ASHRAE LV-11-010-2011 Algorithm for Smoke Modeling in Large Multicompartmented Buildings-Development of a Hybrid Model.pdf

1、2011 ASHRAE 769This paper is based on findings resulting from ASHRAE Research Project RP-1328.ABSTRACTThis paper describes a newly developed hybrid model toenable the simulation of smoke and heat movement in multi-compartment buildings due to a fire. The hybrid modelcomprised two integrated models:

2、a zone and a network model.The two-zone model was developed to simulate fire and smokemovement inside the room of fire-origin and neighboringcompartments. The network model capable of predicting bothmass and energy flow is used to simulate smoke movement intothe rest of the compartments that are far

3、 away from the fire-origin room. The two models were combined to produce ahybrid model that allows an accurate simulation of fire dynam-ics in both the near- and far-field. The paper illustrates the different steps of the developmentof the model. In a subsequent paper, the implementation of thedevel

4、oped hybrid model for different buildings will bediscussed. In order to understand mass and heat transferbetween compartments, two types of vent flows are discussedhere: horizontal flow through a vertical vent, and vertical flowthrough a horizontal ceiling vent. Equations for both types ofvent flows

5、 are presented in the paper.INTRODUCTIONSmoke generated by fires in buildings imposes a greatthreat for occupants. Over the years, many smoke and firemodels have been developed in order to predict smoke andheat movement inside a building during a fire. Examples ofthese models are zone models and net

6、work models. Zonemodels assume that a compartment is divided into twolayers: an upper hot layer and a lower cold layer. Tempera-ture and gas properties are considered to be uniform withineach layer. Zone models provide a reasonably good evalua-tion of the fire dynamics. However, these models often e

7、xpe-rience convergence failures and are computationallyexpensive, especially when dealing with large multicompart-ment buildings. Network models are based on the assump-tion that gas properties are uniform throughout thecompartment (complete mixing). They are computationallymuch simpler than zone mo

8、dels. However, network modelsdo not provide an accurate simulation for the room with thefire and rooms near the fire, as mixing in these rooms cannotbe assumed to be complete. Considering the limitations and strengths of zone andnetwork models, the objective is to develop a hybrid modelthat combines

9、 the accuracy of a zone model near the fire andthe efficiency of a network model sufficiently far away. Differ-ent from other existing fire models, mass and energy transfersfor both the two-zone and network model are considered. Thefundamentals of the hybrid model are presented and discussedin this

10、paper. The coupling between the two-zone model andthe network model is a very important factor in the develop-ment of this hybrid model. Hence, a detailed discussion on thetreatment of the interface is also presented here.LITERATURE REVIEW ON FIRE MODELS Zone models (Chow 1996) were originally based

11、 onexperimentally observed phenomena (Cooper et al. 1982).When there is a fire in a compartment, two layers of gas formin the fire and neighboring compartments due to thermal strat-ification, namely an upper hot smoke layer and a lower cool airAlgorithm for Smoke Modeling in Large, Multicompartmente

12、d BuildingsDevelopment of a Hybrid ModelA. Kashef, PhD, PEng G. Hadjisophocleous, PhD, PEngMember ASHRAE Member ASHRAEX. Zhu D.E. Amundsen, PhDA. Kashef is a senior research officer at the Institute for Research in Construction, National Research Council Canada, Ottawa, Ontario.G. Hadjisophocleous i

13、s a professor in the Department of Civil and Environmental Engineering, X. Zhu is a masters student in the School ofMathematics and Statistics, and D.E. Amundsen is an assistant professor in the School of Mathematics and Statistics at Carleton University,Ottawa, Ontario, Canada.LV-11-010 (RP-1328)20

14、11. American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in ASHRAE Transactions, Volume 117, Part 1. For personal use only. Additional reproduction, distribution, or transmission in either print or digital form is not permitted without ASHRAES p

15、rior written permission.770 ASHRAE Transactionslayer. The zone model assumes that the gas temperatures, aswell as other properties such as smoke and pollutant concen-trations, are uniform within each layer. Smoke propagatesfrom the fire source compartment to its neighboring compart-ments through the

16、 upper stratified hot smoke layer. The energyand mass transfer between the two zones and the surroundingrooms are calculated by solving the mass and energy conser-vation equations for each zone. In the fire source compart-ment, the heat released by the fire causes a strong vertical flow(plume), whic

17、h plays a very important role in smoke propa-gation. Over the years, many zone models have been developedand studied. These models, in general, are designed to simu-late fire and smoke propagation in buildings with a smallnumber of compartments. Among these models, the CFAST(Jones et al. 2005) (an u

18、pgrade of the FAST program) is amulti-room two-zone model that predicts fire and smokemovement within a structure resulting from a user-specifiedfire. The program includes mechanical ventilation, a ceilingjet algorithm, capability of multiple fires (up to 30), heat trans-fer to targets, detection an

19、d suppression systems, and a flamespread model. After being developed and used for many years,it is considered as a benchmark for two-zone modeling.Network models, on the other hand, assume the propertiesof the gas are uniform throughout the compartment. Due tothis assumption, it is valid only for s

20、paces sufficiently remotefrom the fire-origin room where mixing is relatively complete.Computationally, the network model is much simpler than thezone model. This simplicity makes it practical and efficient forthe prediction of smoke spreading in high-rise building.The most advanced and sophisticate

21、d network model isCONTAM (Walton 1997; Dols et al. 2000), which can simu-late flow inside buildings consisting of thousands of compart-ments, including stack effect, wind effect, and forcedventilation. CONTAM was originally developed for indoor airquality applications, but it has become the most ext

22、ensivelyused network model for smoke control analysis.MOTIVATION AND DEVELOPMENT OF THE HYBRID MODELAs already discussed, the two-zone model fire models cansimulate smoke flow and temperatures resulting from buildingfires, but these simulations are limited to relatively fewcompartments. It has been

23、observed that these models oftenexperience convergence failures with simulations over as fewas six compartments (Fu et al. 2002). For compartments faraway from the fire origin, where two-zone models are not effi-cient to predict the conditions inside the compartments, thenetwork model approach is a

24、better candidate.There has been no comprehensive model so far for thesimulation of smoke transport in large, multicompartmentedbuildings, except for a recent approach developed by Floyd etal. (2005). For simulation of smoke in large buildings, anapproximate method has been used that combines two sep

25、a-rate models: (1) a zone fire model (such as FAST Jones 1985)for simulating the near-filed and (2) a network model (such asCONTAM Walton 1997; Dols et al. 2000) for simulatingsmoke flow in the far-field. For example, the zone fire model,FAST, could be used to model the room of fire origin and anumb

26、er of nearby rooms. This simulation provides tempera-ture data that can be used as input data for a network simula-tion of the entire building. The simulations for thisapproximate method are recognized as crude at best, and inparticular only the mass flow is determined over the entirebuilding, the e

27、nergy transfer is limited to the domain of theFAST simulation.The aforementioned analysis indicates that it is necessaryto develop a new model, which can give reliable prediction ofsmoke and heat movements from building fires, especially forhigh-rise complex buildings, but stay within practicalcompu

28、ter capability. A novel hybrid fire model to simulate thefire and smoke propagation in a multi-story building ispresented in this paper. This model is based on the coupling ofa two-zone and a network model and considers both mass andenergy transfer throughout the whole building. In the hybridmodel,

29、the two-zone model is used to model the fire smokemovement in the compartment of fire origin, as well as the firesmoke propagation in the compartments/corridors near the firecompartment where the hot smoke layer is well stratified andthe smoke movement can be reasonably simulated based onthe two-zon

30、e concept. The output from the two-zone model istransferred to the network model as input/sources, which isthen employed for the simulation of smoke movement in thecompartments far from the fire-origin and where mixing isassumed relatively complete. Unlike other existing networkmodels, both mass and

31、 energy transfer are considered in thenetwork model. The application of this model permits areasonable numerical simulation (time and accuracy-wise) ofthe fire process in an entire high-rise building using a standardpersonal computer.The following sections discuss the development of thetwo-zone and

32、network model individually, and the present thegoverning equations for both models.DEVELOPMENT OF THE TWO-ZONE MODELWhen the size of the fire is small compared to the size ofthe compartment in which it develops, it has been observedthat in such a case there was an accumulation of combustionproducts

33、in a layer beneath the ceiling (upper layer), with ahorizontal interface between this upper hot layer and the lowerlayer where the temperature of the gases remained muchcooler. The main hypothesis of the two-zone model is that thetemperature, gas density, internal energy, and pressure areuniform in

34、each layer, but different from one layer to another.Assumptions need to be made in order to use the govern-ing equations that are the base of the two-zone models. Belowis a list of the major assumptions of two-zone models:2011 ASHRAE 771Layers are assumed to be uniform throughout, which isvalid prov

35、ided that the relative difference in temperaturebetween the two layers is large.The fire plume acts as a pump of mass and heat to theupper zone. However, the plume volume is assumed tobe small compared to the upper and lower zones and soit is neglected.The compartment gases in the upper and lower la

36、yersare treated as an ideal gas with a constant molecularweight.The pressure in the compartment is considered uniformin the energy equation.The modeling equations used in the two-zone model takethe mathematical form of an initial value problem for a systemof ordinary differential equations. These eq

37、uations are basedon the conservation of mass, the conservation of energy(equivalently the first law of thermodynamics), and the idealgas law. These equations predict, as functions of time, quan-tities such as pressure, layer height, and temperatures in thetwo layers.Many formulations based upon thes

38、e assumptions can bederived. One formulation can be converted into another usingthe definitions of density, internal energy, and the ideal gaslaw. Though equivalent analytically, these formulations differin their numerical properties. Each formulation can beexpressed in terms of mass and enthalpy fl

39、ow. These ratesrepresent the exchange of mass and enthalpy between zonesdue to physical phenomena such as plumes, natural and forcedventilation, and convective and radiative heat transfer (Jones etal. 2005).A compartment is divided into two control volumes, arelatively hot upper layer and a relative

40、ly cool lower layer. Thegas in each layer has attributes of mass, internal energy,density, temperature, and volume denoted respectively by mi,Ei, i, Ti, Vi, and where i = L for the lower layer and i = U forthe upper layer. The compartment as a whole has the attributeof pressure P. These 11 variables

41、 are related by means of thefollowing seven constraints (density, internal energy, and theideal gas law twice, once for each layer) (Fu et al. 2002): (density) (1)(internal energy) (2)(ideal gas law) (3)(total volume) (4)The specific heat at constant volume and at constant pres-sure Cvand Cp, the un

42、iversal gas constant, R, and the ratio ofspecific heats, , are related by and.Four additional equations obtained from conservation formass and energy for each layer are required to complete theequation set. These equations for each of the 11 variables aresummarized in Table 1. The time evolution of

43、these solutionvariables can be computed by solving the correspondingdifferential equations together with appropriate initial condi-tions.Four basic variables, the pressure of the compartment P,the volume of the upper layer VU, the upper and lower layerstemperatures, TUand TL, have been chosen to des

44、cribe thecondition inside the compartment with the correspondingordinary differential equations (ODEs) as follows: (5)(6)(7)(8)In these equations, the pressure is taken as the pressuredifference relative to an ambient reference pressure to mini-mize numerical instability (Jones et. al. 2005).COMBUST

45、ION MODELThe heat release rate in unconstrained combustion canbe obtained by(9)where (H)Fis the effective heat of combustion per unitkilogram fuel in open air, and is the mass pyrolysisrate. In this model, combustion chemistry is considered asfollows:(10)where mPFis the mass of fuel pyrolysis, compo

46、sed of twoparts, mPFminand mTUF. The variable mPFminis assumed to besome harmful species, such as HC1 and HCN, whose mass ismuch less than mPFand will not further be involved in thecombustion process. Thus, mPFminfrom mPFon the left side ofthe above equation directly goes into its right side. In thi

47、simivi-=EiCvmiTi=P iRTi=VVLVU+=CpCv-=RCpCv=dPdt- 1V- hLhU+()=dVUdt-1P- 1()hUVUdPdt-=dTUdt-1CpUVU- hUCpmUTU()VUdPdt-=dTLdt-1CpLVL- hLCpmLTL()VLdPdt-=QNmPFH()F=mPFmPFmPFminmCHO+()mO2+mCRPmCO2mCOmSoot+()mH2OmPFminmTUF+ +772 ASHRAE Transactionsmodel, mPFmincan be composed of as many as three kinds oftox

48、ic species: mtox1, mtox2, and mtox3. The variable mCHOis themass of the fuel pyrolysis excluding mPFmin, and is assumed tobe composed of carbon, hydrogen, and oxygen. The massproduction of CO2, CO, and soot are combined into one newterm mCRP, carbon-related products, and soot is assumed to becarbon

49、only. The variable mTUFis the mass production of totalunburned fuel, which is assumed to have the same elementcomposition as mCHO.PLUME ENTRAINMENTFire-induced buoyant plume entrainment is a very impor-tant factor in modeling fire growth and smoke spread in abuilding. A number of formulas can be found in the literature(Heskestad 1995; McCaffrey 1983). A widely acceptedconsensus on entrainment models for large fires in compart-ments does not