96 M. El-Kafafy et al. Fig. 9.4 Evolution of the resonance frequencies (top) and damping ratios (bottom) of the 10 most dominant drivetrain modes during the tracking period obviously for modes 1,2,4, and 8. This scatter behavior of the modal damping ratios could be explained by two facts. The first fact is that the wind turbine is a multi-physical machine with significant interaction between its subcomponents, and for the instrumented wind turbine the rotor is directly connected to the gearbox. This means that any change in the ambient conditions (e. g. wind speed) will affect the damping characteristics of the rotor and hence the same will happen for the drivetrain unit that is directly connected to the rotor. Therefore, the change in the ambient conditions plays a role in the variation pattern of the modal damping over the tracking period. The second fact that cloud help in understanding the scatter behavior of the tracked modal damping is that most of the identified modes are closely spaced modes. This can be seen in Fig. 9.5 that shows a typical stabilization chart constructed by the Operational pLSCF (Polymax) estimator when applied to one data record in the frequency band of interest. Identifying two closely spaced modes always increases the uncertainty, especially on the damping values. A statistical analysis on the results obtained during the continuous monitoring is shown in Fig. 9.6, which illustrates one box and whisker plot per frequency and damping value of each tracked mode. These statistical results confirm the results shown in Fig. 9.4 where it can be seen from Fig. 9.6 that the frequency estimates are indeed more consistent than the modal damping estimates. In terms of the median value of the damping estimates, the results presented in Fig. 9.6 show a good agreement with the ones published in [14] that were obtained based on a short-term tracking. For the modal damping ratios, it can be seen from Fig. 9.6 that for most of the identified modes the distribution is symmetric around the median and the length of the tails is about the same length as the box. This gives indication that the damping estimates have a normal distribution over the tracking period. Figure 9.7 shows the boxplot for the MAC value calculated with respect to the reference modes set per identified mode. The MAC evaluates the degree of correlation between the identified mode shape and the reference one. All the modes, except modes 2,4, and 8, show a high correlation with the reference mode shapes with a median value of MAC higher than 70%. The low MAC value of those 3 modes (i.e. 2,4, and 8) agrees with their results in terms of the damping estimates where they showed the highest scatter as it was shown previously in Fig. 9.6.
RkJQdWJsaXNoZXIy MTMzNzEzMQ==