An improved 2-pentanone low to high-temperature kinetic model using Bayesian Optimization algorithm

摘要

The ignition delay times (IDTs) of 2-pentanone (methyl propyl ketone, MPK) were measured at equivalenceratios of 0.5, 1.0, and 1.5, pressures of 1 bar and 5 bar, and temperatures ranging between 1227 and1571 K. A MPK low to high-temperature model was constructed on the basis of Pieper model and Fenardmodel. The improvement of the model is that the rate constants of sixteen MPK decomposition and hydrogenabstraction reactions (R1–R16) obtained by the analogy-based method, were globally optimized bythe Bayesian Optimization algorithm using the high-temperature IDTs. The optimized model well predictsthe laminar flame speeds (measured by Li et al.) and the IDTs and species profiles in low-temperature(measured by Fenard et al.). There is no overfitting during the optimization process. Therefore, the optimizationmethod and the optimized MPK model are reliable. The comparisons of the optimized modelwith Pieper model and Fenard model were performed by the reaction pathway analysis and sensitivityanalysis for the predictions of the MPK profile, the low and high-temperature IDTs and the laminar flamespeeds. Because of the difference in the molecular structure between MPK and 2-butanone, remainingthe branching ratio among three MPK decompositions (R1–R3) and the ratio among the rate constantsof the MPK hydrogen abstraction reactions by OH, H and CH 3 unchanged is unnecessary in the developmentof the MPK model using the analogy-based method, while the branching ratio of the hydrogenabstraction reactions at carbon 1, 3, and 5 remains unchanged. The competition among the decompositionand hydrogen abstraction reactions of MPK is intricate and crucial to the low and high-temperatureoxidation. It is necessary to globally optimize the rate constants of R1-R16 based on the multidimensionalexperimental results, such as low and high-temperature IDTs, laminar flame speeds, and species profiles.