Non-premixed Flows Research Articles

Double-pulse imaging using simultaneous OH/acetone plif for studying the evolution of high-speed, reacting mixing layers

A technique for recording the evolution of large-scale structures in high-speed nonpremixed combustion flows is demonstrated. A single frequency-doubled dye laser system that can produce two pulses separated by tens of microseconds is used to simultaneously excite fluorescence from acetone, seeded into the fuel stream, and from OH, which marks the combustion gases. Thus, four images, an OH/acetone pair for each of the two pulses, are recorded on four separate cameras. The accuracy of the double-pulse technique is examined in a low-speed methane-air flame, which is essentially frozen during the measurement time. The average difference, or error, between corresponding pixels in a double-pulse pair of OH images shows little spatial biasing and primarily results from signal shot-noise fluctuations. The double-pulse OH/acetone technique is demonstrated in a nominally two-dimensional, high-speed reacting mixing layer consisting of an ambient-temperature, low-speed hydrogen-containing fuel stream and a high-temperature, high-speed oxidizing stream. Two mixing layer conditions are presented: a low compressibility case, with convective Mach number M c =0.32, and a more compressible case, M c =0.70. Reasonable quality images, having minimum noise levels of 11–14%, were obtained in both cases. The peak OH levels are near 2000 ppm, and the acetone seeding is roughly 3500 ppm. The regions of the flow defined by the interface between the mixing layer and the acetone-seeded fuel stream are relatively stable during the 30–50 μs measurement times and have convective velocities slightly higher than the speed of the fuel stream. The regions of the mixing layer marked by OH move faster and exhibit greater distortion.

Direct numerical simulation of turbulent flows with chemical reaction

Results from full turbulence simulations incorporating the effects of chemical reaction are compared with simple closure theories and used to reveal some physical insights about turbulent reacting flows. Pseudospectral methods for homogeneous turbulent flows with constant physical and thermal properties in domains as large as 643 Fourier modes were used for these simulations. For the case of nonpremixed flows involving a two-species, second-order, irreversible chemical reaction, it is found that the scalar dissipation microscale is only a weak function of the reaction rate and that chemical reaction contributes very little to the decay of the variance of the reactant concentration. Examination of local values of the velocity and concentration fields shows that the local reaction rate is highest in regions of the greatest rates of strain and that vorticity tends to align with the reaction zone. Finally, difficulties associated with the evaluation of multipoint pdf's and with the archival of time-dependent data from the threedimensional simulations are described.

Turbulence effects on chemical reactions in smog chamber flows

Turbulent mixing effects on the reaction rate of a non-premixed flow are presented for a moderately slow second-order irreversible chemical reaction. The turbulent mixing process leads to inhomogeneities in the concentration of the reactants. Chemical reactions are normally highly non-linear and large errors can result from using average concentrations in the computation of mean reaction rates. A brief review of the literature on this problem and its application areas is made with particular emphasis placed on near isothermal flows where the fluctuation in the reaction rate constant can be neglected. The reaction between mixing air jets containing dilute nitric oxide, NO, and ozone, O 3, is studied in a large Turbulent Smog Chamber (TSC) under conditions of high Reynolds number, three-dimensional flow. The measured reactant concentrations, obtained with good time and space resolution, are used to investigate the effects of the reactant species concentration fluctuations on the mean reactant species concentration field through their contribution to the mean reaction rate. Two flow geometries have been studied: two opposed jets issuing into a large chamber and a plume-like source issuing into a turbulent background flow. A three-dimensional finite-difference computation has been carried out for the flow in the chamber using the k- ε- g model of turbulence and mixing. The reacting species concentration field is calculated by considering the transport of a ‘perturbation variable’ equal to the mean of the difference between the species concentration and its corresponding fast chemistry value. A closure for the mean chemical reaction rate based on this quantity is presented and its experimental validation discussed.

Examination of closure models for mean chemical reaction rate using experimental results for an isothermal turbulent reacting flow

Joint probability density functions of reactant species concentrations have been compiled from measurements made in an isothermal, two feed stream, nonpremixed turbulent flow in which a simple chemical reaction proceeds in chemical nonequilibrium. This data is used here to examine four closure methods for the mean rate of species production by chemical reaction which appears as a sink term in the species conservation equation. The four closures considered are based, respectively, on (i) product of mean reactant concentrations, (ii) a full second order closure scheme, (iii) and (iv) two levels of simplification of results from species concentration perturbation analysis. It is found that closure (i) yields estimated mean rates between 1 and 3 times the true mean rate while the closures based on perturbation analysis produce estimates in the narrow range between 0.9 and 1.4 times the true mean rate. Unclosed triple correlations of species concentration fluctuation occurring in closure scheme (ii) are found to be of the same order as other terms in the governing equations of this scheme. It would, therefore, appear hazardous to neglect these triple correlations in order to effect closure under this scheme. It is shown that as the flow chemistry approaches equilibrium closures (iii) and (iv) display the correct behaviour in yielding rate estimates which approach zero while closure (i) may yield non-zero estimates. Quantitative assessment of closure accuracy close to equilibrium must await further experimental studies.

REACTIVE MIXING IN TURBULENT GASES

Abstract Conversion of the ozone-nitric oxide reaction was measured along a non-premixed flow reactor. Conversion was less than the homogeneous limit with a single jet mixer, allowing determination of the mean product of the reactant concentration fluctuations. The liquid decay law was measured with a very rapid reaction and was found, within experimental error, to be equal to the normalized mean product of the reactant concentration fluctuation in the reacting gas. The results imply that the mixing decay law can be taken as the same for gases and liquids, and that the reactive-mixing model developed for liquids also applies to gases. Thus the design procedure developed for liquids may also be used in gases. In addition, since the mixing decay law is easily measured in liquids, a gas phase reactor may be conveniently designed by measuring d in a pilot scale reactor with a very rapid liquid phase reaction.