Current Fire Research Topics at Sandia Sheldon Tieszen April 30, 2010 Sandia National Laboratories Albuquerque, NM 87185 USA [email protected] SAND2010-2789C Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed-Martin company, for the United States Department of Energy’s National Nuclear Security Administration under contract DEAC04-94AL85000

Outline • Gas-phase Validation – Radiative Transfer Equation is the last

• Solid fuels – Metals (sodium, Al/propellants) – Organics (foams, plastics, biomass, composites)

• Liquid fuels – Splash dynamics – LNG

• Numerics – Cantera – FEM progress

Gas-Phase Validation • Have extensive validation data for physics within a fire – Non-reacting & Reacting Turbulent Buoyant Plume Data – Soot & Species Dataset in JP8 – Convection/radiation partitioning in a methanol fire • Also have non-reacting high temperature plate data – Heat transfer to a cone/cylinder calorimeter in calm conditions – Heat transfer to a cone-shaped calorimeter in a cross-wind • Also have a nice outdoor cylindrical calorimeter in cross flow data set – Heat flux to & evaporation rate from liquid pools – Spectral absorption above a liquid pool

• Last remaining validation is for the Radiative Transfer Equation – Two attempts @ 3 years each to get to this point – Will begin preliminary testing this year with full production next year

First Integrated LII & CARS Measurements In Pool Fires To Start This Fall Simplified Radiative Transfer Equation (RTE)

µσ T 4 dI = −µI ds π

Air Ring

Combustion Air Supply 2000 CARS (0.01 cm X 1 cm) TFNS SIMULATION (5-cm cells)

Quantitative Soot Imaging (LII)

1500

1000 Advanced Optical Diagnostics 500 • Coherent anti-Stokes Raman Scattering (CARS) 0 0 • Temperature • Species Mole Fractions • Laser-induced incandescence (LII) • Quantitative soot concentration imaging • Soot radiative properties • Spectrally resolved radiation intensity •Intensity along a path

5 4 3

Temperature MoleFraction Data (CARS) 0.05

0.6

2 1

0.1 0.15 0.2 0.25 MOLE FRACTION, O /N 2

2

Spectrally Resolved Radiation Intensity

0.3

Y (mm)

absorption

PI: Sean Kearney, [email protected]

Water-Cooled Interior Walls

Grated Floor

TEMPERATURE (K)

emission

Exhaust to Precipitator

0.5

5

0.4

4 3

0.3

2 1

0.2 5 4

0.1

3 2

0

1

fv (ppm)

0 0

2

4

6

X (mm)

8

10

Outline • Gas-phase validation – Radiative Transfer Equation Update

• Solid fuels – Metals (sodium, Al/propellants) – Organics (foams, plastics, biomass, composites)

• Liquid fuels – Splash dynamics – LNG

• Numerics – Cantera – FEM progress

Liquid Sodium Fires •

•

• – –

•

Use of sodium coolant has unique safety implications not faced by the current LWR fleet Establish capabilities needed to support metal fires safety analyses for advanced reactors Performed 11 pan and 4 spray sodium fire tests Spray tests: heat flux and plume temperature data, droplet distribution Pool tests: surface heat flux and pan temperature data, exploring quenching phenomenon

Utilizing experimental data to develop advanced fire modeling capabilities

PI: Tara Olivier, [email protected]

Experimental Results

Propellant Fires (NASA-SNL Collaborative Test Series) Graphite Flat Plate Calorimeter (design showing TC locations)

Images pulled from video (same scale on all) prior to ignition

fireball during burn

Downward-facing Propellant Charge Held Above Calorimeter

post-burn glowing deposit (after ~103 sec. burn)

PI: Walt Gill, [email protected]

Post-test Calorimeter with Deposit

Deposit Removed from Plate: – 3.038 kg. – approx. 50/50 mix of Al/Al2O3 – fairly uniform thickness

Propellant Fire - Object Coupling FUEGO

CALORE

40

2650 cumulative mass

cumulative mass (g)

35

2600 mass average temperature

30

2550

25

2500

20

2450

15

2400

10

2350

5

2300

0

2250 0

0.25

0.5 0.75 time (s)

1

mass-averaged particle temperature (K)

Cumulative Mass, Mass-averaged temperature

1.25

Total Deposition Rate 40

deposition rate (g/s)

35 30 25

• One-way coupling (Fuego
Calore) • Deposit characteristics (deposition rate, temperature) averaged over time period from 0.25 to 1.25 seconds. These were assumed to persist through the entire burn (>100 seconds). • Not spatially averaged • Results for deposition rate and temperature of the deposited particles were stored in a data file (for each mesh location). This file was read in by Calore and applied at appropriate geometric locations.

20

• Improvements in particle distribution were realized. Still some “islanding.”

dm/dt

15 10 5 0 0

0.25

0.5 0.75 time (s)

1

1.25

PI: Bill Erikson, [email protected]

Deposit Thickness

Temperature

Now the fun starts – what happens to metal substrates with molten Al on them? • Dissolve – Liquid Al is very effective at dissolving other metals

• Transport – What physics is important in molten metal transport – the oxide

• Reaction – Al ignition occurs below the oxide melt temperature due to oxide vapor equilibrium

Metal Dissolution – Iridium Example Aluminum-Iridium •

• •

Phase diagram gives solubility of one metal in another – Liquidus line Rate of dissolution is diffusion limited At temperatures over 2000K, erosion rates are significant

ρ Ir vIr = ρ0 v0 = −

ρ 0 D dX 1 − X 0 dy

0

10-3 10

-4

10-5

vIr [m/s]

forced convection natural convection 1s

10

-6

10

-7

60 s 600 s

10-8 10-9 1000

PI: John Hewson, [email protected]

1500

2000

Temperature [K]

2500

Al & Al Alloy Transport Preliminary Observations

Al 6061

Al 1100

Molten Aluminum relocation is severely limited by the oxide shell. The Shell: • Doesn’t stretch • Collapses • Forms as fast as the molten metal moves PI: Walt Gill, [email protected]

Al ignition temperature below oxide melt is transport related 8E-05 6E-05

mdot [kg/m2/s]

4E-05 2E-05 0

-2E-05 -4E-05

• Result is burning rate for both Al and Al2O plus surface heating rate. PI: John Hewson, [email protected]

1600

1800 2000 2200 Temperature [K]

2400

Measured ignition temperatures and predicted change in sign of mass flux 100000

80000 Pressure [Pa]

– Introduce Al2O with Al as dual “fuel” species at surface. – Carry out traditional analysis but with multiple conserved scalars to obtain expression for burning rate.

increasing oxidizer mass fraction

O2/N2 oxidizer

Al2O3 melts

• Among Al-O system, Al2O has substantial vapor pressure at Al surface. • Standard approach to surface burning: conserved scalars form Spalding Transfer number.

Net surface mass flux of Al2O predicted by conserved scalar model. 0.0001

60000

40000

20000

0 1800

CO2 oxidizer

2000 2200 Temperature [K]

2400

Outline • Gas-phase validation – Radiative Transfer Equation Update

• Solid fuels – Metals (sodium, Al/propellants) – Organics (foams, plastics, biomass, composites)

• Liquid fuels – Splash dynamics – LNG

• Numerics – Cantera – FEM progress

Packing foams generate significant gas pressures in sealed volumes such as shipping containers

When decomposing foam flowed away from the heat flux, the time to pressure (2.5 MPa, 350 psig) increased by a factor of 2 to 10 relative to the time to pressure when decomposing foam flowed toward the heat flux. X-Ray movies show complex physics including erosion and channeling of the foam via liquid. If the liquid hits the hot surface, gas generation rates soar.

PI: Ken Erickson, [email protected]

Direct current Electrical Shorting In REsponse to Exposure FIRE (DESIREE-FIRE) DESIREE-FIRE is a collaborative effort between U.S. Nuclear Regulatory Commission and Electric Power Research Institute – Analyze cable fire effects on DC powered circuits, similar to previous experimental program (CAROLFIRE) • Utilize a 60-cell, 125VDC battery bank with a 13,000A fault potential • Test multiple circuits (e.g., motor starter, switchgear, solenoid valve) with fuses ranging from 5 to 35A • Investigate the reliability of various control cables – Determine failure modes and effects of circuits • Spurious operation of equipment • Duration of equipment actuation • Energetic arcing behavior • Inter-cable shorting PI: Steve Nowlen, [email protected]

Direct current Electrical Shorting In REsponse to Exposure FIRE (DESIREE-FIRE) • Electrical failure was more energetic than AC circuits Live cable hanging after heat exposure

Conductor welded to tray

Copper slag from conductors

Direct current Electrical Shorting In REsponse to Exposure FIRE (DESIREE-FIRE) • Larger fusing contributed to longer duration spurious operation

Spurious Operation

Spurious Operation Fuse Blow

Normal Operating Condition

Normal Operating Condition

SNL has supported the U.S. Nuclear Regulatory Commission on Nuclear Power Plant Fire Protection continuously since 1975

Evaluation of the Wildland/Urban Defensible Space using Computational Fluid Dynamics • The objective of this work is to determine if a 30 ft stand-off distance is sufficient to reduce the likelihood of ignition of a house surrounded by both a thinned and unthinned forest. • The Sandia Computational Fluid Dynamics Fire code, SIERRA/Fuego, was used to investigate four cases with a house surrounded by forest.

• The results indicate that the current recommendation of a 30 ft stand-off distance of a thinned forest by the National Wildland/Urban Interface Fire Protection Program and Institute for Business and Home Safety does reduces the potential for house ignition. PI: Anay Luketa, [email protected]

Evaluation of the Wildland/Urban Defensible Space using Computational Fluid Dynamics

Case 1

Case 2

Case 3

Case 4

Carbon Fiber/Epoxy Composite Fires • •

Beginning work on composite material fires. What is the heat flux & duration in such fires? – Very long durations – like a charcoal briquettes lit with starter fluid? – Very high heat fluxes – can a forge-like environment develop in a cross wind?

PI: Alex Brown, [email protected] – Experiments PI: Amanda Dodd, [email protected] - Modeling 21

Outline • Gas-phase validation – Radiative Transfer Equation Update

• Solid fuels – Metals (sodium, Al/propellants) – Organics (foams, plastics, biomass, composites)

• Liquid fuels – Splash dynamics – LNG

• Numerics – Cantera – FEM progress

Transportation Accidents • This work builds on some validation testing performed to a high-speed impact of a water filled tank. • Liquid deposition, particle sizing, and video data

PI: Alex Brown, [email protected]

Transportation Impact Validation • Geometry fidelity was found to be most significant, and coupling methodology was also important Track Key: Predictions Significantly High Predictions Mostly High Predictions Accurate Within Data Mixed Results Predictions Mostly Low Predictions Significantly Low

9

8

7

6

5

4

3

2

1

Target

Total Deposited Mass Fraction

0.16 0.14 0.12 0.10 0.08 0.06 0.04 0.02 0.00 Case1

Case2

Case3

Case4

Case5

Case6

Case7

Data

Medium Fireball Video

The Big Test

Outline • Gas-phase validation – Radiative Transfer Equation Update

• Solid fuels – Metals (sodium, Al/propellants) – Organics (foams, plastics, biomass, composites)

• Liquid fuels – Splash dynamics – LNG

• Numerics – Cantera – FEM progress

Interactions with Cantera •

Cantera is an open-source effort for modeling chemistry: thermodynamics, kinetics, transport properties, etc. –

•

Effort initiated by Dave Goodwin, Cal Tech.

Sandia participation is currently strong. –

Focus on implementing non-ideal solution thermodynamics and transport, surface reactions, condensed phase reactions and phase transformations.

•

Looking for other participants ([email protected])

•

Potential for flamelet libraries, soot, condensed-phases, optical properties, etc.

Testing of Numerical Strategies for Coupled Physics • Example: HFEM fully coupled viscous FSI • Essence of the method is to write a compatibility boundary condition for continuity of normal stress in conjunction with the no-slip fluids condition

PI: Stefan Domino, [email protected]

Disparate Physics Requires Flexible Coupling Strategies Arbitrary coupling defined in input file – Fully coupled: Begin Region myEqSet Activate continuity equation with Mass Adv Activate momentum equation with Mass Adv Diff Src Activate mixture_fraction equation with L_Mass Adv Diff End Region myEqSet

– Loosely coupled: Begin Region myCtntEqSet Activate continuity equation with Mass Adv End Region myCtntEqSet Begin Region myMomEqSet Activate momentum equation with Mass Adv Diff Src End Region myMomEqSet Begin Region myZEqSet Activate mixture_fraction equation with L_Mass Adv Diff End Region myZEqSet

Coupling Strategy Trade-Space Example • LES-based buoyant plume nonlinear convergence comparison within a time step • In some cases, segregated schemes are extremely inefficient at reaching desired strict NL tolerance • However, if strict NL tolerance is not required, then segregated schemes may be significantly faster and have a higher radius of convergence. • Use segregated as preconditioner and then couple to maximize benefit of both?

Fluids – CVFEM, Other Mech - FEM Hybrid Finite Element Method • Question: Can CVFEM pressure stabilization (2nd and 4th order) work favorably in the context of a GFEM discretization (call it HFEM)? • First Test problem: Difficult two dimensional variable density low Mach flow using MMS; dofs: u, v, p and Z; density = f(Z)

X-component of velocity

Density

Velocity vectors colored by Density

Order of Convergence • Second order convergence demonstrated for all of the primitives; difference in absolute errors is negligible

Velocity Mixture Fraction Second Order

10-3

Velocity Mixture Fraction Second Order

10-3

10-4

ln(Loo)

ln(Loo)

10-4

10-5

10-5

0.2

0.4

0.6

0.8

1

ln(h)

Loo error norms for HFEM 4th scheme

0.2

0.4

0.6

0.8

1

ln(h)

Loo error norms for CVFEM 4th scheme

Low Mach Variable Density MMS • Timing Results to reach 1e-15 NL tolerance • Full Newton; iterative linear solver (gmres/ddilut R0=1e-5) – – – –

CVFEM 2nd order; 7 iterations; average Itns 126; 19.1 s HFEM 2nd order; 7 iterations; average Itns 131; 19.3 s CVFEM 4th order; 7 iterations; average Itns 111; 31.9 s HFEM 4th order; 7 iterations; average Itns 126; 34.5 s

• The trend shows that regardless of discretization approaches, the algorithmic behavior is consistent • The new GFEM method out performs standard GFEM stabilization techniques, e.g., PSPG (Hughes et al).

Questions?