Efficient Cartesian Grid CFD-Based Methods for Aeroelastic Analysis of Wind Turbines

Wind power plays an increasingly important role in satisfying global clean electricity generation needs, and consequently there is a significant need to predict the unsteady loading related to turbine configuration, layout and off-design wind conditions. Many current design tools omit the fluid-structure interactions that drive costly fatigue loads, and thus, the research community has started using forensic/scientific-discovery level HPC CFD solvers to investigate these. Unfortunately, while HPC analysis tools that couple blade motion, flexibility, aerodynamics and fluctuations in the turbulent atmosphere have shown impressive predictive capabilities, they are too complicated and expensive for routine industrial use. These challenges and associated computational demands are exacerbated further, when considering the interactions of multiple aeroelastic turbines comprising a wind farm and there is need to better model the interaction phenomena and understand why many design codes overpredict wind farm power by 20% or more. Current design tools employ either low-order models that lack fidelity, or body-fitted CFD grids that are time consuming and require significant expertise in CAD idealization/simplification and grid generation. This paper describes an ongoing effort that seeks to addresses this bottleneck by integrating and enhancing emerging autonomous CFD methods and highly efficient Large Eddy simulation-based wake solvers into a robust CFD-based aeroelastic analysis for wind turbines/farms. The goal of the work is to develop a mid-fidelity CFD-based approach that offers significantly better resolved blade loading than current CFD actuator line/surface methods, but at a much lower barrier to entry (computational and labor cost) than forensic/scientific-discovery level HPC CFD solvers. This article summarizes the work to date pertaining to the assembly of a preliminary integrated analysis, validation and verification activities, and steps to reduce the level of expertise to use the tools.