,
STATENS
PROVNING
SANSTALT
Nation
al
Tes
ting I nstitute
:
/ - -
CEILING JETFire Technology
\\
I
/
/ / ' / / /
PLUME / ' / I II
FLAME /\I
- - -
---_.;'
II
\1,
~"-BURNERBjörn 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 00BJÖRN 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: 48I 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 factoracceleration 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/ fordeve-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 IFront
wiew
I II
II
l 1--r-1-C ) C) "'Figure 2.1. Fire test room.
240
1
---
r
0
BURNER I Il
80l
".,
Plan wiew
Measures in cm C) '° rn2. 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 2D
= PITOT TUBEX
= 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 00J
GAS SAMPLING 200I
350 180*
--
370 570 ~ - - -675-
-
-+
-·
875 975I
1 085 1 315I
1 570 1 720 1 795I
,-~ 1 875 1 930 1 970 D = PITOT TUBE X=
THERMOCOUPLE*
FOR CORNCAL 22 AND 23 THIS POSITION WAS AT 205 mmFigure 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 350Figure 2.4. Positions of thermocouples in the center
of the room (the central tree).
12 60 60
Doorway
60 60 90 90 90 90 CeilingI
I
I II
I
-
r-
-
7~
_
--
-- - ~ 884-
Ooorway
120I
-I
,r-
l
90 90 . . F'loor Figure 2.5. (Over)90 180
1
-
- - -)<' - - - - - - ~ i,_ 60 -;~- - c - - - -t
-
~~9)
_
-
b11
_
12 0I
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
60r
~
~2~
_
µi_
_
60 55 60 60r~i---
-
-
-
i
60 60r--
~t-
53Rear 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 62Figure 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 40TEMPERATURE
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 70Figure 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 CHINJFigure 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 10419
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 3726 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 .0JI
L
2.0J
I
)
1.5 1. 5 1.0 1. 0 0 . 5 0 .50.9
+---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 I2
.0
1.5l.0
0.50
.
0
4,---~--~--~
--0.0
109.0
200.0
309.0
400.0
TEtPERA1URE
[C
J
Corn cal 20 Figure 3. 4. (Over)HEIGHT
[HJ
2.
5
-t----~
---'---'---
----1-2
.0
1.5r
1.0
0.50.0-t----.----
-~-
---1-0.0
109
.0
208.0
309.0
400.0
IDPERATURE
[CJ
Corncal 2 1 N NHEIGHT [HJ HEIGHT [HJ
2.5
+--_
__._
__
_...___
__
__.___----t-2.
5---~-~--~---2
.0
I
2
.0
1.
5
1. 51.0
1.
0
0.
5
0.50.0
---,---~
--~---t-0
.
0
----~-
~---0
.
0
100.0
200.0
308.0
400.00.0
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 JETI
I
I
PLUME II
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 arecal-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, lessHEIGHT
[HJ 2. 0 +-__L____j___l._.L-+---'--___.__.___-:---r1.5
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 . 00.5
0. 0-+-~~r.1~~..+
-2.5
0.0
RATE
OF
HASS
FLOM
[KG/S
/tr2
J
Corncal 212.5
N CJHUGHT
[HJ
2. 0-1-- --L---'---'- --'----+---'---'--'---'---1. 5 1. 0 0 . 5I
0.0--~~~-~-~~--~--2
.5
0.0
RATE
OF
HASS
FLOM
[KG/S/tr2J
Co rncal 222.5
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 232.5
Figure 3 . 8 . Ma ssflow d istr ibution in the doo rway during Corn cal 2 0 -23. tv IDtuGHT [HJ 2_0 ___ ...__ __ ..._ _ __. __ _ 1.5 1.0
0
.
5
0.0I
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 1.00.5
0.0-4.0
-2.0
0.0
2.0
4.0
VELOCID
[11/SJ
Corncal 21 w 0HEIGHT
[HJ
HEI
GHT
[HJ2.0,
__
.!__ _ _f-_
____l __---1-2.0
__
_
...____--l---'--- --+-1.5 1. 5 1.0 1. 00.5
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, 13I
1-'Ij I-'·2
.
0
I
lQ ~ rj (l)1~
w 3.
C: ,0 I-' I.0
l
i
0 •z 00
0
1
~
1~
<
(l) rjI
I
II
~ iI
-I
.0J
~f
-1
'
1.2 . 3,4,5.6. 7I
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 / m2sJ3.0
I
I 9. 1 0, 11.12.13. 142.0
~
+
-~ 3 . C: 1 81.0
/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 NVl 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 33Figure 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 doorwayand 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 netC 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"=
wheat 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 approxi-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 lthe 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 doorwayrevealed 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 discontinuityR
=
=
doorway height (m)=
ratio of absolute upper tolower 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) 2acceleration 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 + T0 0 p 00
where
6
andm
are measured values. Discharge cGefficients0 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 43where 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
=
MI
(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 givesC
w
=
0. 088 for T00
=
290 K. The heightz
0 should probablybe 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 assumedp
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
=
0.56e 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 centere
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