Case C1 - Rotor fault detection

The aim of this case was to propose methods for on-line detection of rotor short-circuit faults and other faults in hydro generators. This case was carried out in close collaboration with HydroCen, and the work was carried out by NTNU postdoc Mostafa Valavi and master student Kari Gjerde Jørstad [1], [2] in close collaboration with the hydropower plant operators and MonitorX industry partners Eidsiva, Vattenfall and Statkraft. In the first stage of their work, an idea was evaluated to use available SCADA data for fault detection. However, simulation of the generator in healthy and faulty state by FEM (finite element method, electromagnetic field simulation) showed that this idea is not feasible. Thus, a new fault detection method was proposed, and its feasibility was evaluated in the second stage of the work.

The new method uses spectral analysis of stator voltage and current for fault detection. The results of the spectral analysis are illustrated for two examples in the figure, where the frequency spectrum of both a generator with healthy rotor winding and a rotor winding with faults are shown. In a case of an inter-turn short-circuit, in addition to the amplitudes at 50 Hz and its odd multiples, sideband harmonics appear at each side of the main harmonics. These sideband harmonics could be used as indicator for fault detection. The method requires a much higher data resolution (voltage or current) than usually available through the SCADA-system, and a sampling frequency of at least 500 Hz is recommended. However, the data collection must not necessarily be continuous, but samples of at least 2 seconds could be collected regularly, e.g. once in a day or week.

Frequency spectrum of induced voltage at no-load, healthy vs. 1 turn short-circuited (left), and frequency spectrum of stator current at full-load, healthy vs. 20 turns short-circuited (right). Courtesy of K. G. Jørstad [1]. PSD: power spectral density.

The detection of rotor inter-turn short-circuits was primarily investigated, but also detection of other types of faults, including eccentricity and bearing faults, were studied. A detailed description of the work and the results can be found in references [1] and [2].

[1] K. G. Jørstad, "Modelling, simulation and online detection of rotor fault in hydrogenerators," Norwegian University of Science and Technology, Trondheim, 2016.
[2] M. Valavi, K. G. Jørstad, and A. Nysveen, "Electromagnetic Analysis and Electrical Signature-Based Detection of Rotor Inter-Turn Faults in Salient-Pole Synchronous Machine," IEEE Transactions on Magnetics, vol. 54, no. 9, pp. 1–9, Sep. 2018.