Abstract

Research of low to intermediate Reynolds numbers flows (1000 < Re < 1000000) has become essential in the last two decades due to increasing interest in small Unmanned Aerial Vehicles (UAVs), small wind turbines, and exploration of planets such as Mars. Simulation of many of these applications need an accurate prediction of aerodynamic forces within five percent accuracy. Therefore, developing high accurate Computational Fluid Dynamics (CFD) tools is essential to design and optimize these systems.
Simulation of a Vertical Axis Wind Turbine (VAWT) on Mars is considered as the motivating application in this project. It is motivated by current plans of sending humans to Mars in the next two decades, and the need to investigate sustainable way to generate power on the planet.
VAWTs have a simple geometry but the flow structure around the blade is known to be one of the most complex flow in aerodynamics. Separation, vortex shedding and dynamic stall frequently occurs on the turbines’ blade. Therefore a tool that accurately simulates the flow around wind turbines, accurately, is needed.
Large eddy simulation has proven to be a reliable turbulence model with the capability of simulation flows with regions of separation and transition to turbulence. Growth of the computational capability along with development of more accurate numerical methods and new advanced LES models has permitted the use of wall-resolved LES.
In the current dissertation, Wall-Adapting Local Eddy-Viscosity (WALE) simulation is used to simulate flow around a vertical axis wind turbine. Using a second-order accurate discretization technique, both in space and time, with a low-dissipative method, enforced by an adjustable upwinding factor, achieves the required accuracy in Large Eddy Simulation. The proposed approach enables us to accurately predict the shear stress and pressure distribution on the blade. Therefore, dynamic stall location is spotted precisely.
The potential to increase the performance of small VAWTs by using Morphing blades are very promising. An in-house code has been extended to simulation flow around dynamically morphing blades. Therefore, Arbitrary Lagrangian Eulerian (ALE) is used to preserve the second order accuracy of the numerical scheme under a dynamic mesh, and a combination of spring and diffusion methods is used to adjust the mesh dynamically around deforming blades. A morphing blade scenario is presented to show the new capability developed.