Comparative study of hybrid multi-timescale and g-scheme methods for MARCS with detailed chemical kinetics

To develop a multiscale adaptive reduced chemistry solver (MARCS) for computationally efficient modeling of a reactive flow, the Hybrid Multi-Timescale (HMTS) method and the G-Scheme have been evaluated and compared for both homogeneous auto-ignition and 1-D premixed spherical propagating flame calculations with detailed chemical kinetics of hydrogen, methane, dimethyl ether, and n-heptane. It is demonstrated that the dependence of CPU time on the number of species is linear and third-order, respectively, for HMTS and G-Scheme method. The CPU Time of G-Scheme increases dramatically when the number of species of the detailed mechanisms is increased due to the huge computation cost of matrix inversion and reaction mode decomposition. Specifically, the G-Scheme method is faster at the induction stage of ignition and the near-equilibrium condition due to the large integration time step determined by the method adaptively. The HMTS method is faster near the ignition point and for a large kinetic mechanism due to the fast integration at a small base time step. Therefore, the present results suggest that it is possible to develop an MARCS for computationally efficient modeling of combustion by adaptively selecting the HMTS method and the G-Scheme according to local combustion regimes and mechanism sizes and integrating with the co-related dynamic adaptive chemistry and transport method (CO-DACT).