Application of ISIS computer code to gas flares under varying wind conditions

Application of ISIS Computer Code to Gas Flares Under Varying Wind Conditions Ahti Suo-Anttila Alion Science & Technology Albuquerque, NM [e-mail] Joseph D. Smith Alion Science & Technology Owasso, OK [e-mail] OUTLINE • Introduction • ISIS Overview • Applications • Mock fuselage engulfed in fire • Flare Tips in Cross Wind • Conclusion 2006 AFRC International Symposium Introduction: Flare Tips in Cross wind Flame bends over in wind and licks downwind side of flare stack Air egression into stack tip can lead to internal burning and possible explosive conditions Air egression into flare stack P. Gogolek, CANMET Energy Technology Centre - Ottawa Natural Resources Canada 2006 AFRC International Symposium Isis-3D • Based on Computational Fluid Dynamics with radiative heat • • • transfer Contains one-dimensional conduction modules to simulate response of simple engulfed objects Container Analysis Fire Environment (CAFE) related utility developed at Sandia (links Isis-3D to PATRAN finite element code). Linked system is capable of modeling complex, three-dimensional objects engulfed in fires. 2006 AFRC International Symposium Isis-3D designed to: • Provide reasonably accurate estimates of the total heat • • • transfer to objects from large fires Predict general characteristics of temperature distribution in object Accurately assess impact of variety of risk scenarios (wind, % flame coverage, thermal fatigue for given package geometry, etc.) Reasonable CPU time requirements on "standard" desktop LINUX workstations (i.e., P4 processor, 1 GByte RAM) 2006 AFRC International Symposium Trade-Offs • Sacrifice generality (large fires only) in favor of quick turnaround time and quantitative accuracy • Reaction rate/radiation heat transfer models apply only to large fires • Models aimed at making Isis-3D predictions "goodenough" when coarse grids employed 2006 AFRC International Symposium Computational Techniques • Variable density version of PISO pressure-based algorithm • 1st and 2nd order Finite difference Methods • Porosity method for curved surfaces (i.e., structured grid with • body fitted computational cells) Uses acceleration factor (TF) to model long burn times Portion of cell inside solid Portion of cell outside solid Cell fractional flow areas on all faces 2006 AFRC International Symposium Radiation Inside Large Fires • High soot volume fractions make large fires non-transparent (optically thick) and radiates as a cloud (radiatively diffuse) • Fire volume located where soot volume fraction fSoot >fmin = fFlameEdge • fFlameEdge determined to be 0.4x10-6 (0.4 ppm) in current work based on comparisons with large fire experiments Snapshots of Isis-3D calculated fire surfaces for three different experiments 2006 AFRC International Symposium Radiation Outside of Large Fires • When fSoot < fFlameEdge = non-participating • • • • medium View factors from the fire to the un-engulfed surfaces of the object calculated at each time step, but neglect fire shadowing Radiation view factor from object surface to surroundings calculated at each time step εFireSurface = 1 (fire is black body radiator) Radiation from the fire surface to the surroundings employed Tsurround = (constant ambient temperature) 2006 AFRC International Symposium Diffuse Radiation Within Fire • Calculated indirectly using a Rossland effective thermal conductivity 16σT 3 kR = >> k Air 3β R • σ = Stefan-Boltzman Constant • T = local temperature • βR= local extinction coefficient. Dependent on local species concentrations 2006 AFRC International Symposium Combustion Chemistry Model • Variant of Said et al. (1997) turbulent flame model • Relevant Species • • • • • F = Fuel Vapor (from evaporation or flare tip) O2 = Oxygen PC = H20(v) + CO2 C = Radiating Carbon Soot IS = Non-radiating Intermediate Species • Four model reactions • Model options include eddy dissipation effects and local equivalence ratio effects 2006 AFRC International Symposium Reactions Involving Fuel • Incomplete Fuel Combustion (soot producing) • 1 kg F + (3.0-2.6S1) kg O2 → S1 kg C + (4.0-3.6S1) kg PC + (42-32S1) MJ • Combustion Soot Mass Parameter, S1 = 0.05 • Endothermic Fuel Pyrolysis (soot producing) • 1 kg F + 0.3 MJ → S2 kg C + (1-S2) kg IS • Cracking Parameter, S2 = 0.15 2006 AFRC International Symposium Reactions Not Involving Fuel • Soot Combustion 1 kg C + 2.6 kg O2 → 3.6 kg CO2 + 32 MJ • Combustion of Intermediate Species 1 kg IS + 3 − 2.6 S 2 4 − 3.6 S 2 41.7 − 32 S 2 kg O2 Æ kg PC + MJ 1 − S2 1 − S2 1 − S2 • Coefficients chosen so that complete combustion of C and IS produce same species and thermal energy as direct combustion of fuel 2006 AFRC International Symposium Reaction Rate Model • Arrhenius rate model • Consumption of primary reactant increases on mass fraction of reactants fRi and temperature T in volume ⎡ = −C ⎢ dt ⎢⎣ df R1 N Π i ⎤ −T / T f Ri ⎥ e A ⎥⎦ • Coefficients C and Activation Temperatures TA must be determined for all four reactions 2006 AFRC International Symposium Parameter Values Reaction Fuel Combustion Fuel Cracking Soot Combustion Intermediate Species Combustion R1 F F C IS N 2 1 2 2 TA 8000 K 42,730 K 26,500 K 8000 K C 1013 1014 108 1013 • Activation Temperatures TA were taken from Said et al. (1997) • Coefficients C were chosen based on comparison with experiments by Gritzo et al. (1998) 2006 AFRC International Symposium Test Facility 2006 AFRC International Symposium Test Facility 2006 AFRC International Symposium Details of Test Facility: Pool 18.28 m 6.09 m 1 LE 3.66 m 2 3 4 L 5 C 3.66 m 6 7 8 R • 18.9 m diameter JP8 fuel pool • Pipe suspended 0.6m above RE Thermocouple Ring Locations • Culvert Pipe • 18.9-m diameter fuel pool leeside of pool Pipe axis perpendicular to predominant wind Wind speed and direction measured 2 and 10 m above the ground on two upwind poles x θ z 30 m 30 m Predominant Wind Direction Southwest Anemometer Pole South Anemometer Pole 2006 AFRC International Symposium Details of Test Facility: Culvert Pipe Culvert Pipe Thermocouple Ring Locations • 18 m long, 4.7 m diameter, 1.6 mm 18.28 m • • • • 6.09 m 1 LE 3.66 m 2 3 4 L 5 C 3.66 m 6 7 8 R RE • L = (2 + 3)/2 • C = (4 + 5)/2 • R = (6 + 7)/2 t wt Predominant Wind direction walls 8 interior TC rings Rings LE and RE had 4 TC's Central rings had 8 TC's Average data from interior rings lt w l wb lb b 2006 AFRC International Symposium 19-m pool Fire with Strong Winds 9.5 m/s wind Pipe • Suo-Anttila and Gritzo (2001) • Measured temperature of culvert pipe suspended over leeside of a JP8 pool 2006 AFRC International Symposium Isis-3D Model: Domain • 60 m square base; 15 m high • 16,500 elements, highly refined near pipe • Each pipe grid linked to 1-D conduction module (see next slide) LX = 60 m NX = 31 Y LZ = 60 m NZ = 28 X Z LY = 15 m NY = 19 Pool Location Pipe 2006 AFRC International Symposium Isis-3D Model: 1-D Conduction Submodel Pipe Wall Fire Heat Flux Air Gap Insulation Nichrome TC (output) Strap • εsurface = 0.9 • Neglects axial and azimuthal conduction • Simulated TC's and nichrome straps were spread uniformly on the interior surface 2006 AFRC International Symposium Test Conditions Wind Speed and Direction Speed Direction 16 180 14 12 90 s [m/s] 10 θ 8 0 6 4 -90 2 0 0 • • • • 200 t [sec] 400 600 -180 0 200 400 600 t [sec] Test was 11 minutes in duration Average Wind speed (Savg) = 9.5 m/s (25 mi/hr), Average direction (θavg) = -110 No sustained speed or direction changes during tests Velocity Boundary for inlet/pressure for outlet 2006 AFRC International Symposium Fuel Pool • Employed an average fuel flux of 0.072 kg/m2s Gritzo et al. (1997). • Apply a two zone pool model, w = D/19 w D • Outer ring flux 3 X inner circle flux to model increased heat transfer and entrainment (somewhat arbitrary) 2006 AFRC International Symposium Simulation • Fire rose from all points of the pool simultaneously • High wind speed (fire volume exits through right side • • • and top) Fire engulfs the entire pipe length Upper windward surface and right end were periodically uncovered Strong downstream recirculation zone increases mixing and fire temperature 2006 AFRC International Symposium Average Pipe Temperature Rise ΔT = TAVG(t) - TAVG(0) 1200 1000 800 ΔTavg [K] 600 400 200 0 0 200 400 600 t [sec] • Temperature rise roughly proportional to total energy delivered to pipe • CFD based Tool must accurately predict this for risk analysis work 2006 AFRC International Symposium Total Heat Delivered vs. FFlameEdge 1200 1000 800 ΔTavg [K] 600 Exp 3 Data 400 Isis-3D: Fs,m=0.3 ppm, TF = 12 Isis-3D: Fs,m=0.4 ppm, TF = 12 Isis-3D: Fs,m=0.5 ppm, TF = 12 200 Isis-3D: Fs,m=0.4 ppm, TF = 1 0 0 200 400 600 t [sec] • Accelerated (low specific heat) simulations, TF = 12 • Fire size and heat transfer increase as FFlameEdge decrease • Accelerated simulations respond to fire puffing but are close to un-accelerated • result FFlameEdge = 0.4 ppm gives best agreement 2006 AFRC International Symposium Computational Time • 2.4 GHz Linux PC with 0.5 Gb RAM • TF = 1, 670 sec fire duration, 40 hour simulation • TF = 12, 56 sec fire duration, 3.5 hour simulation 2006 AFRC International Symposium Inner Ring Temperatures Isis-3d Measurements C-b C-wb C-w C-wt C-t C-lt C-l C-lb L-b L-wb L-w L-wt L-t L-lt L-l L-lb R-b R-wb R-w R-wt R-t R-lt R-l R-lb 1500 T [K] 1300 1100 900 700 500 300 1700 C-b C-wb C-w C-wt C-t C-lt C-l C-lb L-b L-wb L-w L-wt L-t L-lt L-l L-lb R-b R-wb R-w R-wt R-t R-lt R-l R-lb 1500 1300 1100 T [K] 1700 900 700 500 300 0 0 200 400 600 200 400 600 t [sec] t [sec] • Calculated temperatures at all locations rose @ t = 0 • System reached steady state after t ~ 300 sec, T = 1000 to 1600 K • Calculated temperatures were in the correct range but not correct at • individual locations Compare temperature distributions at t = 400 sec 2006 AFRC International Symposium Local Temperature Isis-3d Wind Top 1500 1300 Top Data Top 1700 1700 1500 Wind Top Lee Top 1300 1100 1100 900 900 700 700 500 300 Wind Wind Bot Lee Top Leeward Lee Bot left end left middle right right end 500 300 Wind Leeward Left End Left Center Right Right End Wind Bot Lee Bot Bottom Bottom • Coolest near the top, hottest on leeside: caused by flame tilt and • • downstream recirculation zone Nearly uniform at each axial position Isis-3D max temperature closer to the ground than data 2006 AFRC International Symposium Application: Rail Car engulfed in fire 2006 AFRC International Symposium Application: Multi-Tip Ground Flare Flames merge due to side air demand from sides of rows Test Problem • 124 TPH Propane • Wind Fence surrounds ground flare • Specified open area in fence • Individual Flare Tips located above grade • No wind conditions • Low pressure/non-momentum driven flames 2006 AFRC International Symposium Application: Multi-Tip Ground Flare Low flow - buoyancy driven flames 2006 AFRC International Symposium Application: Utility Flare in Purge Conditions Wind 2006 AFRC International Symposium Conclusions • Buoyant flames important class of problem in gas flaring • ISIS-3D Details: • Combustion chemistry parameters and the soot volume fraction define fire edge (based on large fire measurements) • Provides reasonably accurate estimates of total heat transfer from large fire to engulfed object under steady wind conditions • ISIS-3D predictions verified against test conditions for pool fire engulfing external body 2006 AFRC International Symposium Conclusions • Can link ISIS-3D to other codes (CAFÉ) for risk assessment for • • variety of wind conditions and arbitrary configurations Can conduct ISIS-3D simulations in reasonably short CPU times on low cost standard workstations ISIS-3D has been applied to sample flare conditions for utility flare and multi-tip flare field with wind fence • • • • Evaluate effect of tip/row spacing on air demand Evaluate radiation effect on external structures (i.e. fences) Evaluate effect of flame interaction on flame height Evaluate effect of purge conditions on air ingression 2006 AFRC International Symposium