Fire propagation over combustible exterior facades exposed to intensified flame in Japan

With regard to fire safety for exterior walls of a building, fire-resistance performance is considered, according to the current Building Standard Law of Japan. And it was revealed that the fire safety is not specifically regulated from the viewpoint of reactionto-fire performance, such as fire propagation caused by combustible materials or products installed on the exterior side of fire-resistant load-bearing walls. Actual fire incidents in the world have shown that massive façade fire could occur at the exterior side of building wall even when the wall itself is fire resistant. In previous studies of the authors, a test method of façade fire was proposed for evaluating the vertical fire propagation over an external wall within the same building [1, 2]. Based on these studies, new domestic standard test method was established in Japan as JIS A 1310: 2015, “Test method for fire propagation over building façades” at the end of January 2015 [3]. But there was the argument that heat output of burner inside the combustion chamber was not sufficiently high in the previous study. In this paper, results of fire tests on combustible façades are discussed from the viewpoints of different strength of flame exposing facade. In this research, it was clearly found that JIS A 1310 with heat output of 900kW could be applicable for evaluating fire propagation behaviour over various types of combustible exterior façades.


INTRODUCTION
According to the current Building Standard Law of Japan, it is not restricted to install combustible materials over their exterior walls as long as fireproof buildings retain their required fire resistance performances.However, installation of combustible material over the exterior wall could cause largescale fire propagation.In previous studies of authors, façade test method for evaluation of fire propagation over combustible cladding was proposed, using various types of combustible façade specimens [1,2].Heat output of gas burner in a combustion chamber was set 600 kW, because temperature in the chamber was about 1000 • C and temperature at lintel was about 900 • C, which are near at real fire condition.a e-mail: yoshioka-h92te@nilim.go.jpThis is an Open Access article distributed under the terms of the Creative Commons Attribution License 2.0, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.However, some severe fire incidents occurred at high rise building covered with aluminium composite panel in Korea and Dubai, and a fire propagation also occurred at a building coated with resin paint in Japan even thought it was a small fire in scale.Compared with these combustible façade fires in actual cases, flame propagation at fire test with some combustible façade in previous study were smaller.It was considered that heat output in the previous study could not verify fire propagation over combustible façade properly.
The purpose of this paper is to investigate properties of fire propagation over various types of combustible façade specimens which could be applied in actual buildings with higher heat output from combustion chamber, compared with the previous study and to verify the proper heat intensity on JIS A 1310 as the test method for evaluating properties of fire propagation over combustible exterior façades as the result of the investigation.

TEST METHOD
Tests were performed according to JIS A 1310, which is vertical and façade-shaped simulating ejected flame from opening after flash over.Test facility of JIS A 1310 is shown in Fig. 1.As for test procedures, a propane burner is ignited when the test is initiated, and continues to supply a certain amount of propane gas constantly for 20 minutes by using a flow meter.For setting flow rate of propane gas, the heat output of propane burner was controlled by the oxygen consumption method in advance.A video camera was located diagonally opposite to the specimen's surface to visually observe fire propagation behaviour over combustible façade surface.
Unlike ISO 13785-2 [4], specimen's size is smaller (from lintel to specimen's top is about 2.7m) and opening size is not rectangular but square (about 0.9m square).And specimen's shape is flat surface without wing-wall unlike ISO 13785-2.

FLAME PROPERTIES WITH NONCOMBUSTIBLE FACADE
As a calibration test, a fire test was performed with a non-combustible façade which surface was covered with 25 mm thick ceramic fibre blanket.Heat flux meters were installed at HF2, HF3 and HF4, and thermocouples were installed at the position indicated in Fig. 1 (TC1 to TC5).
The author conducted experiments according to JIS A 1310 with a heat output of 600 kW in the previous study.In this research, fire tests were conducted with a higher heat output than the previous study (about 900 kW).
Figure 2 shows appearance of venting flame in stable condition at a calibration test.It seems that flame in the case of 900 kW was higher, thicker and along the façade surface longer than flame in the case of 600 kW. Figure 3 shows comparison between heat output of 600 kW and 900 kW about average façade surface temperatures and façade surface incident heat fluxes in stable condition.At 0.5 m and 0.9 m upward from the lintel, temperature and incident heat flux on façade surface with 900 kW are much higher than that of 600 kW, especially at 0.9 m upward from the lintel, incident heat flux on façade surface was about twice as much of 600 kW.The difference between temperatures and incident heat fluxes on façade surface of 600 kW and 900 kW is relatively smaller at 1.5 m or higher distance from the lintel.

SPECIMENS
Table 1 summarizes overall information on façade specimens.All the specimens are installed upon substrate, calcium silicate boards.Attached "-a" at end of specimen's number were conducted experiments at heat output of 600 kW, while attached "-b" at end of specimen's number were at 900 kW.
As for an exterior thermal insulation system, which is composed of EPS boards, glass fibre mesh, adhesive (mortar based) and finish coat (cement-based), the authors conducted fire tests for two different thicknesses of EPS (100 mm and 200 mm).As for wooden façade, which have recently been used for decorative purposes at exterior façade, two types of specimens were used.The difference is mainly whether applied chemical treatment to enhance fire-retardant performance or not.Specimens Nos. 5 and 6 are thin aluminium composite panels.Both surfaces are covered by aluminium facings.The core material of No. 5 is polyethylene, while that of No. 6 is composed of 30% weight polyethylene and 70% weight aluminium hydroxide (Al(OH) 3 ) for fire-retardant purpose.The composite panel of No. 6 holds various fire-safety certifications around the world.No. 7 is coated with water-resistant exterior finish paint, coated over the substrate.Only specimen Nos. 5 and 6 have a vertical joint at the center line of the façade, in order to make a weak point which is possible even in an actual building.

FIRE PROPAGATION BEHAVIOUR IN COMBUSTIBLE FAÇADE
Observation of fire propagation behaviour is an essential element to understand fire performance of combustible façade.As follows, we show fire propagation behaviours in each specimen.Figure 4 shows fire propagation behaviours in specimens of exterior thermal insulation system with 100 mm thickness of EPS.Although fire propagation at No. 1-a was slightly visible from 4 to 5 min after the start of the test, fire propagation at No. 1-b was easily visible from 1 to 5 min.
Figure 5 shows fire propagation behaviours in specimens with exterior thermal insulation system with 200 mm thickness of EPS.Different from the case of 100 mm thick EPS, fire propagations were observed at both specimens (No.

No.2-a 600kW
No.2-b 900kW  Figure 7 shows fire propagation behaviours in specimens of wooden façade with chemical fireretardant treatment.In spite of different flaming properties, remarkable fire propagations were not observed at both specimens (No. 4-a and No. 4-b).
Figure 8 shows fire propagation behaviours in specimens of aluminium composite panel with polyethylene core.At No. 5-a, fire propagation occurred 10 min after the start of the test and only the vertical joint of panels located on the centre line of the specimen were seriously damaged.On the other hand, at No. 5-b, fire propagation occurred 3 min after the start of the test and many parts of the specimen were burned out 8 min after the start of the test.Figure 9 shows fire propagation behaviours in specimens of aluminium composite panel with fireretardant polyethylene core.Severe fire propagation did not occur with No. 6-a.On the other hand, with No. 6-b, although fire propagation occurred to some extent at 5 min after the start of the test, severe fire propagation, such as observed at No. 5-b, did not occur.
Figure 10 shows fire propagation behaviours of specimens coated with resinous paint.At No. 7-a, slight fire propagation occurred 3.5 min after the start of the test.On the other hand, at No. 7-b, fire propagated to the top of the specimens immediately and larger fire propagation occurred, but flames mostly disappeared just 5 min after the start of the test.
Compared with the experiments conducted by heat output of 600 kW, clearly different and more severe fire propagation behaviours were observed in some specimens, such as wooden façade without 02002-p.6fire-retardant treatment, aluminium composite panel with polyethylene core and rein paint.On the other hand, no severe fire propagation occurred even though intensified flame in specimens with fire-retardant treatment such as No. 4 and No. 6.It is concluded that fire propagation could occur in the fire test at heat output of 900 kW, while no severe flame propagation observed at the fire test with heat output of 600 kW.from the lintel (chosen among TC1 to TC5).Flame height was 0 m if the peak temperature at TC1 didn't exceed 500 • C. With respect to flame height, flame heights with 900 kW were obviously higher because original flame height with no combustible façade was higher.For the distance from the lintel to specimen's top was about 2.7 m, maximum flame height were 2.5 m even though flame was likely to spread over the specimen's top, especially in the case of 900 kW.Difference between 600 kW and 900 kW about flame height was greater at specimens with resin paint.

COMOPARISION OF 600 KW AND 900 KW
With respect to heat flux at specimen's top (HF6), most of the specimens with heat output of 600 kW had about 10 kW/m 2 expect for aluminium composite panel with polyethylene core and wooden façade with no fire-retardant treatment.On the other hand, specimens with heat output of 900 kW had different values at HF6 in each specimen.Especially, at specimens of aluminium composite panel and resin paint, at which larger fire propagation were observed, relatively high heat fluxes were measured.Heat fluxes at specimen's top with heat output of 900 kW showed clear difference from specimens exposed of 600 kW.

CONCLUSION
As a result of conducting JIS A 1310 test with intensified flame, it becomes clearer to distinguish between combustible exterior façades with higher performance and those which propagates fire easily compared with the previous study.
Indeed JIS A 1310 with heat output of 600 kW could classify combustible exterior façades in which severe fire propagation occurred (specimens such as aluminium composite panel with polyethylene core, resin paint and wood with no treatment for fire retardancy) and those in which severe fire propagation didn't occur though ignition and combustion occurred by flame from the opening (specimens such as aluminium composite panel with fire-retardant polyethylene core and wood with treatment for fire retardancy), but there were combustible façade like rein paint at which flame propagation didn't occur with heat output of 600 kW.JIS A 1310 with heat output of 900 kW could make differences more remarkably in terms of flame propagation behaviour and heat flux measured at the specimen's top.In this research, it was clearly found that JIS A 1310 with heat output of 900 kW could relatively make fire performance clear of various types of combustible exterior façades.

Figure 1 .
Figure 1.Cross section and elevation of test facility of JIS A 1310 including positons of thermocouples and heat flux meters.

Figure 2 .
Figure 2. Photograph of flame venting through the opening at a calibration test.

Figure 3 .
Figure 3. Calibration test result about relationship between distance above the opening and façade surface temperatures and façade surface incident heat fluxes.

No
2-a and No. 2-b).Although fire propagation at No. 2-b was observed from 1.5 min after the start of the test, fire propagation at No. 2-a was not visible until 5 min.02002-p.4 2 nd International Seminar for Fire Safety of Facades, Lund (Sweden), 2016

Figure 6
Figure 6 shows fire propagation behaviours in specimens of wooden façade with no fire-retardant treatment.Although fire propagation at No. 3-a was occurred from 13.5 min after the start of the test, fire propagation at No. 3-b easily occurred from 1 min and continued until the end of the experiment.Figure 7 shows fire propagation behaviours in specimens of wooden façade with chemical fireretardant treatment.In spite of different flaming properties, remarkable fire propagations were not observed at both specimens (No. 4-a and No. 4-b).Figure8shows fire propagation behaviours in specimens of aluminium composite panel with polyethylene core.At No. 5-a, fire propagation occurred 10 min after the start of the test and only the vertical joint of panels located on the centre line of the specimen were seriously damaged.On the other hand, at No. 5-b, fire propagation occurred 3 min after the start of the test and many parts of the specimen were burned out 8 min after the start of the test.

FiguresFigure 10 .
Figures 11 and 12 show correlation of respectively, peak flame height and peak incident heat flux on the specimen's top (HF6 in Fig.1) between test results with heat output of 600 kW and 900 kW.Flame height means the height from the lintel to the point which measured over 500 • C and was most distant