Sunteți pe pagina 1din 10

J. Loss Prev. Process Ind. Vol. 9. No. 3. pp.

231-240,

19%

Copyright 0 1996 Elsevier Science Ltd


Printed in Great Britain. All rights reserved
0950-4230/96 $15.00 + 0.00

SO950-4230(96)00014-9
ELSEVIER

Protection
pool fire

of equipment

engulfed

in a

Eul&lia Planas-Cuchi*, Joaquim Casal*, Antonio Lancia? and


Leo Bordignon?
*Centre dEstudis de Rise Tecnolbgic (CERTEC), Department of Chemical
Engineering of the Universitat Politkcnica de Catalunya-Institut
dEstudis Catalans,
Diagonal 647, 08028 Barcelona, Catalonia, Spain
i_Tecsa Ricerca & Innovazzione, via Aldo Moro 1, Scanzorosciate (BG), Italy
Received 1 I February 1996
When process equipment is engulfed in a fire, the protection system should extinguish the fire
and simultaneously cool the equipment. Therefore, the behaviour of equipment engulfed in a
fire, as well as its eventual influence on the fire, should be known. In this article, experimental
data from pool fires of hexane (4 m2) and kerosene (12 mZ) are discussed. The evolution of the
fire up to its full engulfment is studied, as is the evolution of temperatures at different points
of a tank engulfed in it. The heat release rates under the different operating conditions are
studied. A method is developed to calculate the maximum combustion rate, both for the
developing stage and for the fully developed fire. The hindering effect of the tank on the fire is
discussed and quantified in terms of efficiency of hindered combustion. Copyright 0 1996
Elsevier Science Ltd
Keywords: pool fire; foam and water; cooling; extinguishing;

Certain industrial installations are designed according to


an extremely compact lay-out. Off-shore oil platforms
and some sections of chemical process plants exhibit a
high density of equipment which, in the event of certain
incidents, can lead to dangerous situations and can significantly complicate the management of emergency procedures. In such installations, moreover, there are usually large inventories of hazardous materials, and
protection systems with a high degree of reliability and
efficiency must be provided. If flammable materials are
processed or stored, a good fire protection system must
be installed. Such a system should be able to cover two
objectives simultaneously: extinguishing or at least controlling the fire and, furthermore, cooling the structures
and the equipment. This second scope is very important,
in some cases even more than the first one, as the effect
of fire on certain equipment can result in further loss of
containment of flammable substances and hence in the
escalation of the accident scenario.
In order to cover these two objectives there are various possibilities. In the conventional approach, cooling
and extinguishing are addressed by two different systems: water deluge systems and foam systems, the specific features of water and foam thus being used separately.
Water is the most commonly used fire extinguishing
agent; it has a high specific heat and very high latent heat
of vaporization, and furthermore it is usually available at

modelling

a low cost. However, the use of water also has several


disadvantages, the main one resulting from its relatively
high density, higher than that of most hydrocarbon fuels.
Furthermore, hydrocarbons are also immiscible with
water; therefore, in the event of a pool fire water will
neither cover the burning fuel - thus extinguishing or
reducing the fire - nor mix and dilute it. The hydrocarbon will maintain a lighter burning layer, and in some
cases water may even flash (if it enters a hot mass of
fuel) and spread fuel. If a cover for the burning surface
is required - and this will help significantly in the control
of the fire - foam solutions must be used. However,
foam agents are much more expensive than water and,
moreover, they are available in much smaller amounts.
One interesting approach consists, therefore, in
using a single deluge system which discharges foaming
solution for a given minimum time and then continues
discharging water. Such systems, known as foam and
water deluge systems, should discharge a kind of foam
which is fluid enough to behave like water for cooling
purposes
but stable enough to possess useful
extinguishing properties in spill fires. Foam and water
deluge systems are quite advantageous, especially for
off-shore oil platforms and processing plants, due to the
possibility of achieving fast fire control, the simplification of plant structure and operating procedures and
the resulting reduction in the costs of the system.
This work discusses the behaviour of a typical piece
of equipment (a tank) during the development and

231

13adSaJ ~I!M sJalawo!peJ aql40 uo!ysod

alnpow leluaw!Jadxa aql 01


(q) :sgJoa u! sa!Jas Isal ale3s-lln4 aq~ ~04 pasn dn-yoow WJOpeld aw 40 aJnvnJ3s (a) L alnhj

( p!Ml 01 - 0)

PZ J~HUJO!pe~

1 zw/Ml 01 - 0)
gz Jal~wo!P~~

1 $J/MI OZ - 01
LZ Jwawo!pe~

yuei atjl~o ap!s $saM


aql uo pale301 Saldno3oWJaq~ aql 01 pUOdSaJJO3 SasaqluaJed
u! sJaqwnu aql SaldnO3OWJaql yuel 40 Guyo!gsod
2 arn@j

we3

zz

u.

9 16%

PZ

3-M

GLCLL

LZ

QEJauat)

*atug 30 uoypun3 v SE aIdnoDowayl

pun SUIU 11~ 103 pallold

y3ea 103

SBM (s.xaIyuuds aq~ 30 uoyzlado

ayl 01 snoyAaJd Isa] arfl 30 a)nuytu lsly aq %.ynp)


-wadwa~30

uoynIoAa

AqliqicmpoldaJ

aqi spawamwatu

put! A1yqcgal

aq

amI

Itz)uawI.radxa 30

A3paA 01 laplo

aaJy1 ayl 30 uogsod

III

aw pue 11s auofzmouu1

qll~

II? paIpzlsuj slalawolpvl

~1 30 aDy.ms B ~I!M)

-seaw

Iood

I? qii~)

*(amwadwal

asay

aql s~oys

30 szgsua$atznzq~ aqJ

auasolay qiy
auexaq

~I!M

mo3 pm2 @.I p


pawoyad

aJaM

SE%) slu!od pzrrralxa mo3 IV pam

wM amle.wIwai

a~owaqwg

!(z

am8?& gut.3 am
*g

2g amaDw ss3aL - mu

pm

vds vsxm vdsoavs SKYWWIayl4 PadoIaAap


,su1.1o3~ld

110

aloqs33o

PW mod,
u-681~~1
-aww3 aql U~IM

103

SI-HL

uralsAs

IDafoti

383

a&yap

JawM

aw 30 7~0~

paKiallo9 alaM mp ptz,uatuuadxg


7.uoo3 30 lady e

-Inwys pw

(q)l a.4n8!d *lawi Quo 8uIsn auop sm uoymgxa aql 5 1


pm pi sisal II! ~vq~aluasqo 01 wvodw!
s! I! tl a1qv.z
ul paluasa.td as\? sisal @!a
@I

30 aayns
sisal mod

ye la !yzvyweuejd

(a.ry aql qsp&gxa

pw)

11100~ 01 laplo

Ilids aq

laAo3

uy pal~no~d

djsnoaw)

aq 01 luawdyba

ayl 30 am3ms av 01~0 dpDaI!p pinjj &u~ooD aw a&q:,


-sip speaq hds a.xow .IO au0 qD!qM us su~a$sks asoqj
-a*! stualsrCs dwds Java pw ureo3 30 paluasald 0s~ s!
dpnls v .paprAoJd azrr $u ~1 pm $u p %U!.uW?au.I
auas
-olay put! amxaq 30 saIy Iood u10.13paugqo tmp pmar_u
-uadxg *pa3@ua sr oy qD!qM ui aJy Iood a 30 uoyugxa

:aJ!J load e u! paj@ua

juawdinba

40 uogc)aJoJd

ZEZ

Protection

of equipment

engulfed

in a pool fire: E. Planas-Cuchi et al.

233

Table 1 Description of the selected experiments

Test number

(m*)

Pool area
Fuel

Extinction agent

Sprinklers configuration

S.A.R3*
(Lmin-.m-*)

Deluge flow
(I.min-)

12
13
14
15
88
89
90
91

4
4
4
4
12
12
12
12

Hexane
Hexane
Hexane
Hexane
Kerosene
Kerosene
Kerosene
Kerosene

Foam
Foam
Water
Water
Foam
Foam
Foam
Foam

two
two
two
two
two
two
two
two

14.2
14.2+
14.2+
14.2
14.2+
14.2+
6.5
6.5

246
296
296
246
296
296
112
112

Surface

application

rate

speaking, the results for the hexane pools (4 m2) showed


a practically linear trend, essentially identical for the first
three tests; run number 15 showed slightly higher values
with respect to the rest of the results, although the trend
was similar. For the kerosene pools the results were
parabolic rather than linear; this can be attributed to the
fact that the development of fire over the whole surface
of the pool (in this case 12 m2) was slower.
The data obtained from the radiometers have been
used to plot the variation of radiation intensity as a function of time (Figure 3). As can be observed, after
approximately 40 s the fire can be considered to be fully
developed, and with a practically stationary regime (the
time required to reach this condition was more than 60 s
for the 12 m2 kerosene pool fires); typical oscillations
exist, which must be related to the turbulence of the fire
and to the variability of the shape, size and orientation
of flames as a function of time. Later on, the decrease
of radiation intensity shows the effects of the start-up
of the extinction system. The action of the foam-water
mixture caused a significant decrease in the size of the
flames, which finally led to the complete extinction of
the pool fire as it became covered with a stable layer
of foam.

Evolution of temperature
As the fire proceeds, the equipment engulfed in it - a
tank in this work - will undergo an increase in temperature which will be a function of the evolution and size
4.5
4
3.5
T3
E 2.5
Radiometer 26

j2

-Radiometer

27

z! 1.5
1
0.5
0
0

50

SPKOl125
SPK0/125+SPY0/90
SPK0/125+SPY0/90
SPKOf125
SPK0/125+SPY0/90
SPK1/90+SPY0/90
SPK0/125
SPKl/l ?O

100

150

200

250

time (s)
Figure3
Variation of the radiation intensity as a function
time for test number 13 (4 m*, hexane)

of

of the flames, and also of the location of the measuring


point; those zones exposed to flame impingement will
undergo stronger heating than those which are not in
direct contact with them. On the other hand, after the
start-up of the cooling/extinguishing system, different
behaviour will be observed depending on the existence
or otherwise of a water/foam layer on that particular
zone of the equipment surface. The analysis of the
experimental results indicates the existence of five different situations or steps4, which are discussed in the
following paragraphs.

Initial step (development of the jire)


The first step corresponds to the development of the fire
and takes place over the first minute (approximately).
Figure 4 shows the variation of temperature as a function
of time, at diverse measuring points, for two of the tests.
Temperature increases approximately linearly with time,
the slope depending on the location of the point and on
the impingement of the flames on that point. In one
minute, the temperature at the bottom of the tank
increases from room temperature to more than 350C
while in the lower zone of the sides it does not exceed
250C and at the top the maximum temperature is
125C.
From the data plotted in Figure 4(b) it can be seen
that the temperatures registered by the thermocouples
located in the upper sides of the tank show a degree of
scattering. This must be attributed to the fact that the
fire was not exactly symmetrical owing to turbulences
and the arrangement of the module walls. Thus, although
theoretically all the thermocouples should show the same
behaviour (as in Figure 4(a)), the induced wind and the
radiation from the heated walls caused these differences
between the measurement points.
For the 12 m2 pool fires (Figure 4(c)), the scattering
is less noticeable and the temperatures are higher,
because in this case the engulfment of the tank in the
fire is more complete, and therefore the impingement of
the flames on its walls has a greater influence. As in
Figure 4(a), the bottom of the tank (Figure 4(d)) for
the 12 mz pool fires shows the same behaviour for all
the thermocouples.
The evolution of the temperature at different points
of the tank as a function of time has also been plotted
in Figure 5. The experimental points show clearly how
the bottom of the tank undergoes the most intense heating action, due to the greater exposure to the flames.

Protection

234

of equipment

engulfed

8) 400 I

in a pool
b) 300 I

350 -300 --

JiE&
17 11

5
.

250
A

A Themwouple

0 Thermocouple 5

50

0 Thermocouple 11

A htmocouple

17
04

O-1
0

10

20

40
ti:

50

!FA
5

0
A
A
0

f
j

0
8

40

50

90

a
0

10

20

30
time(s)

during the first minute. (a) Test number 13 (4m *, hexane), tank bottom; (b) test number 13 (4 m*,
(c) test number 91 (12 m*, kerosene), tank bottom; (d) test number 91 (12 m*, kerosene), upper tank lat-

I92an

+t=6a

-a-t-wr
300.. -A-tear

42

-et421
-A-t**

200.250..

150..

-o-t=9or

im..
50 ..

04
60

im

150

200

HeisM (cm)
Figure5
Evolution of
instants as a function
kerosene)

04

Operation of the extinguishing

90

0 llwrrmcoupb 15

50

time (8)
Tank temperature
upper tank laterals;

50

150
100

30

40

0 Thermocouple
13

20

60

A
A

10

50

0 Ttwmpcoupb 7

-200

04

40

)-f~
8

O
0

30
time (8)

250

20

I
10

a Therwcoupb 11

(*)

60

3501
17 11

0 Thermocouple 15

100 --

19

. Thermocouple 9

150 --

Figure4
hexane),
erals

m Thermocouple 3

l-

0 Thermocouple 13

200 --

50 --

0 Thermocouple 7

g250 --

et al.

fire: E. Planas-Cuchi

tank temperature
at different time
of height. Test number 89 (12 m*,

system

The evolution of the fire during its development and during the extinguishing process can be seen schematically
in Figure 6, obtained from a film. In the initial steps the
fire increases rapidly, soon engulfing the whole tank. The
shape of the flames is variable, as usually happens with
relatively large fires. As flame length increases, so do
the turbulence and mobility of the flames.
From the moment at which the extinction system is
started, a significant increase in the turbulence of the
flames is observed, together with the existence of wind
from the west side; as a consequence, the flames move
towards the east side (see Figure 6)4. Therefore, the west
side of the tank is less influenced by the flames (the
surface area undergoing flame impingement decreases
significantly) and consequently the temperatures measured here are lower.
After reaching a maximum, the temperature at the
various measuring points of the tank starts to decrease;
in the plot of temperature versus time, a negative slope
is obtained. This situation corresponds to the moment at
which the cooling effect of water/foam starts, i.e. when

Protection of equipment
Test no12
Time = 10s

235

engulfed in a pool fire: E. Planas-Cuchi et al.

Test no12
Time = 20 8

8
A 300
2 250
a
[ 200
j
Test no12
Time = 30 s

150

Test no12
Time = 40 s

50

100

150

200

250

300

350

400

tlrne (s)
Figure 7 Stages of the temperature evolution at the lower tank
laterals (thermocouple 6). Test number 12 (4 m2, hexane)
Test no12
Time = 50 s
n4

Test no12
Time = 60 s

Test no12
Time = 120s

Test no12
Time = 1802

Figure 6 Evolution of the fire during its development and during the extinguishing process. Test number 12 (4 m*, hexane)

the mixture reaches that measuring point. In any event,


while the tank surface continues to be exposed to flame
impingement,
temperature
will decrease
slowly,
although the size of the fire decreases significantly after
the start-up of the deluge system.
Later on, as the pool gradually gets covered by the
foam, the height of the flames gets less and less and they
no longer impinge on the tank wall. Now the cooling
action of the water is stronger and the temperature
decreases dramatically, with a steep negative slope in
the plot of temperature versus time. Finally, when the
fire is extinguished and the extinguishing system is
stopped, a slight increase in temperature is again
observed due to the heating caused by the radiation from
the hot walls of the experimental module.
This process establishes in a relatively clear way the
existence of five stages in the evolution of temperature at
the various points of the tank wall (Figure 7)4. The first
one, which corresponds to the initial development of the
fire, is characterized by a rapid increase in temperature;
in the plot of temperature versus time a practically

straight line is obtained with a steep slope (the heating


rate depends on the location - at the bottom of the tank
values of the order of 3.6Cs were registered). In the
second stage, which corresponds to a fully developed
pool fire, temperature continues to increase but at a
slower rate and with a trend which is no longer linear;
in this stage a considerable wind appeared in all the tests.
The third stage initiates with the start-up of the
cooling/extinguishing system. The temperature starts to
decrease - its maximum value having been reached at
the transition between these two steps - gradually, at an
approximately constant rate. The fourth stage corresponds to a situation in which, due to the blanketing
action of the foam over the pool, the magnitude of the
flames has decreased considerably and they no longer
reach the point in the tank where the temperature is
being measured. Now the heat balance is clearly influenced by the dominant mechanism of the foam/water
cooling, with a relatively weak heat input by radiation;
thus, temperature decreases abruptly, following an
approximately linear trend. Finally, once the fire is extinguished, gentle heating is observed.
The existence of these five stages was observed
only in certain conditions during the experimental tests.
In other cases some of the intermediate stages were not
found, depending on the point at which the temperature
was being measured. This happened, for example, when
the influence of flames on the tank wall ceased at the
same moment that the cooling mixture reached that particular point. Stage 3 does not exist and just after reaching
its maximum value the temperature decreases abruptly
to very low values, and afterwards continues to fall very
gradually (see Figure 8). For the whole system, this situation would correspond to an immediate extinction of
the pool fire. This case, which would be the best one
from the point of view of controlling the emergency, in
practice implies conditions which can only be found at
certain points of real installations.
At the bottom of the tank, however, flame impingement existed practically throughout the experimental
test, finishing only when the fire was extinguished. Furthermore, this zone was never reached by the cooling
mixture in the various tests that were carried out. This

236

engulfed in a pool fire: E. Planas-Cuchi et al.

Protection of equipment
350

overall heat release rate and of the heat released by convection are discussed.

300
250

Overall heat released


The heat released during each test was measured as a

g
B
e! 200
i

150

I-
100

50
_

5
01
0

50

100

150

200

250

300

350

400

time (s)
Figure 8 Stages of temperature evolution at the lower tank laterals (thermocouple 12). Test number 12 (4 m2, hexane)

is why slope changes are smoother and therefore the


temperature values are higher up to the extinction of the
fire (Figure 9).

Heat released from the pool fire


Probably the best way to follow the development and
evolution of the pool fire is to measure the heat released
by the combustion. For a pool fire of a given fuel, the
heat release rate is a function of the combustion rate and
of the combustion efficiency. The heat release rate will
rise rapidly in the initial moments, while the fire is
developing and the size of the flames is increasing, and
after that it will have an approximately constant value
once the fire is completely developed. Nevertheless, for
pool fires of a relatively large size, this second stage which would correspond to a stationary state if no
changes or external actions were imposed on the system - is characterized by significant variations in flame
size and shape.
The evolution of the heat release rate as a function
of time will be therefore a way to identify the moment
at which the pool fire is completely developed. In the
following paragraphs the experimental results of the
700

Heat released by convection


The heat released from a fire is essentially divided into

two fractions, one transferred by convection and the


other transmitted by radiation. Both contributions can be
determined separately and independently from Q,,. The
heat released by convection can be obtained from the
flow rate and temperature of the gases produced in the
combustion:

.
Q

I,
=

c-cm

iv*moke*

cpx AT,,, .

(1)

During the experimental tests, these data were measured


by the calorimeter and treated in such a way as to give
continuously the value of Q,,,,. The trend of the evolution of the heat released by convection as a function of
time is similar to that corresponding to the overall heat
release rate: for the pool fire of 4 m* a gradual increase
in the heat released, with a steep slope during the first
30 s; later on, a more gradual increase is observed - with
some scattering caused by the turbulence and irregularity
associated with the phenomenon - until the maximum
value (steady state) is reached at approximately 60 s.

Ooo
I

l-

!Gi4
5

600

function of time by using the adiabatic calorimeter. The


plot of overall heat release rate versus time (Figure 10)
shows a steep rise at first, with a characteristic change
in slope at approximately 30 s; from this moment the
heat release rate still increases, but more gradually, up
to a maximum-value (fully developed fire). For a 4 m*
pool fire of hexane, the heat release rate increased from
zero at ignition (t = 0) to rates of the order of 5 100 kW
after 30 s. For the kerosene pool fires (12 m*), the heat
release rate increased to rates of the order of 16 600 after
60 s. When the deluge system was activated, the heat
release rate decreased rapidly until the extinction of the
fire. In those cases in which only water was used for the
extinction (tests 14 and 15), the heat release rate also
decreased after the activation of the deluge system, but
a few seconds later the fire revived.

6000 +

500

g! 400

B
B

p 300
8
200
100
0

hod ,

c
-F

50

100

150

200

250

3o 50

100

150

200

250

time (s)

time (s)
Figure9
Temperature
evolution
at
the
tank
(thermocouple 5). Test number 91 (23 m2, hexane)

bottom

Figure 10 Total heat release rate evolution


time for test number 13 (4 ma, hexane)

as a function

of

Protection

of equipment

engulfed

After this moment, a sudden fall can be observed due to


the start-up of the extinguishing system.
The 12 m2 pool fires (Figure 11) show an increase
in the heat released, with a steep slope, during the first
60 s, after which the extinction system is activated. In
this case the steady state was not reached, i.e. the fire
was still developing when the sprinklers were activated.
It can be observed that after that moment the heat release
rate displays an approximately constant value for a certain period, with a steep fall afterwards; this can be attributed to the difficulties in extinguishing the fire.
For hexane pool fires of 4 m*, the heat released by
convection was approximately 57% of the overall heat
released. In the case of kerosene, this fraction was
approximately 65%. The kerosene fires gave rise to
larger amounts of dark smoke, i.e. the fire was less bright
and the amount of heat transmitted by radiation was correspondingly lower.

in a pool

fire: E. Planas-Cuchi

et al.

237

In this equation, the first term on the right-hand side


corresponds to the heat transferred by conduction; this
is a boundary effect, significant only for pools with a
few centimetres in diameter and, therefore, negligible
here. The second term is the heat transferred by convection, also negligible for the large pools studied in this
work. For pools with a diameter of 1 m or more, the
radiation mechanism is the only one which is really
important. Therefore, in this work the rate at which heat
is transferred to the pool is given by:

ri =
-(n&/4)

F(q - c)(l

- eekd) .

The theoretical combustion rate can be obtained from


this expression and from the heat required to evaporate
the fuel. As combustion rate increases with pool size,
for a pool of infinite diameter, the maximum combustion
rate is obtained:

Discussion
Most communications published on hydrocarbon pool
fires deal with fully developed fires, and very little information is to be found on the initial unsteady state, when
the fire is developing. This is why there is no equation
available in the literature to predict the evolution of
flame temperature during the initial development of a
fire. Moreover, when relatively large equipment is
engulfed in a pool fire, it should have a disturbing effect
on the flames as compared to a free surface pool fire;
again, this aspect has not been studied by those authors
who have worked in this field5-8.
In the following paragraphs this effect is studied. In
order to calculate the theoretical (maximum) combustion
rate in both an unsteady and a steady state, a semiempirical expression is derived from the treatment of experimental data. Accordingly, the heat transfer rate from the
fire to the pool is given by the following expression:
rd 214

Heat bv convection

and for a pool fire of finite diameter:


.
m f=h;max
N

(1 -emM).

From equation (4) it can be inferred that during the


initial development of the fire (unsteady state) the combustion rate will change with time, as will flame temperature. Moreover, flame temperature will also change
with height.
The experimental data have shown that during the
development of a pool fire the flame temperature
changes with time according to an expression of this
type:
Tf

@A=

t
(b

(f-5)

&)

where a and b are parameters which can depend on


height. The values obtained for the different operating
conditions are given in Table 2.

Heat by conduction
+

oF(G

Table 2

1 - emkd).

c)(

Values a and b for different operating conditions

keat by radiation

7000
0 0

t)C b
6000 --

0"

08

coo

%O

Q,

80

Hexane
(4 m2)

Kerosene
(12 m*)

0.000851
~0.0034+0.021 h)

(0.000465-0.000188h~
(0.0347+0.0114h)

5000 --

Equation (5) was used to calculate the flame temperature just above the pool surface (h = 0). However, it
must be taken into account that this temperature corresponds to the central axis of the flame. Flame temperature decreases over the radius, from flame axis to periphery; at the flame boundary, the temperature is
approximately 50% (in terms of temperature increase
with respect to room temperature*) of the temperature

4000 --

00
0

3000 --

0
0

2000 -0
1000

--

rprinklers
diwUon

0
0

ocr

30

60

00

120

150

180

210

240

time (s)
Figure 11 Heat released by convection
for test number 90 (12 m2, kerosene)

as a function of time

* rR - T, = 0.5 x (Tr- TJ, then r, = 0.5 x CT,+ T,). The average


temperature at a given height can be determined, as a function
of temperature at flame boundary and room temperature, from
the following expression: T,,= (T, + T,)/Z. In this work, a room
temperature of 293K has been taken.

238

Protection of equipment

engulfed in a pool fire: E. Planas-Cuchi et al.

at the axislo. Thus, the average flame temperature at a


given height is approximately
T,, = 0.75 . Tf + 73.25

(7)

By introducing this equation, together with equation (6),


into equation (5), it is possible to obtain a semiempirical
expression for the calculation of the combustion rate during the non-stationary regime:
riz(t) =

FU

AK + c,(T,-T,)

0.75(b+at) + 73.25 4

[(

-I

(8)

(l-e-).
1

The heat release rate can be obtained as

*n = m; AH, .
Q t,,fa,

(9)

As in this case the view factor value between the


fire and the pool is unity and the mass loss rate can be
determined from equation (8); the rate at which heat is
released can be calculated from the following relationship:

When there is equipment engulfed in the fire, this equipment has a hindering or obstructing effect on the fire,
thus a hindering factor can be defined as the ratio
between the heat released in the combustion of a pool
fire with an obstacle (for example, a tank) and that
released in the same pool fire with a free surface:
Heat released in the combustion
of a pool with an obstacle
5=
Heat released in the combustion of the same
pool without any obstacle
(12)
5 would have a maximum value of 1 in those cases in
which the influence of the obstacle is negligible, and
decreasing values as the hindering effect increases (i.e.
as the size of the equipment as compared to that of the
pool increases).
Finally, the efficiency of hindered combustion can
be defined as the ratio between the heat released in the
combustion of a hindered pool and the maximum tbeoretical heat which could be released if the combustion
was complete and non-obstructed:
%bst

This expression gives the maximum heat release rate,


i.e. that corresponding to the combustion of all the fuel
evaporated. However, in an accidental fire, not all the
fuel is burnt, i.e. combustion efficiency is not 100%. Figure 12 shows the experimental values for a fire of 4 m2,
as well as those calculated by the method proposed in
the aforementioned paragraphs. As can be observed,
experimental values are fairly lower than the theoretical
ones. Taking into account the fact that the combustion
can be hindered by the existence of the engulfed equipment, two different combustion efficiencies can be
defined. First of all, the combustion efficiency,
Heat released in the combustion
of a free surface pool
= Heat released in the case of complete combustion
of all the fuel evaporated
(11)

12000

6000

(13)

5 =

Heat released in the combustion


of a hindered pool
Heat released in the complete combustion of all *
the fuel evaporated in a free pool
Figure 13 shows the value of q&r for the data
shown in Figure 12 (pool fire of 4 m2 with a tank
engulfed in it); the experimental data corresponding to
the initial moments have a certain lack of accuracy due
to the variability of fire spreading and increasing in this
first step. Therefore, the VdUeS of qObSIare significant
only over the region of Figure 13, in which they are
approximately constant. For this set of values, the
efficiency of hindered combustion is approximately
47%, considerably lower than the values found in the
literature for free surface pool fires.
The hindering effect of equipment engulfed in the
fire can be observed from the data plotted in Figure 14,
corresponding to a pool fire of hexane. For this hydro-

0
0

E
10000
s
E
0000

77 x

co
-I

14000

8
I

0
0
0
0
o

+f

4000

00

0
0

2000
0

pi

lo

20

30

40

50

60

time(s)
Figure 12 Total heat release rate, experimental and theoretical, as a function of time for test number 13 (4 m*, hexane)

10

20

;5

30

40

50

60

time (s)
Figure 13 Efficiency of hindered combustion
time for test number 13 (4 m*, hexane)

as a function of

Protection

of equipment

engulfed

12000
E

10000

c
t
P
;

8000
6000

0
0

10

20

30

40

50

60

time(s)
Figure 14 Total heat release rate as a function of time for tests
number 13 and 9 (4 m2. hexane. with and without tank
respectively). The horizontal line cor;esponds to the theoretical
heat release rate at steady-state, calculated according to Babrauskass equation

carbon, combustion efficiency in free surface pool fires


of 92% has been proposed in the literature. Figure 14
shows experimental data corresponding to the heat
release rate for the pool fire with and without a tank
engulfed in it. Furthermore, the maximum theoretical
heat release rate (77= 100%) has also been plotted, calculated according to the method explained in the previous
paragraphs. To check this method, the following data
can be considered: the heat release rate at steady state
(pool fire of hexane, 4 m*) calculated from the combustion rate predicted by Babrauskass equation
is
13 045 kW; by applying the new method proposed here which can be used even during the unsteady state - a
value of 13 807 kW is obtained at f = 60 s (after reaching
the stationary state). The small difference (5.5% with
respect to the highest value) confirms the validity of the
method. In Figure 14 it can be seen that for a free surface pool fire, the steady state would be reached after
approximately 45 s, with a combustion efficiency of
approximately 88%.

Conclusions
The experimental data obtained have allowed the study
of the evolution of a pool fire from its beginning up to
its complete development. As the sizes of the pool fires
were relatively large (4 m* and 12 m*), both the shape
and size of the flames were relatively variable. This was
even more important after the start-up of the deluge system, as a relatively strong wind was induced. In these
conditions, the fire was moved towards one of the sides
and this gave rise to an asymmetrical evolution of the
temperatures of the equipment engulfed in the pool fire,
with flame impingement essentially on the bottom and
on one of the sides.
Concerning the heating of this equipment, its temperature increased following an approximately linear
trend, at a velocity which depended on the location of
the measuring point. Thus, the highest temperatures were
measured at the bottom of the tank (approximately
300C in 1 min). In this zone, there was direct contact

in a pool fire: E. Planas-Cuchi

et al.

239

between the flames and the tank wall practically


throughout the run, and the cooling fluid never reached
this zone. Instead, as height increased the temperature
reached was lower and the effect of the cooling fluid
(after the start-up of the deluge system) was far greater.
At a given point, the complete process followed five
different steps. The first one corresponded to strong heating, followed by a second stage in which the temperature
still increased but at a slower rate (in this step, induced
wind was already noticeable). Immediately after the
start-up of the deluge system, the temperature started to
decrease, but very slowly (third step); there was still
flame impingement, although the height of the flames
was already decreasing. Once the cooling fluid reached
that particular point, the temperature decreased dramatically (fourth step) up to the extinction of the fire. Afterwards, the temperature increased slightly due to the radiation from the hot walls of the experimental module. In
certain zones of the tank only some of these steps were
observed, depending on the existence or otherwise of
flame impingement and on the contact or lack of it with
the cooling fluid.
Concerning the heat released by the fire, the experimental results were significantly lower than the theoretical values which should be expected. A method has been
developed to calculate the maximum combustion rate,
both for the developing stage (unsteady state) and for
the fully developed fire. This has allowed comparison
between the theoretical values and the experimental data
with and without a tank in the fire. It has been found
that the heat release rate is significantly lower when there
is a tank engulfed in the fire; the existence of this equipment is in fact an obstacle for the flames. Thus, both a
hindering factor and an efficiency of hindered combustion have been defined. The values of qobsr are significantly lower than those corresponding to the combustion efficiency of a free surface pool, this being a clear
effect of the hindering consequence of the engulfed
equipment.
The results obtained show the advantages of a system with the objective - from the first moment - of simultaneously cooling the equipment and extinguishing the
fire. Covering these two aspects with a single deluge system can be very useful in providing a rapid solution to
certain emergencies which can be encountered in the
process industries.

Acknowledgements
Two of the authors @P.-C. and J.C.) gratefully
acknowledge partial financial support from the Universities and Research Commission of the Generditut
(Catalan government).

References
1. Wighus, R. J.

Loss Prev. Process hi. 1994, 7, 305-309


2. Foam and Water Deluge Systems for Off-Shore Oil Platforms,

Project Summary Report, Ek Contract: TH-X125/89-IT


3. Fixed Water Surav and Deluge Protection for Oil and Chemical
Plant, IRI Info-m&ion IM.12r2.1.2, 3 June, 1991
4. Casal, J., Planas, E., Lancia, A. and Bordignon, L. in Fire engineering and emergency planning. Research and application, ed: R.
Barham, Chapman and Hall, London, 19%. PD 209-217
5. Aydemir, N. k., Magapu, V. K., Sousa, A. k: M. and Venart, J.
E. S. J. Hazardous Materials 1988, 20, 239-262

Protection of equipment engulfed in a pool fire: E. Planas-Cuchi et al.

240

6. Bainbridge, B. L. and Keltner, N. R. J. Hazardous Materials 1988,


20,214o
7. Beynon, G. V., Cowley, L. T., Small, L. M. and Williams, I. J.
Hazardous Materials 1988,20, 227-238
8. Birk, A. M. J. of Hazardous Materials 1988, 20, 197-225
9. Mudan, K. S. hog. Energy. Combust. Sci. 1984, 10, 59-80
10. Heskestad, G. F. Safety J. 1984, 7, 25

11. Tewarson, A. SFPE handbook of fire protection engineering,


Chap. 13, Section 1, 1.179-1.199, 1st edn, NFPA, 1990
12. Babrauskas, V. Fire Tech. 1983, 19, 251

Nomenclature
ah

CP
d
F
H

Ah,
AH,
k

Parameters in equation 5
Specific heat of the generated gases (kJ kg- . K-1)
Pool diameter (m)
View factor (dimensionless)
Convection coefficient (kW me2 . K-)
Heat of combustion (kJ kg-)
Heat of vaporization (kJ . kg-)
Extinction coefficient (m-l)

TL1
Tb
Tf
TR

Thermal conductivity of the pool walls (kW . m-r K-r)


Combustion rate (kg. m-* s-r)
Maximum combustion rate (kg. m-*. s-l)
Flow rate of the gases produced in the combustion
(kg . m-* s-l)
Heat transfer rate (kW)
Heat transfer rate per unit surface (kW . m-*)
Heat released by convection (kW . mdZ)
Total heat release rate emitted during the fuel combustion (kW . m-*)
Time (s)
Room temperature (K)
Average temperature of flames at a given height (K)
Boiling temperature of the fuel (K)
Centreline flame temperature (K)
Temperature at flame boundary (K)
Temperature rise of the smoke (K)
Combustion efficiency
Efficiency of hindered combustion
Stefan-Boltzmann constant (= 5.67 . 10-r? kW m-* aK4)
Hindering factor

S-ar putea să vă placă și