The Effect of Spray Refinement
on Flame Extinguishment
Alexander Snegirev, A. Lipjainen, V. Talalov
St-Petersburg State Polytechnic University
Russia
Acknowledgements
• Microsoft Research (FireEx-MicroTEST project 2007–2009)
• Science and Higher Education Committee of the St.-Petersburg Administration (A. Snegirev, 2009, and A. Lipjainen, 2008)
• RFBR 10-08-08019-з
All models are wrong, but some are useful
(George E. P. Box)
1 University of Brighton, 19 April 2010
• Water mist fire suppression system (WMFSS) is a rapidly growing technology of fire protection
• Controversial debates in fire engineering community: completely opposite conclusions on water mist efficiency
• A likely reason is that WMFSS is applied in different regimes
• To classify the regimes of the flame-spray interaction, governing dimensionless criteria should be identified
• A validated CFD methodology is required to complement and, wherever possible, replace full-scale tests
• URANS: Yoon, Kim, DesJardin, Hewson, Tieszen and Blanchat, 2007, 2008
• LES studies of reacting multiphase flows with spray evaporation: Xia, Luo, Kumar, 2008; O’Grady and Novozhilov, 2009
Motivation
2 University of Brighton, 19 April 2010
Objectives
• To develop and validate an LES methodology to predict flame suppression by water sprays
• To identify a dimensionless criterion suitable to distinguish fine and coarse sprays taking into account their structure and dynamics and governing criteria reflecting basics of flame-spray interaction regimes
• To assess the effect of spray refinement on flame suppression
University of Brighton, 19 April 20103
Water as an extinguishing agent
Fire suppression concepts
Total flooding Surface application
Halocarbons
Inert gases
Water mist
Foams
Powders
Water jetWater spray
Aerosols
Damage due to excessive water
Not suitable for combustible liquids
Motivation
to refine
water spray
4 University of Brighton, 19 April 2010
• What happens to the spray when it is refined due to increased pressure?
• It may become not a ―surface application‖ agent, yet not a ―total flooding‖ one
Modelling methodology
• Fire3D model and code
• Turbulence
• Combustion
• Thermal radiation: Monte Carlo + WSGG (CO2, H2O, soot)
• Stochastic Lagrangian spray model
2
3 2SGS S
GC S C
S 2 ij jiS S S
1
2
ji
ij
j i
uuS
x x
min ,S Sl y C
2
2
O
,
O
min , 1fuel fuel ext vap
SGS
Yr C Y P
s
1
SGS S
Pr
t i
t i
gG
x
Subgrid turbulence production
due to buoyancy
Probability of local extinction due to
excessive vapour concentration
5 University of Brighton, 19 April 2010
Buoyant flame modelling
Simulations of a 260-kW flame above a 1-m diameter burner (fuel is CH2O): a) — steady mean temperature field (URANS, mesh 64×64×80); b) —
instantaneous resolved temperature; c) — time averaged temperature; d) — flow vortical structure
as shown by the instantaneous iso-surface Q = S2–Ω2 = const (LES, mesh 88×88×128)
a) b) c) d)
Time averaged axial temperature (a) and
velocity (b). Case of a 260-kW flame above a 1-m
diameter burner (fuel is CH2O)
a)
b)
6 University of Brighton, 19 April 2010
Spray model components
• Droplet movement
• Droplet dispersion by turbulence
• Droplet heating
• Droplet evaporation
• Inter-phase exchange
• Nozzle performance (initial droplet size and velocity distribution)
and
• Flame extinction
• Spray-pool interaction
7 University of Brighton, 19 April 2010
The spray model
,
, ,
31
4
p i Dp i i p i i i
p l l
du Cu u u u g
dt d
,
,
p i
p i
dxu
dt
2/324 11 Re ,Re 1000
Re 6
0.424 ,Re 1000
p p
pD
p
C
Droplet movement
Droplet dispersion
, min , 2 3p SGS p S SGSl k
*
*
*
ln 1 ,
,
SD S D p p
D p pp
S D p p
ll u u
u u
l u u
3/4 3/2
S S
S
l Cl
C k
2
2
1exp
22
ii
uf u
2 3SGSk i i iu u u
Subgrid eddy life time
model
SGS SGSk S
4
3
p l
D
D p
d
C u u
ρl >> ρg, spherical drag law
Spherical drag law
8 University of Brighton, 19 April 2010
The spray model
Droplet heating
,
0 ,
p
p p vap p p boil
p l
p boil
dmdT q h T T T
m c dtdt
T T
, ,
boil
p
T
vap p vap boil p lT
h T h c T dT
, ,p p conv p radq q q
, ,NuPr
g
p conv p P g pq d c T T
1/2 1/3Nu 2 0.6Re Prp
2 4
,4
p rad p p p
Gq d T
4
0
G Id
Droplet evaporation
Shln 1 ,
Pr
,
p g m p boil
p
p
p boil
vap boil
d B T Tdm
qdtT T
h T
1/2 1/3Sh 2 0.6Re Scp
, ,
,1
vap sat p vap
m
vap sat p
Y T YB
Y T
The
classical
model with
Le ≈ 1,
cP, vap ≈ cP,g
9 University of Brighton, 19 April 2010
Convective
and
radiative
incident
flux
The spray model
Inter-phase exchange
p p
M
n dmr
V dt
p
V p p
n dr m u
V dt
p
H p l p
n dr m c T
V dt
Source terms in gas
transport eqs
Spray atomization (cone angle φ)
00
2
,0 ,0
,0 2
1exp
22
p p
p
VV
V Vf V
50
50
exp ln 2
ln11 erf
2 2
v
v
d
d
dR d
d d
,0 02p lV P
Random choice of
velocity direction
and magnitudeRR or LN initial droplet
size distribution
10 University of Brighton, 19 April 2010
The spray model
Spray-surface interaction• Droplet splash evaporation• Droplet escape
Two extreme
limits
Flame extinction
2
2
ln1 1exp
22
vap vap
vap
vap vapvap
Xf X
X
lnvap vapX 2 2ln 1 0.3vap vap vapX X
,
, , vap cr
ext vap vap vap crX
P P X X f X dX
2
,ln 21erfc
2 2
vap cr vap vap
vap
X X
Probability of
the super-
critical vapor
concentration
Presumed
PDF for
vapor mole
fraction to
allow for
turbulent
fluctuations
0.27
11 University of Brighton, 19 April 2010
Nozzle spray characterization
00
2
l
PV
220
0 0 0 04
w
DQ V K P D P
1/32
1/350 0 0
1/3 1/3
0 0 0
1Wev l
l
d V DC C
D P D
2 2
0 050 1/3 2/3
0
v
w
D Dd
P Q
K-factorWater flow ratePressure drop Nozzle diameter
Initial droplet
velocity
Initial median
droplet diameter
Initial median
droplet diameter
depends on
pressure and
flow rate
[Yoon et al, 2004]
[SAND2007 3220]
12 University of Brighton, 19 April 2010
Model validation
• Fire Laboratory for Accreditation of Models and Experiments (FLAME) – Sandia Labs, California, USA
• 2.5 MW pool fire (JP8 – C11H21)
• 30º cone angle, 5 m height
• Spray just covers the pool
90º nozzle
0P , kPa lQ , l/min 0V , m/s 50vd , mm We , 103 Re , 10
3
172.4 53.0 14.7 0.870 26.3 143.9
1310.0 147.6 41.0 0.370 204.1 400.9
13 University of Brighton, 19 April 2010
Model validation
Extinguishing a 2500-kW flame above a 2-m diameter burner. Left to right: 0.05, 0.10, 0.15,
0.20, 0.25 s after nozzle activation (two right plots show the same time instant). Light colour
surface — vapour mole fraction 1%, dark colour surface — temperature 1000 K. Instantaneous
resolved temperature is shown in the axial plane
Low-pressure
spray:
no suppression
High-pressure
spray:
suppression
14 University of Brighton, 19 April 2010
Fine
sprayCoarse
spray
When the spray is really fine?
• Initial droplet size is a key parameter that switches extinguishment regimes
• Single value of characteristic droplet diameter is not enough to classify spray type (NFPA 750: dv99 < 1 mm)
15 University of Brighton, 19 April 2010 00:28
Fine spray is special
• Spray structure and dynamics changes dramatically due to spray refinement
• Focused jet instead of wide angle cone
• Faster evaporation, vapor cloud surrounds jet
Coarse, medium and fine water sprays in no flame conditions (flow rate 10 l/min, log-normal initial droplet size distribution, σ = 0.48): a) ― dv50 = 0.5 mm, b) ― dv50 = 0.2 mm, c) ― dv50 = 0.08 mm. Iso-surface shows vapour
mole fraction of 0.004 (a) and 0.01 (b and c)
Coarse spray:
maintains a
cone
shape until for a
long distance
away from the
nozzle; the
vapour cloud is
inside the spray
near its axis
Fine spray:
forms a thin,
focused jet
surrounded
by the
vapour cloud
S0=0.6 S0=0.15 S0=0.035
16 University of Brighton, 19 April 2010
Fine spray is special
Photographs of atomizer sprays.
Parameters of water
upstream of the atomizer: 8 MPa,
20°C and 170°C
17 University of Brighton, 19 April 2010
20°C
170°C
Time and length scales
• The deceleration (stopping) time is much smaller than the evaporation time
Evaporation timeDeceleration time
Deceleration length
≡ Stopping distance
18 University of Brighton, 19 April 2010
Spray fineness criterion
• The deceleration length can be estimated neglecting evaporation:
• Deceleration length depends not only on d0 but also on V0
• Governing criterion:Spray fineness number
0 0 04
3
ld
d V dL f
0 00Re
V d
0
0
Re0
0 Re 00
ReReRe
Re Re 24D
df
C
0d
f
LS
H L
Stokes number
19 University of Brighton, 19 April 2010
Regimes of flame-spray interaction
• Spray fineness number distinguish qualitatively different regimes of spray interaction with the flame
• Large Stokes number => Spray drag number, SD
• Small Stokes number => Spray momentum number, SM
Extinguishing a 260-kW flame above a 1-m diameter burner (0.2 s after nozzle activation):
a) — coarse spray, dv50 = 0.5 mm; b) — medium spray dv50 = 0.2 mm; c) — fine spray 0.08 mm.
Wide-angle nozzle (120° cone), nozzle height 3.0 m above the floor, water flow rate 10 l/min.
LES, mesh 88×88×128. Light colour surface — vapour mole fraction 0.01.
Coarse spray:
Drops penetrate
inside the flame
and evaporate
there =>Spray droplet drag
Plume momentumDS
Fine spray: vaporized jet suppresses the flame provided jet momentumis high enough
Spray momentum
Plume momentumMS
S0=0.6 S0=0.15 S0=0.035
20 University of Brighton, 19 April 2010
Coarse spray
Fine spray
a) b) c)
The effect of spray refinement
260 kW, 1 m diam., wood volatiles CH2O, mesh
88×88×128, water flow rate 10 l/min
Coarse
spray,
dv50 =
0.50 mm
Medium
spray,
dv50 =
0.20 mm
Fine spray, dv50 = 0.08 mm Time step 0.05 s
No suppression
Suppression
21 University of Brighton, 19 April 2010
Conclusions and future work
• The spray is fine when the Stokes number is low. The Stokes number is the spray fineness number
• Fine spray focuses in a narrow jet => the entire spray momentum is accumulated in the jet suppressing the flame
• This is neither a surface application, nor a total flooding regime
• Two opposite cases of coarse and fine sprays represent two physically different regimes of spray-flame interaction
• The structure and dynamics of the coarse spray is determined by the drag force due to droplet friction in the gas flow => Governing criterion is the dimensionless spray drug
• The structure and dynamics of the fine spray is governed by the total spray momentum => Governing criterion is the dimensionless spray momentum
22 University of Brighton, 19 April 2010
Conclusions and future work
• Drop size reduction (due to higher pressure) may cause faster flame suppression at lower water supply rate
• Necessary conditions:
– sufficient spray momentum and
– proper spray orientation
• Unresolved theoretical problems: flame extinction, interaction with fuel surface, attenuation of radiant heat
Future work
• Further validation against full-scale measurement data
• More sophisticated spray atomization model
• A universal spray model?
23 University of Brighton, 19 April 2010
Experience is what you get when you
didn’t get what you wanted(Randy Pausch)