Acoustic noise and vibration of switched reluctance machines

• Main Author: Long, Stephen Andrew
• Published: University of Sheffield 2002
• Subjects: 621.46
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.436696

This thesis describes a systematic investigation into the sources of acoustic noise and vibration in switched reluctance machines, and encompasses the vibrational behaviour of the stator, the influence of control parameters, and an evaluation of the effectiveness of active vibration cancellation. The influence of leading design parameters, such as the width and number of poles and the yoke thickness, and geometric asymmetries, such as lamination notches in the stator core, and the effect of the stator windings, the frame, the end-caps and the mounting assembly, on the natural frequencies and modes of vibration are investigated, Chapter 3. Both two-dimensional and three-dimensional finite element analyses are employed, the predicted results being validated by measurements on various experimental models, which consequently highlights the limitation of the finite element technique for highly complex structures with discontinuities in their fabrication. The influence of the mass and stiffness of the laminated stator core and the stator windings on the natural frequencies and vibration modes is investigated, and effective material properties are deduced for the analyses. It is found that the number of poles and lamination notches on the stator influence the number of vibrational modes which occur in the audible frequency range due to the introduction of dual natural frequencies, viz. symmetrical and anti-symmetrical modes, which are shown to separate further in value as the asymmetries become more profound. As the diameter of the stator yoke is reduced the natural frequencies increase, whereas increasing the thickness of the yoke and the adding of a frame and end-caps significantly increase the natural frequencies. The effect of the stator poles is to significantly reduce the stator natural frequencies, which are irrespective to a variation to the width of the poles, a variation in their mass being annulled by the resulting change in stiffness. Similarly, it is shown that the winding mass and stiffness offset each other so that their influence is also relatively small, whereas, although quantification of the damping is not within the aims of this thesis, it is apparent that the windings introduce a high level of damping which consequently limits the magnitude of the vibrations and hence acoustic noise. Finally, the laminated nature of the core is quantified and is shown to affect the effective material properties compared to an equivalent solid core, and to increase the effective damping. Previous investigations have studied the influence of the drive control parameters, but generally limit the analysis to either the frequency or time domain or to measurements of the sound pressure level, and are generally carried out in isolation. Therefore. the influence of alternative operating modes and their associated control parameters on the acoustic noise and vibration of an SR machine is thoroughly investigated. the results being analysed in both the frequency and time domains, and compared with measurements of the sound pressure level, Chapter 4. The noise and vibration which results when the SR machine is operated under both voltage and current control. with both hard and soft chopping techniques, and various switching angles, and for various sampling and switching frequencies, is measured. The influence of speed and load is also investigated, and the vibration and noise are also investigated under single pulse mode operation. It is found that hard chopping results in a noisier operation than with soft chopping due to increased current ripple, especially under current control. The noise and vibration is clearly shown to differ under current control compared to voltage control and single pulse mode, due to the random switching of the phase voltages resulting in wideband harmonic spectra, thereby increasing the levels of all the mechanical resonances. Further, it is found that the noise and vibration increase with both speed and load. In general, the increases in noise and vibration are attributed to an increase in the rate of decay of current at phase turn-off, regardless of the control parameter under investigation. Finally, the effectiveness of active vibration cancellation for nOIse reduction is investigated under typical operating modes in Chapter 5, which, for the first time, is analysed in both the frequency and time domains, and validated by measurements of the sound pressure level. It is found that active vibration cancellation is less effective for machine stators which have more than one dominant vibration mode within the audible frequency range, since the technique is only capable of applying active cancellation for a single vibration mode, thus any further resonances remain unaffected. Further, during chopping control, especially current control which results in random switching, it has been shown, for the first time, that the effective time-delay varies to that applied, thus rendering the technique less effective. This is found to be attributed to the asynchronism of the final chopping edge and point of phase turn-off, therefore preventing the vibrations from being excited in anti-phase, as explained in section 5.6.