SEARCH · Engineering Papers
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.
The focus of this work was on kinetic modeling. The specific objectives were: detailed modeling of soot formation in premixed flames, elucidation of the effects of fuel structure on the pathway to soot, and the development of a numerical technique for accurate modeling of soot particle coagulation and surface growth. Those tasks were successfully completed and are briefly summarized.
A method of systematic optimization, solution mapping, as applied to a large-scale dynamic model is presented. The basis of the technique is parameterization of model responses in terms of model parameters by simple algebraic expressions. These expressions are obtained by computer experiments arranged in a factorial design. The developed parameterized responses are then used in a joint multiparameter multidata-set optimization. A brief review of the mathematical background of the technique is given. The concept of active parameters is discussed. The technique is applied to determine an optimum set of parameters for a methane combustion mechanism. Five independent responses - comprising ignition delay times, pre-ignition methyl radical concentration profiles, and laminar premixed flame velocities - were optimized with respect to thirteen reaction rate parameters. The numerical predictions of the optimized model are compared to those computed with several recent literature mechanisms. The utility of the solution mapping technique in situations where the optimum is not unique is also demonstrated.
Isolated droplet combustion processes have been the subject of extensive experimental and theoretical investigations for nearly 40 years. The gross features of droplet burning are qualitatively embodied by simple theories and are relatively well understood. However, there remain significant aspects of droplet burning, particularly its dynamics, for which additional basic knowledge is needed for thorough interpretations and quantitative explanations of transient phenomena. Spherically-symmetric droplet combustion, which can only be approximated under conditions of both low Reynolds and Grashof numbers, represents the simplest geometrical configuration in which to study the coupled chemical/transport processes inherent within non-premixed flames. The research summarized here, concerns recent results on isolated, single component, droplet combustion under microgravity conditions, a program pursued jointly with F.A. Williams of the University of California, San Diego. The overall program involves developing and applying experimental methods to study the burning of isolated, single component droplets, in various atmospheres, primarily at atmospheric pressure and below, in both drop towers and aboard space-based platforms such as the Space Shuttle or Space Station. Both computational methods and asymptotic methods, the latter pursued mainly at UCSD, are used in developing the experimental test matrix, in analyzing results, and for extending theoretical understanding. Methanol, and the normal alkanes, n-heptane, and n-decane, have been selected as test fuels to study time-dependent droplet burning phenomena. The following sections summarizes the Princeton efforts on this program, describe work in progress, and briefly delineate future research directions.
Understanding and modeling of turbulent combustion are key problems in the computation of numerous practical systems. Because of the lack of analytical theories in this field and of the difficulty of performing precise experiments, direct simulation appears to be one of the most attractive tools to use in addressing this problem. From Sep. 1989 - Sep. 1990, the following problems were studied using direct simulation but also experiments and theoretical models: (1) the influence of curvature on premixed flame fronts; (2) the construction of a model for the flame stretch based on direct simulation and multifractal analysis; and (3) the simulation of the interaction between random flow field and a flame front. The general objective was to improve knowledge of turbulent combustion but also to use this information for turbulent combustion models. The transfer of direct simulation results towards models is also in progress. A submodel to evaluate the flame stretch was derived from the present study and incorporated in the Coherent Flame model used in France for piston and aircraft engines.
The nature and formation of smoke and its combustion were investigated. Smoke, which consist of tiny mesomorphous crystals tightly packed into popcorn-ball-like particles that agglomerate to give filaments, was found to contain about 5 percent of the hydrogen originally present in the fuel. Factors affecting smoke formation were studied in both diffusion flames and premixed Bunsen flames. It is suggested that smoking tendency increases with increasing stability of the carbon skeleton of the molecule, as determined by relative bond strength.
The burning of liquid hydrocarbon fuels accounts for a significant portion of global energy production. With predicted future increases in demand and limited reserves of hydrocarbon fuel, it is important to maximize the efficiency of all processes that involve conversion of fuel. With the exception of unwanted fires, most applications involve introduction of liquid fuels into an oxidizing environment in the form of sprays which are comprised of groups of individual droplets. Therefore, tremendous benefits can result from a better understanding of spray combustion processes. Yet, theoretical developments and experimental measurements of spray combustion remains a daunting task due to the complex coupling of a turbulent, two-phase flow with phase change and chemical reactions. However, it is recognized that individual droplet behavior (including ignition, evaporation and combustion) is a necessary component for laying the foundation for a better understanding of spray processes. Droplet combustion is also an ideal problem for gaining a better understanding of non-premixed flames. Under the idealized situation producing spherically-symmetric flames (produced under conditions of reduced natural and forced convection), it represents the simplest geometry in which to formulate and solve the governing equations of mass, species and heat transfer for a chemically reacting two phase flow with phase change. The importance of this topic has promoted extensive theoretical investigations for more than 40 years.
A large-eddy simulation / Reynolds-averaged Navier-Stokes (LES/RANS) methodology is used to simulate premixed ethylene-air combustion in a model scramjet designed for dual mode operation and equipped with a cavity for flameholding. A 22-species reduced mechanism for ethylene-air combustion is employed, and the calculations are performed on a mesh containing 93 million cells. Fuel plumes injected at the isolator entrance are processed by the isolator shock train, yielding a premixed fuel-air mixture at an equivalence ratio of 0.42 at the cavity entrance plane. A premixed flame is anchored within the cavity and propagates toward the opposite wall. Near complete combustion of ethylene is obtained. The combustor is highly dynamic, exhibiting a large-scale oscillation in global heat release and mass flow rate with a period of about 2.8 ms. Maximum heat release occurs when the flame front reaches its most downstream extent, as the flame surface area is larger. Minimum heat release is associated with flame propagation toward the cavity and occurs through a reduction in core flow velocity that is correlated with an upstream movement of the shock train. Reasonable agreement between simulation results and available wall pressure, particle image velocimetry, and OH-PLIF data is obtained, but it is not yet clear whether the system-level oscillations seen in the calculations are actually present in the experiment.
Selective Two-Photon Absorptive Resonance Femtosecond-Laser Electronic-Excitation Tagging (STARFLEET), a non-seeded ultrafast-laser-based velocimetry technique, is demonstrated in reactive and non-reactive flows. STARFLEET is pumped via a two-photon resonance in N2 using 202.25-nm 100-fs light. STARFLEET greatly reduces the per-pulse energy required (30 μJ/pulse) to generate the signature FLEET emission compared to the conventional FLEET technique (1.1 mJ/pulse). This reduction in laser energy results in less energy deposited in the flow, which allows for reduced flow perturbations (reactive and non-reactive), increased thermometric accuracy, and less severe damage to materials. Velocity measurements conducted in a free jet of N2 and in a premixed flame show good agreement with theoretical velocities and further demonstrate the significantly less-intrusive nature of STARFLEET.
This study concentrates on developing a better understanding of triple flames. We relax the assumption of zero heat release, address the issue of stabilization, and, in order to investigate the role that heat release plays in flame propagation in partially premixed combustion, we return to a simple flow field and investigate the behavior of flames in a laminar environment. We solve the compressible Navier-Stokes equations in a two-dimensional domain. At the boundaries, we use an inflow boundary condition on the left and nearly-perfect reflective boundary conditions, required to avoid pressure drift, at the outflow and sides. After the flow and flame are initialized, the mixture fraction is varied at the inlet from its uniform stoichiometric value to a tanh profile varying from zero to one. As the mixture fraction gradient reaches the flame surface only the centerline is exposed to the stoichionetric mixture fraction and locally maintains the planar flame speed and reaction rate. Above this point the mixture is fuel rich, and below fuel lean. As a result, these regions of non-unity equivalence ratio burn less, the reaction rate drops, and the local flame speed is reduced. The excess fuel and oxidizer then combine behind the premixed flame along the stoichiometric surface and burn in a trailing diffusion flame. Thus the 'triple' flame refers to the fuel-rich premixed flame, the fuel-lean premixed flame, and the trailing diffusion flame. Due to heat release, the normal velocity across the flame is increased, whereas the tangential component remains unchanged. Far-field flame speed, local flame speed, and their differences are shown as a function of the local mixing thickness. It was also determined that the lateral position of the flame affects stabilization, and the distribution of the reaction rate along the premixed wings of triple flames affects flame propagation.
The mean reaction rate in flamelet models for turbulent premixed combustion depends on two basic quantities: a mean chemical rate, called the flamelet speed, and the flame surface density. Our previous work had been primarily focused on the problem of the structure and topology of turbulent premixed flames, and it was then determined that the flamelet speed, when space-averaged, is only weakly sensitive to the turbulent flow field. Consequently, the flame surface density is the key quantity that conveys most of the effects of the turbulence on the rate of energy release. In flamelet models, this quantity is obtained via a modeled transport equation called the Sigma-equation. Past theoretical work has produced a rigorous approach that leads to an exact but unclosed formulation for the turbulent Sigma-equation. In the exact Sigma-equation, it appears that the dynamical properties of the flame surface density are determined by a single parameter, namely the turbulent flame stretch. Unfortunately, the turbulent flame stretch as well as the flame surface density is not available from experiments, and, in the absence of experimental data, little is known on the validity of the closure assumptions used in current flamelet models. Direct Numerical Simulation (DNS) is the alternative approach to get basic information on these fundamental quantities. In the present work, three-dimensional DNS of premixed flames in isotropic turbulent flow is used to estimate the different terms appearing in the Sigma-equation. A new methodology is proposed to provide the source and sink terms for the flame surface density, resolved both temporally and spatially throughout the turbulent flame brush. Using this methodology, our objective is to extract the turbulent flame stretch from the DNS data base and then perform extensive comparisons with flamelet models. Thanks to the detailed information produced by the DNS-based analysis, it is expected that this type of comparison will not only underscore the shortcomings of current models, but also suggest ways to improve them.
Ignition and propagation of turbulent flame front in jet flow investigated for combustible hydrogen- air mixture
Flame propagation in supersonic premixed flows of hydrogen and air ignited by coaxial, hot gas, pilot jet
Open laboratory turbulent flames used for investigating fundamental flame turbulence interactions are greatly affected by buoyancy. Though much of our current knowledge is based on observations made in open flames, buoyancy effects are usually not considered in data interpretation, numerical analysis or theories. This inconsistency remains an obstacle to merging experimental observations and theoretical predictions. To better understanding the effects of buoyancy, our research focuses on steady lean premixed flames propagating in fully developed turbulence. We hypothesize that the most significant role of buoyancy forces on these flames is to influence their flowfields through a coupling with the mean and the fluctuating pressure fields. This coupling relates to the elliptical problem that emphasizes the importance of the upstream, wall and downstream boundary conditions in determining all aspects of flame propagation. Therefore, buoyancy has the same significance as other parameters such as flow configuration, and flame geometry.
Open laboratory turbulent flames used for investigating fundament flame turbulence interactions are greatly affected by buoyancy. Though much of our current knowledge is based on observations made in these open flames, the effects of buoyancy are usually not included in data interpretation, numerical analysis or theories. This inconsistency remains an obstacle to merging experimental observations and theoretical predictions. To better understanding the effects of buoyancy, our research focuses on steady lean premixed flames propagating in fully developed turbulence. We hypothesize that the most significant role of buoyancy forces on these flames is to influence their flowfields through a coupling with mean and fluctuating pressure fields. Changes in flow pattern alter the mean aerodynamic stretch and in turn affect turbulence fluctuation intensities both upstream and downstream of the flame zone. Consequently, flame stabilization, reaction rates, and turbulent flame processes are all affected. This coupling relates to the elliptical problem that emphasizes the importance of the upstream, wall and downstream boundary conditions in determining all aspects of flame propagation. Therefore, buoyancy has the same significance as other parameters such as flow configuration, flame geometry, means of flame stabilization, flame shape, enclosure size, mixture conditions, and flow conditions.
The chemical kinetics of cool flames in mixtures of normal alkanes, oxygen, and nitrogen recently have been reduced to obtain simplified asymptotic flame-structure models that can be used to compare theoretical predictions with experimental measurements. A key step that affects flame structures strongly, namely the abstraction of a hydrogen atom from the alkyl radical by an oxygen molecule, had not been taken into account in those models. This deficiency has now been remedied for diffusion flames. The present contribution reports the corresponding remedy for premixed-flame analyses. In the new model, non-zero chemical reaction rates are restricted to narrow regions in the vicinity of the maximum temperature, whereas previously they extended throughout the preheat zone. Comparisons of the new predictions with experimental burning-velocity measurements seem favorable.
A simplified chemical-kinetic cool-flame mechanism for n-alkanes developed recently is applied to premixed laminar cool-flame deflagrations. The present contribution derives the structure of the associated freely propagating cool premixed flame controlled by the low-temperature chemistry and develops formulas for calculating the corresponding laminar burning velocity. Application of activation-energy asymptotics reveals finite-rate chemistry without heat release occurring throughout the preheat zone and leakage of both fuel and oxygen through the thin heat-release zone, there being zones of consumption of an intermediate species on each side of the heat-release zone, thicker than that zone but still thin compared with the preheat-zone thickness. Predicted laminar burning velocities are compared with recently reported measurements for n-dodecane, performed in a newly deigned high-pressure ignition apparatus, resulting in reasonable agreement.
Turbulent non-premixed stoichiometric methane-air flames modeled with reduced kinetics have been studied using the direct numerical simulation approach. The simulations include realistic chemical kinetics, and the molecular transport is modeled with constant Lewis numbers for individual species. The effect of turbulence on the internal flame structure and extinction characteristics of methane-air flames is evaluated. Consistent with earlier DNS with simple one-step chemistry, the flame is wrinkled and in some regions extinguished by the turbulence, while the turbulence is weakened in the vicinity of the flame due to a combination of dilatation and an increase in kinematic viscosity. Unlike previous results, reignition is observed in the present simulations. Lewis number effects are important in determining the local stoichiometry of the flame. The results presented in this work are preliminary but demonstrate the feasibility of incorporating reduced kinetics for the oxidation of methane with direct numerical simulations of homogeneous turbulence to evaluate the limitations of various levels of reduction in the kinetics and to address the formation of thermal and prompt NO(x).
Recent measurements and predictions of the properties of homogeneous (gaseous) laminar and turbulent non-premixed (diffusion) flames are discussed, emphasizing results from both ground- and space-based studies at microgravity conditions. Initial considerations show that effects of buoyancy not only complicate the interpretation of observations of diffusion flames but at times mislead when such results are applied to the non-buoyant diffusion flame conditions of greatest practical interest. This behavior motivates consideration of experiments where effects of buoyancy are minimized; therefore, methods of controlling the intrusion of buoyancy during observations of non-premixed flames are described, considering approaches suitable for both normal laboratory conditions as well as classical microgravity techniques. Studies of laminar flames at low-gravity and microgravity conditions are emphasized in view of the computational tractability of such flames for developing methods of predicting flame structure as well as the relevance of such flames to more practical turbulent flames by exploiting laminar flamelet concepts.