High-resolution numerical simulation of rotating detonation waves with parallel adaptive mesh refinement

Simulations of rotating detonation engines are still dominated by solvers on uniform or statically refined meshes. Here, simulations of premixed rotating detonation waves are conducted using the block-structured adaptive mesh refinement (SAMR) technique. The studied configurations include both a two-dimensional unrolled model with a discretely injected hydrogen-air mixture and a three-dimensional annular model with non-premixed and partially premixed hydrogen-air mixtures. The computations employ a generic solver within the parallel Cartesian adaptive mesh refinement framework AMROC, which has been extended to accommodate curvilinear meshes. A second-order accurate finite volume method for the Navier–Stokes equations is utilized, along with grid-aligned Riemann solvers for thermally perfect gas mixtures. Detailed, multi-step chemical kinetic mechanisms are employed and incorporated with a splitting approach. A study into mesh dependency is undertaken, providing an assessment of the influence of local mesh refinement and adaptation criteria on the simulation results. The analysis reveals the formation of a multi-wave structure and transient heat release patterns, indicating the presence of an irregular cellular structure with enhanced local heat release as the detonation propagates through the injection jets. The ability to resolve sub-scale phenomena down to the cellular structures, intrinsic to detonation propagation, demonstrates the benefit of the SAMR approach. Further simulations are conducted to investigate the effects of partial premixing on rotating detonation. Additionally, a workload distribution analysis demonstrates how the on-the-fly partition strategy in AMROC alleviates computational imbalances. Parallel scaling tests exhibit linear acceleration in solving rotating detonation engine problems, highlighting the efficiency of the parallel adaptive mesh refinement technique in capturing the primary features of these simulations. • An adaptive curvilinear viscous detailed-chemistry solver for RDE is described. • Cell savings around 90% allow for resolution of sub-scale cellular structures. • RDE configurations running on H 2 -air and C 2 H 4 - O 2 - N 2 are studied in 2D and 3D. • The significance of local resolution for reliable predictions is demonstrated. • The influence of refinement criteria and parallel solver performance are assessed.