Blending of Algebraic Transition Model and Subgrid Model for Separated Transitional Flows

Abstract

Accurate and efficient prediction of separated transitional flows is always a big challenge for computational fluid dynamics (CFD), especially in engineering applications. The recently proposed multiple hybrid Reynolds-averaged Navier–Stokes/large-eddy simulation (HRLES) versions (such as DDES, SAS, and LDKM) of the transition model have shown good performance for such flows. But in terms of computational costs, they are still not saving enough. This is mainly because the above closures are driven by nine partial differential equations (PDEs) for a three-dimensional flow. Along with this, a novel hybrid closure is proposed that combines the SA model with BC algebraic transition function and the WALE subgrid model. It only consists of six PDEs and requires less than 30% CPU time per iteration as compared with the aforementioned counterparts. After calibration and basic validation for widely used zero pressure gradient flat plate, decaying isotropic turbulence test cases, and semi-infinite S809 wing, the new hybrid closure is applied to the flow over a circular cylinder from to . The mean drag results are well in line with the HRLES versions of model, but with less computational costs. More detailed simulations are also implemented for the circular cylinder at and rod-airfoil configuration at . The results further indicate that the new closure can capture the physics associated with the mean and fluctuating velocity of shedding vortex, frequency, and power spectral of vortex-induced noise.