• Baker, James
• Wrobel, Rafal
• Mellor, Phil
• Williamson, Sam
• Yon, Jason

Abstract

It is well understood that an electric machine’s output performance is limited by its losses and thermal behavior. For novel prototype machines, hardware testing processes are an important part of quantifying these parameters. For some machines, effective characterization may be accomplished using a series of static and simple prime-mover tests. The resulting data permits calibration of loss- and thermal-models. These can then be used to predict on-load performance. Fault-tolerant machines based on single layer winding arrangements are designed to minimize interaction between windings or module-groups. This paper demonstrates that, for such a machine, the losses measured during simple DC and primer-mover tests may be used to infer performance during both ‘healthy’ and ‘faulted’ operating modes. Under faulted conditions the total machine loss is expected to be a combination of module-specific and common losses, which can be directly deduced from hardware tests. This paper discusses the accuracy of loss superposition when applied to a 180 kW multi-channel, fault-tolerant aerospace machine. From observations following faulted and healthy dynamometry tests, there exists close correlation between full-load performance and estimates made from the superposition of losses under discrete operating modes.

Topics

• impedance spectroscopy