Fire : Full Scale Tests

,

STATENS

PROVNING

SANSTALT

Nation

al

Tes

ting I nstitute

:

/ - -

CEILING JET

Fire Technology

\\

_I

/

/ / ' / / /

PLUME / ' / I _I

I

_{FLAME /}

\I

- - -

---_.;'

I

I

\1,

~"-BURNER

Björn Sundström

Ulf Wickström

FIRE: FULL

SCALE TESTS

CALIBRAT

I

ON OF TEST

ROOM - PART

I

NORD-TEST

PROJECT 1

43: 78, 2

Technical Report SP

-

RAPP

1981

: 48

Borås,

Sweden

1981

<

<

V) '"'

2

NATIONAL TESTING INSTITUTE Division of Fire Technology P.O. Box 857, 501 15 BORÅS Tel . int. +

46

33-10 20 00

BJÖRN SU

NDSTRÖM

U

LF

\I/I

CKS

TRÖM

FIRE

:

FULL

SCA

L

E TES

T

S

CALIBRATIO

N OF TEST RO

OM

-

PART I

NORD-

TEST PROJECT

143:78,

2

g

BORÅS, SWEDEN 1981 ~ u [ & "' ~ SP-R,i\PP 1981: 48

I TABLE OF CONTENTS ACKNOWLEDGEMENTS NOMENCLATURE ABSTRACT l INTRODUCTION 2 EXPERIMENTAL DESIGN 2.1 Test Compartment 2. 2 Gas Burners 2.3 Temperature 2. 4 Heat Flux 2. 5 Gas Velocity ~.6 Gas Analys is 2.7 Instrument Positions 3 TEST RESULTS

3.1 Temperature and Heat Flux 3.2 Mass Flow Rates

3.2.l General Flow Patterns

3.2.2 Measured Vertical Distribution 3.2.3 Measured Horizontal Distribution 3.2.4 Mass Balance

3.3 Heat Flow Rates

3.3.l Heat Release Rate by Combustion, _Qc 3.3.2 Heat Loss Rate by _{Convection, Qa} 3.3.3 Heat Loss Rate by Conduction, _Qw 3.3.4 Heat Loss Rate _{by Radiation, Qr} 3. 3. 5 Heat Balance of Fire Room

4 THEORETICAL MASS FLOW

4.1 Two Layer Model

4.2 Arbitrary Temperature Distribution 4. 3 Plume Flow

4.4 Wall Flow

5 SUMMARY

III

ACKNOWLEDGEMENTS

The work reported here is within the project "Fire Hazard -Fire Growth in Compartments in the Early Stage of Develop-ment (Pre-flashover) '' which i s a joint project between the Lund Institute of Technology and the National Testing

Institute. The project is financed by the Swedish Fire

Research Board, Nordtest (project 143-78,2) and the National Testing Institute. Earlier publications within the project are;

Sundström, B., "Brand: Storskaleförsök - bakgrund och försöksuppställning", National Testing Institute, SP-RAPP 1980:01, Borås 1980.

Sundström, B. and Wickström, U. , "Fire: Full Scale Tests -Background and Test Arrangements", National Testing

Inst itute, SP-RAPP 1980:14, Borås 1980, or ISO/TC 92/WG 2 & 4 N200 N388.

Magnusson, S.E. , "Recent Developments in Mathematical Model ing of Fire Growth", ISO/TC 92/WG 4 N386.

Pet tersson, O. , "Fire Hazards and the Compartment Fire Growth Process - Outline of a Swedish Joint Research Program", FoU-brand, No. 1, 1980, or Department of Structural Mechanics, Lund Institute of Technology, Report No. R 80-5, Lund 1980.

Sundström, B., "ISO Ignitability Test - Round Robin Test Results", National Testing Institute, SP-RAPP 1981:16, Borås 1981, or ISO/TC 92/SC 1/WG 2 N9.

Kaiser, I. and Holmst edt, G., "ISO Spread of Flarne Test -Round Robin Test Results", National Testing Inst i tute, SP-RAPP 1981:17, Borås 1981.

IV

Holmstedt, G., ''Rate of Heat Release Measurement s with the Swedish Box Test", National Test ing Institute, SP-RAPP 1981:30, Borås 1981.

NOMENCLATURE A C C p F g L':.H Hd k k e

.

m R T C x,

z

w p a SUBSCRIPTS i 0 00 f

=

=

=

=

=

=

=

=

=

=

=

=

=

=

=

=

=

=

=

=

=

=

=

=

=

area discharge coefficient gas specific heat view factor

acceleration of gravity heat of combustion

doorway height

thermal conductivity entrainment coefficient

mass flow rate

pressure difference heat release rate by combustion

heat loss by convection

through the doorway heat loss by radiation through the doorway

heat loss by conduction into the surrounding structure Tf/Too

temperature

length coordinates

thermal discontinuity height neutral plane height

emissivity fuel property gas density Stefan-Bolzmann constant in out ambient f ir e , fuel (m2) ( ) (kJ/kg K) ( ) (m/s2) (kJ /kg) (m) (W/mK) ( ) (kg/s) (N/m2, Pa) (W) (W) (W) (W) ( ) (°C, K) (m) (m) (m) ( ) ( ) V

ABSTRACT

Calibration tests of a large-scale fire room with inert walls have been conducted. Agas burner giving constant heat input was placed at the rear wall simulating an ignition source. Gas flows, temperature, radiation and

co

2 and CO concentrations were thoroughly measured particularly in the doorway. Theoretical analysis with available room fire models were also carried out and measured and calculated data were compared.

l

An upward gas flow observed along the walls heated by radiation from the fire plume and the hot gas layer was briefly discussed. This wall flow, not considered in

existing two layer models , was shown in these experiments to develop a layer between the neutral plane and the hot fire gases. The layer had moderate temperature and con-tained very little combustion products, which was ve ri-fied by measuring

co

2 concentration along the height of the doorway.

2

1 INTRODUCTION

This report presents the progress of an experimental

study to develop a lagre-scale room fire test. The project is named "Fire Hazard - Fire Growth in Compartments in

the Early Stage of Development (Pre-flashover)" and is

carried out jointly by the Division of Structural Mechanics at Lund Institute of Technology, and the Divisions of

Fire Technology and Chemical Analysis at the Swedish

National Testing Institute. The ultimate goal of the project is to develop a test method in full scale for

surface lining materials as wel l as furniture and other

products from which the behavior of the tested material

or product at areal fire can be predicted /1/. To achieve

that goal, i t is necessary to have reliabl e mathematical

medels so that results from tests could be used to predict

the fire behavior at a variety of other conditions.

Full-scale room fire experiments are often used for evalu

a-tion of the growth characteristics of materials and building

systems. In spi te of this, there is stil l no standard version of a room fire test, and there is therefore great

need to develop such a test method. The exact details

of a standard room fire testare presently being debated by

working groups within the American Society of Testing

Materials (ASTM) /2/. Within the ISO a working group

under

se

1 TC/92 has recently been proposed /3/ for

deve-l oping a similar method.

The purpose of a large-scale standard test would be to

evaluate the fire performance of materials or objects

under actual in-use situations. The contribution of a

specimen to the fire growth, within a previously

calibra-ted compartment can then be used to rate materials and

also to evaluate the validity of existing small scale

The current thinking favors a test compart~ent with a single door and the ignition source placed in one corner of thc

interior, away from the door. Various ways of gradi ng test specimens are plausible. The most straight forward and simple way is to just measure time to flash-over whcn flames start to come out the opening. More information of the properties of a specimen are, however, achieved i f the rate of burning could be continously measured, and the room could be used as a full-scale calorimeter. Rate of heat release can be measured in several ways. In thi s report the energy balance of a fire room has been studied under controlled condit ions using agas burner and inert walls, and the energy flow out the opening has been measured.

A test room of lightweight concrete with dimensions equal to that discussed by working groups within ASTM has been built. Several introductory experiments have been carried out and some are reported in /4/. Experiment s with non -combust ible wall l inings and controll ed gas burners were then carried out as well as tests of combustible wall linings with liquid heptane burners as ignition sources. The fire behavior of upholstered furni ture /5/ has al so been investigated in the test room and the results from large and small scale experiments have been compared. A burn room of the same dimensions has been invest igated at the University of Cal ifornia, Berkeley, /2/ for

testing of wall l inings. A methane gas burner placed in a corner was then used as ignition source. This report contains an experimental and theoretical study of the heat and mass balance of a room fire at quasi-steady state conditions. Gas was burnt in the fire room at two levels of intensity, 125 and 250 kW, and m easure-ments of temperature, thermal radiat ion, gas veloci ty, and

co

2 and CO concentrations were recorded at several positions in the doorway and inside the compartment.

4

2 EXPERIMENTAL DESIGN

2. 1 Test Compartment

Fire tests were conducted in a room of 15 cm thick light

-weight concrete with internal dimensions 2.4 m x 3.6 m

anda height of 2.4 mas shown in figure 2.1. A doorway 0.8 m

x 2.0 m was made in one of the short sides of the room. The dimensions of the room is in agreement with a standard test currently discussed within ASTM-committees.

I I

I

I

I

I

I I

Front

wiew

I I

I

I

I

l

1--r-1-C ) C) "'

Figure 2.1. Fire test room.

240

1

---

_r

0

BURNER I I

l

80

l

"

.,

Plan wiew

Measures in cm C) '° rn

2. 2 Gas Burners

Propane gas was burnt in circular sand bed diffusion

burners. Two types of burners with the diameter 30 and

5

40 cm were used. By measuring the mass loss of the gas bottle the rate of heat release could be controlled.

Tests were conducted at two levels of heat release rates,

125 and 250 kW, with the smaller and the larger burner,

respectively. During each test the rate of heat release level was kept constant. The heat of combustion of p

ro-4

pane, 6H , was assumed to be 4.64 x 10 kJ/kg. The bur

-c

ners were placed on the floar at the center of the rear wall as shown in figur e 2.1.

2.3 Temperature

Fine wire (0.25 nun diameter) t hermocouples of Chromel

Alumel type were used for measuring gas temperature.

Surface temperature of the surrounding structure was

measured with thermocouples welded to t hin 0.3 mm copper

discs of diameter 12.5 nun. Temperature in the surrounding

structure was measured by accurately placing thermocouples

in cylinders made of the same type of l ightweight concrete

as the wal ls, see section 2.7.

The welding point of the thermocouples were placed along

the cylinder axis and the wires were drawn along t he

expected isotherms to the periphery of the body before

bending and going out. A surface thermocouple was also

installed at the fire exposed end of the cylinder.

The thermocouple readings in the gas phase differ

some-what from the actual temperatures because of radiation

error. An attempt was made to estimate this error. The

and by setting the sum of the convective and radiative

transfer rates equal to zero a correction temperaturs 6

was calculated. Maximum correction occur at low gas ve lo-cities when the gas flow around t he thermocouple is

dominated by natural convection. By assuming the thermo

-couple having the shape of a hori zontal cylinder of twice the wire diameter (0. 50 mm) a Nussel t nurnber could be

calculated to be approximately 0.9. The emissivity of the t hermocouples is extremely difficul t to estimate. In this case where the thermocouples were soot y the emissivity was assumed equal to unity. As the errors because of radiation are difficult to estimate and are

smal l except for zones with low gas velocity temperatures

throughout this report are given wi thout corrections.

2.4 Heat Flux

Total heat flux, convect ion and radiation, was measured

with a Gardon-type instrument called Medt herm heat flux

meter. Pure irradiation, excluding heat transfer by

convection, was measured with radiometers of the type

of "Gunners" /6/, an instrument measuring irradiation

from a semi-sphere, and with the Medtherm heat flux

meter equipped with a sapphire window.

The inaccuracy of the Medtherm heat f lux met er is esti

-mated to be+ 3 % according to t he manufacturer .

2.5 Gas Veloci ty

The fire induced gas flow through the opening was m

ea-sured with bidirectional pitot tubes /7/, see figure 2.2. Near t he floor level at a height of 50 mm a fan anem o-·t1et er was used.

7

Figure 2.2. Pitot tube and thermocouple, for measurement of gas flow, and temperature, respectively.

Because the pressure differentials obtained from the probes are very small sensitive and costly manometers must be used. To reduce the number of manometers needed, a pressure scanner with micro valves was constructed.

The pressure transducer hasa resolution of 0.01 Pa, corresponding toa gas velocity of approximately 0.1

m/s at room temperature. The relation between pressure and velocity included corrections for Reynolds number according to calibration curves reported in /7/. The inaccuracy of the measurements is estimated to be less than 0.5 m/s.

The pressure difference is measured over a time

interval in order to achieve an average gas velocity thereby reducing the rapid fluctuations of turbul ent flow. The time interval can be altered; long intervais

normally increase accuracy but reduce the scanning speed.

The pressure scanner is controll ed by a micro processor which also processes the data before digital ly

The pitot tubes were placed in the doorway, 10 along the cent~rline and~ on a vert ically movable arm for measuring the horizontal flow distribut ion. Thus a total of 64

measuring points f?r the gas flow were achieved by moving the arm up and down during the tests. Two tests at each level of heat release rate were carried out; one test

with all pitot tubes in the vertical plane of the out-side of the door-jamb and the other test with al l tubes positioned in the plane of the inside of the door-jamb.

The measurements were checked by comparing the total

mass flows in and out of the room.

2.6 Gas Analysis

Fire gas was collected with a tube and pumped into a container to which instruments for measuring the concentration of carbon dioxide

(c

o

2) and carbon monoxide (CO) were connected. The tube was mounted

on the arm for the movable pitot tubes with the

inlet near the middle of the doorway.

2.7 Instrument Positions

The positions of thermocouples and pitot tubes in the

doorway are shown in figure 2.3a-b.

The horizontal arm was moved ver t ically <luring the steady

state period of the experiment s and measurements were

taken at heights given in figure 2.3b. This figure also

shows the ga~ sampling tube. Height of thermocouples in

the center of the room (the central tree) are shown in

figµre;2.4. Instrument s in the ceiling, the wall s and the floor are shown in figure 2.5. Positions of the thermocouples for measuring surface and internal wall temperature in cylinders placed in t he cei l ing and the rear wall are shown in figure 2.6.

Ooorway

see

n

from outside

10 9 8 7 6 ~ _5_ ~ -4

~

3 2

D

= PITOT TUBE

X

= THERMOCOUPLE l [han no P -tube 2 -3 4 5 6 7 8 9 10 TC 1 - - -2 · -3 4 5 -6 -7 8 9 10

-*

FOR CORNCAL 22 AND 23 THIS POSITION WAS AT 205 mm.

He1ght:ibove floor ( rn;n l 180

*

-675 - - -875 975 1 085 - - - - -1 315 1 570 1 795 1 875 -1 930 ---

-Figure 2.3a. Positions of pitot tubes and thermocouples along the vertical centerl ine of the doorway.

10

Ooorway

seen

from outs

ide

r

11,12 13

,J

~~ TUBE FOR Height above floor (mm)

l

50 00

J

GAS SAMPLING 200

I

350 180

*

--

₃₇₀ 570 ~ - - -675

-

-

-+

-·

875 975

I

1 085 1 315

I

1 570 1 720 1 795

I

,-~ 1 875 1 930 1 970 D = PITOT TUBE X

=

THERMOCOUPLE

*

FOR CORNCAL 22 AND 23 THIS POSITION WAS AT 205 mm

Figure 2.3b. Positions of pitot tubes and thermocouples

on a vertically movable arm for measuremen'

of horizontal distribution of flow rates and gas concentrations.

28 27 26 25 24 23 22 21 20 19 18 17 16 15 / x

=

THERMOCOUPLE (han no 15 16 17 < - -- · 18 19 20 21 22 ~ --23 · -24 25

-26 ' - -· 27 28 Height above floor (mm) 170 670 - - -970 1 120 1 270 1 4 20 1 5 70 1 720 1 970 - ~ - -- -2 100 -- -2 -200 2 250 2 300 2 350

Figure 2.4. Positions of thermocouples in the center

of the room (the central tree).

12 60 60

Doorway

60 60 90 90 90 90 Ceiling

I

I

I I

I

I

-

r-

-

7~

_

--

-- - ~ 884

-

Ooorway

120

I

-I

,r-

l

90 90 . . F'loor Figure 2.5. (Over)

90 180

1

-

- - -)<' - - - - _{- -} ~ i,_ 60 -;~- _{- c - - - -}

t

-

~~9)

_

-

b11

_

12 0

I

_I

'

-

--t

4(

:8)

_

--

-

-·

f:140

_

)

___

-Side walls. Channel numbers in brackets

refer t o t he left side wall as seen from

the doorway. 30 90 60

r

I

60

r

~

~2~

_

µi_

_

60 55 60 60

r~i---

-

-

-

i

60 60

r--

~t-

53

Rear wall Front wall

13

120

Figure 2.5. Positions of instruments seen from the inside

in the ceiling, the floar and the walls. Legend: x

=

surface thermocouple, 0

=

t hermo-couples in cylinder, ~

=

radiometer type "Gunners" and 0

=

"Medtherm" heat flux meter. Channels 81, 85 and 86 are "Medtherrn" meters for total heat flux. Channel s 80 and 84 are for radiative heat flux only (sapphire window) .

14 7 .~ 1 2 3 4 S 6 - - - - -* -X X- X -X- . X - - - . - -</J SO mm POSITION 1 2 3 4

s

6 7 - -f - - -· - - -

-REAR -WALL 63 64 65 66 67 68 69 CEI LING S6 S7 S8 59 60 61 62

Figure 2.6. Cylinder, diameter 50 mm and length 150 mm, with thermocouples for temperature measurements

15

3 TEST RESULTS

Four tests denoted Corncal 20-23 were conducted. Power

.

input (heat release rate of burner), Q , burner diameter C

and position of thermocouples in the doorway were varied

according to table 3.1. Each test was run more than one hour. At that t ime stabilized conditions for the gas flow

and temperature were established, although the temperature

within the surrounding structure was stil l rising, see

figure 3.1 and 3.2. Observe that time zero is notat

ignition but when measurements were started.

Table 3.1 - List of tests

Test Corncal 20 21 22 23 3.1 125 250 250 125 Pitot tubes position out side door-jamb Il Il inside door-jamb Il Il

Temperature and Heat Flux

Burner diameter (cm) 30 40 40 30

Measured temperature and heat flux for the test s Corncal

20 and 21 are given in table 3.2 and 3.3, respectively.

The measured radiation values were deemed erroneous.

(Wat er probably condensated on t he sapphire window of the

Medtherm heat flux meters.) These results are not given for t he other tests as they are identical to the first

TEMPERATURE

C

CJ

400 300 200 100 0 0 20 40

TEMPERATURE

CCJ

400 300 200 100 0 0 10 20 / 60 80 TIME CMINJ 30 TIHE 40 CHINJ / ~,HUT OFF DL:l l L' l H.lNGE OF GAS Bu r '::' 100 50 60 J l>

l

I

I

120 70

Figure 3.1. Temperature versus time in ceil ing cylinder

during Corncal 20 and 21. Observe that time

zero is notat ignition. The surface

thermo-couples shows the highest temperature and the

others follow consecutively having positions as

TEMPERATURE CCJ 17 400 300 200 SHUT OFF OUE TO / CHANGE OF GAS BOT r: CC -100

---=

====:::::::=:====~

~~~======

=

-~

~

I I 0

-

,

I 0 20 40 60 80 100 TIME CMINJ TEMPERATURE CCJ 400 300 200 100 0 0 10 20 30 40 50 60 70 TIME CHINJ

Figure 3.2. Temperature versus t ime in wal l cylinder

during Corncal 20 and 21. Observe t hat t ime

zero is no t a t igni tion. The sur face thermo

-coupl e shows t he highest t emperature and the

others follow consecut ively having posit ions

18

Tablc 3.2 - Test results from Corncal 20

Temperature Temperature Radiant heat flux

no (OC) no (OC) no (kW/m2) 1 20 31 173 80

_*

2 21 32 202 82

_*

3 21 33 200 83

_*

4 23 34 196 84

_*

5 40 35 235 6 62 36 243 Total heat flux 7 139 37 233 no (kW/m2) 8 202 38 93 9 210 39 178 81 3.0 10 193 40 173 85 3.3 15 30 41 229 86 5.5 16 66 42 100 17 66 43 174 18 66 44 185 19 78 45 229 20 108 46 119 21 181 47 563 22 211 48 515 23 222 49 194 24 233 50 300 25 233 51 194 26 233 52 108 27 233 53 108 28 233 54 181 29 185 55 187 30 173 70 104

19

Table 3. 3 - Test results from Corncal 21

Temperature Temperature Radiant heat flux

(OC) (OC) 2 no no no (kW/ m ) l 26 31 296 80

_*

2 28 32 330 82

_*

3 29 33 325 83

_*

4 35 34 325 84

_*

5 76 35 372

6 119 36 392 Total heat flux

7 260 37 381 no (kW/m2) 8 283 38 186 9 304 39 300 81 9.3 10 290 40 300 85 8.4 15 48 41 370 86 12.6 16 120 42 200 17 120 43 300 18 131 44 325 19 162 45 375 20 247 46 234 21 314 47 647 22 331 48 600 23 345 49 342 24 355 50 450 25 355 51 371 26 355 52 203 27 355 53 212 28 355 54 300 29 300 55 450 30 293 70 196

20

in the doorway. Gas temperature distrib itions at the central t ree are given in figure 3. 3 for Corncal 20 and 21. Tcmperature profiles measured at the center of the doorway are shown in figure 3.4.

Temperature is shown without correction for radiation.

Near the neutral layer, where the gas velocities are small, the radiation correction was calculated according to

sect ion 2.3 and was found to be in t he same order of magnitude as the temperature rise. Away from the neutral

plane the gas veloci ty increases and t he radiation

correc-tion reduces rapidly to typical ly jus t a few degrees in the outflow.

3.2 Mass Flow Rates

3.2.l General Fl ow Patterns

Hot gas is lighter then gasat ambient temperature. There -fore t he hot gasat the upper part of the fire compartment

is driven out by the gravity forces and is replaced by cool

air from the surroundings. Figure 3.5 shows schematically

a side-view of the flow pattern in the room. Note how

the height of the neutral plane, ZN, increases when going

outward from the compartment. The difference in t his height was approximately 8 cm when measured at inner

and outer side of the door-jamb at the 250 kW level, which

corresponds toan angle of elevation of about 30°. At the

125 kW level the corresponding angle was measured to be

about 20°. Observe that these values are very rough est

i-mates and are given here just to give an idea of order

HEIGHT

[NJ

HEIGHT

[NJ

2.

5

2

.5

I

2 .0J

I

L

2.0J

I

)

1.5 1. 5 1.0 1. 0 0 . 5 0 .5

0.9

+---r----r---~

0.0

___

r---- ~~ -

-+--0

.

0

100.0

200.0

300.0

400.0

0.0

100.0

200.0

TEWERATIJRE

[CJ

IDPERATIJRE

[CJ

Co rncal 20 Co r ncal 2 1 Figu r e 3.3. Gas temperature distribution at the ce ntral tree during Corncal 20 and 21. H eight of ceiling equal 2.4 m.

300.8

400.0

N I-'

HEIGHT

[HJ

2

.

5

1

I I I I

2

.0

_1.5

_l.0

_0.5

₀

.

0

4,---~--~--~

--0.0

109.0

200.0

309.0

400.0

TEtPERA

1URE

[C

J

Corn cal 20 Figure 3. 4. (Over)

HEIGHT

[HJ

2.

5

-t----~

---'---'---

----1-2

.0

_1.5

r

1.0

0.5

0.0-t----.----

-~-

---1-0.0

109

.0

208.0

309.0

400.0

IDPERATURE

[CJ

Corncal 2 1 N N

HEIGHT [HJ HEIGHT [HJ

2.5

+--_

__._

__

_...___

__

__.__

_----t-2.

5---~-~--~---2

.0

I

2

.0

1.

5

1. 5

1.0

1.

0

0.

5

0.5

0.0

---,---~

--~---t-0

.

0

----~-

~---0

.

0

100.0

200.0

308.0

400.0

0.0

1

00.0

200.0

308.0

400.0 IDffRATURE [CJ TEl'PERATURE

[CJ

Co rncal 22 Cornc al 23 Figure 3.4. Gas te mperature distr ibution i n the doorway du ring Corncal 20 -23. Height of doorway e qual 2 . 0 m. N w

--~

/ CEILING JET

I

_I

_I

PLUME I

I

FLAME

'

I

_'I

--'

.

\

mp

\1,

\1

mt~'

Zo

Tco

Figure 3.5. Two layer mod el.

Tt

---L ~ mi L ~ ~/

ZN

t\J ,i,.

25

Figure 3.6 shows schematically the horizontal flow pattern

near the doorway. Upstream the contraction, almost

irrota-tional or potential flow can be expected while down-stream

the contraction a flow jet is developed. As the velocity

is proportional to the distance between the streamlines,

figure 3.6 implies that the velocity upstream the door-jamb

is very high near the edges, while down-stream the

vela-city is more evenly distributed.

POTENTIAL FLOW

JET STREAM

Figure 3.6. Streamlines in the doorway showing potential

flow upstream and a jet stream after the

orifice.

The general flow pattern as described above suggests

that i t is not possible to get an accurate estimation of

the total mass flow rates in and out of the fire com

-partment by using measurements at the centre of the door

-way onl y. Consequentl y i t has been experienced here that

the flow rates in and out of the compartment calculated

using measurements at the center of the doorway only may

26

Moreover although the test set-up was symmetrical an

unexpected whirl in the test compartment as shown in

figure 3.7 was discovered when analyzing the data after

the tests. This whi rl was superi nposed on t he flow pattern

and made t he calculation of total mass flow rates in and

out the compartment uncertain as measurement s were taken

at one side of the center of the doorway only.

-

-

_-~

...

--

_-

_-

-Figure 3.7. Schemat ic view of hori zontal whirl in the

27

3. 2. 2 Measured vertical distribution

The vertical distribution of the mass flow rates and gas

velocities as shown in figure 3.8 and 3.9, respectively, are

the weighted average values over hal f the door width. The

flow on the other half i s assumed equal and symmetrically

distribut ed. The flow between measuring points is assumed

to vary linearly. At the door edge surface the flow rate

is assumed to be zero.

3.2.3 Measured horizontal distribution

Measured horizontal profiles of gas mass flow rates at

various height s at the outer (Corncal 20 and 21) and

the inner (Corncal 22 and 23) sides of the door-jamb are shown in figure 3.10.

3. 2. 4 Mass balance

By using the vertical mass flow rate distributions as described in section 3.2.2 the total mass flow rates

in,

m.,

and out,

m,

of the fire compartment are

cal-l 0

culated by integrating over the height.

The height of the neutral plane, ZN, is obtained by l inear interpolation; the results are shown in table 3.4. The calculated mass flow rates into the compartment are

10 to 15 per cent greater than the flow rates out for

all the four tests which probably depends on the whirl

and the assumption of symmetri c flow as described in

section 3.2.1. The average values of the total mass

flow rates,

m,

at similar test deviates, however, less

HEIGHT

[HJ 2. 0 +-__L____j___l._.L-+---'--___.__.___-:---r

1.5

1.0

0.5

0. 0-1---~~ --r---r-l----r----.---.----.---r

-2.5

0.0

2.5

RATE

OF

HASS

FLOM

[KG/S/tr2J

Corncal 20 F igure 3.8. (Over)

H

UGHT

[

HJ

2.0

4--_.____._----L_

~

-~~~~~-;-1.

5

1 . 0

0.5

0. 0-+-~~

r.1~~..+

-2.5

0.0

RATE

OF

HASS

FLOM

[KG/S

/tr2

J

Corncal 21

2.5

N CJ

HUGHT

[HJ

2. 0-1-- --L---'---'- --'----+---'---'--'---'---1. 5 1. 0 0 . 5

I

0.0--~~~-~-~~--~--2

.5

0.0

RATE

OF

HASS

FLOM

[KG/S/tr2J

Co rncal 22

2.5

H

EIG

HT

[HJ 2. 0-t-- --'-- ---'---''--'---+---1...-L___._!___L_----+-1.5 ₁.0 _{0.5 0.} 0-r-- -.---.----.----r ---, l--.---.--r--r---+

-2.5

0

.

0

RATE

OF

HASS

FL

OM

[KG/S/tr2J

Corncal 23

2.5

Figure 3 . 8 . Ma ssflow d istr ibution in the doo rway during Corn cal 2 0 -23. tv ID

tuGHT [HJ 2_0 ___ ...__ __ ..._ _ __. __ _ 1.5 1.0

0

.

5

0.0

I

I

I

I

I

-4

.0

-2.0

0.0

2.0

4.0

VELOCID

QVSJ

Corncal 20 Figure 3.9. (Over) tuGHT [HJ 2.0 ---~---~---r 1.5 1.0

0.5

0.0

-4.0

-2.0

0.0

2.0

4.0

VELOCID

[11/SJ

Corncal 21 w 0

HEIGHT

[HJ

HEI

GHT

[HJ

2.0,

__

.!__ _ _

f-_

____l __

---1-2.0

__

_

...____--l---'--- --+-1.5 1. 5 1.0 1. 0

0.5

0.5

0.04----~---+---.---+- 0.0;---.--- --+--

.+

-4.0

-2.0

0.0

2.0

4.0

-4.0

-2.0

0

.

0

2

.0

4.0

VElOCID

[H/SJ

V

ElO

C

ID

[

H/

SJ

Corncal 22 Co rncal 2 3 F igure 3.9. Gas velocity distrib ut io n i n th e doorwa y d u r ing Co r n ca l 2 0 -23 . w f--'

RATE OF MASS FLOW [ kg/m2s J

3.0

9 . 10 . 11, 12 . 1 4, 13

I

1-'Ij I-'·

2

.

0

I

lQ ~ rj (l)

1~

w 3

.

C: ,0 I-' I.

0

l

i

0 •z 0

0

0

1

~

1~

<

(l) rj

I

I

I

I

~ i

I

-I

.0J

~f

-1

'

1.2 . 3,4,5.6. 7

I

I

-2

.

0J

I

-3.0-+---~----<

0.0

10.0

20.0

30.0

40.0

D ISTANCE FROM DOORJAMB [ cm J Corncal 20 RA TEOF MA SS FLOW [k g / m2sJ

3.0

I

I 9. 1 0, 11.12.13. 14

2.0

~

+

-~ 3 . C: 1 8

1.0

/8 . D I lL 0 C: w . ...

I

~

0.0

I

-1.0

_-

2.0J

1.2. 3, 4_._5__.~. 7

-3.0

-t---r

-

r-

---.---1

0.0

10.0

20

.

0

30.0

40.0 OIS T ANCE FROM DOOR JA MB [ cm J Corn cal 21 w N

Vl N E 3 0 _ J lL V) V) <( L lL 0 w

~

Vl N E 3 0 _ J lL V) V) <( L lL 0 w ~ 0:: ;::! ,, .. ; a,· Wl'\1:JOOO :JO 1l31N3J a:, \

\

).'J'Ml:!000 :JO l:J31ND a:, -

\

J

"5:> <n I "5:> <n I E u co L <( -, 0:: 0 0 D L 0 0:: lL w LJ z ~ V) D E u co L <( --. 0:: 0 0 D L 0 0:: lL w LJ :z: ~ 1/l D N N rl ro {) i:: H 0 u 33

Figure 3.10. Horizontal mass flow distribution in the

doorway during Corncal 20-23. The flow is

given at levels numbered 1-14 starting from

below and going upwards as given in figure

34

Table 3.4

-

Total mass flow rate through the doorway

and height of neutral plane , m = (in.

+m )

12.

l 0

.

Test _Qc Measuring m. m m _ZN

l 0 Corncal (kW) points (kg/s) (kg/s) (kg/s) (m) 20 125 outside 0.78 0.70 0.74 1. 20 21 250 outside 1. 01 0.89 0.95 1.17 22 250 inside 0.99 0.87 0.93 1. 09 23 125 inside 0.81 0.69 0.75 1.14

Table 3.5 shows the calculated mass and energy flows in

t he doorway based on center line measurements only. The

values using several measuring points divided with the

corresponding center line values are given in parenthesis.

Note that these values varies considerably at both sides

of unity.

Table 3.5 - Mass and energy flow rates in the doorway

based on centerline measurements only. Figures

in brackets are the inverse ratio of these

values and the corresponding values using

several measuring points at each level .

Test Corncal 20 21 22 23 m. l (kg/s) 0.58 (1.34) 0.76 (1.33) 0.86 (1.15) 0.75 (1.08) mo (kg/s) 0.83 1. 07 0.91 0.75 (0.84) (0.83) (0.96) (0. 92) ...!... 113 ( 0. 9 2) 233 (0. 89) 214 ( l . 02) 119 (0.99)

35

3.3 Heat Flow Rates

-

---

---The energy released by combustion leaves the room by

convection and radiation through the doorway, and by conduction into the surrounding structure. The heat

ba-lance of the room can therefore be expressed as

where

=

=

( 3 . 1)

heat released by combust ion

heat loss by convection through the doorway

heat loss by conduction into the surrounding

structure

heat loss by radiation through the doorway.

As nearly quasi steady-state conditions are studied, the

rate of heat stored in the gas volume is negligable.

The object of the analysis was to estimate the accuracy of the measurements by evaluating and inserting the energy

terms in equation (3 .1).

3.3.1 Heat_Release_Rate_by_CombustionL_Qc

The gas to the burner was provided from gas bottles which

were placed on scales measuring the weight loss and thus

the burning rate mf. The rate of heat release was

cal-culated as

Q

= 6H mf where 6H (=46.4 MJ/kg) i s the net

C C C

36

3.3.2 Heat_Loss_Rate_by_ConvectionL_Q

0

At steady state burning the surrounding structure gradual ly

heats up and the heat loss decreases. The gas temperature

increases, however, and thereby the heat loss by

convec-tion out t he doorway. As the measurements were not taken

unt il approxirnately one hour after ignition the heat loss to the walls were substantially reduced. The heat loss

rate by convection at this state was t herefore as much

as 80- 90 per cent of the total heat release rate, see

table 3.6.

Table 3.6 - Heat balance of fire compartment. Nurnbers in

brackets are heat loss terms norrnalized to the

heat released by the burner

Q

.

IQ

=

Q

+

Q

+Q

C O W r

.

Test Qc _Qo Corncal (kW) (kW) 20 125 104 (0.83) 21 250 208 (0.83) 22 250 218 (0.87) 23 125 118 (0.94)

.

Qw (kW) 19 ( 0 .15) 32 (0.13) 32 (0.13) 19 0.15)

.

Qr (kW) 6 (0.05) 12 (0.05) 12 (0.05) 6 ( 0. 05)

I

Q

(kW) 129 ( 1. 03) 252 (1.01) 262 (1.05) 143 (1.14)

37

The heat flow in the doorway was calculated as the

product of mass flow, temperature rise and specific heat

capacity. The latter assumed equal to that of air. By

then integrating over the entire doorway the total heat

.

loss by convection Q _-o was obtained. The data obtained

by the pitot tubes and thermocouples mounted on the

movable arm as well as the center line array was used

for calculating

6.

0

3.3.3 Heat_Loss_Rate_by_ConductionL_Qw

The heat conduction to the surrounding structures was

calculated as where Q" w k =

=

Q"

=

w

heat conduction per unit area (W/m2)

thermal conductivity (W/mK)

temperture gradient at the surface (K/m)

The thermal conductivity of lightweight concrete of density

3

500 kg/m was assumed equal to 0.15 W/mK which is

approp-riate for temperatures near 250°c; the conductivity varies

l inearly between 0.12 and 0.20 W/mK in the range of zero

to 50o0c /8/.

The t emperature distribution was measured with several

thermocouples in the cylinders placed i n the ceil ing and

the r ear wall (see section 2.3), and t he gradient at the

38

surface temperature gradient at other places of the

sur-rounding structure was estimated by assuming that the

temperature distribution is similar and that the gradient

is proportional to the temperature rise. As the surface

temperature was measured at several points a good esti

-mation of the heat loss by conduct ion could then be

achieved.

The total heat loss by conduction to the surrounding

struc-ture was obtained by adding the contribution from a

number of subareas with assumed constant heat flux. The

values of

Q

when calculated at t ime of analysis appr

oxi-w

mately after one hour of constant burning was 19 and 32 kW

for burning rates of 125 and 250 kW, respectively. The

errors of these results are assumed to be in the magnitude

of 10-20 per cent.

3.3.4 Heat_Loss_Rate_by_RadiationL_Qr

.

The radiation loss out the doorway, Q was calculated by r

adding the contribution from a number of smaller areas

of the surrounding surfaces:

where 0 € . l A. l F. l T

=

=

= = = = o L i - 4 s . A. F. T. l l l l

the Stefan - Bolzmann constant

(=5. 67 X 10-S W/m2 K4)

emissivity ( )

2

area (m)

view factor ( )

The assumptions of uniform temperature of each surface and emissivity equal to unity was made. As the flames were optically thin their contribution to Q was

neg-r

lected in this rough estimation.

39

Thus calculated radiation out the door-way was 6 and 12

kW for the experiments with the heat release rates of 125 and 250 kW, respectively. The inaccuracy of these values are probably in the magnitude of 10 per cent.

3.3.5 Heat Balance of Fire Room

Fire heat loss components of the fire room measured in

the four tests are summerized in table 3.6. The sum of

them are al l greater than the heat release rate of the

burner but, except from Corncal 23, less than 5 % greater. 3.4 Gas analysis

The measurements of

co

₂ concentration in the doorway

revealed that combustion products exi sted only down to the measuring l evel of 1.570 m from the floor; at the

level of 1.315 m no

co

2 was detected. As t he neutral plane was in these experiments considerably lower, a

zone of gases heated but not contaminated with combustion

products existed in the outflow. No carbon monoxide was detected implying complete combustion of the propane gas.

40

4 THEORETICAL MASS FLOW

The data of the fire induced air flow were analyzed by

assuming hydrostatic pressure distribution in the enclosure,

and an orifice flow model at t he doorway. A temperature

distribution both according to the generally accepted two layer model /9-11/ and according to the measured

value at the center of the room was assumed. Orifice di scharge coefficients were determined by comparing

calculated and measured data. In addition a plume flow

model /10/ was applied and the entrainment coefficient

was determined.

4.1 Two Layer Model

In the two layer model as shown in figure 3.5 the air is

assumed to be at ambient temperature T₀₀ below a certai n

height,

z

0 , and at a uniform elevated temperature, Tf,

above

z

0. The neutral (zero flow) plane at the opening

i s a t a different height ZN. Conservation of mass relates

heights of ZN and

z

0 . In the present case fuel flow is

only a fraction of a per cent of the doorway flow and

can be neglected. The governing relationship is then as

given by Racket /10/. where N

=

D

=

D (1--) N ( 4 .1)

=

normalized neutral plane height

=

normalized thermal discontinuity

R

₌

=

doorway height (m)

=

ratio of absolute upper to

lower gas temperature

and the flow equation is

where

.

m C g mass flow (kg/s) discharge coefficient ( ) ambient air density (kg/m3) doorway area (m2) 2

acceleration of gravity (m/s )

( 4 • 2)

41

To solve for flows from basic principles a third relation

-ship (the plume equation) would be needed that gives either

N or D. In the present series measured values of N are

available, anda value of D can be calculated and checked

for reasonableness. Direct assessment of Dis, however, hard; as is readily seen in figure 3.3 the vertical te

m-perature distributions, while definitely indicating hot

upper and cold lower zones, are not sharp enough to

accurately assign a thermal discontinuity height.

The thermal discontinuity height

z

0 solved from equation (4.1) and the ideal mass flow rate m/C according to

equation (4.2) are given in table 4.1. The hot gas layer temperature was then calculated as Tf

=

Q

/m c + T

0 0 p 00

where

6

and

m

are measured values. Discharge cGefficients

0 0

for the flows in and out the doorway was then obtained

by setting the ideal mass flow rates equal to the measured values.

42

Table 4.1 - Calculated thermal discontinuity height,

z

_{0 ,}

ideal hot layer temperature, Tf=Q /m c , and

0 0 p

mass flow rates, ~/C, and discharge coefficients

Test Corncal 20 21 22 23

for t he flow in and out the doorway.

125 1. 09 166 250 1. 06 251 250 0.92 268 125 0.99 188 m/C (kg/s) 0.94 1. 05 1. 21 1. 07 C. l (-) 0.67 0.85 0.72 0.56

The results show outflow coefficients C ranging from

0 C 0 (-) 0.83 0.96 0.82 0.75

0.75 to 0.96 and inflow coefficients from 0. 56 to 0.85.

The average of C is 0.84 and the average of C. is 0.70.

0 l

Li tterature values /9/ give dischargecoefficients based

on pipe-flow technology equal 0.68. Tu and Babrauskas

/12/ report values of C. and C as low as 0.56 and 0.66,

l 0

respectively. Their values as well as those reported in

/9/ and here all show C. < C .

l 0

4. 2 Arbitrary Temperature Distribution

Instead of using the two layer model with discrete

tem-perature zones for calculating the hydrostatic pressure,

measured values of the gas temperature in the fire

com-partment may be applied / 13/. The hydrostatic pressure difference, interior minus exterior, referenced to the neutral plane and assuming an ideal gas is then given

6p(z)

=

g p ₀₀ (z - T _(X) 2 dz f T ( z)) 0 43

where z is the height coordinate. The theoretical gas

velocity is then (26p/p)1/ 2 where p is calculated using

air at the local temperature.

Using the gas temperature measured in the center of the room (see figure 3.3) and integrating numerically above

the neutral layer gives the theoretical mass flow rates

out the compartment as 0.92 and 0.97 kg/s for heat release

rates of 125 and 250 kW, respectively. (The flow below

the neutral layer was not calculated as the corresponding

temperatures are uncertain due to radiation from the f ire plume.)

Corresponding discharge coefficients are 0.76 and 0.91. These values are higher than what is reported in the

litterature based on pipe-flow technology theory (~ 0.7).

Steckler et al /14/ who measured the t emperature dist

ri-bution in a stagnant zone near the doorway reported values averaging 0.75 and within + 0.04.

4.3 Plume Flow

In simple two layer room fire models as described in

section 4.1 the gases from the hot upper layer and the

cold l ower layer are assumed not to mix along their in

ter-face except at the source plume. During steady state conditions all of the compartment outflow should equal

the plume flow where it intercepts the hot gas layer. The

44

( 4 . 3)

where subscripts f refers to the source (burner) con

di-tions. The fuel constant w is defined as / 13/:

Moof l\Hc

w

=

M

I

(l + rc T)

00 p 00

where M

00f/M00 i s the ratio of the molecular weight of the

flame gas to that of air; AH is the heat of combustion;

C

r is the air- fuel weight ratio and c the specif ic heat.

p

In the calculations reported here M ₀₀ f/M ₀₀ was assumed

equal to 0.824 which corresponds to burning at stochiometri;

c was assumed equal to 1.11 kJ/kg K. The propane pro-p

perties L'IH

=

46.4 x 103 kJ/kg and r

=

17. 2 then gives

C

w

=

0. 088 for T

00

=

290 K. The height

z

0 should probably

be corrected by substracting the burner distance above

the floar and adding a virtual sour ce distance to account

for finite burner area /12/. Since the latter i s not

well known and since the corrections are of opposite sign,

neither correction has been appl ied. Equation (4.3) can

now be solved for the only unknown value, k , the entrai n-e

ment coefficient .

Hence k i s equal to 0.20, 0.23, 0. 27, and 0.23 for Corncal

e

20-23, respectively. The plume flow,

m,

was then assumed

p

equal to the average mass flow rates in the doorway (see

table 3.4). The position of the burner close t o a wall

was neglected in the analysis (Zukoski et al / 15/ has

shown that a wall placed adjacent t o a circular burner

has negligable effect on the plume entrainment) . The

ob-tained values of k are considerably lower then the k

=

0.56

e e

reported by Tu and Babrauskas /12/ but higher then the

k ~ 0.16 obtained when regions of a plume far away from

e

the burner are considered /16/. Mc Caffrey and Rocket /13/

got values of k

=

0.37, 0.32, and 0.26 for the center

e

burner, wall burner, and corner burner configurations , respectively.

45

This large discrepancy of k _e demonstrates that the plume

flow model is a weak link in the room fire analysis.

A superimposed wall flow phenomenum that would reduce the

importance of the plume flow and thereby t he magnitude

of t he entrainment coefficient as calculated above will

be discussed in the next section.

4.4 Wall Flow

The high values of the plume entrainment coefficient

obtained from analysis of compartment fire using equation

(4 .3) may be explained by the upward flow that develops

along the walls when t hey get heated by radiation from the

flames and the hot upper gas layer. As the combustion

products are concentrated to the upper quar ter of the

doorway height (the neutral plane lies approximately at

half of that height) and as the heat flux by conduction

between the gas layers i s small, the intermediate layer

46

5 SUMMARY

Calibration tests of a fire room with inert wall s have been conducted with a propane burner <luring steady state conditions. Temperature, t hermal radiation, gas flow, CO

and

co

2 concentrations have been measured and theoreti -cally calculated values have been discussed and compared

with test results.

Since gas flow rates through an opening in a room containing a fire represent a key element in the f i re-growth process, it is important to descri be these f low rates accurately. A thorough survey was therefore carried out in the doorway. Velocity and temperature were measured at 64 positions

in the doorway along the center l ine and at several other points. Heat and mass flows through the doorway were

calculated by integration over the openi ng, and heat and

mass balances for the entire compartment were then analyzed.

It was found that the total heat loss terms matched well

with the heat input from the propane burner.

A larger discrepancy was, however, found when comparing measured rates of mass flow in and out of the compart -ment. Although the experiment s were conducted indoors,

it was found from measured wal l temperature t hat a steady

horizontal whirl had existed. Thus the neutr al plane in

the doorway was not hori zontal and the assumption of

symmetric flow patterns was not quite correct . The presence

of such a whirl complicates the measurements of mass flows in and out the compartment .

For estimating heat release rates in a standard test method i t is desirable to complete t he measurements with oxygen depletion analysis of the exhaust gases. A system

for such measurements including a hood for collecting

exhaust gases is therefore being developed and wil l be

47

In the theoretical study of the fire compartment a static

pressure model was used to determine opening flows from

the temperature profile data. Two methods were used to

establish these flows. One method util izes the assurnption

of a hot upper layer with uniform temperature defined as

the weighted average temperature of the fire gases in the

doorway. The neutral plane was obtained by interpolation

of the gas velocity measurements. These two quantities

were then inserted inta the two-layer pressure equation

according to Racket /10/ to determine the ideal flow rates.

The other method involved an integration over the opening

height of the differences between the static pressure

profiles at each side of t he opening. Pressure profiles

were determined from the vertical t emperature profil e

in the cent er of the room via the ideal gas law. The

neutral-plane height was obt ained by interpolating t he

experimental mass flow rate data. Mass flow rates obtained by these two theoretical methods and by measurements of

temperature and velocity were then compared.

A plurne flow theory was applied and entrainment

coeffi-cient s were calculated by cornparing with measured mass

flow rates out t he doorway. High values was then obtained

which al so has been reported by others who have done

similar room fire experiments. An explanation may be

that the f low upward the walls is neglected by the two

layer model s applied. This theory i s supported by the

fact that combustion products were concentrated to t he

upper quarter of the doorway height while the neutral

REFERENCES

/1/ Pettersson, 0., "Firc I-Iazards and the Cornpartment Fire Growth Process - Outline of a Swedish JoinL

Research Program'' . FoU-brand, No. 1, 1980, or

Department of Structural Mechanics, Lund InslitulL:

of Technology, Report No. R 80-5, Lund 1980.

48

/2/ Williarnson, R.B. and Fischer, F., "Fire Growth

Experiments - Toward a Standard Room Fire Test" ,

HSS-79-48, Department of Civil Engineering and

Lawrence Berkeley Labora tory, Uni vers i ty of Cal i fonu,,,

Berkeley 1979.

/3/ ISO, "Fire Tests - Building Material s - Corner Wc111/ Room Type Test", ISO/TC 92 N581, 1981.

/4/ Sundström, B. and Wickström, U., "Fire : Ful l Scale

Tests - Background and Test Arrangements", National

Testing Instit ute, SP-RAPP 1980:14, Borås 1980, or ISO/TC 92/WG 2 & 4 N200 N388.

/5/ Andersson, B. and Magnusson, S.E. , "Brand i stopp

-möbler - en experimentel l studie", Department of

Structural Mechanics, Lund Institute of Technology, Report No. R80- 4, Lund 1980 (in Swedish) .

/6/ Gunners, N.E., "Methods of Measur ement and Measuring

~quipment for Fire Tests'', National Swedish Testinq Institute for Material Testing, Fire Engineering

Laboratory 1967:1, 1967.

/7 / McCaffrey, B. and Heskestad, G. , "A Robust Bidir8•.:

t ional Low - Velocity Probe for Flame and Fire

Application", Combustion and Flame, Val 26, 125-12,,

49

/8/ Brand- och Rökspridning längs fasader och i

vcntjla-tionskanaler, delrapport 2, Stockholm 1970 (in

Swedish) .

/9/ Prahl, J. and Ermnons, H.W., "Fire Induced Flow t hrough

an Opening", Combustion and Flame, Vol 25, 369-385,

1975.

/10/ Racket, J .A. , "Fire Induced Gas Flow in an Enclosure",

Combustion Science and Technology, Vol 12, 165-175,

1976.

/11/ Quintiere, J.G. , "Growth of Fire in Building

Com-partments", Fire Standards and Safety, ASTM STP 614,

131-167, 1977.

/12/ Tu, K.-M. and Babrauskas, V. , "The Calibration of

a Burn Room for Fire Tests on Furnishings", Nat.

Bur. Stand. (U.S.) Tech. Note 981, 1978.

/ 13/ McCaffrey, B.J. and Racket J.A., "Stat ic Pressure

Measurement s of Enclosure Fires", Journal of Research

of the Nat. Bur. Stand., Volume 82, No 2, September

-October 1977.

/14/ Steckler, K.D., Quintiere, J.G. and Rankinen, W.J. ,

"Fire Induced Flows through Room Openings - Flow

Coefficients", to be present ed at the Eastern Stat es

Combustion Institute Meeting, November 12-14, 1980.

/ 15/ Zukoski, E.E. , Kubota, T. and Cetegen, B., Entrainmen~ in Fire Plumes", NBS-GCR 80- 294, 1981.

/16/ Tamanini, F. , "An Improved Version of the k-e-g

Model of Turbulance and Its Application to Axisyrrunet·

ric Forced and Buoyant Jets", Report 22360-4,

RC77-BT-4, Factory Mutual Research Corp., Norwood