Electromagnetic flux monitoring for detecting faults in electrical machines

Abstract

The ability of the electromagnetic flux measured in various locations of a 35-kW cage induction motor to provide useful information about faults was investigated. The usefulness of this monitoring parameter was assessed in comparison with some other electrical parameters used for fault detection, such as stator phase current and circulating currents between the parallel branches of the stator winding.

The following faults were investigated in this thesis: a turn-to-turn short circuit in the stator winding; rotor cage-related faults (breakage of rotor bars); static and dynamic eccentricity, and bearing fault.

The relevant fault signatures of the studied electrical parameters were obtained from measurements and/or from numerical electromagnetic field simulations in steady state. These signatures were analysed and compared in order to deduce the most appropriate quantity for the detection of a specific fault. When and where possible, the accuracy of different fault signatures issuing from numerical electromagnetic field simulations was validated by experiments. This investigation is essential since, following a good agreement, it may be assumed that if a monitoring system cannot detect and diagnose an artificial fault from the virtual measurement signals, it is hardly likely to work with real electrical machines, either. In this respect, the numerical methods of analysis limited the present study to such faults that affect the electromagnetic field of a machine.

On the exclusive basis of data obtained from simulations, a study of the modifications brought by various stator winding designs to some of the asymmetrical air-gap electromagnetic flux density harmonics responsible for the detection of various faults was carried out.

The analysis of a core fault (insulation fault in the stator lamination) artificially implemented in the numerical electromagnetic model of the machine in terms of finding a suitable parameter to sense such a fault was also studied in this work.