SEARCH · Search NASA
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
In modeling turbulent combustion, decoupling the chemistry from the turbulence is of great practical significance. In cases in which the equilibrium chemistry model breaks down, laminar flamelet modeling (LFM) is a promising approach to decoupling. Here, the validity of this approach is investigated using direct numerical simulation of a simple chemical reaction in two-dimensional turbulence.
Future aerospace propulsion concepts involve the combination of liquid or gaseous fuels in a highly turbulent internal air stream. Accurate predictive computer codes which can simulate the fluid mechanics, chemistry, and turbulence combustion interaction of these chemical reacting flows will be a new tool that is needed in the design of these future propulsion concepts. Experimental and code development research is being performed at Lewis to better understand chemical reacting flows with the long term goal of establishing these reliable computer codes. The approach to understanding chemical reacting flows is to look at separate simple parts of this complex phenomena as well as to study the full turbulent reacting flow process. As a result research on the fluid mechanics associated with chemical reacting flows was initiated. The chemistry of fuel-air combustion is also being studied. Finally, the phenomena of turbulence-combustion interaction is being investigated. This presentation will highlight research, both experimental and analytical, in each of these three major areas.
The secondary effects in turbulent combustion instabilities leading to flashback are investigated, including those due to buoyancy and contraction at the combustor outlet. Experiments were conducted in an oblong, rectangular cross-section combustion tunnel, where the effects of a bluff-body flames holder were generated by a rear-facing step behind a streamlined inlet nozzle. The results of experiments leading to flashback with the step mounted at the bottom of the combustion chamber were compared to those of experiments in which it was located at the top. Irrespective of the flow obstructions introduced downstream, the critical equivalence ratio for flashback was consistently lower with the step at the bottom, indicating that buoyancy was enhancing the growth of the recirculation zone that pushed the flame upstream and caused flashback. The contraction at the end of the combustion chamber had a promoting influence on the process of vortex pairing, re-enforcing the influence of the trailing vortices over that of the recirculation vortex system, and thereby curbing the tendency to flashback. Provided that the flow velocity was low, however, the characteristic features of combustion instabilities leading to flashback in the absence of contraction could still be established in its presence.
In combustion computations, it is known that the predictions of chemical reaction rates are poor if conventional turbulence models are used. The probability density function (pdf) method seems to be the only alternative that uses local instantaneous values of the temperature, density, etc., in predicting chemical reaction rates, and thus is the only viable approach for more accurate turbulent combustion calculations. The fact that the pdf equation has a very large dimensionality renders finite difference schemes extremely demanding on computer memories and thus impractical. A logical alternative is the Monte Carlo scheme. Since CFD has a certain maturity as well as acceptance, it seems that the use of a combined CFD and Monte Carlo scheme is more beneficial. Therefore, a scheme is chosen that uses a conventional CFD flow solver in calculating the flow field properties such as velocity, pressure, etc., while the chemical reaction part is solved using a Monte Carlo scheme. The discharge of a heated turbulent plane jet into quiescent air was studied. Experimental data for this problem shows that when the temperature difference between the jet and the surrounding air is small, buoyancy effect can be neglected and the temperature can be treated as a passive scalar. The fact that jet flows have a self-similar solution lends convenience in the modeling study. Futhermore, the existence of experimental data for turbulent shear stress and temperature variance make the case ideal for the testing of pdf models wherein these values can be directly evaluated.
Briefs of the 1994 annual progress reports of the Research Fellows and students of the Center for Turbulence Research are presented. Subjects covered include turbulence combustion, large eddy simulation, Reynolds-averaged turbulence modeling, turbulence control, postprocessing, sound generation, and turbulence physics.
Acoustic emission spectra have been obtained for non-premixed turbulent combustion from two small diameter laboratory gas burners, two commercial gas burners and a large gas burner in the firebox of a Babcock-Wilcox Boiler (50,000 lb steam/hr). The changes in burner size and firing rate represent changes in Reynolds number and changes in air/fuel ratio represent departure from stoichiometric proportions. The combustion efficiency was measured independently through gas analysis. The acoustic spectra obtained from the various burners exhibit a persistent shape over the Reynolds number range of 8200-82,000. The spectra were analyzed for identification of a predictable frequency domain that is most responsive to, and readily correlated with, combustion efficiency. A simple parameter (consisting of the ratio of the average acoustic power output in the most responsive frequency bandwidth to the acoustic power level of the loudest frequency) is proposed whose value increases significantly and unmistakably as combustion efficiency approaches 100%. The dependence of the most responsive frequency domain on the various Reynolds numbers associated with turbulent jets is discussed.
This paper provides a detailed discussion of the theory and application of probability density function (pdf) methods, which provide a complete statistical description of turbulent flow fields at a single point or a finite number of points. The basic laws governing the flow of Newtonian fluids are set up in the Eulerian and the Lagrangian frame, and the exact and linear equations for the characteristic functionals in those frames are discussed. Pdf equations in both frames are derived as Fourier transforms of the equations of the characteristic functions. Possible formulations for the nonclosed terms in the pdf equation are discussed, their properties are assessed, and closure modes for the molecular-transport and the fluctuating pressure-gradient terms are reviewed. The application of pdf methods to turbulent combustion flows, supersonic flows, and the interaction of turbulence with shock waves is discussed.
Viewgraphs are presented on computation of turbulent combustion, governing equations, closure problem, PDF modeling of turbulent reactive flows, validation cases, current projects, and collaboration with industry and technology transfer.
Turbulent combustion can not be simulated adequately by conventional moment closure turbulence models. The difficulty lies in the fact that the reaction rate is in general an exponential function of the temperature, and the higher order correlations in the conventional moment closure models of the chemical source term can not be neglected, making the applications of such models impractical. The probability density function (pdf) method offers an attractive alternative: in a pdf model, the chemical source terms are closed and do not require additional models. A grid dependent Monte Carlo scheme was studied, since it is a logical alternative, wherein the number of computer operations increases only linearly with the increase of number of independent variables, as compared to the exponential increase in a conventional finite difference scheme. A new algorithm was devised that satisfies a restriction in the case of pure diffusion or uniform flow problems. Although for nonuniform flows absolute conservation seems impossible, the present scheme has reduced the error considerably.
Turbulent combustion can not be simulated adequately by conventional moment closure turbulent models. The probability density function (PDF) method offers an attractive alternative: in a PDF model, the chemical source terms are closed and do not require additional models. Because the number of computational operations grows only linearly in the Monte Carlo scheme, it is chosen over finite differencing schemes. A grid dependent Monte Carlo scheme following J.Y. Chen and W. Kollmann has been studied in the present work. It was found that in order to conserve the mass fractions absolutely, one needs to add further restrictions to the scheme, namely alpha(sub j) + gamma(sub j) = alpha(sub j - 1) + gamma(sub j + 1). A new algorithm was devised that satisfied this restriction in the case of pure diffusion or uniform flow problems. Using examples, it is shown that absolute conservation can be achieved. Although for non-uniform flows absolute conservation seems impossible, the present scheme has reduced the error considerably.
A unique experiment makes it now possible to obtain detailed images in microgravity showing how an individual vortex causes the wrinkling, stretching, area increase, and eventual extinction of a premixed flame. The repeatable, controllable flame-vortex interaction represents the fundamental building block of turbulent combustion concepts. New information is provided that is central to turbulent flame models, including measurements of all components of flame stretch, strain, and vorticity. Simultaneous measurements of all components of these quantities are not possible in fully turbulent flames but are possible in the present axisymmetric, repeatable experiment. Advanced PIV diagnostics have been used at one-g and have been developed for microgravity. Numerical simulations of the interaction are being performed at NRL. It is found that microgravity conditions greatly augment the flame wrinkling process. Flame area and the amplitude of wrinkles at zero-g are typically twice that observed at one-g. It is inferred that turbulent flames in microgravity could have larger surface area and thus propagate significantly faster than those in one-g, which is a potential safety hazard. A new mechanism is identified by PIV images that shows how buoyancy retards flame wrinkling at one-g; buoyancy produces new vorticity (due to baroclinic torques) that oppose the wrinkling and the stretch imposed by the original vortex. Microgravity conditions remove this stabilizing mechanism and the amplitude of flame wrinkling typically is found to double. Microgravity also increases the flame speed by a factor of 1.8 to 2.2. Both methane and propane-air flames were studied at the NASA Lewis drop tower. Results indicate that it is important to add buoyancy to models of turbulent flames to simulate the correct flame wrinkling, stretch and burning velocity.
Although the different regimes of premixed combustion are not well defined, most of the recent developments in turbulent combustion modeling are led in the so-called flamelet regime. The goal of these models is to give a realistic expression to the mean reaction rate (w). Several methods can be used to estimate (w). Bray and coworkers (Libby & Bray 1980, Bray 1985, Bray & Libby 1986) express the instantaneous reaction rate by means of a flamelet library and a frequency which describes the local interaction between the laminar flamelets and the turbulent flowfield. In another way, the mean reaction rate can be directly connected to the flame surface density (Sigma). This quantity can be given by the transport equation of the coherent flame model initially proposed by Marble & Broadwell 1977 and developed elsewhere. The mean reaction rate, (w), can also be estimated thanks to the evolution of an arbitrary scalar field G(x, t) = G(sub O) which represents the flame sheet. G(x, t) is obtained from the G-equation proposed by Williams 1985, Kerstein et al. 1988 and Peters 1993. Another possibility proposed in a recent study by Mantel & Borghi 1991, where a transport equation for the mean dissipation rate (epsilon(sub c)) of the progress variable c is used to determine (w). In their model, Mantel & Borghi 1991 considered a medium with constant density and constant diffusivity in the determination of the transport equation for (epsilon(sub c)). A comparison of different flamelet models made by Duclos et al. 1993 shows the realistic behavior of this model even in the case of constant density. Our objective in this present report is to present preliminary results on the study of this equation in the case of variable density and variable diffusivity. Assumptions of constant pressure and a Lewis number equal to unity allow us to significantly simplify the equation. A systematic order of magnitude analysis based on adequate scale relations is performed on each term of the equation. As in the case of constant density and constant diffusivity, the effects of stretching of the scalar field by the turbulent strain field, of local curvature, and of chemical reactions are predominant. In this preliminary work, we suggest closure models for certain terms, which will be validated after comparisons with DNS data.
A hybrid Reynolds averaged/assumed probability density function (PDF) approach was developed and applied to the study of turbulent combustion in a supersonic mixing layer. This approach was used to address the laminar-like treatment of the thermochemical terms that appear in the conservation equations. Calculations were carried out for two experiments involving H2-air supersonic turbulent mixing. A SPARK combustion model was employed. In general, the approach shows modest improvement over previous calculations.
The present symposium on combustion discusses reaction kinetics, NO(x) kinetics, premixed, diffusion, and nonsteady flames, turbulent combustion, hazardous waste, fluidized bed combustion, coal boilers and furnaces, engines, heterogeneous kinetics, heterogeneous, droplet, and microgravity combustion, and high-temperature synthesis. Attention is given to reactions of biphenyl, methylnaphthalenes, and phenanthrene with atomic oxygen in the gas phase, the oxidation of ortho-xylene, the effects of water on combustion kinetics at high pressure, and the formation and measurement of N2O in combustion systems. Topics addressed include large ions in premixed benzene-oxygen flames, the structure and kinetics of CH4/N2O flames, the propagation of unsteady hydrogen premixed flames near flammability limits, and the basic structure of lean propane flames. Also considered are OH measurements of piloted diffusion flames of nitrogen-diluted methane near extinction, waste combustion, preferential oxidation of carbon surfaces, and reburning mechanisms in a pulverized coal combustor.
The study of flame-vortex interactions provides one of the means to better understand turbulent combustion, and allows for canonical configurations that contain the fundamental elements found in turbulent flames, These include concentrated vorticity, entrainment and mixing, strain and nonequilibrium phenomena, diffusion and differential diffusion, partial premixing and diluent effects, and heat release effects. In flame- vortex configurations, these fundamental elements can be studied under more controlled conditions than is possible in direct investigations of turbulent flames. Since the paper of Marble, the problem of the flame-vortex interaction has received considerable attention theoretically, numerically and experimentally. Several configurations exist for study of the premixed flame/vortex ring interaction but more limited results have been obtained to date for the diffusion flame/vortex ring case. The setup of Chen and Dahm, which is conceptually similar to that of Karagozian and Manda and Karagozian, Suganuma and Strom where the ring is composed of fuel and air and combustion begins during the ring formation process, is used in the current study. However, it is essential to conduct the experiments in microgravity to remove the asymmetries caused by buoyancy and thus obtain highly symmetric and repeatable interactions. In previous studies it was found that the flame structure of the vortex ring was similar to that obtained analytically by Karagozian and Manda. Dilution of propane with nitrogen led mainly to a reduction in flame luminosities, flame burnout times were affected by both fuel volumes and amount of dilution, and a simple model of the burnout times was developed. In this paper, a discussion on reacting ring displacement and flame burnout time will be given, and the flame structures of vortex rings containing ethane and air will be compared to those of propane reacting in air.
This study involves the development of numerical and physical modeling in spray combustion. These modeling efforts are mainly motivated to improve the physical submodels of turbulence, combustion, atomization, dense spray effects, and group vaporization. The present mathematical formulation can be easily implemented in any time-marching multiple pressure correction methodologies such as MAST code. A sequence of validation cases includes the nonevaporating, evaporating and_burnin dense_sprays.
Planar laser-induced fluorescence (PLIF) excitation/detection methods have been applied to obtain spatial distributions of the hydroxyl [OH] reacting intermediary and hydrocarbon [HC] primary species in laminar and turbulent combustion reactions. In this report, broadband and narrowband excitation/filtering techniques are explored to identify an optimal experimental configuration yielding significant fluorescent signal with low absorption losses. The combustion environments analyzed include 1) a laminar non-premixed methane/air flame and 2) a turbulent, non-premixed Jet-A/air fueled flame within a lean flame tube combustor. Hydrocarbon-based fuel and OH were excited via the R1 (1), R1(10) and R2(7) transitions of the A(sup 2)Epsilon(+) X(sup 2)pi(1,0) band using a broadband Nd:YAG pumped optical parametric oscillator (OPO) and narrowband Nd:YAG/dye laser with ultraviolet frequency extension (UVX) package. Variables tested for influence on fluorescent signal and absorption characteristics were excitation line, laser energy, exciting linewidth, combustion reactants, and test flow conditions. Results are intended to guide the transition from a dye/UVX laser to an OPO system for performing advanced diagnostics of low-emission combustion concepts.
Flame-vortex interactions are canonical configurations that can be used to study the underlying processes occurring in complicated turbulent reacting flows. The elegant simplicity of the flame-vortex interaction permits the study of these complex interactions under relatively controllable experimental configurations, in contrast to direct measurements in turbulent flames. The ability to measure and model the fundamental phenomena that occur in a turbulent flame, but with time and spatial scales which are amenable to our diagnostics, permits significant improvements in the understanding of turbulent combustion under both normal and reduced gravity conditions. In this paper, we report absolute mole fraction measurements of methane in a reacting vortex ring. These microgravity experiments are performed in the 2.2-sec drop tower at NASA Glenn Research Center. In collaboration with Drs. Chen and Dahm at the University of Michigan, measured methane absorbances are incorporated into a new model from which the temperature and concentrations of all major gases in the flame can be determined at all positions and times in the development of the vortex ring. This is the first demonstration of the ITAC (Iterative Temperature with Assumed Chemistry) approach, and the results of these computations and analyses are presented in a companion paper by Dahm and Chen at this Workshop. We believe that the ITAC approach will become a powerful tool in understanding a wide variety of combustion flames under both equilibrium and non-equilibrium conditions.