Workshop on Modeling of Under-Ventilated Compartment Fires

Transcription

1 Workshop on Modeling of Under-Ventilated Compartment Fires - August 30, André Marshall, James G. Quintiere, Arnaud Trouvé Department of Fire Protection Engineering University of Maryland, College Park, MD (USA) Slide 1

2 Origin Small workshop inviting current and potential sponsors Overwhelming response prompted us to expand our vision Later invitations were extended to additional fire research laboratories, consulting firms, universities, and government agencies Slide 2

3 Outline Dynamics of poorly-ventilated compartment fires Main features (flame location, smoke layer composition, air vitiation) Computer-based modeling of compartment fires Main features of the zone and CFD modeling approaches Specific challenges associated with under-ventilated fires Presentation of UMD research program: Main component corresponding to a NIST-supported program (CFD modeling, FDS) Second component corresponding to an in-house/edf-supported program (zone modeling, MAGIC and BRI) Presentation of zone/cfd physical sub-models proposed to describe air vitiation and flame extinction effects Program of the workshop Slide 3

4 Dynamics of Poorly Ventilated Compartment Fires Flame location: (1) near the fuel source; (2) near the vents Smoke Flame Smoke Flame Fuel Air Fuel Air Vent (1) Over-ventilated combustion, i.e. small values of the Global Equivalence Ratio: GER 1 Vent (2) Under-ventilated combustion, i.e. large values of the Global Equivalence Ratio: GER 1 Slide 4

5 Dynamics of Poorly Ventilated Compartment Fires Smoke layer composition: (1) products of complete combustion mixed with air; (2) products of incomplete combustion Flame Smoke Products Excess air Smoke Flame Products Excess fuel (CO, soot) Fuel Air Fuel Air Vent (1) Over-ventilated combustion, i.e. small values of the Global Equivalence Ratio: GER 1 Vent (2) Under-ventilated combustion, i.e. large values of the Global Equivalence Ratio: GER 1 Slide 5

6 Dynamics of Poorly Ventilated Compartment Fires Air vitiation: a concern as GER increases to near-unity values Flame Flame Fuel Smoke Products Excess air Air Y, Fuel T F F Oxidizer Y, T O2 O 2 Y T Y O2 O2, a O Ta 2 Vent Local effects: Reduced burning intensity Flame extinction events Global effects: Oscillatory combustion Total flame extinction Incomplete combustion products Slide 6

7 Slide 7 Computer-based Modeling of Compartment Fires

8 Computer-based Modeling of Compartment Fires Objectives: Response of compartment system to prescribed fire loading (response characterized in terms of: temperature, smoke transport, CO emissions, heating of the building structure, activation/efficiency of fire protection/suppression systems, etc) Response of compartment system to initial fire source (response includes the simulation of freely-evolving fire loading, i.e. initial fire growth, fire spread, transition to flashover, possible flame extinction, etc) Two approaches: Zone modeling Computational Fluid Dynamics (CFD)/field modeling Slide 8

9 Computer-based Modeling of Compartment Fires Zone modeling approach: Decomposition of the fire compartment into 2 control volumes corresponding to the ceiling and floor layers Wall Boundaries (ceiling, side walls, floor) Upper Layer zone (mass, energy conservation statements) z D Lower Layer Flame Zone & Fire Plume Vents Upper layer: Flame zone Fire plume Ceiling smoke layer Lower layer Floor air layer Slide 9

10 Computer-based Modeling of Compartment Fires CFD modeling approach: High-resolution decomposition of the fire compartment into several hundred thousands control volumes Wall Boundaries (ceiling, side walls, floor) Upper Layer Lower Layer Flame Zone & Fire Plume grid cell (mass, momentum, energy conservation statements) Inputs to CFD: Physical models Numerical algorithms Computer power (PE speed/memory, parallel computing) Slide 10

11 Computer-based Modeling of Compartment Fires Features of the zone modeling approach: Computationally cheap (2 control volumes) System-level view point (unlimited in problem size and scope) Well-suited for studies of long time impact of fire loading (i.e. building fire resistance) Limited accuracy (large use of empirical correlations) Features of the CFD modeling approach: Computationally expensive (more than 100,000 grid cells) More limited view point (in problem size and scope) Moderate to high accuracy (based on first principles) Slide 11

12 Computer-based Modeling of Compartment Fires Specific challenges associated with the simulation of underventilated fires: Flame location [zone modeling] Smoke layer composition (CO and soot emission) [zone/cfd modeling] Air vitiation (flame extinction) [zone/cfd modeling] Focus of current work and today s presentations! Slide 12

13 Modeling of Under-Ventilated Compartment Fires Main component of UMD research program: Sponsor: NIST/BFRL ( ) Objective: adapt CFD modeling approach (FDS) to treat air vitiation effects and local/global flame extinction events Ph.D. students: Z. Hu, Y. Utiskul, J. Williamson Faculties: A. Marshall, J. G. Quintiere, A. Trouvé Slide 13

14 Modeling of Under-Ventilated Compartment Fires Joint computational/experimental program; 3 sub-components: Extinction modeling Validation Fundamental combustion experiments (local flame extinction physics) (A. Marshall) Compartment fire experiments (global fire dynamics), zone modeling (J. G. Quintiere) Slide 14 Model development, FDS implementation, verification/validation (A. Trouvé)

15 Modeling of Under-Ventilated Compartment Fires CFD modeling approach: Flammability diagram (in terms of the oxidizer stream properties): T c T O2 1,700 K Flammability Limit T st = T c Y, Flame Fuel T F F Oxidizer Y, T O2 O 2 T a Slide 15 = 300 K 0 Non-Flammable Domain Y O 2, c Flammable Domain 0.17 Y O2 (Lower Oxygen Index ~ 15%) Flammable conditions: Y Y O O 2 2, c T c T O ( 2 T c T a ) 0

16 Modeling of Under-Ventilated Compartment Fires CFD modeling approach: Starting FDS expression for the heat release rate (HRR): q& eq d &ω F Y F ν ~ ~ = ( ) Z 1 st 2 t ρ Z Z Z st Sc δ ( ) t H F Equilibrium chemistry model Modified expression with flame extinction capability: Y T T O2 c O2 & d = H ( ( )) YO c Tc T 2, a Heaviside function q q& eq d Flame extinction factor Slide 16

17 Modeling of Under-Ventilated Compartment Fires CFD modeling approach: Y eq O 2 Estimation of the oxidizer stream properties: (, ) ~ Z < OC OC Z st Z ~ T ~ ( ), ~ Z RC RC Y, T O2 O 2 Z st RC RC ~ Z > Z st FC FC FC Y O T O 2 2 Search algorithm aimed at identifying oxidizer cells (OC) around simulated flame location Use oxygen level and gas temperature in OC cells to estimate oxidizer stream properties Slide 17 OC Oxidizer side RC RC Flame RC Fuel side

18 Modeling of Under-Ventilated Compartment Fires Second component of UMD research program: Sponsor: in-house and EDF (France, ) Objective: adapt zone modeling approach (MAGIC, BRI) to treat radiation-enhanced and oxygen-limited enclosure fires (freely-evolving fuel MLR) Staff: V. Lecoustre (MS student, ENSMA, France), T. Mizukami (visiting scientist, BRI, Japan) Faculties: J. G. Quintiere, A. Trouvé Slide 18

19 Modeling of Under-Ventilated Compartment Fires Joint computational/experimental program: Validation Zone model development, MAGIC/BRI implementation, verification/validation (J. G. Quintiere, A. Trouvé) Compartment fire experiments (global fire dynamics) (J. G. Quintiere) Slide 19

20 321 Modeling of Under-Ventilated Compartment Fires Zone modeling approach: Modified expression for the fuel mass loss rate (MLR): Y ε & m & = ( F 4 O, ( ) 2 LL qul+ w σta m& F, FEF + ) YO 2, a H v free-burn value Air vitiation Modified expression for the heat release rate (HRR): Q& = ( m& FEF ( m& P Y F H F ) H O 2, LL O 2 Flame extinction factor ) if if GER 1 GER 1 A { F fuel source area Thermal feedback due to smoke layer and walls Slide 20

21 Modeling of Under-Ventilated Compartment Fires Zone modeling approach: Estimation of the lower layer properties: dyo2, LL ( ρllzd A) = m& & in O2, a O2, LL e O2, LL dt Upper Layer Y O, LL ( Y Y ) m ( Y Y ) doorway mixing between upper and lower layers 2 O 2, UL z D Lower Layer m& e Slide 21

22 Modeling of Under-Ventilated Compartment Fires Zone modeling approach: Model expression for doorway mixing rate (Quintiere & McCaffrey 1980): m& e = m& in k m T T a UL z N z z N D W W 0 n Upper Layer z D Lower Layer m& e Slide 22

23 Slide 23 Program of the Workshop

24 Program of the Workshop Experimental study of under-ventilated compartment fires. Analysis and parametrization (Yunyong Utiskul) Identification of different flame regimes when going from over- to under-ventilated fire conditions Identification of main control parameters Construction of experimental database for CFD model validation Slide 24

25 Program of the Workshop Zone modeling of under-ventilated compartment fires with MAGIC and with BRI (Vivien Lecoustre, Tensei Mizukami) Adapt zone models to more elaborate descriptions of combustion (MLR with thermal feedback and flame extinction capability) Comparisons with UMD experimental database Slide 25

26 Program of the Workshop CFD modeling of under-ventilated compartment fires with FDS (Hu Zhixin) Adapt CFD models to more elaborate descriptions of combustion (HRR with flame extinction capability) Comparisons with UMD experimental database Slide 26

27 Program of the Workshop Fundamental combustion experiments in support of FDS modeling (Justin Williamson) Identification of flame extinction conditions Evaluation of role of strain rate Comparisons with flammability map used in proposed flame extinction model for FDS Flammable conditions: χ st ( H st χ ) st, ext = T T ref st ref st χ ref st, ext T T st exp( 1 + H st βh st ref ( T T st ) / T ref st ) Non-dimensional measure of enthalpy deficit due to air vitiation Slide 27

28 Conclusion UMD research program on modeling of under-ventilated compartment fires: Team-based effort and weekly meetings Two components: CFD modeling with FDS (main component supported by NIST); zone modeling with MAGIC/BRI Work-in-progress: end-of-summer 05 Today s focus on flame extinction model. Ongoing work on smoke layer composition: Tracking of unburnt fuel that leaks across the flame without burning CO formation model Slide 28