Professional Documents
Culture Documents
Wp030 Computational Study Effects of Dropsize Distribution in Fire Suppression
Wp030 Computational Study Effects of Dropsize Distribution in Fire Suppression
Corresponding Author
Introduction
As long as fire suppression and extinguishment has protected our lives and property, we have been devising
ways to deliver the suppression agent to the fire. Ever since the fire sprinklers were invented, the technology and
complexity of extinguishment methodology has grown at a truly exponential rate. Today, there are literally hundreds, if not thousands of ways to dispense the extinguishing agent. The innovation and diversity of these systems is
seen to be driven by application to solve specific fire problems.
With spray nozzles, extensive design work over many decades has produced highly engineered devices in innumerable types and styles. Engineered sprays have been developed for many thousands of applications and industries, and this has also been largely driven by specific needs. Many spray styles developed originally for other purposes have been applied to fire protection. Setting aside pneumatic atomization or dry chemical extinguishment
sprays, consider that hydraulic atomizing and impingement nozzles are available in hollow cone, full cone, spiral,
flat, or even square and oval spray patterns. The selection that is commercially available is very broad and can
create considerable confusion for engineers looking to integrate the proper spray nozzle into an extinguishment or
suppression system. With single orifice, cluster heads, and spray angles ranging from 0 to 360 in common use, the
fire system designer must determine which type of nozzle works in different fire hazard applications.
With each general application and set of installation parameters, approval agencies and the Authority Having
Jurisdiction (AHJ) require only specific coverage or spray angle and flow density, which are believed to be the most
relevant parameters of spray performance in various situations.
As the applications become more specific to water mist fire suppression, the additional factor of drop size is
taken into consideration. However, there is no current water mist fire protection standard that requires listings to
include basic drop size information such as the Volume Median Diameter (VMD) or Sauter Mean Diameter (SMD).
Rather, water mist systems must be listed for specific fire hazards requiring live fire performance validation, a time
consuming and financially burdensome endeavor.
The potential alternative to physical testing is to use advanced computational analysis to design, build, performance test, and certify a fire suppression system before a pipe is ever laid. Computational simulation of water mist
fire suppression though is an incredibly complex undertaking. Modeling the physics of water droplet and flame interaction, heat absorption and evaporation, vapor displacement, fuel combustion, and temperature reduction are just
part of what needs to be analyzed to accurately predict the fire suppression capabilities of a water mist protection
system.
One of the more important criteria, however, was also the most overlooked and ignored. In the past, in majority
of simulation efforts, water drop size distribution statistics were simplified down to the minimal amount of data required to convey the maximum information possible. In some cases a single parameter, such as VMD, was the only
statistic used.
Perhaps this is because drop size information can be considered to be confusing and overly complex. This is
generally the case with fire protection when discussing sprinkler and nozzle drop size statistics. After all, the research focuses on system suppression and extinguishment characteristics and not spray nozzles. But, is it proper and
accurate to fully characterize a water mist or other spray nozzle in such an abridged manner?
Fire protection designers have begun to realize that nozzles cannot be simplified down to a single number such
as a representative diameter. Whether that specification is the VMD, SMD or DV0.9, it is not an accurate representation of the entire spray field or spray distribution. Despite this recognition, there are such limited standards and certification guidelines to work from that the system designers and modelers are basically picking whatever statistic is
readily available to them without appreciating the inherent limitations and problems of doing so. Can parameters be
used that are sufficiently meaningful to completely and properly characterize the spray?
Studies have been conducted in recent years incorporating more comprehensive particle size characterization
within numerical simulations. [1]. Many studies however are not adapting the simulations to account for the variation of particle size distributions between different types of spray nozzles, though it is well understood within the
industry how nozzle type, flow rate, pressure and other variables can dramatically change the resulting drop size
distributions.
Our purpose here is to utilize computational fluid dynamics (CFD) to assess the fire suppression sensitivity to
realistic drop size distribution in hydraulic atomizer nozzles. Performed analysis consisted of three different types
of hydraulic atomizer nozzles with identical supply pressure and flow capacity conditions under a realistic geometrical and heat release rate environment.
Hydraulic Atomizers
In previous work by the authors [2,3], several hydraulic atomizer nozzles were tested at a typical operation
pressure(s) for each style nozzle. See Figs. 1-3 and Table 1. For the purposes of this initial study, the liquid flow
rates were normalized between each nozzle to make the spray characteristics the main variables on which suppressive and evaporative conditions would depend on.
Swirl type atomizer. See Figure 1. This atomizer features an internal core through which the liquid flow is directed with some tangential velocity component. The liquid is then forced through an exit orifice in a hollow cone
pattern.
Flow Rate
Velocity
Spray Angle
Dmin (DV0.01)
Dmean (DV0.50)
Dmax (DV0.99)
n*
Units
Swirl Type
Atomizer
Full Cone
Nozzle
Spiral
Nozzle
m3/s
m/s
deg.
m
m
m
--
1.367 x 10-4
36
78
29
96
165
3.5
1.367 x 10-4
28
68
44
334
823
2.1
1.367 x 10-4
33
120
50
151
249
3.8
(1)
mum, mean and maximum diameters (Dmin, Dmean and Dmax respectively) which were based on Rosin-Rammler derived DV0.01, DV0.5, and DV0.99 respectively.
Ceiling, floor and Pressure Outlet boundaries had Escape DPM-BC's. The droplets that made a contact with
such boundary vanished in the computational domain without any further effect to the momentum and energy. The
fire pool had a Wall-Jet DPM-BC. In this BC, droplets impinge onto the wall and depending on their momentum,
form a small liquid film or reflect. This DPM-BC is appropriate for high-temperature walls [5,6].
Initially, steady state simulations were performe (Figure 5 below) to assess the 2D axisymmetric simulation
with natural convection. Heating (no spray) and cooling (spray active) periods were computed.
In transient simulations, non-spray case (heating period) was computed for 30 seconds with iterative time advancement while the time step (dt) was 0.01 seconds. The three cases had spray periods each 15 seconds long and
reheat periods each 5 seconds long (see Figures 7-9). They were performed with the same time advancement as
non-spray case. The droplets were injected with the same time step as the main run simulations (dt=0.01 s). In the
transient simulations, especially in heating up period, formation and rise of air classical thermals [7] were observed.
Conclusions and Recommendations
Our purpose here has been to utilize computational fluid dynamics to assess the fire suppression sensitivity to
realistic drop size distributions. Three different hydraulic atomizer nozzles were used with identical supply pressure
and flow capacity conditions to examine the heat absorption capabilities and evaporation rates.
Via comparison, Figures 7-9 shown that the swirl type atomizer with the smallest VMD provided the highest
evaporation rate and the most efficient heat absorption. Though the full cone and spiral nozzles have much greater
drop sizes and those particles show the ability to penetrate the high heat zone, the smaller particle sizes generated by
the swirl type atomizer greatly reduce the overall heat, providing a better overall level of fire suppression.
Further work will be conducted to broaden the scope of these fire suppression simulations. This study was limited to axisymmetric simulation of a surface heat release rate environment. Additional studies are expected to incorporate combustion simulations in a 2D and further progress into 3D environment with multiple injection points.
The authors urge caution when including particle size distributions in fire suppression modeling. It is important
to consider the different types of spray nozzles that can be used in various fire suppression applications and the critical variations in spray field properties that affect the results for a system in application. In addition, probing deeper
with differences within the spray field such as velocity, momentum, radial distance, fluid pressure, and nozzle discharge coefficient may provide some insight when systems or models do not perform as expected.
References
1. Blanchard, E., Boulet, P., Desanghere, S., Cesmat, E., Numerical Simulation of Radiation Propagation Through
Water Mist, Ninth International Water Mist Conference, London, United Kingdom, September 2009
2. Tanner, G.A. and Knasiak, K.F., Spray Characterization of Typical Fire Suppression Nozzles, Third International Water Mist Conference, Madrid, Spain, September 2003
3. Tanner, G.A. and Knasiak, K.F., Water Mist Simplification Effects on Fire Suppression Modeling: A Challenge
to the Industry, ILASS-Americas, 20th Annual Conference on Liquid Atomization and Spray Systems, Chicago,
May 2007
4. LeFort, G., Marshall, A.W. and Pabon, M., Evaluation of Surfactant Enhanced Water Mist Performance, Fire
Technology, Volume 45, Number 3, 2009
5. ANSYS FLUENT 12.0 - User's Guide, ANSYS, Inc., Northbrook, IL, 2009
6. ANSYS FLUENT 12.0 - Theory Guide, ANSYS, Inc., Northbrook, IL, 2009
7. Bejan, A., Convection Heat Transfer, 2nd Ed., New York: John Wiley and Sons, 1995
Figure 6. Initial heating period (no spray). Top row shows temperature contours in whole domain. Middle row shows temporal temperature contours at center
axis and at the fire pool edge. Bottom row shows temporal temperature traces at various heights for center axis and the fire pool edge.
Figure 7. Cooling and reheating periods for swirl type nozzle. Top row shows temperature contours in whole domain. Middle row shows temporal temperature
contours at center axis and at the fire pool edge. Bottom row shows temporal temperature traces at various heights for center axis and the fire pool edge.
Figure 8. Cooling and reheating periods for full cone type nozzle. Top row shows temperature contours in whole domain. Middle row shows temporal temperature contours at center axis and at the fire pool edge. Bottom row shows temporal temperature traces at various heights for center axis and the fire pool edge.
Figure 9. Cooling and reheating periods for spiral type nozzle. Top row shows temperature contours in whole domain. Middle row shows temporal temperature
contours at center axis and at the fire pool edge. Bottom row shows temporal temperature traces at various heights for center axis and the fire pool edge.
10