Enhancing chemical kinetic mechanism optimization using independent component analysis

The rate constants in chemical kinetic mechanisms are defined by Arrhenius parameters, which often exhibit correlations among parameters, posing challenges for efficient mechanism optimization. The present study introduces a novel approach, REaction Parameter space Reconstruction via Independent Component Analysis (REPRICA), which facilitates the selection of physically admissible Arrhenius parameters using Independent Component Analysis during optimization. REPRICA is integrated into an evolutionary algorithm-based optimization framework, and its applicability is demonstrated through the optimization of a chemical reaction mechanism for NH 3 -based mixtures, targeting widely various experimental datasets such as ignition delay times and species concentration profiles. The results show that, compared to the previous method, the probability of selecting valid Arrhenius parameters is significantly increased and a robust chemical kinetic mechanism with improved predictive accuracy is obtained efficiently. • A novel scheme using Independent Component Analysis, REPRICA, is proposed for chemical kinetic mechanism optimization. • The REPRICA method enhances the quality of Arrhenius parameter selection and significantly reduces the number of penalties during optimization. • Enhanced parameter selection synergizes with the elitist selection strategy of evolutionary algorithms, enabling more efficient optimization. • The REPRICA-based optimization is generally applicable to chemical kinetic mechanisms and can contribute to improving their accuracy.