A Room Fire Screening Test Procedure F R E D L. F I S H E R , F R E D E R I C K W. M O W R E R , and ROBERT BRADY WILLIAMSON

University of California, Berkeley Full-scale room fire economics necessitate a screening test procedure requiring only a small amount of material to quantitatively assess heat release rate. This test could determine which materials may justify full-scale testing and those which fail early. Such a procedure is described, sample results are given, and it is suggested that correlation of screening test results with fun-room and bench-scale test methods could improve the evaluation of the pre-flashover fire spread characteristics of materials. INTRODUCTION H A S B E E N recognized since the early 1970s that small-scale test I Tmethods are unable to accurately predict the fire growth potential of certain types of interior finish materials, particularly synthetic polymers. This has led to an interest in full-ecale room fire tests for this purpose. In 1982, a Room Fire Test Method was approved for publication for information only, in Part 18 of the A S T M Book of Standards. 1 This proposed test method appears very promising as a means of providing a realistic assessment of the fire growth characteristics of interior finish materials in their end-use configuration. Until smaller, less expensive tests are developed the ASTM test will be a primary test method for evaluating interior finish materials. The Room Fire Test is now standardized to a test compartment 2.4 m {8 ft} wide, 3.7 m {12 ft} long, and 2.4 m {8 ft} from floor to ceiling. The ignition source is a propan~fired sand burner placed in one corner of the compartment. The test is conducted b y igniting the burner and measuring the following data: temperature 100 mm (4 in.) below the ceiling, the rate of heat release {RHR} b y oxygen depletion in the stack, the heat flux to the floor and the light transmission through the smoke in the stack. Once the test 238

Room Fire

239

WALL SAMPLES D

OUTLINE OF ~ CEILING SAMPLE :

TI

~.~

1

20 m

,, ,i__,: I GAS ____ BURNERJ t

I

216 '~

~22

l 8'

I mt8

I

_1_

~,

~1_

3'

_1_

3'

~1 _l

I-

t2'

- I

Figure 1. Plan of Standard Room Fire Test compartment showing specimen location for screening test and burner.

compartment has been constructed and the instrumentation has been calibrated, the test is only slightlymore difficultto conduct than other standard fire tests, such as A S T M El19. However, from a material manufacturer's standpoint, there is one problem with the current version of the test method, namely, a large amount of material is needed for each test. Since the R o o m Fire Test Method is a full-scalesimulation, the entire ceilingand three walls of the compartment must be lined with the material to be tested. Thus, product development is likelyto be difficultand costly because nearly 33 m 2 {350 sq ft) of each formulation is required for every room fire test. DESCRIPTION OF SCREENING

A ROOM TEST

FIRE

A room firescreening test protocol is a possible solution to this problem. This screening uses the same equipment and instrumentation as the fullscale room firetest method except that it employs two ignition source programs and a reduced sample size. The screening test requires 1.82 m s (20 sq ft) of test material for each of the two runs. The test material is placed on the walls and ceilingforming the ignition source corner, as shown in Figure 1. Each of the two walls comprising the corner contains a strip of test material that measures 0.3 m (I ft} in width and 2.44 m {8 ft)in height. The ceilingpanel measures 0.6 m by 0.6 m

240

Fire Technology

(2 ft by 2 ft). The remaining surfaces of the room are lined with 13 mm (89 in.) thick gypsum wallboard. The test material can be applied to a gypsum wallboard substrate if this simulates anticipated end-use conditions. This test specimen size provides a large enough area to evaluate the ignition and vertical flame spread characteristics of the material under consideration. The screening test utilizes two different ignition source exposures, each characterized by its gross rate of heat release (RHR). The 44 k W {gross) exposure which produces a 1-1.3 m (3-4 ft) flame height is used to evaluate the ignition and vertical flame spread characteristics of the test material. The 44 k W exposure is a square wave pulse that continues for a period of 15 minutes. The second test run utiliT~san ignition source program identical to that described in the 1982 A n n u a l Book of A S T M Standards. ~ This exposure begins at 44 kW {gross) rate of heat release {RHR) and increases every 30 seconds in 44 kW increments until a gross RHR of 176 kW is reached. The program continues at the 176 kW level for an elapsed time of 15 minutes. The purpose of this exposure is to evaluate the characteristics of the material under the same ignition condition as required by the current draft of the test method. Should this ignition source program be changed in a future version of the Room Fire Test, the screening protocol would probably also be changed. We will refer to this second ignition source exposure as the 176 kW exposure. RATIONALE FOR THE SCREENING TEST PROTOCOL The screening test protocol was developed as a means of predicting the fire growth behavior of materials subjected to the Room Fire Test Method without the need for lining the entire room. If m~terial costs are not a consideration, little is gained by not simply conducting the Room Fire Test Method. A flowchart is shown in Figure 2 to illustrate the logic and rationale in evaluating a material using the Screening test protocol. The first step of the flowchart indicates exposure of the material to the 44 kW (gross) ignition source RHR. The 44 kW source represents the burner RHR prescribed during the first 30 seconds of the current draft of the Room Fire Test. In the screening test this exposure is used to evaluate the ignition and flame spread characteristics of the test material. If the material produces f l a m e spread to the ceiling, no further consideration is given to such a material since it would likely lead to room flashover under the conditions of the Room Fire Test Method. If the material does not spread flames to the ceiling, it is then advanced to the next step in the flowchart: exposure to the 176 kW ignition source. The 176 kW level represents the ignition burner RHR reached in the Room Fire Test after an elapsed time of 90 seconds. The first three diamonds following the 176 kW box contain criteria used to denote room

Room Fire

241

flashover. If any one of these criteria is reached during the screening test then the material can obviously be retired from further consideration. The last diamond in the flowchart, RHR > 300 kW, indicates an estimate of the msximllm R H R that a test material can produce during the screening test with the likelihood of not causing room flashover when subjected to the conditions imposed by the R o o m Fire Test Method. If none of the conditions shown in the diamonds occur, then the material can be considered for evaluation using the full R o o m Fire Test Method. It must be stressed that successful completion of the screening test does not guarantee that the material will not cause flashover when subjected to the R o o m Fire Test Method. EXPERIMENTAL

RESULTS

A number of experiments have been conducted using the screening test protocol described above. A comparison of four different materials discussed in this paper is shown in Table 1. Three of the materials were subjected to both the 44 kW and 176 kW ignition exposures. The fourth material, 6 mrn ( 88 in.) A-D plywood, was subjected only to the 176 kW ignition source program. The results from both the screening test and the fully lined Room Fire Test are presented for the plywood in terms of the total rate of heat release and the average ceiling temperature. The RHR and the average ceiling temperature histories

TEST I TESTl

MATERIAL I ~ NUIV~t: 44 kW IGMTIONSOURCE EXPOSURE CEILING? Figure 2. Flow chart showing the logic used in conducting the screening test protocol

[

SCREEN'NGTEST I NUMBERS" 17GkW ~ONITIONSOURCE EXPOSURE

I-- YES I~

OFROOM? NO

YES

MAT 'tAL

~ 2:0iimS?

Y

FULL:JCALE ROOMFIRETEST

242

Fire Technology

recorded in each exposure for two of the materials are shown in Figure 3 through Figure 8. It should be noted that the data acquisition system is activated three minutes before the ignition source is lit. This procedure is employed in order to detect the presence of any baseline drift or instability in the instrumentat i o n used in the test procedure. The RHR curves and the average ceiling temperature histories in the screening tests performed on glass fiber insulation batts faced with polyvinylchloride (PVC) film are shown in Figure 3 and in Figure 4 respectively. The energy release contribution of the PVC vapor barrier and the binder in the batts is minimal and is partially obscured by the noise band present in the instrumentation system. The contribution of the glass fiber insulation and vapor barrier is somewhat more apparent during the 176 kW exposure. However, in both cases the RHR and average ceiling temperature histories are similar to those recorded during calibration tests in which gypsum wallboard was placed behind the burner. The RHR curves and the average ceiling temperature histories of a foilfaced fiber-reinforced foam plastic insulation board show there is very little contribution when it is exposed to the 44 kW ignition source, nor is there any apparent flame spread observed from the 44 kW ignition exposure. However, there is a steady contribution with the 176 kW ignition source of approximately 25 kW, and the ceiling temperature was over 600 ~ C during the first three minutes after the ignition source had reached its fall value. The ceding temperatures in the other quadrants and the center of the room are approximately the same as recorded in the experiments with the glass fiber insulation shown in Figure 4. The foam board insulation used in these experiments is covered with an aluminum foil racer of approximately 0.04 mm (1.5 rail) thickness, and it contains glass fibers which prevent fissures from forming under the severe heating conditions of the ignition source. The third material, unmodified fiber reinforced plastic (FRP), is much more flammable than the other two materials. The RHR curves, as shown in Figure 7, for the FRP are almost identical for both the 44 and the 176 kW exposures, and the ceiling temperature histories shown in Figure 8 are also almost identical for each exposure. Note that the FPR produced fiashover conditions in the compartment within one minute after the ignition source was activated. The final material, 6 mm (88 in.), A-D, Douglas fir plywood, was subjected to the 176 kW burner program in two screening tests with almost identical results. It was also used in several full-scale room fire tests. Results of one screening test are shown in Figure 9 and those of a full-scale room test are shown in Figure 10. Note that the reproducibility between experiments is fairly accurate. The same data were recorded during a full-scale Room Fire Test of plywood with approYimately the same moisture content as that in the screening tests. Note, however, that the Room Fire Test which was fully lined with plywood had to be extinguished quickly.

~,

0",

~.~ ~o

"~ i

•0

D"~ G R E E S

DEGREE8

C

C

.t= ;=

~ r

..,

il

K i L.O~AT1-8 ....

b~ 4==

l

==

Fire Technology

244 3eo

27e

18e

IF-

II" (I

E Z 0 J

0 J "~

~so

E

12e

u a or ~

9

,

,

,

. S

.

.

.

. 1~

~e

i i'li,

+ stnm~

.

.

.

. . ~s

.

9 2,3

TIME

T IME

( M~.-44JTE~ )

(MINUTE~)

(b)

(a)

Figure 5. The R t t R curves for a foil-faced fiber-reinforced foam plastic insulation board are shown (left) for 44 k W exposure; and (right) for 176 k W exposure using the screening test protoco I

7~

7~

5S~

See 0

~SO

9J

hl

o lu

I11

lee

/

O

T II"IE

TIME

(MZNUTE~; }

(a)

(b)

Figure K Ceiling temperature histories recorded during screening tests of the foil.foead fiberreinforced foam plastic insulation corresponding to the RtZR curves shown in Figure &

Room Fire

245

.+ -.,,l

II

I

i

,..

|

|

t

~

~+

i

1

i

,

i

,

i

1

,

+

UI

II

~ M O ~ I ~

,0

<~,,, Q

It

m,, m,,,

l,,,

++j+

.~-

ii

+

.

,

!

i

i

!

| ~&.LI::;IMO"I I :>4

s

s

.

~,-.,

246

Fire Technology

)

J

t~_~,t

. . . .

~ ....

t ....

t,

.l*,x,l.,,=,t

....

0

I ....

I ....

9 ....

I ....

t,,,

,/,

il.l,.I

8~3~0~G

/ ~J

w~ Z

.,

,,

,

247

Room Fire

/

!

~

~

o

~

o

~

.~

Fire Technology

248

OI

/ /.' 2

/

/

/

~ . = = .

~i~M03I~

L~J

Room Fire

249 DISCUSSION

Table 1 provides a comparison of the behavior of four materials under the conditions of the screening test. The maximum RHR, under the 176 kW exposure, ranges from 887 kW for the F R P to only 60 kW for both the vinyl faced glass fiber insulation batte and the foam insulation board. The 6 mm (88 in.} thick plywood samples are intermediate in value at about 389 kW. These peak values correspond to R H R per unit area figures of 35.9 kW per m 2 for the glass fiber insulation batts and the foam insulation board, 531 kW per m ~ for the F R P and 233 kW per m 2 for the plywood. The estimate of 300 kW as the peak R H R permissible during the screening test (see Figure 21 corresponds to a R H R per unit area of 165 kW per m 2. Given this value, it is clear from Table 1 that the 6 mm plywood and the F R P material would be expected to cause flashover if either were subjected to the Room Fire Test Method. The results presented in this paper and the results from previous experiments2 confirm that the 6 mm thick plywood will cause room flashover. The F R P material has been evaluated using a geometry very similar to that of the Room Fire Test Method and has been found to cause very rapid flashover. The F R P is an interesting material because it produces nearly identical results when exposed to the 44 kW square wave burner program and to the 176 kW stepped program. Clearly the 44 kW ignition source is more than adequate to cause ignition of the FRP. Once ignited, the F R P R H R is so great, and occurs so rapidly, that the additional heat produced by the ignition source at the 176 kW level is negligible when compared to that produced by the FRP. Samples of the plywood from the same lot used to conduct the screening tests and the room fire tests described in Reference 2, were evaluated by the National Bureau of Standards, Center for Fire Research using their NBS II heat release rate calorimeter? The NBS II calorimeter, using an 80 kW per m 2 exposure, recorded a peak R H R of 253 kW per m 2. This value is between the 260 kW per m z and 205 kW per m z, values obtained during the two 176 TABLE 1. Comparison of Room Fire Screening Test Results

Maximum R H R

Maximum Average Total Energy Release Ceiling Temperature During 15 Minute Test

(kw)

Material

(~ c)

( MJ)

9

44 k W

176 kW

44 kW

176 k W

44 kW

176 kW

G h s s Fiber Insulation faced with PVC film

19

60

130

380

0.6

15.4

Glass Reinforced foil faced foam

34

59

144

410

9.4

25.5

Fiberglass Reinforced Plastic

942

887

821

730

70.4

66.3

6 m m thick A-D Douglas Fir Plywood

--

--

675*

--

97.9*

* Average of two t e s t s

389*

250

Fire Technology

kW room fires screening tests. We do not suggest t h a t the apparent agreement between the results from the NBS I I calorimeter and the room fire screening test represent correlation between these two methods of testing. Correlation of the results from room fire tests, room fire screening tests and bench-scale tests could provide a viable methodology for rapidly and inexpensively evaluating certain classes of materials before incurring the cost of a final evaluation using the Room Fire Test Method. However, it should be recognized t h a t correlation between bench-scale test results and the room fire test m a y never be viable for composite materials which require the full-scale dimensions of the Room Fire Test, or the room fire screening test, in order to evaluate the propensity for delamination and fall-off caused by thermally induced mechanical stress. The use of a bench-scale method, such as the Cone Calorimeter, ~ m a y prove especially useful as a means of predicting the behavior of relatively homogeneous materials during the Room Fire Test. The Cone Calorimeter method is particularly suited for this purpose since it provides for test runs at different heat flux levels which can be conducted quickly and in which the specimen R H R is determined nonthermally b y measurement of oxygen consumption. The first step in such a correlation would be an investigation of the spatial variation in the wall heat flux immediately adjacent to the gas burner ignition source employed by the Room Fire Test Method. Once such a heat flux map were available, t a r g e t values for use with the Cone Calorimeter could be developed. I t is recognized t h a t the current level of understanding of turbulent flamespread is not sufficient to produce a purely theoretical correlation between the results from bench-scale experiments and the Room Fire Test Method. This lack of understanding does not, however, preclude the development of a first order correlation between small-scale tests and the Room Fire Test Method based on a combination of current theoretical understanding coupled with empirically generated knowledge. REFERENCES American Societyfor Testing and Materials,Annual Book of A S T M Standards, Part 18, Philadelphia, Pa. {1982). 2Fisher, F. L. and Williamson, R. B., Intralaboratory Evaluation of a Room Fire Test Method, NBS-GCR-83-421, National Bureau of Standards, Washington, D.C. {1983}. 3Parker, W. J., Personal Communication. 4Babrauskas, V., Development of the Cone Calorimeter -- A Bench-Scale Heat Release Rate Apparatus Based on Oxygen Consumption, NBSIR 82-2611, National Bureau of Standards, Washington, D.C. {1982). ACKNOWLEDGMENT:We wish to gratefully acknowledgethe assistance of many t~ople in this research. Bill MacCrackenand Peter Tierney helped with the construction and the conduct of the experiments. Pete Goodevewrote the software for the date acquisition system. Cecile Grant helped with the editing of this paper and Claire Johnson did the. typing. , L Financialsupport for part of the research was provided by the Center tor vire ltesearcn at the National Bureau of Standards under Grant Number NBSONADA1072.