for the Timoshenko

beam theory and less for the Modified Gamma method.

22
Table 5: Mean, coefficient of variation (CoV), characteristic value (Char.) and verification index ρPERP,E of the
bending modulus of elasticity estimated with single- and dual-parameter optimization procedures, for loads
perpendicular to plane in the moment-critical configuration.
Theory Modified Gamma Timoshenko Shear Analogy
Layup
Parameters single dual single dual single dual
Mean [MPa] 11474 11403 10254 9671 10866 10243
57- CoV [-] 0.09 0.06 0.11 0.12 0.10 0.20
3layer Char. [MPa] 9294 9921 8101 7350 8693 6453
ρPERP,E [-] 1.04 1.04 0.93 0.88 0.99 0.93
Mean [MPa] 12556 12070 11293 10760 11954 12367
132- CoV [-] 0.06 0.03 0.06 0.08 0.06 0.20
3layer Char. [MPa] 11049 10742 9825 9038 10520 7915
ρPERP,E [-] 1.14 1.10 1.03 0.98 1.09 1.12
Mean [MPa] 11677 11566 10812 10610 11145 11885
298- CoV [-] 0.07 0.05 0.07 0.10 0.07 0.21
7layer Char. [MPa] 10042 10294 9190 8488 9473 7369
ρPERP,E [-] 1.06 1.05 0.98 0.96 1.01 1.08
Mean [MPa] 11893 11676 10778 10335 11312 11461
All Char. [MPa] 10109 10392 8946 8268 9502 7220
ρPERP,E [-] 1.08 1.06 0.98 0.94 1.03 1.04

Table 6: Mean, coefficient of variation (CoV), and characteristic value (Char.) of the rolling shear modulus
estimated with single- and dual-parameter optimization procedures, for loads perpendicular to plane in the
moment-critical configuration.
Theory Modified Gamma Timoshenko Shear Analogy
Layup
Parameters single dual single dual single dual
Mean [MPa] 72 74 64 199 68 161
57-
CoV [%] 0.09 0.23 0.11 0.56 0.10 0.80
3layer
Char. [-] 58 44 51 58 54 26
Mean [MPa] 78 95 71 140 75 86
132-
CoV [%] 0.06 0.20 0.06 0.42 0.06 0.69
3layer
Char. [-] 69 61 62 55 66 18
Mean [MPa] 73 77 68 92 70 64
298-
CoV [%] 0.07 0.17 0.07 0.32 0.07 0.51
7layer
Char. [-] 63 52 58 45 60 20
Mean [MPa] 74 82 67 137 71 96
All
Char. [-] 63 52 56 52 60 21

3.2 Strength properties

Table 7 presents the statistical results of the failure load as measured experimentally for each layup

in moment-critical and shear-critical configurations, as well as the mean section moment and shear

force at the point of failure. The shear-critical configuration specimens fail at a higher force, as

expected, which is reflected to the mean section shear force at failure, while the mean section

23
moment at failure of the three-layer specimens in the shear-critical configuration is close to that

observed in the moment-critical configuration.

From the verification indices in the moment-critical configuration shown in Table 8 it can be observed

that the Modified Gamma method and the Timoshenko beam theory predict a rolling-shear failure

mechanism for the two 3layer layups (ρPERP,τ > ρPERP,σ) and on average the normal strength verification

for the 132-3layer layup is not satisfied. Meanwhile, the Shear Analogy method predicts a bending

failure mechanism with the highest normal strength estimates among the three theories.

Table 7: Mean, coefficient of variation (CoV), and characteristic value (Char.) of the failure load and Mean section
moment and shear forces from MoR tests for loads perpendicular to plane
Layup 57-3layer 132-3layer 298-7layer
moment- shear- moment- shear- moment- shear-
Parameters
critical critical critical critical critical critical
Mean [kN] 34.9 62.6 49.7 87.9 127.8 164.4
CoV [-] 0.12 0.08 0.11 0.03 0.05 0.03
Char. [kN] 26.5 52.6 38.3 78.2 113.7 146.3
Mean section moment [kNm] 6.0 5.4 19.4 17.1 115.0 74.0
Mean section shear [kN] 17.5 31.3 24.9 44.0 63.9 82.2

Table 8: Verification indices ρPERP,σ and ρPERP,τ for MoR tests for loads perpendicular to plane in the moment-
critical configuration
Theory Modified Gamma Timoshenko Shear Analogy
Layup ρPERP,σ [-] ρPERP,τ [-] ρPERP,σ [-] ρPERP,τ [-] ρPERP,σ [-] ρPERP,τ [-]
57-3layer 1.18 < 1.32 1.10 < 1.35 1.34 < 1.35
132-3layer 0.95 < 0.99 0.88 < 1.01 1.09 > 1.01
298-7layer 1.15 > 1.05 1.10 > 1.06 1.33 > 1.06
All 1.09 < 1.11 1.02 < 1.13 1.25 > 1.13

Regarding the results for the shear-critical configuration shown in Table 9 the shear strength

verification indices are similar among the theories but very different among the three layups with an

average value of 1.9 and a lower bound of 1.3-1.4 for the 298-7layer layup. In some specimens of the

3layer layups in the moment-critical configuration a rolling shear failure was developed instead or in

parallel to the bending failure mechanism. For these two layups, the normal stresses from the shear-

critical configuration are higher with respect to the stresses from the moment-critical configuration.

This is certainly a result that requires a more thorough investigation on a possible interaction between

shear- and moment-induced failure mechanisms. Again, the findings of Ehrhart et al. [27] correlate

24
well with the experimental results, predicting correctly an increased rolling shear strength for the first

layup, although they do not predict the difference between the other two layups.

Table 9: Verification indices ρPERP,σ and ρPERP,τ for MoR tests for loads perpendicular to plane in the shear-critical
configuration
Theory Modified Gamma Timoshenko Shear Analogy
Layup ρPERP,σ [-] ρPERP,τ [-] ρPERP,σ [-] ρPERP,τ [-] ρPERP,σ [-] ρPERP,τ [-]
57-3layer 1.36 < 2.45 1.08 < 2.66 1.61 < 2.57
132-3layer 1.15 < 1.90 0.91 < 2.07 1.36 < 2.00
298-7layer 0.85 < 1.31 0.71 < 1.38 1.04 < 1.36
All 1.10 < 1.83 0.89 < 1.96 1.32 < 1.92

4 Results and discussion for loads in plane

4.1 Elastic properties
The experimental results of tests for loads parallel to the plane for each layup are presented in Table

10 in a normalized concept, considering the exact Timoshenko beam theory as given in Appendix A.

The first column shows the effective bending stiffness calculated from the local stiffness measurement

as shown in Eq. (A.1), normalized by the bending stiffness of the net cross section, with width equal

to the sum of the thicknesses of the layers running parallel to the beam, Σt0, considered as a single

isotropic material with E = 11000 MPa. The second column shows the effective shear stiffness

calculated from both measurements as shown in Eq. (A.2), normalized by the shear stiffness of the

gross cross section considered as a single isotropic material with G = 690 MPa. Finally, the third column

shows the effective bending stiffness calculated from the global stiffness measurement neglecting

shear deformations as shown in Eq. (A.3), normalized by the bending stiffness of the net cross section.

The effective bending stiffness based on the global measurement and the infinite shear rigidity

assumption is again lower than the one based on the local measurement, with the exception of test

No. 6, and has a lower CoV. This is consistent with the results from the specimens under loads

perpendicular to plane. Similarly, the reliability of the effective shear stiffness required to equate local

and global predictions is very low.

25
Table 10: Effective elastic properties, expressed as percentage of the stiffness of the gross section: results of
MoE tests for loads in plane in the moment-critical configuration. EIeff,lo and EIeff,glob = local and global effective
bending stiffness; GAeff = effective shear stiffness; CoV = coefficient of variation; Char. = characteristic value
Layup 57-3layer 130-3layer 298-7layer
GAeff EIeff,glob GAeff EIeff,glob GAeff EIeff,glob
Test EIeff,loc EIeff,loc EIeff,loc
(EIeff,loc) (1/GAeff=0) (EIeff,loc) (1/GAeff=0) (EIeff,loc) (1/GAeff=0)
No. [%] [%] [%]
[%] [%] [%] [%] [%] [%]
1 99.8 52.4 94.3 101.5 38.5 93.9 107.6 40.3 99.0
2 137.4 22.6 115.7 120.2 80.0 114.9 102.0 37.8 93.8
3 107.4 24.0 94.4 121.4 37.0 110.2 99.9 94.0 96.6
4 118.8 20.3 100.7 120.5 35.1 109.0 109.6 25.0 95.9
5 107.0 24.8 94.4 103.4 74.8 99.1 92.5 63.0 88.2
6 95.6 -77.2* 99.4 107.8 78.7 103.5 106.4 35.2 96.9
7 98.8 69.9 94.7 121.5 27.5 107.0 101.8 29.3 91.4
Mean
108.5 31.6 98.8 113.4 48.7 105.1 102.7 42.1 94.5
[%]
CoV
0.13 0.52 0.08 0.08 0.45 0.07 0.06 0.46 0.04
[-]
Char.
81.4 9.8 84.0 94.1 17.5 90.4 90.4 15.2 84.1
[%]
*
Value excluded from Mean and CoV calculations

4.1.1 Single-parameter identification

The probability density functions of the elastic modulus from the single-parameter identification

procedures are shown in Figure 6 for the three layups and the two implemented theories. Note that

the estimates from the local deformation are the same for both Bogensperger and Flaig and Blass

theories, since the effective shear stiffness does not affect the results.

Table 11 shows the verification indices ρPAR,E for the bending modulus of elasticity. It can be observed

that the differences are small and all estimates are verified. The Bogensperger theory yields lower

estimates when accounting for the global deformation while that of Flaig and Blass yields on average

slightly higher estimates when considering the global deformation. The two theories, however, predict

quite different effective shear stiffness as shown in Table 12 with the Bogensperger theory yielding

values closer to the experimental ones, which were shown in Table 10.

26
(a) (b)

Figure 6: Bending modulus of elasticity from MoE tests for loads in plane in the moment-critical configuration,
based on the theory of (a) Bogensperger and (b) Flaig and Blass.

Table 11: Verification indices ρPERP,E for the MoE tests for loads in plane in the moment-critical configuration.
kGlob = estimate based on global modulus; kLoc = estimate based on local modulus.
Theory Bogensperger Flaig and Blass
Layup kGlob kLoc kGlob + kLoc kGlob kLoc kGlob + kLoc
57-3layer 1.03 1.08 1.05 1.05 1.08 1.07
130-3layer 1.12 1.13 1.12 1.20 1.13 1.16
298-7layer 1.01 1.03 1.02 1.10 1.03 1.06
All 1.05 1.08 1.06 1.11 1.08 1.10

Table 12: Mean effective shear stiffness predictions, expressed as percentage of the shear stiffness of the gross
section, for loads in plane in the moment-critical configuration
57-3layer 132-3layer 298-7layer
Theory
[%] [%] [%]
Experimental results
31.6 48.7 42.1
(MoE)
Bogensperger theory 79.0 51.1 47.7
Flaig and Blass theory 47.8 22.2 20.2

4.1.2 Dual-parameter identification

The results obtained for the three layups are listed in Table 13 only for the slip modulus of crossing

areas K estimated according to the Flaig and Blass theory, since for the Bogensperger theory one

should calibrate the parameters (p1, p2) and would thus need an additional deflection point other than

the local and global deflections. The modulus of elasticity is not presented because it was equal to the

one based on the local stiffness from the single-parameter identification. A mean value of 7.5

N/mm2/mm was found among the three layups that is similar to the mean value of 7.6 N/mm2/mm

27
reported in [18] from similar tests. As observed in other cases, the CoV for these estimates is quite

high above all for the 57-3layer layup.

Table 13: Mean, coefficient of variation (CoV) and characteristic value (Char.) for the estimate of the slip modulus
of crossing areas with dual-parameter optimization procedures based on the theory of Flaig and Blass
57-3layer 132-3layer 298-7layer All
Mean [N/mm3] 3.5 11.8 10.6 7.5
CoV [-] 1.03 0.30 0.28 -
Char. [N/mm3] 0.4 6.0 5.6 2.3

4.2 Strength properties

Table 14 presents the statistical results of the failure load as measured experimentally for each layup

in moment-critical and shear-critical configurations, as well as the mean section moment and shear

force at the point of failure, for loads in plane. Similar to the case of loads perpendicular to plane, the

mean section moment for the three-layer layups in the shear-critical configuration is close to that

observed in the moment-critical configuration.

Table 14: Mean, coefficient of variation (CoV) and characteristic value (Char.) of the failure load from MoR tests
for loads in plane
Layup 57-3layer 132-3layer 298-7layer
moment- shear- moment- shear- moment- shear-
Configuration
critical critical critical critical critical critical
Mean [kN] 20.5 93.8 38.5 114.7 114.4 269.0
CoV [-] 0.16 0.11 0.11 0.17 0.13 0.26
Char. [kN] 14.6 73.2 30.4 79.1 85.8 150.6
Mean section moment [kNm] 18.5 18.3 28.9 26.3 103.0 93.3
Mean section shear [kN] 10.3 46.9 19.3 57.4 57.2 134.5

The strength verification indices from the MoR tests for the moment-critical and shear-critical test

configurations are listed in Table 15 and Table 16, respectively. These tables show also the prevailing

shear failure mechanism for each theory based on Eqs. (13) and (14).

The normal strength verification indices shown in Table 15, which are the same for both theories, are

on average equal to unity. This verifies the assumptions that only the longitudinal layers (parallel to

the beam) contribute to the bending stiffness while the transverse layers effectively connect the

longitudinal ones without loss of the bending rigidity.

28
Based on the results shown in Table 16 the Flaig and Blass theory predicts a failure mechanism (4)

that is a combination of rolling shear and torsional shear in the longitudinal boards at the points were

the external forces are applied. For this mechanism, a consistent verification index of 1.2-1.3 is

computed for all layups. Instead, the Bogensperger theory yields a verification index that has a greater

variation among the three layups (1.1-1.9) predicting a shear mechanism for the 57-3layer layup and

a torsional shear mechanism for the other two layups.

Table 15: Verification indices ρPAR,σ and ρPAR,τ for MoR tests for loads perpendicular to plane in the moment-
critical configuration
Theor Verifications
Bogensperger Flaig and Blass Verifications Eq. (14)
y Eq. (13)
Layup ρPAR,σ [-] ρPAR,τ [-] ρPAR,σ [-] ρPAR,τ [-] (1) (2) (1) (2) (3) (4)
57-
0.95 > 0.38 0.95 > 0.25 0.38 0.14 0.16 0.18 0.18 0.25
3layer
132-
1.04 > 0.46 1.04 > 0.49 0.41 0.46 0.17 0.20 0.38 0.49
3layer
298-
1.02 > 0.60 1.02 > 0.69 0.55 0.60 0.18 0.24 0.54 0.69
7layer
All 1.00 > 0.47 1.00 > 0.44 0.44 0.34 0.17 0.21 0.33 0.44

Table 16: Verification indices ρPAR,σ and ρPAR,τ for MoR tests for loads perpendicular to plane in the shear-critical
configuration
Theor Verifications
Bogensperger Flaig and Blass Verifications Eq. (14)
y Eq. (13)
Layup ρPAR,σ [-] ρPAR,τ [-] ρPAR,σ [-] ρPAR,τ [-] (1) (2) (1) (2) (3) (4)
57-
0.98 < 1.94 0.98 < 1.26 1.94 0.74 0.81 0.94 0.89 1.26
3layer
132-
0.78 < 1.19 0.78 < 1.28 1.08 1.19 0.45 0.52 1.00 1.28
3layer
298-
0.96 < 1.06 0.96 < 1.21 0.96 1.06 0.31 0.41 0.95 1.21
7layer
All 0.90 < 1.34 0.90 < 1.25 1.26 0.97 0.48 0.59 0.94 1.25

5 Summary and conclusions

The results from a series of four-point bending tests on CLT panels have been presented along with

the estimated mechanical properties based on the Modified Gamma method, the Timoshenko beam

theory and the Shear Analogy method (references given in [5]) for loads perpendicular to plane and

on the methods of Bogensperger et al. [16] and Flaig and Blass [18] for loads in plane.

29
For loads perpendicular to plane, the results indicate that:

 The Shear Analogy method, among the three theories implemented, provides the most

reliable results with respect to the expected values both for elastic and strength properties,

adopting a variable connection efficiency factor γ for every cross section with the combination

of two beam elements.

 All three theories provide similar estimates of the moduli of elasticity when based on the

global stiffness measurements, whereas for evaluations based on the local stiffness

measurements the Modified Gamma method overestimates and the Timoshenko beam

theory underestimates the properties with respect to the Shear Analogy method, since a

constant γ is assumed. However, the local stiffness should always remain as the reference one

since it is not dependent on the rolling shear modulus.

 The values of E0 = 11000 MPa and fm,k = 24 MPa can be considered as reliable estimates for

structural design of CLT panels made of boards strength graded as C24. Instead the value of

fvR,k = 0.8 MPa for the rolling shear strength is a conservative estimate, as demonstrated from

the shear verification indices shown in Table 9, and the actual strength seems to be inversely

proportional to the thickness of the CLT panel with a significant variation among the three

layups. Instead, the values recommended by Ehrhart et al. [27] correlate better with the

experimental findings and should be considered as initial estimates in similar studies in the

future.

For loads in plane, the results designate that:

 The bending stiffness and strength estimates are consistent among the three layups (Table 11

and Table 15) and the values of E0 = 11000 MPa and fm,k = 24 MPa can be reliably considered

for structural design of CLT panels made of boards strength graded as C24. Note that the

bending stiffness and the normal stress profiles were the same for both considered theories,

30
 which administered only the effective shear stiffness and produced different estimates only

for the modulus of elasticity based on the global deformation.

 The Flaig and Blass theory, along with the proposed characteristic resistances and the

verification based on five different stress resultants, provides a consistent approach for the

prediction of the failure load in shear accounting for a combination of rolling and torsional

shear failure mechanism.

 The Bogensperger theory, considering two stress resultants and two verifications in shear and

torsional shear, predicts a shear strength with significant variability, which is yet similar to

that observed for loads perpendicular to plane (inversely proportional to the panel thickness).

Summarizing the results and findings deducted in this study, it can be stated that all presented

theories have been found to be applicable for the determination of the mechanical properties of CLT

panels loaded perpendicular and parallel to plane. Mean values of the elastic modulus parallel to grain

and characteristic values of the bending strength can be estimated with great confidence whereas the

shear failure mechanisms and the associated strength values are estimated with lower reliability due

to the variation of the characteristic values for each layup. Future studies should focus on the

understanding of this variability in strength of shear failure mechanism and on the refinement of the

analytical methodologies.

Acknowledgements
The authors would like to thank Paolo Burato, Paolo Pestelli and Graziano Sani for performing the

laboratory tests presented in this paper.

31
Appendix A: Local and global deformation calculus
The effective bending stiffness based on the local stiffness measurement kloc is given by:

a  s2
EIeff ,loc  kloc  (A.1)
16

where a, s and Lsp are shown in Figure 3. The effective bending stiffness based on the global stiffness

measurement kglob is given by:

3  Lsp2  4  a2
EIeff ,glob  (A.2)
 1 1 
48    
 k  a 2  GA
 glob eff 

where GAeff is the effective shear stiffness of the cross section, which in turn can be also evaluated so

that local and global bending stiffness values are equal. In this case the effective shear stiffness is:

a
GA eff  (A.3)
 1 3  L sp2  4  a2 
2  
k 3  kloc  s2 
 glob

References
[1] Brandner, R. (2013). “Production and Technology of Cross Laminated Timber (CLT): A state-
of-the-art Report”, in: Harris, R., Ringhofer, A. and Schickhofer, G. (ed.) Focus Solid Timber
Solutions – European Conference on Cross Laminated Timber (CLT), May 2013, Graz, Austria.
[2] 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.
[3] EN 16351:2015 Timber Structures - Cross Laminated Timber - Requirements. European
Standard, European Committee for standardization.
[4] Eisenberger, M. (1994). “Derivation of Shape Functions for an Exact 4-D.O.F. Timoshenko
Beam Element”, Communications in Numerical Methods in Engineering, 10, 673-681.
[5] Thiel, A. and Schickhofer, G. (2010). “CLTdesigner – A Software Tool for Designing Cross
Laminated Timber Elements: 1D-Plate-Design”, 11th World Conference on Timber
Engineering, Riva del Garda, Italy.
[6] EN 1995:2008-06 Eurocode 5: Design of timber structures - Part 1-1: General - Common
rules and rules for buildings. European Standard, European Committee for standardization.
[7] Gagnon, S. and Pirvu, C., eds. (2011). “CLT Handbook: Cross-laminated timber”, Canadian
ed. Special Publication, FPInnovations, Canada.
[8] Blass, H.J. and Fellmoser, P. (2004). “Design of Solid Wood Panels with Cross Layers”, 8th
World Conference on Timber Engineering, Lahti, Finland.

32
[9] Joebstl, R.A., Moosbrugger, T., Bogensperger, T. and Schickhofer, G. (2006). “A Contribution
to the Design and System Effect of Cross Laminated Timber (CLT)”, CIB-W18/39-12-5,
Florence, Italy.
[10] Unterwieser, H. and Schickhofer, G. (2013). “Characteristtic Values and Test Configurations
of CLT with Focus on Selected Properties”, in: Harris, R., Ringhofer, A. and Schickhofer, G.
(ed.) Focus Solid Timber Solutions – European Conference on Cross Laminated Timber (CLT),
May 2013, Graz, Austria.
[11] Schickhofer, G., Brandner, R. and Bauer, H. (2016). “Introduction to CLT, Product Properties,
Strength Classes”, Joint Conference of COST Actions FP1402 and FP1404: Cross Laminated
Timber - a competitive wood product for visionary and fire safe buildings, KTH, Stockholm,
Sweden.
[12] EN 1194:1999-04 Timber structures – Glued laminated timber – Strength classes and
determination of characteristic values. European Standard, European Committee for
standardization.
[13] Sturzenbecher, R., Hofstetter, K. and Eberhardsteiner, J. (2010). “Cross Laminated Timber:
A Multi-Layer, Shear Compliant Plate and its Mechanical Behavior”, 11th World Conference
on Timber Engineering, Riva del Garda, Italy.
[14] Gsell, D., Feltrin, G., Schubert, S., Steiger, R. and Motavalli, M. (2007). “Cross-Laminated
Timber Plates: Evaluation and Verification of Homogenized Elastic Properties”, Journal of
Structural Engineering, ASCE, 133(1), 132-138.
[15] Steiger, R., Guelzow, A., Czaderski, C., Howald, M.T. and Niemz, P. (2012). “Comparison of
Bending Stiffness of Cross-Laminated Solid Timber Derived by Modal Analysis of Full Panels
and by Bending Tests of Strip-Shaped Specimens”, European Journal of Wood and Wood
Products, 70, 141-153.
[16] Bogensperger, T., Moosbrugger, T. and Silly, G. (2010). “Verification of CLT-Plates Under
Loads In Plane”, 11th World Conference on Timber Engineering, Riva del Garda, Italy.
[17] Joebstl, R.A., Bogensperger, T. and Schickhofer, G. (2008). “In-Plane Shear Strength of Cross
Laminated Timber”, CIB-W18/41-12-3, St. Andrews, Canada.
[18] Flaig, M. and Blass, H.J. (2013). “Shear Strength and Shear Stiffness of CLT-Beams Loaded In
Plane”, CIB/W18/46-12-3, Vancouver, Canada.
[19] Vessby, J., Enquist, B., Petersson, H. and Alsmarker., T. (2009). “Experimental Study of Cross-
Laminated Timber Wall Panels”, European Journal of Wood and Wood Products, 67, 211-
218.
[20] Andreolli, M., Tomasi, R. and Polastri, A. (2012). “Experimental Investigation on In-Plane
Behaviour of Cross-Laminated Timber Elements”, CIB-W18/45-12-4, Växjö, Sweden.
[21] Brandner, R., Bogensperger, T. and Schickhofer, G. (2013). “In Plane Shear Strength of Cross
Laminated Timber (CLT): Test Configuration, Quantification and Influencing Parameters”,
CIB-W18/46-12-2, Vancouver, Canada.
[22] DIN 4074-1 (2003). “Strength grading of wood - Part 1: Coniferous sawn timber” German
Institute for Standardization.
[23] Mathworks (2011). “MATLAB – R2011b Release,” Mathworks, Natick, MA.
[24] EN 14358:2006-12 Timber structures – Calculation of characteristic 5-percentile values and
acceptance criteria for a sample. European Standard, European Committee for
standardization.
[25] EN 338:2009-10 Structural timber – Strength classes. European Standard, European
Committee for standardization.
[26] Fellmoser, P. and Blass, H.J. (2004). “Influence of Rolling Shear Modulus on Strength and
Stiffness of Structural Bonded Timber Elements”, CIB/W18/37-6-5, Edinburgh, United
Kingdom.
[27] Ehrhart, T. Brandner, R., Schickhofer, G. and Frangi, A. (2015). “Rolling Shear Properties of
some European Timber Species with Focus on Cross Laminated Timber (CLT): Test

33
 Configuration and Parameter Study”, International Network on Timber Engineering
Research (INTER 2015), Šibenik, Croatia.

34

View publication stats