Fire behaviour of aluminium-wood joints with tolerance gaps

Abstract

Tolerance gaps or slips in wood connections are unavoidable, for reasons of constructability and the effects of natural shrinkage in timber elements with changing moisture content. During a fire, these gaps may lead to increased heat transfer through the connection. Aluminium connectors are becoming more popular due to their high malleability and availability, but they are particularly vulnerable to elevated temperatures. Thus, the objective of this study is to investigate the effect of tolerance gaps on the fire performance of aluminium connectors in beam-to-column/wall shear connections. An experimental campaign was designed to study the temperature evolution of the aluminium connectors during standard fire exposure for 1 mm and 6 mm tolerance gaps, as well the mitigation effects of additional intumescent fire protection in a 6 mm tolerance gap connection. The results showed a clear and consistent impact of the connection gap size on the temperature evolution of the aluminium connectors. For the larger 6 mm gap, the temperature of the connector increased much faster, reaching 286 ± 36 °C after 80 minutes, at which time the connector with a 1 mm gap had only reached 97 ± 1 °C. The addition of intumescent protection in a 6 mm gap case led to lower temperatures in the connection after 40 minutes of fire exposure, in comparison to an equivalent tolerance gap without fire protection. This study shows that tolerance gaps can lead to a significant reduction in the capacity of aluminium connectors, but this may be mitigated with additional fire protection.

Full text

SiF 2022– The 12th International Conference on Structures in Fire
The Hong Kong Polytechnic University, Nov 30 - Dec 2, 2022
FIRE BEHAVIOUR OF ALUMINIUM-WOOD JOINTS WITH TOLERANCE GAPS
Hugrún María Friðriksdóttir
1
, Finn Larsen
2
, Ian Pope
3
, Ana Sauca
4
, Alexandru Radulescu
5
, Frank Markert
6
ABSTRACT
Tolerance gaps or slips in wood connections are unavoidable, for reasons of constructability and the effects
of natural shrinkage in timber elements with changing moisture content. During a fire, these gaps may lead
to increased heat transfer through the connection. Aluminium connectors are becoming more popular due
to their high malleability and availability, but they are particularly vulnerable to elevated temperatures.
Thus, the objective of this study is to investigate the effect of tolerance gaps on the fire performance of
aluminium connectors in beam-to-column/wall shear connections. An experimental campaign was designed
to study the temperature evolution of the aluminium connectors during standard fire exposure for 1 mm and
6 mm tolerance gaps, as well the mitigation effects of additional intumescent fire protection in a 6 mm
tolerance gap connection. The results showed a clear and consistent impact of the connection gap size on
the temperature evolution of the aluminium connectors. For the larger 6 mm gap, the temperature of the
connector increased much faster, reaching 286 ± 36 °C after 80 minutes, at which time the connector with
a 1 mm gap had only reached 97 ± 1 °C. The addition of intumescent protection in a 6 mm gap case led to
lower temperatures in the connection after 40 minutes of fire exposure, in comparison to an equivalent
tolerance gap without fire protection. This study shows that tolerance gaps can lead to a significant
reduction in the capacity of aluminium connectors, but this may be mitigated with additional fire protection.
Keywords: engineered timber; connections; aluminium connectors; wood; tolerance gaps; intumescent;
fire resistance; fire protection
1 INTRODUCTION
Engineered timber construction materials are increasingly appealing due to their mechanical properties,
aesthetic qualities, and low embodied CO2 emissions in comparison with more conventional materials.
These materials, such as cross-laminated timber (CLT), laminated veneer lumber (LVL), and glulam,
consist of modified timber products that are manufactured to increase the strength and stiffness of the
elements in comparison to sawn timber. However, the fire performance of engineered timber structures
remains a concern. A key issue is the performance of metallic connections between exposed timber
elements, which can be particularly vulnerable to fire if not adequately protected [1, 2]. A common practice
1
MSc Civil Engineering graduate, Denmark’s Technical Univeristy
e-mail: hugrunmaria@gmail.com
2
Wood Specialist, DTI – Danish Technological Institute
e-mail: fla@teknologisk.dk, ORCID: https://orcid.org/0000-0001-6498-9012
3
Research Consultant, DBI – The Danish Institute of Fire and Security Technology
e-mail: ipo@dbigroup.dk, ORCID: https://orcid.org/0000-0002-3130-6868
4
Scientific Researcher, DBI – The Danish Institute of Fire and Security Technology
e-mail: as@dbigroup.dk, ORCID: https://orcid.org/0000-0001-5992-998X
5
Research Consultant, DBI – The Danish Institute of Fire and Security Technology
e-mail: ara@dbigroup.dk
6
Associate Professor, Denmark’s Technical University
e-mail: fram@dtu.dk, ORCID: https://orcid.org/0000-0002-1396-2810

for protecting metal connectors is to embed them between the adjoining engineered timber members, so
that they are insulated by the surrounding wood and – in case of a fire – by the development of a char layer.
The adjoining members shield the connector from direct exposure to radiative and convective heat transfer
from the fire, and the relatively lower thermal conductivity of the wood and char can insulate the connector
from the high external temperatures. This solution requires a sufficient cover thickness of the wooden
elements around the connector, and minimal gap width between the elements, such that convection and
radiation through the gap is negligible.
However, ensuring that the joint is completely tight is not feasible, since wood is a natural material that
may shrink or swell, and tolerance gaps are unavoidable for constructability. Gaps between members vary
in size depending on the type of connection and the extent of any shrinkage or swelling of the wood [3, 4].
The best feasible gap width for connectors that are recessed in holes carved into the wood is 1 mm, but
even with this design value, the in-use gap width could foreseeably increase to 6 mm – thus potentially
exposing the metal connectors to high temperature gases. In the production phase, often an initial gap of at
least 1 mm is made, so that the connection is not too tight and difficult to mount. Subsequently, shrinkage
of the wood materials due to changes in moisture content (MC), typically falling from 12-15 % MC down
to 6-8 % MC (approximately 8 % change) [5], may result in a longitudinal shrinkage of 4-6 mm for a beam
element with a length of 5-9 m, or approximately 2-3 mm at each end. This shrinkage is typically
incorporated as a slip possibility in the connections. Furthermore, when the connectors are recessed into a
wood-based column or wall, there may also be shrinkage in the depth of the recess around the connector
due to aforementioned moisture changes, leading to shrinkage perpendicular to the wood fibres of 1-2 mm.
Altogether, this could lead to a gap of 6 mm, even for a designed 1 mm gap.
While the fire performance of these joints has been studied for steel connectors [6, 7, 8], there is a lack of
information on the behaviour of aluminium connectors, despite them being widely available and attractive
due to their high malleability. For steel connectors, the predominant failure mode is a loss of embedding
strength in the timber surrounding the steel [6]. However, aluminium loses 77 % of its yield strength at
300 °C [9], whereas the yield strength of carbon steel is relatively unaffected at this temperature. Thus,
exposure to hot gases via tolerance gaps may lead to earlier failure of the aluminium connector itself.
Palma et al. [7, 8] showed that increasing the tolerance gap between the beam and the column has a negative
influence on the fire resistance of beam-to-column shear connections. This study presented the results of
an extensive experimental program on the fire resistance of timber shear connections, where a total of nine
different connection typologies were tested. Only two of the considered connections were commercially
available, while the others were custom-made. All were steel dowel connections except one, which was an
aluminium dovetail connection (AW-6082, EN 755-2 [10]), composed of two interlocking parts that were
separately screwed into each member. This aluminium connector was not recessed, so it was exposed
around its perimeter by an 18 mm gap between the column and the beam. The member was subjected to
30 % of its ambient shear capacity during the test and failed after 34 minutes of standard fire exposure. The
failure occurred in the connector itself, but no temperatures of the metal connectors were reported during
the experiment, which is an important parameter to monitor to understand how the fire performance of
aluminium-wood connections may be improved.
For timber elements of a given size, it is important to know the extent to which the size of the tolerance
gap, charring of the timber, and any additional protection, can affect the temperature evolution of the
aluminium connection. For joints with large tolerance gaps or exposed connections, the use of fire
protective insulation or intumescent materials is relatively straightforward and generally anticipated in
design. However, for small gaps of 1-6 mm, the need for fire protection of aluminium connectors is not
established, and solutions are limited by the constructability of the joint. The aim of this study is to
investigate the effect of tolerance gaps on the fire behaviour of aluminium connectors in a wood joint.
Small-scale experiments were performed in order to investigate the temperature evolution in such
connections, char depths, and the estimation of the shear capacity of the aluminium connectors when
exposed to elevated temperatures. Two intumescent fire protection measures are proposed, and fire tested,
and the results are discussed in this paper.

2 EXPERIMENTAL METHODS
The experimental campaign, presented in Table 1, consisted of four small-scale fire resistance tests. The
use of standard fire exposure for testing wooden elements may not be representative of the most severe fire
conditions in reality, particularly since the limited availability of oxygen in a standard furnace will inhibit
flaming combustion or oxidation of the char layer around the joint [11], and because there is no
consideration of cooling-phase heat transfer. Nevertheless, standard fire exposure can allow comparison of
the fire performance of different designs, and the convective heat transfer induced is particularly relevant
to the vulnerabilities created by tolerance gaps.
In all four tests, samples consisted of a 180 × 366 mm glulam beam with a length of 488-493 mm, connected
by an embedded aluminium connector to 500 × 450 mm LVL walls of 160 mm thickness. The connectors
used in this experiment were commercially available LOCK75215 aluminium connectors from Rothoblaas,
made from alloy AW6005A-T6 (EN 755-2 [10]). Three specimens (used for the Tests 1-3) had 1 mm and
6 mm tolerance gaps at each end, with no additional fire protection, as presented in Figure 1. In Test 4, a
specimen with additional fire insulation around the connector was used to investigate the effectiveness of
this protection. The specimen in Test 4 had 6 mm tolerance gaps on both ends, with a 1 mm thick
intumescent flexible gasket ‘fire stripe’ [12] surrounding the perimeter of the connector, as presented in
Figure 2. The fire stripe’s ‘reaction to fire’ is classified as B-s3, d0 in accordance with EN 13501-1 [13].
At both ends of the specimen, the width of the stripe was 20 mm, with a cover distance of 30 mm from all
edges of the beam. The only difference between the cases at each end of the specimen in Test 4 was that at
one end the fire stripe was inlaid into a channel carved 1 mm into the surface of the beam (Case I), whereas
at the other end, the fire stripe was applied directly to the unaltered surface of the beam (Case II).
No external load was applied to the specimen when exposed to ISO 834 [14] standard fire, but the
temperatures of the aluminium connectors were monitored and their reduced load carrying capacity was
estimated. As shown in equation (1), the reduced capacity of the aluminium connectors () was
estimated as the product of the shear capacity of the aluminium connector at ambient temperature
() and the reduction factor of the 0.2 % proof strength at elevated temperatures for
aluminium alloy AW6005A () as defined in EN 1999-1-2 [9] .
     (1)
Additionally, in Test 4, the in-depth temperature of the glulam beam was measured in ten locations at either
end of the beam, at a depth of 140 mm from the top of the beam (see Figure 5).
Table 1. Small-scale experimental campaign.
Tests 1-3:
Test 4:
Gap size
Fire protection
Gap size
Fire protection
1 mm
None
6 mm
Case I: fire stripe carved 1 mm
into the surface of the beam
6 mm
None
6 mm
Case II: fire stripe glued to the
surface of the beam

Figure 1. Cross-sections at the joint between the LVL wall and glulam beam (left), and down the length of the specimen
(right), for Tests 1, 2, and 3.
Figure 2. Cross-sections at the joint between the LVL wall and glulam beam (left), and down the length of the specimen
(right), for Test 4.
2.1 Testing equipment
The DBI small-scale electrical furnace with opening dimensions of 500 mm × 500 mm, made from a steel
frame with ceramic fibre insulation and concrete interface around the opening was used to provide the
ISO 834 temperature-time exposure to the specimens. Nine electrical heating rods drive the furnace, with
a power of 3×400V/32A at 50 Hz, controlled by three metal-sheathed gas phase thermocouples in the
middle of the furnace. Compressed air was supplied to the furnace as needed throughout the experiment to
keep the pressure at 20 Pa, controlled with a venturi valve.
The oxygen concentration within the furnace was measured in Test 2, using a 4900 CEMS Analyzer from
Servomex [15]. The emission analyser was calibrated based on the ambient oxygen concentration in the
laboratory. A steel rod pipe connected to a rubber tube extracted the gases from inside the furnace to the
emission analyser.
Agilent data acquisition apparatus was connected to the thermocouples and the data acquisition took place
within the Agilent Benchlink software [16]. The temperature time histories were recorded with a sampling
interval of 5 seconds.

2.2 Instrumentation of the specimens
The temperatures in the connection area were measured with thermocouples (TCs) at the locations shown
in Figure 3 for Tests 1-3, and in Figure 4 for Test 4. These thermocouples were fixed in the gap between
the beam and the wall on the left (L), right (R), bottom (B) and top (T) sides of the connector, and also
between the aluminium connector and the wall, along the mid-line of the connector (M). During Test 4,
additional in-depth TCs were added to the beam, as shown in Figure 5, inserted 140 mm in from the top of
the beam at different positions around the screws and the fire stripes. These thermocouples were of type K,
with an exposed junction formed by twisted wires of 0.5 mm diameter each, while the rest of the cable was
insulated with fibreglass. Three Inconel sheathed type K thermocouples of 2 mm diameter were placed
10 mm below the beam, to measure the gas temperatures near the surface of the beam at each end, and at
the centre of the beam.
Figure 3. Location of TCs in the
connection area - Tests 1, 2 and 3.
Figure 4. Locations of TCs in the
connection area - Test 4.
Figure 5. Locations of TCs in the beam -
Test 4 (top view).
2.3 Specimen preparation
A cavity of 75 mm × 240 mm, with a depth of either 22 mm or 16 mm, was carved into each LVL wall, at
the centre, 144 mm above the bottom edge. The depth of the cavity was 22 mm for the walls where the
aluminium connector was to be flush with the surface of the LVL wall specimen. Considering that the
connectors had a slip tolerance of 1 mm, this allowed a gap between the wooden elements of 1 mm. For the
cases that the connector was to protrude 6 mm out of the LVL wall specimen, the cavity had a depth of
16 mm. In Test 4, the fire stripe was attached to the beam around the connector (see Figure 6 a)).
Pre-drilled holes with a diameter of 5 mm were made in the LVL wall element to allow the insertion of
thermocouples from the exterior of the sample. Each thermocouple was then bent 50 mm along the wall
surface so that the measuring junction was at the desired location, as shown in Figure 6 b). The pre-drilled
holes were sealed at the exterior of the wall specimen with a high-temperature resistant silicate-sealant, to
prevent the smoke inside the furnace from passing through the thermocouple holes. For Test 4, an additional
ten holes of 2 mm diameter were drilled in the beam, and TCs were inserted as shown in Figure 6 c).
A 50 mm thick mineral wool insulation was fastened on top of the specimen and 25 mm thick mineral wool
insulation was placed on both sides of the specimen, to create an enclosed space. A 51 mm diameter hole
was made in the top insulation to allow the exhaust of smoke produced during the test.

Figure 6. a) fire stripe installed around connector on beam in Test 4, b) aluminium connector installed in LVL wall, along with
TCs, c) in-depth TCs installed in the top of the beam in Test 4.
3 RESULTS
3.1 Tests 1-3 with unprotected tolerance gaps
Figure 7 a) shows the averaged temperature measurements of selected thermocouples during Tests 1-3.
Maximum and minimum temperatures are also shown, bounding the average values for the thermocouples.
Gas temperatures from near the centre and each end of the beam were averaged across the three tests. There
was no clear difference between the gas temperatures in each location, and the average was close to the
standard value for ISO 834 exposure for most of the test. An initial spike in gas temperatures after two
minutes corresponded to ignition of the timber sample, after which the oxygen concentration in the furnace
rapidly decreased to zero as an excess of pyrolysis gases was produced by the specimen. In Test 2, the
oxygen concentration measured within the furnace had fallen below 3 % within five minutes, and continued
to fall after this time. The lack of available oxygen within the fuel-rich atmosphere in the furnace prevents
oxidisation of the char layer and flaming combustion, which might otherwise make the influence of the gap
on the connection performance even more severe.
The temperatures at the outer edges of the gaps (averages of thermocouples L2, R2, and B3) were similar
at either end, which further supports the assumption of equivalent external heating conditions surrounding
each connection. Therefore, the difference in temperatures at the connectors can only be explained by the
enhanced heat transfer path provided by the larger air gap, increasing the flow of hot gases around the joint.
Figure 7 b) shows the averaged temperatures of the aluminium connectors (M1, M2, and M3) measured in
all three tests, along with the reduced capacity of the connector  during the ISO 834 exposure.
For the larger 6 mm gap, the temperature of the connector increased much faster, exceeding 150 °C after
60 minutes, and reaching up to 300 °C by around 80 minutes of exposure in all tests – at which time the
connector with a 1 mm gap had only reached 100 °C. At 60 minutes, these temperatures corresponded to
reductions in the shear capacity of the connectors of 6 % and 22 % for the connections with 1 mm and
6 mm gaps, respectively. At 75 minutes, the temperatures corresponded to reductions of 7 % and 57 % for
the connections with 1 mm and 6 mm gaps, respectively. As temperatures behind the connector rise above
150 °C after 60 minutes for the connections with a 6 mm gap, this indicates that the surface of the wood
behind the connector has begun to pyrolyse.
a)
b)
c)

Figure 7. a) averaged temperatures from Tests 1-3 for thermocouples L2, R2, and B3, around the perimeter of the gap at the
connection; thermocouples M1, M2, and M3, behind the aluminium connector; and the gas phase thermocouples beneath the
beam. b) averaged temperatures behind the connector, and corresponding reduction in shear capacity for each gap width.
Darker lines indicate average temperatures, while faded lines bound the maximum and minimum values.
Figure 8 shows the specimen, the connections, and the screws at the end of Test 2, immediately after the
specimen has been removed from the furnace. When the test is finished, the specimen continues burning
during its removal from the mobile furnace until it is extinguished with water. Therefore, the pictures show
the state of the specimen after the water is applied, and some smouldering can still be observed in cracks.
No charring occurred directly behind the aluminium connector for the connections with a tolerance gap of
1 mm, while charring was observed for the 6 mm gap connections. Furthermore, the tolerance gap had
increased from 6 mm to approximately 12 mm at the top of the connection, near the connector, due to
shrinkage of the char on either side. Discolouration of the connector and screws at the connections with a
6 mm gap indicates exposure to high temperature pyrolysis gases, or contact with the char and tar from the
timber. This discoloration was not observed for the 1 mm gap. The discoloured part of the screws for the
6 mm gap was measured, with the greatest length of discoloration being 38 mm and the smallest 24 mm.
The discoloured distance is also an indication of the char depth behind the connector.
a)
b)
c)
d)
Figure 8. Test 2 specimen after the test: a) top view of the specimen; b) 1 mm gap connection; c) 6 mm gap connection;
d) 1 mm and 6 mm gap connections and the adjacent screws.
0
200
400
600
800
1000
020 40 60 80
Temperature (°C)
Time (min)
ISO 834 Gas - Avg (Test 1-3)
L2,R2,B3 - 1 mm L2,R2,B3 - 6 mm
M1,M2,M3 - 1 mm M1,M2,M3 - 6 mm
0
10
20
30
40
50
60
70
0
50
100
150
200
250
300
350
020 40 60 80
Shear Capacity of Connector (kN)
Temperature (°C)
Time (min)
M1,M2,M3 - 1 mm M1,M2,M3 - 6 mm
Rv.alu.k.fire - 1 mm Rv.alu.k.fire - 6 mm
1 mm gap
6 mm gap
1 mm gap
6 mm gap
a)
b)

3.2 Test 4 with additional fire protection of the gaps
Figure 9 a) shows the average temperatures of selected thermocouples from Tests 1-3, compared to those
from Test 4. The average gas temperatures measured in Test 4 were slightly higher than those measured for
the 6 mm gap in Test 1-3, which could partly explain the higher average temperatures of thermocouples
L2, R2, and B3, around the perimeter of the gaps at the connections in Test 4, compared to Test 1-3. This
could also be due to activation of the intumescent fire stripe, creating an insulated boundary that reduces
heat losses from the perimeter of the gap into the connector. Nonetheless, the eventual temperature rise of
the connectors with an unprotected 6 mm gap (Tests 1-3) is less than that for the protected gaps in Test 4.
Figure 9 b) shows the average temperatures of the aluminium connectors measured in all four tests, along
with the reduced capacity of the connector during the ISO 834 exposure. In the time interval 0-40 minutes,
it can be observed that the average temperatures of the aluminium connectors for the 6 mm tolerance gaps
with fire stripes in Test 4 followed the same trajectory as the corresponding temperatures in Tests 1-3 for a
6 mm tolerance gap without a fire stripe. These temperatures begin to diverge significantly after 40 minutes,
with the unprotected connector being hotter by approximately 75 °C after 60 minutes, and more than 100 °C
by 80 minutes of exposure.
In Test 4, there was no clear difference between the average connector temperatures in the cases with the
fire stripe applied directly to the surface of the beam or inset into a channel around the connector. This
indicates that both methods of application are similarly effective in reducing the heat transfer to the
connector, although additional tests are needed to verify this. In comparison with the unprotected 6 mm
gap, the fire stripes did mitigate the temperature rise in the connector, but these temperatures were still
significantly higher than for the unprotected 1 mm gap. Even so, the added protection limited the reduction
in shear capacity of the connector to approximately 12 % after 60 minutes and 22 % after 75 minutes.
Figure 9. a) averaged temperatures from Tests 1-3 compared with Test 4, for thermocouples L2, R2, and B3, around the
perimeter of the gap at the connection; thermocouples M1, M2, and M3, behind the aluminium connector; and the gas phase
thermocouples beneath the beam. b) averaged temperatures behind the connector, and corresponding reduction in shear
capacity for each unprotected gap width (Tests 1-3) or protection with a fire stripe inset into a carved groove (FS in) or applied
directly onto the surface of the beam (FS out).
0
200
400
600
800
1000
020 40 60 80
Temperature (°C)
Time (min)
ISO 834 Gas - Avg (Test 1-3)
Gas - Avg (Test 4) L2,R2,B3 - 6 mm
L2,R2,B3 - 6 mm - FS in L2,R2,B3 - 6 mm - FS out
M1,M2,M3 - 6 mm M1,M2,M3 - 6 mm - FS in
M1,M2,M3 - 6 mm - FS out
0
10
20
30
40
50
60
70
0
50
100
150
200
250
300
350
020 40 60 80
Shear Capacity of Connector (kN)
Temperature (°C)
Time (min)
M1,M2,M3 - 1 mm M1,M2,M3 - 6 mm
M1,M2,M3 - 6 mm - FS in M1,M2,M3 - 6 mm - FS out
Rv.alu.k.fire - 1 mm Rv.alu.k.fire - 6 mm
Rv.alu.k.fire - 6 mm - FS in Rv.alu.k.fire - 6 mm - FS out
a)
b)

Figure 10 shows the temperatures on the inner and outer sides of the fire stripe for Test 4, and at the
corresponding locations for the unprotected joints in Tests 1-3. The results for each location have been
averaged for Tests 1-3, and presented individually for Test 4. No consistent differences were observed
between temperatures measured on the left or right side of the connector. Temperatures on the inner side
of the fire stripe, at the left and right edges of the connector (L1/R1), were slightly higher for the case with
a fire stripe inset into a carved channel than for the case where the fire stripe was directly applied to the
beam. Temperatures on the outer side of the fire stripe (L2/R2) were very similar in either of these cases,
and slightly higher than the corresponding temperatures for tolerance gaps without protection. As
mentioned previously, this may be due to the insulating effect of the intumescent fire stripe reducing heat
losses into the connector. However, additional tests with fire stripes are required to verify that these are
consistent results, rather than artifacts of experimental variability.
After 10-15 minutes, the temperatures measured by L1 and R1 on the inner sides of the fire stripes started
to decrease from an initial peak of 200-250 °C. The temperature decrease could be explained by swelling
of the fire stripe closing the gaps around the connector, limiting further convective heat transfer. In contrast,
the corresponding temperatures for the unprotected 6 mm gap (Figure 10 a)) continue to increase steadily
after this time, and end up more than 100 °C hotter after 60 minutes. Even with the protective effects of the
fire stripes, the temperatures at the edge of the connector for the unprotected 1 mm gap (Figure 10 b)) are
significantly lower.
Figure 10: Temperatures on the inner sides of the fire stripe – at the left and right edges of the connector (L1/R1) – and on the
outer sides of the fire stripe (L2/R2), for a fire stripe inset into a carved channel (FS in) or applied directly onto the surface of
the beam (FS out), compared with corresponding temperatures for a) an unprotected 6 mm gap, and b) an unprotected 1 mm gap.
The in-depth temperatures measured in the beam in Test 4 are presented in Figure 11. Due to potential
errors in the drilling angle over the 140 mm depth of the holes, there is some uncertainty in the actual
location of the thermocouple tip within the beam. While care was taken to reduce these errors, these in-
depth temperatures should be considered as indicative, rather than exact. The temperatures measured closest
to the screws, by thermocouples S-L1, S-R1, and FS-R2, remained below 100 °C up until 50 minutes of
exposure for both Case I and Case II, and did not exceed 300 °C up to 80 minutes. This suggests that there
was no charring of the timber during this period at a depth of approximately 55 mm from the surface of the
beam. The temperatures in the beam at a distance of 10 mm behind the fire stripe and 40 mm from the outer
0
200
400
600
800
1000
020 40 60 80
Temperature (°C)
Time (min)
Gas - Avg (Test 1-3) Gas - Avg (Test 4)
L1 - 6 mm R1 - 6 mm
L1 - 6 mm - FS in R1 - 6 mm - FS in
L1 - 6 mm - FS out R1 - 6 mm - FS out
L2 - 6 mm R2 - 6 mm
L2 - 6 mm - FS in R2 - 6 mm - FS in
L2 - 6 mm - FS out R2 - 6 mm - FS out
0
200
400
600
800
1000
020 40 60 80
Temperature (°C)
Time (min)
Gas - Avg (Test 1-3) Gas - Avg (Test 4)
L1 - 1 mm R1 - 1 mm
L1 - 6 mm - FS in R1 - 6 mm - FS in
L1 - 6 mm - FS out R1 - 6 mm - FS out
L2 - 1 mm R2 - 1 mm
L2 - 6 mm - FS in R2 - 6 mm - FS in
L2 - 6 mm - FS out R2 - 6 mm - FS out
a)
b)

surface, i.e., FS-R1 and FS-L1, increase more rapidly, reaching 100-200 °C by 20 minutes and 300 °C by
the end of the test for both cases. Considering that the fire stripe has a nominal activation temperature of
150 °C, this supports the conclusion that the fire stripes would have activated within the first 20 minutes,
but that the wood behind the fire stripe was charring by the end of the test.
Figure 11. Temperatures measured within the beam in Test 4, at distances of 40 or 55 mm from the side of the beam for a)
Case I, with the fire stripe inset into a carved groove (FS in) or b) Case II, with the fire stripe applied directly to the unaltered
surface of the beam (FS out).
Figure 12 shows the connections and screws after the fire exposure of Test 4. These images confirm that
the wood directly behind the connector and surrounding the screws was uncharred, but that the char layer
had just reached the boundary of these areas by the end of the test.
a)
b)
c)
Figure 12. Test 4 specimen after the test: a) Fire stripe carved into the surface of the beam, b) Fire stripe on the surface of the
beam; c) section through the beam, 105 mm from the bottom edge of the original specimen (Case 2).
4 DISCUSSION
In all four tests, aluminium connectors with timber cover distances of 60 mm from the bottom and 53 mm
from the sides were exposed to ISO 834 standard fire temperatures for more than 60 minutes. The
corresponding reduced capacity of the aluminium connectors during fire was calculated using the measured
0
100
200
300
400
500
600
700
020 40 60 80
Temperature (°C)
Time (min)
S-R1 (55 mm) - FS in S-L1 (55 mm) - FS in
FS-R1 (40 mm) - FS in FS-R2 (55 mm) - FS in
FS-L1 (40 mm) - FS in
0
100
200
300
400
500
600
700
020 40 60 80
Temperature (°C)
Time (min)
S-R1 (55 mm) - FS out S-L1 (55 mm) - FS out
FS-R1 (40 mm) - FS out FS-R2 (55 mm) - FS out
FS-L1 (40 mm) - FS out
a)
b)

temperatures of the aluminium and the reduction factor of the 0.2 % proof strength at elevated temperatures
for aluminium alloy AW6005A () as defined in EN 1999-1-2 [9]. Temperatures measured in the gas
phase beneath the beam and around the perimeter of the connections indicate that the thermal exposure
conditions were comparable in each case.
The results of these tests showed a clear and consistent impact of the connection gap size on the temperature
evolution of the aluminium connectors. For the larger 6 mm gap, the temperature of the connector increased
much faster, and reached 286 ± 36 °C after 80 minutes of exposure in all tests, at which time the connector
with a 1 mm gap had only reached 97 ± 1 °C as shown in Figure 7 b). Furthermore, the temperatures of the
aluminium connectors for the larger 6 mm gap were on average 35-40 °C higher than for the 1 mm gap
until 45 minutes of fire exposure. Further evidence of the impact of the gap size was provided by the
observation of significant charring behind the connector with the unprotected 6 mm gap, which was not
seen for the 1 mm gap.
When additional intumescent fire protection stripes were used in the 6 mm gap, the temperature evolution
of the aluminium connectors followed the same trajectory as the for the unprotected 6 mm gap during the
first 40 minutes of exposure, as shown in Figure 9 b). The insulating effect of the fire stripes became
apparent after 40 minutes, as the rate of temperature increase for the unprotected 6 mm gap became much
higher. By 77 minutes, the temperatures of the protected connectors by the fire stripes were more than
100 °C lower than for the unprotected connectors. In terms of the shear capacity of the connectors, the
elevated temperatures at 60 minutes induced an average reduction in capacity of 22 % for the unprotected
6 mm gap, 12 % for the 6 mm gap with a fire stripe, and only 6 % for the 1 mm gap. By 75 minutes, these
reductions had increased to 57 %, 22 %, and 7 %, respectively. This shows the significance of the gap width
for the fire performance of the connection, and the value of the additional fire protection in cases where a
gap of more than 1 mm is unavoidable, particularly for exposure durations beyond 60 minutes. This might
be even more critical for exposures with higher oxygen availability, in which additional insulation may
provide further protection against flaming or oxidation of the char around the connection.
The fire stripe also prevented the wood around the aluminium connectors from charring, by reducing the
heat transfer to the connector and the surrounding wood. This can be observed in Figure 12, in comparison
to a connection with no additional fire protection in the 6 mm gap, as shown in Figure 8 c). The in-depth
measurements in the beam during Test 4, as presented in Figure 11, show that temperatures in the wood
closest to the screws did not reach 300 °C during the test, but they did exceed 100 °C after 50 minutes had
passed. Therefore, the base of the char layer had not reached the wood around the fasteners within these
exposure periods, but the embedding strength of this wood is likely to have decreased significantly at such
elevated temperatures.
No significant difference was observed in the effectiveness of the two fire stripe application methods in
Test 4, but further tests are planned to confirm this with greater certainty. Nevertheless, other considerations
such as ease of construction, or durability, may be more important in determining whether the fire stripe
should be applied directly to the unaltered surface of the beam or inset into a carved channel. Further tests
are also planned to investigate the critical failure mode of the connections (e.g., embedment failure or
aluminium connector failure) in fire, due to the effects of tolerance gaps. These tests will include external
loading under exposure to elevated temperatures, and will be performed at a larger scale.
5 CONCLUSIONS
The objective of this study was to investigate the effects of tolerance gaps on aluminium connectors in a
wood joint during fire exposure. Tolerance gaps are unavoidable in connections between timber elements,
due to constructability and natural shrinkage or swelling of the wood over time, and this can allow hot gases
to travel around the connector during a fire. This convective heat transfer can be critical for aluminium
connectors, since the reduction in strength of the connector at elevated temperatures occurs much earlier
for aluminium than for steel, which may result in premature failure of the connector during a fire. Tolerance
gaps of 1 mm and 6 mm were chosen as they are representative of the minimum gap size range that is
currently feasible. The results of this study showed a clear and consistent impact of the connection gap size

on the temperature evolution of the aluminium connectors. In all cases, the larger gap size resulted in faster
temperature rise of the connector, corresponding to a more severe reduction in capacity.
The addition of an intumescent fire stripe around the connector with a 6 mm tolerance gap improved the
performance of the connector after 40 minutes of fire exposure, in comparison with an unprotected gap of
equivalent size. There was no clear difference between the effectiveness of fire stripes applied directly to
the unaltered surface of the wood or inset into a carved channel around the connector.
To further investigate the critical failure mode of the connections in fire, loaded tests are necessary to
conclude whether an embedment failure will occur prior to failure of the aluminium connector. For this
purpose, a large-scale test is planned, in which the same connection configuration is subjected to a constant
load during fire exposure until failure.
ACKNOWLEDGMENTS
The authors would like to thank Rothoblaas for supplying the connectors and fire-stripes, and Stora Enso,
for supplying the wood materials. The authors are also grateful to Lennart Schou Jensen from DBI for his
assistance with the tests.
REFERENCES
1. Barber D (2017) Determination of fire resistance ratings for glulam connectors within US high rise timber
buildings. Fire Saf J 91:. https://doi.org/10.1016/j.firesaf.2017.04.028
2. Barber D, Gerard R (2015) Summary of the fire protection foundation report - fire safety challenges of tall
wood buildings. Fire Sci Rev 4:5. https://doi.org/10.1186/s40038-015-0009-3
3. McKenzie WMC, Zhang B (2007) Design of Structural Timber: To Eurocode 5. Palgrave Macmillan
4. Buchanan A, Östman B (2022) Fire Safe Use of Wood in Buildings: Global Design Guide, 1st ed. CRC
Press, Boca Raton
5. Rijsdijk JF, Laming PB (1994) Physical and related properties of 145 timbers: information for practice.
Kluwer Academic Publishers, Dordrecht ; Boston
6. Maraveas C, Miamis K, Matthaiou ChE (2015) Performance of Timber Connections Exposed to Fire: A
Review. Fire Technol 51:1401–1432. https://doi.org/10.1007/s10694-013-0369-y
7. Palma P, Frangi A, Hugi E, et al (2016) Fire resistance tests on timber beam-to-column shear connections. J
Struct Fire Eng 7:41–57. https://doi.org/10.1108/JSFE-03-2016-004
8. Palma P, Frangi A, Hugi E, et al (2014) Fire resistance tests on beam-to-column shear connections
9. EN 1999-1-2: 2007 Design of aluminium structures - Part1-2: Structural fire design (The European Union
Per Regulation 305/2011, Directive98/34/EC, Directive 2004/18/EC. EN. Brussels: CEN, 2008.
10. EN 755-2:2016: Aluminium and aluminium alloys - Extruded rod/bar, tube and profiles - Part 2: Mechanical
properties EN 755-2:2016: Aluminium and aluminium alloys - Extruded rod/bar, tube and profiles - Part 2:
Mechanical properties.
11. Lange D, Sjöström J, Schmid J, et al (2020) A Comparison of the Conditions in a Fire Resistance Furnace
When Testing Combustible and Non-combustible Construction. Fire Technol 56:1621–1654.
https://doi.org/10.1007/s10694-020-00946-6
12. Rothoblaas Product brochure. Intumescent Thermo-Inflatable Flexible Gasket: Fire stripe. Rothoblaas.
Retrieved October 18, 2022, from https://www.rothoblaas.com/products/airtightness-and-
waterproofing/tapes-and-sealants/tapes-and-profiles/fire-stripe#documents
13. EN 13501-1: 2018 Fire classification of construction products and building elements - Part 1: Classification
using test data from reaction to fire tests.
14. ISO 834. “Fire-resistance tests - elements of building construction”. International Standard ISO, 834 (1975).
15. SERVOPRO 4900 Multigas Product brochure. Servomex. (n.d.). Retrieved October 18, 2022, from
https://www.servomex.com/resources/product/4900-multigas/
16. Agilent 34970A/34972A Data Acquisition / Switch Unit. User’s guide. Agilent Technologies, Inc. 2009-
2012