A Computationally Efficient Turnkey Approach to Turbulent Combustion Modeling: From Elusive Fantasy to Impending Reality

摘要

The fundamental shortcoming of all current approaches to turbulent combustion modeling is the reliance on extensive a priori knowledge of a combustion system to be simulated. In traditional manifold-based turbulent combustion models such as "flamelet"-like models and Conditional Moment Closure, combustion processes are constrained to a single asymptotic mode of combustion, so the dominant mode of combustion must be identified a priori, if even possible for the multi-modal combustion processes typical of many practical combustion systems. Even in theoretically more general turbulent combustion models such as the Transported Probability Density Function approach and the Linear Eddy Model, their extreme computational cost requires the use of highly reduced chemical mechanisms that are valid over only a narrow region of thermochemical state space, which must be identified a priori. In order to enable the widespread adoption of computational simulations for turbulent combustion, a computationally efficient turnkey approach to turbulent combustion modeling is desperately needed. Such an approach must accommodate (1) multi-modal combustion, (2) non-adiabatic combustion, (3) multi-stream combustion (including multi-component fuels), and (4) pollutants, yet the approach must still be computationally efficient, ideally competitive with traditional manifold-based combustion models. In this paper, recent and ongoing efforts toward the development of such a model are discussed. The modeling framework relies on multi-dimensional manifold equations that are capable of accommodating non-adiabatic multi-modal combustion in its most general form. To accommodate multi-stream combustion, multiple mixture fractions can be introduced as additional manifold parameters. With such a complex model, online computation of the manifold and online convolution against an appropriate (subfilter) PDF manifold during LES or RANS calculations becomes necessary. To accommodate pollutants, the manifold can be cast as an equilibrium manifold for fast processes with slow pollutant processes modeled with separate explicit transport equations or as fully integrated non-equilibrium manifold, albeit at increased computational cost in the latter approach. Finally, a number of remaining issues and future directions are identified including the integration of compressibility effects into reduced-order manifold turbulent combustion models and the unification of turbulence modeling with reduced-order manifold turbulent combustion models.