Far-field drag prediction in CFD applied to sailplane geometries

Abstract

Discrepancies between Computational Fluid Dynamics (CFD) and experimental drag prediction calculations led to a renewed investigation into CFD drag prediction methods. From the recent High-Lift Prediction Workshop in 2017, a concern was raised that highlights the need for a deeper understanding of the fundamental physics underlying Reynolds- Averaged Navier-Stokes (RANS) models, especially in simple configurations. This dissertation focused on CFD drag force calculations for a sailplane airfoil and a streamlined body of revolution similar to a sailplane fuselage. The near-field method is commonly used in commercial CFD codes and is susceptible to numerical errors causing spurious drag. To address these errors and to improve computational efficiency, another drag prediction method known as the wake-integration method and referenced as the far-field method, was explored. This method utilises post-processing techniques based on momentum-based methods. At the time of this study, the exact location of the integration plane in the flow field behind the aerodynamic body cannot be determined mathematically. Furthermore, the effects of the integration boundaries and terms are not explored in depth for non-zero angles of attack. The dissertation will provide guidelines for the near-field surface integration and the farfield momentum-based wake integration drag prediction methods when applied to sailplane aerodynamics. This includes meshing procedures, inlet, outlet, upper and lower boundary locations, and the physics setup. For the far-field method, recommendations are provided for the integration plane location, integration boundaries and the effects of the integration terms for zero and non-zero angles of attack. The approach followed in this dissertation includes a mesh independence study and a validation of the far-field integration method. Subsequently, a method was proposed to determine the integration boundary location for a sailplane airfoil and a streamlined aerodynamic body such as a sailplane fuselage. A Python script was developed that utilises the composite trapezoidal rule that enables the numerical integration of flow variables and calculates the drag at the determined integration plane location. This process is very timeconsuming and provides limited insight towards the effects of the integration location and boundaries. Therefore, a more advanced method was developed within Simcenter STARCCM+. This method utilises the integrated field functions and surface integral reports within Simccenter STAR CCM+ combined with macros, leading to significantly faster integration and interpretation of wake data. Sufficient near-field and far-field results were achieved using the SST k − ω turbulence model coupled with the γ −Reθ transition model with the constitutive option set to cubic to allow for improved turbulence prediction in the boundary layer. The near-field method was more susceptible to spurious drag and mesh refinement than the far-field method, making the far-field method more robust. For a chosen integration location where the static pressure first stabilises, sufficient results were obtained for a 2D (Two-Dimensional) validation case using only the momentum term while neglecting the pressure and viscous stress terms. Utilising the method developed within Simcenter STAR-CCM+ to automatically integrate over all the Trefftz plane locations without integration boundaries, the calculated drag tends towards the experimental results and by applying the integration boundaries at 99.9 % of the free stream velocity results in a drag count difference of less than four drag counts at any integration station.