Assessment of Predictive Capability of Hybrid RANS/LES Turbulence Models for Thermofluid Applications

Abstract

In this study, we have assessed performance of URANS model, various hybrid RANS/LES turbulence models such as detached eddy simulation, Nichols-Nelson HRLES model, dynamic HRLES (DHRL) model, as well as LES for two classes of problems: (a) heat transfer due to subsonic swirling flow subjected to a sudden expansion leading to cylindrical chamber, and (b) flow separation due to oblique shock wave-turbulent boundary layer interaction (STBLI). The results are assessed using the heat transfer characteristics, separation and reattachment characteristics, and capability to predict flow unsteadiness. The study indicates that URANS can predict large scale flow features reasonably well. However, it fails to resolve turbulence. PANS improves TKE prediction, hence, improves heat transfer prediction. Among the hybrid RANS/LES models, DHRL coupled with ILES is capable of providing accurate prediction of flow separation/reattachment characteristics for boundary layer flows. For free-shear dominated flows, implicit LES performs better compared to the explicit LES models.