1、4779 Zone Modeling Simulations on Retail Shop Fires Q. Ku ABSTRACT Retail shop fires of different sizes and ventilation condi- tions will be simulated with the zone model FIRE WIND. Emes to flashover in those shops are estimatedfiom 1760 simula- tions under four sets of NFPA $fires. From the results
2、, corre- lations for theflashover time with the geometry of the shop are derived. Effects of operating thesmoke controlsystem at differ- ent ventilation rates are also simulated. Integrating the sprin- klersystem with the smoke controlsystem in a shop is discussed by referring to their relative time
3、s of operation. Correlation equations on the flashover time with other parameters useful to building authorities are derived. INTRODUCTION There are many small retail shops in the big halls of Hong Kong (now the Hong Kong Special Administrative Region HKSAR). Those places are of higher fire risk tha
4、n halls with- out small shops storing large amounts of combustibles. For this analysis, the shops were protected by sprinkler and smoke control systems. Some of them (Law 1990; Beever 1991, 1995; Bressington 1995, 1999) might be called “cabin design” if located in big terminal halls where there are
5、difficulties in installing smoke control systems for the entire space. A point to consider in a small retail shop is the likelihood of flashover and its consequences. This will be affected by the fire load density associated with the nature of the shop, the design of smoke control and sprinkler syst
6、ems, and fire safety management. The local upper limit on fire load density is 99,561 Btu/ft2 (1135 MJm-2) (Fire Services Department 1998), which is being criticized (Jones 1998). Note that the sprinkler systems might not be as reliable as experienced locally. Even if it works, the fire will not be
7、extinguished W.K. Chow, PhD Member ASHRAE immediately. A “small” shop might become fully involved, especially if it stores highly combustible products such as alcohol. Putting in other factors such as poor management of retail shops-as in storing excessive goods, blocking the sprinkler heads, and co
8、vering the smoke control vents by goods-a very big fire could result. This paper will analyze flashover times in small retail shop fires (Chow 1997a, 1997b; Kiu and Chow 2000) with the fire protection engineering software FIREWIND (1 997). Correlation equations between the time to flashover and the
9、geometry of the fire are presented in this paper. Those equa- tions are useful in understanding the possibility of flashover. Smoke spreading out to the hall from the shop fire is another Concern. Earlier analysis (e.g., Chow 1997a) showed that a big plume will be induced from a small shop. Therefor
10、e, a smoke exhaust system has to be provided in the shop, but this would affect performance of the sprinkler system in control- ling a fire. Integrating the two systems will be discussed in this paper by referring to their relative times of operation. FLASH OVER The flashover criteria during a compa
11、rtmental fire was reviewed by Thomas (1995), whose work has been updated frequently as in Peacock et al. (1999) and Chow et al. (2003). In those works, a heat balance equation for the compartment was set up, as reported by Thomas (1995), for the analysis. The term describing heat gain is basically t
12、he heat release rate of the fire. At the initial stage of a compartmental fire, the fire is fuel-controlled and the heat gain depends on the temperature. As the fire proceeds, it will be changed to a ventilation- controlled fire with the burning rate becoming independent of the compartment temperatu
13、re. Heat loss includes the heat Q. Kui is a graduate student and W.K. Chow is chair professor of architectural science and fire engineering and director of the Research Centre for Fire Engineering in the Department of Building Services Engineering, Hong Kong Polytechnic University, Hong Kong. 02005
14、ASHRAE. 407 transfer through the wall surfaces and the heat taken away by the air. This term is roughly proportional to the compartment temperature. The difference between the heat gain and the heat loss will determine the rate of rise of the compartment temperature. Quasi-steady states can be found
15、 when the heat gain is the same as the heat loss. Plotting the two terms against the aver- age smoke temperature would give something similar to a Semenovs diagram in classical thermal explosion theory (Thomas 1995) for determining the criteria for flashover. From the intersection points of these tw
16、o terms, the likelihood for flashover can be judged by practical criteria (Thomas 1995) on radiation heat flux of 3 17.46 Btuft2 (20 kWm-2) at floor level or upper layer temperature of 932F (500C) to 11 12C (600C). All this is summarized in the literature as reported by Thomas (1 995). FIRE SIMULATI
17、ONS Using a two-layer zone model is good enough to study the likelihood of flashover in a compartment fire (Peacock et al. 1999; Chow et al. 2003). There have been numerous experi- mental validations of this kind of model. The flashover time for fires in a “small retail shop” will be studied using t
18、he fire protection engineering tool FIREWIND. The fire environment in “shops” inside a big hall will be predicted first. Single shops with different dimensions in a mall were considered. The length L was taken as 13.12 ft (4 m), 19.68 ft (6 m), 26.24 ft (8 m), and 32.80 ft (10 m); width was taken as
19、 13.12 ft (4 m), 19.68 ft (6 m), 26.24 ft (8 m), and 32.80 ft (10 m); and height H was 6.56 ft (2 m), 9.84 ft (3 m), 13.12 ft (4 m), and 16.40 ft (5 m). Usually, a fire resistance rating of two hours is required for the walls. Natural ventilation conditions with opening heights H, from 4.92 ft (1.5
20、m) to 8.20 ft (2.5 m) and width W, from 4.92 ft (1.5 m) to the shop width (Le., up to 32.81 ft IO m) were assessed. The number of openings varied from one to four, depending on the design conditions. Larger ventilation area will increase the air intake rate, which gives higher burning rate. The mini
21、mum heat release rate for flashover in that shop will then be increased. A fire of size 3.281 ft (1 m) by 3.281 ft (1 m) was located at the center. Four sets of heat release rate curves were used with the initial stage following NFPA slow, medium, fast, and ultra-fast t2 fires (NFPA 2000), then at c
22、utoff values of 17.065 M Btuh (5 MW), 25.598 M Bhdh (7.5 MW), and 34.13 M BWh (10 MW) until 2000 s, and then dropped to O Btu/h (O MW) at 2100 s. A total number of 1760 simulations were performed with FIREWIND for the four sets of fires on 11 cases with 40 configurations. The initial temperature was
23、 taken to be 68F (20C). Flashover in the enclosure was said to occur when the upper layer gas temperature reached 932F (500C) as listed in the Users Manual (FIREWIND 1997). Values of the flash- over time tffor some cases are shown in Table 1. The following can be observed. Values of tf increased as
24、the dimension of the shop increased, i.e., 952 s for a shop 13.12 ft (4 m) x 13.12 ft (4 m) x 6.56 ft (2 m) to 1245 s for a shop 13.12 ft (4 m) x 19.69 ft (6 m) x 6.56 ft (2 m) with one opening of 13.12 ft (4 m) x 4.92 (1.5 m). The flashover time tf slightly changed with the ceiling height for a slo
25、w t2 fire with cutoff value of 17.065 M Btiuh (5 MW). Value of tf increased to 997 s for a 13.12 ft (4 m) x 13.12 ft (4 m) x 16.40 ft (5 m) shop when there was one opening of 13.12 ft (4 m) x 4.92 ft (1.5 m). Note that in the above case, $-was 952 s when the ceiling height was 6.56 ft (2 m). Flashov
26、er did not occur for shops longer than 26.24 ft (8 m) and wider than 26.24 ft (8 m) for a slow t2 fire. Very similar results for $-were found with the medium t2 fire, fast t2 fire, and ultra-fast-t2 fire with other opening arrangements, as shown in Table 1. Results were similar for the 23.891 M Btuh
27、 (7.5 MW) cutoff fires. Flashover occurred in very few cases (as shown in Table 1) when the opening width was equal to the shop width, even for the big fire with cutoff value of 34.13 M BWh (10 MW). CORRELATION RELATIONS The flashover criteria in a “shop fire” were studied using the predicted result
28、s of the two-layer zone model of FIREWIND. Note that those flashover criteria are only criteria for practical application without physical basis (Thomas 1995; Peacock et al. 1999). Predicted flashover time ?(in s) for the above simula- tions is correlated with floor areaAfin the range of 172.04 ft2
29、(16 m2) IAf b O L - G m N ZI t ASHRAE Transactions: Research 409 4 a vi W vi O r- Cu r- W m - m vi LCI 3 x M vi m 3 ASHRAE Transactions: Research - W z - N 4 - d O N - m O N 3 m 3 0 e 3 3 N - r- O N - N O N - W 4 - O N d n 2 N 00 3 W 2 io 2 N 2 I- z O I- 3 m 00 3 W O CY W O N _. io C N 3 r- d I 3 2
30、O e; r- 2 N 2 d d 3 O N m O W r- O N - Cu io N - 3 I- N I + 3 m N - 00 N N 00 00 N O 00 N N O m cr, 2 + - 3 d -4 z m 2-i 00 W N io 00 N 00 I- N + io W io o d W d N iD W N Wm ww 3mm io000 wmd I W m 9 - N I- W N m m - m 00 m d d - c4 2 I 4- - O m d - m 00 d 00 d io - W N io - d N io - N W io - m io io
31、 - D x - W z t O O I- - 3 m W io io W O CI 00 ir, 2 + N O 00 cr, . z W 3 - - O m m I O W 00 - cr, 2 m W W r- W W - I - 3 2 3 - r- 2 3 m 00 00 + b, 00 r- r- m vi d d io 2 3 N 00 2 D 00 3 3 io 00 2 3 N 2 io io + - 3 O W 3 m N 2 O 00 00 O r- m 2 3 3 O 00 d 2 3 3 Pl 00 2 W 4; ri3 00 r- 2 00 I- io 3 - W
32、W 2 I.!: - ,- 3 . - n 00 Z 3 v W X d ir, 2 3 CI W t - 6 N. W L I I n N 2 “ 3 h N v 2 I.! 3 h N 2 I.! 3 h N W 2 I.! 3 h N o! z I.! 3 n N 2 I.! 3 I N 3 ASHRAE Transactions: Research 41 1 It can be seen inat ine fiashover time increased with the shop height H. This is more reasonable than the simulatio
33、ns reported (Chow 1997a, 1997b) previously. Further, values of the flashover time ?were shorter than the duration of burning (2000 s) of the fire. The thermal pene- tration time might be quite long for some materials such as thick timber. But it is possible that flashover occurred before heat penetr
34、ated through the wall reaching steady-state. SMOKE EXTRACTION IN THE SHOP It is important to install a smoke control system in a shop with the effect of operating the system simulated by FIREWIND. A shop 13.12 ft (4 m) long, 13.12 ft (4 m) wide, and 9.84 ft (3 m) high, with a 6.56 fi (2 m) x 8.20 ft
35、 (2.5 m) opening was taken as an example. Extraction rates of 40, 80, 120, and 180 air changes per hour (ACH) were simulated. The system is expected to operate at time top. Two values of top were considered by a smoke detector: The system will operate when smoke layer thickness falls a typical “down
36、stand“ depth of 1.48 ft (0.45 m) as specified in the local building fire code (Buildings Department 1996). Value of top was 37 s in the shop. The system will operate when the smoke layer tempera- ture is heated up to the operation temperature for sprin- kler head of 154.4“F (68C). The corresponding
37、top would be 150 s. Note that a 154.4“F (68C-rated) sprin- kler head might not be actuated immediately even if the smoke layer temperature reaches 154.4“F (68C). The response time index (Heskestad and Smith 1982) has to be considered. As discussed later in this paper, it is no good for the smoke con
38、trol system to be operated first if both systems are installed. Results of the smoke layer temperature and smoke layer interface height with the extraction system operated at the two operation times are shown in Figures 2 and 3. It is observed that smaller extraction rates would not have any effect
39、on the smoke layer temperature and smoke layer interface height. The smoke layer can be kept at a higher position only if the extraction rate is sufficiently high, say, increased to 40 ACH. The time to operate the smoke control system is impor- tant. For earlier operation of the extraction system at
40、 37 s, when the smoke layer fell by 1.48 ft (0.45 m), the smoke layer could be kept at a higher level within 500 s. But if the system was operated at a later time, at 150 s, the smoke layer thickness was already 4.27 ft (1.3 m). Higher extraction rates are required for keeping smoke at a high level
41、at the early stage of the fire. For example, 180 ACH would bring smoke up to 9.19 ft (2.8 m) before 500 s. Operating the smoke control system with extraction rates less than 60 ACH would not be useful. I“ 1300 - 1200 . 1100 - 1000 IO 20 30 40 50 60 70 (1076) (2152) (3228) (4304) (5380) (6156) (7532)
42、 Floor area Ar I mz (n) Figure 1 Flashover time for slow $fire. 03m(9%ft) - 3 m (9.84 fi) : lined o 4 m (13.12 n) - 4 m (13.12 fi) : nned + Sm(l6.4R) - 5 m (10 n) :nitrd INTEGRATING SPRINKLER SYSTEM WITH SMOKE EXHAUST SYSTEM Sprinkler systems are also installed in a shop for control- ling the fire.
43、Basically, there are at least four mechanisms (Chow and Fong 1991) by which a sprinkler water spray inter- acts with a fire: Direct cooling of the burning object. Prewetting the walls, floors, and objects not yet ignited. Cooling of the smoke layer. Displacing oxygen from the burning object. The fir
44、st two mechanisms are very important in control- ling a fire. Before discharging water, the sprinkler heads must first be activated. The thermal sensitivity of a sprinkler head is very important, and response time index is one of the param- eters to specifj how fast a sprinkler head would operate. T
45、he time for the sprinkler to operate might be quite long as it takes time to transfer heat from the surrounding hot smoke. For a sprinkler head with response time index RTI and temperature rating Ts (154.4“F 68“C in common design) subjected to hot gas at temperature Tg moving toward it with speed Vg
46、, the time constant T of the sprinkler head is given by (Heskestad and Smith 1982): z = RTI V;“. (2) The time tops taken for the sprinkler head to operate, i.e., rising temperature from an initial value of To to Ts (say, 154.4“F 68“C), is (Ts-To) = (Tg-To)(l-e -fops/j (3) For smoke with Tg of 212F (
47、lOOC), Vg of 1 ms-, and initial temperature To at 68F (20“C), putting these numerical values to Equation 3 gives tops = 0.916 RTI. (4) 41 2 ASHRAE Transactions: Research 9.84 Timet/s -No extraaion O40ACH 7r-80 ACH +12OACH *180ACH (a) Smoke layer temperature. Figure 2 Extraction system operated at 37
48、 s. ” O 500 lo00 1500 2OOO Time t I s -No extraction O40 ACH +120 ACH c180ACH *- 80 ACH (a) Smoke layer temperature. Figure 3 Extraction system operated at 150 s. ”imet/s (b) Smoke layer interface height. -No extraction 040ACH + lut ACH c 180 ACH +t- 80 ACH 31 -1 9.84 i $1 O 6.56 y i ft 3.28 O 500 l
49、o00 1500 2Ooo Timet/s (b) Smoke layer interface height. -No extraction 040 AM ASHRAE Transactions: Research 41 3 KTI for a normal sprinkier head is about 724.52 fi!”s!” (400 m1/2s”2) and for an early suppression, fast response (ESFR) sprinkler head is about 126.79 ft1/2s12 (70 m12s12). Putting RTI in Equation 4, tops for a normal sprinkler head is about 366 s and for an ESFR sprinkler head is 64 s. That means even if the smoke temperature is increased to 392F (200C) (it took longer than 400 s for the example shown in the above sect
