The slip ring (see Figure 1) is a crucial component in the generator of a wind turbine, which electrically connects the moving parts of the generator to the static components. During service the ring is subject to severe wear from continuous contact with the brushes both because of debris collected by the brushes and the continuous change of direction of the generator axis that follows the wind direction. The wear is usually uneven so that the ring becomes increasingly out of round in time i.e. defining the ring circumference by the function R(Θ) in polar coordinates with the pole on the slip ring axis of rotation, R varies with Θ and increasingly so during the service time (see Figure 2).
Development Activities. The use of a shock accelerometer for the continuous in service monitoring for wear of the slip ring on a wind turbine generator is assessed and supporting results are presented. Five wear defects in the form of out of round circumference acceleration data with average radial dimensions in the range 5.9µm to 25µm were studied (see Figure 2 for examples). A theoretical model of the acceleration of a point on the circumference of a ring as a function of the defect profile was developed as part of the project. For all defect profiles it is shown that the acceleration is proportional to the square of the ring rotation frequency. So the potential signal to noise ratio in the measurement of the ring acceleration increases with frequency squared. Correspondingly the potential sensitivity in the detection of defects through acceleration measurement increases with frequency squared.
Acceleration data as a continuous function of time has been obtained for ring rotation frequencies in the range 370-3827 revolutions per minute, which span the range of frequencies arising with the variation of wind speeds experienced under all in service conditions. A statistical analysis of the root mean square of the time domain acceleration data as a function of the defect profiles, rotation speeds and brush contact pressure has been performed. The detection performance is considered in terms of the achievement of a signal to noise ratio exceeding 3 (99.997% defect detection probability). Under all conditions of rotation speed and pressure this was achieved for average defect sizes as small as 10µmm, i.e. only 0.004% of the ring diameter.
These results form the basis of a very sensitive defect alarm system. However the defect sizes do not correlate significantly with the acceleration and brush pressure data. Furthermore, whilst the measured root mean square acceleration follows an overall increasing trend with frequency for all defects at all brush pressures, it does not follow the square law behaviour of the ring circumference acceleration. Instead there is a strong oscillatory component. Therefore further work is in hand to model these effects quantitatively, with the ultimate aim of sizing out of round defects with the proposed technique.
Conclusions: Defects with an average radial depth > ~= 10µm and maximum depth of > ~= 0.11mm (which is only 0.05% of the ring diameter) can be detected with a S/N ratio > ~= 3 i.e. a detection probability > ~= 99.997%. Smaller defects with average depths of 5.9µm and 0.035mm average could be detected to the same reliability level at high rotation speeds. In principal the method is suitable for an automated warning system whereby an alarm is triggered when a defect exceeds a pre-decided level. However defect sizing cannot currently be achieved from calibration curves.
For more information, please contact us.