«We were able to show that SFRA is more sensitive than the RSO method our company has been using, in addition to establishing objective evaluation criteria for rotors.» Pascal Fröhlich, Academic employee, EMIS Electrics GmbH What was the motivation behind the thesis? Johannes Ohde: When we serviced the rotors of the machines at one of our sites, we previously tested them with impulse voltage at the respective winding access points. This recurrent surge oscillograph (RSO) measurement feeds several defined impulses into the connected winding. After measuring the response in the time domain, we reverse the measurement and injection direction and compare both responses again in the time domain. This method is very subjective, especially in the evaluation, so we were looking for a sensitive procedure that allows objective evaluation criteria. Pascal, can you explain your approach to frequency response analysis? Pascal Fröhlich:In collaboration with OMICRON, I immediately found that the sweep frequency response analysis is a highly sensitive method for detecting interturn short circuits in rotor windings. However, a reference measurement is usually required to compare results. This procedure is applied to transformers where a comparison to a fingerprint measurement detects damages in the winding and other parts. This is an obstacle for service providers like us because reference measurements are hardly ever available to us. Based on my experience with the RSO method, I investigated the possibility of applying the procedure of swapping the injection and measurement point using the SFRA principle. If you imagine the rotor winding’s terminals, a frequency response with the injection and the reference signal in terminal “A” and the response measurement in terminal “B” are recorded first. Then, the frequency response in terminal “A” is measured, and the injection and reference measurement occur in terminal “B”. The thesis shows these are practically identical for an intact winding, especially in comparatively low frequencies up to 600 kHz. If there are interturn short circuits, you can detect clear deviations. This enables the advantages of the SFRA on rotor windings without need of a fingerprint measurement. How large are these differences? Pascal Fröhlich: They are not particularly large. However, the smallest differences can still be identified because the curves are identical when the winding is intact. An increased intensity of the error also indicates an increased degree of deviation of the two characteristic lines from one another. At this point, however, a machine-specific distinction must be made since differences in the frequency response do not always occur at the same point or with the same intensity. How can I imagine the effect on the measurement result of a possible error in the winding? Pascal Fröhlich: On the one hand, these are identifiable to the eye by all test engineers, but then we run the risk of ending up in subjectivity again, as mentioned at the beginning. 38
RkJQdWJsaXNoZXIy NTkxNzY=