A run time combustion zoning technique towards the EDC approach in large-scale CFD simulations

Abstract

Purpose – This paper aims to present a novel run time combustion zoning (RTCZ) technique based on the working principle of eddy dissipation concept (EDC) for combustion modeling. This technique selectively chooses cells in which the full reaction mechanism needs to be solved. The selection criterion is based on the concept of differentiating between combustion and the non-combustion zone. With this approach, considerable reduction in computational load and stability of the solution was observed and even the number of iterations required to achieve a stable solution was significantly reduced. Design/methodology/approach – Computational fluid dynamics (CFD) simulations of real life combustion problems such as industrial scale flares, fuel fired furnaces and IC engines are difficult due to the strong interactions of chemistry with turbulence as well as the wide range distribution of time and length scales. In addition, comprehensive chemical mechanisms for hydrocarbon combustion may include hundreds of species and thousands of reactions that are known in detail for only a limited number of fuels. Even with the most advanced computers, accurate simulation of these problems is not easy. Hence, the modeler needs to have strategies to either simplify the chemistry or to improve the computational efficiency. Findings – The EDC turbulence model has been widely used for treating the interaction between turbulence and the chemistry in combustion problems. In an EDC model, combustion is assumed to occur in a constant pressure reactor, with initial conditions taken as the concentration of the current species and temperature in the cell. With these assumptions, EDC solves the full or simplified reaction mechanism in all the grid cells at all iterations. Originality/value – This paper presents a novel RTCZ technique for improving the computational efficiency, when the EDC model is used in CFD modeling. Considerable reduction in computational time and stability of the solution can be achieved. It was also observed that the number of iterations required to achieve a converged solution was significantly reduced.