A method for glue bond quality testing

of flange/web adhesive connections

of wooden I-beams

Nordtest project no. 1583-02

Erik Serrano, Mikael Fonselius, Carl-Johan
Johansson and Kjell H. Solli

A method for glue bond quality

testing of flange/web adhesive
connections of wooden I-beams
SP Rapport 2004:02
Bygg och Mekanik
Borås 2004
 2

Abstract
A test method for assessing the quality of glue bonds in the web-flange joint of I-beams
has been developed. The method is based on a shear test of small pieces cut from the
beam. The investigation has included shear tests on different specimen sizes, including
the influence of the wood density, the wood moisture content and the influence of
different pre-treatments (boiling) in order to detect possible gluing errors. A limited finite
element analysis was also performed. The main conclusion is that pre-treatment of the
specimen by boiling is necessary in order to detect gluing errors. Another confirmation is
that the wood failure percentage can be used as a good indicator of the bond quality.
Finally, a draft test method for a possible standard is given.

Key words: quality control, test method, I-beam, shear test, engineered wood product

SP Sveriges Provnings- och SP Swedish National Testing and

Forskningsinstitut Research Institute
SP Rapport 2004:02 SP Report 2004:02
ISBN 91-7848-
ISSN 0284-5172
Borås 2003
Postal address:
Box 857,
SE-501 15 BORÅS, Sweden
Telephone: +46 33 16 50 00
Telex: 36252 Testing S
Telefax: +46 33 13 55 02
E-mail: [email protected]
 3

Contents
Abstract 2

Contents 3

Summary 5

1 Introduction 7
1.1 Background 7
1.2 Aim 7
1.3 Current method used by a manufacturer 7
1.4 Work content and disposition of the report 8

2 Materials and methods 9

2.1 Beam materials 9
2.2 Test set-ups 9

3 Task 2 – Influence of specimen thickness 11

3.1 General remarks 11
3.2 Test programme 11
3.3 Test results 11
3.4 Finite element study 12
3.4.1 FE-model 12
3.4.2 FE-Results 14
3.5 Conclusions – Task 2 15

4 Task 3 – Influence of moisture content 16

4.1 General remarks 16
4.2 Test programme 16
4.3 Test results 16

5 Task 4 – Influence of pre-treatment 18

5.1 General remarks 18
5.2 Test programme 18
5.3 Test results 19

6 Discussion and conclusions 23

6.1 Gluing errors 23
6.2 Detecting errors 23
6.2.1 General 23
6.2.2 Without pre-treatment 23
6.2.3 With pre-treatment (boiling) 23
6.2.4 Decision based on shear strength or wood failure percentage 24

References 25

Appendix A: Wooden I-beams – Test method for evaluation of

flange/web glue bond quality A1
 4

Preface
This report was prepared as the final project report for the Nordtest project 1583-02
“Träbaserade lättbalkar. Utveckling av metod för bestämning av limfogskvalitet” (Wood
I-beams. Development of a method for the determination of adhesive bond line quality”).
The project has been performed as a co-operation by SP – Swedish National Testing and
Research Institute, VTT –Technical Research Centre of Finland and NTI – Norwegian
Institute for Wood Technology. SP has had the responsibility of the final compilation of
the results and editing of this report.

Borås, October 2003

Erik Serrano
 5

Summary
An important aspect of the production of wooden I-beams is the control of the quality of
the adhesive joint that bonds the web to the flange. A test method needs to be developed,
since the existing ones as described, for example in EN 392 (Anon.1995), cannot be used
due to the geometry of the joint. The development of such a test method is the aim of the
current project.

The work reported here has included different investigations in order to establish a
reliable test method. The influence of the test piece thickness was studied in order to
obtain a test specimen size of appropriate dimensions. The influence of the moisture
content and of different pre-conditioning climates was also investigated. Finally, in order
to be able to detect less severe gluing errors, the influence of boiling the specimens prior
to testing was investigated. In total 800 individual shear tests have been performed, each
being evaluated at least in terms of shear strength and wood failure percentage.

Based on the findings, it is recommended that a test method based on a compressive shear
test of the bond between web and flange (push-through) be used. It is recommended that
the test pieces should be 20 mm in thickness, which was found to be a good compromise
between the strive for a large enough specimen in order not to cause to much damage
when cutting it and the strive for small specimens resulting in lesser load levels and more
uniform stress distribution.

It is furthermore recommended that the testing be preceded by repeated boiling of the

specimens in water. This pre-treatment has shown to be enough to predict the most severe
gluing errors. An alternative pre-treatment to use is boiling the specimens only once, but
to use a stricter pass/fail criterion.

When evaluating the test results, it is recommend that the wood failure percentage be
used as a pass/fail criterion in combination with a minimum shear strength criterion. The
level for the wood failure percentage is set to 70% for repeated boiling as a pre-treatment.
This means that, in order to accept the quality of the bond line, the wood failure
percentage should be 70% or more. For the alternative pre-treatment of boiling only once,
the level is set to 80% wood failure. The reference shear strength value to be met, should
be determined at the products initial type testing.

A test method, which can be used as starting point for future standardisation, can be
found in Appendix A.
6
 7

1 Introduction
1.1 Background
Different types of engineered wood products (EWP) have gained importance in recent
years. A large part of these EWP are I-beams, where the web is adhesively bonded into
grooves machined into the flanges. Other EWP include such products like oriented strand
board (OSB) and Laminated Veneer Lumber (LVL).

In the US the increased use of I-beams is evident from the increase in number of
producers, which is fast growing. During the 1990’s, the number of plants producing I-
beams increased in number from 16 to 43 (Zylkowski 2000). The use of I-beams has, to a
large extent, replaced the use of solid timber for floor and roof structures.

1.2 Aim
A test method for the web/flange connection of I-beams needs to be developed, since the
existing test methods for glue bonds, as described for example in EN 392 (Anon.1995),
cannot be used due to the geometry of the joint. Typically the web is bonded into a
slightly wedge-shaped groove, see Figure 1. The development of such a test method is the
aim of the current project.
47

4
47

20

Figure 1. Example of web-to-flange geometry of a wood I-beam.

1.3 Current method used by a manufacturer

At the present, one manufacturer uses an in-house method to determine whether the
adhesive bond has an acceptable quality. The method is based on a shear test of a thin
slice of the complete I-beam cross-section. A loading device applies a compressive force
on the web material, as indicated in Figure 2. This method has served as a starting point
for further development within the present project.
 8

Figure 2. A steel plate is used to push through the web material in shear. Only part of the
cross-section and the loading device is shown.

1.4 Work content and disposition of the report

The project has been divided into a number of tasks, of which Tasks 2, 3 and 4 relate to
testing and evaluation of tests using a proposed test method. The work contents of these
tasks are described below.

• Task 2
o The influence of the test piece thickness. Normally, size effects should be
investigated for any kind of material or structural test. A finite element
study was also conducted in this task.
• Task 3
o The influence of the moisture content. Different pre-conditioning
climates are used in order to investigate their effect on the glue bond
quality.
• Task 4
o The influence of pre-treatment (boiling) of the specimens prior to testing
in order to detect any problems with glue bond quality.

Chapter 2 gives a general description of the materials and methods used for the respective
tasks. Chapters 3-5 describe the results from the tests performed in tasks 2-4 and the FE-
simulations performed in task 2. Chapter 6 gives a general discussion on the project
results and includes some final remarks and recommendations. Appendix A, finally,
presents the draft test method as proposed by the project partners.
 9

2 Materials and methods

2.1 Beam materials
For this project it was decided to use two different beam types, manufactured in Sweden
and in Norway. The Swedish beam is manufactured by Masonite Beams AB. The
Norwegian beam is manufactured by Forestia AS and is sold under the commercial name
“Rantibjelken”. The two beams are based on the same principles of an I-beam, but
differing in their material composition and geometrical shape of the web-to-flange joint.
In Figure 3 is shown two cross-sectional views of the beam types investigated.

The beams from Masonite AB are made from an 8 mm thick HDF-board (K40) and
47×47 mm2 solid timber flanges (K24 or K30). The Rantibeam consists of a 10 mm thick
chipboard of structural class and 47×47 mm2 solid timber flanges of an especially graded
class. Both products are subject to technical approvals in their respective countries. The
Masonite beam has a web-to-flange joint that is wedge shaped, while the Rantibeam has a
V-shaped groove in the edge of the web, and a corresponding shape in the flange, see
Figure 3.

Figure 3. The Masonite beam (left) and the Rantibeam (right).

Since the parameters to be studied included the density of the wood flanges, the beams
where selected in order to obtain different density groups (high-medium-low). Another
parameter studied was the influence of gluing errors and the possibility to detect them.
Therefore, a number of beams were produced were the mixing of the adhesive was
deliberately wrong (the recommended adhesive/hardener mixing ratio was altered) and
also a series were the amount of glue applied differed from the amount recommended by
the manufacturer.

2.2 Test set-ups

Due to the fact that the different beams have differently shaped web-to-flange joints, the
test setups must be adapted correspondingly. The main principle remains the same,
however. A loading device is used to push through the web material from its connection
to the flange. The test piece is placed on a self-aligning steel plate, which has a cut shaped
as the web-to-flange joint. A loading device with a corresponding shape is used to push
through the web material. The precise geometry of the loading device and the cut in the
supporting plate is determined such that the distance between the bond line (i.e. the
 10

intended shear plane) and the loading device is 0.5-1 mm. The parts used in obtaining this
test set-up are shown schematically in Figure 4.

2
1

Figure 4. Schematic of test set-ups used with test specimen (1), loading device (2) and
support plate (3). The support plate must be self-aligning.
 11

3 Task 2 – Influence of specimen thickness

3.1 General remarks
The basic idea behind this task is to investigate any possible size effect. The original idea
was to use the same thickness as that used in EN 392 (75 mm), it turns out, however, that
this is not possible due to the crushing strength of the web materials. It was therefore
decided to test three different specimen thicknesses (in the beam axis direction): 10, 20
and 30 mm. The outcome from this task establishes the reference thickness to be used in
the rest of the project tasks. The tests performed included the two beam types, different
densities and a deliberately induced gluing error.

In addition to the original test plan, a limited finite element study was performed. This
study is presented in section 3.4.

3.2 Test programme

Task 2 included the tests as indicated in Table 1. A total of 450 individual tests were
included in this task.
Table 1. Test programme for task 2.

RH Thickness Number of
Description Tested by
(%) (mm) specimens
MB-1, Low density, thickness 1 65 10 30 SP
MB-1, Low density, thickness 2 65 20 30 SP
MB-1, Low density, thickness 3 65 30 30 SP
MB-2, High density, thickness 1 65 10 30 SP
MB-2, High density, thickness 2 65 20 30 SP
MB-2, High density, thickness 3 65 30 30 SP
RB-B1, Low density, thickness 1 65 10 30 NTI
RB-B1, Low density, thickness 2 65 20 30 NTI
RB-B1, Low density, thickness 3 65 30 30 NTI
RB-A1, High density, thickness 1 65 10 30 NTI
RB-A1, High density, thickness 2 65 20 30 NTI
RB-A1, High density, thickness 3 65 30 30 NTI
RB-C1, Gluing error, thickness 1 65 10 30 NTI
RB-C1, Gluing error, thickness 2 65 20 30 NTI
RB-C1, Gluing error, thickness 3 65 30 30 NTI

3.3 Test results

The test results from the tests are summarised in Table 2 and Table 3 for the Ranti beam
and the Masonite beam respectively. The density and the moisture content were measured
on specimens cut from the vicinity of the specimens used for the shear tests (Ranti) or
from the 30 mm thick specimen (Masonite).
 12

Table 2. Shear test results. Ranti-beam.

Thickness Shear strength Wood failure Density Moisture content

(mm) (MPa) (%) (kg/m3) (%)
Mean Std. Mean Std. Mean Std. Mean Std.
dev. dev. dev. dev.
Low density
10 9.00 0.88 92 7
20 7.30 0.59 90 8 493 6.4 13.0 0.13
30 6.35 0.48 91 6
High density
10 8.71 1.21 90 13
20 7.78 0.71 86 11 576 13.4 13.6 0.13
30 6.55 0.54 86 16
50% hardener
10 9.52 2.17 83 18
20 7.66 1.28 81 21 470 24.7 13.1 0.13
30 6.60 1.00 83 20

Table 3. Shear test results. Masonite beam.

Thickness Shear Wood failure Web material Density Moisture

strength failure content
Mean Std. Mean Std. Mean Std. Mean Std. Mean Std.
dev. dev. dev. dev. dev.
(mm) MPa MPa % % % % kg/m³ kg/m³ % %
Low density
10 4.85 1.04 99 3 67 24 - - - -
20 5.13 0.70 100 1 57 17 - - - -
30 5.44 0.33 100 2 52 15 350 11 11.7 0.6
High density
10 5.78 0.72 98 4 68 21 - - - -
20 5.99 0.63 99 3 65 20 - - - -
30 4.89 0.39 100 2 71 23 398 7 11.7 0.5

3.4 Finite element study

3.4.1 FE-model
The geometries of the specimens were modelled using the nominal dimensions as
provided by the manufacturers. The cut in the flange and the corresponding part of the
web were modelled with the same dimensions, giving a perfect fit. However, since the
bond line thickness is finite, there will be a mismatch between the two parts. The
adhesive bond lines were assumed to be of 0.1 mm thickness. Details of the two
specimens are shown in Figure 5 and in Figure 6. The loading devices were modelled
with rigid plates.
 13

47

4
47

20

Figure 5. Nominal geometry and solid model of the Masonite specimen (symmetric half).

47
47

16

10

Figure 6. Nominal geometry and solid model of the Ranti specimen (symmetric half).

The test specimens were modelled using linear elastic materials for the flanges and the
webs and with a nonlinear fracture-softening model for the adhesive bond line. For the
bond line one must define both fracture energy and strength, for both pure shear and for
pure normal deformation (peel stress). The adhesive layer is characterised not only by its
shear properties, but also by its strength and fracture energy perpendicular to the bond
line. When acting in a combined state of deformation including both shear and normal
deformation, the bond line model accounts for different strength and fracture energy as
compared to the values for uni-axial states. The local strength and fracture energy of the
adhesive layer was set to 18 MPa and 1250 J/m2, respectively, which correspond to a
brittle adhesive, such as a phenolic resorcinol. The bond line thickness was set to 0.1 mm.
The specific elastic constants used in the simulations are summarised in Table 4 and in
Table 5.
Table 4. Material parameters used in the finite element study.

Beam Eweb (MPa) Gweb (MPa) Eflange,0 (MPa) Eflange,90 (MPa) Gflange,0/90 (MPa)
Masonite 5000 2100 12000 400 800
Ranti 3000 1300 13000 430 810
 14

Table 5. Material parameters used in the finite element study.

Shear strength Peel strength Fracture energy, Gf,I Fracture energy, Gf, II
(MPa) (MPa) (J/m2) (J/m2)
18 6 550 1250

The loading was applied by the use of rigid surfaces and the contact modelling
capabilities of the software used (ABAQUS 6.3). The flange piece was assumed to
interact with a rigid surface, which is constrained in all directions. The loading piece was
modelled with a rigid surface interacting with the web part. The loading piece was
prescribed to move in the loading direction while constraining all other directions. The
coefficient of friction was set to μ=0.6 for all cases.
3.4.2 FE-Results
The results from the FE-simulations are summarised in Table 6, which gives calculated
load-bearing capacities and corresponding nominal shear strengths. Figure 7 givas, as an
example the deformation of the Masonite specimens. The load-deformation curves for the
two specimens are shown in Figure 8.
Table 6. Results from FE-simulations. Loads and areas are for the symmetric half.

Beam-Length Max load, Pmax Shear area, A Shear strength, fv Compressive

(N) (mm2) (MPa) stress (MPa) 1
MB-10 mm 2533 191 13.3 42
MB-20 mm 2802 382 7.3 47
MB-30 mm 3226 573 5.6 54
RB-10 mm 1944 179 10.9 26
RB-20 mm 2449 358 6.9 33
RB-30 mm 2822 536 5.3 38

Figure 7. The deformation during the course of loading of the Masonite beam. The
deformations are scaled a factor of 20, for clarity.

1
Calculated under the assumption that the applied load is uniformly distributed over the web
material. The areas were set to 60 mm2 and 75 mm2 for the symmetric halves of the Masonite- and
Ranti- beam, respectively.
 15

Figure 8. Load-deformation from nonlinear FE-analyses. Masonite-beam (left) and Ranti-

beam (right).

3.5 Conclusions – Task 2

The test results from the tests with the Ranti beam show a clear trend: thicker test
specimens tend to give a lower shear strength prediction. The results from the tests with
the Masonite beam, however, are more inconclusive. The wood failure percentages are
higher for these than for the Ranti beam tests. Also, the web material failure percentage is
high when the failure is outside the bond line, as can be seen in Table 3. This is
interesting in terms of the design of the test method. Possibly, the Masonite-beam has had
a larger amount of web material failure, and thus the influence of the bond specimen
thickness is not as obvious. The failure of the web material could be a compressive
failure, but the state of stress at testing is complex and the tensile strength of the material
perpendicular to the web plane could also play a role here.

The tests with the Ranti beam included a series whit only 50% of the recommended
amount of hardener being used. The resulting shear strength of the bond line was not
affected by this gluing error. However, the wood failure percentage was slightly lower for
the specimens with gluing error.

The FE-analyses have shown that it is possible to predict a size effect, similar to the one
seen in the Ranti-beam tests. The FE-analyses were performed using material data, which
have not been calibrated to the present test material. Therefore, the results can only be
used for a qualitative comparison. The FE-analyses have also shown that the influence of
bending of the test specimen is of no importance, and therefore cannot explain any
influence of specimen size.

The specimen should be more than 10 mm thick, in order to avoid that a large fraction of
the bond line is damaged when cut from the beam. However, the specimen thickness
should not exceed approximately 20 mm, in order to avoid failure of the web material.

As a result from this task it was decided that the specimen thickness should be 20 mm, for
the remaining tests.
 16

4 Task 3 – Influence of moisture content

4.1 General remarks
This task relates to the influence of pre-conditioning at different relative humidity levels,
resulting in different moisture contents. The levels of air relative humidity chosen were:

45%: room-climate, dry specimens

65%: standard climate, reference
85%: humid climate, wet specimens

All specimens tested were 20 mm in thickness, as decided from the outcome of task 2.

4.2 Test programme

The tests in this task are summarised in Table 7. A total of 45 specimens, each 20 mm in
thickness, were included in this task. The specimens were uniformly distributed, such that
each beam of three different beams used was represented by 5 specimens. The tests were
performed by VTT.
Table 7. Test programme for task 3.

Description RH No of specimens Tested by

MB-3-45, dry 45% 15 VTT
MB-3-65, reference 65% 15 VTT
MB-3-85, wet 85% 15 VTT

4.3 Test results

The density and the moisture content were determined for 5 specimens from each beam.
Both the mass and the density were measured at the moisture content indicated together
with the results in Table 8.

Table 8. Test results – density and moisture content (MC)

Beam Density MC
Mean Std. dev. Mean Std. dev.
kg/m³ kg/m³ % %
MB-3-1 486 4 14.8 0.1
MB-3-2 465 3 14.7 0.1
MB-3-3 475 6 14.8 0.0

The results from the shear tests are given in Table 9. Two results from series 65% RH
were excluded due to the presence of a knot.

The glue line for the beams was without any flaws, which is indicated by the high wood-
failure percentage – above 90%. The failure was mainly in the web material. The web
material shear strength is higher at 45% RH than at 65% RH, as expected.
 17

Table 9. Shear test results, influence of moisture content.

Series Shear strength Wood failure Web failure

Std. Std. Std.
Mean Mean Mean
dev dev dev
MPa MPa % % % %
MB-3-45 6.65 0.45 98 3 90 10
MB-3-65 5.51 0.53 92 11 79 11
MB-3-85 5.50 0.29 98 4 86 9
 18

5 Task 4 – Influence of pre-treatment

5.1 General remarks
This task relates to the effect of different pre-treatments (boiling) of the test specimens.
The tests in this task were all performed with the specimen thickness 20 mm. The tests
have been performed using three different pre-treatments and with two types of gluing
error, with the following notations:
65%: standard 20°C / 65% climate, used for reference
100%: boiling for two hours followed by storage in 20°C water for two hours,
and
100+100%: boiling for four hours + drying for 18 hours (60°C) + boiling for four
hours+ storage in 20°C water for two hours. This pre-treatment is
required for LVL according to prEN 14374.

Error 1: Adhesive to hardener ratio 100/2, a large gluing error (Masonite beams).

Error 2: Adhesive to hardener ratio 100/4, a small gluing error (Masonite beams).

Error 3: 50% of the recommended amount of glue used (Ranti beams).

Error 4: 50% of the recommended amount of hardener, a large gluing error (Ranti
beams).

5.2 Test programme

This task included the test series indicate in Table 10, making a total of 305 individual
tests. Tests were performed at VTT and SP.

A total of 270 specimens were manufactured from 15 Masonite beams. All specimens
were 20 mm in thickness. Five of the beams (three for series MB-5 and two for series
MB-8) were without any deliberately induced gluing errors, which means that the
adhesive/hardener ratio was 100/15. Five of the beams (three for series MB-6 and two for
series MB-9) were manufactured with a mixing ratio of 100/4, and the last five (three for
series MB-7 and two for series MB-10) had a ratio of 100/2. The specimens were
distributed in the following test series:

MB-5 (3 series): No gluing error, mixing ratio 100/15, reference, 3 x 15 specimens

MB-6 (3 series): Gluing error, mixing ratio 100/4, 3 x 15 specimens
MB-7 (3 series): Gluing error, mixing ratio 100/2, 3 x 15 specimens
MB-8 (3 series): No gluing error, mixing ratio 100/15, reference, 3 x 15 specimens
MB-9 (3 series): Gluing error, mixing ratio 100/4, 3 x 15 specimens
MB-10 (3 series): Gluing error, mixing ratio 100/2, 3 x 15 specimens

A total of 135 specimens were manufactured from 3 Ranti beams. All specimens were 20
mm in thickness. One of the beams (RB-A3) was without any deliberately induced gluing
error. One of the beams (RB-C2) was manufactured with a correct mixing ratio but with a
too small amount of mixed glue (50%). The last beam (RB-C3) had a mixing ratio with a
reduced amount of hardener (50%). The specimens were distributed in the following 9
test series:
 19

RB-A3 (3 series): No gluing error, reference, 3 x 15 specimens

RB-C2 (3 series): Gluing error, 50% of glue amount, correct mixing ratio
RB-C3 (3 series): Gluing error, 50% of hardener, correct amount of mixed glue

Table 10. Test programme for task 4.

Description RH Number of tests Tested by

MB-5, reference 65% 15 VTT
MB-5, boiling 100% 15 VTT
MB-5, 2×boiling 100+100% 15 VTT
MB-6, gluing error 1, reference 65% 15 VTT
MB-6, gluing error 1, boiling 100% 15 VTT
MB-6, gluing error 1, 2×boiling 100+100% 15 VTT
MB-7, gluing error 2, reference 65% 15 VTT
MB-7, gluing error 2, boiling 100% 15 VTT
MB-7, gluing error 2, 2×boiling 100+100% 15 VTT
RB-A3, reference 65% 15 VTT
RB-A3, boiling 100% 15 VTT
RB-A3, 2×boiling 100+100% 15 VTT
RB-C2, gluing error 3, reference 65% 15 VTT
RB-C2, gluing error 3, boiling 100% 15 VTT
RB-C2, gluing error 3, 2×boiling 100+100% 15 VTT
RB-C3, gluing error 4, reference 65% 15 VTT
RB-C3, gluing error 4, boiling 100% 15 VTT
RB-C3, gluing error 4, 2×boiling 100+100% 15 VTT
MB-8, Reference 65% 15 SP
MB-8, Boiling 100% 15 SP
MB-8, 2×boiling 100+100% 15 SP
MB-9, Gluing error 1, reference 65% 15 SP
MB-9, Gluing error 1, boiling 100% 15 SP
MB-9, Gluing error 1, 2×boiling 100+100% 15 SP
MB-10, Gluing error 2, reference 65% 15 SP
MB-10, Gluing error 2, boiling 100% 15 SP
MB-10, Gluing error 2, 2×boiling 100+100% 15 SP

5.3 Test results

The densities and the moisture contents were determined from 5 specimens from each
Masonite beam and from 15 specimens from each Ranti beam. The mass and the volume
of the specimens were determined at the moisture content indicated. The results are given
in Table 11.
 20

Table 11. Test results, density and moisture content.

Series Density MC
Std. Std.
Mean Mean
dev. dev
kg/m³ kg/m³ % %
MB-5 475 10 14.8 0.1
MB-6 442 30 14.0 0.1
MB-7 451 32 14.1 0.1
MB-8 483 16.7 13.7 0.7
MB-9 396 17.6 13.9 0.8
MB-10 381 16.1 14.4 0.8
RB-A3 396 7 13.7 0.1
RB-C2 515 9 13.9 0.1
RB-C3 478 5 13.8 0.1

The shear strength results are indicated in Table 12 - Table 14. One result was excluded
from series R-1 of Table 12 and two results from series R-1 of Table 13 due to the
presence of knots.
Table 12. Test results, influence of gluing error. 20/65.

Series Shear strength Wood failure Web failure

Std. Std. Std.
Mean Mean Mean
dev. dev. dev.
MPa MPa % % % %
MB-5 5.51 0.53 92 11 79 11
MB-6 6.52 0.68 93 5 19 8
MB-7 5.61 0.56 84 7 11 12
MB-8 4.61 0.77 96 4 84 12
MB-9 3.46 0.84 52 27 46 36
MB-10 4.90 0.93 93 1 24 25
RB-A2 8.80 2.51 88 28 16 29
RB-C2 3.51 0.38 50 4 50 6
RB-C3 8.14 2.69 90 19 15 17
Table 13. Test results, influence of gluing errors, Boiling 2 h + storage in 20° water for 2 h.

Series Shear strength Wood failure Web failure

Std. Std. Std.
Mean Mean Mean
dev. dev. dev.
MPa MPa % % % %
MB-5 1.83 0.24 96 2 81 10
MB-6 1.78 0.24 75 16 39 13
MB-7 1.40 0.24 39 16 15 9
MB-8 2.04 0.12 100 1 99 3
MB-9 1.50 0.28 30 16 100 0
MB-10 1.61 0.25 66 17 69 21
RB-A2 3.10 0.62 90 20 82 23
RB-C2 1.77 0.21 48 12 94 9
RB-C3 3.01 0.46 91 10 47 4
 21

Table 14. Test results, influence of gluing error, Boiling 2 h + drying in 60°C for 18 h +
boiling 4 h + storage in 20°C water for 2 h.

Series Shear strength Wood failure Web failure

Std. Std. Std.
Mean Mean Mean
dev. dev. dev.
MPa MPa % % % %
MB-5 1.72 0.27 94 8 91 8
MB-6 1.30 0.12 64 9 50 11
MB-7 0.91 0.22 41 14 26 11
MB-8 1.36 0.17 100 0 95 9
MB-9 0.78 0.15 50.0 19 73 23
MB-10 1.15 0.18 77 13 53 21
RB-A2 2.35 0.23 93 8 92 8
RB-C2 1.44 0.21 49 2 92 15
RB-C3 2.61 0.46 94 10 67 22

The beam used for the series without gluing error in series MB-8, was taken from beams
previously sent to SP for production control. It turned out that this beam, unfortunately,
had a significantly higher wood density, as can be seen in Figure 9, showing the
lengthwise density variation. As additional examples of the lengthwise variation of the
properties measured the results from series MB-8, MB-9 and MB-10 are shown in Figure
10, Figure 11 and Figure 12 for the three pre-treatments investigated.

Density 20°C 65 %RH

600

500
MB-8 reference
Density (kg/m3)

400
MB-9 large gluing error
300

200 MB-10 Small gluing

error
100

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Figure 9. Dry densities for test series MB-8 – MB-10, all for testing without prior boiling.
The peak value for the MB-5 series was due to a specimen containing a knot.
 22

20°C 65 %RH

8.00

Shear strength (MPa)

7.00 MB-8 Reference
6.00
5.00
MB-9 Large gluing
4.00
error
3.00
2.00 MB-10 Small gluing
1.00 error
0.00
1 3 5 7 9 11 13 15
Specim en num ber

Figure 10. Shear strength after pre-conditioning at 20/65.

100% RH (Boiling once)

8.00
Shear strength (MPa)

MB-8 Reference
6.00

MB-9 Large gluing

4.00
error

2.00 MB-10 Small gluing

error
0.00
11

13

15
1

S pe c im e n num be r

Figure 11. Shear strength after boiling once.

100+100% RH (Boiling tw ice)

8.00
Shear strength (MPa)

MB-8 Reference
6.00

4.00 MB-9 Large gluing

error
2.00 MB-10 Small gluing
error
0.00
1 3 5 7 9 11 13 15
Specim en num ber

Figure 12. Shear strength after boiling twice.

 23

6 Discussion and conclusions

6.1 Gluing errors

The aim of this investigation has been to develop a test method for detecting possible
gluing errors. Thus, it is of course important to define the precise meaning of a gluing
error. Obviously, applying only half the recommended amount of the mixed adhesive is
an error. Likewise it is to be regarded as an error if only a small fraction of the hardener
amount as recommended by the adhesive manufacturer is used. Both these gluing errors
have been used here. However, the amount of hardener to use as specified by the adhesive
manufacturer is often given with wide safety margins, such that in some cases using only
half the recommended amount can still give acceptable strength. On the other hand, if an
amount of hardener distinctively different from the one recommended by the
manufacturer is used, then the adhesive is not the same as the one, which has passed the
various criteria necessary for structural adhesives.

6.2 Detecting errors

6.2.1 General
The shear tests performed have shown that:
1. Often, there is no significant difference between the average shear strength of the
reference specimens and the ones with a small gluing error, for the case of no
boiling prior to testing.
2. For many cases, when using boiling as a pre-treatment, there is a significant
difference in average shear strength.
3. Generally, the wood failure percentage is not a good indicator for gluing errors
for the case without boiling. Small gluing errors cannot be detected in this way.
After boiling, however, it is possible to detect also small gluing errors. The wood
failure percentage typically drops to approximately 70-80% from almost 100% in
the reference pieces.
4. The method of boiling twice is a very sensitive method for detecting the large
gluing errors. The difference in average shear strength is much easier to detect
after boiling twice as compared to only boiling once.
6.2.2 Without pre-treatment
It is a general conclusion from this study that, apart from extremely serious gluing errors,
it is difficult to detect gluing errors without prior pre-treatment. Two exceptions though:
for the Ranti-beam both the shear strength and the wood failure percentages are low for
series RB-C2, 50% of adhesive amount, and also for the series MB-9 (13% of correct
hardener amount) it was possible to detect the defect. No other gluing errors were
detectable.
6.2.3 With pre-treatment (boiling)
As to what errors are detectable with boiling prior to testing. Of course, the errors that
were detected without boiling are generally also detectable after boiling. For the Ranti-
beam it has not been possible to detect the error with 50% hardener, which could indicate
large safety margins from the adhesive manufacturer in terms of mixing ratio. It is
generally possible to detect the remaining gluing errors for the Ranti and the Masonite
beam, especially if the wood failure percentage is taken into account.

For some cases it is enough to boil only once, such as for the test series RB-C2 and MB-
7. It is however interesting to note that after boiling only once, in the tests of series MB-8
 24

and MB-9 the gluing error with a low hardener content was detected, but this is not the
case for the nominally equal tests of series MB-6.

Boiling only once is probably a useable method, if combined with repeated boiling for
those cases where a low wood-failure percentage is detected.

6.2.4 Decision based on shear strength or wood failure

percentage
Although the amount of wood failure is a somewhat subjective measure, it is still better to
use than the shear strength. Firstly, the main interest here lies in asserting that the
adhesive strength is good enough, which means better than the wood material. Secondly,
it is difficult to decide on a shear strength value to be used for every combination of
wood-based materials and adhesive. A generally acceptable wood failure percentage
could be 70% and above if repeated boiling is used. As an alternative, an 80% level of
wood failure could be used in combination with boiling only once. Finally, the use of the
wood failure percentage as a decision- making criterion makes the use of a load-
measuring device unnecessary. Of course, it could be of interest to use both a shear
strength minimum value and the wood failure percentage as criteria. However, in cases
where the shear strength is tested in another way it should suffice with using only the
wood failure percentage.
 25

References
Anon. (1987). NT Build 327 Light weight beams of wood: Load bearing capacity and
rigidity.

Anon. (1995). EN 392 Glued laminated timber – Shear test of glue lines

Anon. (2001). EOTA Guide line 011

Anon. (2003). prEN 14374 Timber structures - Structural laminated veneer lumber -
Requirements.

Axelsson, M., Johansson, C.-J.(2000). Light composite wood based beams. Nordic
Round Robin test.

Fonselius, M., Johansson, C.-J., Solli, K. (1994). Dimensionering av träkonstruktioner

genom provning. SP Rapport 1994:60.

Zylkowski, S. (2000). Engineered wood products in North America. Presentation to

COST Action E13
 A1

Appendix A: Wooden I-beams – Test method for

evaluation of flange/web glue bond quality

Foreword
This method was prepared by Nordtest. It is aimed as a compliment to EOTA Guideline
011 – Light Composite Wood-based Beams and Columns.

1 Scope
This document describes a method for measuring the glue bond quality of the adhesive
bond between flange and web in wooden I-beams.

2 Normative references
None
3 Definitions
For the purpose of this standard, the following definitions apply:

test specimen a piece cut perpendicular to the length of the beam

containing part of one flange and the web

wood I-beam composite wood based beam with I-shaped cross section
wood failure rupture in or between wood fibres
wood failure percentage the percentage of wood failure area in relation to sheared
area

4 Symbols and abbreviations

The following symbols and abbreviations are used in this standard:

fv nominal shear strength

fv, ref reference shear strength
Fu ultimate load
Af shear area
αwf wood failure percentage
αwf, ref reference wood failure percentage

5 Bond line shear test

5.1 Principle
The principle is to apply a compressive force to the test piece, resulting primarily in a
state of shear of the bond line so that inadequate bonding quality will result in failure in
the very bond line. Failure outside the bond line is thus regarded as an indicator of
adequate bond quality. The wood failure percentage and the shear strength are used as
measures of the glue bond quality. The method described gives the possibility to measure
both these parameter, but for some applications there might be alternative methods
 A2

available for evaluating the shear strength. For such situations only the wood failure
percentage needs to be determined.

5.2 Apparatus
5.2.1 Testing machine
For measuring the shear strength, a calibrated testing machine capable of applying a
compressive force to the shearing tool, mentioned in subsection 5.2.2, is needed. The
inaccuracy of the measurement of the maximum load should be less than ± 3%. If only
the wood failure percentage is to be estimated, the applied load needs not to be measured.
5.2.2 Shearing tool
Figure 1 illustrates a shearing tool. The cylindrical bearing shall be self-aligning so that
the test piece is loaded at the end grain with stress field uniform over the whole surface.

Test
specimen

Spherical
bearing

Figure A1. Shearing tool with specimen

5.3 Specimen preparation

5.3.1 Test piece
Take care in preparing the test piece to make the loaded surfaces smooth and parallel to
each other and perpendicular to the grain direction. The test piece shall be of the form
shown in Figure 2. The thickness (in the beam axial direction) should be 20 mm.

Figure A2. Example of test piece.

5.4 Procedure
5.4.1 Measurements
Measure the dimensions from which the sheared area is determined to the nearest 0.5 mm.
 A3

5.4.2 Test specimen pre-treatments

Either of the below pre-treatments A or B should be used.
A. Boiling in water at 100°C for 2 hours and stored in water at 20°C for 2
hours before testing.
B. Boiling in water at 100°C for 4 hours + drying for 18 hours (60°C) +
boiling for 4 hours + storage in 20°C water for 2 hours before testing.
5.4.3 Specimen positioning
Place the test piece in the shearing tool so that it is loaded in the direction of the grain.
The bond surfaces shall be positioned so that the distance between them and the sheared
plane is at least 0.5 mm but not exceeding 1 mm.
5.4.4 Loading conditions
The loading shall be undertaken at a constant rate and so that failure occurs after 20
seconds.
5.4.5 Test results, data acquisition and results evaluation
If the shear strength is to be determined, measure the maximum force during the complete
course of loading. Determine the shear strength fv in MPa with two significant digits
from the following equation

Fu
fv=
Af

where

Fu is the ultimate load

Af is the nominal fracture area in. This is set equal to the bond line area of the
web-to-flange joint.

After failure, estimate the amount of wood failure percentage rounded off to the nearest
figure divisible by 5.

5.4.6 Evaluation of bond quality

The bond quality is regarded as adequate if the following criterion is fulfilled:

fv > fv, ref and αwf > αwf, ref

The value of fv,ref should be determined at the initial type testing.

The value of αwf, ref is:

αwf, ref = 80% for pre-treatment method A according to section 5.4.2 or

αwf, ref = 70% for pre-treatment method B according to section 5.4.2.
 A4

5.5 Report
When reporting the results the following should be included:

1. A reference that the testing was performed in accordance with this method
2. The number of test specimens used
3. The shear strength
4. The percentage of wood failure
5. Any significant deviation from the procedures described in this document
6. Any other observation that might have influenced the results

Bibliography
Anon. (2001). EOTA Guide line 011