Fin Weicheng, Vln ZhenghulHuo Ran, Chen Yuming, Zhao Hongbo(State Key Lab of Fire Scienee , USTC, Anhui, China, 230026)
ABSTRACT
A new model (F-Z model) is proposed for compartment fire model ing, which is aeombination of field model and zone model being currently available. The mainidea of the F-Z model is to use field model for rooms (in which the two-layerassumption of zone model is inadequate.) of fire origin, with strongventilation or the rooms in which a detail simul.tion is required, and to usezone model for other spaces in the compartment. The general structure and theway to deal with interface between field and zone models within the frameworkof F-Z model are described and demonstrated by performing two- and three
-dimens ional F-Z modeling for a two-room compartment fire.Also, the case iselperimentally studied. Predictions by the F- Z model are in qualitativeagreement with 8lperimental data, which shows that the F-Z model is promisingin compartment fire model ing.
INTRODUCT ION
Computer modeling has drawn much attention from engineers and scientists inf ire research(1l. There are maiDly two kinds of models available at presentfor compartment fires . They are zone model and field model. The zone model isactive, being widely developed, which is based on two-layer assumption (al ina compartment fire. The zone model produces a fairly reasonable simulationin many eases(lIl . However, the two-layer phenomenon does not generally elistin some special eases, such as in a room of f ire or igin or wi th strongnntilation. Also, eurrent zone model is limited in its simulation of fluidflow, heat transfer and flame spread, whieh are of importanee to manyapplications (deteetion, sprinkler aetivation, eomplel geometries, et e.}. Thefield model(4l ean in prineiple give the distributions of gas veloeity,temperature, eoneentration of various species and their variation with timein a eompartment fire by solving a set of partial differential equations.However, the field model needs a large amount of storage and run time of aeomputer. It is still impossible for applying the field model to a high risebuilding in the foreseeable future.It is advisable to develop a new model, which retain the ad,antage of bothfield and zone models. The new model presented here is a combination of fieldand zone models, and so called as the Field-Zone model (F-Z model in short) .The main idea of the F-Z model is to use field for rooms of fire origin, with
continuity, momentum,solved together with thebe cast into a general form
~.
strong ventilation or rooms in which a detail simulation is needed, and touse zone mGdel for the others connected with the rooms mentioned above. Bythe F-Z model compartment fire will be able to be predicted well under thecapab ility of present computer. The main difficulty in construction of theF-Z model is to explore a way to treat an interface, of which on both sidesthe field model and zone model are performed respectively.Within a framework of the F-Z model either one- , two- or three- dimensionalfield modeling can be adopted for predictions of fire behavior in thosespecially-interested rooms. Accordingly, they can be called respee t i te lr asone-, two- or three-dimensional (I- , 2- or 3-D in short) F-Z model. In thepresent paper two cases have been studied in different ways. First. both 2- DF-Z mode l and Zone model are carried out for prediction of smoke movement ina two-room compartment, and the predictions obtained by the two models arecompared. Secondly, the 3-D F-Z model is used in a small-scale system, whichis experimentally studied as well. Comparisons are made of predictions andexperime ntal data.
F-Z MODEL AND EXPERIMENT
The F-Z model consists of three parts. They are field model, zone model, andthe inter face treatment .
1. FIELD MODEL
Fire is a complex phenomenon. It consists of fluid flow, heat and masstransfer, chemical reac t ions and the ir interac t ions. Most of real - fire areturbulent processes . Models are usually needed to describe the processes ~: ~ fturbulence transport , turbulent combustion, heat radiation, vaporization --ofliquid fuel or pyrolysis of solid fuel , soot formation and consumption, etc.Mod~ls are various with different generality, reliability, flexibil }ty - andeconomy. Therefore comprom ise must be made according to the requiremen tl instudy, when people choose a model .In the present study the governing eq uations ofenergy and chemical species are constructed andbuoyancy-modified k- E turbulence model. which cangiven on Table 1. [Ill [8l
2. ZONE MODEL
Various zone models are ava ilable in literature. A moderate- level zonemodel £7] is adapted for the present study.
3, INTERFACE TREATMENT
Boundary conditions must be given in the interface, which ~. are needed forsolving the governing equations in field modeling, Zero gradient condition isadaptable for all dependent variables except for the velocity component
-3 01-
perpendicular to the interface, which is determined by an assumption of massconservation in the grids adjacent to the interface. Variation of pressurewith height of the interface should be taken into account, which is given bya formula,
NPN"PO-~ P 19h 1
,r i" 1
where the subscript 0 denotes a value on the ground, and N is index of a gridbeing referred to.To perform zone modeling in the connected room, the flow rates of mass andenergy cross the interface and their variation with time are required,which are obtainable from appropriate summation of the predictions ofvelocity, temperature, concentration of-. species and density in the interfaceby field model ing. Taking calculation of energy flow rate, E, as an example,
where i refers to the index of a grid in the interface; N the total number ofgrid in the interface; Cpl,TI' P"UI' and a, stands for specific heat,temperature, density, velocity of the gael and the area of the grid boundaryrespect ively.
TABLE I .PARTIAL DIFFERENTIAL EQUATIONS
EQUATIONS. l%l r. i S " ~ .: :,~~.J ~[.O< . )•CONTINUITY 1 0 0 I.' :, ." ;j ii '.;'('.~ 'l'. " ~ . " '.
" , ,
ap + a ( Ou) + s. ( ov), . .•
+ 0.( aw)u u P .11
- ax ox IJ'/lox . iiy p.nax az . P~lIax
+PR
v v p.1I e» + a ( au) + a ( . av) a ( ow)- oy ax lJ.nay oy 1J.IIOy + az P'"oy
',<,
s» +'a ( OU)+ a ( aV)+ 0(' OW)w w p./I I .- az ax IJ./laz ay P .naz az p.na;,, ..\ I
Ie IC r. G,+G-p't
c r 1:/ K[C \,(G + G, ) (1 + C ,R,)- O~pl:l ';j; ,
\- I: •. ,
Hi
H r A ., . SA to: ~ ~ / -ri . " "'p " :. ,\;'
-3020-
..:
I
tendency ofpredictionsFig. 7. The
4. EXPERIMENT
The exper imental set - up is sketched in Fig 1. A electric heat sourcewithcont ro lled power is cent ra lly located in room A. 24 thermocouples areplaced in room A and B , and li nked with a 24 -CH THERMALS AID and a comp ute rfor tempera ture measurement and data processing.
THERHOCOUPLF
, '. .";j
• 2 t~ 6 ' '~IU-CH THERH,4LS ,;U
II C v n, ' . .. 1<1 J J
I R OOH A ' L :R OOH B tlI ,EL ECT R ONI C BAL4N~~1 'H5"~ 10' 15 6 0 441506200660 ,~ WEIGHTING SYST E' I ~' OPRN DOO R- ~
i HEAT ISORCE , ~aO o 1 8 0 SL OT
I AnoOo250180 __
L _ ______._ ':" __ ' _ I
Fig 1. SKETCH MAP OF TIlE EXPERIMENT
RESULTS AND DISCUSS IONS
Two cases have been studied with the F-Z model .The fi rst is to predicted the smoke movement in a two-room compartment by the2-D F-Z model and the zone model respectively. The geometry and the size ofthe comp ar tment are shown in Fig. 2 The predictions of temperature
distri but ions in three sections of the fire-origin room are given in Fig. 3. Atypica l saddle- shape cure shown in Fig.3 is due to entrainment of surround ingair by the fir e plume. Fig. 4 shows the hot- layer de ve lopmen t in the roomnelt t o the room of fi re origin by the F-Z model and zone model res pect ivelyThey give qu al i t at ively-agreed results . The second st udy is to compare theresuh s of hot -gas movement in a small-scale compartment (see Fig. i) I
which are obtai ned by 3-D F-Z mode ling and elperiment . The predicti on ofveloci t y dis t ri but ion in a typi cal section is given in Fig,5 ,which shows theflow f ie1di n de ta i 1.Fig. 6 gives the res ults in roomA. It can be seen that thepredic t ions and experime nt i s s iminar, although systematic overare shown. Temperature dis tr ibutions in the room B are given inlayeri ng phenomenon can be seen from the elperiment data.It cln be seen from the experiments (Fig. 7) that gas layer ing phenomeuon isnot cl ear in a room with a heat source, whereas the two-layer assumption isclose to elpe r iment in the conn ected room.Predic t ions by F-Z mode l are in qual itat ive agreement with experiment .Howeve r, the temperature is over predicted, which may be ..ttr ibuted t o theassump t ions of walls be ing adiabatic, and t o ne glect of therma l capac i ty ofwalls and the isolat or of'the el ectric al heat er.
-303-
Here we may draw the following conclusions.Two-dimensioned F-Z modeling and zone modeling have been performed forprediction of the smoke movement in a two-room compartment. The F-Z model cangive the flow field, heat and mass transfer in detail, but needs largercomputer capacity(storage and run time) .Three-dimensional F-Z modeling as well as experimental measurement have beenmade to study the hot gas movement in a small- scale two- room system,Qualitative agreement has been found in comparison of predictions withexperiment data, although some discrepancies exist between them.The treatment of interface between field and zone modeling within theframework of the F-Z model is adequate in the present study, but needs to befur ther veri fie d.The F-Z model combines comprehensibility and simplicity, and is full ofpromise in compartment fire modeling.
23.0lG.O1II1EI.1 '.0FIG 4. VARIATION OF TEMP WITH 'TIME
ol66 .
354.
- 304-
L
0.6 0.8( M)
0.2 0.4HEIGHT
288.0 -+-o==j==!c:::::::=jO....'-r--.,0.0
290.0
292.0
296.0
300.0.,...----------,
1"\ 298.0:..:
1294
•
0
0.60.4(1)
--:l==:l::::::1t1""" J
-=~~~~=t=~
0.2HEIGHT
460.0.,...---------,
440.0c 420.0
400.0
~< 300.036e.0
~1- 340.0~ 320.0
300.0280.0 -1----r-----r----I
0.0
FIG 6. CURVE OF TEMP-HEIGHT IN ROOM A FlO . 7 . CURVE OF TEMP-HEIGHT IN ROOM B
REFERENCE
I. Journal:R .Friedman:Survey of Computer Models for Fire and Smoke, FactoryMutual Research Corp.,Norwood,Mass. , 19902. Journal :Shen Hao and"Fan Weicheng:Available Safe Egress Time In EnclosureFires, Proceeding of National Symposium on Combustion, 1987, Academic JournalPress , Be ij ing3. Journal:W. W; Jones and R.D.Peacok:Refinement and Experimental verificationof a Model for Fire Growth and Smoke Transport, proceedings of the secondinternational Symposium on Fire Safety Sc(ence,Hemisphere Publishing, New York
, N. Y. , 1989, pp. 897-906~. Journal :R.S. Levine and H. E. Neson :Full-Scale Simulation of a Fatal fire andComparison of Results with Two Mutiroom Models, NISTIR 90- 4268, NationalIns t ihie of Standards and Technology, Gai therburg, Md., 1990.6. Journal :L. Y. Cooper: A Buoyant Source in the Lower of Two Homogeneous, StablyStratified Layers, 20th Int. Sym. Com. 1984 PP.1667-16736. Journal :W. WJones:A Muticompartment Model for the Spread of Fire, Smoke andToxic Gases, Fire Safety Journal, 9 (1986) PP.66-797. Journal : Launder, B.E and Spalding, D. B: Mathematical Models of Turbulence
, Academic, New York, 19728. Journal : Launder , B. E and Spalding,D. B:The numerical Computation of TurbulentFlo~Comp.Methds Appl.Mech.En~ ,Vol . 3.p . 269, 1974
