Ricerc@Sapienza

In the design of modern, high-performance switched reluctance machines, the highly restrictive sets of constraints and requirements severely limit the number of feasible solutions. In order to improve the chances of attaining a successful design, this work proposes a novel and fully analytical approach to the preliminary design analysis. Initially, the correct number of independent design variables is identified. Subsequently, constraints and requirements are introduced one by one, in order to progressively discard all of the unfeasible candidates. At the end of this process, the actual design space is attained, whose main characteristic is to be populated only by feasible candidates. A design case study shows how the proposed methodology allows: 1) to verify the design feasibility; 2) to reduce the number of candidates by orders of magnitude; and 3) to gain useful insight into the design problem and thus select the most convenient strategy to finalize it. As a result, a highly effective design process is attained, so that considerable computational resources can be saved.