Abstract
Our perception of a flame is strongly grounded in gravity’s influence. From our every interaction with fire from the first birthday candles we blew out, we each build an intuitive understanding of how a flame interacts with the hot air rising via buoyant convection. As researchers, our perceptions of how flames respond to our controls are unconsciously biased by this intrinsic buoyant flow.
Introduction
Our perception of a flame is strongly grounded in gravity’s influence. From our every interaction with fire from the first birthday candles we blew out, we each build an intuitive understanding of how a flame interacts with the hot air rising via buoyant convection. As researchers, our perceptions of how flames respond to our controls are unconsciously biased by this intrinsic buoyant flow.
It has been my experience that many renowned researchers come into the microgravity field of study with preconceived notions of how their flames will respond when gravity is removed, only to be flummoxed when something completely unexpected happens. When gravity is removed, suddenly new realms of fluid mechanics, heat transfer, and chemical reactions become accessible, which is one of the really rewarding aspects of studying microgravity combustion. This is a space-age, still evolving field that is rich for study. It also does not hurt that some of the experimental platforms for microgravity combustion are literally out of this world.
Microgravity combustion does have its challenges with diagnostics, however. Diagnostics have to be small and low power due to launch weight and cooling constraints for flight experiments, and for some drop tower tests they must also be able to withstand repeated 65 g impacts and vacuum conditions. Due to these serious constraints, microgravity flame diagnostics have lagged behind laboratory diagnostic systems. One can imagine the laser alignment issues associated with experiment packages that flex as gravity is removed and flex again during the high-gravity impacts. Recent advances in digital cameras have revolutionized flame imaging so that the dimmest blue flames can be captured even with high-speed movies. Many previously “invisible” microgravity flames can now be imaged, such as cool flames.
This section is organized by general topic, with subsections noted in parentheses: gaseous fuel flames (spherical and gas jets), liquid fuel flames (droplet, candles, and pool fires), and solid fuel flames (spherical, rods, and thin and thick flat sheets).
For further study of microgravity combustion, a good resource is Microgravity Combustion: Fire in Free Fall, by Howard D. Ross, ed., Academic Press, 2001. It provides an excellent framework of understanding and a number of color plates of older microgravity combustion graphics.
4.1 Gaseous Fuels
4.1.1 Cellular Instabilities in Premixed Flames
For many combustible mixtures and ambient conditions, instead of a smooth flame surface (Figure 4.1 top), irregular cells can spontaneously develop and grow over the flame surface as the premixed flame propagates outward, as shown in Figure 4.1 middle and bottom.
Figure 4.1 High-speed Schlieren images for spark-ignited spherically expanding flame under three different conditions demonstrating cellular instabilities that developed: (top): Rich H2/O2/N2 mixtures, equivalence ratio φ = 1.5, P = 1 atm, Le > 1; (middle) Rich H2/O2/N2 mixtures, φ = 1.5, P = 5 atm, Le > 1; (bottom) Lean H2/O2/N2 mixtures, φ = 0.4, P = 2 atm, Le < 1.
The presence of these cells increases the total (wrinkled) flame surface area and consequently the global flame propagation speed. Furthermore, with the continuous generation of these cells, the increase in the propagation speed can be accelerative, leading to the hypothesized scenarios that a wrinkled laminar flame can first transition to a turbulent flame, and eventually to a detonation wave.
There are two instability mechanisms that are intrinsic to the propagation of laminar premixed flames, namely the hydrodynamics, Darrieus-Landau (DL), instability, and the diffusional-thermal (DT) instability.
These instabilities are demonstrated in Figure 4.1: in the top row, the flame does not develop wrinkles because the pressure is not high and the deficit reactant, O2, is also a heavy molecule (Le > 1). In the middle row, the flame exhibits DL instability at high pressure, but not the DT instability (Le > 1). In the bottom row, the flame does not exhibit DL instability (P = 2 atm) but does exhibit DT instability because the deficit reactant, H2, is a light molecule (Le < 1).
This work was supported by the National Science Foundation.
4.1.2 Flame Balls
In 1944, Ya. B. Zeldovich predicted the possibility of stationary, steady spherical flames (“flame balls”) occurring in premixed gases that are supported exclusively by diffusion of reactants to a reaction zone and diffusion of thermal energy and combustion products away from this reaction zone. Flame balls are valid equilibrium solutions to the governing conservation equations for any combustible mixture, but they were predicted to be unstable (i.e., they would either expand beyond their equilibrium radius or collapse and extinguish) and thus not physically observable.
Forty years later, seemingly stable flame balls were accidentally discovered in drop tower experiments at NASA-Glenn, but the short test times of drop tower experiments and g-jitter effects in parabolic aircraft flight experiments precluded a definite conclusion regarding their stability. This led to the development of the Structure Of Flame Balls At Low Lewis-number (SOFBALL) experiment that flew on Space Shuttle missions STS-83, STS-94, and STS-107. These experiments confirmed the prediction that radiative losses could stabilize flame balls near extinction limits, but it led to a number of surprising observations including mutual repulsion of adjacent flame balls, their extreme sensitivity to tiny gravitational disturbances (on the order of 10−6 g), and the uniformity of thermal power among stable flame balls (1–2 watts per ball over a wide variety of mixtures, pressures, and number of balls).
Figure 4.2 is a schematic of a flame ball illustrating representative temperature and concentration profiles.
Figure 4.3 is an image of flame balls in a 7.5% H2 – 15% O2 – 77.5% SF6 mixture taken on the STS-107 mission. Width of field of view is 20 cm. (Flame balls appear to be of different sizes because of [1] differing distances from the camera and [2] differing points in their progress toward extinction at sufficiently small radii.)
This work was supported by NASA.
References
4.1.3 Quasi-Steady Microgravity Spherical Ethylene Diffusion Flame
This is a color image of an ethylene–air diffusion flame in microgravity. The pressure was 1.01 bar and the ethylene flow rate was 1.51 mg/s. The blue flame sheet and yellow soot are clearly visible. The image was recorded 1.3 s after ignition in the 2.2 s drop tower at NASA Glenn. The camera was a Nikon D100 digital still camera with a resolution of 6 megapixels. The scale is indicated by the 6-mm spherical burner.
The Flame Design International Space Station flight experiment examines the soot inception and extinction limits of spherical microgravity flames. It seeks to improve the understanding of soot inception and control to enable the optimization of oxygen-enhanced combustion and the “design” of non-premixed flames that are both robust and soot free. Tests are conducted with various concentrations of both the injected fuel (i.e., ethylene or methane) and the oxygen-enhanced atmosphere to determine the role of the flame structure on soot inception. The effects of flow direction will be assessed with inverse spherical flames. Flame design explores whether the stoichiometric mixture fraction can characterize soot and flammability limits for non-premixed flames like the equivalence ratio serves as an indicator of those limits for premixed flames.
This work was supported by NASA.
Reference
4.1.4 Effect of Microgravity on Sooty Co-flow Laminar Diffusion Flames
The study of the influence of microgravity on the sooting behavior of axisymmetric co-flow laminar diffusion flames was performed in the Microgravity Science Glovebox, on board the International Space Station during expedition 29/30. Examples of the differences between 1 g and μg flames are illustrated in the photographs above. Figures 4.5a and 4.5b show images of a pure methane flame stabilized in normal and microgravity, respectively. The fuel nozzle has a diameter of 3.23 mm and is surrounded by a 76 mm × 76 mm square duct co-flow. The average flow velocities were 46 cm/s and 18 cm/s for fuel and co-flow, respectively. The images appear green because of the BG-7 color filter that was added to the imaging setup to balance the RGB signal of the color detector.1
The cross-sectional soot temperature (in Kelvin), shown in Figures 4.5c and 4.5d, was derived from the collected color images employing an Abel deconvolution and using the color ratio pyrometry technique. Soot volume fraction, shown in Figures 4.5e and 4.5f, was then obtained given the measured temperature and by performing an absolute light intensity calibration.
Due to the lack of buoyancy, in μg a taller and wider flame is produced because the reactant mixing is limited and the inward convection is reduced. The sooty region broadens, following the broadening of the flame, and the peak soot volume fraction increases. The enhanced soot production results in increased thermal radiation losses and hence lower flame temperatures. For these flames, the peak soot temperature in μg is shown to be ~200 K lower than its 1 g counterpart.
Compared to the 1 g flame, the μg counterpart is more diffusion-controlled and has a thicker diffusion layers with more soot production in the wings. From 1 g to μg, the peak of soot volume fraction redistributes from the flame centerline to the wings and the soot growth mode is believed to change from an inception-dominated mode to a surface growth–dominated one.
From accompanying numerical simulations it was seen that, due to the density gradients, the hot flow under normal gravity conditions is accelerated up to 200 cm/s at the far downstream location, while the μg maximum velocity is only 90 cm/s (at the upstream location around the tip of the burner, assuming a parabolic flow profile). The reduction in axial velocity causes longer residence times, allows more time for soot particles to grow, and results in enhanced soot volume fraction.
This work was funded by NASA.
Reference
4.1.5 Steady and Pulsed Sooting Gas Jet Diffusion Flames in Microgravity
A fundamental issue in turbulent gas jet diffusion flames is the behavior of large-scale structures that dominate the dynamics of these flames. The research employs a fully modulated fuel injection approach to study both isolated and interacting flame structures. Buoyancy effects are found to be important for these flames, and hence experiments are conducted in microgravity to suppress these effects. The images show ethylene turbulent gas jet diffusion flames in microgravity obtained in the 2.2 s Drop Tower at NASA Glenn. Ethylene is injected at a Reynolds number of 5000 through a 2 cm id nozzle into a chamber filled with a blend of 30% oxygen in nitrogen. Different degrees of interaction between injected flame puffs may be established by varying the injection time for each puff and the time between puff injections.
Figure 4.6 is an image of the sooting, approximately steady-state, diffusion flame. Figure 4.7 shows a sequence of the pulsed case where the fuel is injected for a time of 40 ms. This case studies the dynamics of an isolated flame structure, hence no more fuel is injected until the flame puff is no longer visible. The time between images in Figure 4.7 is 33 ms and the first image is taken approximately 40 ms after injection. The vertical extent of all images is 56 cm. The length of the flame to burnout is related to the ratio of the injection flow rate and radial diffusion rate of oxygen. The reduced burnout length of the pulsed flame indicates better mixing with the surrounding air. The heat release rate and radiative profiles of the flame may be controlled by varying the degree of interaction between the flame structures. The interaction between flame structures also impacts their downstream convection rate.
This research was supported by NASA.
References
4.2 Liquid Fuels
4.2.1 Soot Shell Formation in Microgravity Droplet Combustion
The image sequence shows a free-floated decane droplet in microgravity from deployment (first image) to hot flame extinction (last image). Immediately after ignition a dense soot cloud forms close to the droplet (second image). As time proceeds, the soot shell moves slowly outward, as the Stefan flow from the droplet, which pushes the soot away from the droplet, is opposed by the thermophoretic forces on the soot particles, which push the soot toward the droplet (third and fourth images).
As the soot shell expands, the more uniform dense cloud transitions into discrete aggregate particles, and the soot shell expels a very large aggregate particle that has penetrated through the flame due to its increased drag (fifth and sixth images). Once it penetrates the flame, thermophoretic forces and the Stefan flow are both pushing the soot away from the droplet.
Near the radiative extinction of the hot-flame (the last image in the sequence), the symmetric soot cloud is much less dense (the corresponding color of the flame is very dim blue) and located far from the droplet surface.
When the hot flame extinguishes, the soot shell moves rapidly away from the droplet, as the Stefan force is no longer opposed by the thermophoretic force. In this test, a prolonged period of cool-flame burning of the remaining large droplet follows hot-flame extinction.
The experiments reveal a number of interesting aspects of the dynamics of soot particles.2 These particles normally migrate around the soot-diameter position as they aggregate and eventually move out through the flame. The soot passing through the flame is very evident in the back-lit view because the particle velocity rapidly increases after the particle traverses the reaction zone as a consequence of the suddenly favorable temperature gradient for enhancing the outward thermophoretic velocity, in contrast to the retarding thermophoretic effect for particles inside the flame.
The lower gas density in the vicinity of the flame also helps increase velocities there. Soot particles, thus, are helpful markers of gas motion once thermophoretic effects are subtracted; temperature gradients can be inferred from thermopherotic effects.
Many aspects of soot production and soot-particle histories require further attention. As with many fundamental scientific investigations, this work has uncovered an appreciable number of additional areas worthy of further investigation.
Details of the Flame Extinguishment Experiment are available in Ref. 1.
This research was supported by NASA.
References
4.2.2 Cool-Flame Supported Microgravity Droplet Combustion
The image sequence shows a free-floating n-dodecane droplet in microgravity from the moment of ignition to cool flame extinction. For each instant in time there is a backlit image of the droplet and an orthogonal image of the flame. The hot-flame images are the CH* chemiluminescence and the cool flame images are the formaldehyde chemiluminescence.
Immediately after ignition there is a small bright hot flame (the igniters are still visible in the droplet view). Shortly after ignition a dense soot shell forms close to the droplet. The soot particles increase in size and the soot shell grows with time. This corresponds with rapid growth of the hot flame. The flame gets continuously dimmer as it grows and eventually extinguishes due to excessive radiative energy loss (relative to combustion heat release). The hot flame burned for approximately 12 s in the test above, and the images are in 4-s intervals.
Following radiative extinction of the hot flame, a cool flame forms. The cool flame is invisible to the naked eye and required the intensified camera with a formaldehyde filter to visualize the flame. The cool flame temperature is much lower than the hot flame and lies closer to the droplet surface. The soot shell present at hot flame extinction rapidly expands when the hot flame extinguishes, quickly growing in diameter and eventually out of the camera field of view.
The cool flame only burns a fraction of the fuel vaporized from the droplet surface. The rest of the fuel vapor transits the cool flame and can collect far outside the cool flame and recondense to form a fuel cloud that is visible in the backlit view as a darkening of the background and also on a color camera view of the test.
The cool flame burns until the droplet reaches a critical size when it extinguishes leaving a small residual droplet. The cool flame in the image sequence burned for approximately 30 s after the hot flame extinguished, and the images are in approximately 10-s intervals.
Details of the experiment are in Ref. 1, and the cool flame discovery is detailed in Ref. 2.
This research was supported by NASA.
4.2.3 Flame Spread over a Randomly Distributed Droplet Cloud Aboard Kibo/ISS
The left three images show the flame spread behavior over an n-decane 2D droplet array in microgravity aboard the Japanese Experiment Module, Kibo, on the International Space Station (ISS). In this test, 97 droplets were randomly distributed at intersections of a 30 × 30 square lattice with 14 µm silicon carbide fibers placed in a combustion chamber. The spacing between fibers was 4 mm. The initial droplet diameter at ignition was about 1 mm.
One droplet was ignited by a hot-wire igniter to initiate the flame spread at atmospheric pressure (first image). The burning behavior was observed by a digital video camera through the window. The flame spread starts from a small spherical flame around a single fuel droplet and then spreads across the lattice of droplets (second image), until finally a yellow large-scale group flame appears (third image). A blue flame propagating a mixture layer around an unburned droplet is also seen near leading flame regions (top left droplet of third image, for example). The fundamental flame-spread mechanism is similar to that of a linear droplet array in microgravity shown in the small fourth image. However, more dynamic flame-spread behavior appears in randomly distributed droplet clouds. The flame-spread behavior and group-flame formation were investigated for different numbers of droplets and the initial droplet diameters.
Percolation theory has been applied to determine the physics that govern the local flame spread between droplets and the flame spread across the droplet cloud, leading to group combustion.
This research will bridge the gap between the combustion of a small number of droplets and spray combustion.
This research was funded by JAXA under the project entitled “Elucidation of flame-spread and group combustion excitation mechanism of randomly distributed droplet clouds (Group Combustion).”
References
4.2.4 Candle Flames in Normal and Microgravity
Figure 4.14 shows that candle flames in normal earth gravity (left) and in microgravity (right) have very different flame shapes, sizes and colors. Figure 4.15 is the numerical simulation of the two cases (with the gravity vector pointed to the left). The visible flames are represented by the fuel reaction rate contour of 5 × 10–5 g/cm3 s. Upper halves of Figures 4.15(a) and (b) are the velocity vectors, and the lower half shows the streamlines and the oxygen mass flux. The normal-gravity flame is elongated due to buoyancy-induced flow of the combustion products as shown in Figure 4.15(b). Oxygen is entrained into the flame reaction zone primarily by convection and to a lesser extent by diffusion. In the microgravity flame, flow is generated by the Stefan flow resulting from wax evaporation. Comparing Figures 4.15(a) and (b), the magnitude of the flow is much smaller in microgravity (notice the different scales between the two images). In microgravity, Figure 4.15(b) indicates oxygen is supplied to the reaction zone entirely by molecular diffusion. Because these rates are slow, the required fuel supply rate is also small. The flame standoff distance from the wick is therefore large compared with that of the normal gravity case as shown in Figure 4.14. Since the microgravity flame has lower burning and heat release rates, the percentage of radiative heat loss from the flame is larger. The resulting lower flame temperature is believed to be the main reason for a visibly soot-free blue flame shown in Figure 4.14. Details of the experiment are in Ref. 1. Computed candle flames as a function of different gravity levels can be found in Ref. 2. Near-limit candle oscillations were reported in Refs. 1 and 3. Modeling of heat and mass transfer inside the porous wick and wick-trimming effect can be found in Ref. 4.
