A comparison between two optimized TFPM geometries for 5 MW direct-drive wind turbines

This paper presents the method and result of optimizing a 5MW Transverse Flux Permanent Magnet (TFPM) generator using a genetic algorithm method. Because the main disadvantage of a TFPM generator is the price, a reduction in costs has to be considered in order to make it more attractive for the industry. The approach presented in this paper focuses on a reduction in mass of active materials, which constitute the generator, because the price of the machine is in direct relation with the amount of materials used for the construction. This strategy is applied for two types of TFPM geometries in order to asses which behaves better when subjected to optimization and which provides the best
result. In order to obtain the mass of active materials the entire analytical design has to be covered, making the analytical design to behave as a cost function for the optimization program. The innovation presented by this article consists on the fact that an optimization algorithm is constructed to be capable of solving any kind of cost function, even entire design procedures, as it is presented here.