| dc.description.abstract | This thesis aims to optimize the total power output of a wind farm with 11 turbines by adjusting individual tilt angles for each turbine, and looking at what scenarios provide optimal tilts. This is done using PyWake to simulate the wind farm by implementing site and turbine conditions, as well as both Jensen (NOJ) and Bastankhah Gaussian (BG) wake deficit models. Three optimization algorithms, simplicial homology global optimization (shgo), Dual Annealing (DA), and Differential Evolution (DE) were compared during the initial simulations, using -30° and 30° as tilt bounds. The initial simulations included two wind directions of 180° and 190°, multiple wind speeds, and distances between turbines. Furthermore, the comparative simulations were done using only BG and DE, as well as a reduced amount of distances and wind speeds. These simulations consisted of changing turbines, increasing turbine power, multiple wind directions, and adding more rows. From the results, it can be seen that there seems to be a power gain for optimized tilts, and it becomes more prominent at either a low wind speed, below 8 m/s , or low distances between turbines, below 6D distance. It can also be concluded that the greatest power gain compared to 0 tilt came from BG, at a 3D distance and a wind speed of 9 m/s , with a gain of 3.71 MW, corresponding to 28.6%. Furthermore, it can be concluded that the furthest up and downstream turbines typically only resulted in optimized tilt angles of 0, while the remaining turbines had various tilts, and regularly reached maximum or minimum tilt. | |
