A Mathematical Model on Interaction of Smoke Layer with. Sprinkler Spray

Transcription

1 A Mathematical Moel on Interaction of Smoke Layer with Sprinkler Spray K.Y. Li a,b, L.H. Hu a, R. Huo a, Y.Z. Li a, Z.B. Chen a, X.Q. Sun a, S.C. Li a a State Key Laboratory of Fire Science, University of Science an Technoloy of China, Hefei 37, China b Collee of Environmental Science an Enineerin, Southwest Jiaoton University, Chenu 63, China Corresponin author: Tel: ( ; Fa: ( ; aress: hlh@ustc.eu.cn, huoran@ustc.eu.cn, steveli@mail.ustc.eu.cn Postal aress: State Key Laboratory of Fire Science, University of Science an Technoloy of China, Hefei, Anhui, 36, China Submitte September, 7 Revise February, 8

2 ABSTRACT A mathematical moel was evelope for preictin the ownwar escenin behavior of the buoyant smoke layer uner sprinkler spray. The behavior of the smoke layer was etermine by consierin the interaction between the ra force of the sprinkler spray an the buoyancy force of the hot smoke layer itself in the spray reion. The smoke layer may be pulle own with its thickness increase at the center of the spray reion ue to the coolin an ra effects of the sprinkler spray, thus to form a ownwar smoke loin plume. In the mathematical moel evelope in this paper, the critical conition uner which the smoke layer lost its stability, as a serious concern, was preicte. Aitionally, the lenth of the ownwar plume, which was rarely investiate before, was also further calculate. Full scale eperiments were carrie out to valiate the moel. Results showe that the preictions, incluin the critical conition an the lenth of the plume, by the mathematical moel aree well with that observe an measure in the eperiments. The lenth of the ownwar plume was shown to increase with the sprinkler operatin pressure by an approimately linear correlation. KEYWORS Smoke layer; sprinkler spray; interaction; smoke loin plume lenth; ra force; buoyancy force

3 INTROUCTION Automatic sprinkler systems are require to be installe in many builin such as hotels, factories an shoppin malls. The sprinkler spray systems, which can irectly control or suppress the fire, are very reliable in protectin builins aainst fire [-5]. But on the other han, the buoyancy of the hot smoke layer, which supports the stratification, ecreases ue to the coolin effect by the water spray. The ra force prouce by the water roplets also pulls the stratifie smoke layer ownwar. These two factors both can lea to the loss of stability of the smoke layer stratification. Uner such a conition, the smoke layer will fall own to a lower level, until it reaches the floor, resultin in smoke loin. The smoke loin is a risk to evacuation an fire fihtin [-]. However, this behavior was not consiere in the most popular zone fire moels, such as the latest version of CFAST (6.. A physical moel shoul be evelope to aress this behavior. The interaction of smoke layer with sprinkler spray was stuie by Bullen [] in 974. A smoke layer assume with a constant thickness was consiere. The sprinkler spray was taken as water roplets with constant iameter calculate from the sprinkler pressure. A physical parameter known as the ra-to-buoyancy ratio was calculate for the entire smoke layer to assess its stability. Moran an Baines [, 6] further inclue the convective heat transfer from the smoke layer to the sprinkler spray in the moel, which was inore by Bullen. Numerical moelin was conucte later by Alpert [7], Chow [8-], Hoffmann [] an Gariner []. Interaction of sprinkler spray with ventilation was also stuie by Heselen [3], Hinkley [4] an McGrattan [5]. More recent work ha also been reporte by Heskesta [6] an Cooper [7, 8]. The behavior of the smoke layer uner a sprinkler spray was stuie by Cooper in 995 with a physical moel evelope. In Cooper s moel, it was consiere that the

4 smoke layer element of unit volume below the sprinkler nozzle was pulle own by the ra force of sprinkler roplets an pushe up by its own buoyancy. The shape of the spray reion was assume to be a cone with the ape at the sprinkler nozzle. The temperature of the smoke layer element was etermine by the heat transfer process between the smoke layer an the spray roplets. As reporte by Cooper [7, 8], the smoke was ownwar-buoyant in the upper layer of the spray reion because its temperature was less than that of the smoke outsie the reion. So the smoke layer element below the sprinkler nozzle shoul move ownwar to penetrate the interface, escen into the lower cool air layer an form a ownwar smoke-loin plume uner the sprinkler spray. However, in some eperiments reporte [-5], the smoke layer woul still remain stable an no ownwar smoke-loin plume was seen uner certain spray conitions. On the other han, the smoke layer thickness was not even consiere in Cooper s moel, but it will certainly have a major contribution. Althouh there were numerous stuies on the interaction of smoke layers with sprinkler sprays, they were mostly focuse on the critical conition uner which the smoke layer will lose its stability. It shoul be also note that the hot smoke will become upwarly-buoyant after penetratin the smoke layer interface as the ensity of the lower ambient air reater than the smoke bein pulle own [7, 8]. So, the ownwar smoke flow pulle own by the sprinkler spray ecelerates below the smoke layer interface an may stop before reachin floor level. The question arisin then is: what will the lenth of the smoke loin plume finally be, or how far will the smoke layer be pulle own by the sprinkler spray? However, this has rarely been investiate. A mathematical moel is evelope in this paper to escribe the smoke-loin behavior of a smoke layer uner a sprinkler spray. The critical conition uner which 3

5 the stable smoke layer stratification will be lost is etermine. The lenth of the smoke-loin plume, uner the conition that the smoke layer loses its stability, is preicte with smoke layer thickness an temperature, an sprinkler operatin pressure consiere. Finally, full scale eperiments were carrie out to valiate the mathematical moel. MATHMATICAL MOEL Assumptions. The viscous force erive from the eformation of the smoke layer is inore as it is too small relative to the ra force of the roplets an the buoyancy force of the smoke layer.. The water ensity istribution (W in the horizontal cross section of the sprinkler spray is uniform. 3. Small oscillations of smoke layer interface are not consiere, althouh they are observe in the eperiments. 4. The behavior of the smoke layer uner a sprinkler spray was consiere by takin a column element of the smoke layer (unit base area δ an fie heiht, h-, (that was covere by the spray into account, as shown in Fiure. The smoke layer behavior is escribe as this column element bein pulle own by the ra force of the sprinkler spray roplets an at the same time pushe back up by its own buoyancy, base on the fact that the similar phenomenon was observe urin the eperiments. 5. The ownwar vertical velocity of the smoke layer was nelecte when consierin the ra force ue to the relative vertical velocity between the roplets an the smoke layer, since vertical velocity of the smoke layer is very small in relation to the velocity of the roplets. 4

6 6. It was mentione by Cooper [7] that recirculation flows woul be cause by the ownwar smoke loin plume an lea to some increase of the thickness of the stable smoke layer. The comple recirculation flows inuce by roplet ra is not taken into account in the current analysis. ra force of the sprinkler spray roplets As shown in Fiure, the arkene reion of the smoke layer is that affecte by the spray. The initial thickness of the smoke layer is h. Initial velocity of the smoke layer is assume to be zero. The vertical ra force cause by a spray roplet is then epresse as [, 3]: ( = k v ( k ( C A = ( where ( is the vertical ra force of a sinle water roplet at coorinate ( N, v is the vertical velocity of the roplet ( m / s, ( is the ensity of the smoke 3 ( k / m, C is the ra coefficient an is taken to be.6 when Re is ~ [-6] an A is the central cross section area of the roplet ( m. The momentum equation of the roplet is: m k v = m v t = m v v (3 where m is the mass of the roplet ( k, is the acceleration of ravity ( m / s. Interatin the equation (3, the vertical velocity v can be epresse as: v m ep( k = + C (4 k m The constant C is etermine by the vertical velocity bounary of the roplet. 5

7 Accorin to the eperimental results by Sheppar [9], the velocity of a roplet. m below the sprinkler is approimately.4~.6 times p 6. An averae value of.5 is taken here. As the velocity irection of the spray roplets when leavin the sprinkler is vertically ownwar, the constant C for the roplets is euce to be: C p = (.5 6 m k k ep( m (5 where p is the operatin pressure of the sprinkler ( MPa, is ensity of the 3 water ( k / m an =. m. The envelope curve of the spray reion is approimately paraboloi accorin to the NFPA3HB [] an the eternal shape curve is efine as: y = E (6 thus the horizontal cross section area of the spray reion at coorinate is: S( = πe (7 where S ( is the area of the cross section ( m. The manual of the sprinkler an eperimental observations inicate that the wette area at 3 m below the ceilin is a circle with raius of approimately 3 m [, 9-]. Then the coefficient E is euce to be 3. The water ensity istribution can be assume to be uniform in the cross section area [, ]. The roplet number, n (, in the sub-volume of the smoke layer column element of heiht at coorinate position is then epresse as: M n( = (8 m S( v where M is the ischare mass flow rate out of the sprinkler nozzle ( k / s : 6

8 K p M = (9 3 6 where K is the flow rate coefficient of the sprinkler ( L /(min bar an taken to be 8 for the sprinkler with nozzle iameter of.7 mm [9, ]. The ra force of the roplets on the smoke layer column element at coorinate is then euce to be: M k m k ' ( = n( ( = + C ep( S( m k m ( where: k m 3 3 ( C = [ ( Cπ π ] = ( with is the iameter of the roplet ( m. The iameters of ifferent roplets are assume to be same with a mean iameter m here, which was calculate by equations ( - (4 [, 3, 9, ]: ( 3 m = Cm nwe ( U We = n (3 σ w U M = (4 π / 4 n where U is the initial velocity of the water when ischarin out of the sprinkler nozzle ( m / s, σ w is the surface tension of water which is taken to 3 be 7.8 N / m, We is the Weber number an n is the iameter of the sprinkler nozzle ( m. The coefficient C m is taken to be.33 for sprinkler with nozzle iameter of.7 mm [, 3]. As shown in Fiure, the total ra force of the sprinkler spray roplets on the 7

9 8 smoke layer column element with unit area δ is: m k C k m m k S M h h + = = ep( ( '( (5 where h is the thickness of the smoke layer ( m. Substitutin equation (7 an equation ( into equation (5 ives: C C C C E M h + = 4 ( 6 ep( ( ( 3 π (6 Buoyancy force of the smoke layer column element The buoyancy force, ( ' B, of the sub-volume of the smoke layer column element with a small heiht of at coorinate position varies with its temperature an is taken as: T T T B ( ( ] ( [ ( ' = = (7 where is air ensity at ambient temperature ( 3 / m k, ( T an T is the smoke temperature (K an the ambient temperature (K respectively. The total buoyancy force on the smoke layer column element with unit area δ is then calculate to be: T T T B B h h = = ( ( ( ' (8 ( T T in equation (8 can be substitute by the averae temperature rise of the smoke layer, assumin that the temperature ecays linearly with heiht in the smoke layer [4, 5]. The equation (8 can be simplifie as: ( ( ( h T T T T T T B h + = = (9

10 where T is the averae temperature rise of the smoke layer (K. Smoke loin behavior The roplets irectly below the sprinkler in the spray reion have the maimum vertically ownwar velocity while leavin the sprinkler nozzle, as reporte by Sheppar [9]. So, the ra force on the smoke layer column element shoul also be maimal at this position, with coorinates = an y =, in the spray reion (Fiure (a. The smoke layer shoul first lose its stability here. So, the smoke layer column element with coorinate y = was only consiere here for the smoke layer behavior. This initial ra force to this column element is: M 3 ( C πe ( C 6 ( C + C ep( 4 h = ( with C bein a constant which can be etermine by equation (5. The initial buoyancy force irectly below the sprinkler is: T = h ( T + T B The smoke layer column element irectly below the sprinkler woul move ownwars if > B, which represents the instability of the smoke layer. As shown in Fiure (b, when the element moves ownwar by a istance S, the ra force on the column element chanes to be: S h M C C S + 3 ( 4 6 ( C S E C = + ep( π 4 3 ( 4 ( The kinetic enery equation of the column element is then taken as: 9

11 J = S S S B S (3 where J is the kinetic enery of the column element which is etermine by the s strenth of the ra force SS an the strenth of the buoyancy force BS. Substitutin equation ( into equation (3 ives: S S + h 3M ( C C 4 6 ( hv = ( + C ep( B S S 4πE 3 ( C 4 (4 where V is the velocity of the smoke layer column element ( m / s. The value of S ecreases as the smoke layer column element moves ownwars. The element beins to ecelerate when B > S an its velocity finally ecreases to zero at a certain coorinate position. The maimum istance that the element can move own can be euce from equation (4 by takin V to be zero: S = S S h + S 3M ( C 4πE 4 3 ( C + C 6 ( C ep( 4 S B (5 The 3r-orer Simpson numerical metho is applie for calculatin S an S in equation ( an equation (5. The lenth of the ownwar smoke loin plume L shoul be: L = S + h (6 EXPERIMENTS The eperimental apparatus is shown in Fiure 3. It consiste of two parts: the burnin cabin an the sprinkler cabin. As shown in Fiure 3, pool fires were burne in the burnin cabin to enerate an initial stable smoke layer in the upper reion of the sprinkler reion. The burnin cabin was 4 m lon, m wie an.5 m hih. Si air

12 supply openins with lenth of.8 m an heiht of.4 m were locate on both sies of the cabin. The sprinkler cabin was 4. m lon, 4. m wie an 4. m hih. A smoke curtain. m hih was installe below the top of the cabin to maintain an initial stable smoke layer with thickness of. m. A measurement aue 4. m hih with a resolution of +.m was place in front of the cabin for measurin the lenth of the ownwar smoke loin plume as shown in Fiure 3(b. The uncertainty of the observe lenth of the smoke loin plume was estimate to be less than +.5 m. 4 thermocouple trees were istribute in a circle of iameter. m with the sprinkler at the centre. The vertical interval of the thermocouples is.3 m. Bare bea K type thermocouples were use with uncertainties estimate to be of less than + o C. The thermocouples were protecte by sale steel waterproofin caps which are use for avoiin the influence of the water roplets on the temperature measurement of the thermocouple bea. As shown in Fiure 3(c, ZSTP-5 Sprinkler with nozzle iameter of.7 mm was use for the tests. The sprinkler is mae by Copper Alloys with the flow rate coefficient of 8. The sprinkler was installe in the central of the sprinkler cabin roof as a stanar penant. A pressure reuction valve an pressure transucer an transmitter were installe on the pipe to control the sprinkler operatin pressure with an accuracy of +. MPa. A iital vieo camera was use to recor the tests process so as to etermine the lenth of the ownwar smoke loin plume. In total, 9 tests were conucte with 3 ifferent fire heat release rates. iesel oil was use as the fuel for of the pool fires. The heat release rate of the pool fires was etermine by the mass loss rate measure by an electronic balance with accuracy of +. The heat of combustion of the iesel oil was taken to be 4 kj / k. The combustion efficiency was taken to be.8 accorin to previous measurements in

13 the ISO 975 Room Calorimeter [6]. The sprinkler spray was ischare at 5 s after inition when the upper part of the sprinkler cabin was fille with a stable smoke layer. The total burnin time of each test was about 4 s. The operatin pressure of the sprinkler was varie from.3 to.3 MPa. RESULTS AN ISCUSSION The stability of the smoke layer uner sprinkler spray The smoke layer woul remain stable when the operatin pressure was relatively low. Uner this conition, the two zone structure of the smoke layer was not broken as the smoke layer temperature was relatively hih. The interface between the smoke an the air was clear in the sprinkler cabin. The thickness of the smoke layer increase a bit ue to the ilution effects of the water roplet to the layer. It was less than.3 m in all of the tests uner this conition. The eperimentally measure ata are summarize in Table, with, B an the smoke loin plume lenth calculate. Fiure 4 presents typical photos of the smoke layer in test A an C. As shown in the Fiure, there was no ownwar smoke plume which woul have inicate the instability of the smoke layer. In these two tests, as well as in test B an C, the smoke layer remaine stable at the top of the sprinkler cabin. This was in accorance with the mathematical moel as were calculate to be less than B in these tests as shown in table. The smoke layer was shown to lose its stability when > B. Uner this conition, the operatin pressure was relatively hih an the smoke layer temperature was relatively low, the stable two zone structure of the smoke-air layer was broken an a ownwar smoke loin plume was shown. The plume penetrate the interface an brouht the smoke to the lower part of the cabin.

14 Fiures 5(a - (f present typical photos of tests C3-C8 (sprinkler operatin pressure varyin from.7 MPa to.3 MPa. As shown in these Fiures, the smoke layer became unstable. Part of the smoke layer in the spray reion move ownwars an forme the smoke loin plume. The hiher the sprinkler operatin pressure, the reater the lenth of the ownwar plume was. Accorin to equations (5 (6 an equations ( (6, the ra force of the sprinkler spray ecreases with the raial istance away from the centerline of the spray reion. Thus the shape of the ownwar smoke loin plume shoul be like an inverse bowl. This is clearly seen urin the eperiments, as shown Fiure 5. The volume of the ownwar smoke plume also enlare horizontally as the operatin pressure increase. In Table, > B is foun for tests C3- C8. It can be seen from above that the relative manitue of an B calculate can be reare as a criterion for stability of the smoke layer uner sprinkler spray. The smoke layer remains stable when < B an becomes unstable when > B. As shown in Table, for fire types of A an B with relatively low heat release rate, the averae smoke layer temperatures were also relatively low. The ownwar smoke loin plume finally reache the floor when the sprinkler operatin pressure was increase from.7 MPa to.9 MPa, respectively, while for type C fire with a relatively hih heat release rate, it was.3 MPa. It was clearly shown that hiher sprinkler operatin pressure was neee for pullin own the smoke to the floor level for smoke layer with hiher temperature. Lenth of the ownwar plume The lenth of the ownwar smoke loin plume calculate by equations ( - (6 was compare with that eperimentally measure value in Fiure 6. Accorin 3

15 to the mathematical moel, with the increase of the operatin pressure an thus the ischare mass flow rate M, the ra force S of the sprinkler spray increase, which will lea to the increase of the movin istance an thus the lenth of the ownwar plume. This tren was in accorance with what was measure in the eperiments. The plume lenths preicte quantitatively by the mathematical moel aree fairly well with the eperimental ata. However, the calculate values were a bit lower than the eperimental values ue to the fact that the water ensity istribution is not uniform in the horizontal cross section of the spray reion, which was not taken into account in the mathematical moel. The real istribution is that the water ensity in the center of cross section is a bit larer than that in the outer reion []. Thus the current mathematical moel shoul slihtly unerestimate the lenth of the ownwar plume. The ownwar movin istance S was correlate with the operatin pressure in Fiure 7(a for tests C3 - C8. It was shown that the variation of S with the operatin pressure fitte the followin linear reression formula with correlation coefficient of.99: S 7.3p.63 (7 = This inicates that the ownwar movin istance S (or the lenth of the ownwar plume L increases linearly with the operatin pressure p. Fiure 7(b presents the variation of the averae temperature rise of the smoke layer in the sprinkler cabin with the operatin pressure. It was shown that the temperature rise reuce while the operatin pressure increase. However, with the increase of the sprinkler operatin pressure, the reuction to the smoke layer temperature seeme to be less effective. So, it can be rawn from above that a optimal operatin pressure shoul be selecte for the sprinkler, to achieve maimum coolin of the smoke layer while at the same time 4

16 ensurin that the sprinkler spray oes not break the stability of the smoke layer reatly an pull own the smoke layer to the floor lever to threaten the safety of the people. For eample, the optimal operatin pressure, as can be seen from Fiure 7, shoul be about.7 MPa for the case of the type C fire in this stuy. CONCLUSIONS A new mathematical moel was evelope in this paper to eamine the interaction of smoke layer with sprinkler spray. The ra force of the sprinkler spray roplets an the buoyancy force of the smoke layer were numerically calculate. The critical conition uner which the stability of the smoke layer stratification lost an in aition the lenth of the ownwar smoke loin plume was preicte. Full scale eperiments were carrie out to valiate the mathematical moel. The preictions of the mathematical moel were shown to aree well with the eperimental ata an observations. The smoke layer remains stable when < B an becomes unstable when > B. With the increase of the sprinkler operatin pressure, the lenth of the ownwar smoke loin plume increase monotonously linearly, but the cool effect to the smoke layer was shown to be less effective. The lenths of the ownwar smoke loin plume preicte by the mathematical moel aree fairly well with the eperimental ata, althouh were a bit lower ue to the fact that water ensity istribution is not uniform in the horizontal cross section of the spray reion but was inore in the mathematical moel. This factor will be inclue to improve the current moel in the future work. The implementin the current mathematical moel into a zone fire moel is also onoin an will be reporte later. 5

17 ACKNOWLEGEMENT This work was supporte by the Natural Science Founation of China (NSFC uner Grant No an Anhui Provincial Natural Science Founation of China uner Grant No

18 NOMENCLATURE A : central cross section area of the roplet, m ; B : buoyancy force on the smoke layer column element below the sprinkler, N ; B '( : buoyancy force of unit volume at coorinate, N ; B : total buoyancy force on the smoke layer column element with unit area, N ; C : ra coefficient; C m : coefficient for calculatin the mean iameter; : iameter of the roplet, m ; m : mean iameter of all roplets, m ; n : iameter of the sprinkler nozzle, m ; ( : vertical ra force of a sinle water roplet at coorinate, N ; : initial ra force on the smoke layer column element below the sprinkler, N ; S : ra force on the smoke layer column element irectly below the sprinkler when move ownwar with a istance of S, N ; '( : ra force of unit volume at coorinate, N ; : total ra force on the smoke layer column element with unit area, N ; E : coefficient of curve equation for the eternal shape of the spray reion; : acceleration ue to ravity, ms ; h : initial thickness of the smoke layer, m ; J : kinetic enery of the column element, J ; k : coefficient for calculatin (, ms ; K : flow coefficient of the sprinkler L /(min bar ; L : lenth of the smoke loin plume, m ; 7

19 m : mass of the roplet, k ; M : ischare mass flow rate of the sprinkler nozzle, ks ; n ( : water roplet numbers in unit volume at coorinate ; p : operatin pressure of the sprinkler, MPa ; S : movin istance of the smoke layer column element irectly below the sprinkler, m ; S ( : area of the horizontal cross section of the spray reion at coorinate, T ( : smoke temperature, K; m ; T : averae temperature rise of the smoke layer, K; T : ambient temperature, K; U : initial velocity of the water when ischarin out of the sprinkler nozzle, ms ; v : vertical velocity of the roplet,; ms V : velocity of the smoke layer column element, ms ; We : Weber number; : coorinate of the ape of the smoke layer column element, m ; Greek symbols : ensity of the water, : ensity of the smoke, 3 km ; 3 km ; : ensity of the air at ambient temperature, 3 km ; σ w : surface tension of water, Nm ; 8

20 REFERENCES [] Moran H P. Heat Transfer from a Buoyant Smoke Layer Beneath a Ceilin to a Sprinkler Spray -A Tentative Theory. Fire an Materials, 979; 3: [] Bullen M L. The Effect of a Sprinkler on the Stability of a Smoke Layer Beneath a Ceilin. In: Fire Research Note 6, Fire Research Station, Borehamwoo, UK, 974; -. [3] Chow W K, Yao B. Numerical Moelin for Interaction of a Water Spray with Smoke Layer. Numerical Heat Transfer, ; 39: [4] Chow W K, Ton.A C. Eperimental Stuies on Sprinkler Spray-Smoke Layer Interaction. Journal of Applie Fire Science, 995; 4: [5] Zhan C F, Huo R, Li Y Z. Stability of Smoke Layer uner Sprinkler Water Spray. ASME s 5 Summer Heat Transfer Conference, San Francisco, CA, 5. [6] Moran H P, Baines K. Heat Transfer from a Buoyant Smoke Layer Beneath a Ceilin to a Sprinkler Spray -An Eperiment. Fire an Materials, 979; 3: [7] Alpert R L. Numerical moelin of the interaction between automatic sprinkler sprays. Fire Safety Journal, 985; 9: [8] Chow W K, Fon N K. Numerical Simulation on Coolin of the Fire-inuce Air Flow by Sprinkler Water Spray. Fire safety Journal, 99; 7: [9] Chow W K, Fon N K. Numerical Stuies on the Interaction of Sprinklers an the Hot Layer. Architectural Science review, 993; 36: 3-. [] Chow W K, Cheun Y L. Simulation of Sprinkler-hot Layer Interaction Usin a Fiel Moel. Fire an Material, 994; 8: [] Hoffmann N F, Galea E R, Markatos N C. Mathematical Moelin of the Fire Sprinkler Systems. Applie Mathematical Moelin, 989; 3: [] Gariner A J. The Mathematical Moelin of the Interaction Between Sprinkler Sprays an the Thermally Buoyant Layers of the Gas from Fires. Ph issertation, South Bank Polytechnic, Lonon, Unite Kinom, 989. [3] Heselen A J M. The Interaction of Sprinkler an Roof Ventin in Inusttial Builins: the Current Knowlee. Builin Research Establishment, Borehamwoo, UK, 984. [4] Hinkley P L. The Effect of Vents on the Openin of the First Sprinklers. Fire Safety Journal. 9

21 986; : -5. [5] McGrattan K B. Hamins A an Stroup, Sprinkler, Smoke & Heat Vent, raft Curtain Interaction - Lare Scale Eperiments an Moel evelopment. NISTIR 696-, National Institute of Stanars an Technoloy, Gaithersbur, 998. [6] Heskesta G. Sprinkler/Hot Layer Interaction. In: NISTGCR-9-59, National Institute of Stanars an Technoloy, Gaithersbur, 99. [7] Cooper L Y. The Interaction of an Isolate Sprinkler Spray an a Two-Layer Compartment Fire Environment. Phenomena an Moel Simulations. Fire Safety Journal, 995; 5: [8] Cooper L Y. The Interaction of an Isolate Sprinkler Spray an a Two-Layer Compartment Fire Environment. International Journal Heat an Mass Transfer, 995; 38: [9] Sheppar T. Spray Characteristics of Fire Sprinklers. Ph issertation, Northwestern University, Evanston, [] NFPA3HB. Automatic Sprinkler System Hanbook. Eition. USA, National Fire Protection Association,. [] Chow W K, Shek L C. Physical Properties of a Sprinkler Water Spray. Fire an Material, 993; 7: [] Yu H Z. Investiation of Spray Patterns of Selecte Sprinklers with the FMRC rop Size Measurin System. Fire Safety Science Proceeins of the First International Symposium, International Association for Fire Safety Science, 986, pp [3] Chan T S. Measurement of Water ensity an roplet Size istributions of Selecte ESFR Sprinklers. Journal of Fire Protection Enineerin, 994; 6: [4] Hu L H, Huo R, Yan R X, He W H, Wan H B an Li Y Z. Full scale eperiments on stuyin smoke sprea in a roa tunnel, Fire Safety Science Proceeins of The Eihth International Symposium, Sep., 5, Beijin, China, P [5] Hu L H, Huo R, Chow W K, Wan H B, Yan R X. ecay of buoyant smoke layer temperature alon the lonituinal irection in tunnel fires, Journal of Applie Fire Science, Vol. 3 (, P , 4-5

22 [6] Yi L. Stuy on Smoke Movement an Manaement in Atrium Builin, Ph. thesis, University of Science an Technoloy of China, Hefei, 5.

23 Fiure Captions Fiure : Schematic view of interaction of smoke layer with sprinkler spray Fiure : Pullin own of smoke layer column element by sprinkler spray Fiure 3: Eperimental ri an the sprinkler Fiure 4: Typical photos of stable smoke layer for tests A an C Fiure 5: Photos of smoke loin plume for test series of C3-C8 with increasin sprinkler spray pressure Fiure 6: Eperimental an calculate values of L varyin with the increase of p Fiure 7: Variation of movin istance of the smoke layer column element an smoke layer temperature rise with sprinkler operatin pressure

24 Table : Summary of the tests Pool fire type Pool size ( m Test No. HRR (kw Sprinkler operatin pressure ( MPa Ambient temperature ( K Averae temperature rise of the smoke layer ( K Lenth of the smoke Measure loin plume ( m B Calculate A A.5 A A A A B B B.36 B B B B C C C C.64 C C C C C

25 Spray reion Envelope curve of the spray reion Fiure : Schematic view of interaction of smoke layer with sprinkler spray 4

26 (a Initial conition with ra force (b The smoke layer column element move own with a istance of S Fiure : Pullin own of smoke layer column element by sprinkler spray 5

27 Sprinkler Smoke curtain Burnin cabin Air supplyin openin Sprinkler cabin Fire source Thermocouple (a Schematic view of the eperimental ri Gaue Smoke curtain Burnin cabin Air supplyin openin Sprinkler cabin (b Photo of the eperimental ri (c Photo of the sprinkler Fiure 3: Eperimental ri an the sprinkler 6

28 (a A (b C Fiure 4: Typical photos of stable smoke layer for tests A an C 7

29 (a C3 (b C4 (c C5 ( C6 (e C7 (f C8 Fiure 5: Photos of smoke loin plume for test series of C3-C8 with increasin sprinkler spray pressure 8

30 Lenth of the plume (m Eperimental Calculate Operatin pressure (MPa (a Pool fire A Lenth of the plume (m Eperimental Calculate Operatin pressure (MPa (b Pool fire B Lenth of the plume (m Eperimental Calculate Operatin pressure (MPa (c Pool fire C Fiure 6: Eperimental an calculate values of L with the increase of p 9

31 Movin istance S (m Operatin pressure (MPa (a Movin istance of the element 3 8 Temperature rise ( Operatin pressure (MPa (b Temperature rise of the smoke layer Fiure 7: Variation of movin istance of the smoke layer column element an smoke layer temperature rise with sprinkler operatin pressure 3