Cool Flames

ABSTRACT: 

Our work has explored the role of convection and diffusive fluxes of heat and species on the stability and spatio-temporal evolution of low-temperature reactions, cool flame(s), and auto-ignition associated with low-pressure propane-oxygen premixtures in a closed, static reactors.  Natural convection plays an important role in all terrestrial, Lunar, and Martian-based, unstirred, static reactor cool flame and low-temperature auto-ignitions.  At near-zero gravity, the effects of convection are suppressed.  To systematically vary the effects of convection without varying the mixture stoichiometry, reactor pressure, or vessel size, we have studied cool flames experimentally in a closed, unstirred, static reactor subject to different gravitational accelerations (Terrestrial - 1g, Martian - 0.38g, Lunar - 0.16g, and reduced-gravity - ~10^–2g).  NASA's KC-135A reduced-gravity aircraft can vary the effective gravitational acceleration (g) from near-zero to terrestrial levels.  During free-fall and tailored parabolic flight profiles, we record temperature and pressure histories in addition to intensified video imaging during the available flight times. 

Numerical studies use a modified Gray-Yang model augmented with diffusive fluxes of heat and species.  The model captures modes of slow reaction, cool flames, and multi-stage ignition at g-levels between (and including) 0g and 1g.  These numerical predictions are then compared qualitatively with changes in the empirical induction period(s), pressure history, and temperature and species concentration distributions.  The trends agree well on a qualitative basis.  Additionally, the incompressible assumption limits the range of validity of the numerical results to cases with weak temperature rises.

RESULTS:

Summary of Experimental Results

With the first ever imagery of cool flames at low gravity, the effects of convection are clear.  Pressure and temperature histories show the increase of gravitational acceleration increases the induction time, dampens the pressure amplitude, and reduces the initial temperature excursion.  The development of empirical ignition diagrams at 10^-2g and 1g provide baseline data for model refinement and validation.

Representative results presented show the evolution of the visible light emission using an equimolar n-butane:oxygen premixture at temperatures ranging from 320-350^oC (593-623K) for sub-atmospheric pressures.

Summary of Numerical Results

A two-dimensional, axi-symmetric semi-circular domain was used to model the experimental reactor.  A skeletal chemical kinetic Gray-Yang model developed previously for a one-dimensional, reactive-diffusive system by Fairlie and co-workers is extended in the two dimensional domain.  The model captures the effect of buoyant convection and diffusion on the induction time(s), temperature and pressure histories.  Temperature versus time plots show oscillations as the rate of heat generation and heat loss oscillate.  The amplitude of the temperature excursion and subsequent dampening of the oscillations is observed at all g-levels.  The onset and direction of the induced recirculating flow at partial and Earth gravities is predicted and shown to be oscillatory as well.

PRESENT AND FUTURE WORK:

We recently completed quantification of cool flame propagation speeds associated with low-temperature propane oxidation [Foster and Pearlman, 2006].  A low, signal-to-noise ratio of the recorded images limited the resolution of the experimental results, yet the results are adequate to compare with numerical model predictions of Fairlie and co-workers [2005].  Interestingly, the model predicts the radial progression of the flame from the center outwards followed by a contraction of the flame after it approaches the wall.  Computed differences in the light emission profiles between the experimental and numerical results are perhaps due to limitations in the low-light sensitivity of the ICCD camera, emission from species other than excited formaldehyde, as well as limitations of the reduced model.

Additional work will be done to determine the temperature, pressure, and mixture composition dependence on the flame propagation speed.  Improved techniques to reduce the noise in addition to line-of-sight corrections with use of the Abel transform will be pursued.  To determine if cool flame experiments conducted in a larger diameter vessel (> 10 cm) will achieve a steady propagation speed, numerical analyses similar to those conducted by Fairlie and co-workers [2005] in larger vessels will be conducted.

Also, a Gas Chromatography (GC) study will be conducted to determine the bulk stable species  concentrations in 1g studies which will then be compared with those reported in prior literature.  Note that all of the present work has been conducted at sub-atmospheric initial pressure conditions.  Since the GC sample must be at least atmospheric pressure in order to purge the column, the post-reaction gases will be pressurized with a pump or a diluent will be added to increase the pressure.  Bulk measurements will be made of the following species: propane, n‑butane, CO, CO₂, H₂, O₂, CH₄, and H₂O in the post-reaction gases for select tests.  In addition, analysis of a sample of the reactant mixture will be taken to quantify the initial mixture composition.

With respect to the velocity field, Particle Imaging Velocimetry (PIV) will be conducted in the laboratory to quantify the two-dimensional velocity field.  Note that no prior experimental data has been reported on the internal convective flows in a static, unstirred reactor.  Additionally, the PIV results will confirm that the initial gas flow is quiescent in a small fraction of the induction period.

The role of diffusive transport on low and intermediate temperature chemistry has also been recently explored using an existing four-step n-heptane mechanism, previously tuned for elevated pressures [Muller et al., 1992].  The energy and species equations were augmented with diffusive fluxes for heat and species and solved numerically for the spatio-temporal temperature and species concentration distributions.  The ignition delay time was also tabulated and compared with that associated with a zero-dimensional case without diffusive transport.  Further work will be done to explore the validity of the heptane mechanism at lower pressures by comparison of the results with those predicted with detailed n-heptane mechanisms (e.g., EXGAS; Curran et al., 1997).  Additionally, due to significant interest in elevated pressure studies, the model will also be extended to include natural convection.  The nondimensionalized equations will be run with Le <, =, and > 1 and the effect of Le of the flame radii, flame propagation speed, flame quenching, and flame enhancement will be explored.

Lastly, the spectral emission of propane cool flames will be measured.  These results will validate the flame imaging methods used in the current study.  A sufficiently sensitive spectrographic camera or spectrometer with long-integration times will be necessary due the extremely weak light intensity.

Fairlie, R., Griffiths, J. F., and H. Pearlman. (2000) “A numerical study of cool flame development under microgravity conditions.” Proc. of the Combustion Institute 28: 1693-1699.

Fairlie, R., Griffiths, J. F., Hughes, K. J., and H. Pearlman. (2005) “Cool flames in space: experimental and numerical studies of propane oxidation.” Proc. of the Combustion Institute 30: 1057-1064.

Müller, U. C., Peters, N., and A. Liñan. (1992) “Global kinetics for n-heptane ignition at high pressures.” Proc. of the Combustion Institute 24: 777-784.