Optimization of Multibrid Permanent-Magnet Wind Generator Systems

• Published in: IEEE transactions on energy conversion Vol. 24; no. 1; pp. 82 - 92
• Main Authors: Li, Hui, Chen, Zhe, Polinder, Henk
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.03.2009; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Analytical models; Cost function; Design optimization; direct-drive; Gears; Genetic algorithms; multibrid concepts; permanent-magnet (PM) generator; Power generation; single-stage; Studies; Synchronous generators; Wind energy generation; Wind power generation; Wind turbines
• DOI: 10.1109/TEC.2008.2005279
• ISSN: 0885-8969

This paper investigates the cost-effective ranges of gearbox ratios and power ratings of multibrid permanent-magnet (PM) wind generator systems by using a design optimization method. First, the analytical model of a multibrid wind turbine concept consisting of a single-stage gearbox and a three-phase radial-flux PM synchronous generator with a back-to-back power converter is presented. The design optimization is adopted with a genetic algorithm for minimizing generator system cost. To demonstrate the effectiveness of the developed electromagnetic design model, the optimization results of a 500-kW direct-drive PM generator and a 1.5-MW multibrid PM generator with various gear ratios are, respectively, compared with those from other methods. Then, the optimal design approach is further employed for a range from 750 kW up to 10 MW. The optimization results of PM generator systems including direct-drive and multibrid wind turbine configurations are obtained, and the suitable ranges of gear ratios for different power ratings are investigated. Finally, the detailed comparisons of the most cost-effective multibrid PM generator system and the optimized direct-drive PM generator system are also presented and discussed. The comparative results have shown that the multibrid wind turbine concept appears more cost-effective than the direct-drive concept.