1、140 2008 ASHRAE ABSTRACTThis paper presents a study on the impact of variousparameters on fire plumes. The characteristics of these plumesplay a major role in the effectiveness of atrium smoke exhaustsystems. An increase of the air entrainment rates in fire plumescan render an atrium exhaust system
2、ineffective and endangerbuilding occupants. There are a number of parameters thatmay cause such an increase. In this study, a CFD model wasused to evaluate the impact of a number of parameters on thefire plume and the smoke layer interface. Parameters consid-ered include the wind velocity and its im
3、pact on flame tilt angle,position of make-up air opening, and area of the fire. Theresults show that these parameters affect plume characteristicsand the interface height. INTRODUCTIONAn atrium within a building is a large open space createdby an opening or series of openings in floor assemblies, th
4、usconnecting two or more floors of the building. The roof of theatrium is closed, and the sides may be opened to all floors, tosome of the floors or closed to all or some of the floors by fire-or non fire-resistant construction. There may be two or moreatria within a single building, all interconnec
5、ted at the groundfloor or on a number of floors.By interconnecting floor spaces, an atrium violates theconcept of floor-to-floor compartmentation, which is intendedto limit the spread of fire and smoke from the floor of fireorigin to other floors of the building. With a fire on the floorof an atrium
6、 or in any space open to it, smoke can fill the atriumand connected floor spaces. The smoke management approachdescribed in codes is based on maintaining the smoke layerinterface at a specified distance above the highest walkingsurface in an atrium. The associated smoke exhaust capacityrequired to p
7、rovide a large clear height could be substantial.The effectiveness of an atrium smoke management systemmay be affected by obstructions in the smoke plume (Hanselland Morgan 1) or the presence of a pre- existing stratificationlayer in the atrium (Hinckley 2). In the former case, smokemay be directed
8、to adjacent spaces or mixed with the air withinthe zone in which tenable conditions are required. In the lattercase, smoke produced by the fire may not reach the ceilingwhere it could be exhausted by a smoke management system.Also, in this case, smoke buildup could occur at a height atwhich it can m
9、igrate into the communicating spaces.Generally, smoke is recognized as the major killer in a firebecause it often migrates quickly to building locations remotefrom the fire space, exposing occupants to toxic gases, heatand thermal radiation, and reduced visibility 3. The reduc-tion in visibility is
10、a major threat in atrium fires as it affectsoccupants who are not located in the fire area and can causedisorientation and increase the time required for evacuation.The effectiveness of an atrium smoke managementsystem depends on the amount of smoke entering the upper-hot layer. Atrium smoke exhaust
11、 systems are designed toexhaust smoke at a specified mass flow rate to maintain theinterface at the required height. For the majority of atria inNorth America, the design flow rate is determined based oncalculations of entrainment rates in axisymmetric plumes. The buoyant axisymmetric plume is the m
12、ost commonlyused plume in fire safety engineering. An axis of symmetry isassumed to exist along the vertical centerline of the plume, andair is entrained horizontally from all directions along theplume height 4. The equation describing the rate of upwardmass flow of the plume is written in Eq (1). T
13、his equation,Parameters Affecting Fire PlumesJian Zhou George Hadjisophocleous, PhD, PEMember ASHRAEJian Zhou is a student and George Hadjisophocleous is a professor in the Department of Civil and Environmental Engineering at CarletonUniversity, Ottawa, Canada. NY-08-019 (RP-1300)2008, American Soci
14、ety of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in ASHRAE Transactions, Volume 114, Part 1. For personal use only. Additional reproduction, distribution, or transmission in either print or digital form is not permitted without ASHRAEs prior written perm
15、ission.ASHRAE Transactions 141which neglects the effects of the plume virtual origin, is thepreferred equation for atria because Z is much larger than thevirtual origin correction 3. (1)Where= Upward mass flow rate of plume at height Z, kg/s= Convective heat release rate, kW (0.60.7 )= Heat release
16、rate, kWZ = Height above fuel, mZ1= Mean flame height, m Effect of Wind on FlameThe impact of cross-flow on flames has been studied byseveral investigators. Figure 1 illustrates a general schematicof the problem, in which wind with a horizontal velocity ublows across the flames causing the flames to
17、 tilt by an angle. Using experimental data from wood crib fires, Thomas5 developed the following correlation for the flame tilt angle:(2)Where:u = Wind velocity, m/s= Mass flow rate per area, g/(s m2)D = Diameter of fire, m= Density of ambient air, g/m3 g = Acceleration due to gravity (9.8 m/s2)The
18、American Gas Association (AGA) 6 proposed thefollowing correlation to determine the flame tilt angle :(3)Where(4)u = Wind velocity, m/s= Mass flow rate per area, g/(s m2)D = Diameter of fire, m= Density of ambient air, g/m3 Both of these correlations are similar in the sense that theparameters invol
19、ved are the same: the diameter of the fire, thewind velocity, and the mass flow rate per unit area.Yi et al. 7, using a zone model studied the impact ofdifferent positions of make up air supply on the performanceof a mechanical exhaust system. Three scenarios with differ-ent relative positions for p
20、roviding make-up air duringmechanical exhaust were considered: smoke layer interface isabove, within and below the air inlet. The predictions by thezone model agreed well with the experimental findings. Theystate that when the position of the air supply is lower than thesmoke layer, a minimum smoke
21、layer interface height could bemaintained for a given fire size and extraction rate. When theair supply is above the smoke layer interface, make-up airwould enter the smoke layer directly and mix with the smoke.Smoke temperature would be reduced significantly and a safesteady height of smoke layer c
22、ould not be attained for this situ-ation. When the air inlet is at the interface height, the averagetemperature rise of the smoke layer would be lower than thecase with the air inlet located below the smoke layer.This paper focuses on identifying the main parametersthat affect plume characteristics,
23、 and investigating their effecton the plume. Parameters to be studied include: cross-flow,height of cross flow and fire area. A study of the effects ofmake-up air velocity on the plume and interface height ispresented in 8. DESCRIPTION OF THE MODELThe numerical simulations for this study were done u
24、singthe Computational Fluid Dynamics Model Fire DynamicsSimulator (FDS) (9, 10. FDS, which was developed by theNational Institute of Standards and Technology (NIST), isused extensively for fire applications. Although FDS includesboth Direct Numerical Simulation (DNS) and Large EddySimulation (LES) a
25、pproaches, LES is the technique used formost fire applications. Combustion is modeled using a mixturefraction model that assumes that all chemical reactions occurinfinitely fast so that fuel and oxygen never exist at one loca-tion at the same time. Boundary Conditions The following boundary conditio
26、ns were used in thesimulations:mp0.07Qc13Z530.0018Qcfor Z Z1+=mpQcQQZ10.166Qc25=()cos 0.7ugmD ()13-0.49=mcos 1 for U* 1=cos1U*-12for U* 1=U*=ugmD ()13-mFigure 1 Flame inclination due to wind.142 ASHRAE Transactions Solid walls: All solid walls were modeled as solid wallscovered with gypsum boards. F
27、loors: Floors were modeled as 200 mm thick concrete. Ceiling vent: For the cases with smoke exhaust, a con-stant mass flow rate was defined throughout the ceilingarea based on the mass flow rate required to maintainthe interface height at 0.8H, where H is the height of theatrium.Wall openings: The m
28、ake-up air opening on the wallwas assumed to be a passive opening.NUMERICAL SIMULATIONSThe purpose of the numerical simulations performed wasto determine the impact of various parameters on the fireplume. A number of parameters were investigated such as thecomputational grid size, impact of cross fl
29、ow, impact of open-ing location above ground, and area of the fire. These simula-tions are discussed in the following sections. Computational GridA number of preliminary simulations were performed inorder to determine the optimum size of the grid, which will yieldacceptable results. For this, an atr
30、ium with a square crosssectional area of 10 m x 10 m and a height of 10 m was consid-ered. The atrium had a 3 x 3 m opening in one wall, and theexhaust flow rate was set to 4.88 m3/s, which is the required flowrate to maintain the hot layer of a 1-MW fire at 8 m. The inletarea was 9 m2resulting in a
31、n inlet air velocity of 0.54 m/s. A firewith a heat release rate of 1 MW was located at the atriumcenterline and 2.5 m away from the opening. Three differentgrids were employed for the entire space of the atrium with sizesof 0.5 m, 0.25 m, and 0.125 m.Temperature, concentration of CO, and concentrat
32、ion ofCO2 profiles at the quarter points of the atrium and at thecenterline of the fire are compared at various heights.Figure 2 shows the temperature profiles with height at thecenterline of the plume (at X=7.5, Y=5.0). The temperaturepredicted on the three different grids vary significantly atlowe
33、r heights. The maximum temperature on the fine grid isabout 3 times higher than that on the coarse grid. This is dueto the mixture fraction combustion model that was used forsimulating the fire. In the mixture fraction model it is assumedthat the reaction takes place on an infinitely thin flame shee
34、twhere both the fuel and oxygen concentrations go to zero. So,a fine grid captures the flame sheet better than a coarse gridand provides a more accurate flame temperature. The temper-ature difference between the results of the different gridsdecreases with height. Temperatures at heights over 7.5 m
35、arevery close.Figure 3 shows the CO2concentration profiles withheight at the centerline of the plume. As with the temperatureprofiles, the profiles show that there is a large difference in theconcentration determined using the three different grids atheights below 6 m. Above this height, the concent
36、rations arevery close to each other. Figures 4 and 5 show the temperatureand CO2concentration profiles for the three different grids atthe quarter points of the atrium. The results indicate that themedium and fine grid sizes produce very similar profiles.The results of these runs show that FDS is se
37、nsitive to gridsize, especially in the region near the fire. The profiles showthat a 0.5 m grid produced results that are very different fromthose of the fine grids. The results produced by the 0.25 m gridand the 0.125 m grid do not differ significantly especiallyoutside the fire area. From these re
38、sults it is clear that if theconditions in the fire plume need to be computed accuratelythen it is necessary to use a grid with a maximum size of 0.125m. A 0.25-m grid produces acceptable results in areas awayfrom the combustion area.Impact of Wind on FlameTo better understand the interaction betwee
39、n a cross flowand a fire, this section describes a number of simulations doneusing FDS to investigate the effect of a cross flow on theflames. For these runs, a compartment 30 meters in length, 10meters in width and 6 meters in height was used. The wall andceiling of the compartment were made with g
40、ypsum board andthe floor was 200 mm thick concrete. The 10 m (W) x 6 m (H)sides of the compartment were open. At one of the openings,a constant velocity flow was specified, while the other sidewas defined as a passive opening. The fire was placed on theFigure 2 Temperature profiles with height at th
41、ecenterline of the plume.Figure 3 CO2profiles with height at the centerline of theplume.ASHRAE Transactions 143ground, 7.5 m from the left opening and along the centerlineof the compartment. The velocity of entry air was given thefollowing values: 0.5 m/s, 1.0 m/s, 1.5 m/s, and 2 m/s. The firesize w
42、as: 0.5 MW, 1 MW, and 5 MW. The mesh used had 120cells in the X direction, 40 cells in the Y direction and 24 cellsin the Z direction giving a cell size of 0.23 x 0.25 x 0.25 m.To evaluate the impact of the cross flow on the flames, theinclination angle of the flame in the Y direction was estimatedf
43、or each simulation. The results obtained from FDS are shownin Table 1. They are compared with calculations employingthe Thomas 5, Equation 2,and American Gas Association(AGA) methods 6, Equation 3. Figure 6 shows a comparison of the tilt angle obtainedfrom FDS for the fire sizes and cross flow veloc
44、ities consid-ered and the results obtained using the correlations of Thomas5 and AGA 6. These values are also shown in Table 1. Ingeneral the results show that an increase of the cross flowvelocity results in an increase of the flame inclination and thatas the fire size increases the tilt angle decr
45、eases. The figuresdemonstrate that FDS and the Thomas correlation for the 0.5-MW and 1.0-MW fires the flames tilt even with a 0.5 m/svelocity. For all fire sizes, the AGA method produces no tiltwith when the velocity is 0.5 m/s. The FDS results are in closeragreement with the results of the Thomas c
46、orrelation. Impact of Opening LocationA number of simulations were conducted to determine theimpact of the height of the opening used to supply make-up airon the fire plume. For these simulations, an atrium was consid-ered with dimensions of 10 meters in length, 10 meters inwidth and 10 meters in he
47、ight. A fire of 1 MW was located 2.5meters from the air supply opening. An opening 5-m wide and3.33-m high was placed on one wall of the atrium at threedifferent heights supplying the specified make-up air. The make-up air velocity for all simulations was set to 1m/s. The exhaust velocity of 0.195 m
48、/s was specified through-out the atrium ceiling. This velocity corresponds to an exhaustflow rate of 20.85 kg/s, which was determined using equation1 with a heat release rate of 1 MW and an interface height of8 m.Three simulations were performed using FDS. Figure 7shows the plume shape for the diffe
49、rent opening locations.From Figure 7 (a), it can be seen that the flame tilt angle isabout 10 when the opening is at the bottom. The incoming airaffects only the base of the flame and causes a small flame tilthowever it does not affect the smoke plume much. Figure 7 (b)shows that when the opening is at 3.33 m from the floor, theincoming air disrupts the fire plume causing mixing with thesurrounding air, bringing smoke to the low