Intermittency-Based Dynamic Temporal-Scale Hybrid Model for Separation Flows with Transition

High-Reynolds-number boundary-layer separation (e.g., circular cylinder flow and hump separation) is sensitive to the boundary-layer state. While large-eddy simulation (LES) can consistently capture the laminar-to-turbulent transition within the boundary layer and the downstream separated shear layer, it does so at a very high computational cost. Coupling an intermittency-based Reynolds-averaged Navier–Stokes (RANS) model with LES can reduce grid requirements, provided that a natural RANS-to-LES transition is ensured. This paper incorporates a dynamic temporal filtering method into the [Formula: see text]-based transition model, replacing the original [Formula: see text]-scale equation with a new dynamic temporal scale equation, which can be viewed as an LES with a temporal filter scale width [Formula: see text] that continuously approaches the energetic temporal scale [Formula: see text]. Furthermore, the [Formula: see text]-based RANS mode is naturally maintained in the freestream region to preserve calibrated turbulent decay. The proposed method has been validated using subcritical-to-supercritical circular cylinder flows and the hump-separated flow, demonstrating its capability to achieve accurate predictions from the boundary layer to the downstream separated shear layer.