1、4734 Smoke Control for Retail Shops with Cabin Design W.K. Chow, Ph.D. Member ASHRAE ABSTRACT It is notfeasible to provide dynamic smoke exhaust for the entire hall in some terminals or malls with a large volume space. The cabin design is commonly used with a ceiling roof covering the retail areas,
2、giving smaller shops. Active jre protection systems, including dynamic smoke exhaust andjre suppression systems, are expected to be installed. In this paper; design aspects of the smoke control system in such retail shops will be discussed. It is found that jre suppression systems should be provided
3、. The new concept of using a water mist system, as reported earlier; will be further conjrmed. Full-scale burning tests are useful in assessing the performance of water mist systems in such a shopjre. The research $re jeld model Fire Dynamics Simulator FDS is a firs (a) Cabin Figure 1 Geometry appli
4、ed to simulate thejre environment with and without the roof and the roof with downstands. INTRODUCTION Big halls are found in malls, airports, railways, and bus terminals, where it might not be feasible to protect the entire hall space with a dynamic smoke exhaust system. However, small retail shops
5、 packed with combustibles (e.g., Chow 2002) are found there. Fire safety in these small retail shops should be considered carefiilly, especially for those in public trans- port terminals where the passenger loading is extremely high during rush hours. The “cabin” design (Law 1990; Beever 1991; Bress
6、ington 1995) is commonly used with a ceiling roof constructed above the retail area with higher fire risk. This gives a smaller protected area in those retail shops, as shown in Figure 1. This is similar to the situation for enclosed exhi- Much bigger protected space ci, (b) Hail W. K. Chow is chair
7、 professor of architectural science and fire engineering and director of the Research Centre for Fire Engineering in the Department of Building Services Engineering, The Hong Kong Polytechnic University, Hong Kong. 41 7 02004 ASHRAE. bition spaces within a large exhibition hall. Currently, in the U.
8、S., these spaces are required to be sprinklered rather than have a smoke control system. As a “bare cabin” (Chow 1997) might lead to rapid flashover in some cities such as Hong Kong, fire protection in such a small protected area would include the following: ing the same air change rate of 12 such a
9、s those required in many fire codes (e.g., FSD 1998) in the cabin and hall, gives Vcob. 3600 Phu 3600 = 12 - hall cab Dynamic smoke exhaust system Fire suppression system to limit the heat release rate A design fire is required to determine the smoke produc- tion rate in the space concerned. Some fi
10、gures were recom- mended in standards and guides (e.g., NFPA 1995). However, there were always arguments, as very low figures, such as 500 kW (525 Bids), had been used in a huge terminal hall! There were some data on total heat release rates (e.g., Chow 1997) for burning combustibles in small retail
11、 shops with smoke exhaust and a sprinkler. For example, design fires deduced from large-scale fire tests in a sprinklered calorimeter were reported (Garrad and Smith 1999; Grant et al. 2000). But these tests were started from a small fire, such as an “igniter,” due to a short-circuited electrical ap
12、pliance, a litter bin, or a gas burner, as used in some standard fire tests. There is no radi- ation heat flux applied to test the samples, as in a cone calo- rimeter (e.g., Babrauskas and Grayson 1992). Results are useful for understanding how a fire grows, develops to flash- over, and then spreads
13、 to adjacent areas, but this will not give the contribution of materials, nor their assemblies, to a fire under flashover condition (Chow et al. 2003a). The heat release rate measured would not be too high, as only a small amount of the combustibles were ignited, and, for most cases, fire suppressio
14、n systems might be operated to reduce the resultant heat release rate. The tests then become demonstra- tions of the operational sequence of the systems under the tested fire. This is not good enough for understanding the actual heat release rate and the possibility of igniting the combustibles unde
15、r flashover condition. The situation should be reviewed as the number of fires other than those due to acci- dents that were reported. Rigs similar to an “industry calorimeter” in Sweden (e.g., Mnsson et al. 1994) should be developed to burn an actual retail shop for studying how much heat would be
16、released. This is expensive but necessary with the concept pointed out previously. There, burning tests in small cabins were performed in a new fill-scale burning facility, the PolyUIHEU Assembly Calorimeter (Chow et al. 2003a; Chow 2001a), developed as a collaboration project between the Harbin Eng
17、i- neering University (HEU) and The Hong Kong Polytechnic University (PolyU). Preliminary results on heat release rate in flashover shop fires will be reported in this paper. ON DYNAMIC SMOKE EXHAUST Smoke exhaust in a cabin would take advantage of using a smaller exhaust rate to compare with that i
18、n the entire hall. Taking subscripts cab and hall for the space volume V(in m3) and volume flow rate ofthe exhaust system V (in m3s-), keep- or Taking Qfre (in kW) as the heat release rate of the fire to give temperature rise AT (in OC) and p (in kg.m”) as the air density, conservation of heat gives
19、 and For the same Ofire, “cab - vhall - hall AThall vcab cab The following can be observed from the above two key equations: A much lower smoke exhaust rate is required in the cabin, i.e., only a ratio of ( vcab I vh,/) in comparing with the value in the hall. For a hall of length 40 m (132 ft), wid
20、th 20 m (66 t), and height 20 m (66 ft), Vharr is 16000 m3 (575,000 fi3). A cabin of length 10 m (33 ft), width 4 m (13.2 ft), and height 4 m (13.2 ft) gives vcab of 160 m3 (5,750 fi3. Therefore, (Vcab I Vhar) is only (i / 100). However, the ratio of the temperatures rise ATcab i ATha/ will also be
21、increased by a ratio of (Vhau / vcab), say (100 I 1) as in the above example. Note that normally, the smaller the compartment, the less the minimum heat release rate required for flashover (Thomas 1981). A post-flashover fire will be ventilation-controlled, with the heat release rate depending on th
22、e ventilation factor of the opening. Retail shops used to have large openings for easy access, and the heat release rate can be very high, There- fore, fire control systems (e.g., Law 1990; Beever 199 1 ; Hume 1997) or fire suppression systems (e.g., Chow and Ya0 2001) have to be installed in the ca
23、bin to limit the heat release rate. In this way, Qfire in cabin and hall areIlot the same, say as denoted by ofire cab and 0f;i-e hall 1 (3) 418 ASHRAE Transactions: Research The same temperature rise can be achieved if vhuf, ofire cub = vcab. Qjre hull . (4) But it is unlikely that (Qfi, cab o, hal
24、l) can be reduced to the same ratio as (V, Vhull) for operating a fire “control” system. Fire suppression systems, such as those using water mist, are preferred, as pointed out before. Note that a sprinkler system is only possible for “control,” not to “suppress” nor to “extinguish” the fire. The co
25、mbustibles would keep on burning with the heat release rate being non- zero. However, performance of a water mist fire suppression system depends on the fire scenario and should be demon- strated by experiments (Chow and Ya0 2001; Chow et al. 2003b). FULL-SCALE BURNING TESTS Several full-scale burni
26、ng tests were carried out to study the heat release rates that resulted from burning combustible items in a retail shop under flashover condition at the PolyU/ HEU Assembly Calorimeter (Chow et al. 2003a, 2003b; Chow 2001a). A set of results on a newsstand of length 3.6 m (12 fi), width 2.4 m (8 ft)
27、, and height 2.4 m (8 fi), with a door ofheight 2 m (6.6 ft) and width 0.8 m (2.6 ft), was taken to illustrate how big is the heat release rate and how effective is the fire suppres- sion system. A total amount of 90 kg of books, magazines, and news- papers were placed on display boards of width 2 m
28、 (6.6 ft) and height 2.2 m (7.3 ft) at the back, left, and right walls of the shop. Flashover was onset by a gasoline pool fire first. Transient results on heat release rate are shown in Figure 2a. All books, magazines, and newspapers placed on the display boards were consumed completely. The peak h
29、eat release rate was up to 3.2 MW (3.36 x IO3 Btu/s). Time I s (a) No water mist discharged Figure 2 Results on heat release rates. The effect of water mist discharged from a low-pressure fire suppression system at 1.2 MPa (12 bar) operating pressure was also demonstrated (Chow et ai. 2003b). Arrang
30、ements of combustibles were the same as above. Water mist was discharged manually at 105 s after ignition when the heat release rate was about 3 MW (3.15 x lo3 Btu/s)-very close to the peak value under this configuration. Results on heat release rate are shown in Figure 2b. The heat release rate ros
31、e initially after discharging water mist. The fire then appeared to be extinguished about 99 s later. No flames were observed, and the heat release rate reduced to very low values. After turning off the water supply, the fire ignited again 20 s later, and all books and magazines on the display board
32、s were consumed. This point should be further investigated. Results on the discharging time required for different fire scenarios will be reported later. However, it is well demon- strated that a water mist system can “suppress” such a fire effectively. SIMULATIONS WITH A FIELD MODEL Fire field mode
33、ls, or applications of computational fluid dynamics (CFD), are now very popular in studying smoke movement in buildings (e.g., Cox 1995). A CFD program was developed at the Building and Fire Research Laboratory, National Institute of Standards and Technology (NISS), to simulate the transport of smok
34、e and hot gases in an enclosure fire. This program, Fire Dynamics Simulator (FDS) version 3.01 (McGrattan et al. 2000), was based on years of research effort (Baum and Rehm 1984; McGrattan et al. 1998) and should be a key element in fire safety engineering. There, smaller scale chemical reactions we
35、re modeled by large eddy simulations. The larger scale of buoyancy-induced turbulence structure was simulated directly by solving a set of hydrody- namic equations with only low mach number flow. A series of Water mist discharged Water mist stopped E. -7 Em-/ ! . p! 8 ! 8 - b Time I s (b) With water
36、 mist discharged ASHRAE Transactions: Research 41 9 ScemioS3: With i m downstands (a) Scenario S1: With a fire only Figure 3 Geometry of the hall. validations and verifications were carried out during the past few years (e.g., McGrattan et al. 1998). Combustion in FDS was modeled by the mixture frac
37、tion combustion model when large eddy simulations are used. This software is now applied to study the fire environment with and without cabin design. Three scenarios were simu- lated: Scenario S1 to study the plume generated by a pool fire in a hall was simulated. The hall is of length 40 m (132 fi)
38、, width 40 m (1 32 fi), and height 40 m (1 32 It). There is a ceiling with the four sides opened. A 5 MW (5.25 x lo3 Btu/s) fire of length 1 m, width 1 m, and height 0.5 m was located at the center of the hall, as in Figure 3a. A fine grid system (48 x 48 x 96) was assigned with simu- lations up to
39、100 s. The time steps were adjusted automat- ically on the software to satisfy the convergence criterion during the calculation. Simulations were carried out in a personal computer, requiring about 50 hours of CPU time. Results at 100 s on the predicted velociy vectors and temperature contours are s
40、hown in Figures 4 and 5. Scenario S2 was simulated in the same hall but with a shop roof of area 3 m by 3 m placed above the fire, as in Figure 3b. This is part of the cabin design where fire shutters can be activated to enclose the shop. Sprinklers installed would be activated to suppress a fire, a
41、nd smoke extraction systems would operate to extract smoke generated. However, these suppression effects are not studied by FDS in this paper. Results on the fire environment at 100 s predicted by FDS are shown in Figures 4 and 5 as well. It is well demonstrated that the presence of the shop roof, i
42、.e., scenario S2, would lead to a biggerplume (e.g., Chow 2002; Chow 1997), in comparison to scenario S 1 without it. These predictions agreed with the earlier CFD studies based on the Reynolds averaging Navier-Stokes (RANS) equationmethodwith k- temperature under the roof is hotter and so flashover
43、 is more likely to occur. Therefore, combustible materials inside should be tested under the flashover condition (Chow et al. 2003a) to give the heat release rate, not just starting from a pilot flame. A fire suppression system should be installed to reduce the heat release rate to lower acceptable
44、values. Smoke exhaust systems are required to extract smoke away from the retail areas. CONCLUSION Fire safety provisions are normally designed for protec- tion against accidental fires. However, the number of arson fires over the world appears to be increasing (Chow 2001b; SCMP 2003). The general p
45、ublic is now very concerned about the hidden fire hazard, even when they are traveling on an 420 ASHRAE Transactions: Research (a) S1: With a fire only 5 ms + (a) S1: With a fire only (b) S2: With a shop roof (b) S2: With a shop roof 30 (i (c) S3: With downstands (c) S3: With downstands Figure 4 Elo
46、city vectors for the three scenarios at I O0 s. Figure 5 Temperature contours for the three scenarios at 100s. ASHRAE Transactions: Research 421 underground railway (SCMP 2003). Perhaps it is the right time to review the fire safety codes, and the government is doing that at the moment. Shopping mal
47、ls with small retail shops are always crowded with people. For those malls having dificul- ties installing a smoke management system for the entire hall space, cabin design has to be used. This concept, reviewed above, is that a fire suppression system such as water mist is necessary. Otherwise, a v
48、ery big plume will form as demon- strated by fire simulation with the field model FDS version 3.01. Obviously, a heat release rate database for local combus- tibles should be developed (e.g., Babrauskas and Grayson 1992; Peacock et al. 1994), and carrying out full-scale burning tests is necessary. R
49、esults are useful for specifying the design fire for smoke management system (e.g., NFPA 1995) in those big halls. Fire safety management (Della-Giustina 1999) can be recommended based on such studies. For example, books and magazines are suggested not to be displayed vertically. Fire suppression systems should be installed inside the cabin. A water mist system is an appropriate candidate as pointed out before (Chow and Ya0 2001). However, perfor- mance of water mist fire suppression systems should be eval- uated by full-scale burning tests (Chow et al. 2003a,
