2014 年 56 巻 176 号 p. 156-165
Performance and accuracy of several time integration methods for chemical reaction equations are comprehensively investigated, aiming at an efficient reacting flow simulation with large detailed chemical kinetics. In this study, a modified CHEMEQ2, dynamic multi-timescale method (MTS), and two Runge-Kutta-based methods (R-K-Chebyshev and R-K-Fehlberg) are considered as currently available and possible explicit time integration methods, while VODE is used as a reference implicit time integration method. Ignition problems for three hydrocarbon systems (CH4, n-C7H16, and n-C10H22) and an internal combustion engine model with n-C7H16 are simulated. The results for both problems show that the modified CHEMEQ2 shows the best performance for all the conditions in not only the accuracy, but also the robustness, while MTS gives less performance and the two Runge-Kutta-based methods cannot work even for the ignition problems with hydrocarbon systems. It is also found that the two explicit time integration methods (CHEMEQ2 and MTS) reduce nearly 20-100 times computational time compared to the reference implicit time integration method.