Colloquia

Samenvatting

Mitigation of climate change poses one of the biggest challenges in modern times. Across the globe, efforts are made to reduce greenhouse gas emissions, to limit global warming. These efforts have sparked the interest in renewable energy technologies, such as wind turbines, which harness the wind’s kinetic energy using large rotors. The cross-sectional shape of a rotor blade, i.e. the airfoil, heavily impacts the aerodynamic performance of the wind turbine, and must be designed with care.

The design of airfoils is often guided by Computational Fluid Dynamics (CFD): numerical tools capable of solving problems involving flows around solid surfaces. Panel methods are a subset of CFD, able of solving potential flow problems, which are heavily simplified with respect to reality. This allows for fast computations, but limited use cases. Complex flows are solved using other CFD methods, which tend to be computationally demanding. Demand arises for intermediate options: methods capable of solving more complex flows with reasonable accuracy, while maintaining a low computational demand.

The research presented here focuses on the extension of a standard 2D panel code, to allow evaluation of more complex problems. These include unsteady flows, severely separated flows and airfoils with blunt trailing edges. The numerical method is designed such that extension to 3D remains viable.

The developed panel method computes the pressure distribution and the lift force for an airfoil in arbitrary motion. Source and doublet distributions are employed and a Dirichlet boundary condition is applied on the airfoil surface. Wake panels with doublet distributions are shed over time, and are convected with the local flow velocity; the wake shape develops naturally. A non-linear Kutta condition and viscous vortex cores are used to increase the accuracy of the pressure distribution and wake geometry. Blunt trailing edges and severe flow separation problems are simulated by allowing non-zero normal flow on respectively the trailing edge base, and in the separation region. Here, we instead prescribe the pressure or velocity. The flow separation scheme requires the separation point to be known a priori.

Good agreement was found between the developed panel code and analytical solutions, existing potential flow solvers (Xfoil), and experiments, for a variety of test cases. It shows great potential for further development and extension to 3D. Some challenges persist though, as currently not all combinations of input parameters yield a well posed problem, and a solution may not always exist. This remains a topic for further research.