Impact of operator splitting schemes on detonation convergence

In reacting flow simulations, operator splitting allows one to treat the impact of chemical reactions, which represents an extremely stiff dynamical system, with a stiff time integration scheme in isolation by considering each cell in the domain as a local reactor. As such, in standard operator split implementations, the coupling due to convection and diffusion is misrepresented. Despite its widespread use, this decoupling has been shown to introduce splitting errors that may eventually lead to unphysical results that can corrupt simulations, such as incorrect autoignition or extinction times. In this presentation, we focus on isolating the impact of these splitting errors in convection-dominated compressible reacting flow problems, namely detonations. More specifically, through a grid convergence analysis of quantities of interest in the detonation structure, various integration schemes (Strang splitting, balanced splitting, spectral deferred correction) are compared with fully coupled unsteady baselines and steady state detonation profiles.