Premixed-gas flames - Paul Ronney - University of Southern California
Transcription
Premixed-gas flames - Paul Ronney - University of Southern California
Premixed-gas flames Paul D. Ronney Department of Aerospace and Mechanical Engineering University of Southern California, Los Angeles, CA 90089-1453 USA [email protected] Keywords: Microgravity; premixed-gas combustion; radiation; reabsorption; flammability limits; ignition; instability; flame stretch; flame balls; cool flames; turbulence. Reference: Ronney, P. D., “Premixed-Gas Flames,” in: Microgravity Combustion: Fires in Free Fall (H. Ross, Ed.), Academic Press, London, U.K., 2001, pp. 35-82. 1. INTRODUCTION Premixed-gas flames occur in mixtures of fuel, oxidant and inert gases that are intimately mixed on the molecular scale before combustion is initiated. Examples of premixed-gas flames include Bunsen flames, gas appliance stoves and gasoline-fueled internal combustion engines. Accidental explosions that occur in mine shafts and chemical refineries are also premixed-gas flames that undergo a transition to a detonation (a combustion wave propagating at supersonic speeds, driven by a leading gasdynamic shock), sometimes with disastrous consequences. Thus, an understanding of premixed-gas flames is necessary for both energy conversion and fire safety applications. This chapter discusses the studies performed to date and future challenges related to premixed-gas flames at µg. Perhaps the most important property of premixed-gas flames that distinguishes them from non-premixed flames (e.g., gas-jet flames, liquid fuel droplet flames, fire spread over solid fuel beds) is the fact that (with the exception of “flame balls” discussed in section 5.3) in premixed flames the flame front propagates relative to the gas. This is because premixed flames are not constrained to follow a contour of stoichiometric composition, whereas with nonpremixed flames, the fuel and oxidant must mix in stoichiometric proportions before chemical reaction can occur. The propagation speed of the premixed flame with respect to the unburned gases is called the burning velocity, SL. As indicated in recent reviews of µg combustion (Sacksteder, 1990; Law and Faeth, 1994; Anon., 1995; Kono et al., 1996; Ronney, 1998) when SL is low, i.e., comparable to or lower than convection velocities induced by buoyancy or forced flow, gravity may have a significant effect on the burning characteristics of premixed-gas flames. 2. COMPARISON OF TIME SCALES To estimate under what conditions gravity can affect premixed-gas flames, and thus µg experiments might be enlightening, we compare the time scales for chemical reaction (tchem), buoyant convection in inviscid flow (tinv), buoyant convection in viscous flow (tvis), heat loss to tube walls via conduction (tcond) and radiant heat loss from the burned gases (trad). Premixed flame structure is determined by a balance between chemical reaction and diffusion of heat and reactants over a zone of thickness d, thus tchem ≈ d/SL. d in turn can be estimated as a/SL, where a is the thermal diffusivity of the gas, thus t chem ≈ a/SL2. A buoyant transport time scale can be estimated as d/U, where d is a characteristic length scale of the flow and U is the velocity induced by buoyancy. U in turn can be estimated as (gd(Dr/r))1/2, where r is the density and Dr is the density change across the flame front. Since Dr/r ≈ 1 for flames, U ≈ (gd)1/2 and thus the buoyant time scale for inviscid flow becomes d/(gd)1/2 = (d/g)1/2. For flames propagating in tubes or stabilized on a burner, d would be the tube or burner rim diameter. For viscous flow, d cannot be independently specified; instead d ≈ n/U, where n is the kinematic viscosity. Combining the relations d ≈ n/U and U ≈ (gd)1/2 leads to U ≈ (gn)1/3 and thus t vis ≈ d/U ≈ (n/U)/U ≈ (n/g2)1/3. In general a Prandtl number (Pr ≡ n/a) should appear in this estimate, but for gases Pr ≈ 1. tcond can be estimated as the ratio of the flame temperature to the rate of temperature decay due to conductive heat loss, i.e., Tf/(dT/dt) ≈ Tf/(rCph(Tf-T∞)), where Cp is the constant-pressure heat capacity, h the heat loss coefficient in the tube = 16l/d2, l the thermal conductivity, Tf the flame temperature and T ∞ the ambient (wall) temperature, thus t cond ≈ d2/16a, where we have assumed (Tf-T∞)/Tf ≈ 1, which is reasonable for practical flames. t rad for optically-thin gases can be estimated as Tf/(dT/dt) ≈ Tf(L/rCp), where L = 4sap(Tf4 - T ∞4) is the radiative heat loss per unit volume of gas, s the Stefan-Boltzman constant and ap the Planck mean absorption coefficient of the gas, thus t rad ≈ {g/(g-1)}{P/4sap(Tf4 - T ∞4)} where g is the specific heat ratio and P the pressure. For illustrative purposes two sets of time scales are generated, one for near-stoichiometric hydrocarbon-air flames and the other for very weakly burning lean hydrocarbon-air flames near the flammability limits, both at P = 1 atm. For the former case, SL = 40 cm/s, Tf = 2200K, a = n = 1.5 cm2/s and ap = 0.56 m-1. For the latter case, SL = 2 cm/s, Tf = 1500K, a = n = 1.0 cm2/s and ap = 0.83 m-1. For both cases g = 980 cm/sec2, g = 1.35, T∞ = 300K and d = 5 cm (a typical diameter for burner or flame tube experiments.) The estimated time scales for these flames are shown in Table 1. < Table 1 near here > 2 Several observations can be made based on these simple estimates: (1) Buoyant convection is unimportant for near-stoichiometric flames because both t vis and tinv are much larger than tchem (2) Buoyant convection strongly influences near-limit flames at earth gravity because in this case both tvis and tinv are comparable to or smaller than tchem (3) Radiation effects are unimportant at earth gravity because buoyant convection is a much faster process (both tvis and tinv are much smaller than trad) (4) Radiation effects will dominate the behavior of flames with very low SL since t rad and t chem are comparable for the slower flame, but these effects can only be observed at low gravity because of (3) (5) For the representative conditions used here, the apparatus size, e.g., the tube diameter, must be larger than about 2.6 cm if one is to observe radiation-induced extinction, otherwise conduction losses to the tube wall will be comparable to radiative losses. (6) Many phenomena associated with radiative loss effects can be studied in drop tower experiments, with test durations of 2 to 10 s, since these times are typically larger than trad. As a result, combustion science has probably benefited more from the utilization of short-duration drop tower experiments than any other microgravity science discipline. (7) Since tinv ~ g1/2 and tvis ~ g1/3, aircraft-based µg experiments at g ≈ 10-2 go, where go is earth gravity, may not provide sufficient reduction in buoyancy effects to observe radiative effects. (8) Since tvis ~ n1/3 ~ P-1/3 and t rad ~ r/L ~ P1/P1 ~ P0, t vis/trad ~ P-1/3. Thus, the radiative time scale is similar at all pressures, but at high pressures buoyancy effects interfere more strongly with radiative effects. (9) At Reynolds numbers Ud/n ~ (gd3/n2)1/2 ≡ Grd1/2, where Grd is a Grashof number, of the order 103 or larger, thus Grd of the order 106 or larger, buoyant flow at 1g will necessarily be turbulent, thus it is difficult to study steady laminar flames in large systems at 1g (≈ 10 cm for the property values employed in the examples given here). A key aspect of these predictions is that only for mixtures with large t chem and thus low SL will buoyancy effects be significant. Low SL implies mixtures highly diluted with excess fuel, oxidant, or inert gas. It is well known that combustible gases will not burn if sufficiently diluted. The composition delineating flammable from nonflammable mixtures is called the flammability 3 limit. Much of the early µg research on premixed-gas flames was concerned with flammability limits, thus our discussion will begin with this topic. The importance of flammability limits due to radiative losses (tchem ≈ trad) in µg experiments cannot be overstressed because it leads to a new type of limit at low flow velocities and long residence times in addition to the high-velocity, short residence time limits that are well known from earth-based experiments. This dual-limit behavior permeates many of the phenomena discussed below. 3. FLAMMABILITY LIMITS 3.1 Buoyancy effects Despite many years of study, there is still controversy surrounding the mechanisms of flammability limits, including the effects of hydrodynamic strain and flame front curvature (collectively called “flame stretch”), buoyancy, heat losses and flame chemistry. A standardized measurement of flammability limits using a cylindrical tube of 5 cm diameter and 180 cm length, called a Standard Flammability Limit Tube (SFLT), was proposed long ago (Coward and Jones, 1952). The tube is filled with combustible gas, ignited at an end open to the atmosphere and propagates toward a closed tube end. Mixtures are defined as flammable if they sustain flame propagation throughout the tube. The data compiled in Coward and Jones (1952) show that flammability limits are different for flame propagation in the upward, downward and horizontal directions of propagation. This indicates that buoyancy effects play a role in these limits. Apparently all flammability limit studies show that burning velocity at the flammability limit (SL,lim) is nonzero. Computations (Lakshmisha et al., 1988; Giovangigli and Smooke, 1992) have shown that there is no purely chemical flammability limit criterion for planar unstretched flames; without losses SL decreases asymptotically to zero as dilution increases. In Lakshmisha et al. (1988), it was shown that the solutions of the unsteady planar one-dimensional adiabatic premixed flame equations do not predict any flammability limit for lean CH4-O2-N2 mixtures. In Giovangigli and Smooke (1992), the steady version of these equations was solved for CH4-air and H2-air mixtures and a similar conclusion was reached. Both are important findings, since it is possible in principle that unsteady effects could suppress limits of steady flames or cause limits to occur for mixtures that are flammable as steady flames. Together these works show that one must identify loss mechanisms of the appropriate magnitude to explain flammability limits. These computations also show that since d increases as SL decreases, an ever-larger computational domain is needed to model progressively weaker mixtures. If the domain is too small, an apparent extinction limit is observed that is purely a computational artifact. Consequently, loss mechanisms such as those discussed below are needed to explain limit mechanisms. The resulting predictions of SL,lim indicate that usually SL,lim depends only weakly 4 on chemical reaction rate parameters. Thus, limit mechanisms may be inferred by comparing predicted and measured SL,lim without detailed chemical knowledge. The mixture composition at the limit affects SL,lim only weakly through Tf, thus, comparing predicted and measured limit compositions is not especially enlightening; comparisons of SL,lim values is much more useful. Consequently, this discussion will emphasize comparisons of predicted and measured values of SL,lim. For upward propagation, experiments (Levy, 1965) show that at the flammability limit the rise speed of the flame, Sb,lim, is the same as the rate of rise of a hot gas bubble up the tube: (1). Sb, lim = 0.33 gd Equation (1) is consistent with the inviscid rise speed discussed in Section 1. This rise speed is dictated by hydrodynamics alone. It represents a minimum speed for flame propagation, though it does not in itself indicate an extinction mechanism. Subsequently, it has been shown theoretically (Buckmaster and Mikolaitis, 1982a) how the hydrodynamic strain at the tip of a flame (Fig. 1a) rising at this rate could cause extinguishment (see Section 5.2 for further discussion of strain effects). These predictions can be expressed in the form (using temperature averaging of the transport properties) < Figure 1 near here > SL ,lim Ê ga 2 ˆ = 2.8 f Á ˜ Ë d ¯ 1/ 4 Èb Ê 1 ˆ ˘ T ; f ≡ exp Í Á1 - ˜ (1- e )˙ ; e ≡ • Î4Ë ˚ Le ¯ Tf (2), where b = E/RgT f is the non-dimensional activation energy, E is the overall activation energy of the heat-release reactions, Rg is the gas constant and Le the Lewis number defined below. Note that f = 1 when Le = 1. As might be expected, the functional form of Eq. (2), except for the Lewis number effect, can be obtained by setting t inv = t chem. Note that Sb,lim is different from SL,lim; Sb,lim is the observed flame front propagation rate in the laboratory frame whereas SL,lim is the value of SL for a planar steady flame in the mixture at the flammability limit and thus is a property of the mixture (Williams, 1985). Because the rising flame is curved, its area is greater than the cross-sectional area of the tube. Mass conservation dictates that the ratio of the flame area to tube cross-sectional area be equal to the ratio of Sb to the mean SL averaged over the flame surface. Consequently, in general Sb,lim > SL,lim, and there is no contradiction between Eqs. (1) and (2). 5 In Eq. (2), SL,lim is seen to be strongly dependent on the Lewis number, defined as Le ≡ Thermal diffusivity of the bulk mixture Mass diffusivity of the scarce reactant into the bulk mixture (3) This parameter can be interpreted as the ratio of the rate of diffusion of thermal enthalpy from the flame front to the unburned gas to the rate of diffusion of chemical enthalpy (in the form of scarce reactant) from the unburned gas to the flame front. When Le differs from unity, flame stretch (Section 5.2) causes changes in the rates of transport of chemical and thermal enthalpy that in turn affects the temperature at the flame front. For flame in mixtures with low Le that are concave towards the burned products (as in a rising flame cap in a tube or an expanding spherical flame) the increase in the rate of chemical enthalpy to the flame front is greater than the increase in the rate of thermal energy loss, and thus the curved flame will burn more intensely than a planar flame in the same mixture. Since heat release reactions in most combustible mixtures have high activation energies, these changes in flame front temperature lead to large changes in reaction rate at the flame front and thus large changes in the local, instantaneous propagation rate. For downward propagation, high-g centrifuge experiments (Krivulin et al., 1981) show that SL,lim ~ g1/3. SL,lim is apparently independent of Le, which is reasonable since downwardpropagating near-limit flames in tubes are nearly flat and unstrained, hence SL ≈ Sb. An extinction mechanism is suggested by observations (Jarosinsky et al., 1982) of a sinking of a layer of cooling burned gas near the walls overtakes the flame front and “suffocates” the flame (blocking the flame front from the fresh unburned gases) is an important factor in the extinction processes (Fig. 1b). Detailed numerical computations (Patnaik and Kailasanath, 1992) support this mechanism also, though the experimentally-observed g1/3 scaling was not tested. The g1/3 scaling can be obtained by setting tchem = tvis, leading to SL,lim ≈ (ga)1/3 (4). The a 1/3 scaling has been confirmed by experiments on flames in tubes using a wide range of diluent gases and pressures (Wang and Ronney, 1993), and together with the g1/3 scaling found in centrifuge experiments (Krivulin et al., 1981) support the mechanisms proposed here. Both upward and downward limit mechanisms indicate that as g Æ 0, SL,lim Æ 0 also, implying that arbitrarily weak mixtures could burn at g = 0, albeit very slowly. As discussed in the next section, an additional factor, namely heat losses, prevents arbitrarily weak mixtures from burning even in the absence of gravity. 6 3.2 Radiative heat loss Numerous authors (Williams, 1985; Spalding, 1957; Buckmaster, 1976; Joulin and Clavin, 1976; Jarosinsky, 1983; Aly and Hermance, 1981) have considered how dilution may lead to extinction via heat loss due to radiation or conduction. Increasing dilution decreases the flame temperature, which in turn increases tchem more than the characteristic loss time scale since the former is generally a stronger function of temperature than the latter, i.e., exponential vs. algebraic. Consequently, increasing dilution increases the impact of heat losses, which eventually leads to a flammability limit. An estimate of SL,lim for radiative losses has been given in Williams (1985), that, after temperature-averaging of transport properties, becomes approximately SL ,lim = 1.2bLl f 1 r • Cp Tf (5), where lf is the thermal conductivity evaluated at T = T f and L is the rate of radiative heat loss per unit volume of gas discussed in Section 2. The functional form of Eq. (5) can be obtained by equating tchem and trad. Using the same set of property values as employed in section 2, Eq. (5) yields SL,lim ≈ 2.3 cm/s for lean-limit hydrocarbon-air mixtures at 1 atm, which is practically identical to that predicted using the same gas radiation data (Hubbard and Tien, 1978) but a detailed numerical model of chemistry and transport (Lakshmisha et al, 1990). Such small values of SL,lim are not usually observed experimentally; note that Eq. (2) yields SL,lim ≈ 3.3 cm/s for upward propagation of CH4-air flames and Eq. (3) yields SL,lim ≈ 7.8 cm/s for downward propagation. Thus, radiation effects do not dominate at earth gravity because of buoyant convection (tinv < trad and tvis < trad), however, at reduced gravity in sufficiently large diameter tubes, radiation effects can be anticipated to be significant. It is important to note that Eqs. (2), (4) and (5) apply only if the apparatus size is large enough that conductive heat losses to the walls of the apparatus are not significant. Flame quenching via conduction losses occurs when tchem ≈ tcond, leading to SL ~ a/d or (Spalding, 1957; Joulin and Clavin, 1976). Pelim ª 40; Pelim ≡ SL ,lim d a• (6), where Pelim is the Peclet number at the flammability limit and a ∞ is the thermal conductivity evaluated at T = T∞. Thus, for flames in small-diameter tubes or in gases with large a (e.g., low pressures and light inert gases such as helium), even at µg it may be difficult to observe a limit due to radiative loss rather than one due to conductive loss. 7 3.3 Microgravity experiments The earliest study of premixed-gas flames at µg is probably that dating back to 1980 (Krivulin et al., 1980), in which an aircraft flying low-gravity trajectories was employed to study the effects of buoyancy on flammability limits. The limits for lean H2-air and rich C3H8-air mixtures in a 20 liter cylinder of equal diameter and length, ignited by a 17J spark at the center of the chamber (that produced spherically expanding flames), were found to be 7.0% for H2 / 8.6% C3H8 at µg (for propagation throughout the chamber), compared to about 4.0% H2 / 9.9% C3H8 for upward propagation at 1g and 8.5% H2 and 8.0% C3H8 for downward propagation. Thus, with the definitions employed the limits at µg were found to lie between those for upward and downward propagation at 1g. Surprisingly, these authors did not report SEFs (Section 5.1) for the rich C3H8-air mixtures (Le ≈ 0.87) nor flame balls (Section 5.3) for the lean H2-air mixtures (Le ≈ 0.3), both of which were readily observed in later µg experiments employing a similar apparatus. At about the same time, lean CH4-air mixtures were studied (Strehlow and Reuss, 1981) in a SFLT using a 2.2 second drop tower facility. A 5.22% CH4 mixtures was determined to be probably flammable whereas the flammability of a 5.10% CH4 was uncertain because it had not extinguished nor reached a steady SL within the µg time available. The corresponding values of SL for the experiments reported in Stehlow and Reuss (1981) have been inferred (Ronney and Wachman, 1985) and are given in Table 2. Using a 12 liter cylindrical chamber with ignition by a 5 J spark at the center of the chamber, limit mixture compositions and near-limit values of SL have been measured (Ronney and Wachman, 1985) (Table 2) that are very close to the SFLT results (Strehlow and Reuss, 1981), suggesting that at least for lean CH4-air flames, µg may provide a means to obtain a fundamental flammability limit, i.e., one that is independent of the apparatus. Moreover, the value of SL,lim in both cases is close to the theoretical and computational prediction of about 2 cm/s as discussed in the previous section. However, it should be noted that the SFLT results may be somewhat fortuitous, for in 5 cm tubes, the estimated SL,lim due to conduction losses (Eq. 6) is 1.6 cm/s, which is comparable to the estimated SL,lim due to radiative losses (Eq. 5) of 2.3 cm/s. Thus, conductive losses to the tube wall and radiative losses were probably of nearly equal importance in the SFLT experiment, whereas in the expanding spherical flame experiment (Ronney and Wachman, 1985), conductive losses were negligible. < Table 2 near here > 8 The aforementioned SFLT experiments (Strehlow and Reuss, 1981) were later extended Strehlow et al., 1986) using an aircraft to obtain longer µg durations. The flammability limits for 1g upward, µg and 1g downward propagation were 5.25%, 5.25% and 5.85%, respectively, for CH4-air mixtures and 2.15%, 2.06% and 2.20% for C3H8-air mixtures. The corresponding values for spherically expanding C3H8-air flames at µg are 2.02%, 2.09% and 2.07% (Ronney, 1988a). Thus, again the µg limits are very similar in different apparatuses whereas the 1g limits are different, indicating again that µg provides the closest approximation to an apparatusindependent limit. Strehlow et al. (1986) reported that the 1g upward limit was due to stretch at the flame tip as discussed in Section 3.1, whereas both the µg and 1g downward limits were said to be due to heat losses to the tube wall. These claims are difficult to evaluate because values of SL,lim were not reported, but clearly since the downward limit occurs at a higher fuel concentration than the µg limit, SL,lim must be higher than the value corresponding to Pe = 40, and thus it is not a true quenching limit. Instead, the heat loss to the tube wall causes the sinking layer of combustion products to form as discussed in Section 3.1 which leads to extinction, but in itself this heat loss does not cause extinction in the manner described by the analysis leading to Eq. (6). Finally, the predictions of Eq. (5) have been compared to experiments at µg (Ronney, 1988b; Abbud-Madrid and Ronney, 1990) and good agreement is found for a wide range of pressures, fuels, and inert gases when Le not so high nor pressure so low that an ignition limit rather than a flammability limit is reached (because minimum ignition energies increase as Le increases or pressure decreases.) Thus radiative heat loss appears to be the cause of flammability limits when extrinsic heat losses, e.g., due to conduction, buoyant convection and hydrodynamic strain (Section 5.2) are eliminated. This is one instance where µg experiments have enabled observation of a phenomenon that can probably never be observed at 1g. 3.4 Radiation reabsorption effects The radiation effects described in the previous section are only valid if the gases are “optically thin,” that is, there is no reabsorption of emitted radiation. This probably cannot be true in systems of very large size, at very high pressures, or in mixtures with a high concentration of sufficiently strong absorbers. With this motivation, lean CH4-air mixtures seeded with inert solid particles have been studied (Abbud-Madrid and Ronney, 1993) to increase ap to values sufficiently high that reabsorption effects were observable in a laboratory-scale combustion apparatus. Since solids emit and absorb as black- or gray-bodies, whereas gases radiate in narrow spectral bands, a particle-seeded gas can emit and absorb much more radiation than a particle-free gas. Data reported in Abbud-Madrid and Ronney (1993) on flame shapes, propagation rates, peak pressure, maximum rate of pressure rise, and thermal decay in the burned gases indicated that at low particle loadings, the particles act to increase the radiative loss from the gases, 9 whereas at higher loadings, reabsorption of emitted radiation becomes significant, which in turn acts to decrease the net radiative loss and augment conductive heat transport. For example, for 5.25% CH4-air mixtures, as the particle loading was increased, SL decreased at first then increased to a value above that of particle-free mixtures (Fig. 2). In a leaner mixture (5.15% CH4) at particle loadings of 0.00 g and 0.75g, the burning velocities were 1.70 and 1.30 cm/s, respectively. The latter value is noteworthy because it is lower (by about 15%) than any value attainable in particle-free lean CH4-air mixtures at 1 atm, which, according to Eq. (5), indicates that the net radiative loss is lower than in any particle-free mixture. In principle, SL,lim may be reduced to zero in optically-thick gases, though only minor decreases in SL,lim have been seen in experiments performed to date. Based on these trends, one could speculate that for apparatuses in which the absorption length (i.e., the inverse of the absorption coefficient ap) is much smaller than the system size, flammability limits might not exist at µg conditions because emitted radiation would not constitute a loss mechanism. < Figure 2 near here > The maximum increase in SL attainable through radiation effects is related to the ratio of the blackbody emissive power of the particle-laded gas per unit area, s(Tf4 - T ∞4), to the total enthalpy flux per unit area of the particle-free flame, r ∞CpSLT f. The ratio of SL with radiative reabsorption to that in the particle-free mixture (m) is given by (Joulin and Deshaies, 1986) 4 4 1 - w s (Tf - T• ) B m = exp m ; B ≡ b g r •C p SLTf ( ) (7) where B is the Boltzman number and the symbols w and g indicate, respectively, albedo (scattering to attenuation ratio), and a constant ( 3 < g < 2, depending on w). Equation (7) shows that away from flammability limits, where SL is large, B will be small and thus m will be close to unity, indicating that radiative effects do not affect SL substantially. The predictions of Eq. (7) have not been tested experimentally to date, though Eq. (7) is consistent with detailed numerical computations (Ju et al., 1998a). When particles are used to decrease ap, the heat capacity of the particles must be also be considered since this reduces Tf. Adding particles to the gas might enable one to suppress flammability limits, but only if the heat capacity of the particles were low enough to avoid a significant decrease in Tf. Even for spectrally-radiating gases, computations (Ju et al., 1998a) using a detailed statistical narrow-band radiation model show that flammability limits may be extended remarkably by considering reabsorption, for example in CH4-O2-N2 mixtures seeded with CO2 10 (Fig. 3). Note that the equivalence ratio f (defined as the fuel to oxygen ratio of the mixture divided by the stoichiometric ratio) at the flammability limit is 0.682 for optically-thin conditions vs. 0.442 with reabsorption. The latter value is even lower than the computed value for CH4-air despite the fact that CP is much higher for CO2 than N2, and thus the adiabatic T f is lower with CO2. Still, in Ju et al. (1998a) it was found that there are two radiative loss mechanisms that lead to flammability limits even with reabsorption. One is due to the difference in composition between reactants and products; if a product of combustion that radiates significantly is not present in the reactants (e.g., H2O), radiation from this species that is emitted upstream cannot be reabsorbed by the unburned mixture unless, by coincidence, its spectrum overlaps completely with the constituents of the unburned mixture. The second is that the emission spectra of most molecules are broader at flame temperatures than ambient temperature, thus some radiation emitted near the flame front cannot be absorbed by the reactants even if they are seeded with that molecule. Via both mechanisms some net upstream heat loss due to radiation will always occur, leading to extinction of sufficiently weak mixtures. These results suggest that fundamental (domain-independent) flammability limits due to radiative heat loss may exist at µg, but these limits are strongly dependent on the emission-absorption spectra of the reactant and product gases and their temperature dependence, and cannot be predicted using gray-gas or optically-thin model parameters. In fact because of the spectral nature of gas radiation, very significant reabsorption effects were found in domains as small as 1 cm even for mixtures with aP-1 ≈ 24 cm. This is because ap is a mean absorption coefficient weighted by the Planck function and averaged over all wavelengths, whereas for some wavelengths the spectral absorption coefficient is much higher (by orders of magnitude) than the mean. < Figure 3 near here > 4. FLAME INSTABILITIES 4.1 Low Lewis number (cellular) instability It is well known that gravity affects the stability of plane flames through the RayleighTaylor effect, which states that when a dense fluid lies on top of a lighter fluid, gravity is a destabilizing influence on the interface, whereas the opposite configuration is stabilizing. Darrieus (1938) and Landau (1944) showed that, independent of gravity effects, the density decrease across the flame front by itself causes wrinkling to be encouraged because the pressure and density gradients become misaligned if a small perturbation to the flat front occurs, leading to a torque on the fluid that acts to encourage the wrinkling. The combined effect of Rayleigh-Taylor wrinkling (or stabilization) and Darrieus-Landau wrinkling can be written in the form of a 11 dispersion relation for the growth rate of infinitesimal disturbances (s) as a function of the wavenumber of the wrinkle (k) (Pelcé and Clavin, 1982): s e 2 ga f ˆ ˘ 1 È 1 - e 2 ÊÁ ˜ kSL = 1 + e -1 Í 1+ e Ë 1 - kS 3 ¯ - 1˙ L Î ˚ (8). Note that for g = 0 all wavenumbers are unstable (s > 0) and that only for downward propagating flames (g < 0) can some wavenumbers, namely small ones (long wavelength disturbances), be stabilized by gravity. Besides thermal expansion driven and buoyancy driven instabilities, premixed-gas flames are subject to diffusive-thermal instabilities when Le differs from unity (Williams, 1985; Clavin, 1985) due to the imbalance of the diffusion rates of thermal energy and reactants. For Le less than a critical value (Lec ) slightly less than unity, cellular flames occur, which are characterized mathematically by a growth rate that is real (no imaginary component) and maximum at a finite wavenumber. Physically this instability occurs because wrinkled regions that are concave toward the burned products (i.e., pointing toward the unburned gas) have a higher local burning velocity than the flat flame (see the discussion of curvature effects in Sections 3.1 and 5.1) and wrinkled regions pointing toward the burned gas have a lower local burning velocity than the flat flame, thus the wrinkling is encouraged. Conversely, for Le > Lec , wrinkling is discouraged. Since this instability is dependent on diffusional effects that occur on the scale of d, it has no effect on very long wavelengths and at short wavelengths it is so dominant that diffusion damps out all wrinkling. In the absence of buoyancy and thermal expansion, the dispersion relation for adiabatic flames is given by solutions of the relation (Joulin and Clavin, 1979) [ ] A(1 + A) = b ( Le - 1) s (1 + 1 + A ) - A / 2 ; A ≡ 4(s + k 2 ) (9), Inspection of Eq. (9) shows that Lec = 1 - 2/b since only for Le > 1 - 2/b is s > 0 for any value of k. The effects of thermal expansion, buoyancy and diffusive-thermal instabilities are shown schematically in Fig. 4. Diffusive damping suppresses instability at large k (small wavelengths) while buoyancy suppresses instability at small wavenumbers (for downward propagating flames). Thus, the growth rate is maximum at an intermediate wavelength and negative at short and long wavelengths, leading to well-defined cellular structures. Even without buoyancy, s exhibits a maximum, though in this case there is no mechanism to suppress growth entirely at small k. 12 < Figure 4 near here > Cellular structures observed at µg can be expected to be different from those observed at 1g for at several reasons. First, mixtures that can be studied at µg may not be flammable at 1g for the reasons discussed in section 3.1. Also, for upward propagation in a confined tube the resulting hydrodynamic strain (Section 5.2) suppresses much of the fine-scale wrinkling (large k) caused by the diffusive-thermal instability (Rakiv and Sivashinsky, 1987) (though the flame is highly wrinkled on the scale of the tube diameter due to buoyancy effects). These factors may affect the extinction mechanisms significantly because diffusive-thermal instabilities are an important factor in flammability limits. In particular, it has been shown (Joulin and Sivashinsky, 1983) that in low-Le mixtures the wrinkling leads to an extension of flammability over that of planar flames. This extension of flammability reaches its ultimate limit in flame balls, to be discussed in Section 5.3. Despite the rich behavior associated with premixed flame instabilities and the likely influences of buoyancy, very few experimental studies of cellular flames at µg have been conducted. Experiments (Dunsky and Fernandez-Pello, 1990) on the effects of buoyancy on rich C3H8-air flames (Le ≈ 0.8), stabilized above a porous plug through which the combustible gases were emitted, have shown that the flow induced by the plume at 1g had a substantial influence on the overall flame shape and thus the size and shape of the cells, but otherwise the cells were not significantly influenced by buoyancy. At first this seems surprising considering that t chem/tinv = (a/SL2)/(d/g)1/2 ≈ 0.3 when d is based on the observed cell size (typically 1 cm), however, as Dunsky noted, their experiments were performed on flames stabilized above a porous plug that are not free to respond to buoyancy-induced flow to nearly the extent that freely propagating flames can because of heat loss to the plug and the flow-straightening effect of the plug. Very lean H2-air mixtures have been studied (Ronney, 1990) in a 12 liter cylindrical chamber with spark ignition at the center of the chamber and observed cellular patterns in the resulting expanding quasi-spherical flames (Fig. 5). While no quantitative information on cellular flame structure was reported in (Ronney, 1990), it is evident that freely-propagating flames at µg provide a means to study diffusive-thermal instabilities without buoyancy effects and minimal influence of strain, heat loss to burners, etc. < Figure 5 near here > 4.2 High Lewis number (pulsating) instabilities At sufficiently high Le, the diffusive-thermal theory (Eq. (9)) predicts flame fronts with traveling-wave or pulsating characteristics (Re(s) > 0, Im(s) ≠ 0) (Joulin and Clavin, 1979). An 13 important application of this high Le instability is to the combustion of lean mixtures of heavy hydrocarbons such as octane in air (Le ≈ 3), which is relevant to lean-burning automotive engines because diffusive-thermal instabilities affect the wrinkling and thus propagation rates of turbulent premixed flames (Williams, 1985) such as those occurring in automotive engines. The high Le instability has been studied experimentally (Pearlman and Ronney, 1994; Pearlman, 1997) using flames in lean C4H10-O2-He mixtures (Le ≈ 3.0) propagating downward in a tube open at the ignition end and closed at the other end, i.e., a SFLT though with larger diameters. Two types of instabilities were observed at 1g: (1) a rotating spiral wave (Fig. 6a) and (2) a pure radial pulsation (Fig. 6b). The spiral waves occurred only very near the flammability limit, whereas the radial pulsations could occur in these mixtures and also mixtures farther from the limits. For mixtures sufficiently far from the limits, only stable flames were observed. For µg drop tower experiments (Pearlman and Ronney, 1994) the pulsating mode was observed but no spiral flames were seen; instead a flame consisting of six rotating azimuthally-distributed bright and dark striped zones was found (Fig. 6c). < Figure 6 near here > The existence of nonuniform modes of propagation near the limits but not farther from the limits is consistent with theory (Joulin and Clavin, 1979) that indicates a wider range of unstable k and Le exists for mixtures with greater impact of heat loss, i.e., closer to the flammability limits. While no definitive explanation of the observed differences between the 1g and µg flame instabilities has been advanced, the impact of gravity is probably to cause a reduction in the effective heat loss, since theory (Kaper et al., 1987) shows that for the high-Le instabilities, the influence of buoyancy and heat loss on instability can be combined into a single parameter whose value decreases with increasing gravity for downward propagating flames. Thus at 1g, the effective heat loss (from the standpoint of instability behavior) is lower. The stability of various spinning modes of flame propagation having different numbers of stripes. Predictions (Booty et al, 1987) for the parameters corresponding to the experimental conditions at µg is that the first modes to bifurcate from the steadily-propagating solution as the combined buoyancy/heat loss parameter (or Le) is increased correspond to a "four-headed" spinning wave and a "one-headed" spinning wave, which are stable modes. Further increases in the buoyancy/heat loss parameter indicate that a six-headed mode may appear. Since the buoyancy/heat loss parameter is higher at µg than 1g for downward propagating flames, the µg flame is farther into the unstable regime and thus might exhibit the higher-order instability mode, i.e., the six-headed mode. 14 5. STRETCHED AND CURVED FLAMES 5.1 Self-Extinguishing Flames Premixed flames are generally not flat and steady nor do they commonly propagate into a quiescent flow. Consequently, flames are usually subject to “flame stretch” (S) defined by (Williams, 1985) S≡ 1 dA A dt (10) where A is the instantaneous flame surface area. The effects of flame stretch on propagation rates and extinction conditions are discussed in various reviews (Williams, 1985; Clavin, 1985). For flames at 1g, buoyancy imposes a flame stretch comparable to tinv-1 or tvis-1. In the absence of gravity, weak flame stretch effects that are insignificant at earth gravity may become dominant. One such example is expanding spherical flames for which, according to Eq. (10), S= 1 d 2 dr* 4pr*2 ) = ( 2 4pr* dt r* dt (11), where r* is the flame front radius. Using Eq. (11), an evolution equation describing the propagation rate of an expanding spherical flame in the presence of heat loss effects was obtained (Ronney and Sivashinsky, 1989): dS 2S + S 2 ln S 2 = -Q dR R (12), where S ≡ dR/dt is the propagation speed divided by SL(r∞/rf), R is the flame radius scaled by bdI(Le, e), I(Le, e) is a scaling function that is positive for Le < 1 and negative for Le > 1 (but of course the physical radius r* is always positive), t is the time scaled by b(d/SL(r∞/rf))I(Le, e) and Q ≡ {bL( Tf)d2}/{l(Tf-T∞)} is the scaled heat loss. The terms in Eq. (12) represent unsteadiness, heat release, curvature-induced stretch and heat loss, respectively. For steady planar flames, Eq. (12) reduces to S2lnS2 = -Q, which has a turning point and thus a flammability limit at a maximum value of Q = 1/e = 0.3679... where S = e1/2. This turning point corresponds to the value of SL,lim given by Eq. (5). Numerical integrations of Eq. (12) for Le < 1 transformed into the time domain are shown in Fig. 7. Figure 7 shows that for Le < 1 the effect of curvature (2S/R) works opposite that of heat loss (Q), allowing mixtures that are non-flammable as plane flames (Q > 1/e) to sustain 15 expanding spherical flames until R (or r*) grows so large that the enhancement of combustion due to curvature is too small relative to the heat loss. Note that when Q is only slightly greater than 1/e, i.e., for mixtures just outside the flammability limit, the extinction radius may be very large. Such behavior of spherically expanding flames, termed “self-extinguishing flames” (SEFs), has been observed experimentally (Ronney, 1985; Ronney, 1988a, b) for a variety of fuels, diluents and pressures in mixtures near the µg flammability limit when Le < 1 but not so low that diffusive-thermal instabilities fragment the flame into a cellular structure. Figure 8 shows a comparison of the temporal behavior of experimentally observed SEFs and normal expanding spherical flames. Experiments (Ronney, 1988a, b; Abbud-Madrid and Ronney, 1990) also show that SEFs do not occur for Le > 1 because in this case both curvature and heat loss weaken the flame. This is consistent with the predictions of Eq. (12) for Le > 1 (R < 0). < Figure 7 near here > While the analysis leading to Eq. (12) is instructive, there are several experimental observations that it does not predict. The most significant of these are that (1) a narrow range of mixtures can exhibit both SEFs and normal flames depending on the spark ignition energy (Eign) and (2) Eign can affect the extinction radius substantially. In regards to (2), according to Eq. (12) the initial conditions have almost no effect on the extinction radius whereas experimentally, values of the ratio of the chemical enthalpy release before extinguishment (Echem) to Eign up to 70,000 have been observed. Thus, SEFs possess a remarkably strong “memory effect” for their initial condition. An example of the effects of Eign and mixture strength on extinction radius is shown in Fig. 8. SEFs occur for conditions just outside those resulting in normal flames (weak mixtures and/or lower Eign), but sufficiently weak mixtures or small Eign result in the conventional non-ignition behavior (Lewis and von Elbe, 1987) where Echem/Eign ≈ 10 (Ronney, 1985). This causes the contours of constant extinction radius seen in Fig. 8 to bend over to horizontal for sufficiently weak mixtures. A third difference between the predictions of Eq. (12) and experimental observations is that experimentally-observed SEFs exhibit a radius increasing roughly with the square root of time, i.e., r* ~ t 1/2 (Fig. 9) whereas Eq. (12) predicts a radius increasing only slightly more slowly than linearly with t. Thus, while Eq. (12) provides insight into SEFs, it does not provide a complete description. < Figures 8, 9 near here > Equation (12) is based on the Slowly Varying Flame assumption where the ratio of flame radius to d is of order b, and the Activation Energy Asymptotics (AEA) analysis is performed in 16 the limit b Æ ∞. Thus, the analysis is not valid on the length scale corresponding to flame kernel radii characteristic of ignition kernels in mixtures with Le < 1, where r* < d (Joulin, 1985). Analysis using AEA but a different scaling approach (Buckmaster and Joulin, 1989) shows that an expanding spherical flame in a mixture with Le < 1, starting from a spherical flame whose structure was that of a flame ball (Section 5.3) would exhibit a radius increasing as t 1/2ln(t), which might be indistinguishable from the apparent t1/2 behavior seen experimentally. Thus, this analysis describes one additional aspect of SEF behavior, however, the memory effects were still not predicted. These analyses, based on AEA, are necessarily subject to certain scaling assumptions and thus are not uniformly valid for the entire range of r* from very small values characteristic of ignition kernels to much larger values characteristic of SEFs near their extinction radius. A numerical model of nonadiabatic spherically expanding flames in mixtures with Le < 1 (Farmer and Ronney, 1990), free of the scaling assumptions inherent in the analytical models, does predict memory effects in good qualitative and fair quantitative agreement with experiments. As discussed in Farmer and Ronney (1990), these memory effects seem to be a result of the character of the temperature and concentration profiles that are established early in the life of the flame but persist to much later times. In particular, at early times the ignition process in mixtures with Le < 1 establishes profiles characteristic of flame balls (Section 5.3) with temperature decaying in proportion to r-1, where r is the radial coordinate, and reactant concentration increasing in proportion to 1 - r-1. These profiles persist on the longer time scale leading to the development of a propagating flame when the profiles are much steeper, i.e., with temperature decaying in proportion to exp(-(r-r*)/d), but the small additional temperature established early on affects the propagation rate until this later stage The calculations by Farmer and Ronney (1990) also show that for small initial radii, all mixtures exhibit extinguishment, which corresponds to non-ignition behavior (Lewis and von Elbe, 1987). This indicates that in mixtures capable of exhibiting SEFs, flames extinguish at sufficiently small curvature due to high stretch rates and at large curvature due to radiative losses. This dual-limit behavior is also observed in many of the other flame phenomena described below. A one-step chemical reaction model was employed in Farmer and Ronney (1990), indicating that, in accord with experimental observations (Ronney, 1988a, b; Abbud-Madrid and Ronney, 1990), the details of the flame chemistry do not affect the qualitative behavior of SEFs. Still, quantitative comparisons between experiment and computation using detailed chemical, transport and radiation sub-models and comparisons with experiment would be instructive, and in fact comparison to experimental observations of SEFs could provide a useful means of testing models of near-limit premixed-gas flames. 17 5.2 Strained flames The effects of hydrodynamic strain on flames has been studied for many years because of the need to understand how turbulence-induced strain affects flame fronts in practical combustion devices such as automotive engines. The most common apparatus for studying strained premixed flames is the counterflow round-jet configuration. Since fresh reactants are emitted from both jets, twin flames (one on either side of the stagnation plane) are produced with a burned gas region between the twin flames. The axial flow velocity, U, is given by -2cz, where c is a constant and z is the axial distance from the stagnation plane, which is independent of the radial coordinate r. The radial velocity, V, is given by cr and is independent of z. The stretch acting on the flame is sum of the extensional strains in the two directions orthogonal to the z-axis, i.e., ∂V/∂x + ∂V/∂y = 2c = -∂U/∂z. Thus, the flame stretch S is simply the magnitude of the axial velocity gradient. The counterflowing jet configuration is popular because ideally S is constant within the entire region between the two jets, so that a single parameter describes the flow, and it is very simple to produce experimentally. The equilibrium location of the flame front is at the axial location where the local axial velocity U is equal to SL for the given S = ∂U/∂z. Thus, as S is increased or SL is decreased (e.g., by reducing the fuel concentration), the flame moves closer to the stagnation plane (smaller U and z). This reduces the volume of burned gas, which affects the impact of radiative loss as discussed in the following paragraph. As with curvature-induced strain, for Le less than/greater than unity, in the counterflow configuration SL is increased/decreased by moderate strain, and for all Le, sufficiently large S extinguishes the flame (Buckmaster and Mikolaitis, 1982b). The combination of the nonmonotonic flame response to S at low Le, plus the reduced volume of burned gas (thus reduced radiative heat loss) at larger S may lead to a variety of extinction behaviors for counterflow flames depending on the relative magnitudes of trad and S-1. Recent µg experiments (Maruta et al, 1996; Guo et al., 1997) on counterflow flames in low-Le mixtures (Fig. 10) have revealed extinction behavior somewhat reminiscent of SEFs and non-ignitions in spherical flames. For large S, the short residence time of reactants within the flame front (~S-1) causes extinguishment (S-1 ≈ tchem) along the “normal flame” branch, which is analogous to non-ignition behavior of flames with small r*. For small S the residence time is large, the impact of heat losses is significant (trad ≈ tchem) and the increase in SL due to Le effects is weak, so radiant heat losses extinguish the flame along the “weak flame” branch, which is analogous to SEFs. Figure 10 shows that extinguishment along the weak flame branch can occur even in mixtures far richer than the lean planar flammability limit. The optimal S that produces the maximum increase in the flammable range (S = 13 s-1) corresponds to a time scale of 0.08 s, which is less than t vis or t inv. Thus, the C-shaped response and the entire weak-flame branch cannot be observed at 1g, however, behavior on this time scale is readily observed in drop-tower experiments. The optimal 18 S is found to be nearly the same for model and experiment (Maruta et al., 1996; Guo et al., 1997), suggesting that the loss rates are modeled well. In contrast, the computed limit composition is leaner that the experimental one, suggesting that the chemical mechanism used is not accurate for weak mixtures. < Figure 10 near here > Interestingly, due to the decrease of radiant loss as S increases, the extension of the flammability limit can also occur for mixtures with Le greater than unity (Ju et al., 1998b, 1999), though for sufficiently high Le, e.g., lean C3H8-air (Le ≈ 1.7) (Fig. 11) it does not occur. No analogous effects occur for SEFs because for the spherically expanding flame there is no mechanism by which flame stretch can affect the total radiative loss. < Figure 11 near here > Theory (Buckmaster, 1997) and numerical simulations (Ju et al., 1998b, 1999) predict that strained premixed flames with radiative loss and Lewis number effects exhibit an even more complex set of behaviors that those described here (Fig. 12). For example there is another branch of solutions, called the Far-Standing Weakly Stretched Flame (FSWSF) in which the flame front is far from the stagnation plane and thus has a very large burned-gas region. Only the FSWSF behavior can be extrapolated to S = 0 to obtain the flammability limit for planar unstretched flames. Additionally, there are jump limits between different modes that are not readily explained based on the simple physical principles outlined here. It is uncertain whether any of these solutions are physically observable, since they have not been identified experimentally and stability analyses have not yet been performed. < Figure 12 near here > 5.3 Flame balls Sections 5.1 and 5.2 discuss cases of stretched flames, where heat and mass transport are influenced by the convective environment, and behavior resulting from differences in the convective environment between 1g and µg conditions. This section discusses a phenomenon for that can occur only in the absence of convection and the curvature of the flame front is the dominant influence. Over 50 years ago, Zeldovich (1944) showed that the steady heat and mass conservation equations admit a solution corresponding to a stationary spherical flame or “flame ball” (Fig. 13), 19 just as the same governing equations in planar geometry admit a steadily propagating flame as a solution for every mixture. In the former case the solutions are characterized by a radius (r*) and in the latter case by SL. The mass conservation equation in a steady spherically symmetric system with no sources or sinks, —⋅(rv) = 0, where r is the density and v the fluid velocity vector, requires that v be identically zero everywhere. The solution to steady diffusion equations — 2T = 0 and —2Yi for the temperature T and species mass fractions Yi in spherical geometry are of the form c1 + c2/r, where c1 and c2 are constants. This form satisfies the requirement that T and Y be bounded as r Æ ∞. For cylindrical and planar geometry the corresponding forms are c1 + c2ln(r) and c1 + c2r, respectively, that are obviously unbounded as r Æ ∞. For this reason theory admits steady flame ball solutions, but not “flame cylinder” or “flame slab” solutions. Zeldovich (1944) showed that for an adiabatic flame ball, the energy and species conservation equations could be combined to infer the temperature at the surface of the flame ball (T*): T* = T• + Tf - T• (13), Le thus c1 = T∞ and c2 = (T* - T∞)r*. Zeldovich also predicted, as was supported much later by more rigorous AEA analyses (Deshaies and Joulin, 1984; Buckmaster and Weeratunga, 1984), that flame ball solutions are unstable and thus probably would not be physically observable, just as planar flames are frequently subject to instabilities that prevent them from remaining planar (Section 4). The unstable flame ball solutions, however, are related to the minimum flame kernel size required for ignition (Joulin, 1985). < Figure 13 near here > Forty years later after Zeldovich, seemingly stable flame balls were discovered accidentally in drop-tower experiments in lean hydrogen-air mixtures (Le ≈ 0.3) (Ronney, 1990) and later in drop tower and aircraft µg experiments using H2-air, H2-O2-CO2 (Le ≈ 0.2), H2-O2SF6 (Le ≈ 0.06) and CH4-O2-SF6 mixtures (Le ≈ 0.3) (Ronney et al., 1994). The µg environment was needed to obtain spherical symmetry and to avoid buoyancy-induced extinction of the flame balls. The following sequence of phenomena was observed as the mixtures were progressively diluted. For mixtures sufficiently far from flammability limits, an expanding spherical front composed of many individual cells (resulting from the diffusive-thermal instability discussed in Section 4.1) was observed that regularly subdivided to maintain a nearly constant cell spacing (cf. Fig. 5). For more dilute mixtures closer to the flammability limits, the cells formed initially did not split but instead closed up upon themselves to form stable spherical flame structures (the 20 flame balls). For still more dilute mixtures all flame balls eventually extinguished. It was inferred that stable, stationary flame balls would probably occur in all combustible mixtures with low Le for mixtures close to the extinction limits, however, the short duration of drop tower experiments and the substantial fluctuations in the acceleration level in aircraft-based µg experiments precluded definite conclusions. Recent experiments performed on the STS-83 and STS-94 Space Shuttle missions (Ronney et al., 1998) confirmed that flame balls can be stationary and stable for at least 500 seconds, which was the entire test duration in these cases. Sample images of flame balls from the space flight experiments are shown in Fig. 14. < Figure 14 near here > These results were found to be qualitatively the same over the range 0.06 < Le < 0.3, with H2 and CH4 fuels, with or without added CF3Br (a chemical inhibitor) and at pressures from 0.5 to 3 atm, indicating that variations in Lewis number over this range, chemical mechanisms, and radiation spectra do not qualitatively influence these phenomena. As predicted by Eq. (13), because T* - T ∞ ~ 1/Le, in mixtures with Le < 1 the flame ball temperature T* can be much larger than the adiabatic homogeneous flame temperature T f. In the case of H2-O2-SF6 mixtures (that have Le ≈ 0.06, the lowest of the mixtures tested to date), values of Tf as low as 465K have been found to exhibit flame balls. This temperature is far below the H2-O2 explosion limit temperature of about 850K at 1 atm (Lewis and von Elbe, 1987), thus such mixtures could not possibly exhibit plane flames. Zeldovich (1944) noted the possibility of heat losses stabilizing flame balls. The apparent experimental discovery of stable flame balls in near-limit mixtures 40 years later motivated additional theoretical studies. Volumetric radiative losses (e.g., due to gas radiation) are predicted (Buckmaster et al., 1990, 1991) to lead to two stationary flame ball radii (Fig. 15), a “large” flame ball that is strongly affected by heat loss and a “small” flame ball that is nearly adiabatic. When the losses are sufficiently strong no solutions exist, indicating a flammability limit.* As the limit is approached, the difference between the radii of the "large" and "small" balls decreases to zero. Stability analyses predict that all small flame balls are unstable to radial disturbances, i.e., the flame will either grow outward from the equilibrium radius (and possibly develop into a propagating flame) or collapse and extinguish. The basic reason is that as the flame ball radius increases, the radius to volume ratio decreases, thus the ratio of total heat release (which is proportional to the flame ball radius) to total radiative heat loss (which is proportional to the *As a result, there are at least four steady solutions to the low Mach number conservation equations for nonadiabatic flames, namely the two solutions to the equation S 2ln(S2)=-Q for planar flames and the two flame ball solutions. 21 flame ball volume) increases, thus the flame ball becomes weaker and shrinks. Conversely, if the radius decreases, the flame ball grows stronger and expands. Thus, flame balls with sufficient volumetric losses can be stable to radial disturbances, but only for the large flame balls because the smaller balls have too little volume and thus too little volumetric loss for this mechanism of stabilization to be effective. Large flame balls with weak heat loss effects, i.e., far from the flammability limits, are predicted to be unstable to three-dimensional disturbances, which is consistent with the observation of splitting cellular flames in these mixtures. A portion of the large flame branch close to the extinction limits is stable to both types of disturbances, which is consistent with the experimental observations. < Figure 15 near here > It has also been predicted (Lee and Buckmaster, 1991) that stable flame balls can only exist for mixtures with mixtures having Le less than a critical value that is less than unity, which explains why flame balls are not observed for mixtures with Le less than but close to unity (e.g. CH4-air) or larger than unity (e.g. C3H8-air), even for near-limit mixtures at µg. Instead, conventional propagating flames are observed under these conditions, and SEFs in sub-limit mixtures when Le is less than (but still close to) unity. The reason is that, according to Eq. (13), for Le > 1, T* < Tf, thus flame balls are weaker than plane flames and cannot benefit from curvature in the manner discussed above for Le < 1 mixtures. All the aforementioned theories assume single-step Arrhenius kinetics with large activation energy, constant thermodynamic and transport properties, and simple radiation properties. Numerical simulations (Buckmaster et al., 1993) of the steady properties of nonadiabatic flame balls in H2-air mixtures employing detailed chemistry, diffusion and radiation models were qualitatively consistent with these theories in that two solutions were predicted for mixtures having fuel concentrations higher than the limiting value. It was shown that for mixtures away from the flammability limit, the large flame ball is highly influenced by radiative loss, thus its temperature is much lower than T* given by Eq. (11) (the cold giant (CG) flame) and the small flame ball is too small for volumetric losses to be significant, thus its temperature is much closer to T* from Eq. (11) (the hot dwarf (HD) flame). These correspond to the upper and lower solution branches shown in Fig. 16; only the CG branch is stable, but the numerical methods used in Buckmaster et al. (1993) allowed the unstable HD flames to be computed as well. These two solutions are somewhat reminiscent of the “normal flame” and “weak flame” branches of the strained flame extinction curves (Section 5.2). Thus, radiative heat losses can quench flames even in mixtures far from the turning-point extinction limits on the solution branches corresponding to large residence times (for strained flames) or large volumes (for flame balls). 22 < Figure 16 near here > The simulation of the steady properties of flame balls (Buckmaster et al., 1993) has been extended (Wu et al., 1998, 1999) to consider dynamical properties. As Fig. 17 shows, when the initial flame radius (ro) is close to the steady flame ball radius (r*), the flame eventually evolves to r*, indicating stable solutions as predicted by the theories (Buckmaster et al, 1990, 1991; Lee and Buckmaster, 1991), whereas for significantly larger or smaller ro the flames eventually quench. As the lean and rich stability limits are approached, the range of ro leading to steady flames narrows to zero. None of these CG-like initial conditions led to steadily propagating flames, but obviously, sufficiently rich mixtures do exhibit propagating flames. Calculations showed that f at the planar lean limit (due to radiative losses) is 0.298, which is higher than the rich stability limit (f = 0.285) of the flame ball. Interestingly, then, for 0.285 < f < 0.298 there are no stable flames of any kind. Additionally, the effects of various chemical and radiation models were tested. It was found that three different widely-used models of H2-O2 chemistry gave widely varying predictions of flame ball radius and radiant emission (Fig. 18), even though all predict the burning velocities of propagating flames in H2-air mixtures very well (Fig. 19). The main chemical step responsible for these discrepancies is the inhibiting step H + O2 + H2O Æ HO2 + H2O, which, over the relevant range of temperatures, vary by a factor of more than two between the H2-O2 reaction mechanisms shown in Fig. 18. Similar discrepancies have been noted for near-limit propagating H2-air flames (Egolfopoulos and Law, 1990a). < Figure 17, 18, 19 near here > Another factor in flame ball properties is reabsorption of emitted radiation, which is an important effect in mixtures diluted with radiatively-active gases such as CO2 and SF6. The Planck mean absorption lengths are of the order 100 cm for H2O at the relevant conditions but 4 cm for CO2 and 0.3 cm for SF6. Since the chamber radius is 12 - 16 cm in the experiments performed to date, H2O is optically thin (negligible reabsorption) in all cases, but CO2 and SF6 are optically thick, and thus at least a portion of their emitted radiation is reabsorbed within the gas, and therefore is not a loss process. An upper bound on diluent reabsorption effects (aP,diluentÆ ∞) can be obtained by neglecting diluent radiation entirely because as aP,diluentÆ ∞ there is no radiative loss from the diluent and furthermore the “radiative conductivity” ≡ 16sT3/3aP approaches zero, thus there is no additional heat transport due to radiative transfer. An example of the differences in predictions obtained using the optically-thin and optically-thick approximations is shown in Fig. 20 for H2-O2-CO2 mixtures. The agreement between predicted 23 and measured flame ball radii is much better when diluent radiation is neglected and the experimental flammability limit composition is bracketed by numerical results with and without diluent radiation. Recent simulations employing a detailed Statistical Narrow Band model (Wu et al., 2000) yield results surprisingly similar to that of the upper bound estimate. These observations strongly suggest that radiative reabsorption effects are needed for accurate numerical simulation in these cases. < Figure 20 near here > 6. "COOL FLAMES" It has been known for over 100 years than some combustible mixtures, particularly rich hydrocarbon-O2-inert mixtures, can exhibit unusual behavior when preheated to moderate temperatures (typically 300˚C - 600˚C) (Mallard and LeChatelier, 1880). The classical apparatus used to study such behavior is a preheated, initially evacuated vessel into which reactants are rapidly introduced. After some induction period, the vessel pressure begins to rise and, depending on the conditions, either a single-stage autoignition that consumes most of the reactants occurs, or one or more propagating fronts or "cool flames" occur that consume only a portion of the reactants propagates through the vessel. Such cool flames occur because for these mixtures, over a range of temperatures the net heat release rate decreases as temperature increases, consequently the more typical self-accelerating thermal runaway is not observed. Such cool flames result from complex thermokinetic interactions that occur within this temperature range. Under favorable conditions, specifically in the presence of heat and radical losses, this leads to oscillatory reaction or multiple cool flames. The study of autoignition and cool flames has many practical applications, particularly to knock in premixed-charge (i.e. spark-ignition, gasoline-fueled) internal combustion engines, which is the limiting factor in compression ratio and thus thermal efficiency. Moreover, proposed homogeneous charge - compression ignition engines (Stanglmaier, 1999) promise to provide high thermal efficiency and low emissions, but rely on repeatable, controlled autoignition which is currently very difficult to obtain. Recently new attention has been focussed on autoignition and cool flame phenomena because of their possible role in the TWA Flight 800 Center Wing Tank explosion (DOT, 1998). Because of its practical importance, numerous reviews of cool flame behavior appear in the literature, e.g., Griffiths and Scott (1987), Scott (1997). More than 30 years ago (Tyler, 1966; Griffiths et al., 1970) buoyant convection was shown to be an important factor in the induction period and cool flame generation, since cool flames are generally only observed at sufficiently high pressures and in sufficiently large vessels 24 that the Rayleigh number Rad ≡ GrdPr is larger than the critical value (≈ 600) for the onset of buoyant convection. For this reason the continuously-stirred tank reactor, in which forced turbulence is used to cause a homogeneous mixture, was developed to suppress temperature and composition gradients. In this system oscillatory behavior can be observed, but spatiallyinhomogeneous cool flames cannot. Experiments by Pearlman (2000a, b) in unstirred vessels showed that at 1g, multiple cool flames can be observed in rich n-C4H10/O2 mixtures whereas at µg, usually only a single flame is observed (Figure 21.) At 1g, the cool flames always originate at the top of the vessel whereas at µg, the flame originates from the center of the vessel. In both cases this is thought to be due to the temperature becoming highest at the location where the cool flames originate. (While the initial temperature in both 1g and µg cases is uniform, at 1g exothermic chemical reaction generates locally higher temperatures in some region and the hotter, less dense gas rises to the top of the vessel, whereas at µg the walls act as a sink for the thermal energy generated by chemical reaction and thus in quasi-steady state there is a net conductive heat flux from the center of the vessel toward the walls.) Pearlman (2000a, b) also found that the induction period (time to the first cool flame or autoignition) is always shorter at µg due to the lower heat losses associated with conductive as compared to convective transport. Furthermore, at µg, when chemically-inert helium is added, which increases both the thermal diffusivity and Lewis number of the mixture, multiple cool flames were observed. When the same mole fraction of argon (which has the same heat capacity as helium but leads to lower thermal diffusivity and Lewis number than helium) is substituted for helium, only a single cool flame followed by homogeneous ignition is observed. < Figure 21 near here > These results clearly indicate the need for sufficiently high heat and/or mass diffusion rates in order to observe multiple cool flames at µg, and that transport due to buoyant convection at 1g can act in a manner qualitatively similar to diffusive transport. Very recent modeling studies (Fairlie et al., 2000) using a one-dimensional, time-dependent code in spherical geometry show qualitative agreement with the experimental observations and provides some additional insight. The following explanation for the effects of transport on the existence of multiple cool flames at µg was proposed. After the passage of the first cool flame, the lowest temperatures are present in the outer part of the reacting mass (nearer the wall in µg, or nearer the bottom of the vessel at 1g), which enables the accumulation of reactive intermediates close to the edge of this mass (due to the negative effect of temperature on reaction rate under these conditions). Chemical reaction and heat release are thus sustained in this region and there is an accompanying transport of the reactive intermediates towards the center, which then augment reaction of the 25 residual fuel sufficiently to lead to development of subsequent cool flame(s). Fairlie et al. also found that when the mass diffusion of the intermediates was suppressed but the thermal diffusion was maintained at a sufficiently high value, oscillations did not occur, indicating there is a key role of the re-supply of the intermediate to the center to help re-establish reaction there (since the cool flames always originate in the center.) Without this diffusion to the center, the system was dormant because intermediates did not survive after the first cool flame passage. In the model, it was found that a low Lewis number of the free radical specie responsible for chain propagation encouraged multiple cool flames but a high Lewis number discouraged cool flames, indicating that transport of radicals is a more significant factor that transport of temperature in enabling cool flames. While the experiments (Pearlman, 2000b) showed that a higher Lewis number encouraged rather than discouraged cool flames, this was in conjunction with the substitution of helium for argon as an inert, and thus both diffusivities increased in this case. Consequently, the overall propensity for cool flame behavior seems to depend more on increasing diffusivities than increasing the ratio of mass to thermal diffusivity. When spatially uniform conditions and a finite volumetric heat loss term are assumed (i.e., infinite diffusivity within the gas, but finite transfer from the gas to the wall) in the model, the entire volume can cool down, leading oscillatory chemical activity (this is expected since the chemical model used by Fairlie et al. was chosen to ensure oscillatory chemical activity under spatially uniform conditions.) Of course, if the heat loss term were infinite, no cool flame or homogenous oscillatory behavior could be observed because the gas temperature would be equal everywhere to the wall temperature. To date, computations and experiments have not yet established the upper and lower limits on Lewis and Peclet number (a measure of the heat loss term that could be defined in this case as tdiff/tchem) for which cool flames, oscillatory homogenous reactions, quenching, etc. are observed. 7. TURBULENT FLAMES All of the phenomena described in the previous sections apply to flames in quiescent or laminar flows. Of course, most practical flames occur in turbulent flows. In general, one would expect that with the large increases in convective transport in turbulent flows as compared to laminar flows, most turbulent flames would not be affected by buoyancy, but exceptions do exist. The impact of buoyancy on turbulent flame speed (ST) has been predicted (Libby, 1989) to be given by the relation 26 ST u' 1 - e gLI = 2.15 + SL SL 0.867 S2L (14), where LI is the integral scale of turbulence, u’ the turbulence intensity and g is positive/negative for upward/downward propagation. Equation (14) indicates that the strongest effects occur for large scale turbulence and small SL. This might be expected since a buoyancy-induced convection velocity on the turbulent wrinkling scale would be proportional to (gLI)1/2, and thus the buoyancy effect term in Eq. (14), gLI/SL2, is simply the square of the ratio of this buoyant convection velocity to SL. It may also be considered the inverse of a Froude number. The effect of the buoyancy parameter gLI/SL2, thermal expansion and diffusive-thermal effects on ST has also been studied (Aldredge and Williams, 1991) for downward-propagating turbulent flames in the limit u’/SL << 1. Still, there have been no experimental tests of these predictions. Flames in a rapidly rotating tube (50 go < g < 850 go) have been examined (Lewis, 1970) and it is found that ST ~ g0.387, though in this case there was no forced turbulence; instead the buoyant flow induced turbulence (with the aid of a perforated plate near the ignition site to initiate turbulence). Experiments at earth gravity in tubes of varying diameter with gases of varying viscosity (Wang and Ronney, 1993) have shown that laminar-flow relation for rising flames in tubes (Levy, 1965) Sb ~ (gd)1/2 applies even in large tubes and in gases with low viscosity where the burned gases are turbulent, and thus the observations (Lewis, 1970) might be simply a manifestation of the buoyant rise speed at very high g. Other investigators (Hamins et al., 1988) compared upward and downward-propagating flames in which a rotating fan was used to generate turbulence. No significant effect of gravity was found, but near-limit mixtures having small SL (which are the mixtures most likely to exhibit buoyancy effects) were not tested and furthermore the turbulence intensity (u’) was not measured. Recently the effects of gravity on flame structure and stabilization of turbulent premixed round-jet flames and flames stabilized downstream of a cylindrical rod (“V-flames”) has been investigated (Kostiuk and Cheng, 1995; Bedat and Cheng, 1996). For these flames, as the flow velocity (U) increases, both the turbulence levels and the volume of buoyant fluid increase, thus buoyancy effects do not necessarily decrease. For example, in lean methane-air mixtures the flame lengths at 1g and µg do not converge as U increases, in fact the difference in flame lengths remains approximately constant (Kostiuk and Cheng, 1985). This was attributed to the difference between the divergent (thus decelerating) flow at µg as opposed to the practically non- 27 divergent flow associated with the accelerating buoyant plume at 1g. In contrast, little effect of buoyancy was found on flame stabilization limits (minimum and maximum U for which the jet flame could be stabilized) for turbulent jets, whereas for laminar jets there was much more effect (Bedat and Cheng, 1996), perhaps because the stabilization mechanism is dominated mainly by the behavior near the flame base and thus is not affected by buoyancy when the jet exit velocity is large (and thus the flow is turbulent). 8. RECOMMENDATIONS FOR FUTURE STUDIES 8.1 Flame propagation in optically-thick mixtures This review of prior literature on premixed-gas flame studies at µg indicate the need for additional studies in some areas. Perhaps the most pressing of these is optically-thick flame propagation. All of the radiative effects discussed above are critically dependent on the degree of reabsorption. To study reabsorption effects requires large systems, high pressures and/or radiatively-active diluents such as CO2 and SF6 that have small n. All of these conditions lead to high Grashof numbers at 1g and thus turbulent flow. Hence, µg experiments provide an excellent opportunity to study reabsorption effects on combustion processes without the additional complication of turbulent flow. Reabsorption effects is a subject of importance not only to µg studies, but also combustion at high pressures and in large industrial furnaces. For example, at 40 atm, a typical pressure for premixed-charge internal combustion engines, ap ≈ 18 m-1, thus aP-1 = 4.5 cm, for the products of stoichiometric combustion. This length scale is comparable to the cylinder radius, thus reabsorption effects within the gas cannot readily be neglected in models of engine combustion and heat transfer. The ratio of radiative conductivity (lR) to molecular conductivity = (16sT f3/3ap)/lf for this example is about 1700. This indicates that even when turbulent transport is considered (which increases lf significantly but has no effect on l R), much of the thermal energy transport will be due to radiation rather than conduction or convection. Understanding of these radiative effects is a relevant issue because simple estimates (Ronney, 1995, 1999) indicate radiative loss may influence flame quenching by turbulence, which limits the use of clean, fuel-efficient lean mixtures in practical engines. For similar reasons, reabsorption cannot be neglected in atmospheric-pressure furnaces larger than aP-1 ≈ 2.2 m. Moreover, many combustion devices employ exhaust-gas or flue-gas recirculation; for such devices significant amounts of absorbing CO2 and H2O will be present in the unburned mixtures, which will further increase the importance of radiative transport. 28 While modeling of radiation effects in flames under optically thin conditions (e.g. Lakshimisha et al., 1990; Ju et al., 1998b, 1999; Wu et al., 1998) is reasonably straightforward, modeling of spectrally dependent emission and absorption is a challenging task because effects of local fluxes depend on the entire radiation field, not just local scalar properties and gradients. Some studies using gray-gas models have been reported (Marchese and Dryer, 1996) but recent studies (Bedir et al., 1997; Ju et al., 1998a) have shown that the accuracy of these methods for assessing reabsorption effects at high pressures or in large systems is questionable, because of the wide variation in spectral absorption coefficient with temperature and species. A useful comparison of various approximate radiative treatments for a nonpremixed flame of small dimension (≈ 1 cm) has been given recently (Bedir et al., 1997). Comparisons for premixed flames and larger, multi-dimensional systems would be useful. Despite the value of investigations of radiative reabsorption effects, only very preliminary experimental studies have been conducted to date (Abbud-Madrid and Ronney, 1990, 1993; Wu et al., 1998). Experimentally, there are at least two ways to control the optical thickness, either by using diluent gases with small absorption lengths (e.g. CO2 or SF6, particularly at high pressure), or by the addition of inert radiating particles. Both approaches have limitations. In the former case the disadvantage is on the modeling side - there is a very complicated spectral dependence of the absorption coefficient on wavelength, which in turn depends on temperature and to a lesser extent on pressure, and there are always spectral regions in which no emission or absorption occurs. In the latter case the problem is mostly on the experimental side, namely in obtaining uniform particle dispersion and possibly complications due to heterogeneous chemical effects on the surface of the particles. 8.2 Flame propagation in tubes In sections 3.3 the relative importance of conductive vs. radiative heat loss on extinction of flames propagating through cylindrical tubes was discussed. A more thorough assessment of the mechanisms of flammability limits in tubes would be valuable, in particular the assessment of the transition from limits induced by conductive loss to limits induced by radiative loss as the tube diameter is increased. It would be expected that a transition from diameter-dependent SL,lim due to conductive losses to diameter-independent SL,lim due to radiative losses would occur when the values of SL,lim predicted by Eqs. (5) and (6) are equal, i.e. d ª 16 l •T f (14), bL 29 where the same temperature-averaging of transport properties used in section 3.2 and 3.3 has been applied here. For the example of lean hydrocarbon-air flames employed throughout this paper, Eq. (14) predicts d ≈ 2.7 cm at the transition. The accuracy of this prediction should be tested for a range of pressures and diluent gases to test the effect of l ∞ and L on this transition diameter. 8.3 Three-dimensional effects Linear stability analyses (Buckmaster et al., 1990, 1991) show that three-dimensional instabilities are important in the development of flame balls from an ignition kernel. To date, it is known that large flame ball branch is linearly unstable to three-dimensional disturbances and exhibits splitting flame balls if the scaled heat loss is sufficiently small (Fig. 15), which will occur mixtures sufficiently far from the extinction limit or if the local enthalpy is increased sufficiently by an ignition source. Still, the transition from splitting flame balls to stable flames has not been analyzed to date nor is there any method to predict the number of flame balls produced from a given ignition source. Modeling using a three-dimensional code (Patnaik et al., 1996) could shed some light on this subject. 8.4 Catalytic combustion The potential benefits of catalytic combustion for reduced emissions and improved fuel efficiency in many combustion systems is well known (Pfefferle and Pfefferle, 1986; Warnatz, 1992). Since catalysis occurs at surfaces, catalysis is inherently a multi-dimensional and/or unsteady process requiring transport of reactants to the surface and heat and products away from the surface. While boundary layer approximations can be incorporated into models of reaction at catalytic processes, the only truly one-dimensional steady catalytic configuration would be a spherical surface immersed in a nonbuoyant quiescent gas - i.e., a “catalytic flame ball.” In this case the radius (r*) is fixed but the surface temperature T* and fuel concentration Y* are unknown. T * and Y* can be related through energy conservation (including surface radiation) and the diffusion equations, leading to the following expression for the surface reaction rate in moles per second (w): w (Y* ,T* ) = r* D* r* Y• (1 - Y* / Y• ) / M; Ê T - T ˆ Ê ser* (T*4 - T•4 )ˆ Y* • ˜ Á1 + = 1 - LeÁ * ˜ Y• l * (T* - T• ) ¯ Ë Tf - T• ¯ Ë (15) where Y∞ is the ambient fuel mass fraction, M the fuel molecular weight, T* the measured surface temperature, e the surface emissivity (not to be confused with the density ratio used earlier in this paper) and the subscript * refers to properties evaluated at the temperature T *. Through 30 varying r*, Y∞, pressure, and diluent gas, w can be determined for a range of surface fuel concentrations and temperatures and compared to models. Of course, the conditions must be unfavorable for the initiation of a propagating flame or a flame ball that stands off from the surface for the catalysis process to be examined. 8.5 Turbulent flame quenching Turbulence may increase ST to values well above SL, however, increasing turbulence levels beyond a certain value increases ST very little if any and may lead to complete quenching of the flame (Abdel-Gayed and Bradley, 1985, Bradley, 1992). This effect is particularly pronounced when SL is small compared to u', e.g., for lean fuel-air mixtures. This indicates that the propagation rates of very lean mixtures cannot be increased ad infinitum merely by increasing u'. Thus lean mixtures, which thermodynamically promise higher thermal efficiencies and lower pollutant emissions, will exhibit unsatisfactory combustion rates in many practical systems. In addition to its long-standing relevance to automotive applications (Heywood, 1988), lean premixed turbulent combustion is now employed in stationary gas turbine applications because NOx emissions can be reduced considerably compared to stoichiometric mixtures (Correa, 1992). The mechanism(s) of flame quenching by turbulence are still not well understood. Recently it has been suggested that radiative heat losses are a likely mechanism, leading to, for hydrocarbons in air at 1 atm, the following predicted quenching criterion (Ronney 1995, 1999): Ka ≈ 0.38 ReL0.76 (16), where Ka ≡ 0.157 ReL-1/2(u'/SL)2Sc-1 is the turbulent Karlovitz number, ReL = u'LI/n is the turbulent Reynolds number and Sc the Schmidt number (≈ 0.7 for most gases). One difficulty in assessing this prediction and especially probing the structure of flames near this quenching threshold is that relatively high u'/SL is required to obtain quenching, thus very high u' for nearstoichiometric mixtures with high SL. Moreover, the power required to generate the turbulence is proportional to (u')3 and the size of the smallest scales of turbulence is proportional to 1/(u')3/4. Consequently, it would be desirable to employ mixtures with low SL and thus low u' to obtain a given u'/SL, but buoyancy effects preclude the use of such mixtures at 1g. A study of turbulent flame quenching at µg using diluents of varying radiative properties, and comparison of these results to Eq. (16) or other models of quenching (Abdel-Gayed and Bradley, 1985, Bradley, 1992), could provide considerable insight into this subject. For example, it may be useful to study the effects of pressure (P) on Ka at quenching, since very few of these data are reported on the literature. The analysis leading up to Eq. (16) predicts that the factor 0.38 is proportional to 31 P-1 because of the effect of pressure on radiative loss per unit volume. This leads to Ka ~ P0.24 for fixed u' and LI whereas published correlations (Bradley, 1992) predict Ka ~ P1 for such conditions. Microgravity experiments could be quite useful for testing and comparing such models. 8.6 Chemical models One of the most important contributions of µg combustion experiments has been an improved understanding of extinction processes. Of course, extinction processes are inherently related to finite-rate chemistry effects. Thus, to obtain closure between µg experiments and model predictions, accurate chemical models are needed. The near-limit flame studies at µg have clearly indicated the inadequacy of our knowledge of the reaction rates of even the relatively simple H2-O2 chemistry (a necessary subset of hydrocarbon-O2 chemistry) under weakly burning conditions (cf. Figs. 18 and 19). In fact, µg experiments, for example SEFs (section 5.1) or flame balls (section 5.3) may prove to be one of the most useful means of determining these rates because the simplicity of the flow environment allows more computational resources to be brought to bear on the chemical part. To date, in practically all comparisons for lean premixed hydrocarbon-air flames the models (Lakshmisha et al., 1990; Guo et al., 1997; Ju et al., 1998a) predict higher SL and leaner flammability limits than the experimental observations (Strehlow and Reuss, 1981; Ronney, 1985; Abbud-Madrid and Ronney, 1990; Maruta et al., 1996). The discrepancy seems to be more than experimental uncertainty or unaccounted heat losses could explain. In contrast, for flame balls (Wu et al., 1998, 1999) and strained premixed H2-air flames at 1g (Egolfopoulos and Law, 1990a) the same chemical reaction mechanisms predict smaller flame balls, lower SL and richer flammability limits than the experimental observations. All of these chemical models faithfully predict the burning velocities of flames in mixtures away from extinction limits. A substantial part of the discrepancy seems to be due to differences in the 3-body recombination rates for the H + O2 + M reactions, and in particular the third-body efficiency of various M species (Wu et al., 1998). These reactions are extremely important in near-limit flames, but of much lesser importance in mixtures away from limits, because of the competition between chainbranching and chain-inhibiting steps near limits (Egolfopoulos and Law, 1990b). Further consideration of the proper rates of these reactions in the intermediate temperature range (11001400K in most cases) would be most welcomed. All practical combustion engines operate at pressures much higher than atmospheric. The relative importance of various elementary reaction steps changes as pressure increases (Egolfopoulos and Law, 1990b). The impact of buoyancy scales as tchem/tvis ~ (ga/SL3)2/3 ~ Pn-4/3, where n is the overall order of reaction (SL ~ Pn/2-1). Since typically n < 4/3 for weak mixtures 32 (Egolfopoulos and Law, 1990b) where buoyancy effects are likely to be important, the impact of buoyancy generally increases with pressure. Also, as discussed in Section 1, the effects of radiative transport are more difficult to assess at higher pressure due to increased interference from buoyant transport. Consequently, further study of µg flammability limits and near-limit burning velocities at high pressure would provide a useful assessment of flame chemistry models at high pressure. 9. CONCLUSIONS A wide variety of premixed-gas flame phenomena in mixtures with low burning velocities are either seen only at µg conditions, or are much more clearly elucidated at µg. At 1g, heat and mass transport affecting these weakly-burning flames is dominated by buoyant convection. When buoyancy is eliminated, transport of thermal energy by diffusive and radiative mechanisms becomes much more important. Correspondingly, transport of chemical species by diffusion also becomes much more important at µg (though there is no analog to thermal radiation for transport of chemical species). Most of the observed changes in flame behavior at µg can be attributed to the increased importance of diffusive and radiative effects. One of the most important results of µg combustion experiments is that they have helped to integrate radiation into premixed flame theory. Although flame radiation has long been recognized as an important heat transfer mechanism in large flames, its treatment has largely been ad hoc because of the difficulty of predicting soot formation. Also, large-scale flames at 1g are inevitably turbulent, leading to complicated flame-flow interactions. µg flames are laminar, often soot-free and have significant influences of radiation. As a result, premixed flames have exhibited dual-limit extinction behavior, with residence time limited extinction at high strain or curvature and radiative loss induced extinction at low strain or curvature. The high-strain limit is readily observed at 1g, and in fact in some cases buoyant flow causes the strain. For weak mixtures these limits converge, but the convergence and the entire low-strain extinction branch can only be seen at µg (Figs. 10 and 11). Similar behavior is also commonly found for non-premixed flames (Maruta et al., 1998). Considering the rapid progress made recently in studies of premixed flames at µg, further advances are certain to occur in the near future. Hopefully this report on the current state of understanding can help motivate and inspire such advances. ACKNOWLEDGMENTS 33 The author’s work on premixed-gas combustion at µg has been supported by the NASA Glenn Research Center under grants NAG3-965, NAG3-1242, NAG3-1523 and NAG3-2124. Discussions with John Buckmaster, John Griffiths, Guy Joulin, Yiguang Ju, Kaoru Maruta, Takashi Niioka, Howard Pearlman, Howard Ross, Gregory Sivashinsky, Karen Weiland, Forman Williams and others have been invaluable in extending the author’s knowledge of this subject and in the preparation of this manuscript. REFERENCES Abbud-Madrid, A., Ronney, P. D. (1990). Effects of radiative and diffusive transport processes on premixed flames near flammability limits, Proc. Combust. Inst. 23, 423-431. Abbud-Madrid, A., Ronney, P. D. (1993). Premixed flame propagation in an optically-thick gas, AIAA J. 31, 2179 -2181. Abdel-Gayed, R. G., Bradley, D. (1985). Criteria for turbulent propagation limits of premixed flames. Combust. Flame 62, 61-68. Aldredge, R. C., Williams, F. A. (1991). Influence of wrinkled premixed-flame dynamics on large-scale, low-intensity turbulent flow, J. Fluid Mech. 228, 487-511. Aly, S. L., Hermance, C. E. (1981). A two-dimensional theory of laminar flame quenching, Combust. Flame 40, 173. Anon., (1995). Microgravity Combustion Science: 1995 Program Update, NASA TM-106858. Bedat, B., Cheng, R. K. (1996). The effects of buoyancy on premixed flame stabilization, Combust. Flame 107, 13-26. Bedir, H., Tien, J. S., Lee, H. S. (1997). Comparison of different radiation treatments for a onedimensional diffusion flame, Combust. Theory Modeling 1, 395-404. Booty, M. R., Margolis, S. B., Matkowsky, B. J. (1987). Interaction of pulsating and spinning waves in nonadiabatic flame propagation, SIAM J. Appl. Math. 47, 1241-1286. Bradley, D. (1992). How fast can we burn?, Proc. Combust. Inst. 24, 247-262. Buckmaster, J. D. (1976). The quenching of deflagration waves, Combust. Flame 26, 151 -162. Buckmaster, J. D. (1997). The effects of radiation on stretched flames, Combust. Theory Modeling 1, 1-11. Buckmaster, J., Joulin, G. (1989). Radial propagation of premixed flames and √t behavior, Combust. Flame 78, 275-286. Buckmaster, J. D., Mikolaitis, D. (1982a). A flammability-limit model for upward propagation through lean methane/air mixtures in a standard flammability tube, Combust. Flame 45, 109119. 34 Buckmaster, J. D., Mikolaitis, D. (1982b). The premixed flame in a counterflow, Combust. Flame 47, 191-204 . Buckmaster, J. D., Weeratunga, S. (1984). The stability and structure of flame-bubbles, Combust. Sci. Tech. 35, 287-296. Buckmaster, J. D., Joulin, G., Ronney, P. D. (1990). Effects of heat loss on the structure and stability of flame balls, Combust. Flame 79, 381-392. Buckmaster, J. D., Joulin, G., Ronney, P. D. (1991). Structure and stability of non-adiabatic flame balls: II. Effects of far-field losses, Combust. Flame 84, 411-422. Buckmaster, J. D., Smooke, M., Giovangigli, V. (1993). Analytical and numerical modeling of flame-balls in hydrogen-air mixtures, Combust. Flame 94, 113-124. Clavin, P. (1985). Dynamic behavior of premixed flame fronts in laminar and turbulent flows, Prog. Energy Comb. Sci. 11, 1-59. Correa, S. (1992). A review of NOX formation under gas-turbine combustion conditions. Combust. Sci. Tech. 87, 327-362. Coward, H. and Jones, C. (1952). Flammability limits of gases and vapors, U.S. Bur. Mines Bull. 503. Darrieus, G. (1938). Propagation d’un front de flamme, paper presented at La Technique Moderne, France. Deshaies, B., Joulin, G. (1984). On the initiation of a spherical flame kernel, Combust. Sci. Tech. 37, 99-116. DOT/FAA/AR-98/26 (1998). "A Review of the Flammability Hazards of Jet A Fuel Vapor in Civil Aircraft Fuel Tanks," June. Dunsky, C., Fernandez-Pello, A. C. (1990). Gravitational effects on cellular flame structure, Proc. Combust. Inst. 23, 1657-1662. Egolfopoulos, F. N., Law, C. K. (1990a). An experimental and computational study of the burning rates of ultra-lean to moderately-rich H2/O2/N2 laminar flames with pressure variations, Proc. Combust. Inst. 23, 413-421. Egolfopoulos, F. N., Law, C. K. (1990b). Chain mechanisms in the overall reaction orders in laminar flame propagation, Combust. Flame 80, 7-16. Fairlie, R., Griffiths, J. F., Pearlman, H. (2000). Cool flame development under microgravity conditions, Proc. Combust. Inst. 28, to appear. Farmer, J. N., Ronney, P. D. (1990). A numerical study of unsteady nonadiabatic flames, Combust. Sci. Tech. 73, 555-574. Giovangigli, V. and Smooke, M. (1992). Application of Continuation Methods to Plane Premixed Laminar Flames, Combust. Sci. Tech. 87, 241-256. 35 Griffiths, J. F., Scott, S. K. (1987). Thermokinetic interactions: fundamentals of spontaneous ignition and cool flames. Prog Energy Combust Sci. 13, 16. Griffiths, J. F., Gray, B.F., Gray, P. (1970). Multistage ignition in hydrocarbon combustion: Temperature effects and theories of nonisothermal combustion. Proc. Combust. Inst. 13, 239-248. Guo, H., Ju, Y., Maruta, K., Niioka, T., Liu, F. (1997). Radiation extinction limit of counterflow premixed lean methane-air flames, Combust. Flame 109, 639-646. Hamins, A., Heitor, M., Libby, P. A. (1988). Gravitational effects on the structure and propagation of premixed flames, Acta Astronautica 17, 503-514. Heywood, J. B. (1988). Internal Combustion Engine Fundamentals, McGraw-Hill. Hubbard, G. L., Tien, C. L. (1978). Infrared mean absorption coefficients of luminous flames and smoke, J. Heat Trans. 100, 235. Jarosinsky, J. (1983). Flame quenching by a cold wall, Combust. Flame 50, 167. Jarosinsky, J., Strehlow, R. A., Azarbarzin, A. (1982). The mechanisms of lean limit extinguishment of an upward and downward propagating flame in a standard flammability tube, Proc. Combust. Inst. 19, 1549-1557. Joulin, G. (1985). Point-source initiation of lean spherical flames of light reactants: an asymptotic theory, Combust. Sci. Tech. 43, 99-113. Joulin, G., Clavin, P. (1976). Analyse asymptotique des conditions d’extinction des flammes laminaries, Acta Astronautica 3, 223. Joulin, G., Clavin, P. (1979). Linear stability analysis of nonadiabatic flames: diffusional-thermal model, Combust. Flame 35, 139-153. Joulin, G., Deshaies, B. (1986). On radiation-affected flame propagation in gaseous mixtures seeded with inert particles, Combust. Sci. Tech. 47, 299-315. Joulin, G., Sivashinsky, G. I. (1983). On the dynamics of nearly-extinguished non-adiabatic cellular flames, Combust. Sci. Tech. 31, 75-90. Ju., Y, Masuya, G. and Ronney, P. D. (1998a). Effects of radiative emission and absorption on the propagation and extinction of premixed-gas flames, Proc. Combust. Inst. 27,.2619-2626. Ju, Y., Guo, H., Maruta, K., Niioka, T. (1998b). Flame bifurcations and flammable regions of radiative counterflow premixed flames with general Lewis numbers, Combust. Flame 113, 603. Ju, Y., Guo, H., Liu, F., Maruta, K. (1999). Effects of the Lewis number and radiative heat loss on the bifurcation of extinction of CH4-O2-N2-He flames, J. Fluid Mech. 379, 165-190. Kaper, H. G., Leaf, G. K., Matkowsky, B. J., Olmstead, W. E. (1987). Dynamics of nonadiabatic premixed flames in a gravitational field, SIAM J. Appl. Math. 47, 544-555. Kono, M., Ito, K., Niioka, T., Kadota, T., Sato, J. (1996). Current state of combustion research 36 in microgravity, Proc. Combust. Inst. 26, 1189-1199. Kostiuk, L. W., Cheng, R. K. (1995). The coupling of conical wrinkled flames with gravity, Combust. Flame 103, 27-40. Krivulin, V. N., Kudryavtsev, E. A., Baratov, A. N., Pavlova, V. L., Fedosov, L. N., Luzhetskii, V. K., Shlenov, V. M., Babkin, V. S. (1980). Combustion of near-limiting gas mixtures under conditions of weightlessness, Dokl. Phys. Chem. (Engl. Transl.) 247, 686-689. Krivulin, V. N., Kudryavtsev, E. A., Baratov, A. N., Badalyan, A. M., Babkin, V. S. (1981). Effect of acceleration on the limits of propagation of homogeneous gas mixtures, Combust. Expl. Shock Waves (Engl. Transl.) 17, 37-41. Lakshmisha, K. N., Paul, P. J., Rajan, N. K. S., Goyal, G., Mukunda, H. S. (1988). Behavior of methane-oxygen-nitrogen mixtures near flammability limits. Proc. Combust. Inst. 22, 15731580. Lakshmisha, K. N., Paul, P. J., Mukunda, H. S. (1990). On the flammability limit and heat loss in flames with detailed chemistry, Proc. Combust. Inst. 23, 433-440. Landau, L. D. (1944). Zhur. Eksp. Teor. Fiz. 14, 240. Law, C. K., Faeth, G. M. (1994). Opportunities and challenges of combustion in microgravity, Prog. Energy Combust. Sci. 20, 65-113. Lee, C., Buckmaster, J. D. (1991). The structure and stability of flame balls: a nearequidiffusional model, SIAM J. Appl. Math. 51, 1315-1326. Levy, A. (1965). An optical study of flammability limits, Proc. Roy. Soc. (London) A283, 134. Lewis, B., von Elbe, G. (1987). Combustion, Flames, and Explosions of Gases, 3rd ed., Academic Press, Orlando. Lewis, G. D., Combustion in a centrifugal-force field (1970). Proc. Combust. Inst. 13, 625-629. Libby, P. A. (1989). Theoretical analysis of the effect of gravity on premixed turbulent flames, Combust. Sci. Tech. 68, 15-33. Mallard, E., LeChatelier, H. L. (1880). C. R. Acad. Sci. Paris 98, 825. Marchese, A. J., Dryer, F. L. (1996). The effect of liquid mass transport on the combustion and extinction of bicomponent droplets of methanol and water, Combust. Flame 105, 104 -122. Maruta, K., Yoshida, M., Kobayashi, H., Niioka, T. (1996). Experimental study of methane-air premixed flame extinction at small stretch rates in microgravity, Proc. Combust. Inst. 26, 1283-1289. Maruta, K., Yoshida, M., Guo, H., Ju, Y., Niioka, T. (1998). Extinction of low-stretched diffusion flame in microgravity, Combust. Flame 112, 181-187. Patnaik, G., Kailasanath, K. (1992). Numerical simulation of the extinguishment of downward propagating flames, Proc. Combust. Inst. 24, 189-195. Patnaik, G., Kailasanath, K., Sinkovits, R. S. (1996). A new time-dependent, three-dimensional 37 flame model for laminar flames, Proc. Combust. Inst. 26, 899-905. Pearlman, H. (1997). Excitability in high-Lewis number premixed gas combustion, Combust. Flame 109, 382-398. Pearlman, H. (2000a). Low-temperature oxidation reactions and cool flames at earth gravity and microgravity. Combust. Flame 121, 390-393. Pearlman, H. (2000b). Cool Flames and Low Temperature Hydrocarbon Oxidation At ReducedGravity - An Experimental Study, submitted. Pearlman, H. G., Ronney, P. D. (1994). Near-limit behavior of high Lewis-number premixed flames in tubes at normal and low gravity, Phys. Fluids 6, 4009-4018. Pelcé, P., Clavin, P. (1982). Influence of hydrodynamics and diffusion upon the stability limits of laminar premixed flames, J. Fluid Mech. 124, 219-237. Pfefferle, W. C., Pfeferle, L. D. (1986). Catalytically stabilized combustion. Prog. Energy Combust. Sci. 12, 25-41. Rakiv, Z., Sivashinsky, G. I. (1987). Instabilities in upward propagating flames, Combust. Sci. Tech. 54, 69-84. Ronney, P. D. (1985). Effect of gravity on laminar premixed gas combustion. II: Ignition and Extinction Phenomena, Effect of Gravity on Laminar Premixed Gas Combustion II: Ignition and Extinction Phenomena, Combust. Flame 62, 120-132. Ronney, P. D. (1988a). Effect of chemistry and transport properties on near-limit flames at microgravity, Combust. Sci. Tech. 59, 123-141. Ronney, P. D. (1988b). On the mechanisms of flame propagation limits and extinction processes at microgravity, Proc. Combust. Inst. 22, 1615-1623. Ronney, P. D. (1990). Near-limit flame structures at low Lewis number, Combust. Flame 82, 1. Ronney, P. D. (1995). Some open issues in premixed turbulent combustion, in: Modeling in Combustion Science (J. D. Buckmaster and T. Takeno, Eds.), Lecture Notes In Physics, Vol. 449, Springer-Verlag, Berlin, pp. 3-22. Ronney, P. D. (1998). Understanding Combustion Processes through Microgravity Research, Proc. Combust. Inst. 27, 2485-2506. Ronney, P. D. (1999). “Flame Structure Modification and Quenching By Turbulence,” Combustion Science and Technology (Japanese edition), Vol. 6 (Supplement), pp. 53-76. Ronney, P. D., Sivashinsky, G. I. (1989). A theoretical study of propagation and extinction of nonsteady spherical flame fronts, SIAM J. Appl. Math. 49, 1029-1046. Ronney, P. D., Wachman, H. Y. (1985). Effect of gravity on laminar premixed gas combustion. I: Flammability limits and burning velocities. Combust. Flame 62, 107-119. Ronney, P. D., Whaling, K. N., Abbud-Madrid, A., Gatto, J. L., Pisowicz, V. L. (1994). Stationary premixed flames in spherical and cylindrical geometries, AIAA J. 32, 569-577. 38 Ronney, P. D., Wu, M. S., Weiland, K. J. and Pearlman, H. G. (1998). Structure Of Flame Balls At Low Lewis-number (SOFBALL): preliminary results from the STS-83 space flight experiments, AIAA Journal 36, 1361-1368. Sacksteder, K. R. (1990). The implications of experimentally-controlled gravitational accelerations for combustion science, Proc. Combust. Inst. 23, 1589-1597. Scott, S. K. (1997). Global behaviour in the oxidation of hydrogen, carbon monoxide and simple hydrocarbons" in Comprehensive Chemical Kinetics, Vol. 35, (M. J. Pilling, Ed.) p 439, Elsevier, Amsterdam, p. 439. Spalding, D. B. (1957). A theory of inflammability limits and flame quenching, Proc. Roy. Soc. (London) A240, 83-100. Stanglmaier, R. H. (1999). Homogeneous Charge Compression Ignition (HCCI): Benefits, Compromises, and Future Engine Applications, SAE paper no. 1999-01-3682. Strehlow, R. A., Reuss, D. L. (1981). Flammability limits in a standard tube, in: Combustion Experiments in a Zero Gravity Laboratory (T. H. Cochran, Ed.), Progress in Aeronautics and Astronautics 73, AIAA, New York, pp. 61-89. Strehlow, R. A., Noe, K. A., Wherley, B. L. (1986). The effect of gravity on premixed flame propagation and extinction in a vertical standard flammability tube, Twenty-First Symposium (International) on Combustion, Combustion Institute, pp. 1899-1908. Tyler, B. J. (1966). An Experimental Investigation of Conductive and Convective Heat Transfer during Exothermic Gas Phase Reactions. Combust. Flame 10, 90-91. Wang, Q., Ronney, P. D. (1993). Mechanisms of flame propagation limits in vertical tubes, Paper no. 45, Spring Technical Meeting, Combustion Institute, Eastern/Central States Section, March 15-17, 1993, New Orleans, LA. Warnatz, J. (1992). Resolution of gas phase and surface combustion into elementary reactions, Proc. Combust. Inst. 24, 553-579. Williams, F. A. (1985). Combustion Theory, 2nd ed., Benjamin-Cummins, Menlo Park. Wu, M.-S., Liu, J. B., Ronney, P. D. (1998). Numerical simulation of diluent effects on flame ball structure and dynamics, Proc. Combust. Inst. 27, 2543-2550. Wu, M. S., Ronney, P. D., Colantonio, R., VanZandt, D. (1999). Detailed numerical simulation of flame ball structure and dynamics, Combust. Flame 116, 387-397. Wu, M.-S., Ju, Y. and Ronney, P. D. (2000). Numerical simulation of flame balls with radiative reabsorption effects,” Paper No. 2000-0851, 38th AIAA Aerospace Sciences Meeting, Reno, NV, January 11-14, 2000. Zeldovich, Ya. B. (1944). Theory of Combustion and Detonation of Gases, Academy of Sciences (USSR), Moscow. 39 Time scale Chemistry (tchem) Buoyant, inviscid (tinv) Buoyant, viscous (tvis) Conduction (tcond) Radiation (trad) Stoichiometric flame 0.00094 s 0.071 s 0.012 s 1.04 s 0.13 s Near-limit flame 0.25 s 0.071 s 0.010 s 1.56 s 0.41 s Table 1. Estimates of time scales for stoichiometric and near-limit hydrocarbon-air flames at 1 atm pressure. 40 Mole % CH4 SL, µg, spherical expanding flame (Ronney and Wachman, 1985) SL, µg, Standard Flammability Limit Tube (Strehlow and Reuss, 1981) 5.87 6.8 6.4 5.63 5.0 4.9 5.45 3.8 4.0 5.33 3.0 3.7 5.22 2.4 2.2 5.10 1.7 1.7 5.07 1.5 --- Table 2. Comparison of burning velocities of near-limit CH4-air mixtures at atmospheric pressure measured at µg. 41 Figure captions Figure 1. Schematic of buoyancy-induced flame extinction in tubes. Left: upward propagation; right: downward propagation. Figure 2. Effect of 0.6 µm SiC particle loading on burning velocity and peak pressure in 5.25% CH4-air mixtures burning at µg (Abbud-Madrid and Ronney, 1993). Chamber volume is 12 liters, initial pressure 1 atm. Figure 3. Predicted values of burning velocity and peak flame temperature in CH4 - (0.21 O2 + 0.49 N2 + 0.30 CO2) mixtures under adiabatic conditions, with optically-thin radiative losses, and with radiation including reabsorption effects (Ju et al., 1998a). Figure 4. Diagram of non-dimensional growth rate for diffusive-thermal, Darrieus-Landau (thermal expansion) and Rayleigh-Taylor (buoyancy-driven) instability mechanisms as a function of wavenumber for an idealized downward propagating flame with b = 10, Le = 0.5, e = 0.8 and ga/SL3 = 0.5. Figure 5. Images of expanding cellular flame fronts in H2-air mixtures at µg (Ronney, 1990). Left: 7.0% H2 in air, 0.4% CF3Br added for improved visibility, 1.18 s after ignition. Right: 7.0% H2 in air, 2.25% CF3Br, 1.67 s after ignition. Field of view is 18 cm x 18 cm in both images. Figure 6. Images of flames propagating in tubes in high Lewis number mixtures. All flames were ignited at the open end of the tube and propagating towards the closed end. Images courtesy of Howard Pearlman. (a) and (b) are axial views, (c) is a radial view. (a) Spiral wave pattern, 0.80% octane - 21.00% oxygen - 78.20% helium, 28.5 cm diameter tube, downward propagation at 1g (b) Target pattern, 1.46% octane - 21.00% oxygen - 77.54% helium, 28.5 cm diameter tube, downward propagation at 1g (c) Spinning 6-arm flame, 1.21% butane - 21.00% oxygen - 77.79% helium, 14.3 cm diameter tube, microgravity (Pearlman and Ronney, 1994). Figure 7. Predicted evolution of non-dimensional flame radius (R) as a function of nondimensional time (t) for expanding spherical flames in a mixture with Le < 1 according to Eq. (12), for various values of the non-dimensional heat loss (Q). 42 Figure 8. Measured minimum ignition energies at 1g and µg as a function of fuel concentration and extinction radii of spherically expanding flames (dashed curves) at µg in mixtures outside the µg flammability limit. Numbers in parenthesis refer to energy release before extinction. Mixtures: NH3-air at 1 atm initial pressure (Ronney, 1988a). Figure 9. Measured flame radius as a function of time for expanding spherical flames at µg in 5.07% CH4-air mixtures at 1 atm for varying values of spark ignition energy (Ronney, 1985). Figure 10. Measured and predicted extinction strain rates for strained premixed CH4-air flames at 1g and µg (Guo et al., 1997). Figure 11. Measured and predicted extinction strain rates for strained premixed C3H8-air flames at 1g and µg (Ju et al., 1998b). Figure 12. Predicted extinction branches for strained CH4-O2-N2-He mixtures (Le ≈ 1.2) with radiative heat loss (Ju et al., 1999). Figure 13. Schematic diagram of a flame ball, illustrated for the case of fuel-limited combustion at the reaction zone. The oxygen profile is similar to the fuel profile except its concentration is nonzero in the interior of the ball. The combustion product profile is identical to the temperature profile except for a scale factor. Figure 14. Enhanced-contrast images of flame balls obtained during the STS-94 space flight experiments (flame balls may appear to have widely varying sizes because of varying distances from the camera). Figure 15. Effect of heat loss (Q) on flame ball radius (R) and stability properties (Buckmaster et al., 1990). Figure 16. Computed effect of mixture strength on flame ball radii in lean H2-air mixtures (Buckmaster et al., 1993). Figure 17. Computed dynamical properties of flame balls in lean H2-air mixtures - eventual fate as a function of initial radius (ro) normalized by the steady radius (r*) (Wu et al., 1999). 43 Figure 18. Predicted flame ball radii and radiant power for H2-air mixtures using three popular chemical models (Wu et al., 1998). Figure 19. Comparison of computed properties of propagating flames for 3 different H2-O2 chemical mechanisms - burning velocity (SL) as a function of equivalence ratio in H2-air mixtures (Wu et al., 1998). A compilation of experimental results from several sources is also shown. Figure 20. Computed flame ball radius at the location of maximum volumetric heat release as a function of H2 mole fraction for steady flame balls for H2-O2-CO2 mixtures with H2:O2 = 1:2 (Wu et al., 1998). Preliminary experimental results from MSL-1 are also shown (filled circles). Figure 21. Autoignition and cool flame behavior of a 67% n-C4H10- 33% O2 mixture at an initial temperature 310oC and an initial pressure of 4.2 psia at 1g (left) and µg (right) (Pearlman, 2000b). 44 Buoyancy-induced flame stretch Cooling combustion products near wall cause sinking boundary layer Direction of flame propagation Direction of flame propagation Flame front Tube walls Tube walls Figure 1. Schematic of buoyancy-induced flame extinction in tubes. Left: upward propagation; right: downward propagation. 45 4.2 2.50 Burning velocity (cm/s) 4 2.25 3.8 Burning velocity 3.6 2.00 3.4 1.75 Maximum pressure (atm) Pressure 3.2 1.50 3 0 2 4 6 Mass of particles in chamber (g) 8 Figure 2. Effect of 0.6 µm SiC particle loading on burning velocity and peak pressure in 5.25% CH4-air mixtures burning at µg (Abbud-Madrid and Ronney, 1993). Chamber volume is 12 liters, initial pressure 1 atm. 46 Figure 3. Predicted values of burning velocity and peak flame temperature in CH4 - (0.21 O2 + 0.49 N2 + 0.30 CO2) mixtures under adiabatic conditions, with optically-thin radiative losses, and with radiation including reabsorption effects (Ju et al., 1998a). 47 Diffusive-thermal Thermal expansion Buoyancy Total 0.075 Growth rate ( s) 0.050 0.025 0.000 -0.025 -0.050 0 0.1 0.2 0.3 0.4 0.5 Wave number (k) 0.6 0.7 0.8 Figure 4. Diagram of non-dimensional growth rate for diffusive-thermal, Darrieus-Landau (thermal expansion) and Rayleigh-Taylor (buoyancy-driven) instability mechanisms as a function of wavenumber for an idealized downward propagating flame with b = 10, Le = 0.5, e = 0.8 and ga/SL3 = 0.5. 48 Figure 5. Images of expanding cellular flame fronts in H2-air mixtures at µg (Ronney, 1990). Left: 7.0% H2 in air, 0.4% CF3Br added for improved visibility, 1.18 s after ignition. Right: 7.0% H2 in air, 2.25% CF3Br, 1.67 s after ignition. Field of view is 18 cm x 18 cm in both images. 49 (a) (b) (c) Figure 6. Images of flames propagating in tubes in high Lewis number mixtures. All flames were ignited at the open end of the tube and propagating towards the closed end. Images courtesy of Howard Pearlman. (a) and (b) are axial views, (c) is a radial view. (a) Spiral wave pattern, 0.80% octane- 21.00% oxygen - 78.20% helium, 28.5 cm diameter tube, downward propagation at 1g (b) Target pattern, 1.46% octane- 21.00% oxygen - 77.54% helium, 28.5 cm diameter tube, downward propagation at 1g (c) Spinning 6-arm flame, 1.21% butane - 21.00% oxygen - 77.79% helium, 14.3 cm diameter tube, microgravity (Pearlman and Ronney, 1994). 50 80 Q = 0.36 Dimensionless radius (R) 70 60 Q = 0.4 50 40 Q = 0.42 30 Q = 0.45 20 Q = 0.5 10 0 0 20 40 60 Dimensionless time ( t) 80 100 Figure 7. Predicted evolution of non-dimensional flame radius (R) as a function of nondimensional time (t) for expanding spherical flames in a mixture with Le < 1 according to Eq. (12), for various values of the non-dimensional heat loss (Q). 51 Figure 8. Measured minimum ignition energies at 1g and µg as a function of fuel concentration and extinction radii of spherically expanding flames (dashed curves) at µg in mixtures outside the µg flammability limit. Numbers in parenthesis refer to energy release before extinction. Mixtures: NH3-air at 1 atm initial pressure (Ronney, 1988a). 52 Figure 9. Measured flame radius as a function of time for expanding spherical flames at µg in 5.07% CH4-air mixtures at 1 atm for varying values of spark ignition energy (Ronney, 1985). 53 Figure 10. Measured and predicted extinction strain rates for strained premixed CH4-air flames at 1g and µg (Guo et al., 1997). 54 Figure 11. Measured and predicted extinction strain rates for strained premixed C3H8-air flames at 1g and µg (Ju et al., 1998b). 55 Figure 12. Predicted extinction branches for strained CH4-O2-N2-He (Le ≈ 1.2) mixtures with radiative heat loss (Ju et al., 1999). 56 T* C ~ 1-1/r Temperature Fuel concentration T ~ 1/r To Interior filled with combustion products Reaction zone Heat & products diffuse outward Fuel & oxygen diffuse inward Figure 13. Schematic diagram of a flame ball, illustrated for the case of fuel-limited combustion at the reaction zone. The oxygen profile is similar to the fuel profile except its concentration is nonzero in the interior of the ball. The combustion product profile is identical to the temperature profile except for a scale factor. 57 3.70% H2 - 96.30% air, 500 s after ignition, diameter 1.0 cm. 3.57% H2 - 96.43% air, 300 s after ignition, diameters 1.0 cm. 5.20% H2 - 10.40% O2 - 84.40% CO2, 50 s after ignition, diameters 0.7 cm. 4.00% H2 - 96.00% air, 15 s after ignition, diameters 1.0 -1.8 cm. Figure 14. Enhanced-contrast images of flame balls obtained during the STS-94 space flight experiments (flame balls may appear to have widely varying sizes because of varying distances from the camera). 58 Dimensionless flame ball radius (R) 15 Unstable to 3-d disturbances 10 Stable Equation of curve: 5 0 0 -2 R ln(R) = Q Unstable to 1-d disturbances 0.05 0.1 0.15 Dimensionless heat loss (Q) 0.2 Figure 15. Effect of heat loss (Q) on flame ball radius (R) and stability properties (Buckmaster et al., 1990). 59 1 Radius (cm) Cold Giant 0.1 Hot Dwarf 0.01 0.001 0.08 0.09 0.1 0.11 Equivalence ratio 0.12 0.13 Figure 16. Computed effect of mixture strength on flame ball radii in lean H2-air mixtures (Buckmaster et al., 1993). 60 Quench Steady flame ball No stable flames Planar flames possible Quench 1 0.1 No flames possible Stretching parameter (c) = or/r* 10 Steady flame balls Steady flame balls Quench Quench 0.08 0.1 0.2 0.3 Equivalence ratio (f ) Figure 17. Computed dynamical properties of flame balls in lean H2-air mixtures - eventual fate as a function of initial radius (ro) normalized by the steady radius (r*) (Wu et al., 1999). 61 10 Upper curves: radius Lower curves: radiant power 8 0.6 6 0.4 4 Yetter 0.2 2 Peters GRI 0 Radiant power (W) Flame ball radius at max OH concentration 0.8 3 3.5 4 4.5 Mole percent H in air 5 0 2 Figure 18. Predicted flame ball radii and radiant power for H2-air mixtures using three popular chemical models (Wu et al., 1998). 62 400 8 GRI chemical mechanism 350 250 4 w/o H O 2 radiation 2 w/ H O 2 radiation 0 200 0.26 0.28 0.3 0.32 0.34 L S (cm/s) 300 6 150 Yetter GRI Peters Expt. 100 50 0 0.1 1 Equivalence ratio 10 Figure 19. Comparison of computed properties of propagating flames for 3 different H2-O2 chemical mechanisms - burning velocity (SL) as a function of equivalence ratio in H2-air mixtures (Wu et al., 1998). A compilation of experimental results from several sources is also shown. 63 Flame ball radius (cm) 1.6 1.2 predictions (w/o CO radiation) 2 0.8 predictions (w/ CO radiation) 2 0.4 0 2 experiments 4 6 8 10 Mole % H 12 14 16 2 Figure 20. Computed flame ball radius at the location of maximum volumetric heat release as a function of H2 mole fraction for steady flame balls for H2-O2-CO2 mixtures with H2:O2 = 1:2 (Wu et al., 1998). Preliminary experimental results from MSL-1 are also shown (filled circles). 64 Cool Flames 5 Pressure 320 T1 315 T2 3 310 T3 2 305 1 300 0 295 -1 290 0 50 100 Time (s) 150 -2 200 325 5 Cool Flame 320 Pressure 315 T (o C) P (psia) 4 4 T1 3 T2 310 2 T3 305 300 1 0 Acceleration 295 -1 290 P (psia), Acceleration (g/gearth) T (o C) 325 -2 10 15 20 30 25 Time (s) 35 40 45 Figure 21. Autoignition and cool flame behavior of a 67% n-C4H10- 33% O2 mixture at an initial temperature 310oC and an initial pressure of 4.2 psia at 1g (upper) and µg (lower) (Pearlman, 2000b). 65 NOMENCLATURE aP A CP d D E g go Grd h LI Le P r rf R Rg SL SL,lim Sc t chem t inv t rad t vis T T ad u’ U Uy Y Planck mean absorption coefficient flame surface area constant-pressure heat capacity characteristic flow length scale or tube diameter mass diffusivity overall activation energy of the heat-release reactions acceleration of gravity earth gravity Grashof number based on characteristic length scale (d) = gd3/n2 heat transfer coefficient in a cylindrical tube = 16l/d2 turbulence integral scale Lewis number (a/D = thermal diffusivity / reactant mass diffusivity) pressure radial coordinate flame radius scaled flame radius (Eq. (12)) gas constant premixed laminar burning velocity burning velocity at the flammability limit Schmidt number = n/D chemical time scale inviscid buoyant transport time scale radiative loss time scale viscous buoyant transport time scale temperature adiabatic flame temperature turbulence intensity convection velocity local axial velocity in counterflow configuration fuel mass faction a b d g l L n r s S thermal diffusivity non-dimensional activation energy = E/RgT f flame thickness gas specific heat ratio thermal conductivity radiative heat loss per unit volume = 4saP(Tf4 - T∞4) kinematic viscosity density Stefan-Boltzman constant flame stretch rate 66 Subscripts f ∞ flame front condition ambient conditions 67