INTER / 49 - 5 - 1

Strength grading of European beech

lamellas for the production of GLT & CLT

Thomas Ehrhart, ETH Zürich, Institute of Structural Engineering, Zurich, Switzerland

Gerhard Fink, Aalto University, Department of Civil Engineering, Espoo, Finland
René Steiger, Empa, Swiss Federal Laboratories for Materials Science and Technology,
Dübendorf, Switzerland
Andrea Frangi, ETH Zürich, Institute of Structural Engineering, Zurich, Switzerland

Keywords: European beech, strength grading, visual grading characteristics, longitu‐

dinal eigenfrequency, strength classes, glued laminated timber, cross laminated tim‐
ber

1 Introduction
Like in several other Central European countries, the share of European beech wood
(Fagus sylvatica L.) in the Swiss forests is very high and steadily growing due to forest
policy and the progressing climate change. Nowadays, European beech is primarily
used in furniture industry and for heating purposes, but not in the construction in‐
dustry. The best way to process beech logs for structural purposes is still being dis‐
cussed around the question: To peel or not to peel? For veneered products such as
LVL, strength grading is of minor importance due to the high degree of homogeniza‐
tion in the final product; however, grading of the raw material is a key step in the
production of glued laminated timber (GLT) and cross laminated timber (CLT).
Previous studies by Frühwald & Schickhofer (2004) and Blass et al. (2005) have out‐
lined the great potential of European beech timber in terms of strength and stiffness.
High mean values of tensile strength ft,0,mean between 62 and 75 N/mm2
(ft,max ≈ 120 N/mm2) and mean values of the modulus of elasticity (MOE) Et,0,mean of
approximately 13300‐14000 N/mm2 (Et,0,max ≈ 20000 N/mm2) were found. However,
the studies also report a large scatter of results, i.e. coefficient of variation COV of up
to 0.50 for ft,0 and 0.25 for Et,0.

1
INTER / 49 - 5 - 1

The parameters to be accounted for and the limits for visual strength grading of
hardwood – including European beech – are specified in DIN 4074‐5 (2008). The
highest visual strength grade LS 13 corresponds, according to DIN EN 1912 (2012), to
the strength grade D 40 (valid for beech wood from Germany). The mechanical and
physical properties of D 40 are listed in EN 338 (2016) (selected values in Table 1).
Table 1 Selected mechanical and phyiscal properties according to EN 338 (2016)
fm,k ft,0,k Em,0,mean Em,0,05 ρmean ρk
Strength grade
[N/mm2] [N/mm2] [N/mm2] [N/mm2] [kg/m3] [kg/m3]
D 40 40 24 13000 10900 660 550

The requirements for the production of GLT are defined in EN 14080 (2013), however
this standard is restricted to softwood species and poplar. According to EN 14080,
the lamellas used for the production of GLT need to be of a certain T‐class, i.e. re‐
quirements with regard to the 5% fractile value of tensile strength ft,0,l,k, the mean
value of tensile MOE Et,0,l,mean and the 5% fractile value of density ρl,k have to be met.
Regarding CLT, production requirements are defined in EN 16351 (2015), again lim‐
ited to softwood species and poplar. So far, there are no European standards availa‐
ble for the production of GLT or CLT made from hardwood (except for poplar).
In Germany, the technical approval Z‐9.1‐768 (2009) regulates the production of
beech GLT for strength classes up to GL 48c. This technical approval represents a ma‐
jor step towards a practical application of beech GLT. However, besides the bending
strength, the other prescribed strength‐ and stiffness properties seem to have been
limited to a rather conservative level and hence, do not encourage designers to apply
beech GLT in practice.
The present study is part of a research project with the objective of providing the
technical basis for the market implementation of high strength GLT made from Euro‐
pean beech timber. Strength grading – especially regarding high strength grades – is
considered to be of utmost importance for the whole project and therefore is fo‐
cused on in the first stage.
In order to facilitate the production of GLT of the strength classes GL 40 (= fm,g,05 of
GLT), GL 48 and GL 55, reliable limits and rules for strength grading of European
beech lamellas into the tensile strength grades T 33 (= ft,0,k of the T‐class), T 42 and
T 50 have to be specified.

2
 INTER / 49 - 5 - 1

2 Material and Methods

2.1 Material
The experimental investigations summarised in this paper were performed on 294
lamellas produced from Swiss grown European beech wood. Planed lamellas with the
dimensions w·t·l = 160 x 25 x 3000 mm3 were ordered in equal shares from four
sawmills located north of the Alps in the Cantons Bern, Jura, Aargau and Zurich
(Figure 1).
This order strategy was chosen attempting to cover the beech population in Swiss
forests adequately and allowing an assessment of the availability of the demanded
quality.

Figure 1 Location of the sawmills in the Swiss Cantons Bern, Jura, Aargau and Zurich
The primary use of GLT and CLT products made from beech is considered to be in
service class 1 according to Eurocode 5 (2004), due to the unfavourable behaviour of
beech when subjected to changes in humidity. Thus, a moisture content (MC) of
8 ± 2% was defined in this study in order to, at least partially, avoid shrinkage effects.
Frühwald et al. (2003) and Ohnesorg & Becker (2006) reported preferable delamina‐
tion behaviour for thinner lamellas. Therefore, compared to GLT from softwood
where lamella thicknesses up to 45 mm are common, a rather low thickness
t = 25 mm of the lamellas was chosen as default value for the whole project.

3
INTER / 49 - 5 - 1

2.2 Methods
2.2.1 Grading criteria
Based on findings by Frühwald & Schickhofer (2004) and Blass et al. (2005), the fol‐
lowing visual indicators were identified in Ehrhart et al. (2016) to have a major influ‐
ence on the tensile strength and stiffness of beech lamellas, and were therefore used
in the present study in order to optimize the grading rules.
 Knots and bark inclusions quantified by the total knot area ratio (tKAR).
 Obvious fibre deviation in lamellas with rift or half rift grain.
 Wavelike annual ring pattern.
 Discolouration decreasing the surface hardness.
Lamellas containing wane, rot, fissures through the thickness, pith and insect damage
were rejected. The presence of redheart was documented but not used in the
strength grading process.
According to Görlacher (1990), the dynamic MOE Edyn was used as physical indicator
when optimising the grading process. Based on the first eigenfrequency f1 for excita‐
tion in the longitudinal direction, the length l and the bulk density ρu at a MC of u %,
Edyn was calculated:

4∙ ∙ ∙

According to findings by Frühwald & Schickhofer (2004), Blass et al. (2005) and
Ehrhart et al. (2016), density does not influence ft,0 and Et,0 markedly. Thus, density
was not considered as indicator in the strength grading process.
For investigations concerning strength grading (see Chapter 3), only characteristics
located within the tested length (free length between the clamping jaws of the test‐
ing machine) were taken into account. However, the entire lamella (l = 3000 mm) and
all present characteristics were considered in questions of economical yield. For both
issues, the chosen approach can be considered as conservative.
2.2.2 Definition of pre‐grades
In order to simplify the testing process, all lamellas were initially assigned to a pre‐
grade (A = highest estimated strengths, B, C or Rejects) based on several visual indi‐
cators and Edyn. In Ehrhart et al. (2016), the requirements and limits for each pre‐
grade are described in detail. Destructive tensile tests were performed on the ran‐
domly selected batches of lamellas listed in Table 2.
Table 2 Number of tested lamellas of the pre‐grades A, B, C and Reject
Pre‐grade A B C Rejects
Number of tested lamellas 50 150 50 47

4
 INTER / 49 - 5 - 1

2.2.3 Destructive tensile tests

Destructive tensile tests according to EN 408 (2010) were performed in order to de‐
termine ft,0 and Et,0 of the lamellas (Table 2). 50 lamellas of each pre‐grade (A, B and
C) and all Rejects were tested with a test length lt = 2060 mm (≈ 13·w, inner distance
between the clamping jaws). Additionally, 50 lamellas of pre‐grade B were tested
with lt = 1580 mm (≈ 10·w) and 1100 mm (≈ 7·w) in order to investigate the effect of
the test length on the results. A detailed description of the test setup, methods and
calculations can be found in Ehrhart et al. (2016).
The MC of the specimens was determined using the oven dry method according to
EN 13183‐1 (2002). Et,0,u and ρu were adjusted to the project’s reference MC of 8%
and the standard reference MC of 12% according to EN 384 (2013). No adjustment is
required for ft,0.
As specified in EN 14358 (2013), ft,0 was assumed lognormally distributed. The maxi‐
mum likelihood method was used to estimate the distribution parameters that best
fit the test data. Et,0 and ρ were assumed normally distributed according to EN 14358.
The distribution types of ft,0 and ρ also correspond to the recommendations provided
by the JCSS ‐ Probabilistic Model Code (2006).

3 Results and Discussion

In the following sections, an overview of the mechanical properties of the investigat‐
ed material is given and the maximum potential and limits of single indicator grading
is discussed. The optimized grading limits as well as the related strength, stiffness and
density values are presented. Furthermore, the influence of the test length lt on the
measured mechanical properties and the influence of the minimum accepted length
lmin on the material yield are investigated. Finally, the potential mechanical properties
of GLT made from European beech are estimated.
3.1 Overall results
In Figure 2, the empirical distribution of ft,0 and Et,0,8 (MC = 8%) is shown. The tensile
strength was investigated on all lamellas (n = 294). Et,0,8 was determined for lamellas
of pre‐grades A, B and C when tested with a length of lt = 2060 mm (n = 147). Due to
large distortions, three lamellas could not be tested.
A mean tensile strength ft,0,mean = 61 N/mm2 confirms the great mechanical potential
of European beech wood. However, the large scatter in strength (COV = 0.48;
ft,0,min = 4.3 N/mm2, ft,0,max = 132 N/mm2) clearly evidences the need for reliable
strength grading procedures and rules. Comparison to previous studies on ft,0 of Eu‐
ropean beech wood by Frühwald & Schickhofer (2004) and Blass et al. (2005) indi‐
cates, that the investigated population covers the full strength‐spectrum.

5
INTER / 49 - 5 - 1

40 30
n = 294 n = 147
COV = 0.48 COV = 0.13
25

Absolute frequency [‐]

Absolute frequency [‐]
30
20

20 15

10
10
5

0 0
0 20 40 60 80 100 120 140 10 12 14 16 18 20 22
ft,0 [N/mm2] Et,0,8 [kN/mm2]

Figure 2 Distribution of the tensile strength ft,0 (left) and MOE Et,0,8 (right) of the tested European
beech lamellas (MC = 8%)
Regarding Et,0,8, minimum, mean and maximum values of 10100, 15200 and
20500 N/mm2 were found (COV = 0.13). These values are just above the values re‐
ported by Frühwald & Schickhofer (2004) and Blass et al. (2005) when adjusting Et,0 to
the standard reference MC of 12%.
3.2 Single indicator grading (tKAR or Edyn): maximum potential and limits
For the identification of high‐strength and stiff lamellas, combined visual and me‐
chanical grading procedures are required as already reported in Frese & Blass (2006).
The influence of varying limits for Edyn (lower limit) and tKAR (upper limit) on the 5%
fractile value of the tensile strength ft,0,05 can be seen in Table 3 (only lamellas tested
with lt = 2060 mm (n = 194) are considered). The bottom row (Edyn,8 = all) represents a
visual grading method exclusively based on the tKAR‐value. The far right column
(tKAR = all) represents mechanical grading using Edyn only.
Both approaches lead to unsatisfactory results: Even for the strictest possible visual
limit (tKAR = 0), ft,0,05 of 50 N/mm2 cannot be achieved. Regarding the exclusive me‐
chanical grading approach, the requirements for the strength grade T 50 are reached
for Edyn,8 ≥ 19000 N/mm2. However, only 16 out of 194 lamellas (8%) exceed this limit.
In contrast, a grading approach combining both visual and mechanical grading meth‐
ods leads to significantly better yield especially for high strength levels. A sample of
94 lamellas (= 70 + 64 ‐ 48 + 54 ‐ 48 + 16 ‐ 14, Table 3) with a characteristic tensile
strength ft,0,05 > 50 N/mm2 can be found and assigned to the strength grade T 50.

6
 INTER / 49 - 5 - 1

Table 3 Number of specimens and level of characteristic tensile strength ft,0,05 depending on the
maximum tKAR value and the dynamic MOE Edyn,8 (MC = 8%)

Edyn,8 tKAR [‐]

2
[N/mm ] 0 ≤ 0.05 ≤ 0.1 ≤ 0.2 ≤ 0.3 ≤ 0.4 ≤ 0.5 all
≥ 20000 4 5 7 7 7 7 7 7
≥ 19000 6 11 14 15 16 16 16 16
≥ 18000 18 28 32 35 36 37 37 37
≥ 17000 35 48 54 60 61 63 63 63
≥ 16000 48 64 73 86 93 95 95 95
≥ 15000 56 74 85 105 116 122 124 124
≥ 14000 70 89 103 126 141 150 152 154
≥ 13000 75 94 110 136 153 165 168 173
≥ 12000 77 97 113 139 158 173 177 185
≥ 11000 77 97 113 140 160 176 181 190
≥ 10000 77 97 113 140 162 178 183 193
all 77 97 113 140 162 179 184 194

≥ 50 42 ‐ 50 33 ‐ 42 26 ‐ 33 < 26
ft,0,05 [N/mm2]

3.3 Optimized grading limits

The visual and mechanical grading rules presented hereafter and summarized in Ta‐
ble 4 are based on the assumption that the target T‐classes are T 50, T 42 and T 33.
Besides the tKAR‐value, limits for several additional visual strength indicators – briefly
introduced in 2.2.1, specified and investigated in Ehrhart et al. (2016) – were defined
to complement and further improve the strength grading process. Additional limits to
assure the processability of beech lamellas (e.g. regarding distortion and fissures) are
defined in DIN 4074‐5 (2008) and purposely not described at this point.
Following the proposed grading strategy, the population initially was split into three
different visual grades (1 = best, 2, and 3) and Rejects. Due to the high correlation be‐
tween tKAR and ft,0, as well as the influence of fibre deviation and further visual char‐
acteristics on ft,0, this led to a first separation into groups of different strengths.
Subsequently, the lamellas were assigned to the T‐classes depending on their visual
grade (1, 2 or 3) and Edyn,8. Besides further improvement of the strength grading pro‐
cess, this allowed for a precise grading regarding Et,0.
Figure 3 shows the cumulated distribution of ft,0 (left) and Et,0,8 (right) of the classes
T 33, T 42, T 50 and Rejects as resulting after the application of the grading rules de‐
scribed above.

7
INTER / 49 - 5 - 1

Table 4 Limits for the visual and mechanical grading into strength grades T 33, T 42 and T 50
Visual grade
Criterion Visual 1 Visual 2 Visual 3
Knots or bark inclusions tKAR ≤ 0.05 tKAR ≤ 0.1 tKAR ≤ 0.2
Obvious fibre deviation ‐ Unlimited Unlimited
Wavelike annual ring pattern ‐ Unlimited Unlimited
Redheart Unlimited Unlimited Unlimited
Discolouration
Surface hardness not reduced Unlimited Unlimited Unlimited
Surface hardness reduced ‐ ‐ Unlimited
Insect damage ‐ ‐ ‐
Pith ‐ ‐ ‐
Required Edyn,8 T 33 T 42 T 50
Visual 1 ≥ 12000 ≥ 14000 ≥ 16500
Visual 2 ≥ 14000 ≥ 16500 ‐
Visual 3 ≥ 16500 ≥ 18000 ‐

The large gap of ft,0 between the rejected sample and the T‐classes shows that the
applied grading criteria are efficient to identify weak lamellas (Figure 3, left). Fur‐
thermore, ft,0 of less than 20% of the Rejects exceeded 40 N/mm2. The clear separa‐
tion between the T‐classes indicates that the chosen grading approach leads to satis‐
factory results.
Regarding Et,0,8, also a clear separation between the T‐classes is found (Figure 3,
right). The distribution of Et,0,8 can be controlled quite precisely due to the high corre‐
lation with Edyn (a coefficient of determination R2 = 0.84 was reported by
Ehrhart et al. (2016)).
1.0

1.0
0.8

0.8
0.6

0.6
cdf [‐]

cdf [‐]
0.4

0.4

T‐class T‐class
T 50 T 50
0.2

0.2

T 42 T 42
T 33 T 33
Rejects Rejects
0.0

0.0

0 20 40 60 80 100 120 140 10 12 14 16 18 20

ft,0 [N/mm2] Et,0,8 [kN/mm2]

Figure 3 Cumulated distribution (cdf) of tensile strength ft,0 (left) and MOE Et,0,8 (right) for the
resulting T classes T 33, T 42 and T 50 (MC = 8%)

8
 INTER / 49 - 5 - 1

In Table 5, the mechanical properties and densities of the resulting T‐classes and Re‐
jects are summarized. Values of ft,0,05 slightly higher than targeted resulted when ap‐
plying the specified grading rules. As no explicit provision concerning the density is
made in the grading process, ρmean and ρk of all T‐classes is on a similar level.
Table 5 Mechanical properties and density of the resulting T‐classes and Rejects for specimens with
a test length of 2060 mm (MC = 8%)
T 33 T 42 T 50 Rejects All
2
Et,0,8,mean N/mm 14200 15400 16900 13200 15200
Et,0,8,05* N/mm2 12300 12900 15500 10400 12100
COV ‐ 0.09 0.11 0.07 0.13 0.13
n ‐ 26 49 41 31 147
ft,0,mean N/mm2 59.7 70.6 89.3 28.7 56.2
ft,0,05** N/mm2 34.7 45.9 54.8 11.0 16.3
COV ‐ 0.30 0.24 0.25 0.44 0.53
n ‐ 26 49 41 78 194
ρ8,mean kg/m3 714 722 739 720 724
ρ8,05* kg/m3 651 683 673 668 668
COV ‐ 0.06 0.04 0.05 0.05 0.05
n ‐ 26 49 41 78 194
*
Normal distribution, ** Lognormal distribution

3.4 Influence of the test length lt

In Figure 4 (left), the cumulated distribution of ft,0 for the three different test lengths
lt = 2060, 1580 and 1100 mm are plotted. Each sample consists of 50 specimens of
the pre‐grade B (n = 48 for lt = 2060 mm due to strong distortion of two lamellas).
1.0

40
0.8

Absolute frequency [‐]

30 lt = 2060 mm
lt = 1580 mm
0.6

lt = 1100 mm
cdf [‐]

20
0.4

lt [mm]
10
0.2

2060
1580
1100
0.0

0
20 40 60 80 100 120 0.00 0.05 0.10 0.15 0.20 0.25 0.30
ft,0 [N/mm2] tkar [‐]

Figure 4 Distribution of tensile strength ft,0 (left) and tKAR‐values (right) for samples subjected to
tensile tests with different test lengths lt

9
INTER / 49 - 5 - 1

The mean values of tensile strength of the samples with test lengths of lt = 2060 mm
(ft,0mean = 66.7 N/mm2) and lt = 1580 mm (ft,0mean = 67.4 N/mm2) were found to be al‐
most equal. However, ft,0mean of lt = 1100 mm is about 13% higher
(ft,0mean = 75.8 N/mm2). This can be due to the different distribution of tKAR‐values of
the groups, i.e. considerably more lamellas of the lt = 1100 mm sample showed tKAR‐
values < 0.05 (78%) compared to lt = 1580 mm (66%) and lt = 2060 mm (71%) (Figure
4, right).
Regarding the general adequacy of a test length lt ≥ 9·w, as prescribed in EN 408, two
contradictory conclusions (i & ii) can be drawn. (i) As the number of obvious weak
points like knots or bark inclusions is much lower in European beech timber com‐
pared to most softwoods (see also Chapter 3.5), the location of failure can usually be
predicted with high reliability if such a weak point is present. Local failure restricted
to a short section of a lamella is observed in many of these cases (Figure 5). Thus, the
actual lt has negligible influence on the tensile strength of lamellas of low visual quali‐
ty as long as weak points are positioned within the tested length of a lamella.
Increasing lengths l of the lamellas, of course, have also impact on the likelihood of
the presence of weak points.

Figure 5 Local failures at obvious weak points

However, (ii) shear failures affecting large parts of lamellas were frequently observed
in lamellas with low tKAR‐values. Depending on the general and local slope of grain,
shear fractures propagate until they reach either the lamella’s edge or the clamping
jaws. In the second case, shear fracture is supressed in the areas close to the clamp‐
ing jaws as the deformations in the direction perpendicular to the lamella’s axis are
restrained (Figure 6). Hence, the shorter lt, the higher the probability that shear fail‐
ure reaches and stops at the clamping jaws, which potentially influences ft,0.

Figure 6 Shear failure over large parts of a lamella reaching and stopping at the clamping jaws

10
 INTER / 49 - 5 - 1

As a consequence, specification of a minimum test length for the determination of ft,0

is of utmost importance for comparability reasons, in particularly when investigating
lamellas of high strength grades.
3.5 Influence of the minimum accepted length lmin on the material yield
In accordance with findings by Glos et al. (2004) and Frese & Blass (2006), the num‐
ber of knots or knot clusters in the investigated beech lamellas (Figure 7, left) was
found to be considerably lower compared to spruce lamellas. Thus, finger jointing
plays a decisive role when aiming at improving the technical as well as the economi‐
cal yield as it allows cutting out single weak sections containing large knots or bark
inclusions.
This study so far focused on the weakest testable section of the entire lamella in or‐
der to grade or reject the lamellas, i.e. single knots or bark inclusions with tKAR‐
values exceeding the defined limits led to downgrading or rejection of the whole la‐
mella. In several cases, large knots or bark inclusions are located close to the end of
lamellas. Thus, cutting them out leads to a significant improvement regarding yield.
Depending on the minimum length accepted lmin (as technically given by the finger
jointing process or economically asked for when considering costs and benefits), a
significant proportion of the initially rejected material can be used when lmin is speci‐
fied shorter than 3 m.
In total 87 lamellas (87 x 3 m = 261 m) were rejected due to insufficient visual ap‐
pearance. Thereof, 68 lamellas showed too large knots or bark inclusions (204 m) and
19 lamellas (57 m) could not be used due to issues like curvature, fissures through
the lamellas thickness and wane. The latter were excluded as definitive Rejects,
which means that permission of shorter lengths does not affect their classification.
40 n = 294 300 Rejects Accepted

250 def. other Vis 1 Vis 2 Vis 3

30
Relative frequency [%]

200
Yield [m1]

20 150

100
10
50

0
0
0 1 2 3 4 5 6 3 2.5 2 1.5
Number of knots or knot clusters Minimum accepted length lmin [m]

Figure 7 Number of knots or knot clusters per lamella (left) and influence of the minimum accepted
length lmin on the yield (right)

11
INTER / 49 - 5 - 1

Figure 7 (right) shows the influence of varying lmin on the yield (in running metre m1).
The group of Rejects is split into definitive and other (see above). Regarding the ac‐
cepted material, a distinction is made between the different visual grades (Visual 1, 2
and 3) according to Table 4. Starting from lmin = 3 m, the significant reduction of ma‐
terial assigned to the group of Rejects for decreasing lmin is obvious. For lmin = 1.5 m,
more than 50% of the initially rejected material can be used in almost equal parts for
the visual grades 1, 2 and 3. Due to the mechanical potential of European beech tim‐
ber, specification of even shorter lengths could be reasonable in order to achieve
higher yield in the strength grades T 42 and T 50.
3.6 Potential mechanical properties of GLT
Numerous models for the derivation of bending strength of GLT (fm,g,k) from the la‐
mellas’ characteristic tensile strength (ft,l,k) and the characteristic strength of finger
joints in bending or tension (fm,j,k or ft,j,k) are available (Brandner & Schickhofer, 2008).
However, the use of most of these models is restricted to softwoods and only limited
findings regarding the laminating effect of beech GLT on a high strength level are re‐
ported in literature. One exception is Frese & Blass (2006), presenting a quadratic
equation for the estimation of fm,g,k of beech GLT.
According to EN 14080 (2012), fm,g,k, the mean value of MOE of GLT (E0,g,mean) and its
density (ρg,mean) can be calculated based on the lamellas’ properties. Although the ap‐
plication of EN 14080 is restricted to softwoods and poplar, it is used at this point to
estimate the properties of beech GLT.
Based on the material properties presented in this study and the above‐mentioned
models it seems realistic to reach the target GLT strength grades. For GL 40, GL 48
and GL 55, expected properties are E0,g,mean ≈ 14300 / 15500 / 17000 N/mm2 and
ρg,mean ≈ 730 / 740 / 750 kg/m3, respectively (MC = 12%). Due to the small variations
between the strength grades, a differentiation regarding the density does not seem
to be appropriate.

4 Conclusions
Characteristic tensile strength values ft,0,k of 33, 42 and 50 N/mm2 can be exceeded
when applying the grading rules presented in this paper. Consequently, the investi‐
gated European beech lamellas can be assigned to the strength grades T 33, T 42 and
T 50. Regarding the mean value of MOE, clear separation between the strength
grades was found, i.e. Et,0,8,mean = 14200, 15400 and 16900 N/mm2, respectively, at
the project’s reference MC of 8%. The density of all grades is on a very similar level
(ρmean,8 ≈ 720 kg/m3) as it is not considered as a strength or stiffness indicator in the
grading process.
The proposed grading strategy bases on two main parameters – the total knot area
ratio (tKAR) and the dynamic MOE (Edyn). Obvious fibre deviation, wavelike annual
ring pattern and discolouration are used as complementing visual indicators to fur‐

12
 INTER / 49 - 5 - 1

ther improve the strength grading process. It is shown that a combined grading ap‐
proach is essential for efficient strength grading, i.e. neither a pure visual nor a pure
mechanical grading approach allow reaching the target strength levels.
If single obvious weak points are present, provision of a minimum test length of 9·w
is not necessary, as failures occur in very restricted areas. In beech lamellas of higher
strength grades, shear failure occurs frequently. The propagation of shear failure is
stopped, or at least influenced, at the clamping jaws. Thus, maintaining a minimum
test length is clearly necessary.
Strict requirements regarding the maximum size of knots and bark inclusions are es‐
sential to reach the target strength classes. In questions of material efficiency, the
minimum accepted length lmin of lamellas has significant impact on the achievable
yield, as the number of knots in European beech timber is substantially lower com‐
pared to most softwoods. The reduction of lmin from 3 m to 1.5 m approximately
halves the volume of rejected material and hence, markedly increases the yield.
Based on the results presented in this study, the target strength classes GL 40, GL 48
and GL 55 appear realistic and reachable. Due to the lack of appropriate models for
the observed strength level and/or timber species, the estimated strength and stiff‐
ness properties for GLT are subjected to uncertainty.

5 Acknowledgement
The authors acknowledge the support of the Swiss Federal Office for the Environment
FOEN within the framework of Aktionsplan Holz.

6 References
Blass, HJ; Denzler, J; Frese, M; Glos, P; Linsenmann, P (2005): Biegefestigkeit von
Brettschichtholz aus Buche; Band 1 der Reihe Karlsruher Berichte zum Ingenieurholz‐
bau; Universitätsverlag Karlsruhe; Karlsruhe, Germany.
Brandner, R; Schickhofer, G (2008): Glued laminated timber in bending: new aspects
concerning modelling; Wood Sci Technol (2008) 42:401‐425; Springer.
Ehrhart, T; Fink, G; Steiger, R; Frangi, A (2016): Experimental investigation of tensile
strength and stiffness indicators regarding European beech timber; Proceedings of
14th World Conference on Timber Engineering WCTE, 2016; Vienna, Austria.
Frese, M; Blass, HJ (2006): Characteristic bending strength of beech glulam; Materials
and Structures (2006) 40:3‐13; RILEM.
Frühwald, A; Ressel, JB; Bernasconi, A (2003): Hochwertiges Brettschichtholz aus Bu‐
chenholz; Abschlussbericht; Universität Hamburg & Bundesforschungsanstalt für
Forst‐ und Holzwirtschaft; Hamburg, Germany.
Frühwald, K; Schickhofer, G (2004): Strength grading of hardwoods; 8th World Con‐
ference on Timber Engineering; Lahti, Finland.

13
INTER / 49 - 5 - 1

Glos, P; Denzler, JK; Linsenmann, PW (2004): Strength and stiffness behaviour of

beech laminations for high strength glulam; CIB‐W18/37‐6‐3, Meeting 37; Edinburgh,
Scotland.
Görlacher, R (1990): Klassifizierung von Brettschichtholzlamellen durch Messung von
Longitudinalschwingungen; Berichte der Versuchsanstalt für Stahl, Holz und Steine
der Universität Fridericiana in Karlsruhe; 4. Folge ‐ Heft 21; Karlsruhe, Germany.
Ohnesorg, D; Becker, G (2006): Glueability of beech wood containing red heartwood;
5th IUFRO International Symposium, Wood Structures and Properties; Sielnica, Slo‐
vakia.
DIN 4074‐5 (2008): Strength grading of wood ‐ Part 5: Sawn hard wood; DIN.
EN 338 (2016): Structural timber ‐ Strength classes; CEN.
EN 384 (2013): Structural timber ‐ Determination of characteristic values of mechani‐
cal properties and density; CEN.
EN 408 (2010): Timber structures ‐ Structural timber and glued laminated timber ‐
Determination of some physical and mechanical properties; CEN.
EN 1912 (2012): Structural timber ‐ Strength classes ‐ Assignment of visual grades
and species; CEN.
EN 1995‐1‐1 (2004): Eurocode 5: Design of timber structures ‐ Part 1‐1: General ‐
Common rules and rules for buildings; CEN.
EN 13183‐1 (2002): Moisture content of a piece of sawn timber ‐ Part 1: Determina‐
tion by oven dry method; CEN.
EN 14080 (2013): Timber structures ‐ Glued laminated timber and glued solid timber
‐ Requirements; CEN.
EN 14358 (2013): Timber structures ‐ Calculation of characteristic 5‐percentile and
mean values for the purpose of initial type testing and factory production control;
CEN.
EN 16351 (2015): Timber structures ‐ Cross laminated timber ‐ Requirements; CEN.
JCSS (2006): Probabilistic model code Part 3 – Part 3.5: Properties of timber.
http://www.jcss.byg.dtu.dk/Publications/Probabilistic_Model_Code.aspx
Z‐9.1‐679 (2009): Allgemeine Bauaufsichtliche Zulassung; BS‐Holz aus Buche und BS‐
Holz Buche Hybridträger; Studiengemeinschaft Holzleimbau e. V.; Wuppertal, Ger‐
many.

14