The present work addresses the question of whether the chemical reaction zone within a turbulent, high-Reynolds-number jet flame is thin-and can be modeled using strained wrinkled laminar flamelet theory-or is thick-and must be modeled using distributed reaction zone theory. The region near the wrinkled instantaneous stoichiometric contour is identified using CH Planar Laser-Induced Fluorescence (PLIF) imaging and the strain on this interface is measured using simultaneous Particle Imaging Velocimetry (PIV) diagnostics. With a separate set of simultaneous images, the relative position of the CH and OH radicals is observed using PLIF. It is found that the CH reaction zone remains thin and rarely exceeds 1 mm, even near the flame tip in a high-Reynolds-number (18,600) jet flame. The mean thickness of the CH reaction layer (5 C H) increases from 0.3 to 0.8 mm in the streamwise direction; this is expected because the scalar dissipation rate is known to decrease with downstream distance and should cause a corresponding increase in 8 CH . However, 8 CH also increases with jet velocity, which is not predicted by theory. Furthermore, the mean strain rates on the stoichiometric contour increase in the streamwise direction, which is contrary to previous predictions. Thus, strain rate does not, in general, scale with the local dissipation rate in a turbulent flame, and this aspect of the counterflow flow analogy is not valid. It is concluded that unsteady laminar flamelet concepts are consistent with most of the present observations, but a method independent of the local dissipation rate is needed to predict the local strain rate.
IntroductionOften, the analogy is made between segments of turbulent nonpremixed jet flames and a steady laminar counterflow diffusion flame that is selected to have the same strain rate as the turbulent flame segment; particularly, this analogy is used in "laminar flamelet models" [1,2]. The flamelet regime, by definition, exists over regions in which the flame front is thinner than the smallest scale of turbulence, and thus the flame front can be modeled as an ensemble of laminar flamelets. For the flamelet analogy to be realistic, several factors must be known: 1. if the reaction layers are sufficiently thin (rather than thick, distributed zones); 2. if the reaction-layer properties (e.g. radical concentrations, thickness) scale as predicted by counterflow theory; 3. how frequently the layers extinguish or merge; and 4. if the strain on the turbulent reaction layers is proportional to the dissipation rate as in counterflow flames [2]. If all of these conditions are satisfied, it becomes possible to apply several promising modeling ideas which draw heavily on flamelet theory, including the original flamelet model [2], the Local Integral Moment (LIM) model [3,4], the Large Eddy Simulation (LES) model [5], and Direct Numerical Simulation (DNS) [6]. To date, lack of information on the structure of turbulent