Kinetic mechanism reduction by using a genetic algorithm

One of the biggest challenges for the simulations of reacting flow systems using detail chemical mechanisms is the large species number and the broad distribution of species time scales. To achieve efficient and accurate modelling of combustion, reduced mechanisms based on a rigorous reduction algorithm are needed. In this study, a new Genetic Arithmetic Reduction (GAR) method for mechanism reduction is proposed and investigated numerically. An optimally reduced kinetic mechanism can be obtained from the maximization of the target function. This mechanism can accurately reproduces the predictions of the reference detailed mechanism within specified tolerances. The GAR method is employed to obtain several reduced mechanisms for n-decane/air oxidation, which are tested via simulation of homogeneous ignition process. The comparisons between results predicted by the detailed and reduced mechanisms demonstrate that the present method can obtain an optimized reduced mechanism at a given species number. The error of the reduced mechanism is found to be controlled by the tolerances in the GAR. By increasing error threshold, more species and reactions can be eliminated. The mechanisms generated by the current GAR method show significant reduction in the CPU requirement. The computation efficiency can be further improved by integrating the reduced mechanism with a dynamic multi-time scale (MTS) method.