MONITORING ECCENTRICITY IN ELECTRICAL MACHINES

 

Parameter estimation, condition monitoring and diagnosis of electrical machines - Peter Vas OXFORD EDITION 2003 

 

1 Detection of static and dynamic air-gap in induction machines

An induction machine can fail due to air-gap eccentricity can occur due to shaft deflection, inaccurate positioning of the movement, and so on. If the air-gap eccentricity is large, then the resulting unbalanced radial forces (unbalanced magnetic pull) can cause rotor-to-stator rub, and this can result in the damage of the stator core and stator windings. It is possible to detect air-gap eccentricity in induction machines by using non-invasive techniques. For this purpose it is possible to utilize the monitored stator currents. Such a technique is now described.

There are two types of air-gap eccentricity: the static air-gap eccentricity and the dynamic air-gap eccentricity. In the case of the static air-gap eccentricity, the position of the minimal radial air-gap length is fixed in space. Thus for example, static eccentricity can be caused by the ovality of the core or by the incorrect positioning of the stator or rotor at the commissioning stage. Assuming that the rotor-shaft assembly is sufficiently stiff, the level of static eccentricity does not change. Due to the air-gap asymmetry the stator currents will contain well-defined components, and these can be detected.

In the case of dynamic air-gap eccentricity, the centre of the rotor is not at the centre of the rotation and the minimum air-gap rotates with the rotor. It follows that dynamic eccentricity is time and space dependent (static eccentricity is only space dependent). For example, dynamic eccentricity can be caused by a bent rotor shaft, wear of bearings, misalignment of bearings, mechanical resonances at critical speed, and so on. Due the dynamic eccentricity, side-band components appear around the slot harmonics in the stator line current frequency spectra.

It can be shown by some changes in the notation that, in general, the frequency components in the stator currents of an induction machine which are due to air-gap eccentricity can be obtained as

                            (1)

where   is the fundamental stator frequency,  is any integer,  is the number of rotor slots, and  is the eccentricity order number, which for static eccentricity is  and for dynamic eccentricity is . Furthermore  is the slip,  is the number of pole-pairs, and  is the harmonic of the stator m.m.f time harmonics (). It follows from that  gives the principal slot harmonic frequency, and due to the stator time harmonics there are  frequency components, if follows from eqn (1) that when the rotor slot number is higher, the resulting frequency components in the stator currents due to eccentricity are increased.

Thus by detecting a stator line current, and by using a frequency spectrum analyzer, it is possible to detect the presence of air-gap asymmetries by utilizing eqn (1). For example, if  Hz,  (four-pole machine), , and , it follows from eqn (1) that by considering , the principal slot harmonic frequency is obtained as  Hz and this is one of the stator current time harmonics which is present due to static eccentricity. The other frequency component due to static eccentricity can be obtained as  Hz. The two line current stator frequency components due to dynamic eccentricity are obtained from eqn (1) by considering  , and thus  and

 Hz are obtained.

The stator current can be monitored by using a simple clip-on current transformer around one of the supply cables to the induction machine. It is possible to utilize fully digital techniques for the monitoring of eccentricity in induction motors. In such a case the monitored line current signal is digitized by using an A/D converter and the spectrum analysis can be performed by using FFT. To minimize leakage, the sampling rate has to fulfill the Nyquist criteria (sampling to be performed at twice or more than twice the highest frequency component). For detecting sidebands around principal frequency components, ZOOM FFT must be used in order to obtain the desired frequency resolution.

It should be noted that when a signal is analyzed by conventional FFT techniques, a spectrum is produced, which covers a range from d.c. to a chosen maximum frequency. The resolution of the spectrum is determined by the size of the transform, that is, by the number of samples used to describe the signal. For example, for a 1 kHz transform,  samples are taken, and the resulting spectrum normally consists of N=400 frequency lines, which are evenly spaced. Thus the line spacing (resolution) is , or in terms of the sampling frequency , it is . It follows that when the analysis is performed in a given frequency range (0-), in case of normal FFT techniques, it is possible to increase the resolution only by increasing the transform size. However, by using ZOOM FFT, the resolution can be increased without increasing the transform size, but only a correspondingly smaller part of the original frequency range can be analyzed at a time. In case of conventional FFT for a given transform size, better resolution can be obtained either by decreasing , and thus losing the high frequency information, or by increasing the transform size (which requires more computational time). When using ZOOM FFT, the increased resolution can be obtained without losing the high frequency information, or increasing the transform size. If the resolution is changed by a factor of N, from , then the length of the signal (in the time domain) increases from , and the high resolution spectrum gives a more detailed spectrum compared to the baseband spectrum. ZOOM FFT can be implemented (digitally) in two ways. Either the increased  resolution is obtained by shifting the frequency range of interest, and then by using digital low-pass filtering, or by recording a longer time signal (increasing the samples by a factor of N) and by transforming it by parts using a smaller transform.

Finally in should be noted that instead of monitoring a line current, it is also possible to detect eccentricity in induction machines by analyzing the locus of the stator current space phasor.

 

2 Non-invasive monitoring of the air-gap of a hydroelectric generator

Air-gap eccentricity can occur on any size or design of a hydroelectric generator. Eccentricity can cause a rotor to stator rub without warning on a generator equipped with only conventional instrumentation. A non-invasive on-line air-gap monitor is now described, which is suitable for installation on almost any generator, and which can simultaneously protect the generator against air-gap failure resulting in rotor-stator contact while providing diagnostic data. The system is non-sensitive to electromagnetic interference and has a 0.2 percent accuracy over a range of air-gap lengths 2 to 40 mm.

Optical sensors are used to determine the air-gap length by measuring four time intervals () between pairs of four pulses produced by two sensor beams each time they pass a pair of retro-reflective strips on the opposing surface.

A LED light source (which can be a laser diode), couples 50-200 continuous microwatts into the optical fiber. The latter is connected with fiber-optic terminals into a four-port splitter/coupler. The splitter divides the light into two paths to the sensor head. In the sensor head, two lenses collimate the light exiting from the fibres into two marrow beams. The beams are projected at a known angle across the air-gap. These beams will intersect each of the pairs of the retro-reflective strips mounted on the opposing surfaces (while the rotor is moving) causing light to be reflected back to the source lenses. The reflected light-pulses are added together by the splitter/coupler and the sum is sent to the photodetector diode. This light is converted into a current by a photodiode and then into a voltage signal by an electro-optical interface card.

By using , the air-gap length can be described by the following linear relationship:

                  (2)