ABSTRACT
The design of electric machines consists of two distinct stages: the dimensioning calculus stage and the verification calculus stage.
Dimensioning includes the choosing of materials, the topology, and the calculus of all geometrical parameters starting from the machine design specifications. Because the relationships between the machine performance indices and the geometrical dimensions are rather involved (and nonlinear), it is impossible to produce a set of inversion functions to distinguish the latter from the former:
pi = fi (X) i = 1,n (13.1)
where pi is the performance index from the design specifications X is the geometric variable and material properties vector fi is the involved (nonlinear) function
In general, there is more than one variable vector X that satisfies pi. Also, for some design specifications there may be no design solutions (e.g., unity power factor for an induction motor) with a certain type of electric machine. By classical nonoptimal design, a possible solution X is found that satisfies most if not all design specifications. The optimal design selects, after mastering a good part of all feasible variable values, the solution that satisfies a given optimum criterion (e.g., minimum initial cost, weight, or maximum efficiency).
