In this work axial flux permanent magnet rotors are studied as parts of small wind turbines in off-grid systems. Trying to increase the power of such turbines brings significant challenges which are mainly connected to the best possible dimensioning of the generators and the following increase of mass and volume. Ιn this theses, we try to improve the optimization algorithms and study and tackle the consequences of power increase.
At the beginning of the work there is an introduction to the concepts of open source technology and intermediate technology, as well as the role that small wind turbines can have in meeting energy needs with respect to the environment and local communities.
A comparison between different optimization methods is carried out. Initially, mass is added as an additional criterion to the objective function which already contains cost and efficiency and the Particle Swarm Optimization (PSO) algorithm is adjusted to optimize the permanent magnet’s dimensions.
Generations of solutions are designed and simulated with four different Neodymium magnet grades (N40, N42, N45, N50) for six different rotor radii (2.4 to 3.9 meters with 0.3 meter per step) and for two different magnet thicknesses of 10 and 20 mm. A comparison of the different thicknesses and the different magnet grades is carried out.
Having found the optimal magnets for each generator we seek the optimal universal magnet which can be satisfactorily used for generators throughout the whole radius spectrum of the study. The universal magnet is then compared with the optimal magnets per rotor radius. Finally, a 5kW generator – designed to operate with a 3 meter diameter turbine – is manufactured, and its performance is compared between laboratory bench tests and finite element analysis simulation results.
Keywords: small wind turbines, permanent magnets, neodymium magnets, axial flux, standalone systems, efficiency-cost-mass optimization, universal magnet, open source, intermediate technology, particle swarm optimization (PSO)
Author: Petros Markopoulos
Responsible PhD: Kostas Latoufis / email@example.com
Supervising Professor: Nikos Hatziargyriou / firstname.lastname@example.org
PDF: Full version (Greek) and Short version (English)