Computational Simulation of Complex Turbulent Diffusion Flames

University Of Washington, Seattle WA

Investigators

Abstract

Although the principles and fundamental equations that describe turbulent combustion are known, their direct numerical solution (DNS) remains impossible for practical problems. The usual approximations are to either time average or to spatially filter the fundamental equations. The first approach involves full time averaging (Reynolds averaged Navier-Stokes equations, or RANS), while the second filters only for the smaller scales, allowing the larger scales to be resolved in space and time (Large Eddy Simulation, or LES). In either case, the semi-empirical modeling assumptions associated with the averaging process are critical, and have been the focus of much debate. These approaches have had substantial successes in the modeling of many flames, principally those dominated by boundary-layer-like flows such as free jets and shear layers, and combustion that is uninterrupted by local extinction. Most practical flames are, however, stabilized by recirculation or swirl, and exhibit extinction events, e.g., standoff from the burner. These phenomena are not well captured by the present models. To improve the modeling one must consider issues such as strongly curved streamlines, buoyancy, strongly varying density, dilatation due to heat release and its influence on the overall dynamics of the flow, and local extinction and reignition of the flame chemistry in regions of high shear rates. In this study, the bluff-body laboratory flames of the University of Sydney are used as a prototype for the study of the modeling of these phenomena. A bluff body at the base establishes a recirculation zone that stabilizes the flame. This experimental work has covered a wide range of conditions between fast chemistry and close-to-blowout behavior. These experiments therefore exhibit many of the challenging phenomena occurring in more complex, practical flames. They are, however, of a simple axisymmetric configuration and have been experimentally well characterized. Thus, these flames are excellent targets for fundamental investigations into the modeling of complex flame behavior. The following research activities are proposed: (a) DNS and theoretical work towards better understanding of (1) the mechanisms of extinction/reignition in recirculation-stabilized jet flames, and (2) the physics and stability of flames dominated by large density differences, dilatation, curved streamlines, and buoyancy; (b) investigation/evaluation of LES subgrid-scale models that incorporate the new information gained on extinction/ reignition and on fluid-mechanics issues specific to recirculating flames; and (c) comparison of LES computations with laboratory data on bluff-body stabilized jet flames to test the performance of the subgrid-scale models developed in this project.