Computationally efficient prediction of statistical variance in the AC losses of multi-stranded windings

In the volume manufacture of electrical machines with multi-stranded windings the disposition of the individual conductors within the slot cannot be precisely defined. This randomness in conductor lay can manifest as a significant variation in the winding AC losses and thermal performance. This paper presents computationally efficient techniques that can be used to quantify the statistical variation in AC loss arising from the conductor placement. The impact of winding temperature and frequency is described using a simplified behavioural model, eliminating the need for FEA analysis at multiple operating points. Concentrated and distributed wound machines are used as case studies to test the methods.