Loading...
Please wait, while we are loading the content...
Similar Documents
Modeling the Effect of Water Spray Suppression on Large-scale Pool Fires
| Content Provider | Semantic Scholar |
|---|---|
| Author | Ș. |
| Copyright Year | 2003 |
| Abstract | In practical fire suppression systems for large rooms or compartments, water sprinklers are often located on or near the ceilings. In this configuration, the droplets from the water spray must be large enough to penetrate the high temperature thermal plume of the tire and reach the pool surface, yet be small enough to evaporate and provide effective suppression of the flame zone. The focus of this research is to investigate computationally the influence of initial drop size and spray system configuration on water suppression of medium to large scale (Le., -100kW) pool fires. The cases addressed are representative of ongoing experiments at the Naval Research Laboratories. The tire scenario examined in this study involves a 123 kW heptane pool fire located in a 3.05m (10 ft) cube enclosed facility. Results from this study indicate that ( 1 ) for large drops (Dd > 150 pm) an initial rise in temperature is observed associated with enhanced turbulent mixing before evaporative cooling takes place; (2) an optimum drop size is found that allows for maximum decrease in gas-phase temperature for one of the spray configurations examined: and (3) a low pressure spray with more nozzle locations appears to provide improved suppression when compared to using a single high pressure nozzle. INTRODUCTION The purpose of this research is to explore the use of numerical modeling for simulation of fire suppression using water sprays. This capability is useful to assess the performance of halon alternatives as driven by the ban on halon production. One attractive alternative for halon is the use of water sprays. The main physical mechanisms for flame suppression using water sprays include the effects of thermal cooling due to evaporation and gas phase heat capacity, oxygen displacement, and radiation attenuation due to the liquid spray. A partial summary of the research in the use of water sprays for fire suppression can be found in the reviews by Tatem et al. [l], Jones et ai. [2], Mawhinney et al. [3], and Grant et ai. [4]. More recent numerical studies in this area can be found in the works of Chow et al. [5], Novozhilov et al. [6], and Prasad et al. 17, 8,9]. Numerical suppression studies of Prasad et al. focused on the optimization of water sprays for the suppression of low Reynolds number (i.e., laminar) jet diffusion flames [7,8] and pool fires [9]. The results of these studies demonstrate that suppression efficiency is highly dependent on the size of the water spray droplets as well as the location of the injection. Optimal suppression was observed when small drops are injected upward through the base of the flame. These conclusions are also supported in the work of Novozhilov et al. [6] and Chow et al. [5] for enclosure fires. Novozhilov et al. [6] show reasonable agreement with experimental data by only considering the thermal effects of the spray, thus indicating that the chemical effects of water vapor on exothermic reaction kinetics are not significant. Lentati et al. [IO] support this idea in their study examining the thermal and chemical suppression effects of water spray in a counter flow diffusion flame. Their results also indicate that the chemical suppression effects of water vapor contribute little (less then 10%) to the overall temperature drop indicating that thermal cooling due to evaporation of the water spray accounts for the majority of the suppression. The complete understanding of water spray suppression of a turbulent fire involves resolving the intimate coupling of liquid evaporation, turbulence, finite rate chemical kinetics, and radiation heat transfer. The objective of this work is to account directly for the first three of these physical processes through subgrid modeling of pool fires for a scenario of engineering interest. 262 Halon Options Technical Working Conference 2-4 May Zoo0 A detailed description of the subgrid models developed to account for thermal cooling (spray submodel) and chemical kinetics effects (PSR submodel) is provided in the Mathematical Model Formulation section. Results are then presented for a square (0.305 m [ 12 in]) 123 kW heptane pan fire, which is representative of tests at the Naval Research Laboratory (NRL) [ I I]." Nutnerical predictions are presented to examine the sensilivity of volume averaged temperature in the test cell to initial drop size specification and suppression system configuration. MATHEMATICAL MODEL FORMULATION The following mathematical description is limited to a summary of the spray and flame extinguishment model formulations used to account for the thermal and chemical effects of a liquid suppressant. respectively. The models are implemented into a general purpose fire simulation code, VULCAN. which is based on the KAMELEON-Fire code [12]. VULCAN uses a RANS based model suite including a k-e turbulence model [ 131, the EDC combustion model [ 141. a soot model [ I 51, and a radiation model [ 161. The gas phase conservation equations are discretized on a staggered, block-structured grid with second-order upwind differencing for the convective terms using a version of the SIMPLE algorithm [ 171. Previous studies using VULCAN for pool fire simulations can be found in references [ 18, 19, 201. SPRAY SUBMODEL The spray submodel is based oil a stochastic separated flow approach 1211. The following transport equations for mass, momentum, and energy are integrated in time for groups of droplets (parcels). " I dt Sc, -dm'l -nD, ,p , -B,,,Sh, In Eq. ( I ) , the Sherwood ( S h , = h#,,,,Dzf /D,,,, )and Nusselt ( N u , = ~ I , ~ D , , / k , )transfer numbers are expressed in terms of film conditions using the Ranz-Marshall correlations [22], I / Z SI?, = 2 log( 1 + B,, ) / B,r, [Re ,, Sc , I" / 3 1 N u , = 2 log(l+B,~,)/B,,,[Re,"'Pr,"'/3] " Mardnghides, A,, Anleitner, R.L., Binnette, C., Austin, E.M. and Sheinson, R.S., Resrrlt.s,fi~~Self' Contui~?rd Totul F/ooding Nulou 1301 A1ter.wtiw Technologies E~uluurion, NRL report. In progress, 2000, draft on file with the author. Halon Options Tcchnicai Workins Conlcrcnce 2-4 May 7000 263 The coefficient of drag (CD) is modeled as that of a sphere using the relations from Shuen et al. ~231, 24(1+Red2”/6)/Re, forRe, < I O 0 0 forRe, > 1000 cD = (0.44 The thermodynamic properties at the droplet film surface are obtained by using a thin-skin approximation [21] where the film temperature is approximated as a weighted average of the droplet and surrounding gas temperatures, Ti = aT, + (1 -a)T,. In this study, (3) a = MIN(C,, Bi,l) where Bi (= h,D, /kf ) is the thermal Biot number. The model constant CBi (set equal to 0.5) defines the transition Bi number for which the droplet can be treated using a lumped capacitance approach (Le., T, z T,). This model is used to account for rapid changes in droplet temperature as the droplet is transported into a flame zone. Attempts to use a simple 1/3 rule weighting [24] results in non-physical values of temperature and species mass fractions. The film is assumed to he at saturation conditions so that a partial pressure can be calculated using a Clausius-Clapeyron relation, Pf = P,cq Explh,,, lR(1 / T , 1 /T,<, )I where heat of vaporization, h,, , is expressed as a function of temperature using Watson’s relation 125, 261 (i.e., 4, =h,,,., [(T, T, ) / (Tc T,, )11).3x). Once the partial pressure is determined, the mass fraction of water vapor is calculated from the ideal equation of state, Y , = MW,?, / MW, (P, / P, 1 + MW,,, I MW, ). Lastly, the mass ( Bnr ) and thermal ( B, ) transfer numbers in Eq. (1) are obtained from their definitions as derived from steady-state droplet analysis [27]: Bn, = Droplet dispersion due to turbulence is implemented using both parcel and subparcel models. The parcel model accounts for the effects of large-scale turbulent eddies perturbing a parcel trajectory and is based on the random walk model of Gosman and Ioannides [28] as modified by Shuen et al. [23]. Turbulent dispersion of the droplets within a parcel is accounted for using the group modeling concept of Zhou and Yao [29] where the spatial distribution of droplets within each parcel is assumed to have a Gaussian distribution. Y , ) /CY, 1) and B, = C , (Tg -T, ) /h Ig . FLAME EXTINGUISHMENT SUBMODEL To account for the first order effects of the water spray on the exothermic chemical reactions in a flame, a subgrid model was developed based on Perfectly Stirred Reactor (PSR) theory as formulated by Glarborg et al. [30] in the CHEMKIN I1 [31] software package. The model is constructed through a sequence of PSR precalculations that map out chemical extinction time scales as a function of temperature and suppressant mixtures. For the current study, the suppressants include H,O, C02, N 2 , and combinations. In general, extinction times are functions of suppressant mixtures, temperature, and pressures and could be used in a tabulated lookup form. However, in order to reduce the storage requirements of such a table, mixing rules are used that allow the mixture mole fraction of suppressant to be determined using the following expression from Saito et al. [32]: 264 Halon Oplions Technical Working Conference 2-4 May Zoo0 |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | https://ws680.nist.gov/publication/get_pdf.cfm?pub_id=909452 |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
