Howard D. Ross

Detailed experiments of flame spread across deep butanol pools

Experiments using simultaneous rainbow schlieren deflectometry, particle image velocimetry (PIV), infrated (IR) thermography, standard flame imaging, and thermocouples are conducted to reveal detailed physics of low-speed, forced-opposed-flow flame spread over liquid pools. Our effort concentrates on the effects of gravity and air flow speed on the flame-spread behavior across a deep, rectangular tray filled with 1-butanol. The microgravity ( μg ) part of the experiment is the first combustion experiment conducted aboard a sounding rocket. Several novel observations are made. Even with forced-air velocities of the same order of magnitude as that induced naturally by buoyancy in normal gravity, the μg flame behavior is completely different. Whereas the normal gravity (1 g ) flames we studied are soot-producing, and pulsate and spread rapidly, the μg flames are free of soot and spread steadily and very slowly. The rainbow schlieren measurements show that heat penetrates far deeper into the pool in μg , suggesting that the major effect of liquid-phase buoyancy is its stratification of the temperature field in 1 g : its absence in μg may lead to a very different liquid-phase surface temperature, flow field, and flame-spread character. The IR thermography reveals hetetofore unobserved liquid surface temperature and side-flow phenomena in both 1 g and μg . PIV has been used for the first time to obtain quantitative, full-field, liquid-phase velocity fields ahead of the flame, revealing significantly more surface flow in μg than in 1 g . A state-of-the-art models predictions are qualitatively confirmed in regard to gravitational effects on flame shape, flame extinction in μg when there is no opposed air flow, and fuel consumption rate. Disagreement is found in the flame spread character in μg , unless hot gas expansion is artificially set to zero.

Smoke visualization of the gas-phase flow during flame spread across a liquid pool

Gas-phase flow ahead of the leading edge of a flame spreading across a pood of flammable liquid is an important heat and mass transfer mechanism that has not been explored in depth experimentally. Detailed knowledge of the flow field near the flame front is necessary for an understanding of how the flame spreads and for use by model developers as verification of predicted flame characteristics and flow patterns. In this paper, results are presented of gas-phase flow visualization obtained by releasing smoke ahead of a flame spreading across 1-butanol into a slow opposed air flow. The smoke filaments were illuminated with two orthogonal laser light sheets and show clearly for the first time the extent of the hypothesized gas-phase recirculation cell in front of the flame, and the lateral flow divergence due to thermal expansion. The cell forms during the crawling portion of the pulsation cycle in normal gravity, and in the quasi-steady regime in microgravity. Comparisons are made to a numerical model for similar cases of flame spread over alcohols. Despite significant differences between the flame spread rates and character in normal and microgravity predicted by the model and seen in the experiment, only a moderate difference between the flow patterns was seen experimentally for the two gravity levels. This is contrary to an earlier hypothesis that greater flow divergence would be measured in low gravity in the absence of a byoyant flow directed toward the flame and challenges the notion that thermal expansion should be reduced in 2-D numerical models to account for 3-D effects. Instead, buoyancy appeared to have its greatest influence on the flow pattern in the trailing portion of the flame. Corroborating previous experimental evidence from liquid-phase diagnostics, flame spread over a liquid in a narrow tray is found to be a 3-D phenomenon.

Flame spread across liquid pools with very low-speed opposed or concurrent airflow

The consequences of forced airflow on flame spread over liquids is to a large extentunexplored, especially for low air speeds on the order of that due to buoyant convection driven by the flame. Yet small gas flows can influence heat and mass transfer ahead of the flame and alter its spreading characteristics. In this paper we present results from normal and microgravity experiments on flame spread over 1-butanol with either an opposed or a co-current forced airflow ranging from 5 to 30 cm/s. A small flow duct was used to control the air flowing over a laboratory-scale fuel tray. Video flame imaging was used to determine the flame shapes and spread rates, and Rainbow Schlieren Deflectometry (RSD) and infrared thermography of the liquid surface were employed to measure the liquid thermal fields during spread. Large differences in the liquid temperature field were noted depending on the flow direction, and for some concurrent flow rates, the initial flame spread rate was affected. Slow opposed airflows serve mainly to alter the flame pulsation frequency, while small concurrent flows eliminate the pulsations in agreement with numerical modeling that indicates a gas-phase recirculation cell is necessary for flame pulsations. In microgravity the flame spreads slowly and uniformly for higher opposed velocities, and a lower bound of between 10 and 20 cm/s is found below which the flame begins to spread and then extinguishes.

Further observations of flame spread over laboratory-scale alcohol pools

Results are presented for flame spread rate and behavior for alcohol pools in normal and microgravity. The normal gravity experiments concentrated on depth effects in the uniform and pulsating regimes for pool depths of 2, 5, and 10 mm. Extra care was taken in achieving a very uniform initial fuel temperature, and a new video/PC-based flame tracking program was introduced that allowed very detailed graphs of flame position vs. time to be generated. A linear dependence of the average spread rate with pool depth was found in the pulsating regime, with a less-than-linear dependence in the uniform regime. Furthermore, the results show the pulsation wavelength is strongly affected by the pool depth, but not by the temperature. Possible mechanisms for this are discussed. In microgravity we extended earlier experiments to deep pools and examined the extinction limits of flame spread with and without an opposed, forced air flow. The previously reported correspondence between pulsating spread in normal gravity and extinction in a quiescent, microgravity environment for shallow, axisymmetric pools was upheld for deep, linear pools. However, with the imposition of a forced, opposed flow, the flame is sustained, thus opposed flow lowers the limiting oxygen index for pools in microgravity.

Gravitational effects on flame spread through non-homogeneous gas layers

Flame propagation through non-uniformly premixed gases occurs in several common combustion situations. Compared with the more usual limiting cases of diffusion or uniformly premixed flames, the practical concern of non-uniform premixed gas flame spread has received scant attention, especially regarding the potential role of gravity. This research examines a system in which a fuel concentration gradient exists normal to the direction of flame propagation and parallel with the gravitational vector. This paper presents experimental and numerical results for flame spread through alcohol/air layers formed by diffusive evaporation of liquid fuel at temperatures between the flash-point temperature and the stoichiometric temperature. A gallery, which had either the top and/or one end open to maintain constant pressure, surrounded the test section. The numerical simulations and experiments conducted include normal and microgravity cases. An interferometer was used, in normal gravity only, to determine the initial fuel layer thickness and fuel concentration distribution before and during flame spread. Both the model and experimental results show that the absence of gravity results in a faster spreading flame, by as much as 80% depending on conditions. This is the opposite effect to that predicted by an independent model reported earlier in this symposium series. Determination of the flame height showed that the flame was taller in microgravity, an effect also seen in the results of the numerical model reported here. Having a gallery lid results in faster flame spread, an effect more pronounced at normal gravity, demonstrating the importance of enclosure geometry. The interferometry and numerical model both indicated a redistribution of fuel vapor ahead of the flame. Numerical simulations show that, despite the rapid flame spread in these systems, the presence of gravity strongly affects the overall flow field in the gallery.

Parametric investigations of pulsating flame spread across 1-butanol pools

The experimental realization of pulsating flame spread across a flammable liquid in microgravity was accomplished for the first time through systematic tests in a forced, opposed flow of oxygen-enriched air across shallow and intermediate-depth pools. In tests with the deeper pools, the sequential transition through all three subflash flame spread regimes—from pseudo-uniform to pulsating to uniform spread behavior—was achieved. Normal gravity tests were performed for many of the same conditions and showed similar behavior. In addition, agreement between a detailed numerical model and experiments in both normal gravity and in microgravity was newly obtained through changes in the mass diffusivity and the addition of a heat loss parameter in the two-dimensional model of flame spread over a subflash pool of 1-butanol. The model now uniquely captures the differences in flame spread character in going from normal to microgravity and provides quantitative agreement in flame spread rate and surface temperature at either gravity level. An earlier hypothesis that lateral thermal expansion in the gas-phase accounted for the initial discrepancy between the two-dimensional model and the three-dimensional experiment is no longer supported: this is because experiments designed to directly eliminate lateral expansion resulted in flame extinction rather than the desired pulsation.

Computational predictions of flame spread over alcohol pools

The effects of buoyancy and thermocapillarity on pulsating and uniform flame spread above n-propanol fuel pools have been studied using a numerical model. Data obtained indicate that the existence of pulsating flame spread is dependent upon the formation of a gas-phase recirculation cell which entrains evaporating fuel vapor in front of the leading edge of the flame. The size of the recirculation cell which is affected by the extent of liquid motion ahead of the flame, is shown to dictate whether flame spread is uniform or pulsating. The amplitude and period of the flame pulsations are found to be proportional to the maximum extent of the flow head. Under conditions considered, liquid motion was not affected appreciably by buoyancy. Horizontal convection in the liquid is the dominant mechanism for transporting heat ahead of the flame for both the pulsating and uniform regimes.