Você está na página 1de 7

Fire Safety Journal 47 (2012) 17

Contents lists available at SciVerse ScienceDirect

Fire Safety Journal


journal homepage: www.elsevier.com/locate/firesaf

Froude-modeling-based general scaling relationships for re suppression


by water sprays
Hong-Zeng Yu
FM Global, 1151 Boston-Providence Turnpike, Norwood, MA 02062, USA

a r t i c l e i n f o

abstract

Article history:
Received 6 November 2009
Received in revised form
24 January 2011
Accepted 21 September 2011

Based on the Froude modeling concept, Heskestad proposed a set of scaling relationships for the
sprayplume interaction for high droplet Reynolds number conditions (10 r Red r500). The droplet
Reynolds number is dened as the ratio of the product of droplet diameter and the absolute value of the
droplet velocity relative to the gas velocity over the gas kinematic viscosity. The aforementioned scaling
relationships have been used widely for scale-modeling of water-based re protection under conditions
within or beyond 10 r Red r500. Recently, it was shown that the same scaling relationships can be
extended to low droplet Reynolds number conditions of Red r 1 except that the droplet size is scaled
with the 1/4-power of the scale ratio, instead of the 1/2-power for 10 rRed r 500. The conditions of
10 r Red r 500 in general prevail in sprinkler applications and the conditions of Red r 1 usually take
place in water mist applications. With the above difference in mind, the Froude modeling is revisited in
this paper to establish a set of general scaling relationships not limited to specic droplet-Reynoldsnumber regimes. The derived general relationships not only reproduce those for Red r 1 and
10 r Red r 500, but also reafrm the previous nding that the scaling relationships are independent
of the scale ratio except for the droplet size, whose scaling requirement varies with the range of Red
values in which the scale-modeling is performed. The published experimental results to date show that
the Froude-modeling-based scaling relationships for sprayplume interaction are a viable tool for scalemodeling of the re suppression or extinguishment by water sprays.
& 2011 Elsevier Ltd. All rights reserved.

Keywords:
Physical scaling
Froude modeling
Water sprays
Fire suppression
Fire extinguishment

1. Introduction
In the 1970s, Heskestad proposed a set of scaling relationships
for the interaction of water sprays and res in geometrically
similar spaces [1] based on the Froude modeling concept. These
scaling relationships were originally intended for sprinkler sprays
where the droplet Reynolds number (Red) is expected to be
between 10 and 500. This condition leads to the requirement
that droplet size is scaled with 1/2-power of the characteristic
length of the modeled space. These scaling relationships have
been satisfactorily validated with re extinguishment experiments conducted in open space under high Red conditions where
high droplet velocity relative to gas velocity was prevalent [2,3].
One requirement of the Froude modeling is to preserve the
characteristics of momentum transfer and heat transfer between
water droplets and gas ow in the space of interest for different
physical scales. It is well known that the momentum transfer and
heat transfer between droplets and gas medium is a function of
droplet Reynolds number. For the re suppression or extinguishment by water mist where droplets move closely with the gas

E-mail address: bert.yu@fmglobal.com


0379-7112/$ - see front matter & 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.resaf.2011.09.006

current, low Red conditions are typically the case. Theoretical


analysis for Red r1 showed that the scaling relationships for
10 oRed o 500 could be broadly applied to the Red r1 condition
except for the droplet size, where the droplet diameter should be
scaled with 1/4-power of the scale ratio [4], instead of the 1/2
power for 10 oRed o500. The droplet-scaling requirement for the
Red r1 condition was conrmed with a series of experiments
conducted in open space and in enclosures by applying water
mist to methane, propylene or propane res [4,5].
When using water sprays for re protection, the prevailing
physical conditions may not strictly meet the conditions of Red r1
or 10oRed o500. To address this issue, the Froude modeling has
been revisited from a general application perspective, resulting in a
set of general relationships, which is not limited to specic Reynolds
number regimes. These general relationships have been provided in
Ref. [5]without the derivation. The theoretical basis and derivation
were presented in the 7th Asia-Oceania Symposium on Fire Science
and Technology [6]. The purpose of this paper is to make available
the theoretical basis and derivation in the public domain for the
above-mentioned general scaling relationships. The experimental
evaluations of the physical scaling for re suppression or extinguishment by water sprays using the Froude-modeling-based scaling
relationships are also reviewed in this paper.

H.-Z. Yu / Fire Safety Journal 47 (2012) 17

Q_ c
Q_ cool
Red
S
,
S
t
T
u
,
u
000
vw
V_ w

Nomenclature
A
Ad
AV
CD
Cp,g
Cp,w
d
,
Fd
,
g
hd
kg
L
Lv
md
_
m
,e
_w
m
_e
M
_w
M

frontal area of water droplet


droplet surface area
total surface area of control volume
drag coefcient
specic heat of gas
specic heat of water
droplet diameter
droplet drag force vector
gravitational acceleration
droplet heat transfer coefcient
thermal conductivity of gas
characteristic length
latent heat of vaporization
mass of water droplet
droplet vaporization rate
water mass ux vector
total vaporization rate of droplets
total water mass discharge rate
droplet number density
droplet Nusselt number
Prandtl number
dynamic pressure of the gas ow

n000
Nud
Pr
Dp

rg
rw
ng

Subscripts
c
cool
d
g
N

2. The general scaling relationships


Before presenting the derivation of the general scaling relationships, two dimensionless parameters need to be dened: one
is the droplet Reynolds number and the other is the Froude
number.
The droplet Reynolds number is dened as
,

Red

d9u d u g 9

ng

where d and u d are the droplet diameter and droplet velocity


,
vector, and u g and ng are the gas velocity vector and gas kinematic
viscosity, respectively.
On the other hand, the Froude number is dened as the ratio of
the momentum force versus the buoyancy force of the gas ow:
Fr

rg u2g
,
gLr1 rg

re convective heat release rate


water mist cooling rate
droplet Reynolds number
scale ratio
total drag force of droplets per unit volume
time
temperature
magnitude of velocity vector
velocity vector
total droplet volume per unit volume
total volumetric water discharge rate
gas density
water density
gas kinematic viscosity

where rg and r1 are the re gas density and ambient density,


respectively, g is gravitational acceleration, L is the characteristic
dimension of the re environment, and ug is the scalar value of
,
the gas velocity vector u g .
For buoyancy-induced turbulent re plumes in geometrically
similar control volumes, the essence of Froude modeling is to
perform Froude number similarity of such ows. By preserving
the ratio of the re plumes momentum force and buoyancy force,
the re plumes in different scales are expected to be similar. If
the ambient temperature and re plume temperature maintain
constant in different scales (i.e. constant densities), the remaining
variables in Eq. (2) are the gas velocity and the characteristic
dimension of the re environment, which can be substituted with
two independent variables: a characteristic length and time. For
instance, the characteristic time could be the time for one
complete gas ow circulation in the control volume, and the
corresponding characteristic length is the circumference of the
control volume in the primary gas ow direction. After water
sprays are discharged into the control volume, the momentum

characteristic quantities
cooling
water droplet
gas medium
ambient

and heat transfer characteristics between water droplets and gas


ow have to be also preserved to maintain the similarities of gas
ow pattern, spray pattern, and thermal eld in different scales.
The above description can be summarized as follows: to maintain
the overall similarity of two-phase ow and thermal eld for
water sprays discharging to the re plume in geometrically
similar spaces, we have to preserve: (1) the Froude number of
the dry gas ow before water sprays are discharged, (2) the
temperatures of the ambient and dry gas ow, and (3) the
momentum transfer and heat transfer characteristics between
water droplets and gas ow. The droplet vaporization characteristic is automatically preserved if the heat transfer characteristic
is preserved.
Therefore, by maintaining constant for the ambient temperature, dry gas ow temperature, and Froude number in different
scales, the following relationships are obtained:
u2g,c pLc ) ug,c pL1=2
c ,

tc p

Lc
pL1=2
c ,
ug,c

2
Dpprg u2g,c pL1=2
c pLc

and
2
5=2
Q_ c prg ug,c T g T 1 L2c pL1=2
c Lc pLc ,

where Q_ c is the convective heat release rate of the re, Lc is the


characteristic dimension, and other symbols are dened in the
Nomenclature.
Assuming that vaporizing droplets are subjected to only
gravitational force and drag force, the equation of motion of a
single water droplet is expressed by
,

dmd u d
, 1
,
,
,
,
mg  C D Arg 9u d u g 9u d u g ,
dt
2

H.-Z. Yu / Fire Safety Journal 47 (2012) 17

knxg rg t 2c

where the mass of the droplet is


md

p
6

rw d ,

24
Red

27
Re0:8
d
12

for

Red r 1,

for

1 o Red r 10,

for

10 oRed r 500:

It has been reported that the drag coefcient for a vaporizing


droplet is very close to that for a non-vaporizing droplet [8].
Therefore, it is reasonable that Eq. (9) is used to estimate the drag
coefcient of a vaporizing droplet in the respective ranges of
droplet Reynolds numbers.
In general, the drag coefcient can be expressed by
k
,
Rex

11

where tc is the characteristic time associated with the gas ow in


a control volume.
The acceleration of the droplet in turn can be expressed with
,
du d

Lc d X d =Lc
:
t 2c dt=t c 2

_ w 9 is the water mass ux vector.


and 9m
After combining Eqs. (10), (19) and (20), we have


2 ,

4rw

kn

00
1x ,
_ w9 , ,
9m
u d u g :
,
9u d 9

9u d u g 9

21

00
2x ,
_ w 9 u2g,c
9m
p
,
Lc
9u d 9

9u d u g 9

22

dt=t c

1x 0
1
,
,


,
 Lc dX d =Lc ,

u
L
dX
=L

u
g
g
c
c
d
 @
A



ug,c 
t c ug,c dt=t c
ug,c
t c ug,c dt=t c

x 1 x
gd

Eq. (12) indicates that u d is scaled with L1/2. As a result

dX d =Lc dd=L=dt=t c
t 2 , 3 knxg rg t 2c 2x 1 x
3
c g
u d
dt=t c
d=L
4 rw Lc g,c
Lc

14

where ug,c is the characteristic buoyancy-induced gas velocity. In


Eq. (14), L is a self-fullling scaling parameter in the second term
on the left-hand side of the equation. As a result, the function of L
in terms of characteristic quantities has to be derived from the
similarity conditions required in the other terms of Eq. (14) so
that the scaled spray pattern will be invariant in different model
scales.
In order to yield the same droplet-gas dynamics, thus reproduce the same relative droplet trajectory pattern in different
model scales, the following coefcients in Eq. (14) have to be
invariant, considering that all terms in Eq. (14) are signicant:
t 2c
constant,
Lc

3rg

To reproduce the gas ow pattern in different


model scales
,
under water discharge, the absolute value of S has to be proportional to the characteristic pressure gradient of the gas ow. Since
ng is invariant, we have

Substituting Eqs. (12) and (13) into Eq. (11), we have


d X d =Lc

20

1 x

13

19

1
,
,
,
,
C D Arg 9u d u g 9u d u g ,
2 00

2 ,

18

where
,
Fd

12

x
pL2x=21
:
c

00
, ,
_ w9
F d 9m
,
,
md 9u d 9

Lc dX d =Lc
,
t c dt=t c

17

Froude modeling requires not only that the relative spray


patterns in different scales have to be reproduced, but also that
the similarity of the gas ow patterns has to be maintained. As
water droplets travel in the gas medium, they exchange momentum with the gas current via the interacting drag force between
the droplets and gas medium. Assuming that the frictional loss is
negligible, the total force exerted by the droplets on the gas
medium per unit volume is

The droplet velocity can be expressed with

dt

dpug,c

x
du d
, dd=dt
, 3 kng rg 1 x ,
, 1x ,
,
g
3u d
d
9u d u g 9 u d u g :
d
4 rw
dt

16

Based on Eqs. (3) and (4), both conditions required in Eqs. (15)
and (17) are self-fullled. In Eq. (16), the gas and water properties
do not change between different model scales because the
temperatures are preserved, as per the modeling requirement.
Therefore, to satisfy Eq. (16), the functional relationship for
droplet size has to be

10

where k and x are constants, which give the best regression of CD


for a designated range of droplet Reynolds numbers, either for
those shown in Eq. (9) or any other range in which the scalemodeling is performed.
Substituting Eqs. (8) and (10) and A pd2/4 into Eq. (7), we
have

,
ud

constant,

Lc
constant:
t c ug,c

2x=1 x

Re0:5
d

CD

1 x

u2x
g,c d

and

and CD and A are the drag coefcient and the frontal area of the
droplet, respectively.
For a spherical droplet, the drag coefcient can be expressed
with power-law functions of the droplet Reynolds number such
as [7]
CD

rw Lc

15

, 00

_ w 9p
9m

,
, x2
1 x ,
9u d 99u d u g 9 u2g,c

Lc

pL1=2
c :

23

Eq. (23) states that the water uxes in different physical scales
are scaled with the square root of the characteristic dimension. As
a result, the number of droplets per unit volume is scaled with
, 00

000

_ w9
9m
,

md 9u d 9

x
pL3x2=21
:
c

24

Consequently, the total volume of water droplets per unit


space volume is scaled with
3

vw pn d pL0c ,
000

000

25

which states that the volumetric concentration of water droplets


is conserved in different scales as the gas temperature.
The total water mass discharge rate can be calculated by
integrating the water uxes under no re conditions with
I , 00 ,
_w m
_ w dA V ,
26
M
A

H.-Z. Yu / Fire Safety Journal 47 (2012) 17

where AV is the total surface area of the control volume. Therefore


V_ w

_w
M

rw

2
5=2
pL1=2
c Lc pLc ,

27

which states that the total volumetric water discharge rate has to
be scaled with L5/2.
Under different Reynolds number conditions, the Reynolds
analogy for heat transfer gives the following general relationship [9]:
Nud

hd d
pPrRe1x
d ,
kg

28

where Nu and Pr are the Nusselt and Prandlt numbers, respectively,


hd is the heat transfer coefcient on the droplet surface and kg is the
thermal conductivity of the gas medium.
Assuming that the droplet is heated up uniformly, the droplet
temperature in the initial heat-up period is approximated with
md C p,w

dT d
Ad hd T g T d :
dt

30

which states that the instantaneous droplet temperature is


reproduced in different physical scales, consistent with the
scaling requirement that the scalar variables be conserved.
Since both the droplet temperature and gas temperature are
_ e , is
conserved, the droplet vaporization rate per droplet, m
proportional to
54x=2 2x
_ e phd d2 pRe1x
:
m
d dpLc

31

When the droplet velocity relative to the gas velocity is small,


the quasi-steady droplet vaporization rate is proportional to d(kg/
Cp,g)ln(1 B) [10], where B (Cp,g/Lv)(Tg  Td). Since Cp,g and Lv are
approximately constant and temperature is conserved in the
modeling, B can be regarded as constant; thus, the droplet
vaporization rate is only proportional to the droplet diameter
for Red r1.
When Red r1, the value of x is 1 (cf. Eqs. (9) and (10)). As a
result, Eq. (31) indicates that the droplet vaporization rate is
proportional to droplet size, consistent with the expectation for
situations where the relative velocity between the droplet and gas
is small. On the other hand, with x 1/2 for 10oRed r500,
_ e pRe1=2
Eq. (31) gives m
d, which reproduces the functional
d
relationship previously reported for Red 4 20 [11].
_ e , is the
The total vaporization rate in the control volume, M
product of number of droplets per unit volume, domain volume,
and vaporization rate per droplet. Therefore
Z
2x 54x=2 2x 3
_ e n000 m
_ e dVpL3x6=2
M
Lc
Lc pL5=2
32
c
c :
V

Consequently, the total cooling rate, Q_ cool , in the control


volume is scaled with
Q_ cool pL5=2
c :

33

Furthermore, the droplet lifetime, td, is


3

td

x
m
d
L32x=21
c
p
p 54x=2
pL1=2
c ,

2x
_e
_e
m
m
L

Red r 1

10 o Red r500

Drag coefcient

pRe1
d

Dimension
Time
All scalar parameters except droplet
number density
Droplet number density
Velocity
Ventilation rate
Fire convective heat release rate
Total water discharge rate
Water ux
Total water cooling rate
Droplet diameter

S1
S1/2
S0

pRed
S1
S1/2
S0

S  3/4
S1/2
S5/2
S5/2
S5/2
S1/2
S5/2
S1/4

S  3/2
S1/2
S5/2
S5/2
S5/2
S1/2
S5/2
S1/2

Scaling parameters

1=2

Any Red
regime
pRex
d
S1
S1/2
S0
S(3x  6)/(2 2x)
S1/2
S5/2
S5/2
S5/2
S1/2
S5/2
S(2  x)/(2 2x)

29

By normalizing Eq. (29), we have


dT d =T 1
2
0
pt c d Re1x
d pLc ,
dt=t c

Table 1
Scaling relationships for different Red regimes.

34

which is consistent with the dependence of time scale on length


scale for the gas ow, as shown earlier.
Finally, it is worth mentioning that the individual gas concentrations are also conserved in different physical scales. Since the
convective heat release rate of the re is proportional to the chemical
heat release rate for the same fuel, the rates of produced combustion
products are expected to be proportional to L5/2
c . Since the water

vapor generation rate is scaled with L5/2


c , the total water vapor
generation rate in the control volume from both combustion and
vaporization is therefore proportional to L5/2
c . Since the entrainment
rate to the re is proportional to the product of the surface area
(pL2c ) and the entrainment velocity (pLc1=2 ), the rate of air entrained
to the re is therefore proportional to L5/2
c . Since the concentration is
proportional to the ratio of the generation rate and the entrainment
rate, the gas concentrations are therefore conserved.
Table 1 summarizes the general scaling relationships derived
above and the scaling relationships for Red r1 and 10 oRed r500
obtained by substituting x with 1 and 12, respectively. As shown, all
the scaling relationships are identical except for droplet number
density and droplet size. In the table, S denotes the scale ratio.
To properly implement the above scaling relationships, the
range of Red values should be rst estimated for the droplet ow
conditions for the intended physical scales. The power index x in
Eq. (10) is then determined for the Red range of modeling, which
in turn provides the corresponding droplet-scaling requirement.
Besides the droplet size is scaled as described above, the characteristics of the starting droplet size distribution and spray angle
should also be preserved in different scales.

3. The status of experimental evaluations of re suppression


applications
Although the modeling described above is performed strictly
from the view point of uid mechanics and heat transfer for the
interaction between re plume and water sprays in open space, it
does preserve the re environment condition in different scales,
encompassing the gas ow pattern, water spray pattern, thermal
eld, gas species concentration and volumetric water droplet
concentration. Therefore, it is expected that the re suppression
propensity could be reasonably reproduced in different scales as
long as the environment in which the re is present is reproduced. Nevertheless, before applying the scaling relationships
listed in Table 1 to the actual re protection problems, the
impacts of the following factors not considered in the modeling
need to be evaluated experimentally: radiation transfer, control
volume with solid boundaries (e.g. enclosure), and combustible
type as in the gas, liquid or solid phase.
To date, experiments based on the scaling relationships for
10oRed r500 [2,3] and Red r1 [4,5] have been conducted to
address most of the above factors, and the results show that re
suppression and extinguishment can be reproduced reasonably
well for scale ratios up to 10. The key experimental conditions
and results reported in Refs. [25] are briey reviewed below.

H.-Z. Yu / Fire Safety Journal 47 (2012) 17

Ref. [2] demonstrated that the propensity of methane re


extinguishment with single water sprays could be reproduced
quite well in open space. The experiments were conducted in two
physical scales of 1 and 10.
The diameters of the Scale-1 and Scale-10 sandbox-type gas
burners were 102 mm and 914 mm, respectively, where 914 mm
was slightly off from the ideal value of 1020 mm. The heat release
rates used in the Scale-1 experiments ranged from 1.1 to 4.4 kW,
and the corresponding heat release rates used in the Scale-10
experiments ranged from 365 to 1530 kW.
Both the Scale-1 and Scale-10 water sprays had a spray angle of
about 601, and exhibited similar gross droplet size distributions. The
volume-median droplet diameters of the Scale-1 and Scale-10 sprays
were about 176 and 380 mm, respectively, following more closely to
the proportionality of S1/3 than the ideal S1/2 for 10oRed r500.
For each scale, the experiments were conducted by positioning
the nozzle vertically downward on the axis of the gas burner. The
nozzle was 0.61 m above the burner surface in the Scale-1
experiments and 6.1 m in the Scale-10 experiments. Based on
the reasonably well-correlated re extinction results, it was
concluded in Ref. [2] that, between scale ratios of 1 and 10, the
imperfect scaling of radiation heat transfer and slight inaccuracy
in the scaling of droplet size did not seem to adversely affect the
scaling results.
3.2. Extinguishment of methane res and pool res in open space [3]
Ref. [3] extended the scaling experiments presented in Ref. [2] to
different nozzle heights above the methane burner. Experiments
were also conducted for heptane pool res to explore if the pool re
extinction results could also be correlated similarly as the methane
results. The experiments were conducted in scales of 1 and 3.
The diameters of the Scale-1 and Scale-3 methane burners
were 102 and 305 mm, respectively. The Scale-1 methane res
ranged from 1.1 to 16.8 kW, and the corresponding Scale-3
methane res ranged from 17.2 to 262 kW. The Scale-1 pool re
tests were conducted with a 99-mm diameter pan producing a
heat release rate of 2.4 kW, while the Scale-3 pool was 305 mm in
diameter producing about 43 kW.
Except for the Scale-1 nozzle described in Ref. [2], the spray
information of the other nozzles used in this investigation was
not reported except for the nozzle designations, spray angles and
orice diameters. However, it was stated that all the nozzles
produced ow rates proportional to the square root of the water
pressure in the investigated range.
As in Ref. [2], each experiment was conducted with a single
nozzle discharging vertically downward over the center of the gas
burner or heptane pool. For each nozzle height and heat release
rate, the water pressure was gradually increased until re
extinction occurred. The results showed that the extinction data
of the heptane pool res are consistent with those of the methane
res, except that the heptane pool res need higher water
discharge rates to achieve extinction.

The experiments were conducted in Scales 1, 3 and 9 by


introducing water mist into the ame by re entrainment. The
sandbox-type burners were 102, 305 and 914 mm in diameter.
For both the methane and propylene res, the re heat release
rates ranged from 1.1 to 3.3 kW in the Scale-1 experiments, 17.1 kW51.4 kW in the Scale-3 experiments, and
275824 kW in the Scale-9 experiments. Under free-burn, the
more radiative propylene res yielded radiation fractions of 0.24,
0.34 and 0.39 for the Scale-1, -3 and -9 res, respectively,
as opposed to 0.17, 0.23 and 0.27 for the corresponding
methane res.
Three sets of water mist nozzles were selected to produce
water mist discharges closely meeting the scaling requirements
for Red r1, where the droplet diameter is scaled with S1/4. The
volume-median droplet diameters of the Scale-1, -3 and -9 sprays
were 60, 73 and 90 mm, respectively. In each experiment, four
water mist nozzles were positioned equidistant to and around the
burner and in the plane of the burner surface. The water mist
sprays were directed upward in co-ow orientations with the re
plume so that the delivery of water mist to the ame was mainly
by entrainment.
Fig. 1 shows the cooling results obtained from the methane
and propylene re experiments, where the Scale-3 results are
used as the benchmark for comparison. Therefore, the Scale-1
results are scaled up by K 35/2 and the Scale-9 results are scaled
down by K (1/3)5/2 in the gure. The reasonably good agreement
between the methane and propylene re data indicates that the
impact of re radiation propensity is insignicant for the experiments conducted for scale ratios up to 9. The results also conrm
that the water mist cooling of re gases is indeed scaled with the
5/2-power of scale ratio as presented in Table 1.
3.4. Spray cooling of propane res in enclosures [5]
The impact of enclosure on Froude modeling applicability was
evaluated with propane re cooling experiments conducted in two
geometrically similar enclosures of 3:1 ratio under the Red r1
condition. The two enclosures measured 1.22 m  1.22 m  1.22 m
and 3.66 m  3.66 m  3.66 m, respectively. The ventilation effect
was evaluated by conducting the experiments with two door opening sizes for each enclosure: 0.30 m  0.61 m high and

70
C3H6 Scale 1, K=3
C3H6 Scale 3, K=1
C3H6 Scale 9, K=1/3
CH4 Scale 1, K=3
CH4 Scale 3, K=1
CH4 Scale 9, K=1/3
Regression

65
60
55
50

K5/2Qcool (kW)

3.1. Extinguishment of methane res in open space [2]

45

+40%

40
35
30
25
-40%

20

3.3. Spray cooling of methane and propylene res in open space


[4,12]

15
10
5

Two series of experiments were presented in Refs. [4,12] to


quantify the water mist cooling of methane and propylene res in
open space under the Red r1 condition. The objectives of these
experiments were (1) to validate the scaling requirement of
droplet diameter for the Red r1 condition; and (2) to evaluate
the impact of re radiation propensity on the scaling results.

0
0

10

20

30

40

K5/2Q

50
c

60

70

80

90

(kW)

Fig. 1. Comparison of water mist cooling rates obtained from the methane and
propylene re experiments conducted in open space.

H.-Z. Yu / Fire Safety Journal 47 (2012) 17

60

50
45

50

+13%

45

+13%

40
35
30

1.22x1.22x1.22 m enclosure, no extinction, S=1


3.66x3.66x3.66 m enclosure, no extinction, S=3
1.22x1.22x1.22 m enclosure, extinction, S=1
3.66x3.66x3.66 m enclosure, extinction, S=3
regression (slope=0.771)

55

Qcool/ S5/2 (kW)

Qcool/S5/2 (kW)

60

1.22x1.22x1.22 m enclosure, no extinction, S=1


3.66x3.66x3.66 m enclosure, no extinction, S=3
1.22x1.22x1.22 m enclosure, extinction, S=1
3.66x3.66x3.66 m enclosure, extinction, S=3
regresssion (slope=0.718)

55

-13%

25
20

40
35

25
20

15

15

10

10

-13%

30

10

15

20

25

30

Qa/S

5/2

35

40

45

50

55

60

10

(kW)

15

20

25

30

35

Qa/S

5/2

(kW)

40

45

50

55

60

Fig. 2. Scaled re cooling rates of the Scale-1 (circles) and Scale-3 (squares) experiments for the larger door openings (the graph on the left) and the smaller door openings
(the graph on the right). Closed symbols represent the data obtained from tests in which the re could persist indenitely; open symbols represent the tests in which re
extinction occurred.

0.61 m  0.61 m for the Scale-1 enclosure and 0.91 m  1.83 m high
and 1.83 m  1.83 m for the Scale-3 enclosure.
Propane res were established on the 305-mm and 914-mm
diameter burners in the Scale-1 and Scale-3 enclosures, respectively. Four corresponding pairs of propane supply rates were
selected according to the scaling requirement, producing
1050 kW in the Scale-1 enclosure and 150800 kW in the
Scale-3 enclosure.
Two sets of nozzles were selected to produce the water mist
sprays in the Scale-1 and Scale-3 enclosures. The Scale-1 and
Scale-3 water mist sprays exhibited similar droplet size distributions, and had a spray angle of about 601. Each Scale-1 spray
discharged 0.18 lpm with a volume-median droplet diameter of
62 mm, and the values of the Scale-3 spray were 2.86 lpm and
88 mm. In each enclosure, nine ceiling-mounted nozzles were
arranged in a 3  3 pattern with equal nozzle-to-nozzle and
nozzle-to-wall spacing. To minimize the disturbance to the re
by water mist sprays, the burner in each enclosure was protected
with a metal shield with openings for air access.
The re cooling results obtained from the Scale-1 and Scale-3
experiments are compared in Fig. 2, where the plot on the left is
for the larger door openings and plot on the right is for the
smaller door openings. The symbol S in the gure denotes the
scale ratio with respect to Scale 1. In both plots, the ordinate is
the water mist cooling rate divided by S5/2, and the abscissa is the
re heat release rate divided by S5/2. The use of re heat release
rate instead of convective heat release rate in the abscissa is due
to the fact that the heat loss to the enclosure surfaces during
water mist application was insignicant because the thermal
environment in the enclosure was extremely uniform during
water mist application. As a result, the net re heat input into
the enclosure was essentially the re heat release rate [5]. As for
the open space case described above for the methane and
propylene res, the water mist cooling of propane res in
enclosures can also be scaled with the scale ratio to the 5/2
power. The greater data regression slope for the smaller door
openings is consistent with the expectation that the re cooling
rate will be higher as the enclosure gas temperature is increased
by restricting the enclosure ventilation.

4. Conclusions
Froude modeling was re-visited for the interaction of water
sprays and buoyant re plumes to establish the general scaling
relationships not limited to specic droplet-Reynolds-number
(Red) regimes. The general relationships show that most of the
scaling requirements are identical for different Red regimes
except for droplet size and the droplet number density. However,
the droplet volume concentration is preserved in different scales.
The general relationship for droplet size provides the same
droplet-size scaling requirements reported previously for
10oRed o500 [1] and Red r1 [4]. When physical scaling is
performed in other Red regimes, the corresponding droplet-size
scaling requirement can be determined with the general
relationship.
To date, experiments based on the scaling relationships for
10oRed r500 and Red r1 have shown that re suppression or
extinguishment by water sprays can be scaled reasonably well
with the Froude-modeling-based scaling relationships for scale
ratios up to 10. In these experiments, the impacts of the following
factors were addressed: re radiation propensity, enclosure effect,
and combustible types as in gas or liquid. Additional re suppression and extinguishment experiments for pool res and solid
combustible res in enclosures are currently on-going. The results
will be reported when they are available.

References
[1] G. Heskestad, Physical modeling of re, physical modeling of re, J. Fire
Flammability 6 (1975) 254273.
[2] G. Heskestad, Scaling the interaction of water sprays and ames, Fire Saf. J. 37
(2002) 535548.
[3] G. Heskestad, Extinction of gas and liquid pool res with water sprays, Fire
Saf. J. 38 (2003) 301317.
[4] T.M. Jayaweera, H.-Z. Yu, Scaling of re cooling by water mist under low drop
Reynolds number conditions, Fire Saf. J. 43 (1) (2008) 6370.
[5] H.-Z. Yu, X. Zhou, B.D. Ditch, Experimental validation of Froude-modelingbased physical scaling of water mist cooling of enclosure res, in: Proceedings of the 9th International Symposium on Fire Safety Science, 2008,
pp. 553564.

H.-Z. Yu / Fire Safety Journal 47 (2012) 17

[6] H.-Z. Yu, Froude-modeling-based general scaling relationships for water


sprayreplume interactions, in: Proceedings of the 7th Asia-Oceania
Symposium on Fire Science and Technology, Hong Kong, 2007.
[7] H. Schlichting, Boundary-layer Theory, 6th edition, McGraw-Hill, New York, 1968.
[8] M.C. Yuen, L.W. Chen, On drag of evaporating liquid droplets, J. Combust. Sci.
Technol. 14 (1976) 147154.
[9] J.P. Holman, Heat Transfer, 4th edition, McGraw-Hill, New York, 1976.

[10] A.H. Lefebvre, Atomization and Sprays, Hemisphere Publishing, New York, 1989.
[11] I. Glassman, Combustion, 3rd edition, Academic Press, San Diego, California,
1996.
[12] H.-Z. Yu, A revisit of Froude-modeling-based physical scaling of re suppression by water sprays, in: Proceedings of the Suppression and Detection
Research and Applications: A Technical Working Conference (SUPDET 2009),
Orlando, Florida, February 2427, 2009.

Você também pode gostar