An Introduction to the Uses and Limitations of CFD Modeling for Solving Fire Safety Problems in DOE/Industrial Type Facilities
Jason E. Floyd, Ph.D.
2003 Fire Safety Workshop @ PANTEX April 29 – May 2, 2003
HUGHES ASSOCIATES, INC
FIRE SCIENCE & ENGINEERING
Please Note
This presentation contains several animations. The animations have been removed to reduce file size for downloading. If you would like to receive a copy of the presentation complete with animations, please contact Jason Floyd at 410-737-8677.
DOE Order 420.1
Basic DOE objectives of fire protection programs are to minimize:
of fires or related events Releases impacting health and environment both onsite and offsite Interruption of vital DOE programs Excessive property loss Damage to safety class systems or critical process controls
Occurrences
Implementation Guide for Order 420.1
Use of NFPA or other recognized codes Complete and comprehensive analysis may support “equivalent systems” Unique facilities, nuclear facilities, other high risk facilities require an FHA Recognized that needs of many DOE facilities are not adequately addressed by current codes
DOE Need For Fire Modeling
To determine code equivalence for fire protection Evaluate fire risks of unique DOE facilities Quantitative support of hazard analyses
Methods of Analysis
Hand Calculations/Handbooks
large experience base Large uncertainties, limited applications
Fast,
Lumped Parameter/Zone Model
Fast,
accepted use, entire buildings No local values, geometry restrictions
CFD
Local
values, arbitrary geometric complexity, lowest level of empiricism. Steep learning curve, slow
Why CFD?
The Ability to Give Local Information
AFFF Hose Reels
Carrier Hangar Bay Fire
Complex Geometries
3 MW Lube Oil Fire in a Containment Building (HDR Test 52.14)
Fire Dynamics Simulator (FDS)
Developed by NIST Building and Fire Research Laboratory Companion software called Smokeview for viewing/animating FDS output Large eddy simulation (LES) Single parameter mixture fraction Gray gas, finite volume radiation heat transfer 1D heat conduction through surfaces
Additional FDS Capabilities
Multi-block grids Ignition of remote objects Pool fires with calculated heat release rates* Fire spread and growth over solid fuels* Droplets*
• Fuel spray fires • Conventional sprinklers • Mist sprinklers
Fire suppression by oxygen depletion and fuel cooling or delivered water*
*Level of physical detail in submodels may not support its use for all applications
Mixture Fraction Surface
Virginia Tech Fire Compartment
FDS v2 Simulation
400 kW propane fire in a 50%-scaled ISO-9705 Compartment
Multiblock Grids
Multiple computational grids Can have different node sizes Reduction of active grid cells at the expense of more complex boundary conditions
ISO-9705 Corner Fire
Heat Flux (kW/m2): 10 20 30 40 50 60 70 80 90 100 Heat Flux (kW/m ):
2
10 20 30 40 50 60 70 80 90 100
2.2 2.0 1.8 1.6 1.4 1.2
Z
2.2 2.0 1.8 1.6 1.4 1.2
1.0 0.8 0.6 0.4 0.2 0.0 0.0 0.2 0.4 0.6 0.8 Y 1.0 1.2 1.4 1.6
Z
1.0 0.8 0.6 0.4 0.2 0.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
Y
FDS v3
Lattimer et al., 1999
Methanol Pool Fire
0.028 0.024
Mass Flux (kg/m -s)
0.020
2
0.016
0.012
0.008 Hamins et al., 1999 FDS v3
0.004
0.000
0
0.5
1
1.5
2
Pool Diameter (m)
Heptane Pool, Thin Steel Wall/Floor
Remote Ignition
Compartment:
4mx4mx2m 2 m x1.5 m vent Marinte walls
1 m x 1 m kerosene pool fire along wall opposite door 0.5 m x 0.5 m x 0.5 m PMMA block in corner
Cost Reduction of Experiments
Pre-test computations for experimental design Given a benchmarked experiment, perform parametric studies in a neighborhood of the experiment
Sprinkler Activation
(McGrattan and Stroup, 1997)
67’1”
Draft Curtain
Sprinklers w/10’ Spacing
4’x8’ Roof Vent
2:45 5:48 2:44 2:40
71’2”
2:26 4:20
1:11 1:28
1:09 1:28 2:11 2:06
1:12 1:20
1:11 1:12 2:16 3:30
2:22 2:38
2:36 2:52
~4.5 MW Heptane Spray Fire
Time to Activate
FDS v1 Test
2:34 3:58 DNO 6:52
Droplet Evaporation and Radiation Attenuation
Cold Wall
Hot Wall
Water Mist System Discharge
(Hunt, Floyd, and Cutonilli, 2003)
250
1.2 MW heptane fire Two LN26 Spray Systems nozzles PDA area of the ex-USS Shadwell
FDS 18 Data 18 FDS 54 Data 54 FDS 90 Data 90 FDS 128 Data 128
200
Temperature (° C)
150
100
50
0
0
50
100
150
200
250
300
350
Time (s)
Consequences
Smoke spread/fallout Brand lofting
High Pressure Discharge of Flammable Liquid (Oil Well Blowout)
Limitations
Requires a knowledgeable user
Grid
generation Defining material properties Defining boundary conditions Aware of “proof by pretty picture” danger
Current mixture fraction combustion model
not lend itself to predicting extinguishment Does not properly account for vitiated fires Only represents a single fuel
Does
Difficult to quantify fire properties of solid materials
Limitations
Large inaccuracies in near-field heat feedback to fuel makes spread and growth predictions for solid fuels difficult if not impossible No structural failure submodels though obstacles can be created/removed Sprinklers
Difficult
to quantify spray characteristics No droplet breakup, condensation, or surface wetting submodels (assumes uniform sheet)
Resolution vs. available time (double the nodes per side = 16 fold increase in time)
Potential Benefits for DOE of CFD Modeling
Avoid under-design of fire protection Reduce conservatisms in risk assessments for unusual spaces Examine cost vs. benefits of passive fire protection features (separation distance, heat shields, etc.) Evaluate protection system design for mission critical or one-of-a-kind equipment and public safety for applications where no code guidance is available