ASHRAE LO-09-019-2009 Comparisons of Numerical Predictions and Filed Tests in a Road Tunnel《在公路隧道申请试验和数值预测的比较》.pdf

《ASHRAE LO-09-019-2009 Comparisons of Numerical Predictions and Filed Tests in a Road Tunnel《在公路隧道申请试验和数值预测的比较》.pdf》由会员分享，可在线阅读，更多相关《ASHRAE LO-09-019-2009 Comparisons of Numerical Predictions and Filed Tests in a Road Tunnel《在公路隧道申请试验和数值预测的比较》.pdf（11页珍藏版）》请在麦多课文档分享上搜索。

1、2009 ASHRAE 221ABSTRACTA research project has been conducted at the National Research Council of Canada (NRC) to evaluate the effective-ness of in-place emergency ventilation strategies to control smoke spread in the event of a fire in a road tunnel. Some of these strategies date back to the design

2、of the tunnel (1964). Following a recent fire, the operating instructions were revised. A scientific based evaluation of these operation instructions is the main objective of the research study. The research includes both numerical and experimental studies. The numerical study uses the Computational

3、 Fluid Dynamics (CFD) model Fire Dynamic Simulator (FDS) to investigate smoke ventilation in tunnels. The experimental study is used to provide the necessary initial and boundary conditions for the CFD model.In-situ fire tests were conducted in an operating road tunnel using a fire source of 2 MW (1

4、.9 x 103BTU/s). Temper-ature, airflow velocity, pressure and smoke optical density (SOD) values were measured. These data was used to validate the numerical model against small fires.CFD simulations were conducted to compare with measured field data. Comparisons were made, at the near and the far fi

5、eld of the fire source, of several parameters included: volumetric airflow, temperature, and SOD values The CFD simulations were able to replicate field tests trends. They provided insight into phenomena observed during the in-situ fire tests. This understanding had been used as the basis to improve

6、 the performance of tunnel ventilation system to control smoke spread. The comparison studies showed that both numerical predictions and experimental measurements were, in general, comparable.As such, the validated CFD model can be used to comple-ment the experiments to analyze different fire and ve

7、ntilation scenarios. They offer a predictive tool for the situations where the actual fire tests prove to be cumbersome to conduct.INTRODUCTIONFires are, in general, very complex in nature. Their complexity arises from the fact that the physical and chemical processes (e.g., turbulence, combustion,

8、radiation, etc.) controlling fire and smoke development interact with each other, and with the surroundings. For the systematic design of an effective fire protection system, it is essential that the important transport process controlling fire development is properly understood and that the key com

9、ponents are clearly identified. Because of the mutual interactions of these processes and their coupling with any enclosure, reduced-scale experiments alone are often not sufficient to reproduce full-scale features. Mathematical models supported by full-scale experiments offer a practical solution f

10、or a better under-standing of fire and the fluid dynamics involved.Fire protection design in tunnels as in other building types has been, up to the 1990s, based upon empirical approaches supported by a rationale from experimental studies of fire and experience gained from actual fire disasters. The

11、problem with this approach is that it is not possible to conduct enough exper-iments or fire tests to adequately deal with all possibilities. A good example of the application of empirical techniques to the problems of fire in road tunnels, was given by (Heselden 1976). Improvements in state-of-the-

12、art have traditionally alternated between experimental and mathematical modelling approaches. Experimental approaches include full-scale phys-ical model tests or on-site measurements.Comparisons of Numerical Predictions and Filed Tests in a Road TunnelAhmed H. Kashef, PhD, PEngMember ASHRAEAhmed H.

13、Kashef is a senior research officer in the Fire Research Program at the Institute for Research in Construction, Ottawa, Ontario, Canada.LO-09-019 2009, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in ASHRAE Transactions 2009, vol. 115, p

14、art 2. For personal use only. Additional reproduction, distribution, or transmission in either print or digital form is not permitted without ASHRAEs prior written permission.222 ASHRAE TransactionsWith the advent of more powerful computers, the Compu-tational Fluid Dynamics (CFD) modelling techniqu

15、e is rapidly expanding. CFD models solve the complex differential equa-tions describing the conservation of mass, momentum, enthalpy, species, at several thousand nodes within the tunnel. The mass and momentum equations describe the fluid dynam-ics. Whereas the enthalpy equation and the species conc

16、entra-tions equations depict the heat transfer and the transport of combustion products and unburnt fuel. Thus, CFD models simulate the overall fire environment for a specific fire scenario including ambient conditions prior to a fire and provide local predictions of temperature, velocity, smoke con

17、centration, etc, as a function of time. As such, CFD model-ling approach offers the prospect of a more general predictive tool for the calculation of conditions prevailing during a fire and for the optimization of smoke control and fire protection measures.A major difference between CFD models is ho

18、w the viscosity used in the momentum equation is calculated. The majority of CFD models, e.g. k- models (Wilcox 1993), use turbulence models to approximate the turbulent energy and dissipation produced by the fire. This approach results in a solution to an averaged version of the flow equations.On t

19、he other hand, the Large Eddy Simulation (LES) (McGrattan 2000) approach solves the large scales of motion and models the small scales that are assumed to be universal. The LES approach results in an unsteady solution to the Navier-Stokes equations. Because real turbulent flow situa-tions are inhere

20、ntly unsteady, LES methods can have an advantage in modelling turbulent fire-induced flows.The Fire Dynamic Simulator (FDS) CFD model (McGrattan 2000) is based on the LES approach and solves a form of high-speed filtered Navier-Stokes equations valid for a low-speed buoyancy driven flow. These equat

21、ions are discretized in space using second order central differences and in time using an explicit, second order, predictor-corrector scheme.In FDS, fire is represented using the “mixture fraction-based” combustion model. This model directly simulates large-scale convective and radiative transport p

22、henomena. The small length and time scales physical processes are, on the other hand estimated. The actual combustion process in the fire is not simulated. As such, the model inherently assumed that the reaction of fuel and oxygen is infinitely fast (fuel and oxygen cannot co-exist and they will rea

23、ct at any temperature). The local heat release rate is computed from the local oxygen consumption rate at the flame surface. FDS has been the subject of numerous validations (McGrattan 2000). The vali-dation efforts included: comparison with full-scale tests conducted specifically for code evaluatio

24、n, comparison with engineering correlations, comparison with previously published full-scale test data, comparison with standard test, and comparison with documented fire experience.The main objective of the article is to rationalize the use of the FDS model in tunnel applications. This was achieved

25、 by comparing the FDS predictions against in-situ fire tests measurements conducted using a fire source of 2 MW (1.9 103Btu/s). The validated CFD model was then used to analyze fire scenarios where the actual fire tests prove to be cumber-some to conduct. Kashef and Bnichou (2008) conducted a numeri

26、cal parametric study to assess the performance of the emergency ventilation systems of the road tunnel for a fire size of 30 MW (2.8 104Btu/s).Tunnel Ventilation SystemThe road tunnel (Figure 1), built in 1964, is located in Montreal city, and travels underwater in a north-south direc-tion. The tunn

27、el is 1.8 km (1.1 mile) long with three lanes in each direction, inside two concrete tubes. Two ventilation towers are located at the ends of the underwater section. A control and monitoring centre for the tunnel is located at the North tower. A central section separates the two tubes. Galler-ies lo

28、cated in this section are used to supply air along the tunnel length via openings distributed along the walls and these galleries can also be used as evacuation routes. Doors at various locations along the length of the tunnel provide access to the gallery. There are doors between the galleries prov

29、iding a route to the other roadway. The wall openings have adjust-able dampers to ensure uniform air distribution. The side vents are situated in two rows: upper and lower. The lower and the upper rows are located at heights of 1.0 m (3.281 ft) and 3.9 m (13.1 ft) above the tunnel floor, respectivel

30、y, and at intervals of approximately 6 m (19.7 ft). The two rows of vents are offset by 3 m (9.8 ft).The tunnel ventilation is provided by a semi-transverse ventilation system with local extraction points (Figure 2). The ventilation system is composed of 8 ceiling exhaust fans (4 fans for each roadw

31、ay) and 8 fans that supply air through the Figure 1 General layout of the tunnel.ASHRAE Transactions 223side vents which are uniformly distributed along one wall for each roadway. All fans can operate in reverse mode. There-fore, fresh air may be supplied at either the ceiling (fans VE151 through VE

32、254), or by fans VA101 through VA204 through the side vents. In the exhaust mode, fans VE151 through VE254 can operate at full or half capacity, and in the supply mode they can only operate at full capacity. In the supply mode, fans VA101 through VA204 can operate at: half, three-quarter, or full ca

33、pacity. In the exhaust mode, these fans can only operate at full capacity.FIELD FIRE TESTSThe need for reliable experimental data is common to validation of all fire models. While there is a significant amount of experimental data available from large-scale fire tests in tunnels, many of those tests

34、 were performed for reasons other than validation of fire models. In view of this and due to the fact that the tunnel under investigation is an oper-ating tunnel, it was logical to proceed with conducting full-scale field tests using an appropriate fire size to minimize damage to the tunnel structur

35、e and its components. The main objective of the current article is to validate the use of FDS CFD code for tunnel applications.Four fire tests were conducted at two positions in the North Roadway of an operating road tunnel under two differ-ent ventilation scenarios. It was concluded from the measur

36、ed data that the ventilation system helped to control the temper-atures and produced enough airflow velocity to prevent the occurrence of the backlayering phenomenon. The full details of the experimental work of the study are documented in refer-ences (Kashef et al. 2003, 2004, 2005).Two fire locati

37、ons were selected (Figure 2): one close to the exhaust fans at the north end of the tunnel (Test 1) and one in the middle of the tunnel (Test 2). Two different ventilation scenarios were activated for the two tests. For Test 1, the two fans VE151 and VE153 were operated in the exhaust mode and the t

38、wo fans VA103 and VA201 in the supply mode. In Test 2, a similar ventilation scenario was used except for fan VA103, which was used in the exhaust mode. A clean-burning propane burner system that produces minimal smoke is developed for the in-situ fire tests (Figure 3). The burner is about 1.3 m (4.

39、3 ft) wide and 1.8 m (5.9 ft) long. This system is a compact, portable and convenient heat source that is capable of producing up to 5 MW (4.7 103Btu/s) of Figure 2 Ventilation system of the tunnel.Figure 3 Propane burner system.224 ASHRAE Transactionsheat output simulating a small car fire. Artific

40、ial smoke is used for visualization purposes with four smoke bombs added at 1 min intervals. The heat output of the burner is measured using a gas tube variable area rotometer set for propane gas density with 2% accuracy on full-scale readings.A fire size of 2 MW (1.9 103Btu/s) is selected so as to

41、minimize damage to the tunnel structure and its components and to produce reliable data for the calibration and validation of the numerical models. Temperature and smoke optical density (SOD) were measured (Figure 4). These measure-ments were taken at sixty locations downstream of the fire (Position

42、s 1 through 3) and twenty locations upstream of the fire (Positions 4 through 8). In addition, air speed was measured at selected lower and upper side vents, the four ceil-ing exhaust fans (VE fans), the two portals, the middle cross-section of the tunnel, and inside the evacuation paths located bet

43、ween the two roadways. Upstream is defined in the context of this paper as the direction from which the traffic enters the tunnel.A vane anemometer is used to measure the air velocity and temperature in the range of 0.40 to 25 m/s (79 to 4921 fpm) and 0 to 60C, respectively, with resolutions of 0.1

44、m/s (20 fpm) and 0.1C. The accuracies of measurements are 2% for the velocity and 0.8C (33F) for the temperature.A “Pulsed White Light LED” system built in-house is used to measure the obscuration in terms of OD/m. The system has a wavelength in the range (0.450-0.650 mm, 1.5 103 2.1 103ft) and cons

45、ists of a white light emit-ting diode (LED) source, aperture and detector. A silicon pin diode was used as the detector. The LED was pulsed to mini-mize the affects of ambient light and temperature variations. This LED source approximates the light sensitivity of the human eye. The system is accurat

46、e to 5% from 400 to 700 nm (1.3 106to 2.3 106ft) and from 0.4 OD/m (0.12 OD/ft) to 4 OD/m (1.2 OD/ft).A thermocouple tree was constructed at each fire location to measure the centerline plume and ceiling temperatures (Figure 4e). The thermocouple tree included eleven thermo-couples, distributed vert

47、ically along the tunnel height and horizontally at the ceiling spanning 6 m (19.7 ft) upstream and downstream from the fire. Figure 5 shows the time plots of the vertical and ceiling temperature profiles at the fire location for the tests, respectively. The measurements were recorded for 25 min at e

48、ach location. The fire grew to the maximum heat release (2 MW, 1.9 103Btu/s) in 5 min and was maintained at that level for about 15 min. The activation of emergency ventilation scenarios modified the flow field in the tunnel. Airflow speeds of 2.5 and 4.6 m/s (492 and 906 fpm) in the south-north dir

49、ection were recorded at fire locations for Tests 1 and 2, respectively. As a result, the fire plume was deflected towards the downwind direction. Figure 3 shows a visual angle of deflection of the fire plume of about 25-30 for Test 2. The degree of deflection depends on both the wind speed and plume upward velocity. Raj et al. (1979) suggested the following relationship to calculate the plume deflection of LNG pool fires:(1)where and fare densities of ambient air and fuel vapour respectively, Hcis the heat of combustion, u is the wind speed, and w is th