Design optimisation and performance evaluation of flux switching machines for geared medium-speed wind generator drives

[摘要] ENGLISH ABSTRACT: As wind turbines become large, the cost of energy increases because of the employed drivetrain (geared or direct–drive). Consequently, non–conventional geared medium–speed (MS) generators aregaining relevance, potentially due to better compromise to both the generator and gearbox costs. Thestudy proposes a novel approach for the multi–objective design optimisation (MDO) of two variantsof geared MS flux switching machines (FSMs) in their simple radial–flux structures––the 12–statorslots/10–rotor poles (12/10) and 12/14 machines, with major emphasis on rare–earth–free designs forindustrial–scale applications. Based on finite element analyses (FEA), whereby torque density, torqueripple and power factor are prioritised, multiple design options are provided in different Pareto mapsfor the designer to make informed selections. From an initial optimal design comparison of the 12/10machines based on rare–earth permanent magnets (PMs) in different wind generator drivetrains at 10kW, the MS design yielded the best solution in terms of average torque densities per generator costs.Consequently, the focus on MS drivetrains was intensified to further evaluate the 12/10 and 12/14rare–earth PM–FSMs as their power level shifts from 10 kW to 3 MW. As an indication of potentiallyadopting rare–earth–free designs, an improvement in terms of increased torque densities and reducedtorque ripple values is obtained for the rare–earth designs at 3 MW due to a number of factorssuch as variations in their PM volumes and relative differences in their saliency ratios. Moreover, dueto the optimal behaviour of key design parameters in the MDO environment, the superior performanceof the 12/14 machines at 10 kW are reversed at 3 MW. Disappointingly, based on the samedesign requirements, the nominated rare–earth–free designs such as ferrite PMs and wound–field(WF) coils do not produce better torque ripple effects compared to rare–earth PMs, as should be expectedfor such flux–focusing machines. However, an inherent tradeoff was found between their aspectratios and armature current densities which influence the active mass, especially for industrial–scale designs. Consequently, to ensure the feasibility of the optimal design, it may be needful to appropriatelyrestrict the boundaries of the aspect and split ratios before engaging them in any MDOprocedure. In another instance, it is found that it may be better to pursue MDO problems e.g., of WF–FSMs, by concentrating more on the performance (torque ripple and power factor) than on the cost ofthe machines. Interestingly, it was also found that the cheapest MS generator, even when comparedwith similar conventional wind generators at 3 MW, is the WF–FSM. Eventually, to validate the seriesof FEA prediction made during the study, a 10 kW WF–FSM generator prototype is selected,manufactured and tested, with certain novel implementations. Based on measured no–load, short–circuit, thermal, uncontrolled–normal and overload resistance, as well as current–controlled tests, the design feasibility as well as the conceptualisation of the proposed wind generator drivetrains has beenproven beyond reasonable doubt. In agreement with Chen et al (2011) [44], there is, indeed, a brightfuture for FSMs. The study is concluded with a general conclusion and recommendations for the future.

[发布机构] Stellenbosch University