Physics-based distinction of nonequilibrium effects in near-wall modeling of turbulent separation bubble with and without sweep
Phys. Rev. Fluids 10, 104602 – Published 14 October, 2025
DOI: https://doi.org/10.1103/gfr7-6l2m
Abstract
Pressure gradient-induced separation of swept and unswept turbulent boundary layers, based on the DNS studies of Coleman et al. [J. Fluid Mech. 847, 28 (2018) and J. Fluid Mech. 880, 684 (2019)], has been analyzed for various nonequilibrium effects. The goal is to isolate physical processes critical to near-wall flow modeling. The decomposition of skin friction into contributing physical terms, proposed by Renard and Deck [J. Fluid Mech. 790, 339 (2016)] (short: RD decomposition), affords several key insights into the near-wall physics of these flows. In the unswept case, the spatial growth term (encapsulating nonequilibrium effects) and the TKE production appear to be the dominant contributing terms in the RD decomposition in nonequilibrium zones, i.e., the separated zone and the upstream/downstream pressure gradient zones. Inspection of wall-normal profiles of these terms reveals that only the spatial growth term dominates the energy balance in the inner layer close to the separation bubble. This implies a strong need to incorporate nonequilibrium terms in wall-flux modeling of the unswept separation bubble. Comparison of the streamwise RD decomposition of swept and unswept cases shows that a larger accumulated Clauser-pressure-gradient parameter history in the latter enhances the outer-layer TKE production peak in the adverse pressure gradient zone. The spanwise RD decomposition in the swept case indicates that the downstream spanwise velocity profiles, including those within the separation bubble, largely retain the upstream zero-pressure-gradient characteristics. This seems to ease the near-wall modeling challenge in the separated region, especially for basic models with an inherent log-law assumption. Wall-modeled LES of the swept and unswept cases are then performed using three wall models with varying physical fidelity. Many of the above modeling implications from the DNS are verified by analyzing various LES flow quantities. In particular, extending the RD decomposition to wall models reveals the critical role of the spatial growth term near the separation bubble and demonstrates that the PDE wall model accurately captures this term.
