Dispersion and PD detection in shielded power cable [Feature article]

Dispersion, the variation in propagation velocity with frequency, is one of those phenomena that people in the field of partial discharge (PD) detection have talked about since at least the early 1980s, but which is lacking a formal treatment in the literature. When one of the authors recently published an analytic theory of PD propagation in shielded power cable, both reviewers said "dispersion, dispersion". They believed that dispersion was important but gave no evidence for that assertion. In fact, as shown below, dispersion in shielded power cable has little if any impact on measurement of PD magnitude because the Fourier components of the pulse attenuate to insignificance before they can disperse to a degree that would cause appreciable change in the peak pulse amplitude or integral of the pulse waveform. However, dispersion does cause some distortion of the pulse shape that has implications for PD location, as it has a second-order effect on the timing of the peak PD amplitude relative to other pulses in the pulse train caused by multiple reflections from the ends of the cable. A Gaussian PD pulse in the time domain will have a Fourier spectrum that is also Gaussian in the frequency domain. If the Fourier components in the frequency domain propagate down the cable at differing velocities, the waveform to which they add will vary as a function of distance propagated, and the energy in the PD pulse is likely to spread out in time, which would have an effect on wide band PD detection no matter what the means of detection.