DES vs RANS: The flatback airfoil case

George Papadakis,Vasilis A Riziotis,Marinos Manolesos,Konstantinos Diakakis

doi:10.1088/1742-6596/1618/5/052062

George Papadakis, Vasilis A Riziotis + Show 2 more

Open Access

https://doi.org/10.1088/1742-6596/1618/5/052062

Copy DOI

Abstract

Using flatback airfoils at the root of wind turbine (WT) blades is becoming more popular as the WTs increase in size. The reason is that they provide significant aerodynamic, aeroelastic and structural benefits. However, due to the blunt trailing edge (TE), the wake of such airfoils is highly unsteady and rich in three-dimensional vortical structures. This poses significant challenges on the numerical simulation of the flow around them, given the highly unsteady, three-dimensional turbulent character of their wake. In this work, computational predictions for a flatback airfoil employing both RANS and DES approaches on three successively refined grids up to 25 million cells are compared with available experimental data. Results suggest that even though URANS and DDES are in good agreement in terms of lift and drag, RANS simulations fail to accurately capture the turbulent wake unsteady characteristics.

Highlights

The practice of incorporating flatback airfoils, i.e. airfoils with a blunt trailing edge (TE), at the root of wind turbine (WT) blades is becoming popular as WT rotor diameters increase in size
In order to quantify the effect of grid resolution on the predictions, three grids with varying resolutions are generated, consisting of approximately 5, 10 and 25 million elements, respectively
The same computational mesh is used for both Unsteady Reynolds-Averaged Navier Stokes (URANS) and Delayed DES (DDES) simulations to avoid grid related deviations

Summary

Introduction

The practice of incorporating flatback airfoils, i.e. airfoils with a blunt trailing edge (TE), at the root of wind turbine (WT) blades is becoming popular as WT rotor diameters increase in size. Computational predictions for a flatback airfoil employing both RANS and DES approaches on three successively refined grids up to 25 million cells are compared with available experimental data. Results suggest that even though URANS and DDES are in good agreement in terms of lift and drag, RANS simulations fail to accurately capture the turbulent wake unsteady characteristics.

Results

Conclusion