Reduced-Order Manifold Models for Non-Adiabatic Turbulent Combustion

Abstract

Turbulent combustion is a problem critical to the design of engines, gas turbines, and other practical combustion devices; to control emissions, the design of these technologies must improve, and predictive simulation is critical to achieving this goal. However, many modeling challenges must be addressed to improve the quality of these simulations. For example, an important factor in the prediction of the formation of pollutants is heat loss, but for turbulent combustion models that rely on reduced-order manifolds, assumptions leading to specific manifold models are not well understood. The goal of this dissertation is three-fold: to determine the assumptions made by reduced-order manifold models that include heat loss, to create a framework through which to evaluate and quantify the validity of these assumptions, and to build a general manifold which requires fewer assumptions.

To examine these assumptions, the different heat loss models are then examined in the context of both premixed and nonpremixed turbulent flames. In the case of the premixed flame, boundary enthalpy and source term models for including heat loss are with no prior indication of the assumptions being made. A two-dimensional manifold equation for nonadiabatic premixed combustion is constructed from the ``bottom up'' through a transformation of the physical space governing equations to assess the assumptions, and a time scale analysis is used to characterize the effect of the assumptions. Here, the most restrictive assumption is satisfied, so all manifold models with heat losses considered perform equally well. For the nonpremixed flame, a similar manifold equation for non-adiabatic nonpremixed combustion is derived to evaluate the assumptions in manifolds with heat loss ``top-down'', and the assumptions leading to specific models are identified. Here, the mode of heat loss is shown to have an effect. The reasons the assumptions are only partially met in this case are shown through a similar time scale analysis.

A broader model is then constructed for multi-modal, non-adiabatic combustion. This work contains a detailed derivation, which relies on a three-dimensional manifold equation. Closure for this equation is outlined, and the two- and one-dimensional limits of such a model are shown.