Gravitational Effects on Flame Spread through Non-Uniform Mixtures
A Research Project Sponsored by NASA Microgravity Combustion Science

Principal Investigators

Fletcher Miller and John Easton 

National Center for Microgravity Research

NASA Glenn Research Center

Cleveland, OH

Anthony J. Marchese, Ph.D.
Associate Professor
Department of Mechanical Engineering
College of Engineering
Rowan University
201 Mullica Hill Road
Glassboro, NJ 08028-1701

Office: 235 Rowan Hall
Email address: marchese@rowan.edu
Telephone: (856) 256-5343
Fax:              (856) 256-5241

Project Summary

In 2000, NASA Microgravity Combustion Science awarded a team of researchers at NASA, Rowan University and Akron University a grant to continue to examine the gravitational influence on flame propagation and structure in non-homogeneous, premixed fuel-oxidizer systems.  Often referred to as “layered” mixtures in the literature, they are encountered in a wide variety of fire-hazardous situations such as chemical spills, underground mining operations, building fires, and automobile and aircraft crashes.  The flames in such systems have been found to spread 2-3 times faster than the laminar burning velocity of premixed gases due to the flows induced by the flame itself, to push fuel vapor and combust beyond obstacles and in areas that initially contained no vapor, and to burn in some regions outside the traditionally defined lean limit.  The presence of such flames is well-documented in terrestrial accidents and is a likely fire scenario on board space craft.

Furthermore, the study of these flames, which are neither purely premixed nor diffusion flames, is of fundamental importance.  Textbooks generally focus on these two classic limiting cases, perhaps because of the relative lack of data for simple situations in which flames span a range of fuel concentrations.  By eliminating buoyancy we will achieve flames that are easier to model and conceptualize, and will achieve more stable, lean conditions than normal gravity would allow.  For uniformly premixed gases, there is considerable effort expended to measure the flammability limits, which are different without buoyancy.  Our goal is to generalize these flammability limits by considering fuel concentration gradient and other factors that may affect non-uniform premixed systems.  To do this, we plan experiments in three different geometries, flame spread perpendicular to the fuel concentration gradient along a surface, flame spread perpendicular to the fuel concentration gradient in free space, and flame propagation parallel to the fuel gradient.  Previous research by us has examined the first case for the first time in microgravity.  For the second, we have identified no published research on flames propagating through non-uniform, lean mixtures far from surfaces either in normal or microgravity.  For the case of parallel flame spread, there has been little work (compared to premixed systems) done at 1g by others, and none in 0g.

In the previous grant period (1996-2000), we studied for the first time the spread of flames through non-homogeneous layers in microgravity. Evaporation of an alcohol fuel in normal gravity for a specified period of time formed the fuel layer, and we ignited in either µg or 1g depending on the test.  We constructed both a 1g apparatus and a drop rig and conducted tests with methanol, ethanol, and propanol fuels.  For the 1g studies we developed an interferometry system to measure the fuel vapor concentration before ignition and as the flame spread.  The basic results of the studies to date show that there is a slight increase in the flame spread rate in microgravity as compared to 1g, especially for flames in galleries with no top in which buoyancy can act most strongly.  In agreement with other researchers, the flame speed varied with the initial fuel temperature; surprisingly, it varied relatively little with the length of the diffusion time (i.e., the thickness of the layer). The interferometric measurements showed the redistribution of fuel vapor in 1g, and also revealed the flame burning outside the lean flammability limit.  We successfully numerically modeled the flame spread rate, and predicted the increase in speed in µg, which seemed to be due to an increase of flame-induced velocity in the forward direction.  These are the first detailed numerical results obtained for layered systems.

There are some unanswered questions from the first grant that we propose to address in this work, as well as extend our research into new areas.  First, it seems likely from the experimentally measured rapid flame speeds, as well as from the numerical model, that the flame-induced velocity field is dramatically increasing the flame spread rate.  We propose to measure this velocity field by using particle image velocimetry, and to conduct a detailed comparison with the numerical model for both 1g and µg conditions.  Velocity data do not currently exist in the literature for lean, layered flames.

Second, we will extend a spectroscopic technique that can be used in the drop tower as well as in 1g to measure methanol concentration.  This will verify the 1g results obtained with the interferometer that the flame burns beyond the lean limit in 1g, and permit µg data to be obtained.  The spectroscopy of methanol is unknown to the degree currently needed for these measurements, and the proper absorption line determination will allow measurements in other combustion systems such as fuel sprays.  The University of Akron will pursue fundamental measurements of the methanol spectrum and use a chemometric technique to determine the optimal wavelengths.  We will then purchase and install the proper laser diode in an existing spectroscopic apparatus.

Third, we plan to study two new flame spread situations, that of a flame propagating along a fuel layer far from a surface, and that of flame spread parallel to the fuel gradient.  In normal gravity fuel plumes tend to either rise to the ceiling or sink to the floor due to buoyancy, but in µg such a plume may remain far from walls and result in a new type of flame.  Without the heat loss to a surface, the flame may spread at lower concentrations and represent a greater fire hazard.  Flames spreading parallel to fuel gradients have been seldom studied at 1g, and not at all in 0g, despite their application to fire safety. 

In all of the cases above, the experiments will be conducted in existing apparatus, with relatively minor changes.  We will measure the flame speeds by video tracking, and the concentrations using the interferometer and laser absorption spectroscopy.  For all of these cases numerical modeling will be used to aid in choosing experimental conditions and in interpreting the experimental results.  Rowan University will use the numerical model from the first grant period as a basis for the new geometries to be studied here.  It is a two-dimensional, full Navier-Stokes model with finite-rate chemistry and variable thermophysical properties.

This research will provide data on flame spread applicable to possible fire scenarios on board spacecraft and it will also provide fundamental data on lean flames burning in concentration gradients that has significance to combustors and fire hazards on earth.  It directly supports the long-term HEDS goals 1, 4, and 5.  Using gravity as an independent variable we will compare the results to our numerical model.  Fire spread in non-homogeneous systems is a realistic case, and is especially relevant to cases were an inert diluent gas is used to suppress the fire.  Finally, studies of methanol detection and flame spread are valuable when considering using methanol as a practical fuel such as in automotive applications.

Anthony Marchese, PI

Fred Hovermann,'01,'03, Graduate Student, ME

Fred Hovermann's Master's Thesis

Highlights from Fred's Defense

Marcos Villa-Gonzalez, '04,'05

Picture of Propagating Ethanol Edge Flame from Hovermann's Research 

Picture of Propagating Ethane Edge Flame (Side and Top View) from Villa-Gonzalez' Research

Last updated: October 13, 2005