You are on page 1of 6

BENDING TESTS ON THE CONTINUOUS GLULAM BEAM

Maurice Brunner
1
, Martin Lehmann
2






ABSTRACT: There are few published data on the testing of continuous beams. The valid standards for timber
structures assume that the continuous beam is fully elastic till failure and that it possesses the same bending strength as
simply supported beams. The authors have performed bending tests on glulam beams spanning two spans. The results
indicate that the load-bearing behaviour of the two-span beam may be much more complex than is often assumed in
current design practices. Many of the tested beams exhibited clear signs of plastic behaviour before bending failure, and
some of them exhibited full elastic behaviour till failure. The failure moment of the continuous timber beams was
markedly greater than that of simple beams which were also tested as a control. Apart from the above mentioned plastic
effects, favourable volume effects could help to explain the observation of the increased bending strength of the two-
span beam in comparison to the simple beam.
KEYWORDS: Glulam, two-span beam, higher bending strength, reduced scattering


1 INTRODUCTION
12

The continuous timber beam offers technical advantages
over the simply supported beam. In particular, the
reduced deflections are no longer the dominant
governing factor for the design, thus the bending
strength can be fully exploited. The maximum bending
moment is generally smaller in the continuous beam in
comparison to the simply supported beam of the same
span. Modern production and transportation techniques
make it easier today to use the continuous beam in
practical timber structures. Many timber engineers
prescribe continuous beams where appropriate in order
to reduce member sizes and thus help save project costs.
The continuous timber beam has not been as intensively
researched as the simply supported beam. Most timber
researchers limit their laboratory tests to simply
supported members and structures. In a recent work,
Frese [1] has carried out numerical investigations on
continuous beams. He comes to the conclusion that due
to the volume effect, the bending strength for the
hogging moment will be much higher that for the simply
supported beam. Frese did not consider possible plastic
behaviour in the beam over the central support, and no
bending tests were carried out to support his claims.
In two test series, the authors have tested the continuous
glulam beam. The methods used will be described in this
paper. The test results will be presented and discussed

1
Maurice Brunner, Bern University of Applied Sciences,
Architecture Wood & Civil Engineering, Solothurnstr. 102,
2504 Biel, Switzerland. Email: maurice.brunner@bfh.ch
2
Martin Lehmann, Bern University of Applied Sciences,
Architecture Wood & Civil Engineering, Solothurnstr. 102,
2504 Biel, Switzerland. Email: martin.lehmann@bfh.ch
with the help of simple elastic and plastic calculation
models. The different volume effects which occur in the
continuous beam as opposed to the simple beam will
also be discussed.

2 OBJECTIVES
The authors have researched the two basic assumptions
used by most structural engineers to design continuous
glulam beams:
Do continuous timber beams really exhibit fully
elastic behaviour till failure?
Is the bending strength really the same as for
the simply supported beam?

3 MATERIAL AND METHODS
In two series, the authors performed bending tests on
two-span beams. Parameters included different cross-
sections as well as the length of the steel plate of the
central support. As a control, simply supported beams of
the same size and grades and from the same supplier
were also tested according to EN 408:2010 [2].
As no testing standards for two-span beams exist in
Europe, the EN 408 was adapted: both spans were
loaded in the third points with equal loads, thus the
whole beam had four external loading points (Figure 1).


Figure 1: Test set-up for the two-span beam

F/2 F/2
h
F/2 F/2
L/3 L/3 L/3 L/3 L/3 L/3
L L
Figure 2 shows the pivoted steel plate at one end
support. Both end supports had such plates.

Figure 2: End supports of the two-span beams were on
classical pivoted steel plates

The external loads and the reaction of the central support
were measured. Figure 3 shows the measuring device for
the reaction of the central support. Also visible are strain
gauges.


Figure 3: Measuring device for the reaction at the central
support, strain gauges.

4 RESULTS
4.1 FIRST TEST SERIES
This series of tests was carried out in 2009. All the test
beams were of grade GL24h. They had a width of
100mm and a height of 200mm [3]. The span of the six
simply supported control beams was 4m, whilst the six
continuous beams crossed two spans of 4m each. The
steel plate of the central support was 100mm long.
A visual inspection revealed extensive plastic
deformations in the compressive face of all the tested
two-span beams over the central support (Figure 4).


Figure 4: Plastification in the compressive face over the
central support was observed in all beams of the first test
series.
Figure 5 shows the typical deflection of the two-span
beams. The particular beam shown was an exceptional
case of extreme plastification over the central support:
the resulting load redistribution led to failure in both
fields (marked with stars) instead of over the central
support as would be expected in a fully elastic material.


Figure 5: Deflection of the two-span beam under loading.

Since both the external loads and the reaction at the
central support were measured, it was possible to
calculate the two end support reactions and hence the
prevailing internal forces. Figure 6 shows the moment
growth in a test beam as the external loads in the two
spans 1 and 2 are increased. Also shown is the growth in
the reaction of the central support. After an initial elastic
phase, the hogging moment remains fairly constant
despite the increasing loads because of the forming of a
plastic hinge.

Figure 6: In the elastic domain the moment growth over
the central support of a test beam is directly proportional
to the measured external loads (red and blue) and the
measured reaction of the central support (indigo). The
hogging moment remains constant at peak load because
of the forming of a plastic hinge over the central support.

The load redistribution observed in the all the tested two-
span beams was confirmed by theory: After an initial
fully elastic stage, during which the force increase is
proportional to the deflection, the force-deflection curve
flattens as plastification effects occur in the beam over
the central support (figure 7). The theoretical
calculations are discussed in Chapter 5.

35
30
25
20
15
10
5
0
0 20 40 60 80
m
o
m
e
n
t

o
v
e
r

c
e
n
t
r
a
l

s
u
p
p
o
r
t

[
k
N
m
]
force[kN]
forcespan1
forcespan2
reactioncentalsup.

Figure 7: The measured beam deflection was
proportional to the loads during the initial elastic phase.
The curve flattens as plastification effects occur in the
beam over the central support at higher loads

Table 1 shows that the two-span beams exhibited much
larger calculated failure moments than the simply
supported control beams. The variation coefficient
(COV) was also significantly smaller. In engineering
practice, not the average strength value but rather the 5%
fractile value is the relevant value used for design
purposes, and this value is favourably affected by a low
coefficient variation.

Table 1: The calculated average failure moments in the
two-span beams compare favourably to those of the
simple beam.
Beam Type Average M
U
[kNm] COV
One span 19.9 21%
Two span 33.6
(+69%)
14%

4.2 SECOND TEST SERIES
The second test series was carried out in 2011. As in the
first test series, the timber grade used was GL 24h. The
beams were 160mm wide and 200mm high. The two-
spans were 3.6m each. The most important parameter
investigated was the length of the steel plate of the
central support: 100mm, 150mm and 200mm. For each
of the three central support length parameters, four
beams were tested.
Figure 8 shows the typical beam failure in bending in
two places: first over the central support, followed
almost immediately by failure in one of the spans.

Figure 8: Typical beam bending failure over the central
support and in midspan

All twelve beams failed in bending as expected. Unlike
the case of the first test series, not all the two-span
beams exhibited plastification effects. The length of the
steel plate of the central support seemed to have some
influence on the load-bearing behaviour of the two-span
beams. In one of the beams with a steel plate length of
100mm over the central support, as well as in three of
the beams with a steel plate length of 150mm, a partial
failure over the central support led to a favourable load
redistribution, and severe compressive strains were
observed (Figure 9).


Figure 9: Plastification effects were observed in only four
of twelve beams of the second test series.

The load redistribution observed in the four beams was
confirmed by theory: After an initial fully elastic stage,
the measured central support reaction was observed to
possess a value lying between the theoretical values of a
fully elastic beam and a fully plastified beam. Some
beams exhibited nearly fully plastic failure (figure 10).
The theoretical calculations are discussed in Chapter 5.


Figure 10: The measured central support reaction (red
curve) of this beam shows that after an initial fully elastic
phase (blue curve), it exhibited nearly fully plastic failure
(purple curve under the blue and red curves).

In the remaining eight beams, the failure mode was
elastic, with no tangible signs of plastification. Figure 11
shows an example of failure in the finger joints over the
central support. Figure 12 confirms that there is no load
redistribution as would be expected if there were any
plastification effects: during the entire loading cycle
until the highest failure load was attained, the measured
central support reaction was observed to possess a value
corresponding exactly to the theoretical values of a fully
elastic beam. The theoretical calculations are discussed
in Chapter 5.

0
20
40
60
80
100
120
0 20 40 60
f
o
r
c
e

[
k
N
]
crossheadmovement[mm]
span1
span2
total
0
20
40
60
80
100
120
140
0 20 40
r
e
a
c
t
i
o
n

c
e
n
t
r
a
l

s
u
p
p
o
r
t

[
k
N
]
crossheadmovement[mm]
measured
elastic
plastic

Figure 11: Eight of the twelve beams of the second test
series failed in fully elastic behaviour, often across the
finger joints.


Figure 12: The measured central support reaction (red)
of eight beams corresponds exactly to the theoretical
values for fully elastic beam behaviour (green) till failure
(highest load).

The calculated failure moments are listed in Table 2. As
has already been observed in the first test series, the two-
span beam has a clearly higher average bending strength
than six simple beams, which were also tested as a
control. The variation coefficient is also smaller and thus
more favourable for the two-span beam as compared to
the simple beam.

Table 2: Calculated failure moments of the single and
two-span beams
Beam
Type
Length central
support [mm]
Average
M
U
[kNm]
COV
One span = 36.9 24%
Two span 100 43.9 25%
Two span 150 56.2 7%
Two span 200 45.5 13%

All two-span beams
together
48.5
(+32%)
15%

The beam deflections were also measured and used to
calculate the global E-Moduli of the test specimens with
fully elastic calculation methods. Fig. 13 indicates that
the simple beams exhibited an average E-modulus which
was about 8% higher than that of the two-span beams.
This result cannot be considered to be as significant as
the finding of the higher strength for the two-span beam
as compared to the simple beam. There were too few
beams and the scattering was considerable.


Figure 13: Calculated average E-Moduli in N/mm2 for
the simple and two-span test beams

5 THEORETICAL CONSIDERATIONS
5.1 Overview
The test results indicate clearly that the two-span beam
seems to have a greater bending strength than the control
simple beams of the same grade and supplied by the
same glulam manufacturer. The authors believe that
these findings can be attributed to two main effects.
First, plastification effects were clearly evident in many
of the two-span beams. Secondly, since even those two-
span beams which did not exhibit any signs of
plastification were still of higher bending strength than
the simply supported beam, another plausible
explanation could be found in the so-called volume
effect. These two ideas will be discussed in simple form
in the following.

5.2 Two-span beams with fully elastic behaviour
The calculation of the fully elastic beam for the test set-
up follows standard structural analysis principles. If the
total load F is applied in four equal forces
1
/
4
F at the
third points of both spans of length L, the end support
reactions are
1
/
6
F, whilst the central support reaction is
2
/
3
F. The hogging moment over the central support is
1
/
12
FL, whilst the largest positive midspan moment is
1
/
18
FL and thus much smaller (figure 14).


Figure 14: Calculated support reactions and internal
forces of a fully elastic two-span test beam.

5.3 Two-span beams with fully plastic behaviour
In comparison to steel and reinforced concrete engineers,
many timber engineers are less familiar with plastic
design. Steel is the ideal plastic structural material. A
0
20
40
60
80
100
120
140
0 20 40
r
e
a
c
t
i
o
n

c
e
n
t
r
a
l

s
u
p
p
o
r
t

[
k
N
]
crossheadmovement[mm]
measured
elastic
F /4
F L/12
F L/36
F L/18
M
F L/18
F L/36
F /6
F /12
F /3
F /3
F /12
F /6
V
el F /4 el F /4 el F /4 el
F /6 el 2F /3 el F /6 el
el
el
el el
el
el
el
el
el
el
el
Simple beam

2-span beam
two-span steel test beam under increasing loads will
exhibit an initial fully elastic stage as shown in figure 14
above. When the ultimate moment is attained over the
central support, failure does not occur immediately but
rather, a plastic hinge will develop in the beam at this
place. As the loads increase, the moment over the central
support will remain constant and the smaller moment in
midspan will increase. Failure occurs when a second
plastic hinge is attained at the midspan. The load
redistribution can be calculated by moving the support
moment (-FL/12) of figure 14 down, reducing it by M
whilst keeping the end support moment at zero. The
maximum positive field moment (FL/18) would increase
accordingly by
1
/
3
M (figure 15).

Figure 15: Plastic load redistribution: reducing the
support moment by M will cause the maximum field
moment of the two-span beam to increase by M/3.

The calculation is really simple:
l F
l F l F
M
l F
l F l F
M
l F
M
M
l F M l F
M M


16
1
48 12
16
1
48 3
1
18
48
12 3 18
min
max
min max


Once the redistributed hogging (negative) moment and
the redistributed maximum positive moment in the field
are known to have the same value of
1
/
16
FL, the support
reactions and the other internal forces can be readily
calculated. The results of the load redistribution of the
ideal plastic beam are depicted in figure 16.

Figure 16: Calculated support reactions and internal
forces of a fully plastic two-span test beam.
As has been mentioned above, during the bending tests
the central support reaction and the external loads were
continuously measured till failure. The measured values
were compared to the values calculated according to the
theoretical models for the fully elastic and the fully
plastic beams. The comparison helped to confirm that
many of the two-span test beams exhibited partial
plastification before failure occurred.

5.4 The volume effect
Timber engineers are familiar with the fact that the
tensile strength of timber is much smaller than the
bending strength. This is because in a tensile test the
whole volume of the timber member is under the same
peak stress. In the bending test of a simple beam the
highest tensile stresses are concentrated in a much
smaller volume of the test member (figure 17). The
reasons are first, the peak moment is achieved round
midspan, secondly the critical, maximum tensile stress is
only concentrated at the bottom surface of the beam. As
the weakest links of the timber member (knots, shakes
etc.) are generally uniformly distributed throughout the
body of the timber member, the probability of failure is
greater for the test specimen in a tensile test as compared
to one under bending.

Figure 17: Volume effect I: The entire volume of a tensile
member is under peak tensile stress, whilst in a beam
only a relatively small volume is under peak tensile
stress.

If a two-span test beam is compared to a simple test
beam, it is evident that an even smaller volume of the
two-span beam is under peak tensile stresses. Thus the
two-span beam should have a much higher bending
strength than the simple beam (figure 18).
F /4
F L/16
F L/24
F L/16
M
F L/16
F L/24
3F /16
F /16
5F /16
5F /16
F /16
3F /16
V
pl F /4 pl F /4 pl F /4 pl
3F /16
pl
5F /8 pl pl
pl
pl
pl pl
pl
pl
pl
pl
pl
pl
pl
3F /16
F
N
F

F
F/2
M
F/2
F/2 F/2
FL/4

3



Figure 18: Volume effect II: The volume of a two-span
test beam under peak tensile stress is even smaller than
that in a simple test beam.

6 CONCLUSIONS
The results indicate that the load-bearing behaviour of
the two-span beam may be influenced by many factors
such as possible plastification effects or the width of the
central support. The number of specimens tested was too
small for a statistical conclusion. Much more testing is
needed before the puzzle can be fully understood. What
is clear at present is that the bending strength for the
hogging moment over the intermediate support may be
much higher than that of the simple beam. Apart from
possible plastification effects, a plausible explanation
may also be the volume effect: a much smaller beam
volume is under peak stresses in a two-span beam that in
a simple beam.

ACKNOWLEDGEMENT
The first test series was part of a research project
financed by the Swiss Commission for Technical
Innovation. Two Wood Engineering students N.
Drachsel and L. Stiefel worked on the second test series
within the framework of a semester project. D. Roder
and S. Kraft, two staff members of the Bern University
of Applied Sciences, helped with both tests series.

REFERENCES
[1] Frese, M.: System Effects in Continous Glulam
Beams. in World Conference on Timber
Engineering (WCTE 2010).
[2] CEN/TC 124 "Timber Structures": SN EN
408:2010 Timber Structures - Structural timber and
glulam - SIA Swiss Association of architects and
engineers. 2010
[3] Lehmann, M., Brunner, M.: Ductile Adhesive
Layers for glulam. CTI-Project. 2689-SB-01. Bern
University of Applied Sciences. 2010
[4] Drachsel N., Stiefel L.: Load-bearing behaviour of
two-span beams in timber (Das Tragverhalten des
Zweifeldbalkens aus Holz), student project work,
Bern University of Applied Sciences. 2010
[5] Brunner M., Lehmann M., Kraft S., Fankhauser U.,
Richter K.: Super Glulam: a flexible adhesive layer
boosts the bending resistance of glulam, WCTE
2010, Italy.
[6] T. Valle, T. Tannert, M. Lehmann, M. Brunner, in:
Proceedings of the International Conference on
Fracture 12, Ottawa (Canada) (2009)



F/4
FL/12
FL/36
M
FL/36
F/6 2F/3 F/6

F/4 F/4 F/4


F/2
M
F/2
F/2 F/2
FL/4

FL/18 FL/18

You might also like