A NEWTON-KRYLOV BASED SOLVER FOR MODELING FINITE RATE CHEMISTRY

摘要

To date, computational fluid dynamics (CFD) codes aimed at solving practical engineering problems involving chemically reacting flow have incorporated relatively simple descriptions of the chemical mechanisms involved. Techniques are now available to create reduced mechanisms that faithfully represent detailed chemical descriptions over an appropriate range of conditions using many fewer species. However, including reduced mechanisms into a CFD analysis typically leads to numerical difficulties. In a recent project, a new modeling tool was created that utilizes a combination of state-of-the-art techniques used by Reaction Engineering International (REI) for modeling finite rate chemistry in chemically reacting flows using reduced mechanisms with emerging Newton-Krylov methods for solving systems of non-linear equations. For tests problems ranging from geometrically simple combustion problems to full-scale utility boiler simulations, the Newton-Krylov solver has reduced the CPU time to achieve a solution by up to 60% compared to our traditional Picard iteration method. This paper discusses the implementation of the Newton-Krylov solver into the REI combustion code, the impact of parameters on the performance of the Newton-Krylov solver for solving problems using reduced mechanisms, and demonstration of the Newton-Krylov solver on full-scale utility boiler NOx simulations.