1、2008 ASHRAE 369ABSTRACTThere have been a number of concerns regarding thebalcony spill plume equation provided in North Americanstandards and codes. These include: lack of verification by full-scale experiments and application of the equation for highatria even though it was developed for low-height
2、 atria. As aresult of these concerns, the American Society of Heating,Refrigerating and Air-Conditioning Engineers (ASHRAE)initiated a project to evaluate the balcony spill plume equationused in North American codes and standards (RP-1247). Theprimary research conducted, as part of RP-1247, was a se
3、riesof full-scale experiments conducted to investigate smoke flowin balcony spill plumes and the resulting mechanical exhaustrequirements for an atrium. The research project includedCFD modeling studies to investigate smoke entrainment in thebalcony area and smoke entrainment in high atria. In thisp
4、aper, the results of a CFD model study, which investigated thevariation of plume mass flow rate with elevation in a simulatedatrium, are presented. Results from the CFD model study werecompared to existing engineering correlations. A new corre-lation was also developed for the variation in mass flow
5、 ratewith elevation for a balcony spill plume. This correlationextends the present correlations to high elevations at which noexperimental data is available.INTRODUCTIONUntil recently, atrium smoke management systems weredesigned assuming that the fire would occur on the atriumfloor. It was further
6、assumed that sprinklers would limit thefire size and thus the smoke produced by fires in any compart-ments attached to the atrium. However, some North Americanstandards and model codes call for consideration of smokeexiting from these attached compartments as balcony spillplumes (NFPA 2003; IBC 2003
7、).Atria with attached compartments are commonly found inoffice buildings and shopping malls. For a fire in an attachedcompartment that opens onto a balcony, smoke can flow out ofthe compartment, under the balcony and up through the atriumas a balcony spill plume (Figure 1). As this buoyant plumerise
8、s through the atrium, entrained air increases the mass flowrate of the plume. In order to design the atrium smoke manage-ment system to maintain a safe evacuation path, accuratecalculation of the mass flow rate as a function of elevation isrequired.The primary research conducted, as part of RP-1247,
9、 wasa series of full-scale experiments which investigated smokeflow in balcony spill plumes and the resulting mechanicalexhaust requirements for an atrium. Detailed results of theexperimental studies are provided by Lougheed et al. 2007,Lougheed et al. 2008a and Lougheed et al. 2008b.CFD modeling st
10、udies were undertaken to address twoconcerns with the experimental program: (1) the inability tofully investigate the effect of the parameters that affect airentrainment in the balcony area using full-scale experimentsand (2) the limited distance (5-7 m) between the balcony andthe ceiling. This dist
11、ance exceeds the distance required in theinitial request for proposals but does not span a representa-tive range of North American atria elevations. To addressthese issues, an effort was made to verify a CFD model usingNISTs Fire Dynamic Simulator (FDS) software (McGrattanet al 2002a; 2002b) for det
12、ermining smoke entrainment intoa spill plume. Detailed results of the modeling study, whichinvestigated smoke flow in the balcony area are provided byKo (2006) and summarized in Ko et al. (2008).CFD Investigation ofBalcony Spill Plumes in AtriaC.J. McCartney G.D. Lougheed, PhD E.J. Weckman, PhDMembe
13、r ASHRAEC.J. McCartney is a technical officer and G.D. Lougheed is a senior research officer in the Fire Research Program, National Research Council,Ottawa, Ontario, Canada. E.J. Weckman is a professor in the Faculty of Mechanical Engineering, University of Waterloo, Waterloo, Ontario,Canada.NY-08-0
14、42 (RP-1247)2008, American Society 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 wit
15、hout ASHRAEs prior written permission.370 ASHRAE TransactionsThere is a lack of full- or model-scale data for balconyspill plume mass flow rates at elevations greater than 9 mabove the balcony (Harrison 2004). The CFD modeling study,which investigated the variation of plume mass flow rate withelevat
16、ion in a simulated atrium, was used to extend the corre-lations to higher heights typical of modern North Americanatria. Previous research has demonstrated that CFD simula-tions using FDS can accurately model buoyant plumes(Hadjisophocleous and McCartney 2005; Ma and Quintiere2003). Results from the
17、 CFD model study were compared toexisting engineering correlations. A new correlation was alsodeveloped for the variation in mass flow rate with elevation fora balcony spill plume. This correlation extends the presentcorrelations to high elevations at which no experimental datais available.EXPERIMEN
18、TAL STUDIESThere have been several experimental studies of balconyspill plumes. Most of these were conducted at the BuildingResearch Establishment (BRE) in the UK using 1/10thscalemodels (Morgan and Marshall 1975; Morgan and Marshall1979; Hansell et al. 1993; Marshall and Harrison 1996).Experiments
19、were also conducted at the University of Canter-bury, New Zealand using salt-water modeling (Yii 1998) anda 1/10thscale model similar to that used at BRE (Harrison2004). Reviews of the physical model studies are provided byHarrison (2004) and Lougheed et al. (2007) and a briefsummary is provided by
20、Lougheed et al. (2008a).For RP-1247, a series of experiments were conductedusing a fire compartment that was approximately the full-scaleequivalent of the 1/10thscale physical model used in the earlierstudies. A detailed discussion of the results of the full-scaleexperiments are provided by Lougheed
21、 et al. (2007) andsummarized in Lougheed et al. (2008a, 2008b).BALCONY SPILL PLUME CALCULATION METHODSExisting balcony spill plume correlations are primarilybased on data from one-tenth scale physical models of shop-ping mall atria (Morgan and Marshall 1975; Morgan andMarshall 1979; Hansell et al. 1
22、993; Marshall and Harrison1996). The physical model studies conducted by Morgan andMarshall (1975; 1979) were used to develop the BRE spillplume method. Data from these experiments were also used todevelop a number of different algebraic correlations forbalcony spill plume mass flow rates (Law 1986;
23、 Thomas 1987;Law 1995; Poreh et al. 1998; Thomas et al. 1998). Furtherinformation on the various methods used to estimate the massflow rate in balcony spill plumes are provided by Morgan et al.(1999), Harrison (2004) and Lougheed et al. (2007).The basic assumption in the methods for estimating thema
24、ss entrainment into a spill plume, including those devel-oped by Law, is that it is analogous to a line plume. Based onthis assumption, all the simplified equations for air entrain-ment into the rising plume vary linearly with the height of theplume above the balcony edge:(1)where= mass flow rate at
25、 height zb(kg/s);zb= height above the balcony (m);A = linear coefficient for air entrainment in line plume (kg/sm);B = coefficient defining initial conditions for line plume at balcony edge (kg/s).The coefficients A and B are dependent on parameters such asfire size, opening width, balcony height an
26、d mass flow at theend of the balcony. Law (1986) developed relationships for the coefficients Aand B in terms of fire size, opening width and balcony height.The resulting equation was:(2)where= mass flow rate at height zb(kg/s);= convective heat output (kW);W = length of the spill (m);zb= height of
27、plume above the balcony edge (m);Ap= linear coefficient for spill plume (kg/skW1/3m5/3);Bp= virtual origin term (kg/s).Law (1986) used the results from the initial physicalmodel experiments conducted at BRE (Morgan and Marshall1975; Morgan and Marshall 1979) to determine a value for thelinear coeffi
28、cient and an estimate for the virtual origin basedon the height of the balcony. Law (1995) slightly modified theestimate for both the linear coefficient and the virtual origin toderive the following relationship for the mass flow rate in aFigure 1 Balcony spill plume schematic.mbAzbB+=mbmbApQcW2()1/
29、3zbBp+=mbQcASHRAE Transactions 371balcony spill plume using the experimental data provided byHansell, Morgan and Marshall (1993):(3)where= mass flow rate at height zb(kg/s);= convective heat output (kW);W = length of the spill (m);zb= height of plume above the balcony edge (m);H = height of the balc
30、ony (m).Equation (3) is in the form used in CIBSE (1995), whichwas derived from Law (1995). The principle difference is thatthe total heat release rate was used in the algebraic relationshipin Law (1995) as well as the earlier paper (Law 1986). Theversion provided in Equation (3) was developed assum
31、ing aradiative fraction of 0.35.The equation in NFPA 92B (2005) was derived from therelationship given in Law (1995). However, the NFPA 92Bequation is given in terms of the total heat release rate. Ifconverted to the same form as Equation (3) assuming a radi-ative fraction of 0.35, the linear coeffi
32、cient Ap= 0.41.Several other algebraic relations for the mass flow rate inbalcony spill plumes were developed. Morgan et al. (1999)noted four different methods for calculating smoke productionrates developed at BRE: BRE spill plume method, Method byThomas (1987), Method by Poreh et al. (1998) and Me
33、thod byThomas et al. (1998). All of these relationships have the samegeneral form as Equation (2). The algebraic relationshipsdeveloped at BRE assumed an infinite line plume in the atriumspace. In some cases, additional terms were developed toinclude entrainment into the end of the plume. Harrison (
34、2004) also developed equations for the airentrainment into a balcony spill plume. The equations have asimilar form to those developed by Poreh et al. (1998) andThomas et al. (1998). However, the linear coefficient value of0.2 is higher than that used in the earlier correlations (0.16).The linear coe
35、fficient developed in Harrison (2004) was deter-mined by a fit to a set of physical model experiments and thusincludes air entrainment into the end of the plume. The earliercorrelations were developed assuming an infinite line plumewith an additional term used in some cases to estimate endentrainmen
36、t.The correlations developed by Harrison (2004) werebased on physical model experiments with a downstand at theend of the balcony. The downstand depth was 0, 0.1 and 0.2 mfor the 1/10thscale model. The linear portion of the equationis comparable to correlations used for line plumes (CIBSE1995):(4)wh
37、ere= mass flow rate for line plume at height z (kg/s);= convective heat release rate (kW);Ll= length of longest side of rectangular source (m);z = height above the base of the fire (m).In summary, the simplified methods for balcony spillplumes all assume a linear equation for the air entrainmentwith
38、 height above the balcony Equation (1). However, twodifferent approaches were used to determine the linear coeffi-cient, Ap, and the virtual origin term, Bp. 1. Law (1985 and 1995) used a fit to experimental data todetermine Apand Bp.2. The spill plume methods developed at BRE assumed avertical infi
39、nite line plume in the atrium. Any additionalentrainment at the end of the balcony was included in thevirtual origin term.MODELING THE SPILL PLUME One concern with the experimental program was thelimited distance (5-7 m) between the balcony and the maxi-mum height of the smoke layer. This distance e
40、xceeds thedistance required in the initial request for proposals. However,in comparison to scenarios in many North American atria, thisdistance was a concern.To address this issue CFD modeling was undertakenusing Fire Dynamic Simulator (FDS) software (McGrattanet al. 2002a; 2002b) to investigate the
41、 air entrainment in the farfield. The CFD modeling was not required by the RFP and wasconducted in addition to the primary work undertaken for theproject.The modeling study investigated three areas: (1) thecompartment fire, (2) the full-scale experimental facility and(3) mass entrainment into the ba
42、lcony spill plume in the farfield. A detailed discussion of the CFD modeling is providedin McCartney (2007). In this paper, the modeling of the spillplume in the far field is discussed.HIGH-ELEVATION ATRIUM CFD MODELModel DescriptionA CFD model of a high-elevation atrium was developedto address limi
43、tations of the experimental facility. In particu-lar, the limited height did not allow for investigations of airentrainment at heights in the far field (20 m above thebalcony edge). Figures 2 and 3 show the geometry of the high-elevationatrium CFD model. A 12 m wide by 5 m deep by 5 m high firecompa
44、rtment similar to the experimental facility was locatedat floor level in a 50 m cubic atrium. The fire compartmentopening was varied between 5 and 10 m. A 5 m wide balconyprojected from the fire compartment. Variable height draftcurtains were located at the edges of the fire compartmentopening and r
45、an the full depth of the balcony. A removable1.5 m downstand (fascia) was located at the fire compartmentopening.mb0.36 QcW2()1/3zb0.25H+()=mbQcm10.21Qc1/3L12/3z=m1Qc372 ASHRAE TransactionsAll surfaces were specified as inert and all flow domainboundaries except for the floor were specified as open
46、bound-ary conditions. No smoke layer formed in the atrium and thebalcony spill plume was allowed to flow freely without inter-action with atrium walls or an atrium smoke layer. This is incontrast to the experimental facility where the presence of anatrium ceiling allowed an atrium smoke layer to for
47、m. Themass flow rates in the spill plume were calculated across hori-zontal sections of the flow domain over its entire height at0.5 m intervals.The fire was modeled as a steady release of propane at thecenter of the fire compartment at floor elevation. One isomet-ric computational mesh was defined
48、over the entire flowdomain. Simulation times of 300 s were chosen to achievesteady-state conditions. The fuel was specified as propane anda constant radiative fraction of 0.35 was used.Grid Sensitivity AnalysisA grid sensitivity analysis was performed on the high-elevation atrium model with two obje
49、ctives: to determine themost efficient grid size for a subsequent parametric study andto determine the atrium width and depth required to fullycontain the balcony spill plume. The former ensures accuracywith minimal computational requirements and the latterprevents mass loss from the sides of the flow domain fromaffecting the vertical mass flow rate profiles.For the grid sensitivity analysis simulations, a fire size of2000 kW and an opening width of 10 m were chosen as beingin the middle of their respective ranges. The draft curtain andfascia depths were presumed not to
