Viswanath R. Katta

Flame-vortex dynamics in an inverse partially premixed combustor - The Froude number effects

In this paper we report on a computational and experimental investigation of the transient combustion characteristics of an inverse partially premixed flame established by injecting a fuel-rich (CH4-air) annular jet sandwiched between a central air jet on the inside and coflowing air on the outside. A time-dependent, axisymmetric, reacting flow model is used to simulate the flame dynamics. A global 1-step and a relatively detailed 52-step mechanism are used to model the CH4-air chemistry. Results focus on the dynamic flame structure and flame-vortex interactions at different Froude numbers (Fr), the scaling of the flame flicker frequency, and the global comparison of experimental and computational results. At high Froude numbers (nonbuoyant limit), the computed flame exhibits a steady-state structure, which is markedly different from that of a jet diffusion flame. The flame structure reveals two distinct reaction zones consisting of an inner premixed region followed by two nonpremixed flames at the wings. Methane is converted to CO and H 2 in the premixed reaction zone and these intermediate species provide fuel for the outer nonpremixed flames. Main reaction pathways associated with the double-flame structure are identified. For intermediate Fr, the buoyant acceleration becomes significant, causing a periodic rollup of toroidal vortices. While the rollup process is highly periodic, the flame exhibits steady-state behavior, since vortices are relevant only in the plume region. For Fr < 1.0, the rollup occurs closer to the burner port, resulting in flame-vortex interactions and a dynamic flame. A distinguished characteristic of this flame is the rollup of two simultaneous vortices corresponding to inner and outer diffusion flames, which convect downstream at the same velocity, and interact with the twin flame surfaces, causing flame flicker and stretch. Both numerical and laboratory experiments are employed to obtain a correlation between the Strouhal number (S), associated with the vortex rollup or flame flicker frequency, and the Froude number. Simulations yield a correlation S = 0.56 Fr 038, while measurements yield S = 0.43 Fr -°38, indicating an excellent agreement, considering that the flow conditions in the numerical and laboratory experiments are only globally matched in terms of overall stoichiometry, Fr, and Reynolds number, and not with respect to burner size and jet velocity. Finally, the effects of chemical kinetics on the computed flame structure are examined. Both the time-averaged and the dynamic flame structure, including flame height, peak temperature, and flicker frequency, are found to be influenced by chemical kinetics. However, the scaling of the dominant frequency or Strouhal number with Fr is essentially the same for the two mechanics. In addition, the frequency is found to be independent of the chemical kinetic parameters used in the global mechanism.

Attachment mechanisms of diffusion flames

A common view of the stabilizing mechanism of methane diffusion flames, including jet and porous flat-plate burner flames, is presented. Two-color particle-imaging velocimetry measured the velocity field in the stabilizing region of jet diffusion flames under near-lifting conditions. Computations using a time-dependent, implicit, third-order accurate numerical model, including semidetailed chemical kinetics and buoyancy effects, revealed the detailed structures of the vertical jet diffusion flames and flat-plate burner flames with different orientations of the plate surface and fuel injection. The numerical results are in a good agreement with the measurements in the flame base locations and surrounding velocity fields. In the calculations of both classes of flames, the highest reactivity spot ( reaction kernel ) with peak rates of heat release, oxygen consumption, and water vapor production, was formed in the relatively low-temperature ( 3 +O→CH 2 O+H reaction predominantly contributed to the heat-release rate peak. Heuristic correlations were found between the heat-release or oxygen-consumption rate and the local velocity over a wide range. At a high coflow air velocity in a jet diffusion flame, the flame base shifted slightly downstream before lifting, resulting in a higher reactivity and thereby withstanding at a higher local velocity. On the other hand, in a long horizontal flat-plate flame with downward fuel injection, a recirculation zone was formed ahead of the flame base, resulting in an order-of-magnitude lower local velocity and reactivity. Therefore, the reaction kernel provides a stationary ignition source and sustained stable combustion for incoming reactants, thus holding the trailing diffusion flame in the oxidizing stream.

A reaction kernel hypothesis for the stability limit of methane jet diffusion flames

The lifting limit of an axisymmetric, laminar, co-flow methane-air jet diffusion flame under normal earth gravity has been successfully predicted. Computations of the time-dependent full Navier-Stokes equations with buoyancy were performed using an implicit, third-order accurate numerical scheme and a detailed C 2 chemistry model. A one-step global chemistry model was also used to reveal its deficiencies and to demonstrate the need for “tuning” its kinetic parameters for the studies on flame lifting. The detailed chemistry model resulted in the standoff distance of the flame from the burner rim in good agreement with that measured previously. As the mean co-flow air velocity was increased, at a fixed fuel jet velocity under the near-limit condition, the calculated reaction kernel (peak reactivity spot) in the flame base broadened and rapidly shifted away downstream. As a result, a higher reactivity (heat-release rate, oxygen consumption rate, etc.) at the reaction kernel could be obtained to sustain combustion against a higher incoming flow velocity, or a shorter residence time. The reactivity augmentation is due to a “blowing” effect, which caused enhanced convective and diffusive fluxes of oxygen into the relatively low-temperature (∼1550 K) fuel-lean (equivalence ratio ≈0.55) reaction kernel. Based on these new findings, a reaction kernel hypothesis is proposed for the diffusion flame stability, namely, that a subtle balance between the residence time and reaction time in the reaction kernel is maintained by its continuous movement in the downstream direction in response to the destabilizing effect caused by an increase in the co-flow air velocity, and the overall reaction time eventually exceeds the available residence time at the stability limit. If a secondary stabilizing point is obtained as a result of the transition to a turbulent flame base downstream, the flame lifts off, otherwise it blows off.

Measurements of OH mole fraction and temperature up to 20 kHz by using a diode-laser-based UV absorption sensor

Diode-laser-based sum-frequency generation of ultraviolet (UV) radiation at 313.5 nm was utilized for high-speed absorption measurements of OH mole fraction and temperature at rates up to 20 kHz. Sensor performance was characterized over a wide range of operating conditions in a 25.4 mm path-length, steady, C2H4-air diffusion flame through comparisons with coherent anti-Stokes Raman spectroscopy (CARS), planar laser-induced fluorescence (PLIF), and a two-dimensional numerical simulation with detailed chemical kinetics. Experimental uncertainties of 5% and 11% were achieved for measured temperatures and OH mole fractions, respectively, with standard deviations of < 3% at 20 kHz and an OH detection limit of < 1 part per million in a 1 m path length. After validation in a steady flame, high-speed diode-laser-based measurements of OH mole fraction and temperature were demonstrated for the first time in the unsteady exhaust of a liquid-fueled, swirl-stabilized combustor. Typical agreement of approximately 5% was achieved with CARS temperature measurements at various fuel/air ratios, and sensor precision was sufficient to capture oscillations of temperature and OH mole fraction for potential use with multiparameter control strategies in combustors of practical interest.

Gravity effects on steady two-dimensional partially premixed methane–air flames

Under normal-gravity conditions the flame heat release produces both flow dilatation and buoyancy effects. While it may be possible to minimize gravitational effects in a fully premixed flame by isolating buoyancy effects to the lower-density postflame region or plume, this cannot be accomplished in nonpremixed flames. It is known that partially premixed flames can contain two reaction zones, one with a premixed-like structure and the other consisting of a transport-limited nonpremixed zone (in which mixing and entrainment effects are significant). For these reasons it is important to understand the fundamental interaction between flow dilatation and buoyancy effects in partially premixed flames. A detailed numerical study is conducted to characterize the effect of buoyancy on the structure of two-dimensional partially premixed methane‐air flames. The computational model is validated by comparison with the experimentally obtained chemiluminescent emission from excited-C* free radical species as well as with velocity vectors obtained using particle image velocimetry. Both the experiments and simulations indicate the presence of two reaction zones that are synergistically coupled, with each region providing heat and/or chemical species for the other. While the inner premixed flame is only weakly affected by gravity, the outer flame shows significant spatial differences for the two cases due to buoyancyinduced entrainment, since advection of air into the outer reaction zone increases in the presence of gravity. The presence of gravity induces more compact flames, influences the velocity profiles in the post‐inner flame region, and increases the normal strain rate. Although the spatial differences between the 0- and 1-g flames are more significant on the lean side, the state relationships in that region are relatively unaffected by gravity. On the other hand, the inner (rich-side) reaction zone shifts toward less-rich locations in the presence of gravity, possibly due to the enhanced buoyant mixing. The 1-g flames exhibit a larger energy loss in the form of CO and H2 emissions.

An experimental and numerical investigation of the structure of steady two-dimensional partially premixed methane-air flames

Steady two-dimensional partially premixed slot-burner flames established by introducing a rich fuel-air mixture from the inner slot and air from the two outer slots are investigated. Numerical simulations are conducted using detailed chemistry, velocity measurements are made using particle image velocimetry, and images of the chemiluminescent reaction zones are obtained. Two reaction zones are evident: one in an inner rich-side premixedlike flame and the other in an outer lean-side non-premixed flame. Validation of the predictions involves a comparison of the (1) premixed and non-premixed flame heights, (2) the double-flame structure and (3) velocity vectors. The measured and predicted velocity vectors are in good agreement and show that the flame interface separates smaller velocity magnitudes on the reactant side from large values on the (partially) burned side. The outer flame temperature is higher than that of the inner premixed flame. A substantial amount of methane leaks past the inner flame and reacts in the outer non-remixed zone. The inner flame produces partially oxidized products such as H2 and CO, which provide the fuel for the non-premixed flame. The initiation reaction CH4+H⇔CH3+H2 proceeds strongly at the base of the flame where both the inner and outer flames are connected and at the tip of the inner flame, and it is weak along the sides of both inner and outer flames in accord with the chemiluminescent images. Carbon dioxide formation through the reaction CO+OH⇔CO2+H is more diffuse than methane consumption in the outer flame, because the availability of hydroxyl radicals in that region is limited through oxidizer transport.

Calculation of Multidimensional Flames Using Large Chemical Kinetics

A time-dependent, two-dimensional, detailed-chemistry computational fluid dynamics model, known as UNICORN (unsteady ignition and combustion using reactions), is used for solving complex flame problems. The unique features incorporated in UNICORN for handling extremely large chemical kinetics with ease and efficiency are discussed. A submixture concept that is used for evaluating transport properties is described. This concept increases the computational speed by a factor of five for a 208-species mechanism and is expected to have even higher efficiency with larger mechanisms. An implicit treatment for certain reaction-rate terms applied during the solution of species-conservation equations is described. Moving the reaction-rate source terms to the left-hand side of the partial differential equations eases the stiffness problem that is typically associated with combustion chemical kinetics. Computational speeds are further improved in UNICORN by completely integrating the chemical-kinetics mechanisms with the solution algorithm. A software-generated computational fluid dynamics approach is used to avoid the tedious and near-impossible task of manually integrating a large chemical-kinetics mechanism into a computational fluid dynamics code. Several calculations demonstrating the abilities of the UNICORN code are presented. Chemical-kinetics mechanisms up to 366 species and 3700 reaction steps are incorporated, and simulations for unsteady multidimensional flames are performed on personal computers. Making use of the robustness and efficiency of the UNICORN code, detailed chemical mechanisms developed for JP-8 fuel are tested for their accuracy, and a parametric study on the role of parent species of a surrogate mixture in predicting flame extinguishment is performed. Ease of changing chemical kinetics in the UNICORN code is demonstrated through the investigation of effects of additives in JP-8 fuel.

A Numerical Study of Droplet-Vortex Interactions in an Evaporating Spray,

In this paper, we present the time-dependent axisymmetric numerical simulation of a n-heptane evaporating spray, and investigate the droplet-vortex interactions which determine the structural and dynamic characteristics of a spray jet flow. The spray is formed between a droplet-laden heated nitrogen jet and a coflowing air stream. A detailed, multidimensional, two-phase algorithm is developed for the simulation. Monodisperse spray is introduced into the large vortex structures that are generated by the buoyancy-induced hydrodynamic instability of the heated jet. Results focus on the two-way interactions between vortical structures and droplets, and the dynamics of both non-evaporating and evaporating sprays. The vortex structures cause droplets to disperse radially outward, and this in turn determines the fuel vapor distribution and also modifies the vortex dynamics. Thus, the dynamics and structural characteristics of evaporating sprays are strongly influenced by the two-way transient interactions. The effects of initial droplet size, injection location, and liquid-to-gas mass loading ratio on these interactions are investigated. These studies indicate that the effect of dispersed phase on gas phase is negligible for mass loading ratio less than 0.5. At higher mass loading ratios, the dispersed phase modifies the dynamics of vortex structures but not the time-average behavior for non-evaporating spray, while for evaporating spray it influences both the dynamics and the time-averaged behavior. It is also found that the spray injection characteristics have strong influence on the processes of droplet-vortex interactions.

Triple flame propagation and stabilization in a laminar axisymmetric jet

The propagation of a methane–air triple flame in a partially premixed jet is investigated experimentally and numerically. The flame is ignited with a Nd : YAG laser in a nonuniform jet-mixing layer downstream of the burner. The ignition and flame propagation processes are recorded using a high-speed video camera. The flamefront propagation velocity in laboratory coordinates is inferred from the video images. A comprehensive, time-dependent computational model is used to simulate the transient ignition and flame propagation phenomena. The model employs a detailed description of methane–air chemistry and transport properties. Following ignition, a well-defined triple flame is formed that propagates upstream towards the burner along the stoichiometric mixture fraction line. As the flame propagates upstream, the flame propagation speed, which is defined as the normal flamefront velocity with respect to the local gas velocity, decreases linearly. Near the burner wall, the flame curvature increases to two times the value of its downstream freely propagating counterpart. During the flame propagation process, the curvature-induced stretch dominates over the hydrodynamic stretch and the flame speed decreases with increasing stretch rate in accord with previous measurements. We also examine the dominant reaction rates to follow the transition from a triple flame to a double flame structure.

Photolytic-interference-free, femtosecond two-photon fluorescence imaging of atomic hydrogen

We discuss photolytic-interference-free, high-repetition-rate imaging of reaction intermediates in flames and plasmas using femtosecond (fs) multiphoton excitation. The high peak power of fs pulses enables efficient nonlinear excitation, while the low energy nearly eliminates interfering single-photon photodissociation processes. We demonstrate proof-of-principle, interference-free, two-photon laser-induced fluorescence line imaging of atomic hydrogen in hydrocarbon flames and discuss the methods implications for certain other atomic and molecular species.