Development of a Highly Compact and Robust Chemical Reaction Mechanism for Unsaturated Furan Oxidation in Internal Combustion Engines via a Multiobjective Genetic Algorithm and Generalized Polynomial Chaos

This work focuses on the development of a highly compact and robust chemical reaction mechanism for unsaturated furan oxidation in internal combustion engines. Recently, furans have gathered much attention in the research community as they are second-generation biofuels that do not compete with food supplies, unlike first-generation biofuels. Moreover, they are oxygenated and highly resistant to knocking when compressed. A four-component chemical reaction mechanism, consisting of furan, 2-methylfuran, 2,5-dimethylfuran, and toluene, is developed to simulate the combustion of unsaturated furans under engine-relevant conditions. First, a detailed furan, 2-methylfuran, and 2,5-dimethylfuran reaction mechanism from the literature is selected and reduced under engine-relevant conditions. Sensitivity analysis and the species rate of production are employed to identify the major reaction pathways of the respective furanic components under different conditions. After which, isomer lumping and reaction lumping are used to further reduce the size of the major reaction pathways and the skeletal major reaction pathways for furan, 2-methylfuran, and 2,5-dimethylfuran are included to a compact toluene base mechanism from the literature to form the final mechanism, consisting of only 62 species among 228 reactions. Subsequently, the pre-exponential A factors in Arrhenius equations of the furanic reactions are optimized via an in-house multiobjective nondominated sorting genetic algorithm, and successively, the generalized polynomial chaos is introduced to model the uncertainty of the input rate coefficient and its propagation in ignition behavior. The performance of the different furanic components in the mechanism is then validated against experimental data from the literature. The predicted results are in reasonable agreement with experimental results considering the compact size of the developed mechanism and the challenge of multiobjective optimization.