Smoke production in the cone calorimeter and the room fire test

Fire Safety Journal 17 (1991) 27-43

Smoke Production in the Cone Calorimeter and the Room Fire Test*

B. A.-L. Ostman & L. D. Tsantaridis Swedish Institute for Wood Technology Research, Box 5609, S-114 86 Stockholm, Sweden

ABSTRACT

The smoke production rates for 13 different surface lining materials have been determined in the cone calorimeter at three irradiance levels: 25, 50 and 75 k W / m 2. Two light systems have been used simultaneously, a helium-neon laser and a white light source, showing equal results. Different smoke parameters obtained in the cone calorimeter have been compared with those obtained in a full-scale room fire test. There seems to be a reasonable agreement which must be further studied.

INTRODUCTION

Measurement of smoke production from different materials has so far mainly been carried out in static boxes, of which the NBS Smoke Density Chamber is best known? This test has several disadvantages: the rate of smoke production is hard to follow accurately; the vertical orientation of the specimen excludes relevant testing of thermoplastics; it has no measurement of mass loss and a limited range of irradiance levels. Some of these disadvantages have been overcome in later modifications, but the main problems with a static, accumulative test method still remain. * Paper presented at the conference 'Fire: Control the Heat, Reduce the Hazard', 24-5

October 1988, London, UK. 27

28

B. A-L. Ostman, L. D. Tsantaridis

A dynamic, flow-through system has therefore been proposed 2'3 and is expected to have a better predictive capacity for full-scale and real fires. Such an instrument for small-scale testing is now available in the cone c a l o r i m e t e r : It was originally developed for rate of heat release measurements but also enables the determination of smoke release, time-to-ignition, etc. However, some researchers have pointed out the importance of measuring matured or aged smoke, 3'5 which might not be the case in a flow-through system. Only limited comparisons between static and dynamic conditions have been carried out on a small scale 6 showing less smoke production during dynamic conditions. The relationship of smoke production in small-scale and full-scale fires is of main interest, even if progress so far is limited. 3'7"21The use of a mass loss related smoke parameter as a smoke potentiaP or smoke extinction area 8 has recently shown promising results 2'9 in some cases but has so far been applied to only a few surface linings. 1° The cone calorimeter for smoke measurements includes the determination of mass loss and makes such comparisons easier. Smoke measurements in the cone calorimeter are performed by a H e - N e laser beam 4 in contrast to most earlier measurements. The laser has several advantages such as simple design, high level of beam collimation and simplified theoretical relevance. 2'1~'12 A laser system may, however, create some problems with signal stability and relation to visibility, which is important for escape in real fire situations. The signal stability has been improved by a second controlling p h o t o m e t e r in the cone calorimeter application, but the relation to visibility has not yet been proved. Only one direct comparison between a laser beam and a white light source has been published so far, and was performed under static conditions. H It indicated a slight but significant difference between the two light systems. This study directly compares the laserwith a white light source in the cone calorimeter under dynamic conditions. It also presents smoke production data and smoke potentials for a set of different surface linings and makes a first attempt to relate t h e m to a full-scale r o o m fire test. It is an extension of an earlier preliminary version) 3 A full report on the cone data is also available) 4

EXPERIMENTAL The experiments have been performed in a cone calorimeter (Fig. 1) which is in accordance with the standard v e r s i o n : It is a further development of an earlier version 15 used to test the effect of specimen

29

Smoke production in cone calorimeter and room fire test

fhermocouptes. exhoust duct 110mmid

to fan orifice piote

mensuremenfs to qas anmyzers (02, CO,CO2) from obove: detedor

I ~ photocell

,,', : ~bto.rap

/-"--~ cone heater spark ~ plug ] ]

specimen

holder

FiS. I.

IQser I

The cone calorimeter.

size. New main items are the cone heater, the spark igniter, the hood and the exhaust duct. It is also equipped with two light systems to measure smoke production. The cone heater and the spark igniter with motor was delivered from the University of Ghent, Belgium. A square hood and a circular exhaust duct with 110ram i.d. is connected to the earlier constant volume radial fan, which is situated about 10 m away from the cone heater. This position of the fan is a real advantage since it does not require a high-temperature fan and will not influence the flow or gas measurements. The volume flow can be varied by different dampers. The orifice plate is placed about 825 mm from the curve of the exhaust duct and the straight free section after it is about 1200 mm long. The oxygen concentration is measured by a paramagnetic cell (H & B Magnos 4G from Hartmann und Braun AG, Frankfurt, Germany) and the concentrations of carbon monoxide and carbon dioxide by IR (Siemens Ultramat 22 P from Siemens AB, Stockholm, Sweden). The gas sample is taken from a ring sampler placed about 675 mm after the orifice plate. The gas sample passes a cold trap where moisture is removed, then a filter of loosely-packed glass wool and a tube with water-free CaSO4 for extra drying. The gas then passes through a pump and finally passes a 2-7 # m glass fiber filter. In order to minimize the transient time, part of the flow is wasted after the pump. The smoke is measured by two different light systems placed close together at about 50 mm distance and about 100 mm after t h e gas

30

B. A-L. Ostman, L. D. Tsantaridis

sampler. At first, there is a H e - N e laser of wavelength 633 nm with silicon photodiodes as main beam and reference detectors 4 delivered from Ghent University (model OEMO5P, Aerotech GmbH, Niirnberg, Germany). Then there is a white light source from a 10W tungsten filament lamp for which the beam is made parallel by a lens system. The detector has a spectrally-distributed response that duplicates the human eye (model PIN-10AP from United Detector Techn., Hawthorne, CA, USA). In both cases the smoke release is expressed as smoke production rate in ob x m3/s and smoke potential in ob x m3/g according to Rasbash. 5 The latter parameter is directly proportional to the specific extinction area in m2/g 4 by a factor of In 10. All data are reduced to the same timescale, i.e. minor delays due to gas transport and response of instruments are corrected for. CALCULATIONS The basic smoke parameter used is a quantity called obscura (ob). s One ob is the smoke concentration giving a light absorption of 1 dB/m, which is equivalent to a visibility of about 10 m. Obscura, DE, is defined as DE = (10/L) log (Io/I) (ob) where L is the path length (m), Io is the light intensity in the absence of smoke, and I is the light intensity in the presence of smoke. The smoke production rate, Dsp, is defined as Dsp = DE X 12

(ob x m3/s)

where I2 is the volume flow of gases in the exhaust duct in m3/s. The smoke potential, Do, is defined as Do = Dsp/rh

(ob X m3/g)

where rh is the mass loss rate in g/s. The smoke potential may be expressed equally well as smoke extinction area, SEA (m2/g) SEA = Do x 100 In 10 TESTED PRODUCTS The test products used are listed in Table 1. All of them originate from the same lot which was initially selected and used for several studies on

Smoke production in cone calorimeter and room ]ire test

31

TABLE 1

Tested Wall Lining Materials Material

Thickness (ram)

Density (kg/m 3)

Rigid polyurethane foam Textile wall-coveringon rock-wool Insulating fiber board Expanded polystyrene Medium density fiber board Wood panel (spruce) Paper wall-covering on particle board Particle board Melamine-faced particle board Plastic wall-coveringon gypsum board Textile wall-coveringon gypsum board Paper wall-coveringon gypsum board Gypsum board

30 42 + 0.5 13 49 12 11 10 + 0-5 10 13 13 + 0.7 13 + 0-5 13 + 0-5 13

32 150 250 18 655 450 670 670 870 725 725 725 725

reaction to fire within Scandinavian fire laboratories. ~°,13-2° They were all tested in the horizontal orientation in the cone calorimeter.

T W O L I G H T SYSTEMS F O R S M O K E All smoke measurements in the cone calorimeter were m a d e simultaneously with two light systems, a H e - N e laser and a white light system. In most cases they showed an excellent agreement. In only some cases very narrow peaks had somewhat different heights. This can probably be overcome by more frequent data sampling (5s used here). Representative data are given in Fig. 2. Some significant difference between the two light systems might have been expected due to their different wavelengths of light, n,~2 Differencies should occur especially for smaller smoke particles. One interpretation of our results may therefore be, that the smoke particles formed in the cone calorimeter are larger than some critical level; at least for the materials tested here. One advantage with the laser light system is its simplicity with minimum risks for soot deposits which might occur on the lenses of the white light system when materials with large smoke release are tested. Some material may also produce a sticky smoke which easily deposits, as has been experienced with, e.g. Polymethylmethacrylate (PMMA). However, the signal stability of the laser system might be improved.

B. A-L. Ostman, L. D: Tsantaridis

32

% (obwm3/s)

0.2

INSULATING FIBER BOARD

0.t

.

.

.

.

i

.

.

.

.

t00

i

.

.

.

.

200

i

.

.

.

.

i

300 TIME (a)

.

.

.

.

400

i

.

500

•

,

i

I

BOO

(a) Osp (obm3/s)

0.5 O. 4

|

PARTICLE BOARO

~LAMI~AOED 75 kW/m2

0.3

0.t

0

.

0

t00

.

.

.

P00

.

.

.

.

.

.

.

300

.

.

.

400

.

.

.

.

.

500

(500

TIWE (,,,)

(b) Fig. 2. Comparison of a laser and white light system for measuring rate of smoke p n x l u ~ i o n (D~,) in the cone calorimeter . . . . . White light; - - , laser. (Note the different

scales.)

REPEATABILITY Data presented here are from double tests at 50 kW/m 2 irradiance and from single tests at 25 and 75 kW/m 2. This may be justfiied since the repeatability seems to be acceptable. Figure 3 gives an example of materials with quite low smoke production. At higher smoke production the repeatability is usually improved.

33

Smoke production in cone calorimeter and room fire test Dsp (ob.13/e)

0.2

PARTICLE BOARD

Laser 0.t

,jr

"~

j

Fvl

jv I

i

'..x

,

i

,

0

i

.

.

.

.

i

100

.

.

.

.

i

200

.

.

.

.

t

300

.

.

.

.

i

400

.

.

.

.

i

500

600

TINE (s)

(a) Os~ (obNl3/e)

0.1

6YI~JUM BOARD Laser 0.05

0

i

0

.

.

.

.

i

.

100

.

.

.

I

200

.

.

.

.

i

.

.

300

.

.

i

.

.

.

400

.

i

500

.

.

.

.

i

600

TIME (e}

(b) Fig, 3.

Repeatability in smoke measurements in the cone calorimeter. The figure shows laser data for particle board and gypsum board at 50 kW/m 2.

SMOKE PRODUCTION The smoke production in the cone calorimeter was measured at three irradiance levels from the cone heater 25, 50 and 75 kW/m 2 and with two light systems. As the light systems give equal data (see the preceding section 'Two Light Systems for Smoke') just the white light data are presented here. Figure 4 shows the rate of smoke production and rate of heat release

34

B. A-L. Ostman, L. D. Tsantaridis Dsp (obU3/s]

0.3

TEXTILE NALL-COVERI~ ON GYPflUN BOARD

75 kN/m2

0.2

Nhtte light

iia 5o kW/,,:, O.i

,j, I

,

.

.

0

,

i

,

,

.

iO0

. ,

.

.

i

.

,

.

,

•

200

. •

.

.

.

.

.

.

.

.

.

.

i

i

i

i

300

400

500

600

TINE [s)

(a) 800

8HR [kW/m2]

600

75 kN/m2

4OO

~ 50 kW/m2

TEXTILE NALL-COVERIN6 ON 6YPSUN BOARD

II I ~ 25 kW/m2

200

0

.

0

.

- .- . - - ' ~

- -i " -

loo

,

,

. . ,. . . . .,. . .

i

200

,

L

; . ., . . . . ,. . . . .i . . . . .. . . .. . .. . . .. . . . . .

300 TINE (s)

i

.400

.

.

.

.

i

500

.

.

.

.

t

600

(b) Fig. 4. Rate of smoke production (De) and rate of heat release (RHR)for some fining materials in the cone calorimeter. (Note the different scales in both cases.) for some typical materials. The smoke is released mainly during periods of heat release, but may partly be released somewhat earlier than the heat. The early smoke released before ignition is usually white and different from the smoke after ignition which is darker. In some cases they appear as distinct peaks. The rate of smoke production (ob x m3/s) differs between the heat flux levels, as does the rate of heat release, but expressed as smoke potential (oh × m3/g), it is similar regardless of heat flux levels, see Fig. 5. However, peaks are related to the time-to-ignition which differs with heat flux level.

Smoke production in cone calorimeter and room fire test

35

Dsp(ob*~/s)

0.2 _

PAPER NALL-COVERIN6 ON PARTICLE Bokqo Nhtte

11ght

0.1 ...". 75 kW/m2 ../:" : . . •

.

.'.-

-.

50

:

ii:: I

:"

/,"'.....:~.~?, .':'":,'.,- .....

. . . .

I

0

. . . .

I

~.00

,.,

,

,

200

..::"'"'"'"'"-

:.

. I

'..

. . . .

I

300

. . . .

I

400

. . . .

I

500

600

TINE [s) (c)

400

RIq (kN/12)

320

PAPEg NALL-COVERING ON PkqTI~.E BOkqO

240

75 kW/m2 . . ' . . %

'

• ".

160

80

|

kN/m2

.

.

.

.

t

t00

.

. "

"

I E5 kN/m2 # I .

.

.

.

a

,

200

,.,,'7.,

i

•

•

i

.

.

300 TIME

,

i

L

400

a.

•

.

.

n

500

.

•

.

"

i

I

600

(s)

(d) Fig. ~-co~d.

The smoke potential is directly proportional to the smoke extinction area expressed in m2/g. ' The general appearance of the two curves is very similarY Smoke production data for all materials tested are given at 50 kW/m 2 in Table 2. They include smoke production during smoke peak period (ob x m3), average smoke potential (oh x m3/g) after ignition during a peak in smoke production, and average smoke production per heat release (ob x m3/~1~J). The latter parameter is in the order of 15-35 for most products except a few with ratios 100-200, indicating a higher smoke release in relation to heat release.

B. A-L. Ostman, L. D. Tsantaridis

36 5-

DO (obV,B3/O)

4 •- 50 kW/m2

PLASTIC WALL-COVERING ON GYPSUM BOARD

I ! 25 kN/m2

3

White l i g h t

1 II

II II ii |1

2

:',

,,|!

,~: |,'| ,'~11~/ 0

. . . .

I

,

' " \r

,

,

,

100

I

_r

. . . .

I

200

. . . .

I

300

. . . .

[

400

i

~

i

500

i

I

600

TINE (e)

(a)

~

5-

oo (obm3/gl

4 -

#

~

kN/m2

RIBIn P O L ~ T H A N E White light

r

3 -

FOAN

2

t

0

i

i

,

I

tO0

i

t

,

i

l

200

i

i

i

i

I

i

300

i

,

•

i

.

400

.

.

.

i

500

. . . .

i

600

TINE (e)

(b) Smoke potentials (Do) for some materials in the cone calorimeter. (Note the different scales.) GAS P R O D U C T I O N The production of carbon dioxide and carbon monoxide has also been detected. The curves for the carbon dioxide production rate have almost the same shape as the rate of heat release curves and show a production of about 40 liters CO2/MJ (heat release) for all materials tested (liters at atmospheric pressure and ambient temperature). The curves for the carbon monoxide production rate have a somewhat dissimilar shape compared to the RHR-curves, partly depending on the

For periods where CO is detectable.

Rigid polyurethane foam Textile wall-covering on rock-wool Insulating fiber board Expanded polystyrene Medium density fiber board Wood panel (spruce) Paper wall-covering on particle board Particle board Melamine-faced particle board PlMtic wall-vovering on gypsum board Textile wall-covering o n gypsum board Paper wall-covering on gypsum board Gypsum board

Mater/a/ 2 11 12 52 28 21 27 34 42 10 20 21 34

28.5 2-7 3.5 57-3 14-1 3.5 3.1 6-7 55.0 4.5 2.2 1-0 0.5

3.8 1-7 0-4 5-5 0.4 0-4 0-3 0-4 1-6 1.7 0.7 0-5 0-4

140 25 25 110 35 20 15 20 200 30 25 25 15

20 4 190 130 160 50 40 210 35 6 3 20 2

A verage smoke Average smoke Average smoke Smoke potential production production production after per heat per CO Time to during smoke ignition release production a ignition peak period DO Dsp/RHR D.p/CO (s) (ob X m 3) (ob × m3/g) (ob X m3/Mj) (ob × m3/liter)

TABLE 2 S m o k e and Gas R e l e a s e D a t a in the C o n e C a l o r i m e t e r at 50 k W / m 2

4 20 250 25 200 130 180 250 25 3 40 8 2

Average CO2/C0 production a (liter~liter)

200 90 15 55 2 2 3 4 25 30 25 30 25

Average CO production per mass loss (g/kg)

"..d

B. A-L. Ostman, L. D. Tsantaridis

38 6 -

~

(ml/s)

5 D POLYSTYRENE 4

3

;2

t 25 I

.

.

.

.

I

0

.

.

.

.

100

i

.

.

.

.

i

200

.

.

.

.

300

i

.

.

.

kN/m2

.

400

i

.

.

.

.

500

i

600

TIME (s)

(a) 0.

Usp

(obwm3/s)

o

-

EXPANDED POLYSTYRENE 0.6

White l l g h t

0.4

0.2

25

0

l

0

. . . .

I

100

. . . .

I

200

. . . .

I

. . . .

300

l

400

. . . .

I

500

kW/m2

. . . .

I

600

TIME {s)

(b) 1~o 6. Pnxluction rate of carbon monoxide (CO) in relation to smoke production rate for some materials in the cone calorimeter. The rate is expressed as volume flow at atmospheric pressure and ambient temperature. (Note the different scales in both cases.) low CO production during the test. However, they are quite similar to the curves for the smoke production rate. Some examples are given in Fig. 6. Generally, the peak in CO production seems to appear somewhat later than the peak in smoke production and in heat release. The average ratio of D~,/CO(ob x m3/1) which has been calculated for periods where CO-production is detectable, is dependent on the materials tested (see Table 2). The same is true for the ratio CO2/CO and for the CO production per mass loss (also in Table 2).

39

Smoke production in cone calorimeter and room ]ire test _

CO (ml/s)

MEDIUM DENSITY FIBER BOAF~

0.75

~

0.5

2

f 0.25

/ I

.

.

.

g ; ~11 25 kti/m2

i'N',',,

I

'filial

:, .

I

0

.

.

.

.

100

I

I

,:',:':"',

~

i

I

,

,

200

,

i

.

.

.

.

300

I

.

.

.

.

400

n

.

.

.

.

500

I

600

TIME (sl

(:) Dsp (obl(m3/s}

0.2

MEDIUN DENSITY FIBER BOCci}

Whtte light 0.1

I

0

i

n

n

i

I

100

.

.

.

.

n

200

.

.

.

.

i

.

.

300

.

.

I

400

.

.

.

.

n

.

.

.

.

500

a

500

TINE ($)

Fig. 6----contd.

R E L A T I O N TO THE R O O M FIRE TEST Some different means have been tried to relate the smoke production in the cone calorimeter and in the room fire test to each other. Full-scale room fire data are available for exactly the same lining materials. TM Smoke potential or the equivalent term, smoke extinction area is probably the best parameter for comparing smoke production in small-scale and full-scale fire tests. 2 In full-scale tests, however, the mass loss rate is not usually determined, but by using the effective heat release obtained in small-scale, the full-scale smoke production data

40

B. A-L. Ostman, L. D. Tsantaridis FULL SCALE to~o[ smoke production up to f.o. ob, m 3

15000 L

x

10000-

SO00, (D

/x

O.~q/

•

_

tofo[ smoke production

10

20

/,0

60 ob.m3

(a)

300- FULLSCALE smoke production peak before f o ob. m3/s

A

200-

x 100SHALL SCALE smoke production peak m~50kW/m2 ob. m3/s

e '

'

'

ols

11o

(b) Fig. 7. C o m ~ m between different smoke parameters obtained in the cone calorimeter at 50 kW/m 2 and in the room fire test is for those materials reaching flash over. In diagram (c) t w o materials are laclane because o f uncertain data in full-scale. m, particle board; l , insulating fiber board; I1, medium density fiber board; I'q, wood panel (spruce); A , rigid polyurethane foam; &, expanded polystyrene; ® , g3q3sum board; x, melamine-faced particle board; ~, paper wall-covering on particle board; @, paper wall-covering on g.b.; 0, plastic wall-covering on g.b.; 0 , textile wall-covering on g.b.; +, textile wall-covering on rock-wool,

Smoke production in cone calorimeter and room fire test

41

FULLSCALE smokeprod/heal release / ob.m3/HJ // / /

200

/ 100.

SMALLSCALE smokeprod/heat releose ob.mZ/HJ

/

(c) Fig. 7---contd.

may be converted to smoke potential. TM In an earlier study 13 the full-scale smoke production per heat released (ob × m3/Mj) was converted to smoke potential by multiplying with the average effective heat release (H/g) obtained at 50 kW/m 2 in the cone calorimeter. This effective heat release is constant during major parts of the fire test, but is hard to get very accurately for composite products with a short burning time. Another approach has to be tried for the general case. One possibility might be to use some equations proposed by Rasbash, 5 but the parameters included are hard to get. Another possibility is to use the carbon balance method9 to calculate specific extinction areas. It does not require mass loss of fuel, but mass loss of soot collected on a filter, which is not available for the test data used here. Therefore, only some direct comparisons of smoke production rate, total smoke production and smoke production per heat release are given in Fig. 7. It is obvious that the latter parameter shows best agreement, even if two materials had to be excluded due to uncertain data in the full-scale test. For most materials the smoke production is proportional to the heat release, but for a few others, it is significantly higher. However, the production of smoke and gases is not primarily dependent on materials, but to a high degree on ventilation conditions and sizes and shapes of flames. These parameters must be considered in more elaborate comparisons of the smoke production in small- and full-scale fires. CONCLUSIONS Smoke and gas production can be measured with good ac~tracy in the cone calorimeter; including materials with low smoke and gas release.

42

B. A.L. Ostman, L. D. Tsantaridis

A He-Ne laser and a white light system gives equal results. Generally there seems to be good agreement between smoke parameters obtained in the cone calorimeter and calculated from room fire tests. This needs to be studied further. More accurate means to obtain smoke potentials or specific extinction areas from room fire tests have to be found. One possibility might be to determine the mass of soot produced in room fire tests. ACKNOWLEDGMENT We are especially grateful to Dr G6ran Holmstedt, Lurid University for reviewing the paper, to Mr Bj6rn Sundstr6m, the Swedish National Testing Institute, for supplying additional full-scale data and discussing the results, and to the Swedish Board for Fire Research for financial support to part of this work. REFERENCES 1. ASTM E 662-83, Standard test method for specific optical density of smoke generated by solid materials. Annual Book of A S T M Standards, Vol. 04.07, 1987. 2. Babrauskas, V. & MulhoUand, G., Smoke and soot data determinations in the cone calorimeter. Special Technical Publication (S TP ) 983: Mathematical Modelling of Fires. Am. Society for Testing and Materials, Philadelphia, USA, 1988, pp. 83-104. 3. Quintiere, J. G., Smoke measurements: An assessment of correlations between laboratory and full-scale experiments. Fire and Materials, 6 (1982) 145-60. 4. ASTM E-5 Proposal P 190: Proposed test method for heat and visible smoke release rates for materials and products using an oxygen consumption calorimeter. Annual Book of A S T M Standards, Vol. 04.07, 1987, pp. 1203-19. 5. Rasbash, D. J. & Pratt, B. T., Estimation of the smoke produced in fires. Fire Safety J., 2 (1979/80) 23-37. 6. Drysdale, D. P. & Abdul-Rahim, A. F., Smoke production in fires: Small-scale experiments. Special Technical Publication (STP) 882: Fire Safety Science and Engineering. Am. Society for Testing and Materials, Philadelphia, USA, 1985, pp. 285-300. 7. Rasbash, D. J. & Drysdale, D. D., Fundamentals of smoke production. Fire Safety J., $ (1982) 77-86. 8. Babrauskas, V., Applications of predictive smoke measurements. J. Fire and Flammability, 12 (1981) 51-64. 9. Mulholland, G. W., Henzel, V. & Babrauskas, V., The effect of scale on smoke emissions. Proc. of the Second Int. Symp. on Fire Safety Science. Hemisphere, Tokyo, 1988, pp. 347-57.

Smoke production in cone calorimeter and room ]ire test

43

10. Holmstedt, G., Analysis of fire gases and smoke in SP-Rapp. Swedish National Testing Institute. 1984:22 (1984) (in Swedish). 11. Clark, F. R. S., Assessment of smoke density with a helium-neon laser. Fire and Materials, 9 (1) (1985) 30-5. 12. Mulholland, G. W., How well are we measuring smoke? Fire and Materials, 6 (2) (1982) 65-7. 13. Ostman, B. A-L., Comparison of smoke release rate from building products. Conference book, Fire: Control the Heat, Reduce the Hazard, Queen Mary College. Fire and Materials Centre, London, UK, 1988. 14. Tsantaridis, L. & Ostman, B., Smoke, gas and heat release data for building products in the cone calorimeter. Swedish Institute for Wood Technology Research,..Stockholm. Report 8903013, 1989. 15. Nussbaum, R. M. & Ostman, B. A-L., Larger specimens for determining rate of heat release in the cone calorimeter. Fire and Materials, 10 (1986) 151-60. 16. Andersson, B., Model scale compartment fire tests with wall lining materials. Lund University, Sweden, Report LUTVDG (TVBB-3041), 1988. 17. Magnusson, S. E. & Sundstr6m, B., Modelling of room fire growth--combustible lining materials. Paper presented at ASTM/SFPE Symposium on Application of Fire Sciences to Fire Engineering, Denver, Colorado (1984). Report LUTVDG/(TVBB-3019), Lurid University, Sweden (1984). 18. Sundstr6m, B., Full scale fire testing of surface materials. SP-Rapport 1986: 45, Swedish National Testing Institute, 1986. 19. Wickstr6m, U. & G6ransson, U., A simple model for predicting room fire growth based on cone calorimeter tests. A S T M J. of Testing and Evaluation, November 1987. 20. Ostman, B. A-L. & Nussbaum, R. M., Correlation between small-scale rate of heat release and full-scale room flashover for surface linings. Proc. of the Second Int. Symp. on Fire Safety Science. Hemisphere, Tokyo, 1988 pp. 823-32. 21. Christian, W. J. & Waterman, T. E., Ability of small-scale tests to predict full-scale smoke production. Fire Technology, 7 (4) (1971) 332-44.