Experimental Study of Backdraft in a Compartment with Different Opening Geometries and its Mitigation with Water Mist

Transcription

1 Experimenal Sudy of Backdraf in a Comparmen wih Differen Opening Geomeries and is Miigaion wih Waer Mis WENGUO WENG and WEICHENG FAN Sae Key Laboraory of Fire Science Universiy of Science and Technology of China Hefei, Anhui, , P.R. China ABSTRACT This paper presens he resuls of reduced-scale experimenal ess o sudy backdraf in a reduced-scale comparmen (1.2 m x 0.6 m x 0.6 m), fied wih six end opening geomeries and wo ceiling opening geomeries. The experimenal variables included he fuel flow rae, he ime during which he fuel was burned, and he opening geomeries. The quaniies recorded before backdraf included emperaure and he concenraions of oxygen, carbon dioxide, and carbon monoxide. To quanify he effec of backdraf, he gas velociies in he opening and also he pressures in he comparmen were measured. The effecs of differen opening geomeries on he occurrence of backdraf are discussed. This sudy shows ha he mass fracion of unburned fuel (i.e., he unburned mehane in his sudy), whose criical value varies wih he opening geomery, is a key parameer deermining he occurrence of backdraf. In addiion, he experimenal resuls using waer mis, generaed by a downward-direced pressure nozzle ha was operaed a pressure of 0.2 MPa, o miigae backdraf are presened. The experimenal resuls show ha waer mis is an effecive miigaing acic able o suppress backdraf in a comparmen primarily by means of diluing he gas in he comparmen and reducing he mass fracion of unburned fuel, raher han by a hermal mechanism of cooling. KE WORDS: backdraf, waer mis, comparmen fire, opening geomery INTRODUCTION During fires in a building wih limied venilaion, backdraf may develop ha can produce fire gases conaining significan proporions of parial combusion producs and un-burn pyrolysis producs. If hese accumulae, when an opening is made o he building and air eners, i can lead o a sudden deflagraion. The deflagraion moving hrough he building and ou of he opening is a backdraf. Backdraf coninues o be a hazard ha can kill people and cause he building o collapse [1-2]. Lile research has been done on backdraf. Fleischmann [2-] conduced experimens in a half-scale comparmen wih wo opening geomeries and mehane as fuel. Full-scale FIRE SAFET SCIENCE PROCEEDINGS OF THE EIGHTH INTERNATIONAL SMPOSIUM, pp COPRIGHT INTERNATIONAL ASSOCIATION FOR FIRE SAFET SCIENCE 1181

2 experimens have been conduced by Bolliger [6] o deermine he effec of scaling he comparmen; he resuls were compared wih Fleischmann s work [2]. Gojkovic e al. [7] used naural gas as fuel o sudy backdraf in a comparmen (.2 m x 2.2 m x 2.2 m). Full-scale backdraf experimens have also been performed [8-] in wo differen comparmens o improve naval firefighing acics using Diesel spray as fuel. One comparmen was used o produce safe and reliable backdraf scenarios, which could be used as a basis for conducing backdraf experimens onboard a ship. Anoher comparmen wih differen geomeries and venilaion condiions was adjacen o he backdraf comparmen. They [8-] also ried o preven backdraf using a waer spray. Weng carried ou experimenal and heoreical sudy on backdraf in a comparmen for his Ph. D degree supervised by Prof. Fan, and published some journal papers. Weng and Fan [11-12] esablished a reduced-scale apparaus o sudy he criical condiion of he occurrence of backdraf and he effecs of differen opening geomeries on backdraf. Waer mis was applied o miigae backdraf [13-14]. In heory analysis, a model of backdraf was esablished and he nonlinear dynamical mechanism of backdraf was analyzed [], and numerical models including a subgrid scale laminar flamele model and a parially premixed model are imbedded in FDS3.0 source code for backdraf simulaion [16]. In order o invesigae he graviy curren prior o backdraf, a series of scaled sal waer experimens using flow visualizaion and DPIV (Digial Paricle Image Velocimery) were conduced [17]. This paper presens experimenal resuls exraced from Weng s disseraion in Chinese [18]. EXPERIMENTAL DESIGN AND PROCEDURE The configuraion of he backdraf apparaus is given in Fig. 1. The apparaus is made up of a reduced-scale comparmen, fuel sysem, igniion sysem, waer mis sysem, daa acquisiion sysem. These sysems are described in deail in he following secion. A special comparmen was consruced o wihsand he dangerous overpressures expeced in backdrafs. The experimenal apparaus (1.2 m x 0.6 m x 0.6 m) was roughly 1/4 ha of a residenial room o minimize his hazard and o carry ou backdraf experimens more precisely. Because of he explosive naure of backdrafs, he inernal and exernal surfaces of he comparmen were oally covered wih 2 mm hick sainless seel, welding ogeher o ensure srengh. The filler beween he wo layers of sainless seel was a refracory aluminosilicae blanke ( mm hick), whose hermal conduciviy was 0.36 W/mK a 1 K, o provide he primary hermal resisance for he srucure. An observaion window (quarz glass, 0.4 m high by 0.7 m wide, capable of wihsanding 1 K and kpa) was insalled in one of he long walls. In one of he 1182

3 shor walls and ceiling, shown in Fig. 2, bol holes were buil so ha differen opening geomeries, shown in Fig. 3, could be easily modified by replacing a face plae boled o he comparmen. These end and ceiling openings were covered wih a compuer-acivaed hach, which was opened afer he fire had been burning for a predeermined ime. The angular speed of he hach was ~ 4 o /s. Every effor was made o seal all he consrucion holes o conrol leakage. Mehane Pressure regulaor Thermocouple Gas sample apparaus Igniion apparaus burner Roameer Conrolled valve Species analyzer apparaus Pressure Pressure gauge Waer mis Backdraf comparmen Camera Ceiling hach Bidirecional probes End hach Valve Compuer Conroller Image acquisiion clip Pressure gauge Pressure regulaor Waer k High pressure air Daa acquisiion apparaus Fig. 1. Schemaic of he backdraf experimenal apparaus. A mehane (99.8% pure) burner (0. m square by 0. m high) was placed agains he wall opposie o ha wih he openings, as in Fig. 1. To ignie he combusible mixure in he comparmen, an elecrically heaed meal wire (power 1 W) provided an igniion source. The wire was 0.6 m long and was round around a ceramic cylinder (diam. 0.0 m), which was horizonally fasened on he burner. A downward-direced pressure nozzle (from Lechler GmbH & Co. KG) was posiioned 0.3 m from he wall wih he openings, m from he ceiling, and 0.3 m from he wall wih he observaion window. High pressure air was combined in he waer ank so he waer can be crushed ino millions of waer droples, and his way waer mis was formed. The nozzle was operaed a a pressure of 0.2 MPa in his work. The waer sprays were injeced direcly downward ino he comparmen and he cone angel of he nozzle was 60 o. The flow rae was abou 3.0 ml/s, and he volume mean diameer of he mis was abou 38 µ m, measured using a LDV/APV (Laser Doppler Velocimeer/ 1183

4 Adapive Phase Velocimeer) sysem. Gas sample line 0.1 Thermocouple ree Hach Slo ceiling opening Hach Middle-slo End Opening uni in m Fig. 2. Skech of he reduced-scale comparmen giving he inernal dimensions of he comparmen and he locaions of he insrumenaion Downside-slo end opening Middle-slo end opening Upside-slo end opening Window end opening 400 Door end opening 600 Window ceiling opening 600 Verical middle-slo end opening Slo ceiling opening uni in mm Fig. 3. Skech of eigh opening geomeries for he reduced-scale comparmen. A verical ree of hermocouples was placed hrough he middle of he ceiling. The hermocouples were made from 0.2 mm ype K hermocouple wire wih a sainless seel overbraid. The average bead diameer was 1 mm. The en hermocouples were locaed a 1184

5 0.0 m inervals, wih he highes hermocouple a 0.07 m below he ceiling. The hermal inerface heigh hisory was calculaed from he ime dependen emperaure profiles recorded from he hermocouple ree. The profiles were convered ino he average upper and lower layer emperaures using he mehod of Quiniere, e al. [19] applied o seady sae emperaure profiles. The pressure in he comparmen was recorded using an elecronic pressure ransducer (NOVA echnique) whose calibraed range was 0 o 300 Pa, and response ime was 00 µ s. The pressure po was placed 0.3 m from he wall wih he openings and mouned in he wall opposie o he observaion window a floor level. The ambien pressure reference was aken ouside he comparmen. The flow rae of mehane was measured by a roameer wih an effecive range of 160 o 1600 L/h. The uncerainy in his measuremen was esimaed o be 1% based on a flow es of he sysem. The mass of waer mis was calculaed from he pressure, measured by a manomeer wih an effecive range of 0-1 MPa, and diameer (.4 mm) of waer pipeline. The uncerainy was esimaed o be 1.% based on a sandard es. Coninuous gas samples for measuring he concenraions of O 2, CO 2 and CO were aken hrough a sainless seel sampling ube, locaed 0.9 m from he wall wih he openings, 0.1 m from he ceiling, and 0.1 m from he wall wih he observaion window. Analyzers consised of SIEMENS ULTRAMAT 23 for CO 2 (2% mass full-scale range), O 2 (2% mass full-scale range) and SIEMENS ULTRAMAT 22 for CO (3% mass full-scale range). Unburned fuel (CH 4 in his sudy) could be calculaed based on he equaion for oxidaion. The following assumpions were necessary: (1) ha he upper layer was well-sirred, and (2) ha he overall reacion was: CH 4 + a( O N 2 ) bco2 + cco + dh 2O + en 2 + fch 4 (1) Using hese assumpions, which were reasonable since he fire source was a mehane gas burner, an overall balance on he oxygen and carbon would yield he concenraion of unburned fuel. F = mch / m (4/11) (4/7) 4 oal CO 2 CO (2) Here m CH 4 is he oal mass which has enered he comparmen and m oal is he oal mass of gas in he comparmen.,, are he mass fracions of unburned fuel, F CO2 CO carbon dioxide, and carbon monoxide, respecively. The flow in and ou of he comparmen afer he hach was opened was recorded using bidirecional probes in he opening. The probes (diam. mm) were designed in accordance wih he guidelines given by McCaffrey e al. []. The probes were locaed in he cener of he opening as shown in Fig. 1 and were 37. mm apar for all he opening geomeries, excep he verical middle-slo end opening, where he probes were 118

6 7 mm apar, and he op and boom probes were 37. mm from he soffi and sill, respecively. The pressure difference beween he gases in he comparmen and he air ouside he comparmen was measured using a differenial pressure ransducer (NOVA echnique) whose calibraed range was ± 300 Pa, and response ime was µ s. The mass flow rae hrough he opening can be calculaed using he daa measured by he bidirecional probes and he relaionship given before [2]. Daa from he hermocouple, pressure ransducer and differenial pressure ransducer were recorded using HP E1413 wih a 64-channel high-speed scanning Analog-o-Digial Converer. The sysem was capable of recording each channel imes a second. Video daa were capured using Panasonic DS28, which had 2 frames per second. The lengh and heigh of fire ball which burns ouside he comparmen would be measured according o hese video daa. Before each experimen, a 60s base line was aken o record he iniial condiions. The burner was lef on and he flame was ignied using he elecrically heaed meal wire before he sar of he experimen. A 0s, he hach was closed. Afer a predeermined ime period (he flow ime of he fuel), he gas flow o he burner was erminaed immediaely when he hach was opened. During his ime period, he fire would be from combusion o exincion due o insufficien oxygen. Afer he hach was opened, he meal wire would ignie he combusible mixure in he comparmen if igniion was reached. A he end of he experimens, he meal wire was aken off. EXPERIMENTAL RESULTS AND DISCUSSIONS Criical Condiion of Backdraf 64 backdraf experimenal resuls for eigh opening geomeries of he downside-slo end opening, he middle-slo end opening, he upside-slo end opening, he window end opening, he door end opening, he verical middle-slo end opening, he window ceiling opening and he slo ceiling opening are seen in Ref. [18]. Table 1, aken as an example, is a summary of he backdraf experimens in he reduced-scale comparmen wih he door end opening. Here T a is he ambien emperaure. O 2, CO 2, CO and F are he mass fracions of oxygen, carbon dioxide, carbon monoxide and unburned fuel a hach opening, respecively. T U, T L and h L are he upper emperaure, he lower emperaure and he layer heigh a hach opening, respecively. P max is he peak i pressure measured inside he comparmen a hach opening. m = i in, m = fo ou, m = in, fo m = ou are he oal mass which flows ino and ou of he comparmen afer hach opening and prior o igniion, and ha afer hach opening and prior o flames exiing 1186

7 Table 1. Summary of he backdraf experimens in he reduced-scale comparmen wih door end opening. Run Num. T a (K) Fuel Flow Rae ( -3 kg/s) Burner Flow Time (s) O 2 Species Concenraion CO 2 CO F Comparmen Temperaure (K) and Layer Heigh (m) T U T L h L max P Opening Flow Mass (kg) Fire Ball (m) i (P) m = i m = fo ou m = fo m = Lengh Heigh ou nonoccurrence Table 4. Summary of he backdraf experimens wih waer mis in he reduced-scale comparmen wih door end opening. Run Num. Fuel Comparmen Temperaure Waer Mis Species Concenraion Opening Flow Mass (kg) Fire Ball (m) T Flow Burner (K) and Layer Heigh (m) a Rae Flow P max (K) ( -3 Time kg/s) (s) Time Mass (s) ( -3 i kg) O 2 CO 2 CO F T U T L h (P) L m = i m = fo ou m = fo m = Lengh Heigh ou no waer mis nonoccurrence nonoccurrence in in in in 1187

8 he comparmen, respecively. From Table 1, i is he firs view ha he relaionship beween he mass fracions of O 2, CO 2 and CO, he upper emperaure, he lower emperaure and he layer heigh, he oal mass which flows ino and ou of he comparmen, and he occurrence of backdraf are fuzzy. This able shows ha he peak pressure and he size of fire ball, which indicae he inensiy of backdraf, increases wih he increase of he mass fracion of unburned fuel. So he mass fracion of unburned fuel is a key parameer deermining he occurrence of backdraf. And he more mass fracion of unburned fuel, he more inensiy of backdraf. This conclusion can also be seen from experimenal resuls wih oher seven opening geomeries [18]. Discussions wih Opening Geomeries Comparing he differen opening geomeries, he criical values of he mass fracion of unburned fuels are differen. Figure 4 shows he mass fracion of unburned fuels deermining he occurrence (solid) and nonoccurrence (hollow) of backdraf for eigh opening geomeries. Table 2 gives he corresponding esimaed criical values for he occurrence of backdraf. These differences for eigh opening geomeries are he area of he opening and is locaion. From Fig. 4 and Table 2, he firs impression is ha he criical values of he ceiling openings are lower han for he end openings. Among he downside-slo end opening, he middle-slo end opening, he upside-slo end opening and he slo ceiling opening, whose opening area are he same, bu locaions are differen, he criical value for he slo ceiling opening is he lowes and ha of he downside-slo end opening is he highes. The higher he locaion of he cener of he openings, he lower is he criical value of he mass fracion of unburned fuels. For he middle-slo end opening and he verical middle-slo end opening, provided hey have he same opening area and locaion of heir cener (i.e., boh in he middle of he end wall, only one is horizonal and anoher is verical), he criical value of he mass fracion of unburned fuel wih a horizonal opening is lower han ha of a verical opening. So i is concluded ha i is more difficul for backdraf o ake place in he comparmen wih a verical opening han ha wih a horizonal opening of he same opening area and locaion of heir cener. Comparing he middle-slo end opening wih he window end opening, whose opening locaions are he same, bu he opening area of he former is bigger han ha of he laer, he corresponding criical values of he mass fracion of unburned fuels are inverse. I is clear ha wih a larger opening area a he same locaion of is cener, he criical value of he mass fracion of unburned fuel is lower. The same conclusion is also drawn from comparing he slo ceiling opening wih he window ceiling opening. Bu he criical 1188

9 values for he window end opening and he verical middle-slo end opening, whose locaion ceners have he same locaion, are approximaely he same, in spie of he differen opening areas. The criical value for he door end opening is lower han ha for he window end opening and he verical middle-slo end opening. F F F F F F F F Fig. 4. The mass fracions of unburned fuels deermining he occurrence (solid) and nonoccurrence (hollow) of backdraf wih (a) downside-slo end opening, (b) middle-slo end opening, (c) upside-slo end opening, (d) window end opening, (e) door end opening, (f) verical middle-slo end opening, (g) window ceiling opening, and (f) slo ceiling opening (from op o boom and from lef o righ). From he observaion of large numbers of experimens, i can be concluded ha graviy curren prior o backdraf leads o he differences of he criical values of he mass fracion of unburned fuels deermining he occurrence of backdraf among differen opening geomeries. Because of he invisible fresh air and ho gas, which consiue graviy curren, he experimenal resuls of scaled sal waer experimens using flow visualizaion and DPIV help us inerpre his [17]. In hese sal waer experimens, fresh waer added wih phenolphhalein simulaed he ho gas, and sal waer added wih sodium hydroxide simulaed he fresh waer. Once phenolphhalein mixes wih sodium hydroxide, he produc of he reacion is red. The experimenal deail was described in he lieraure [17]. Table 3, aken as an example, gives he average value of he * nondimensional velociy v = v / βgh1, he nondimensional heigh h = h0 / h1 and he nondimensional mass m * = v h of graviy curren for he downside-slo end opening, he middle-slo end opening and he upside-slo end opening. Here v β, h, h, g are he velociy, he densiy difference, he heigh of graviy curren, he, 0 1 heigh of comparmen, and he acceleraion of graviy, respecively. I is obvious ha he higher he opening locaion, he less is he mass of graviy curren, which indicaes ha he less air eners he comparmen in backdraf experimens, so he less ho gas leaves 1189

10 he comparmen from he mass balance heory. And hen he higher he opening locaion, he less unburned fuel moves ou of opening, so he lower criical value of he mass fracion of unburned fuel deermining he occurrence of backdraf. Table 2. The esimaed criical values of he mass fracion of unburned fuel deermining he occurrence of backdraf for eigh opening geomeries. * Table 3. Average values of v, h and m for he downside-slo end opening, he middle-slo end opening and he upside-slo end opening. Openings Criical values downside-slo end opening 9.0 middle-slo end opening 8. upside-slo end opening 7.1 window end opening 9.8 door end opening 8.8 verical middle-slo end 9.8 opening window ceiling opening 7.7 slo ceiling opening 7.0 downside-sl o end opening middle-slo end opening upside-slo end opening * v h m Miigaion Mechanism of Waer Mis 31 experimenal resuls wih waer mis miigaing backdraf in a comparmen wih eigh opening geomeries are seen in Ref. [18]. This paper only gives he experimenal resuls wih he door end opening, as shown in Table 4. The resuls ha he mass fracion of unburned fuel is a key parameer deermining he occurrence of backdraf, and he more mass fracion of unburned fuel, he more inensiy of backdraf can be seen in Column F, P max and fire ball size. In addiion, from Table 4, i is indicaed ha he injecion of waer mis resuls in he decrease of he mass fracion of unburned fuel, bu no he decrease of emperaure in comparmen including he upper emperaure and he lower emperaure. The mass fracion of unburned fuel in he experimen wihou waer mis is he maximum value in all of he experimens. The more he waer mis injecion mass, he less mass fracion of unburned fuel, and he less inensiy of backdraf. More waer mis makes backdraf be from occurrence o nonoccurrence. Therefore, waer mis is an effecive miigaing acic ha is able o suppress backdraf in a comparmen primarily by means of diluing he gas in he comparmen and reducing he mass fracion of unburned fuel, raher han by a hermal mechanism of cooling. The same conclusions can be drawn from experimenal resuls wih oher seven opening geomeries. 1190

11 CONCLUSIONS This paper repors he resuls of backdraf experimens in a reduced-scale comparmen wih eigh opening geomeries. This sudy shows ha he mass fracion of unburned fuel (i.e., he unburned mehane in his sudy) is a key parameer deermining he occurrence of backdraf. As he mass fracion of unburned fuel increases, he over-pressure in he comparmen also increases and he backdraf becomes more severe. In addiion, he effecs of eigh opening geomeries on he occurrence of backdraf are discussed. The resuls show ha he criical values of he mass fracion of unburned fuel deermining he occurrence of backdraf vary wih differen opening geomeries. The experimenal resuls of waer mis miigaing backdraf in a comparmen are also given in his paper, and he miigaion mechanism is discussed. Waer mis is an effecive miigaing acic able o suppress backdraf in a comparmen primarily by means of diluing he gas in he comparmen and reducing he mass fracion of unburned fuel, raher han by a hermal mechanism of cooling. The more he waer mis injecion mass, he less mass fracion of unburned fuel, and he less inensiy of backdraf. ACKNOWLEDGMENTS The auhors would like o acknowledge he suppor provided by he Naional Naural Science Foundaion of China (Gran No ) and he China NKBRSF projec (No. 1CB409600). REFERENCES [1] Bukowski, R.W., Modeling Backdraf: he Fire a 62 Was Sree, NFPA Journal, 199, 89: [2] Fleischmann, C.M., Backdraf Phenomenon, Naional Insiue of Sandards and Technology, USA, NIST-GCR , [3] Fleischmann, C.M., Pagni, P.J., and Williamson, R.B., Quaniaive Backdraf Experimens, Fire Safey Science Proceedings of he Forh Inernaional Symposium, Inernaional Associaion for Fire Safey Science, 1994: [4] Fleischmann, C.M., Pagni, P.J., and Williamson, R.B., Sal Waer Modeling of Fire Comparmen Graviy Currens, Fire Safey Science Proceedings of he Forh Inernaional Symposium, Inernaional Associaion for Fire Safey Science, 1994: [] Fleischmann, C.M., and McGraan, K.B., Numerical and Experimenal Graviy Currens Relaed o Backdrafs, Fire Safey Journal, 1999, 33: [6] Bolliger, I.B., Full Residenial-scale Backdraf, Universiy of Canerbury, New 1191

12 Zealand, Fire Engineering Research Repor 9/1, 199. [7] Gojkovoc, D., Iniial Backdraf Experimens, Lund Universiy, Sweden, Repor 3121, 1. [8] Gouk, D.T., Williams, F.W., and Farley, J.P., The Developmen and Miigaion of Backdrafs: a Full-scale Experimenal Sudy, Fire Safey Science Proceedings of he Fifh Inernaional Symposium, Inernaional Associaion for Fire Safey Science, 1997: [9] Gouk, D.T., Peaross, M.J., Farley, J.P., and Williams, F.W., The Developmen and Miigaion of Backdraf: a Real-scale Shipboard Sudy, Fire Safey Journal, 1999, 33: [] Williams, F.W., Farley, J.P., Peaross, M.J., and Gouk, D.T., 199 Class B Firefighing Docrine and Tacics: Final Repor, Naval Research Laboraory, USA, NRL/MR/ , [11] Weng, W.G., and Fan, W.C., Criical Condiion of Backdraf in Comparmen Fires: a Reduced-scale Experimenal Sudy, Journal of Loss Prevenion in he Process Indusries, 3, 16: [12] Weng, W.G., Fan, W.C., e al. Experimenal Sudy of Back-draf in a Comparmen wih Openings of Differen Geomeries, Combusion and Flame, 3, 132: [13] Weng, W.G., and Fan, W.C., Experimenal Sudy on he Miigaion of Backdraf in Comparmen Fires wih Waer Mis, Journal of Fire Sciences, 2, : [14] Weng, W.G., and Fan, W.C., Miigaion of Backdraf wih Waer Mis: a Reduced-scale Experimenal Sudy, Process Safey Progress, 3, 22: [] Weng, W.G., and Fan, W.C., Nonlinear Analysis of he Backdraf Phenomenon in Room Fires, Fire Safey Journal, 4, 39: [16] ang, R., Weng, W.G., Fan, W.C., e al. Subgrid Scale laminar Flamele Model for Parially Premixed Combusion and is Applicaion o Backdraf Simulaion, Fire Safey Journal,, 40: [17] Weng, W.G., Fan, W.C., e al. Sudy on Sal Waer Modeling of Graviy Currens prior o Backdrafs using Flow Visualizaion and DPIV, Experimens in Fluids, 2, 33: [18] Weng, W.G., A Simulaive Sudy on Backdraf Phenomenon in Comparmen Fires, Ph. D disseraion, Universiy of Science and Technology of China, 2. [19] Quiniere, J.G., Seckler, K., and Corley, D., An Assessmen of Fire Induced Flows in Comparmens, Fire Science and Technology, 1984, 4: [] McCaffrey, B.J., and Heskesand, G., A Robus Bidirecional Low-velociy Probe for Flame and Fire Applicaion, Combusion and Flame, 2: