Symposium (International) on Combustion最新文献

Among many other methods for simplifying chemical kinetics for laminar and turbulent flame calculations, the method of intrinsic low-dimensional manifolds (ILDM) has shown to be an efficient tool for the development of reduced kinetic schemes. Based on a numerical analysis, it identifies and decouples the fast relaxing timescales of the chemical system. The results, for example, the thermokinetic state of the system or the reaction rates, are then stored in terms of a small number of parameters (mixture fraction, reaction progress variables) for subsequent use in reacting flow calculations. Furthermore, together with the reduced mechanism, information about the coupling of the chemical kinetics with the physical processes (molecular transport, turbulent mixing) is obtained.

In this paper, we present a method that allows an efficient implementation of the ILDM method in flame calculations and overcomes several problems that had been discussed in previous work. It is based on three ingredients: A robust numerical method to calculate the ILDM, a storage scheme that allows an easy implementation in CFD codes, and a model for the coupling of the chemical kinetics with transport processes. In this way, not only the chemistry can be calculated beforehand but also a reduced set of diffusion coefficients. The method is verified by simulations of laminar syngas-air flames with an emphasis on the non-premixed case.

This paper considers ignition and combustion of small (50–100 μm) single particles of magnesium and 50-50 magnesium-aluminum alloy in the atmosphere of carbon dioxide or its mixtures with argon. This investigation is of interest for both basic combustion science and applications to rocket engines, including those using Martian CO₂ as an oxidizer, An experimental setup with an electrodynamic levitator inside a high-pressure chamber was employed. A CO₂ laser was used for heating to ignition of the particles. The laser was switched off after ignition. The experiments were conducted with the oxidizer at room temperature over the range of pressures from 0.1 to 2 MPa. Effects of the CO₂ concentration and pressure on the critical ignition conditions, ignition delay times, and burning times have been determined for Mg particles. The results clearly indicate that ignition of Mg in CO₂ is controlled by chemical kinetics and that its combustion is controlled by diffusion in gas phase. Quantitative disagreement of the observed critical ignition pressures with previous experimental data on ignition of Mg disks in CO₂ is explained by the differences in heat-loss mechanisms. The measured values of the burning rate correlate well with previous experimental results on combustion of 2-mm particles and with a quasi-steady model of Mg particle burning in CO₂. In contrast to pure Mg and Al, particles of Mg−Al alloy did not ignite in CO₂ under the present experimental conditions.

Instantaneous measurements of temperature, major species, OH, CO, and NO are performed in turbulent premixed flames using simultaneous Rayleigh scattering, Raman scattering, and laser-induced fluorescence (LIF). Temperature is determined from Rayleigh scattering, and concentrations of CH₄, O₂, N₂, H₂O, CO₂, and H₂ are obtained from Raman scattering. Linear LIF is used to measure OH and NO, while two-photon LIF is used to determine CO concentrations. The two-photon CO-LIF system provides significant improvements over CO-Raman measurements. This combination of diagnostic techniques is used to investigate the detailed compositional structure of turbulent lean premixed methane-air flames. Three turbulent flames having different equivalence ratios are considered. The ratio of rms velocity to laminar flame speed ranges from approximately 6 to 17. The correlation of major species, OH, and CO concentrations with temperature are similar to those predicted by one-dimensional laminar flame calculations. However, NO concentrations in the leaner flames are higher than the predicted values.

The bimolecular reaction HO+CO ixH+CO₂ involves the intermediate formation of HOCO. As a consequence, the rate coefficient shows a complicated temperature and pressure dependence. An optimized E- and J-resolved rigid activated complex RRKM theory, with simplified E- and J-resolved pressure-dependent collision efficiencies, fits the available experimental data and allows for extrapolations to unexplored conditions. Experiments between 80 and 2370 K, between 10⁻³ and 10³ bar in the bath gas He, and below 1 bar in Ar, N₂, CF₄ SF₆, and H₂O at 298 K, serve as the database. A limiting low-pressure rate constant for HO removal of k_o=[1.0×10¹³ exp(−8050 K/T)+9.0×10¹¹ exp(−2300 K/T)+1.01×10¹¹ exp (−30 K/T)] cm³ mol⁻¹ s⁻¹ and a limiting high-pressure rate constant of k_∞=[1.23×10¹⁵ exp (−7520 K/T)+1.1×10¹³ exp(−1850 K/T)+8.0×10¹¹ exp(−120 K/T] cm³ mol⁻¹ s⁻¹ will reproduce the results. The pressure dependence of the rate coefficient as a function of the temperature is represented for the bath gases He, Ar, N₂, CF₄, SF₆, and H₂O. Rate coefficients for HOCO formation and HOCO dissociation are also given.

This paper is a continuation of our work on edge-flames in premixed combustion. An edge-flame is a two-dimensional structure constructed from a one-dimensional configuration that has two stable solutions (bistable equilibrium). Edge-flames can display wavelike behavior, advancing as ignition fronts or retreating as failure waves. Here we consider two one-dimensional configurations: twin deflagrations in a straining flow generated by the counterflow of fresh streams of mixture; and a single deflagration subject to radiation losses. The edge-flames constructed from the first configuration have positive or negative speeds, according to the value of the strain rate. But our numerical solutions strongly suggest that only positive speeds (corresponding to ignition fronts) can exist for the second configuration. We show that this phenomenon can also occur in diffusion flames when the Lewis numbers are small. And we discuss the asymptotics of the one-dimensional twin deflagration configuration, an overlooked problem from the 70s.

By burning droplets of benzene in a single-droplet combustor and performing phase-discriminating sampling of the liquid and gas phases of the droplet system, we have found that gas-phase pyrolysis products arise in the liquid phase of the droplet. The experiments are conducted at 1000 K and 21 mol % O₂ in the postcombustion gas from an oxygen-rich premixed methane flame. Disruptive burning, which has not previously been reported for a pure hydrocarbon in normal gravity conditions, is observed at the end of the droplet residence time (∼92 ms). Samples of the liquid phase have been taken at various times throughout the combustion lifetime and analyzed by high-pressure liquid chromatography. Compositional analysis using ultraviolet-visible absorbance spectra of the separated components of the samples reveals a wide variety of pure polycyclic aromatic hydrocarbons (PAH), substituted PAH, and cyclopenta-fused PAH. In addition, recent synthesis of new reference standards has enabled identification of cyclopenta-fused PAH—cyclopent[hi]acephenanthrylene, cyclopenta[cd]fluoranthene, and dicyclopenta[cd, jk]pyrene—which have never before been identified as benzene products. Because the droplet remains relatively cold (∼350 K) with respect to the gas phase in the oxygen-deficient zone between the droplet and the flame (∼2000 K), we conclude that these compounds are gas-phase pyrolysis products that are obsorbed into the droplet, rather than products of reactions within the droplet. These heavier species may play a role in observed terminal disruptive burning events by acting as additional droplet components that promote multicomponent effects. Analysis of species concentrations over time reveals the dominance of both ring rupture pyrolysis products such as phenylacetylene, triacetylene, and acenaphthylene, and biaryl pyrolysis products such as biphenyl. These four products in particular represent 70% of the identified mass of absorbed pyrolysis products, which accounts for up to 5% of the droplet mass at the end of its lifetime.

An experimental investigation of the flame response to strain rate in the case of unsteady premixed low-turbulent flames is presented. In order to point out the fundamental aspects of the mutual interaction between combustion and turbulence, measurements of local flame properties (curvature, displacement speed) and tangential strain rate were performed under varying conditions of Lewis number and turbulence.

Three different mixtures (methane/air, propane/air, and hydrogen/air) were successively spark ignited in a vertical wind tunnel. The expanding flame freely propagated in a grid-generated decaying turbulent flow. An advanced field imaging technique coupling high-speed laser tomography and cross-correlation particle image velocimetry (PIV) was used to measure the temporal evolution of local flame stretch exerted by the turbulent cold flow.

Local flame curvature and local displacement speed were calculated from flame-front contours. Curvature probability density functions (PDFs) were negatively skewed, especially for nonunity Lewis numbers, and displacement speed distributions underlined the influence of local stretch and thermodiffusive effects on flame-speed variations. Tangential strain rate was determined by using the velocity field in the neighborhood of the flame front and appears to be independent of the Lewis numbers. A strong correlation between local flame curvature and tangential strain rate was demonstrated, underlining the cold flow effects on the local flame structure. The influences of turbulence and Lewis number were evaluated and compared with numerical simulations. Then, local flame stretch distributions were determined versus time, indicating that a significant proportion of the flame was under compression.

An extended Landau-Levich model of liquid-propellant combustion, one that allows for a local dependence of the burning rate on the (gas) pressure at the liquid-gas interface, exhibits not only the classival hydrodynamic cellular instability attributed to Landau but also a pulsating hydrodynamic instability associated with sufficiently negative pressure sensitivities. Exploiting the realistic limit of small values of the gas-to-liquid density ratio p, analytical formulas for both neutral sability boundaries may be obtained by expanding all quatities in appropriate powers of p in each of three distinguished wave-number regimes. In particular, composite analytical expressions are derived for the neutral stability boundaries A_p(k), where A_p is the pressure sensitivity of the burning rate and k is the wave number of the disturbance. For the cellular boundary, the results demonstrate explicitly the stabilizing effect of gravity on long-wave disturbances, the stabilizing effect of viscosity (both liquid and gas) and surface tension on short-wave perturbations, and the instability associated with intermediate wave numbers for negative values of A_p, which is characteristic of many hydroxylammoninum nitrate-based liquid propellants over certain pressure ranges. In contrast, the pulsating hydrodynamic stability boundary is insensitive to gravitational and surface-tension effects but is more sensitive to the effects of liquid viscosity because, for typical nonzero values of the latter, the pulsating boundary decreases to larger negative values of A_p as k increases through O(1) values. Thus, liquid-propellant combustion is predicted to be stable (that is, stealy and planar) only for a range of negative pressure sensitivities that lie below the cellular boundary that exists for sufficiently small negative values of A_p and above the pulsating boundary that exists for larger negative values of this parameter.

Wire-stabilized premixed methane-air flames have been studied in a grid-generated homogeneous turbulent flow field in order to identify different burning regimes. The planar Rayleigh scattering technique was used with two parallel laser light sheets, which allows the detection of three-dimensional temperature gradients. For a detailed investigation of the flame structure and topology, the modification of the local temperature gradients at different progress variables c due to the turbulent motion was studied by varying the flame stoichiometry and thereby the Karlovitz number Ka while keeping the turbulent Reynolds number Re_t constant at 87 or 134. Because of a nearly Gaussian shaped statistical distribution of the thermal gradients, the 50% median and the width of the distribution are suitable measures used to characterize the flame response. Compared with laminar unstrained calculations, especially very lean flames (<0.55) marked with the highest Karlovitz number (Ka=4.6) revealed a reduction of the flame thickness of about 30%. This is in contrast to the expected burning regime but fits well with laminar strained calculations. Subsequently, detailed investigations were made to examine the influence of curvature on local thermal gradients. It was found that negatively curved cusps (concave toward the reactants) show a steepening of the flame-temperature profile, while positively curved flame elements can be identified by a retardation of the overall reaction process. In terms of a statistical examination, the widths of the thermal gradient distribution conditioned at different reaction progress variables c were regarded, finding a decrease of the spread with increasing Ka independent of Re_t and c. Based on different curvature radii and perturbation frequencies of the detected flames, we assume that in our experiments the flame response depends more on flame curvature than on effects caused by modification of Ka.