LNG POOL FIRE MODELING

1
LNG POOL FIRE MODELING

• Background The MTB Model and the Need for
  Better Modeling Methods
• View Factor Models LNGFIRE as an Example
• Theoretical Fire Models The FDS Computational
  Fluid Dynamics Model as an Example

2
EXERCISE 2

• What determines the radiant (thermal) energy you
  receive from a liquid pool fire?

3
ALPHABET SOUP

• MTB (Materials Transportation Bureau, U. S. DOT
• RSPA (Research and Special Programs
  Administration, U. S. DOT)
• PHMSA (Pipeline and Hazardous Materials Safety
  Administration, U. S. DOT)

4
THE MTB MODEL FOR POOL FIRES

• Promulgated into 49 CFR 193.2057 in 1980s
• Resulted from Review of 1971 NFPA, U. S. Bureau
  of Mines, AGA-Sponsored, ESSO, U. S. Coast
  Guard-Sponsored Work
• d f (A)0.5
• where
• d exclusion distance measured perpendicular
  to flame surface to target
• A horizontal area of impoundment
• f offsite classification factor based on
  radiant flux limit.

5
(No Transcript)
6
THE MTB MODEL (Cont.)

• T Tilt angle 45o (always)
• L Flame Length reduces to 3 D (always)
• D Equivalent Diameter (rectangular as well
  as circular impoundments)
• f

7
INCIDENT FLUX LIMITS
Abbreviated definitions from 49 CFR
193.2057(1980).
8
CRITICISMS OF THE MTB MODEL

• L and T Fixed Specifications are Unsupported
• Point Source Energy Model (used for calculating
  target energy from flame surface and f factors)
  is Inferior to Full Flame Surface Representation
  (cylinder or parallelepiped)
• Average Maximum (black body) Surface Emissive
  Power Specification - Estimated 142.0 kW/m2
  (45,000 Btu/hr ft2) Is Not Consistent with Data
• Full technical discussion in LNGFIRE A
  Thermal Radiation Model for LNG Fires, Gas
  Research Institute, GRI-89/0178, June 29, 1990.

9
CRITICISMS OF THE MTB MODEL (Cont.)

• Surface Emissive Power is Not Constant Varies
  Exponentially With Flame Thickness
• Flame Length Varies With Burning Rate (and,
  secondarily, wind speed) Slight Differences for
  Equilateral Pools and Elongated Trenches
• Flame Tilt Angle Varies With Respect to Wind
  Speed and Dimensions (size and shape) of
  Impoundment
• Flame Drag May be Important Varies With Wind
  Speed
• Elongated Trenches
• Radiation Attenuation Due to Water Vapor.

10
LNGFIRE (1989)

• Currently Referenced Model in 49 CFR 193.2057
• Resulted from Several Years of Effort to Resolve
  MTB Model Criticisms, Including Need to Model
  Elongated Trenches
• Key Research
• Coast Guard (view factors)
• Shell (surface emissive powers)
• British Gas (correlations of flame length, tilt,
  and drag)
• GRI-ADL/British Gas (trench fires)

11
LNGFIRE SUMMARY

• Model Type Semi-Empirical
• Basic Equation
• q F t e qs
• where
• q Incident radiant heat flux at the
  target (kW/m2)
• F Geometric view factor from flame
  surface to the target (non-
  dimensional)
• t Transmissivity of the atmosphere to
  thermal energy (0 to 1)
• e Average emissivity of the flame ()
• qs Maximum effective black body
  radiation of the flame (kW/m2)
• eqs Surface Emissive Power (kW/m2)

12
VIEW FACTOR CONSIDERATIONS
Integration
FdA1?A2 1 / p A2? cosß1 cosß2 dA2/r2
Piecewise
13
VIEW FACTOR REQUIRED COVERAGE

• Vertical and Horizontal Targets
• Targets in the Flame Shadow
• Elevated Flame Bases Relative to Target
• Elevated Targets Relative to Flame Bases

14
FLAME LENGTH CALCULATION

• Lf/D 42 (m / ?a v(gD))0.61
• where
• Lf Flame Length (m)
• D Pool Diameter (m)
• g Gravitational Acceleration (m/s2)
• m Mass Burning Rate (kg/m2s)
• ?a Ambient Air Density (kg/m3)
• Calculation for circular pool.

15
FLAME TILT CALCULATION

• cos ? 1 / vU for U 1
• cos ? 1 for U 1
• where
• U U / Uc
• U Wind Velocity (m/s)
• Uc Characteristic Velocity (m gD /
  ?v)1/3
• m Mass Burning Rate (kg/m2s)
• ?v LNG Vapor Density (kg/m3)
• Calculation for circular pool.

16
FLAME DRAG CALCULATION

• DR (D D)/D 1.5 (Fr)0.069
• where
• DR Drag Ratio (Drag Distance/Diameter)
• D Pool Diameter (m)
• D Extension of the Flame Base Beyond
  Pool Edge
• Fr Froude Number u2 / gD
• Calculation for circular pool.

17
BURNING RATE CALCULATION

• m 0.11 1 exp (-0.46D)
• or
• m 0.11 kg/m2s
• where
• m LNG Burning Rate (kg / m2s)
• D Pool Diameter (m)
• Calculation for circular pool.

18
ATMOSPHERIC TRANSMISSIVITY

• t 1 - aw - ac awac
• where
• aw Absorptivity of Water Vapor
• ac Absorptivity of Carbon Dioxide

19
FLAME SURFACE EMISSIVE POWER

• e qs 190 (1 - e-0.3Df)
• where
• e Flame Emissivity
• qs Maximum Effective Black Body Radiation
  Emissive Power (kW/m2)
• Df Flame Thickness (m)

20
HYPOTHETICAL ZONED FLAME
21
SURFACE EMISSIVE POWERS AS MEASURED
GRI-ADL/British Gas trench fire tests, Test 8,
side view
22
MONTOIR 35 METER POOL FIRES (1987)
Test 1
Test 2
Test 3
23
SURFACE EMISSIVE POWER DATA AND CURVE FIT FOR
LNGFIRE EQUATION
24
LNGFIRE VALIDATION AND MTB MODEL COMPARISON,
DOWNWIND
GRI-ADL/British Gas trench fire tests, Test 4
25
LNGFIRE VALIDATION AND MTB MODEL COMPARISON,
CROSSWIND
GRI-ADL/British Gas trench fire tests, Test 4
26
OTHER REGULATORY MODELS

• FIRES2 British Gas/Advantica
• CORE Gaz de France
• Model Comparison Results to Montoir 35m Scale
  Pool Fire Scenarios, Including Experiments
• The conditions calculations corresponding to
  the Montoir experiments lead to a rather good
  agreement,with relative differences being 10 to
  30 for crosswind and downwind, respectively.
• Debernardy, J. L., Perroux, J. M., Nedelka, D.
  Comparison of LNG Fire Radiation Calculation
  Codes, Gaz de France, 1992.

27
BUT DO SEMI-EMPIRICAL VIEW FACTOR MODELS MEET ALL
NEEDS?

• Irregular Shapes Unconfined Spreading, Flow
  Barriers
• Interaction with Fire Control Measures
• Structures in Flames and Their Interaction with
  Fire Dynamics (e.g., presence of a tank shell)
• Smoke Shielding
• Transient Behaviors

28
THEORETICAL MODELS FOR POOL FIRES FDS

• FDS Fire Dynamics Simulator (Version 4, 2004),
  Developed and Supported by the U. S. National
  Institute of Standards and Technology (NIST)
• Under Development for 25 Years
• Computational Fluid Dynamics (CFD) Model for
  Low-Speed Fire-Driven Flow Emphasizing Heat
  Transport and Smoke
• Time-Dependent ,3-D Spatially Computed
  Differencing Solutions Approximating the Partial
  Differential Navier-Stokes Equations for
  Conservation of Mass, Momentum, and Energy

29
FDS APPLICATIONS

• Low-Speed Transport of Heat and Combustion
  Products from Fires (Thermal Radiation Computed
  Using a Finite Volume Technique Within the 3-D
  Grid)
• Radiative and Convective Heat Transfer Between
  Gas and Solid Surfaces
• Pyrolysis
• Flame Spread and Fire Growth
• Interactions with Fire Suppression and Detection
  Systems

30
FDS MODEL RESULTS

• Within the Fire Plume and Surrounding Air
• Gas Temperature, Velocity, Concentration by
  Species, and Density
• Smoke Concentration and Visibility
• Pressure
• Heat Release Rate per Unit Volume
• Mixture Fraction
• Water Droplet Mass per Unit Volume
• Impingement on Solid Surfaces
• Surface and Interior Temperatures
• Radiative and convective Heat Flux
• Burning Rate
• Others, Including Global Quantities

31
FDS GENERAL MODEL STRUCTURE

• Hydrodynamics Model, Including Navier-Stokes
  Approximation Differencing Equations and
  Turbulence
• Large Eddy Simulation (LES) Course Grids
• Direct Numerical Simulation (DNS) Fine Grids
• Combustion Model, Based on Scalar Quantity
  Mixture Fraction
• Radiation Transport Model, Based on Finite Volume
  Method (FVM) Including 100 Discrete Transport
  Angles
• Geometry (Gridding) Model for One or More
  Rectilinear Grids
• Boundary Condition Definitions, Assessed as
  Thermal as well as Physical Boundaries for
  Controlling Heat and Mass Transfer
• Fire Target Response Models, Including Sprinkler
  and Detectors, and Water Sprays (Lagrangian
  Droplets)

32
FUNDAMENTAL CONSERVATION EQUATIONS
33
SIMPLIED EQUATIONS USED IN FDS
34
VALIDATION FOR POOL FIRES

• Historical Development on Unconfined Fires
• Since 2000 and Revision Code
• Methane Pool Fire 1 m diameter (Xin,et. al.,
  2002)
• Methane and Methanol Pools (Hostikka, et. al.,
  2002)
• Heptane Pools (Hietaniemi, et. al., 2004)

35
COMPUTER REQUIREMENTS

• Recommended Minimum
• Windows-Based PC Running 1 GHz Pentium III, with
  512 MB RAM
• 1 GB Storage per Average Large Simulation
• But Really - The Faster (and Bigger), the Better

36
DOCUMENTATION

• FDS
• SMOKEVIEW
• Website http//www.fire.nist.gov/fds/

37
GENERAL STEPS FOR SETTING UP FDS RUNS

• Input Files
• Setting Time Limits
• Defining Computational Domain (i.e., Grid Mesh)
• Defining Boundary Conditions
• Defining Fire Conditions Via Combustion
  Parameters
• Defining Obstructions, Mitigation Systems
• Running
• Monitoring Progress
• Error Statements
• Output Files
• Point Measurements Within the Domain
• Animated Planar Slices, Boundary Quantities,
  Isosurfaces (SMOKEVIEW)
• Static Data Files

38
POTENTIAL APPLICATIONS OF CFD TO LNG FIRES

• Addressing Limitation of View Factor Models
• Non-Regulatory Cases
• Complex, Progressive Failures
• Complex Consequence Analysis
• Analysis of Hazard Mitigation Measures
• Phenomena Other Than Pool Fires?