ASHRAE.AT-01-11-2-2001TheSmokeHazardfromaFireinHighSpaces.pdf

THIS PREPRINT IS FOR DISCUSSION PURPOSES ONLY, FOR INCLUSION IN ASHRAE TRANSACTIONS 2001, V. 107, Pt. 1. Not to be reprinted in whole or in part without written permission of the American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc., 1791 Tullie Circle, NE, Atlanta, GA 30329. Opinions, findings, conclusions, or recommendations expressed in this paper are those of the author(s) and do not necessarily reflect the views of ASHRAE. Written questions and comments regarding this paper should be received at ASHRAE no later than February 9, 2001. ABSTRACT In a recently completed research project sponsored by ASHRAE (899-RP), the National Research Council of Canada used full-scale physical model studies combined with CFD modeling to investigate the effectiveness of mechanical smoke exhaust systems used for atrium smoke management. The previous publications resulting from this research focused on the impact of plugholing on the effectiveness of a smoke management system. In this paper, the research results are used to investigate the impact on smoke venting of the ceiling jet formed when the plume impinges on the ceiling. Also, alge- braic equations and a computational fluid dynamic (CFD) model were used to estimate smoke properties in the smoke layer. The numerical results are compared with the experimen- tal results. INTRODUCTION In a recently completed research project sponsored by ASHRAE (899-RP), the National Research Council of Canada (NRC) used full-scale physical model studies combined with CFD modeling to investigate the effectiveness of mechanical smoke exhaust systems used for atrium smoke management (Lougheed and Hadjisophocleous 1997; Hadjisophocleous et al. 1999; Lougheed et al. 1999). This research addressed concerns raised by designers and research- ers regarding the possibility of fresh air being pulled into the exhaust inlet for smoke management systems in which the headroom for accumulation of smoke above the highest egress route is minimal. This plugholing of the exhaust inlet by the fresh air from below the smoke layer can decrease the effi- ciency of the smoke exhaust system and can result in an increased smoke layer depth, which may expose building occupants to smoke. In addition to the plugholing phenomena, NFPA 92B (1995) also recommends that the smoke layer depth below the exhaust inlets must be deep enough to accommodate the ceil- ing jet, which is produced when the buoyant plume produced by a fire comes into contact with the ceiling of a compartment. The ceiling jet is the flow of smoke under the ceiling extending radially from the point of fire plume impingement on the ceil- ing. Normally, the temperature of the ceiling jet is greater than the adjacent smoke layer. There has been considerable research on the ceiling jet. However, most of this research has focused on problems related to detector activation (Beyler 1986). Very little research has been conducted to investigate the impact of the ceiling jet on smoke venting using mechanical systems. However, design guides recommend that an atrium smoke exhaust system be designed assuming a minimum smoke layer depth of at least one-tenth the floor-to-ceiling height (Klote 2000). In this paper, the results from the research project on plugholing are used to investigate the ceiling jet phenomenon. Also, the impacts of the ceiling jet and plugholing on the effec- tiveness of a mechanical smoke exhaust system are compared. For some smoke management applications in high atria, in which the ceiling jet and plugholing phenomena can have an impact, the smoke accumulating at the top of the atrium will be appreciably diluted by entrained air, decreasing the hazard posed by the smoke. There are algebraic equations available (NFPA 1995) for use in estimating the smoke properties, both during the smoke filling phase and with venting. Milke (2000) has discussed the smoke hazard during early development of The Smoke Hazard from a Fire in High Spaces Gary D. Lougheed, Ph.D.George V. Hadjisophocleous, Ph.D., P.Eng. Member ASHRAEMember ASHRAE Gary D. Lougheed and George V. Hadjisophocleous are senior research officers in the Fire Risk Management Program, Institute for Research in Construction, National Research Council Canada, Ottawa, Ontario. AT-01-11-2 (RP-899) 2AT-01-11-2 (RP-899) the fire. The algebraic equations and a computational fluid dynamic (CFD) model were used to determine the properties of the smoke in the smoke layer during venting. In this paper, the numerical and experimental results are compared. PHYSICAL MODELS Test Facility Two experimental test facilities were used during the project on plugholing. For the larger of the two facilities, the ductwork for the mechanical exhaust system was suspended below the ceiling of the large-scale test facility (Lougheed et al. 1999). The exhaust inlets were approximately 1.4 m below the ceiling height. As such, the results from the tests using this facility do not provide any information on the effect of the ceil- ing jet. Also, the instrumentation used in this facility was limited to thermocouples used to determine temperature profiles. In the small-scale test facility, a plenum system was mounted above a large room and 32 exhaust inlets were extended through the ceiling into the test compartment (Figure 1). The facility was 9 m 6 m 5.5 m with a door on the west wall near the southwest corner and a door on the east wall near the northeast corner. The interior wall surface of the Figure 1Small-scale physical model. AT-01-11-2 (RP-899)3 compartment was insulated using 25 mm thick rock fiber insulation. The insulation was used for two reasons: first, to protect the walls of the facility so that high gas temperatures could be attained during the tests and, second, to provide a better boundary condition for the CFD runs. A detailed description of the test facility is provided in Lougheed and Hadjisophocleous (1997). A fan was used to supply fresh air into the compartment through openings in the floor around the walls as shown in Figure 1. The openings were designed to maintain the velocity of the incoming air to less than 1 m/s for the maximum air flow expected, which was between 2 and 4 m3/s. Thirty-two exhaust inlets with a diameter of 150 mm were located in the ceiling of the compartment, as shown in Figure 1. These inlets were used to extract the hot gases from the compartment during the tests. All exhaust ducts were connected through a central plenum and ductwork to an exhaust fan. Instrumentation The test facility was instrumented with thermocouples and pitot tubes for velocity measurements. Also, gas inlets were located in the room for extracting gas samples to deter- mine CO2 concentrations at various locations. The location of the instrumentation is shown in Figure 1. A detailed descrip- tion of the instrumentation is provided in Lougheed and Hadjisophocleous (1997). Test Procedure and Parameters Tests using the small-scale test facility were conducted over an extended period of time, using a propane burner fire located at the center of the test facility. During the test period, the propane burner was used to provide a series of steady-state fire conditions. Each fire size was maintained for several minutes allowing the conditions (temperature and CO2 concentrations) in the test room to come to equilibrium. The exhaust system was operated throughout the test. The main parameters, which were varied in the tests, were as follows: 1.Heat release rate. Tests were conducted with the following heat release rates: 15, 25, 50, 150, 250, 300, 400, 500, 600, and 800 kW. 2.Number of exhaust inlets. Tests were conducted with 1, 4, 16, and 32 exhaust inlets with the exhaust inlets 150 mm in diameter. In addition, tests were conducted with 32 exhaust inlets with a 75 mm diameter. 3.Exhaust inlet height. Tests were conducted with the exhaust inlets at heights of 5.5 m (150 mm below the ceiling), 4.5 m, and 3.5 m. 4.Exhaust inlet orientation. Most tests were conducted with the centerline of the exhaust inlet oriented vertically. For comparison, a limited number of tests were conducted with the centerline of the exhaust inlets oriented horizontally. 5.Fan speed. Tests were conducted with two fan speeds for tests with 32 exhaust inlets. Only the low fan speed was used for tests with 1, 4, and 16 exhaust inlets. In order to extend the range of test parameters that could be investigated, extensions were added to the exhaust inlets for some tests. The results of the tests with extensions provided a test arrangement that minimized the impact of the ceiling jet. For these tests, plugholing was the dominant factor, which had an impact on the effectiveness of the mechanical exhaust system. A series of tests was also conducted with the exhaust inlets located at ceiling height. The results of these tests are used in this paper to investigate the impact of both the ceiling jet and plugholing on smoke venting. NUMERICAL MODELING Description of CFD Model The numerical simulations for this project were done using a general three-dimensional computational fluid dynamics model with capabilities in handling laminar and turbulent flows, incompressible and compressible media, multi-component fluids, porous media, Lagrangian particle tracking, reacting combusting flows, conjugate heat transfer, surface-to-surface radiation, rotating frames of reference, and subsonic, transonic, and supersonic flows (ASC 1994). The grid generation features of the model include the ability to handle nonorthogonal boundary-fitted grids, grid embedding, and grid attaching. Turbulence was modeled using the k- turbulence model, which is the model used for most engineering applications and found in most commercial codes. Radiation exchange between the hot gases and the surroundings was modeled using the models diffusion radia- tion model (ASC 1994) with a gas absorption coefficient of 0.1. Fire Modeling The fire was modeled using two methods: a flamelet combustion model (Peters 1984, 1986) and a heat source. In the flamelet model, only two user-defined scalar equations are used: one is the mixture fraction and the other is the variance of the mixture fraction. In this simulation, propane was used as fuel and the following 12 chemical species were used: C3H8, O2, H, O, H2, H2O, CO, CO2, CH3, CH4, C2H2, and C2H4. When using the heat source method, the fire heat release rate was defined as a volumetric heat source in control volumes at the fire location. Computational Grid Results from the tests in the small-scale facility and preliminary numerical simulations indicated that conditions in the room were symmetric, so the room was modeled by considering only one quarter of the room. Two different 4AT-01-11-2 (RP-899) computational grids were employed. When the flamelet model was used, the computational domain was divided into a grid of 21 31 21 control volumes. Additional grid points were embedded around the fire and the exhaust inlets to enable better resolution of the solution in these areas. The total number of grid points for these simulations was 24,014. When the volumetric heat source was used for the fire, the compu- tational domain was divided into a grid of 23 17 24 control volumes. The total number of grid points for these simulations was 9384. Boundary Conditions and Fluid Properties The following boundary conditions were used: Solid walls. The walls of the enclosure were modeled as solid, adiabatic, and hydrodynamically smooth bound- aries. This type of boundary is appropriate as the walls of the enclosure were insulated and had a smooth sur- face. Ceiling vents. At the ceiling vents, a constant mass flow rate was assigned based on the experimental data obtained at the quasi-steady state. Floor openings. At the floor openings, the total pressure was set to 101,325 Pa across the area of the openings. The inlet velocities were free to develop corresponding to the mass flow rate downstream. Inlet air temperature was assumed to be 291 K. Although in the experiments a fan was used to supply fresh air to the room through these openings, the supply air had a very low velocity, and the governing parameter was the flow rate of the exhaust fan, so the assumed boundary condition is acceptable. Radiative and Convective Heat Transfer Simulations were conducted with radiative and convec- tive heat transfer to the walls and ceiling. In these simulations, conduction through the wall materials was also considered. Simulations were also conducted without running the radiation model. For these simulations, it was assumed that 30% of the heat released by the fire was lost to the boundaries of the test facility. Similar results were obtained using the two types of simulations. However, there were considerable savings in computational time using the latter approach. IMPACT OF CEILING JET AND PLUGHOLING ON SMOKE VENTING It is generally assumed that two phenomena have an impact on smoke venting for atria in which there is minimum headroom between the design height and the exhaust inlets. These are the ceiling jet and plugholing. In this section, the two phenomena are described, and the results obtained with the small-scale physical model are used to investigate their impact. Plugholing The objective of a smoke management system, using a mechanical exhaust system, is to maintain the smoke layer above a design height, which is above the highest egress route or opening to a communicating space. However, the plughol- ing of the exhaust inlet by fresh air from below the smoke layer can decrease the efficiency of the smoke exhaust system and result in an increased smoke layer depth, which may expose building occupants to smoke (Figure 2). In the recently completed research project sponsored by ASHRAE, full-scale physical model studies were used, combined with CFD modeling, to investigate the plugholing issue (Lougheed and Hadjisophocleous 1997; Hadjisophocle- ous et al. 1999; Lougheed et al. 1999). This research indicated that, in order to minimize the impact of plugholing, multiple exhaust inlets should be used for the mechanical smoke exhaust system. This is demonstrated in Figures 3-5, which show the temperature profiles in the test facility for tests with heat release rates of 15, 25, and 50 kW, respectively. The numbers of inlet openings were 1, 4, 16, and 32. The temperature profiles were determined with steady fire and smoke exhaust conditions. These profiles show the location of the hot upper layer in the test facility. In previous publications (Lougheed and Hadjisophocleous 1997; Hadjiso- phocleous et al. 1999; Lougheed et al. 1999), a method was developed for estimating the height of the base of the smoke layer and the base of the transition zone. Specifically, these Figure 2Atrium smoke managementplugholing. AT-01-11-2 (RP-899)5 boundaries were assumed to be the heights at which the temperature rise was 80% and 20%, respectively. The temperature profiles shown in Figures 3-5 clearly show that by using 16 or 32 exhaust inlets, a much thinner smoke layer can be obtained. With the small test facility, there was a variation in the volumetric flow rate produced by the exhaust system due to restrictions in the duct system with one and four inlets. However, the impact of the number of openings on the smoke depth was also observed in the tests with the large-scale facil- ity, in which the number of openings did not affect the volu- metric flow rate in the exhaust system (Lougheed et al. 1999). Design Criteria Limiting the mass flow rate through the exhaust inlets can reduce the impact of plugholing. One approach for determin- ing the maximum flow rate through an exhaust inlet is one used in the United Kingdom to minimize plugholing in gravity venting systems (CIBSE 1995). Using mass flow rate, the design criteria is given by (1) where mmax =maximum mass flow rate of exhaust without plugholing, kg/s; Ts=absolute temperature of the smoke layer, K; To=absolute ambient temperature, K; d=depth of the smoke layer below the exhaust inlet, m; =exhaust location factor (dimensionless); C=3.13 (constant). Based on limited information, suggested values for are 2.0 for a ceiling exhaust inlet located near a wall, 2.0 for a wall exhaust inlet located near the ceiling, and 2.8 for a ceiling exhaust inlet far from any walls. This equation may not be practical for high spaces with high exhaust flow rates. For these cases, the smoke is highly diluted by the entrainment of air into the plume, resulting in low temperature rises. Other criteria based on life hazard assessments may be more suitable. Impact of Ceiling Jet The general flow conditions produced by the ceiling jet are shown in the CFD simulation shown in Figure 6. Once the plume hits the ceiling, the smoke flows under the ceiling until it hits the walls where it is initially redirected downward. Subsequently, there is a recirculation flow back toward the center of the test facility. The results shown in Figures 3-5 indicate that there was a significant reduction in the smoke layer depth by increasing the number of exhaust inlets to 16. However, a further increase to 32 had little or no impact even for tests with high exhaust flow rates. In this operating regime, it can be assumed that the Figure 3Temperature profiles for tests at 15 kW. Figure 4Temperature profiles for tests at 25 kW. Figure 5Temperature profiles for tests at 50 kW. mmaxCd5 2 TsTo Ts - - 1 2 To Ts - 1 2 = 6AT-01-11-2 (RP-899) ceiling jet rather than plugholing was the predominant factor determining the minimum attainable smoke layer depth. In Figure 7, the temperature profiles for a series of tests with 16 exhaust inlets are shown with heat release rates of 15, 25, and 50 kW. Using the 80% temperature rise criteria to determine the bottom of the smoke layer, the test results indi- cate that the minimum smoke layer depth was approximately 0.5