Numerical Investigation of Dynamic Stall using Delayed Detached Eddy Simulations

This thesis examines the feasibility of Delayed Detached Eddy Simulations (DDES) in terms of predicting aerodynamic loads and capturing complex flow physics relating to the dynamic stall process at a transitional Reynolds number of 200,000 and Mach number of 0.10, using NASA’s OVERFLOW 2.3 code. This investigation tests the performance and capabilities of state-of-the-art transition models in terms of their ability to capture the underlying viscous mechanisms from which dynamic stall is onset. This work focuses on the events leading up to stall, rather than the analysis of the stalled flow itself, hence the focus will be on the upstroke region. This is due to the modern-day modeling tools being more so calibrated for pre-stall. Furthermore, an accurate prediction of the pre-stall regime could have a significant impact in employing flow control technologies in the future. All grids generated are based on current best practices of Chimera Grid Tools (CGT). While current Large Eddy Simulations (LES) standards recommend a low y + value in the boundary layer, a conservative value of 0.4 is used. Deeper investigations into the flow physics are performed on the 2nd finest grid out of 3 configurations. While by nature it would be expected that the finest mesh would yield the most accurate results, another motivation for this study is to test capabilities performed on a scale parallel with current industry standards, using coarser meshes, and lower fidelity models, resulting in quicker turnaround times. It is found that the overall flow physics is captured in detail even with a mesh composed of roughly 16 million grid points. A leading-edge separation region and formation of a dynamic stall vortex is examined in detail, and results suggest that the role of a laminar separation bubble is significant in the dynamic stall process, which has otherwise been a controversial point in the past regarding this flow phenomenon. Furthermore, it is found that DDES matches up accurately with respect to benchmark LES results for the same case, but with coarser spatial and temporal resolutions. James Coder’s SA Amplification Factor Transport (AFT) transition model yielded the most accurate results when compared with LES results and shows promising results for future viii use in more high-fidelity case studies. Langtry and Menter’s SST correlation-based transition model and its SA counterpart developed by Medida and Baeder are also investigated in detail