1、471 8 Full-Scale Burning Tests of Mechanical Smoke Exhaust in Large Atrium L. Yi W.K. Chow, Ph.D. R. Huo N.K. Fong, Ph.D. Y.Z. Li, Ph.D. Member ASHRAE C.W. Leung C.L. Shi W.Y. Hung H.B. Wang ABSTRACT Removing smoke at the early stage of a buildingfire would assist in evacuation. Mechanical smoke ext
2、raction systems are commonly installed in larger buildings. In designing such a system, makeup air must beprovided to displace the hot smoke. However; air supplied would also provide additional oxygen for combustion. Therefore, it is important to study how air should be supplied. In this paper; the
3、eflciency of mechanical exhaust in an atrium with diferent arrangements for air supply will be discussed. Full-scale burning tests on smoke filling and mechanical extraction were conducted in a full-scale burning facility, the PolyUiUSTCatrium, with inner dimensions of 22.4 m x 11.9 m x 27.0 m, as a
4、 collaborativeproject between The Hong Kong Polytechnic University (PolyU) and University of Science and Technology of China (USTC). Numerical simula- tions with a computational fluid dynamics (CFD) package- thejre dynamics simulator (FOS) developed at National Insti- tute of Standards and Technolog
5、y-were also carried out. The predicted results agreed with experimental measurement from the full-scale burning tests. This study suggested that air inlets should be locatedat a certain height above thejre (i.e., in the PolyU/USTC atrium with a 1.3 MWjre in the center of the flool; locating the air
6、inlets at 3 to 4 m r9.8 to 13.131 higher than the tip of the flame will be better), and their area should be big enough to lessen the mix of makeup air with the smoke plumes. INTRODUCTION More atriums have been built in Southeast Asia and can be found in gymnasia, complex shopping malls, theaters, a
7、nd warehouses. There, fire safey provisions, especially on smoke management systems, should be designed carefully. Because of the large space without compartmentation, it is difficult to confine the fire and smoke to the immediate area of origin. Smoke will spread quickly, creating psychological ill
8、 effects to the occupants and making it difficult to locate exits. There- fore, smoke management systems should be provided in these buildings to protect the occupants from the smoke hazards. Engineering design guides for smoke management systems can be consulted (Klote and Milke 2002; NFPA 2000). M
9、echanical exhaust is commonly used for smoke control, though the efficiency of the system depends on many factors. The exhaust capability affecting the airflow pattern, expressed as mass flow rate or mass flux of the fan (in kg/s or Ib per hour), is a critical parameter in designing the fan and duct
10、 system. For describing how much smoke is produced, air entrainment rate into the smoke plume, which depends on the height of the smoke layer and the heat release rate (HRR) of the fire (Zuko- ski et al. 1980), should be estimated. Performance of the mechanical exhaust system will be affected by the
11、 height of the smoke layer and heat release rate of the fire. To describe the efficiency of an exhaust system, the ratio of exhaust rate to smoke production rate is commonly used. Work reported in the literature on mechanical exhaust systems in atria include smoke development under mechanical exhaus
12、t by Tanaka and Yamana (1985). The effects of different exhaust rates, heat release rates, and exhaust inlet heights were investigated by Lougheed and Hadjisophocleous (1997). There, results of numerical predictions were reported and physical model stud- ies of an atrium with mechanical exhaust (Had
13、jisophocleous and Lougheed 1999; Hadjisophocleous and Fu 1999) were carried out. In recent years, a full-scale experimental study on L. Yi, R. Huo, Y.Z. Li, C.L. Shi, and H.B. Wang are with the State Key Laboratoiy of Fire Science, University of Science and Technology of China, Hefei, Anhui, China.
14、W.K. Chow, N.K. Fong, C.W. Leung, and W.Y. Hung are with the Department of Building Services Engi- neering, The Hong Kong Polytechnic University, Hong Kong, China. 02004 ASHRAE. 267 the effect of air inlets on mechanical extraction was made by Huo et al. (2001a). In this paper, both full-scale burni
15、ng tests and numerical simulations with computational fluid dynamics (CFD) or fire fieldmodeling will be applied to study the effects of the area and position of vents for air supply. Because of the large volume space, temperature rise in the PolyU/USTC Atrium under a 1.6 MW (1.52 x lo3 BWs) fire is
16、 still not high (Chow et al. 2001; Hu0 et al. 2001b). Forma- tion of a smoke layer will take time, and, as a result, a clear smoke layer was not observed at the early stage of a fire. Airflow in the atrium is driven primarily by fire-induced buoy- ancy prior to activation of the smoke exhaust fans.
17、Upon acti- vation of fire protection systems, such as smoke exhaust or sprinkler, it becomes increasingly difcult to predict both the airflow pattern and temperature distribution in the atrium. There were studies on applying two-layer zone models, such as CFAST and FIERAsmoke, and CFD fire “field” m
18、odels to study smoke movement in an atrium. The CFD soft- ware fire dynamics simulator (FDS Version 2), developed at the National Institute of Standards and Technology (NIST), based on large eddy simulation (LES) (e.g., McGrattan et al. 2000), was selected to study airflow induced by the fire in thi
19、s paper too. This model had been applied to studying problems in fire-induced airflow and compared with other CFD approaches (e.g., Yin and Chow 2002). EXPERIMENTAL STUDIES Arrangement Full-scale burning tests were carried out in the PolyU USTC Atrium, built as a 20-year collaborative project betwee
20、n the University of Science and Technology of China (USTC) and The Hong Kong Polytechnic University (PolyU) since 1997 (Hu0 et al. 1998). The geometrical configuration of the atrium is shown in Figure 1. It is 22.4 m (73.5 fi) long, 1 1.9 m (39.0 ft) wide, and 27 m (88.6 ft) high. The vents for air
21、supply in the process of mechanical exhaust could be any combination of the windows and door of the atrium. The size of each window is 1.4 x 1.1 m (4.6 x 3.6 ft). Two fans were installed at the top of the atrium. The dimensions of each ofthe exhaust air ducts are 1.2 x 1.2 m (3.9 x 3.9 ft), and the
22、volume flux of each fan is 15.0 m3/s (3.18 x 1 O4 ft3/min). A fuel bundle of 1.0 x 1.0 x 0.3 m (3.3 x 3.3 x 1.0 ft) was placed near the center of the floor. The ambient temperature was measured to be at 15C (59F). Five tests, labeledT1, T2, T3, T4, andT5, were conducted in the PolyUAJSTC Atrium with
23、 different arrangement of air inlets as listed in Table 1. The labels Ei-j, Ni-j, and Wi-j mean the east, north, and west windows, respectively; i is the floor number; and j is the window number (see Figure 1). In T1 and T2, the vertical positions of the air inlets are the same at 6.5 m (21.3 ft) ab
24、ove the ground, but they are of differ- ent areas. In T1, T3, and T5, the air inlets are of the same area but different vertical positions. The height above the floor of the air inlet centerline in T4 and T5 is 11.0 m (36.1 fi), while the area of the air inlets in T4 is about two-fifths of that in T
25、5. Only those air inlets listed above for each of the five tests were open during the respective individual experiments. Both of the fans were switched on at about 40 seconds after fuel ignition. At that time, the smoke front had already descended to approx- imately 15 m (49 ft) above the atrium flo
26、or. Diesel oil (heat of combustion is 42,000 kJ/kg, 1.8 1 x 1 O4 Btu/lb) of mass 6 kg (13.2 lb) was used in the experiments. Transient mass of the fuel bundle was recorded in each exper- iment. Burning time of each test varied from 180 to 220 seconds. Three thermocouple trees were hung in three come
27、rs of the atrium (see the black cross and dotted lines in Figure 1). Two trees are of type-K thermocouples (TS1 and TS2,27 thermo- couples on each tree at intervals of 1 .Om 3.3 fi) and one oftype T thermocouples (TS3,lO thermocouples at 2.0 m 6.6 fi inter- vals). Velocity at the inlets was measured
28、 with a portable anemometer. Experimental Results Variation of the fuel mass for test T1 is shown in Figure 2. An amount of 4 kg (8.8 lb) of cold water was put into the bottom of the fuel pan to give a horizontal plane. As diesel oil is lighter than water, it was kept above the water while burning.
29、About 8 MJ (7.58 x IO3 Btu) heat was estimated to boil some water. This value can be ignored in comparing with the total heat released by the fuel bundle of 250 MJ (2.37 x IO5 Btu). The burning lasted for about 220 seconds. During the initial 50 seconds, the mass loss rate increased slowly. From 50
30、seconds to 170 seconds, the mass loss rate kept constant at about 0.033 Figure I Geometrical conjiguration of the atrium. ASHRAE Transactions: Research kg/s (0.073 lb/s). Multiplying the loss rate by the heat combus- tion of diesel oil gives the heat release rate of the fire about 1.38 MW (1.31 x lo
31、3 BWs). The steady HRRoffire in T1 can be taken as 1.38 MW (1.3 1 x lo3 Btus). Similar curves were measured for tests T2 to T5, where the HRR was 1.35 MW (1.28 X lo3 Btu/s) with a burning time of 215 seconds;l.41 MW (1.34 x 1 O3 BWs) with burning time of 180 seconds; 1.32 MW (1.25 x lo3 Btus) with b
32、urning time of 200 seconds; and 1.33 MW (1.26 x lo3 BWs) with burning time of 190 seconds, respectively. The flame height in the experiments was observed to be about 2.5 m (8.3 ft). 42 40 38 34 32 O 50 100 150 200 250 300 TimCl5 Figure 2 Variation of the mass of fuel bundle. The velocity measured du
33、ring the test when the fans were operating in normal mode is listed in Table 2. There was one point at each window and four points at the door (average values are presented in Table 2) measured for air velocity. The ambient wind velocity was measured to be less than 0.2 m/s (40 FPM). This value is r
34、elatively small compared with the values at the air inlets when the fans were operating in five tests. The total air makeup rate calculated from the measured velocity was roughly equal to the total air extraction rate (30 m3/s or, 6.36 x lo4 ft3/min) except in T3. It was difficult to get a steady ve
35、locity at the four testing points at the door because of large door area. The smoke layer temperature was observed to be lower than 35OC (95F). The variations of clear height for all of the five tests are shown in Figure 3. Each of the thermocouples in their individually fixed locations above the at
36、rium floor serves the purpose of defining the smoke layer height as a function of time, by virtue of the thermocouples continuous measurement of temperature at that fixed location. The descending smoke layer height can be assumed to coincide with the fixed height of a given thermocouple at the speci
37、fic point in time in which the temperature measured by that thermocouple begins to rise above the observed ambient temperature. Monitoring of the transient vertical temperature distribution would indicate the instantaneous positions of the smoke layer clear height (Hc). Table 1. Parameters for Each
38、Test Height of Test Inlets Duration Test Inlets Used (m) (s) T1 windows E2-1, E2-2, E2-3, E2-4, N2-1, 6.5 220 N2-2, N2-3, W2-1, W2-2, W2-4 windows E2-1, E2-2, E2-3, E2-4 windows E3-1, E3-2, E3-3, E3-4 11.0 200 190 N3-2, N3-3, W3-1, W3-2, W3-4 Table 2. Velocity at Air Inlets Capacity of Ambient 1.32
39、30 15 30 I 1.33 1 ASHRAE Transactions: Research 269 . 30 25 20 15 10 II 5- - c 7 - - - - 4 E rn a w s O O O X O O O x o T1 o T2 a T3 x T4 o T5 O O a o I I l I I O 50 100 150 200 250 Figure 3 Variation of clear height. From Figure 3, the smoke layer started to descend at about 50 seconds in each test
40、. At the early stage when descending (from 50 seconds to 100 seconds), the flow structure was not stable after the ceiling jet reached the wall. Smoke descended quickly and the smoke/clear-air interface was not well defined. Comparing the experimental results of T1 with those of T2, it is observed t
41、hat the smoke layer interface height descended more slowly in T 1 than in T2. Supplying air by the mechanical exhaust system in T1 was more effective than in T2 delaying the descent of smoke with its larger area of air inlets. Similar conclusions can be drawn for tests T4 and T5. In T2 (T4), the are
42、a of air inlets is smaller than that of T1 (T5), so the velocities at the inlets were bigger than that of T1 (T5). It resulted in the mixing of smoke and makeup air being inten- sified, and the efficiency of the mechanical exhaust was reduced. The smoke layer descended at almost the same rate initia
43、lly for tests T1 and T3, as shown in Figure 2. After 100 seconds, it was obvious that the smoke layer descended more slowly in T1 than in T3. When comparing the results of T1 and T5, it is observed that the smoke layer descended more quickly in T5 than in T1. Therefore, the effect ofmakeup air on ex
44、haust of T1 was greater than in T2, T3, and T5, resulting in a slower rate of smoke spread. Low-velocity makeup air ( 1 ds) is required in some design guides such as NFPA 92B for minimizing its effects on the plume. When makeup air is supplied below the smoke layer with higher velocities, the plume
45、motion will be affected (deflected or scattered) by the makeup airflow, leading to more mixing with smoke. When the makeup air is supplied above the smoke layer, it will enter directly into the smoke layer to mix with the smoke. In both cases, more smoke will be produced, and the mechanical exhaust
46、system might become not so efficient. The vertical positions ofthe makeup air should be designed carefully under the conditions of high airflow. With low positions of the makeup air, the plume and flame will be affected by the airflow of makeup air. Stronger mixing of smoke with air in the atrium wi
47、ll result at lower positions rather than at higher positions. Ifthe position ofthe makeup air is too high, the smoke layer will descend below the air inlets earlier. The position of the makeup air should not be too low nor too high above the fire. In the above experiments, supply- ing makeup air at
48、the height about 3.5 m (1 1.5 ft) above the flame tip was found to be more effective in delaying the descending of the smoke layer than in the other cases. Since the thermocouples were placed at different posi- tions, the measured temperature profiles would be different. For example, TS3 was placed
49、under the ductwork of the exhaust fan. Therefore, the smoke layer interface height deduced from the vertical temperature profile measured at TS3 will be different from that of TS 1. The measured clear heights by thermocouples at different locations were not the same for a given test. Considering the maximum danger to occupants, clear heights shown in Figure 3 are the minimum of the three values at each time. The results will be verified in the following section by CFD simulations. NUMERICAL SIMULATIONS Configurations CFD simulations were carried out by the LES code
