Improved pressure-gradient sensor for the prediction of separation onset in RANS models

TL;DR

This paper introduces an improved pressure-gradient sensor and alternative eddy viscosity models for RANS turbulence modeling, enhancing the prediction of flow separation in aerodynamic applications.

Contribution

The paper presents a new adverse pressure gradient sensor and two eddy viscosity models, calibrated with DNS and RANS data, to improve separation prediction in RANS models.

Findings

Enhanced prediction of stall and separation onset.

Better alignment with DNS and experimental data.

Robustness across different airfoil geometries.

Abstract

We improve upon two key aspects of the Menter shear stress transport (SST) turbulence model: (1) We propose a more robust adverse pressure gradient sensor based on the strength of the pressure gradient in the direction of the local mean flow; (2) We propose two alternative eddy viscosity models to be used in the adverse pressure gradient regions identified by our sensor. Direct numerical simulations of the Boeing Gaussian bump are used to identify the terms in the baseline SST model that need correction, and a posteriori Reynolds-averaged Navier-Stokes calculations are used to calibrate coefficient values, leading to a model that is both physics driven and data informed. The two sensor-equipped models are applied to two thick airfoils representative of modern wind turbine applications, the FFA-W3-301 and the DU00-W-212. The proposed models improve the prediction of stall (onset of separation) with respect to the prediction of the baseline SST model.