LAUSR

Multisector machines reveal a high fault-tolerant capability, since failure events can be isolated by de-energizing the faulty sector, while the healthy ones contribute in delivering the required power. This article is focused on the thermal analysis of multisector three-phase machines in healthy and faulty operations. First, a 3-D lumped parameter thermal network (LPTN) of a single sector is developed and finetuned against experimental data, through a genetic algorithm for identifying the uncertain parameters. According to the operating conditions, the varying housing surface temperature affects the heat exchanged to the ambient. Hence, an analytical formula is proposed to adjust the natural convection coefficient value depending on the operating condition. Then, the 3-D LPTN, modeling the whole machine, is built aiming at investigating the thermal behavior during faulty conditions. Finally, the complete 3-D LPTN is employed for predicting the machine thermal performance under several faulty conditions. Furthermore, the current overload experienced by the healthy sector (in order to keep the same torque level as during the pre-fault operation) is determined, in accordance with the magnet wire thermal class. The effectiveness of the 3-D LPTN in predicting the temperature is experimentally demonstrated.