Flame Acceleration and Detonation Re-Initiation in Non-Premixed Rotating Detonation Engines

Reliable detonation onset is essential for rotating detonation engine (RDE) operation, yet initiation mechanisms under continuous non-premixed injection remain unexplored. Through synchronized high-speed imaging (single-pass schlieren, double-pass shadowgraph, and [Formula: see text] chemiluminescence) in an optically accessible methane/oxygen obround RDE, the influence of reactant conditions, centerbody confinement, and channel curvature on flame acceleration, detonation re-initiation, and stability was investigated. After the predetonator wave diffracts into the combustor and decouples, RDE-specific re-initiation pathways rely on flame-driven compression waves, Mach stem development in stratified reactants, and hotspot autoignition within heated turbulent shear layers behind the Mach stem. Centerbody confinement yields tulip-shaped turbulent flames transitioning rapidly to overdriven detonation, accelerated by higher equivalence ratios, whereas hollow combustors exhibit spatially stochastic re-initiation via asymmetric flame–wall interactions; greater channel curvature delays re-initiation via enhancing rarefaction, yet raises incident angles and strengthens the Mach stem. Nevertheless, re-initiation cannot guarantee self-sustained detonation without sufficient energy release for re-establishing detonation after subsequent wave collisions. The hollow combustor extends lean detonability limits and promotes earlier detonation establishment compared to centerbody cases. The curvature transition induces velocity deficits and wave strength disparities, intensifying instabilities in multiwave collision regimes. These insights are vital for extending operational envelopes and enhancing deflagration-to-detonation-transition-based ignition reliability in non-premixed RDEs.