Parallel on-the-fly adaptive kinetics in direct numerical simulation of turbulent premixed flame

Abstract

A new numerical framework for direct numerical simulation (DNS) of turbulent combustion is developed employing on-the-fly adaptive kinetics (OAK), correlated transport (CoTran), and a point-implicit ODE solver (ODEPIM). The new framework is tested on a canonical turbulent premixed flame employing a real conventional jet fuel mechanism. The results show that the new framework provides a significant speed-up of kinetics and transport computation, which allows DNS with large kinetic mechanisms, and at the same time maintains high accuracy and good parallel scalability. Detailed diagnostics show that calculation of the chemical source term with ODEPIM is 17 times faster than with a pure implicit solver in this test. OAK utilizes a path flux analysis (PFA) method to reduce the large kinetic mechanism to a smaller size for each location and time step, and it can further speed up the chemical source calculation by 2.7 times in this test. CoTran uses a similar correlation method to make the calculation of mixture-averaged diffusion (MAD) coefficients 72 times faster in this test. Compared to conventional DNS, the total CPU time of the final framework is 20 times faster, kinetics is 46 times faster, and transport is 72 times faster in this test.