1、232 2009 ASHRAEABSTRACTTemperature and smoke spread in the early stage of a fire were modeled, using computational fluid dynamic techniques, and compared with data obtained from field tests conducted in an operating roadway tunnel in the City of Montreal, Canada. Fire characteristics, including temp
2、eratures and smoke spread over the tunnel were measured during these tests. Two types of fire scenarios were simulated: gasoline pool fires under vehi-cles and gasoline pool fires behind vehicles. The estimated fire size used in the simulations was 650 kW. The initial and bound-ary conditions of eac
3、h simulation were set to mimic the condi-tions of the corresponding test. Comparisons were made to temperature and smoke optical density measurements. In general, favourable comparisons between the numerical predictions and the experimental data were observed. The ceil-ing temperature downstream of
4、the fire decreased with an increase in the distance from the fire source, which is also the case for smoke optical density. The ceiling temperatures produced by the fire behind the vehicle were higher than those produced by the fire under the vehicle. The temperature vari-ation along the central cro
5、ss section of the tunnel shows that the highest ceiling temperature occurs 35 m downstream of the fire because the plume was tilted by the airflow inside the tunnel. Fire location had a significant impact on ceiling temperature development in the tunnel. The airflow conditions at the fire location s
6、ignificantly affect smoke and temperature distributions in the tunnel which will also affect the perfor-mance of detection systems.INTRODUCTIONIn a tunnel environment, development of fire and smoke spread are affected by the fire set-up and ventilation conditions in the tunnel. During normal traffic
7、 operation, smoke can be diluted or pushed away from the detection system by the normal ventilation system, which is designed to maintain acceptable levels of contaminants in the tunnel (Beard and Carvel 2005). It can create conditions that may challenge the ability of detectors to detect and locate
8、 the fire in the early stage if the fire is enclosed in a vehicle or located behind an obstruction. In order to achieve early detection of fires in a tunnel, it is essential to understand how fire develops and smoke spreads during the initial stage of fire under various conditions.An extensive Compu
9、tational Fluid Dynamics (CFD) study was carried out as part of the International Road Tunnel Fire Detection Research Project (Liu et al. 2006a), which aimed at investigating the detection performance of current fire detection technologies. The CFD study included simula-tions of full-scale tests cond
10、ucted by the National Research Council of Canada (NRCC) in the Carleton University labo-ratory tunnel and a series of simulations to examine effects of various fire scenarios and different ventilation schemes. Find-ings of this CFD study were (Kashef et al. 2008 and Ko et al. 2008); Simulated result
11、s exhibited relatively good agreement with laboratory test results.Temperature development inside the tunnel was consid-erably affected by fire scenarios, such that temperature rise near the ceiling was less significant for fires enclosed by a vehicle body than that for open fires.The simulations ag
12、reed with the laboratory test results in that the longitudinal airflow affected the burning behaviour of the fire and smoke spread in the tunnel. Moreover, the impact depended on the relative size of CFD Study of Smoke Movement during the Early Stage of Tunnel Fires: Comparison with Field TestsYoon
13、J. Ko George V. Hadjisophocleous, PhD, PE Ahmed Kashef, PhD, PEMember ASHRAE Member ASHRAEYoon Ko is a PhD candidate in the Fire Protection Engineering Program and George Hadjisophocleous holds the Industrial Research Chair in Fire Safety Engineering and is a professor at Carleton University, Ottawa
14、, Ontario, Canada. Dr. Ahmed Kashef is a senior research officer at the Fire Research Program of the National Research Council of Canada (NRC), Ottawa, Ontario, Canada.LO-09-020 2009, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in ASHRA
15、E Transactions 2009, vol. 115, part 2. For personal use only. Additional reproduction, distribution, or transmission in either print or digital form is not permitted without ASHRAEs prior written permission.ASHRAE Transactions 233fire to the airflow velocity, as well as the fire scenario. In general
16、, the ceiling temperature decreased with an increase of airflow.The development of temperature depended on the venti-lation scheme (longitudinal, semi-transverse and fully-transverse ventilation systems) inside the tunnel. The length of the tunnel did not have a significant impact on the temperature
17、 development near the ceiling close to the fire location. Thus, the results found in the laboratory tunnel scale can reasonably be extrapolated to longer tunnels.In order to further investigate and verify these results, field tests were conducted in an operating tunnel environment. This paper report
18、s the CFD study carried out to simulate the field tests conducted in Tube A of the Carr-Viger Tunnel in Montreal. As well, the paper presents the results of the study and comparisons between model predictions and experimen-tal data.Numerical Simulations of Field testsThe current research employs the
19、 Fire Dynamic Simula-tor (FDS) version 4.07 (McGrattan and Forney 2006), devel-oped by the National Institute for Standard and Technology, to study the fire growth and smoke movement in road tunnels. FDS is based on the Large Eddy Simulation (LES) approach and solves a form of high-speed filtered Na
20、vier-Stokes equa-tions, valid for low-speed buoyancy driven flow. These equa-tions are discretized in space using second order central differences and in time using an explicit, second order, predic-tor-corrector scheme. Turbulence parameters used in simula-tions were 0.2, 0.5, and 0.5 for Smagorins
21、ky constant, turbulent Prandtl, and Schmidt number, respectively. For combustion, FDS uses a mixture fraction method based on equilibrium chemistry. Fire is modelled as the ejection of pyrolyzed fuel from the fuel surface that burns when mixed with oxygen (McGrattan and Forney 2006). Fire modelling
22、and smoke generation are modelled based on specified stoi-chiometric parameters and yields for soot. Three CFD simulations were carried out to simulate field tests conducted in the Carr-Viger Tunnel in Montreal. The aim of this study is to compare the simulation results of smoke movement with actual
23、 test results. Detailed description of the field tests can be found in Liu et al. (2008b). Comparisons were made to temperature and smoke optical density measure-ments.MODEL DESCRIPTIONThree simulations were carried out with variations of fire set-up and location of the fire. Table 1 lists simulatio
24、ns and conditions used for each of the three simulations.Model GeometryThe simulated tunnel section was a 4-lane, 420 m long, 5 m high and 16.8 m wide, as shown in Figure 1. The fire was Figure 1 Model description.Table 1. List of SimulationsSimulation IDNRC TEST #FireScenarioGasolinePan SizePeak HR
25、R(kW)Location of FireAmbient Temperature (C)Tun4VF3 Test 3 UV10.6 m 0.6 m 550650 90 m downstream of the fan 17Tun4VF6 Test 6 BV20.6 m 0.6 m 550650 90 m downstream of the fan 17Tun4VF8 Test 8 BV20.6 m 0.6 m 550650 60 m downstream of the fan 171 Pool fire located under a simulated vehicle body.2 Pool
26、fire located behind a simulated vehicle.234 ASHRAE Transactionsplaced in the first lane, 4.2 m away from the north wall, as marked in Figure 1. The initial and boundary conditions of each simulation were set to mimic the conditions of the corre-sponding test. The boundary condition for walls, ceilin
27、g and floor was concrete. The west end of the tunnel was open. Airflow into the tunnel was specified through the east end corresponding to measured airflow velocities (about 1.3 m/s) of each experiment. The combination of this boundary condi-tion and the air flow of a jet fan resulted in a flow of 1
28、.4 m/s near the fire, which is close to the averaged velocity measured during tests of the fire location.Grid SizeGrid convergence tests were conducted, in which 650 kW fire source (0.6 m 0.6 m) was simulated under natural venti-lation condition. Two different grid sizes were used; 0.3 m and 0.4 m.
29、In addition, one grid setting using two overlapping meshes was also tested. In the setting, 0.3 m grid size was used for the tunnel, and 0.1 m grid size was used for fire area. Temperature variances over time were compared for different grid settings. Although the high resolution in the combustion v
30、olume predicts better in the combustion area, it does not improve the temperature prediction at the far field (Hadjiso-phocleous and McCartney 2005). Since the interest of this study is temperature variance over time at some distance from the fire, the optimal spatial size 0.3 m was selected to save
31、 computation times. The grid size of the middle of the model tunnel was 0.3 m (D) 0.3 m (W) 0.3 m (H) (Figure 1). To save cells for the rest of the long tunnel, the grid stretching technique (McGrattan and Forney 2006) was used. For the rest of the tunnel, a non-isometric grid of 0.3 m (D) 2 m (W) 0
32、.3 m (H) was used since it was found from grid tests that this grid did not affect the temperature results in the middle section where comparisons were made with the experiments.Ventilation ConditionThe tunnel ventilation in the tests was maintained under normal operating conditions. The tunnel has
33、a longitudinal ventilation system that is equipped with four ceiling jet fans (one in each lane spaced at 3.45 m). Detailed description of the jet fan can be found in Liu et al. (2008b). The fans are located in a recess in the tunnel ceiling, at which the maximum height is approximately 9 m. The bas
34、e of the fans is at a height of 6 m. In these simulations, only one jet fan located on the ceiling of the third lane was activated. To model the jet fan in the model tunnel whose ceiling height was 5 m, the horizontal jet fan capacity was calculated taking into account the angle of the jet flow. As
35、shown in Figure 1, the ceiling jet fan was mounted on the ceiling of the third lane. In the model a horizontal flow of 21 m3/s (45,000 cfm) over an area of 1.5 m 0.5 m was defined.During the field tests, air velocity measurements were conducted at a number of cross sections of the tunnel, includ-ing
36、 the section where the fire was located. Using a hand-held velocity meter, measurements were taken prior to and during the tests. Measurements showed that air velocities were rela-tively uniform at the section where the fire was located, and the average air velocity was 1.4 m/s. In the field tests,
37、air velocities were found to be more uniform and stable with an increase in distance from the fans (Liu et al. 2008b). In the simulations, the same longitudinal airflow with a velocity of approximately 1.4 m/s was achieved at the section where the fire was located.Fire ScenariosThe field fire tests
38、in an operating tunnel were carried out using fire scenarios developed in Task 1 of the International Road Tunnel Detection Project for evaluating performance of road tunnel detectors. Two types of fire scenario were simu-lated as in the field tests: a pool fire located under a mock-up vehicle body
39、(UV) and an open pool fire located behind a vehi-cle (BV). The fire size was approximately 650 KW. These fire scenarios are encountered in the majority of tunnel fire inci-dents and presented a challenge to the fire detection systems (Liu et al 2006b). The same fire scenarios were tested in the prev
40、ious laboratory tunnel study as well as in this field tests. Detailed descriptions on these scenarios as well as geometry of mock-ups are provided in Liu et al. (2006b). As in the field tests, the effect of changing fire location was also simulated. The fire source was placed at two different locati
41、ons in the tunnel. For simulation Tun4VF3 and Tun4VF6, the fire was placed, in the first lane, 90 m downstream of the fan, whereas the fire was placed, in the first lane, 60 m downstream of the fan for Simulation Tun4VF8, as shown in Figure 2.Figure 3-(a) shows the vehicle body mock-up used in the f
42、ield tests for a fire under a vehicle scenario in which a vehicle crashed, and the fuel leaks formed a pool under the vehicle body. In the simulations, a steel plate vehicle mock-up was built over the pool pan. The size of the plate was 1.5 m wide by 2.4 m long and was located 0.5 m above the ground
43、. A 0.6 m 0.6 m gasoline pool fire was placed under this obstruction.Figure 2 Schematic of thermocouple and smoke meter locations (Kashef et al. 2008b).ASHRAE Transactions 235For an open gasoline pool fire located behind a vehicle, which is a more general tunnel pool fire scenario (Liu et al. 2006b)
44、, a metal plate (2.5 m wide by 4.2 m high) obstacle simulating the front portion of a crashed truck was placed at a distance of 6 m in front of the pool fire and 0.3 m above the ground, as in the field tests Figure 3-(b). Fire and Smoke ModellingIn FDS, the burning rate of gasoline can be prescribed
45、 by specifying the heat release rate of the fire, or alternatively the burning rate can be predicted based on the energy fed back from the fire. It was found, from preliminary simulations, that simulating the actual burning was sensitive to the thermophys-ical properties of the fuel and boundary con
46、ditions, particu-larly the vent conditions. Therefore, the fire source in this study was modelled by prescribing the heat release rate, elim-inating some complications, and avoiding potential errors due to a combination of factors, such as insufficient grid resolution and uncertainty in the absorpti
47、on coefficient and flame temperature.The growth rate was modified to correspond to the test results so that the burning rates reflected the environment. The ramp-up time parameter was used to control the burning rate such that the fire developed within 60 seconds as in the tests. Satisfactory result
48、s were obtained in modeling these field tests.InstrumentationThe predicted temperature and smoke optical density were compared against the field data. Temperatures were monitored at three different locations corresponding to the thermocouple trees used in the tests. Figure 2 shows schematic of therm
49、ocouple and smoke meter locations. Drop-1 and Drop-2 were placed at the middle of the tunnel with five ther-mocouples spaced at 1.0 m intervals starting 1 m above the tunnel floor. Drop-F was placed above the fire placed in the first lane with four thermocouples spaced at 1.0 m intervals staring 2 m above the pool. The spacing between drops was 15 m for Tun4VF3 and Tun4VF6. For Tun4VF8, Drops 1 and 2 were placed at 30 m and 60 m downstream of the fire, respec-tively. Smoke optical density values were monitored at two heights, 4.4 m and 2.7 m above the tunnel flo
copyright@ 2008-2019 麦多课文库（www.mydoc123.com）网站版权所有
备案/许可证编号：苏ICP备17064731号-1