Available online at ScienceDirect. Cunfeng Zhang a,c, Wanki Chow b, *

Transcription

1 Available online at ScienceDirect Proceia Engineering 6 ( 013 ) The 9 th Asia-Oceania Symposium on Fire Science an Technology Numerical stuies on the interaction of sprinkler an smoke layer Cunfeng Zhang a,c, Wanki Chow b, * a School of Urban Construction an Safety Engineering, Nanjing University of Technology, Nanjing, Jiangsu, China b Research Centre for Fire Engineering, The Hong Kong Polytechnic University, Hong Kong, China c State Key Laboratory of Fire Science, University of Science an Technology of China, Hefei, Anhui, China Abstract Sprinklers are the most reliable metho available for controlling builing fires. Due to the cooling effect by water spray an rag force prouce by the water roplets, sprinkler spray can lea to the loss of stability of the stratifie smoke layer. In this paper, the cooling effect of water spray is stuie. The smoke layer stability an the mass flow rate are also stuie. The Fire Dynamics Simulator (FDS) coe, base on the concept of large ey simulation, is aopte in the present simulation. It has been foun that the temperature ecrease was almost linear to the working pressure of the sprinkler system. The imensionless pressure to the smoke penetration epth can be expresse as power function. The spray has great effect on the smoke movement insie the compartment. 013 Publishe International by Elsevier Association Lt. for Selection Fire Safety an/or Science. peer-review Publishe uner responsibility by Elsevier Lt. of the Open Asia-Oceania access uner Association CC BY-NC-ND for Fire license. Science Selection an Technology. an peer-review uner responsibility of the Asian-Oceania Association of Fire Science an Technology Keywors: Sprinkler; Stability; Mass flow; Drag force; Cooling Nomenclature A surface area (m ) m b mass loss/prouction rate per unit volume (kg/m 3 ) c specific heat (J/kg K) p Pressure (N/m ) C D rag coefficient p sprinkler working pressure (N/m ) C s Smagorinsky constant P ˆ imensionless pressure (N/m ) C sp sprinkler proportionality constant q total heat flux (convective an raiant) (W/m ) roplet iameter ( m) q r raiant heat flux (W/m ) m volume meian roplet iameter ( m) q c heat release rate per unit volume (W/m 3 ) n sprinkler orifice iameter (m) q D i iffusion coefficient for species i (m /s) S penetration epth (m) heat absorption rate by roplets per unit volume (W/m 3 ) fd rag force (N) S ˆij strain rate tensor g gravity vector (m/s ) Ŝ imensionless smoke penetration epth h convective heat transfer coefficient (W/m K); smoke layer thickness (m) t Time (s) h i enthalpy for species i (J) u Velocity (m/s) h m mass transfer coefficient u sp roplet velocity at sprinkler orifice (m/s) h s enthalpy for smoke (J) We Weber number * Corresponing author: Tel: ; Fax: aress: bewkchow@polyu.eu.hk International Association for Fire Safety Science. Publishe by Elsevier Lt. Open access uner CC BY-NC-ND license. Selection an peer-review uner responsibility of the Asian-Oceania Association of Fire Science an Technology oi: /j.proeng

2 454 Cunfeng Zhang an Wanki Chow / Proceia Engineering 6 ( 013 ) h v heat of evaporation (kj/kg) Y mass fraction k thermal conuctivity (W/ m K) Y s water vapor mass fraction at saturation conitions m Mass, mass flow rate (kg, kg/s) Greek Symbols turbulence energy issipation rate (J/ kg s) σ surface tension of water (N/m) gri cell length (m) i,j stress tensor Density (kg/m 3 ) μ ynamic viscosity (N s/m ) σ log-normal istribution coefficient Rosin-Rammler istribution exponent Subscripts roplet i,j i th, j th species; tensor components D rag s smoke 1. Introuction It is require by the fire regulations of China mainlan to install automatic sprinkler systems in builings such as hotels, factories an shopping malls. However, ue to the cooling effect by the water spray, the buoyancy of the hot smoke layer, which supports the stratification, woul ecrease. On the other han, the rag force prouce by the water roplets woul also rag the smoke ownwar. These two factors both can lea to the loss of stability of the stratifie smoke layer. The smoke woul fall own to the floor, which is call smoke logging. It is a risk to the smoke ventilation an human evacuation uring a fire [1-11].Therefore, it is essential to carefully stuy the interaction of a smoke layer with a sprinkler. The problem on the stability of smoke layer uner water spray was first iscusse by Bullen [1]. The smoke layer was consiere as a constant thickness. The sprinkler spray was taken as spherical roplets with constant iameter calculate by the sprinkler pressure. A physical parameter known as the rag-to-buoyancy ratio for the entire smoke layer was taken as the criterion for its stability. Morgan an Baines further evelope the moel by incluing the convective heat transfer between the sprinkler spray an fire smoke [-3]. A istribution function of sprinkler roplets was also inclue by Morgan []. In Cooper s moel, it was consiere that the smoke layer element of unit volume below the sprinkler nozzle was pulle own by the rag force of sprinkler roplets an pushe up by its own buoyancy [4]. Due to the cooling effect an rag force of water spray, the venting system woul not be so effective [5-7]. The smoke behaviour in the compartment woul also be affecte. Due to the rag force of water roplets, the mass flow pattern woul change [8]. There are many numerical simulations on the fire smoke an spray. However, these simulations are mainly stuie on the interaction of fire plume an the spray [9-1]. The stability of smoke is selom stuie in the numerical simulation. There are also some experiments on the interaction of spray an smoke [1-17]. Due to the cost of experiments, few experiments coul be one. To stuy the phenomena on the spray an fire smoke, CFD moel is a goo choice [18]. In this paper, the cooling effect is stuie. The smoke layer stability an the mass flow rate are also stuie. The Fire Dynamics Simulator (FDS) coe, base on the concept of large ey simulation, is aopte in the present simulation.. Mathematical moel an parameters Fire Dynamics Simulator version (FDS) which is evelope by NIST was use here for the numerical simulation [19]. The moel solves numerically a form of the Navier-Stokes equations appropriate for low-spee, thermally-riven flow with an emphasis on smoke an heat transport from fires. The core algorithm is an explicit preictor-corrector scheme that is secon orer accurate in space an time. Turbulence is treate by means of the Smagorinsky form of Large Ey Simulation (LES). Lagrangian particles are use to simulate smoke movement, sprinkler ischarge, an fuel sprays. The moel has been successfully teste for a range of fire-relate problems, most extensively for natural convective flows an smoke movement [0, 1]. The gris use in computations woul have a major impact on the simulation results. In orer to select an appropriate cell size, four gris (30 cm, 0 cm, 10 cm an 7.7 cm) were trie in the computations. The results from 30 cm cell size have the largest eviations in these computations. At the fire plume area, the results from 0 cm cells an 10 cm cells are close to each other. For the places just uner the sprinkler hea, the results are almost same except the 30 cm cells. However, for the place near the oor, the results calculate by 10 cm an 7.7 cm cells are close. So the 10cm cells were use in this paper. The same cells were also chosen by Novozhilov in the simulation of sprinkler interaction with a fire ceiling jet [18].

3 Cunfeng Zhang an Wanki Chow / Proceia Engineering 6 ( 013 ) Moel escription The following is the summary of the moel [19].The flow an energy equations (for ieal gas) are consiere in a weakly-compressible approximation: t ρ + ( ρu) = m b t + + = + + ( ρu) ( ρu u) p ρg fd τ i, j Dp ( ρhs) + ( ρhsu) = + q c q q + τi, j u t Dt (1) () (3) The term q represents the total conuctive an raiant heat fluxes: q = k T hρd Y + q (4) i i i i r There is a term in the energy equation known as the issipation rate,, the rate at which kinetic energy is converte to thermal energy by viscosity: ( ) ε τ ˆ ˆ ij u = μ Sij Sij u (5) 3 Following the analysis of Smagorinsky [], the viscosity is moele: 1/ μ μ ( ) ˆ ˆ LES = ρ CsΔ Sij Sij ( u) (6) 3 where C s is an empirical constant an is a length on the orer of the size of a gri cell. The bar above the various quantities enotes that these are the resolve values, meaning that they are compute from the numerical solution sample on a gri. Water spray is represente by an ensemble of Lagrangian particles. The force term f D represents the momentum transferre from the roplets to the gas. It is obtaine by summing the force transferre from each roplet in a gri cell an iviing by the cell volume f D 1 ρcdπ u u ( u u) = (7) 8 δxδyδz where C D is the rag coefficient, x y z is the volume of the gri cell. The momentum, mass, an energy balances for each roplet are governe by t mu = π ρcd u u( u u) + mg (8) 8 ( ) m t = A h ( Y Y ) (9) m s g T m mc = AhT ( g T) + hv (10) t t

4 456 Cunfeng Zhang an Wanki Chow / Proceia Engineering 6 ( 013 ) Here, the vapor mass fraction of the gas, Y g, is obtaine from the gas phase mass conservation equations, an the liqui equilibrium vapor mass fraction, Y s, is establishe from the Clausius-Clapeyron equation. The most important parameter of the moel is the sprinkler spray patterns. The size istribution is expresse in terms of its Cumulative Volume Fraction (CVF). The CVF for a sprinkler can be well presente by a combination of log-normal an Rosin-Rammler istributions [19] F ( ) = γ 0.693( ) m [ln( / m )] σ 1 1 e ( ) 0 m π σ 1 e ( < ) m (11) where an are empirical constants equal to about.4 an 0.6 respectively. The meian roplet iameter, m, was estimate using the formula reporte by You [3]: m C = (1) W sp n 1/3 e auspn W ρ e = (13) σ where C sp is the sprinkler constant. For the orifice iameters of 16.3 mm, 13.5 mm, 1.7 mm, the constants were approximately 4.3,.9 an.3, respectively. The above escriptions are just the outlines of the FDS moel. The etails on this moel can refer to the Fire Dynamics Simulator (Version 5), Technical Reference Guie [15]... Moel parameters The schematic iagram of the computational omain is shown in Fig. 1. The imensions of this omain are 1 m (Length) 6 m (With) 5 m (Height) with openings at both ens. At the en near the fire, the opening is m (with) m (Height). The air was supplie from this open by natural convection. At the other en, the opening is m (with) 4 m (Height) for natural smoke ventilation. For the simplicity of calculation, the omain surface is set to be aiabatic. The 1 m 1 m rectangular constant fire locate at the floor (location F0) was basically moele as the ejection of gaseous fuel from a soli surface. Two heat release rates were use: 1.0 MW an 1.5 MW. All of the surfaces in the omain were set as non-reacting soli bounary. The initial temperature of walls an ambient air were set to 0 C. Table 1. Sprinkler parameters for simulate pressure Parameter Values Pressure / Bar Flow rate / L/min Droplet volume meian iameter / μm Droplet initial velocity / m/s Upright sprinkler heas with a 1.7 mm orifice iameter (K factor equals to 80 L/(min bar 1/ )) were place 0.1 m apart from the ceiling (Location S0) in the simulation. The spray angles range from 30 to 80 as shown in Fig.. The atomization length of water spray is set to 0. m. For the working pressure of 1 bar, the meian roplet iameter, m, equals to 88 μm an the initial roplet velocity, v, is 5.58 m/s. For the other working pressure, the m an v can be calculate by the following [18]: 1/3 1 m( p) = 88 ( μm) (14) p

5 Cunfeng Zhang an Wanki Chow / Proceia Engineering 6 ( 013 ) v( p ) = 5.58 p ( m/ s) (15) Table 1 shows the information of sprinkler hea. A more etails on the sprinkler parameters can be foun in reference [18]. The p is the sprinkler working pressure in unit of bar. A summary of the sprinkler parameters are list in Table 1. It shoul be note that the escribe parameters are use as the base case, an variations of these are performe to quantify sensitivity of the moel m.0 m 4.0 m 5.0 m 6.0 m 7.0 m 8.0 m 9.5 m m.5 m.0 m SU C m m -1.0 m Air supply Plume flow Smoke flow by sprinkler Smoke Fire vent F0 T0 SL SL1 S0 SR1 SR D Fire plume Transition Area uner spray Ceiling jet irection Fig. 1 (a). Simulation omain layout an locations of thermocouple trees. 10 cm 40 cm Fig. 1 (b). Details of thermocouple tree. Fig.. Definition of spray angles. The thermocouple trees at the locations marke were use to recor temperatures of smoke, as shown in Fig. 1(a). These trees are place 10 cm below the ceiling. As shown in Fig. 1(b), the space of thermocouples in every tree is with the exception of the two thermocouples near the ceiling. They are 40 cm apart from each other. The mass flows at the two openings were recore. For the air supply opening, the net mass flow (m a1 ) was calculate. For the ventilation opening, the total mass flows in (m a ) an out (m D )of the opening were simulate. The mass flows across two planar areas, marke gray in Fig. 1(a), at the height of 3m from the floor were also calculate. The net mass flow across the left area in Fig. 1 (a) (4 m 4 m) is consiere as fire plume mass flow (m p ). The area just uner the sprinkler hea is 6 m 6 m. The mass flows up (m u ) an own (m E ) the area are both recore.

6 458 Cunfeng Zhang an Wanki Chow / Proceia Engineering 6 ( 013 ) Results an iscussion A total of 10 cases were simulate using FDS. There are groupe into two sets ue to ifferent heat release rate of fire: 1 MW an 1.5 MW. Sprinkler hea was set on after the fire starte 90 secons. The working pressure of sprinkle hea varie from 0.5 bar to 1.5 bar. Typical results of smoke evelopment are shown in Figs (a) Before sprinkler activation (b) After sprinkler activation Fig. 3. Slice of smoke temperature at ifferent conitions. (a) Before sprinkler activation (b) After sprinkler activation Fig. 4. Smoke view at ifferent conitions. (a) Detaile smoke figure before sprinkler activation (b) Detaile smoke figure after sprinkler activation Fig. 5. Temperature istribution an velocity vectors. At about 30 s after the fire ignition, a steay smoke layer was forme an the hot smoke flowe out across the upper part of the smoke vent (Figs. 3(a) an 4 (a)). A more etaile figure on temperature istribution an velocity vector was shown in Fig. 5 (a). The contour lines are parallel an nearly horizontal. There was a clear smoke-air interface at about m above the floor. From the ceiling to smoke-air interface, the temperature ecline smoothly. The velocity vector in Fig. 5 is the smoke velocity projecte at the central plan of y = 0. It can be seen from the figure that the ownwar velocity of smoke was at about 0.5~1.0 m/s when the sprinkler hea was not activate. Due to the high temperature of hot smoke, the out flow velocity was greater than 3 m/s, much higher than the ownwar smoke. There are great ifferences when the sprinkler was activate. Due to the cooling effect an the rag force on the smoke, the temperature woul ecrease apparently. The smoke woul not keep on the ceiling an move ownwar. In some situations, the smoke woul not keep layer an the smoke logging [1] woul happene. Whether or not the smoke logging happene was ecie by the smoke temperature an the spray parameters. The results shown in Figs. 3(b)-5(b) came from the case A3. The smoke temperature contour was convex towars the sprinkler hea. Comparing with the conition before sprinkler activation, the smoke temperature after sprinkler activation at the same height was low. The ownwar velocity uner sprinkler spray increase to larger than m/s. The out flow velocity i not vary apparently.

7 Cunfeng Zhang an Wanki Chow / Proceia Engineering 6 ( 013 ) Table. Temperature of smoke layer Case NO. HRR/ MW Sprinkler Pressure/bar Temperature / C Smoke layer Uner sprinkler hea Smoke layer thickness/mm penetration epth/mm A A A A A * B B B B B B *: The smoke escene to the floor Cooling effect For the convenience of iscussion, the omain is ivie into three parts accoring to the horizontal istance to the sprinkler hea: fire plume part, the area uner the water spay an the transition region between them (as shown in Fig. 1). In this part, only the area uner the water spray woul be iscusse. Accoring to the fire regulations of China mainlan, the protection area of a sprinkler hea coul not be larger than 0 m an the istance between sprinkler heas must be less than 4.5 m. The raius of water- wette area is about 3 m for the 5 m high builing. The thermocouple trees S0, SL1, SL, SR1, SR an SU are use to simulate the temperature change of smoke uner the sprinkler. The temperatures of smoke layer uner the water spray were liste in Table. These values were calculate by SL1, SL, SR1, SR an SU. The thermocouples of these trees in the smoke layer were use for average calculation. Case A0 an B0 were the values before the sprinkler hea was activate. The higher the working pressure was, the lower temperature the smoke ha. The temperature ecrease was almost linear to the working pressure (Fig. 6). 3.. Stability of smoke layer Due to the rag force of water spray, once the sprinkler was activate, the smoke woul escen. The smoke air interface surface woul not keep horizontal. The smoke penetration epth was ecie by the fire heat release rate an the working pressure of water spray. When the sprinkler pressure is low, the smoke layer is still stable. The results for the penetration epth were liste in Table. The smoke layer thickness in Table was the thickness before the sprinkler hea was activate. The penetration epth was ecie by the temperature contour lines an the smoke view after sprinkler activation. For the smoke layer thickness calculation, N-percent metho was use. For the penetration epth, just a simple temperature threshol, 40 C~45 C, was use. The smoke view was also use for reference to ai the ecision of epth. For quantitative escription of relationship on the smoke penetration epth an the working pressure an heat release rate of the fire, imensionless pressure ( P ) an smoke penetration epth ( Ŝ ) are efine as followe: ˆ Pˆ = P / ρ V 0 Sˆ = S / h (16) The reference velocity, V 0, is efine as [17, 4]:

8 460 Cunfeng Zhang an Wanki Chow / Proceia Engineering 6 ( 013 ) V gq 1/3 0= ( ) (17) ρ0cth p 0 Temperature / 0 C MW 1.5MW Pressure / Bar S/h Equation y = a*(x-b)^c 0.3 Aj. R-Square Value Stanar Error B a B b E E-1 B c P /ρ V 0 Fig. 6. Smoke layer temperature after sprinkler activation. Fig. 7. Relationship between imensionless thickness an imensionless pressure. where H is the height of the computational omain an Q is the fire heat release rate. The symbol h is the smoke layer thickness, while S is the smoke penetration epth. The results of S ˆ to the imensionless pressure are plotte in Fig. 7. The otte line in the figure is the corresponing fitting curve. As the parameter b in the Fig. 7 is very small, the curve an its corresponing correlation coefficient R for each sprinkler type can be expresse approximately as the below: Sˆ = Pˆ R = (18) 3.3. Mass flow Fig. 8. Conceptual iagram of smoke behaviour Figure 8 is a conceptual iagram of smoke behavior in the compartment uring sprinkler system activation. The fire source an the water application are intentionally separate in orer to explain the influence of roplets of water on smoke behavior. It is also assume that the compartment is in stationary state. The mass conversation coul be expresse as: m + m = m + m + m + m p u f a1 a E m + m = m + m p u D E m = m + m p f e m = m + m + m e E a1 a (19) Here, the m p is the mass flow rate of fire plume, the m u is the mass flow rate into the smoke layer across the interface, the m f is the mass loss rate of combustion material, the m a1 an m a is the total mass flow rate into the omain by the two opening an the m D is the total mass flow in the omain by the smoke vent. Because the value of m f is very small, the burning rate can be ignore. The followe equation can be euce:

9 Cunfeng Zhang an Wanki Chow / Proceia Engineering 6 ( 013 ) m = m + m (0) D a1 a The simulation results for these mass flows were liste in Table 3. The corresponing figures were shown in Fig. 9. The m u an m E are the mass flows across the area just uner the sprinkler hea. It coul be seen that the sprinkler ha some effects on the mass flow. The ownwar flow an the plume increase apparently. The upwar flow m u increase also. The inflow rate m a1, m a an outflow rate m D i not change much except for some particular working pressure. For the HRR 1MW, the particular pressure was 1 bar. An for the HRR 1.5 MW, it was 1.5 bar. It can be euce that the spray has great effect on the smoke movement insie the compartment. However, the inflow an outflow o not change much. The reason is that in the moel calculate in this paper, the sprinkler hea was place too far away from the vent. The water roplets can t move to the area near the vent. The smoke flow near the vent is mainly rove by the smoke buoyancy an the pressure ifference between the computational omain an the outsie. Because the m u an m E are the mass flows across the area just uner the sprinkler hea, substituting these values into the Eq. (19) cannot obtain goo results. It means that in the transition area (Fig. ), the mass flow across the smoke-air interface was significant. The great mass of ownwar flow force by sprinkler spray woul move back into the smoke layer in the transition area. Table 3. Mass flow results Case NO. HRR/ MW Sprinkler Pressure/bar Mass flow/ kg/s m a1 m a m u m D m E m p A A A A A A B B B B B B (a) 1 MW (b) 1.5 MW Fig. 9. Mass flow for ifferent pressures. 4. Conclusions The interaction of fire smoke an the water spray was stuie in this paper. Fire Dynamics Simulator version (FDS) which is evelope by NIST was use for the numerical simulation. A total of 10 cases with ifferent sprinklers an ifferent fire heat release rates were calculate. The following observations are mae:

10 46 Cunfeng Zhang an Wanki Chow / Proceia Engineering 6 ( 013 ) Due to the cooling effect, the temperature woul ecrease apparently once upon the sprinkler was activate. The temperature ecrease was almost linear to the working pressure. Due to the rag force of water spray, once the sprinkler was activate, the smoke woul escen. The imensionless pressure ( P ) to the smoke penetration epth ( Ŝ ) can be expresse as: ˆ Sˆ = Pˆ R = (1) The spray has great effect on the smoke movement insie the compartment. Because the sprinkler hea was place too far away from the vent, the inflow an outflow o not change much. Acknowlegements This work was supporte by the Opening Fun of State Key Laboratory of Fire Science uner Grant No.HZ010-KF08 an the Opening Fun of State Key Laboratory of Builing Safety an Environment uner Grant No.bsbe References [1] Bullen, M. L., The Effect of a Sprinkler on the Stability of a Smoke Layer beneath a Ceiling. Fire Research Note 1016, Fire Research Station, Borehamwoo, UK. [] Morgan, H. P., Heat Transfer from a Buoyant Smoke Layer beneath a Ceiling to a Sprinkler Spray 1-A Tentative Theory, Fire an Materials 3(1), p. 7. [3] Morgan, H. P., Baines, K., Heat Transfer from a Buoyant Smoke Layer beneath a Ceiling to a Sprinkler Spray -An Experiment, Fire an Materials 3(1), p. 34. [4] Cooper, L. Y., The Interaction of an Isolate Sprinkler Spray an a Two-Layer Compartment Fire Environment: Phenomena an Moel Simulations, Fire Safety Journal 5(), p. 89. [5] McGrattan, K. B., Hamins, A., Stroup, D., Sprinkler, Smoke an Heat Vent, Draft Curtain Interaction Large Scale Experiments an Moel Development, NISTIR Builing an Fire Research Laboratory, National Institute of Stanars an Technology, Gaithersburg, MD. [6] Beyler, C. L., Cooper, L. Y., 001. Interaction of Sprinklers with Smoke an Heat Vents, Fire Technology 37(1), p. 9. [7] Hsieh, T. L., CHIEN, S. W., Chung, K. C., Lin, C. H., 00. The Effect of Smoke Vents on the Sprinklers in a Small Compartment, Journal of Applie Fire Science 11(), p. 91. [8] Ota, M., Kuwana, Y., Ohmiya, Y., Matsuyama, K., Yamaguchi, J., 009. A Stuy on Smoke Behavior in a Compartment with Sprinkler System Activation-Simple Preictive Metho on Mass Flow Rates Base on Experimental Stuy, Fire Science an Technology 8(3), p. 88. [9] Chow, W. K., Fong, N. K., Application of Fiel Moeling Technique to Simulate Interaction of Sprinkler an Fire-Inuce Smoke Layer, Combustion Science an Technology 89, p [10] Chow, W. K., Cheung, Y. L., Simulation of Sprinkler-Hot Layer Interaction Using a Fiel Moel, Fire an Materials 18(6), p [11] Soonil, N., Development of a Computational Moel Simulating the Interaction between a Fire Plume an a Sprinkler Spray, Fire Safety Journal 6, p. 1. [1] Hua, J. S., Kumar, K., Khoo, B. C., Xue, H., 001. A Numerical Stuy of the Interaction of Water Spray with a Fire Plume, Fire Safety Journal 37, p [13] Ingason, H., Olsson, S., 199. Interaction between Sprinkler an Fire Vents Full Scale Experiments, Fire Technology SP Report 199:11, Sweish National Testing an Research Institute, Sween. [14] Chow, W. K., Tong, A. C., Experimental Stuies on Sprinkler Spray-Smoke Layer Interaction, Journal of Applie Fire Science 4, p [15] McGrattan, K. B., Baum, H. R., Rehm, R. G., Hammins, A., Forney, G. P., Floy J. E., Hostikka, S., Prasa, K., 00. Fire Dynamics Simulator (Version 3) Technical reference guie, NISTIR Builing an Fire Research Laboratory, National Institute of Stanars an Technology, Gaitherburg, MD. [16] Zhang, C. F., Huo R., Li, Y. Z., 005. Stability of Smoke Layer uner Sprinkler Water Spray, ASME s 005 Summer Heat Transfer Conference, SanFrancisco, CA. [17] Zhang, C. F., Chow, W. K., Huo R., Zhong, W., Li, Y. Z., 010. Experimental Stuies on Stability of Smoke Layer with a Sprinkler Water Spray, Experimental Heat Transfer 3(3), p [18] O Gray, N., Novozhilov. V., 009. Large Ey Simulation of Sprinkler Interaction with a Fire Ceiling Jet, Combustion Science an Technology 181(7), p [19] McGrattan, K. B., Hostikka, S., Floy J. E., Baum, H. R., Rehm, R. G., Mell, W. E., McDermott., R., 010. Fire Dynamics Simulator (Version 5), Technical Reference Guie, Volume 1: Mathematical Moel. Gaithersburg, Marylan, USA, Oct [0] Mell, W. E., McGrattan, K. B., Rehm, R. G., Numerical Simulation of Combustion in Fire Plumes, Proceeings of the Combustion Institute 6 (1), p [1] Xin, Y., Filatyev, S. A., Biswas, K., Gore, J. P., Rehm, R. G., Baum, H. R., 008. Fire Dynamics Simulations of a One-meter Diameter Methane Fire, Combustion an Flame 153 (4), p [] Smagorinsky, J., General Circulation Experiments with the Primitive Equations. I: The Basic Experiment, Monthly Weather Review 91(3), p. 99. [3] You, H. Z., Investigation of Spray Patterns of Selecte Sprinklers with the FMRC Drop Size Measuring System, Fire Safety Science - Proceeings of the First International Symposium, International Association for Fire Safety Science, pp [4] Quintiere, J. G., Scaling Applications in Fire Research, Fire Safety Journal 15(1), p. 3.