Crossflow turbulent production using hybrid Reynolds-averaged Navier–Stokes/large-eddy simulation approaches

Highlights: • DHRL simulation designed for scalar mixing and compared to IDDES. • Normal and inclined ethylene jet in Mach 2 crossflow with momentum ratio of 0.5. • Turbulent productions preserved for resolved and modeled portions of DHRL. • DHRL more consistent with experimental mean and variance ethylene concentration. • Inclined injection requires transfer of inflow RANS production into the interaction region. - Abstract: The objective of the study is to investigate the performance of a newly developed dynamic hybrid Reynolds-averaged Navier–Stokes/large-eddy simulation (DHRL) framework versus a traditional hybrid approach in the context of the crossflow mixing. An independent scalar-based transition parameter is introduced to partition the species turbulent production within DHRL. The models' capability in resolving turbulent structures, agreeing with experimental measurements, and transferring turbulent production between modeled and resolved parts of the domain are discussed. Unsteady simulations corresponding to non-reacting ethylene injected in Mach 2 air crossflow are performed. The simulations correspond to both normal and low-angled injection with a momentum ratio of 0.5. Time-averaged fuel concentrations are compared with experimental Raman scattering data and with simulation results using a traditional improved delayed detached eddy simulation (IDDES) model. For the normal injection case, both DHRL and IDDES show reasonably accurate results, although DHRL is more consistent with the measured concentrations. The inclined jet case requires a large modeled contribution to the unsteady region since jet breakup and transition is less pronounced than the normal injection case. DHRL is able to couple Reynolds-averaged Navier–Stokes (RANS) and large-eddy simulation (LES) model terms more effectively and produce concentration fields consistent with the experiment. IDDES fails to predict unsteadiness and produces concentration distributions with large discrepancies from the experiment.