2. Structure Of Flame Balls At Low Lewis-number (SOFBALL): Results from the STS-83 & STS- 94 Space Flight Experiments Paul D. Ronney, Ming-Shin Wu and Howard G. Pearlman Department of Aerospace and Mechanical Engineering University of Southern California, Los Angeles, CA 90089 Karen J. Weiland Microgravity Science Division NASA Lewis Research Center, Cleveland, OH 44135 Supported by NASA-Glenn

3. 2 FLAME BALLS - history Zeldovich, 1944: purely diffusion- controlled stationary spherical flames with steady radius (r * ) possible since diffusion equations  2 T &  2 Y = 0 (T = temperature, Y = fuel mass fraction) have solutions for unbounded domain in spherical geometry: n Analogous to steady solution for spherical fuel droplet burning Convection velocity V  0 everywhere + Dynamics dominated by 1/r tail (with r 3 volume effects!) - unlike propagating flame where T ~ e -r n BUT CAN THEY REALLY EXIST??? Most people don’t believe it at first n Buckmaster, 1985; Joulin, 1985: adiabatic flame balls are unstable - maybe they can’t exist???

4. 3 Flame balls - history - continued n Ronney (1990): seemingly stable, stationary flame balls accidentally discovered in drop-tower experiment - MAYBE THEY ARE STABLE??? n Confirmed in parabolic aircraft flights (Ronney et al., 1993) n Only seen at microgravity (µg) (needed for spherical symmetry) n Only seen at low Le, near extinction limits n Buckmaster, Joulin & collaborators: window of stable conditions near extinction limits when radiative loss present & Le is low (richer mixtures: 3-d instability, splitting balls, propagating cellular flame front). Why should leaner mixtures (lower temperatures) have more radiative loss??? n Encouraging results, but stability & properties compromised by short duration of µg in drop towers, low quality of µg in aircraft - SKEPTICS NOT CONVINCED n NEED SPACE EXPERIMENT: long duration, high quality µg

5. 4 Practical value of flame ball studies n Improved understanding of lean combustion u Benefit of lean combustion to efficiency & emission reduction in engines well known, but experience shows lean mixtures lead to misfire & rough operation u Need better models of weak combustion - determine ultimate limits of lean operation u Current H 2 - O 2 chemical kinetic models disagree on flame ball properties, but all give similar S L away from limits - need better chemical models of weakly burning flames u H 2 -O 2 essential building block of hydrocarbon-air chemistry n Stationary spherical flame - simplest interaction of chemistry & transport - test combustion models u Motivated > 20 theoretical papers to date n Spacecraft fire safety - flame balls exist in mixtures outside one-g extinction limits

6. 5 Implementation of space experiment n Structure Of Flame Balls At Low Lewis-number (SOFBALL) experiment u Space Shuttle missions MSL-1 (April 4 - 8, 1997) & MSL-1R (July 1 - 16, 1997), u Combustion Module-1 (CM-1) facility n Test strategy u 4 mixture types - 1 atm H 2 -air (Le ≈ 0.3); 1 atm H 2 -O 2 -CO 2 (Le ≈ 0.2); 1 atm H 2 -O 2 -SF 6 (Le ≈ 0.06); 3 atm H 2 -O 2 -SF 6 (Le ≈ 0.06) u Only mixtures near limit, few flame balls, little or no splitting SOFBALL logo n Experimental apparatus u Combustion vessel - cylinder, 32 cm i.d. x 32 cm length u 15 individual premixed gas bottles u Ignition system - spark with variable gap & energy u Imaging - 2 views, intensified video u Temperature - fine-wire thermocouples, 6 locations u Radiometers (4), chamber pressure, acceleration (3 axes) u Gas chromatograph

7. 6 Summary of results n STS-83 - April 4 - 8, 1997 u Shortened mission, 2 tests performed, both successful n STS-94 - July 1 - 16, 1997 u Limited opportunity to make changes from STS-83 u Full mission, 17 tests performed, 16 successful, 10 reburn attempts, 8 successful n STABLE FLAME BALLS OBSERVED FOR ENTIRE EXPERIMENT DURATION (500 s) IN MOST CASES! n 1 to 9 flame balls n First premixed gas combustion experiment in space n Weakest flames ever burned (≈ 1 Watt/ball) (birthday candle ≈ 50 Watts) First 2 flame ball tests in space. It worked the first time!

8. 7 How many flame balls do you get? Why does the number of flame balls usually increase with increasing fuel concentration? There is a 3-d instability of flame balls that occurs only for mixtures away from the limits (see slide: Flame Ball History: continued). The spark ignition source creates some excess enthalpy (above that of the ambient mixture.) This decreases the “effective” heat loss magnitude from the stable mixtures (near the limit) into the unstable region, causing splitting of flame balls. Eventually this excess enthalpy is smeared out by conduction and lost by radiation. The higher the initial fuel concentration, the more the ignition source decreases the effective heat loss into the unstable region and the longer the flame balls stay there, so more flame balls are formed before dropping back into the stable region. Of course, if the mixture is too rich, it never drops down into the stable region and the result is a continually expanding cellular front rather than stable flame balls. False-color flame ball images

9. 8 Results - surprises - #1 of 4 n Very little buoyancy-induced drift - flame balls survived much longer than expected without drifting into chamber walls n Aircraft µg data indicated drift velocity (V) ≈ (gr * ) 1/2 u Gr = O(10 3 ) - V) ≈ (gr * ) 1/2 - like inviscid bubble rise u In space, flame balls should drift into chamber walls after ≈ 10 min, even at 1 µg, due to this drift n Space experiments: Gr = O(10 -1 ) - creeping flow - apparently need to use viscous relation: u With representative property values u Similar to recent prediction (Joulin et al., submitted) u Not yet verified experimentally u Much lower drift speeds with viscous formula - possibly hours before flame balls would drift into walls n Also - fuel consumption rates (1 - 2 Watts/ball) would allow several hours of burning time in some tests with only 1 ball!

10. 9 Results - surprises - #2 of 4 n When more than one flame ball was produced, the balls always drifted apart, at a continually decreasing rate n Flame balls interact by u (A) warming each other - attractive u (B) depleting each other’s fuel - repulsive u Which is more important? For adiabatic flame balls, the two effects exactly cancel, according to a recent analysis (Buckmaster & Ronney, 1998). For non-adiabatic flame balls, fuel effect wins because thermal effect disappears at large spacings due to radiative loss - and all stable flame balls must be strongly influenced by radiative loss

11. 10 Theory of flame ball mutual repulsion due to competition for fuel - comparison with experiments

12. 11 Results - surprises - #3 of 4 n Radiometer data drastically affected by impulses caused by small VRCS thrusters used to control Orbiter attitude u Temperature data moderately affected u Vibrations (zero integrated impulse) - no effect n Flame balls & their surrounding hot gas fields are very sensitive accelerometers! n Requested & received “free drift” (no thruster firings) during most subsequent tests with superb results Without free driftWith free drift

13. 12 Results - surprises - #3.5 of 4 n Flame balls seem to respond more strongly than ballistically to acceleration impulses, I.e. change in ball velocity ≈ 2 ∫g dt n Consistent with “added mass” effect - maximum possible acceleration of spherical bubble is 2g Results - surprises - #4 of 4 n 2 missions, 26 burn tests, 1 atm & 3 atm, N 2, CO 2, SF 6 diluents, 20x range of thermal diffusivity, 2600x range of Planck mean absorption length, 1 to 9 flame balls, yet Every single flame ball, without exception, produced between 1.0 and 1.8 Watts of radiant power !!!!! WHY???

14. 13 Comparison with computation n Computational model (Wu et al., 1998a, 1998b) u 1-d, spherical, unsteady code (Rogg) u Detailed chemistry, transport, radiation u Isothermal, fixed composition at outer boundary u Study evolution over time to steady state or extinction n Comparision of radius, radiant emission & limit composition u Fair-poor (radius), good (radiation & limit) for H 2 -air (see plots at right - compare predicted r * vis to experimental data points) u Very poor for H 2 -O 2 -CO 2, H 2 - O 2 -SF 6 (next page)

15. 14 Comparison of predicted & measured ball radii n Predictions sensitive to chemical mechanism, especially rate of H + O 2 + H 2 O  HO 2 + H 2 O n Different models of H 2 -O 2 chemistry yield very different predictions of flame ball properties, even though all predict burning velocities of planar H 2 -air flames accurately! n Likelihood of radiation reabsorption in mixtures diluted with CO 2 & SF 6 u Not included in radiation model but L planck,CO 2 ≈ 3.5 cm at 300K; L planck, SF 6 ≈ 0.26 cm at 300K u Reabsorption decreases heat loss, widens flammability limits u Agreement much better when CO 2 & SF 6 radiation ignored! (limit of zero absorption length for CO 2 & SF 6 ) (H 2 O radiation still included; optically thin & can pass through CO 2 & SF 6 ) u Need improved radiation models, including reabsorption effects

16. 15 Conclusions n SOFBALL - dominant factors in flame balls:  (1) Far-field (1/r tail, r 3 volume effects, r 2 /  time constant) Flame ball: a tiny dog wagged by an enormous tail u (2) Radiative heat loss u (3) Radiative reabsorption effects in CO 2, SF 6 u (4) Branching vs. recombination of H + O 2 - since stable flame balls can occur only near extinction limits, they are like a “Wheatstone bridge” for near-limit chemistry n General comments about space experiments u Space experiments are not just extensions of ground-based µg experiments u Expect surprises and be adaptable u µg investigators are quickly spoiled by space experiments F “Data feeding frenzy” by some investigators during STS-94 u Caution when interpreting accelerometer data - frequency range, averaging, integrated vs. peak