Dielektrisk diagnostik vid godtycklig vågform

Startdatum 2010-09-01
Slutdatum Projektet är avslutat: 2015-08-31

​Dielectric response measurements is one of the two major techniques for high voltage diagnostics, the other being partial discharge measurements. Dielectric response measurements are used both for basic material studies, to develop and select construction materials, and for apparatus diagnostics. Examples of the latter which have come to a rather wide-spread use are cable and transformer diagnostics, where primarily the extent of treeing and the humidity is of concern, respectively.


With the work of Björn Sonerud, Arbitrary Waveform Impedance Spectroscopy, AWIS, has been developed to a versatile tool for insulation dielectric diagnostics [1]. The primary advantage with AWIS as compared to conventional methods is that a well-defined sinusoidal waveform with precisely known frequency is no longer required to perform high-precision measurements. The lowered demand on controllability of the voltage signal makes AWIS a candidate for field testing and online monitoring of dielectric properties of high voltage components under service voltage.
As AWIS is not developed to become a commercial instrument, rather a measurement technique, a large freedom and responsibility is entrusted to the operator to design the best possible conditions for a specific task. This has enabled the technique to be applied to both multi-μF capacitors and sub-pF treeing samples using similar circuits, albeit with different parameters. Further, when partial discharges are contributing significantly to the current through the test object not only may the additional current caused by the discharges be detected; the charge relaxation current caused by decay of deposited charges can be measurable and compared to the dielectric and the discharge current. This can provide new possibilities for partial discharge studies as all conduction processes, be they fast or slow, are measurable in the same quantity.


Dielectric response measurement techniques under partial discharge activity were also explored in the PhD work of Nathaniel Taylor [2]. He found that the current estimated from PD activity and the loss current as measured with dielectric technique could be similar for some types of PD sources. In other sources, a considerable loss current was measured in half-periods when no PD activity could be detected. For stator bars, the detected PD activity could only account for a small portion of the measured loss current, even though the phase distribution was quite similar. This line of investigations was not pursued further in the thesis; a possible reason is that the work was aimed at low-frequency AC excitation. This is probably not the best waveform for such studies as he discusses in the future work section of the thesis. Semi-square waveforms, such as the ones employed by Björn Sonerud, are most probably a better alternative. With such waveforms, PD are namely occurring mainly when the voltage changes which makes it possible to study the charge relaxation current at phase positions when no discharges occur.


There is a need for a deeper understanding of the factors that influence PD activity in insulation materials. The decay of charges deposited from PDs is clearly one such factor which furthermore may be controlled by additives in the material. Changing the rate by which charges from previous PDs decay can have a considerable influence on the number and magnitude of PDs and thus on the lifetime of the insulation. Presently, effects on insulation lifetime are only observable in time to breakdown tests. While such test will be the finally required proof for the foreseeable time, it would be time effective to have additional observables of the ageing process during material development.


PD occurrence is a natural first candidate for an observable. This has however proven not to be interpreted very easily. Morshuis [3] and, perhaps most recently, Wang [4], have exemplified that both PD number and amplitude may decrease to very low values shortly after an ageing test is initiated. Only after a substantial part of the lifetime has been consumed, the PD parameters return to values that are similar the initial ones. Here, observables based on dielectric response techniques could yield more information as it is possible to observe the total PD current per period even though the individual PD events are not detectable, which is required by most PD pulse detection methods. To our knowledge, no investigations of this kind have ever been performed.


With this background, it seems highly motivated to suggest further work to establish AWIS as a diagnostic tool, both for material and apparatus diagnostics. In particular, investigations of the information that may be gained from AWIS measurements under partial discharge activity in various insulation systems are motivated as they may unravel new understanding and thereby also produce an important scientific contribution. This technique has further the potential of being developed to an alternative to traditional time resolved partial discharge diagnostic and monitoring techniques.


Application of dielectric measurement techniques under PD activity is planned to receive the main attention during the work towards the doctoral degree, as no suggestions for industrial testing have been proposed. Such proposals would however be welcomed and receive positive attention that may steer the project into another direction for some time.


The discharge studies will be limited to laboratory measurements as the capacitive current in most high voltage equipment is so large that PD current contribution often is undetectable, even with AWIS. Thus, the value of these investigations is mainly to support insulation material development. A few such projects are presently pursued at Chalmers and KTH and the suggested studies may have an important impact on the direction of these.
The first step in these studies is to establish the quantities to be studied and techniques for measurement and extraction of these. As this is an almost virgin field, in contrast to the licentiate work, there is not much experience to build on. From Björn Soneruds work is seems however that fast and slow processes may be studied simultaneously with a semi-square waveform. This may be a way to study charge decay during PD activity. The charge decay speed is thought to be an important parameter that control PD activity.


Throughout these studies, a number of test object types should be used in order to avoid focusing the method on possible peculiarities of one type. The studied types should include but not be limited to: corona in air, twisted pairs, cavities and surface discharges.


It is foreseen that the implications of some of the obtained results are not directly obvious. There will thus be a need for some theoretical studies to unveil the full potential of the technique.
With reasonably well established quantities, a number of novel studies can be performed, ranging from generic studies of particular PD sources to ranking of material modifications. If this work is successful, material developers will have been provided with a new and important tool.

References
[1] Björn Sonerud, “Characterization of electrical insulation exposed to arbitrary voltage waveforms”, PhD thesis, Chalmers, 2010. ISBN 978-91-7385-373-6
[2] Nathaniel Taylor, “Dielectric response and partial discharge measurements on stator insulation at varied low frequency”, PhD thesis, Royal Institute of Technology, 2010.TRITA-EE 2010:037, ISBN 978-91-7415-713-0, Chapter 17.
[3] P.H.F. Morshuis, “Partial discharge mechanisms”, PhD thesis, Delft University of Technology, 1993, ISBN 90-6275-931-9
[4] Wang Le, “Physical model of PD behaviour and relevant damage growth from micro-cavities in polyethylene-based material under AC voltage”, PhD thesis, Università di Bologna, 2011

​ELEKTRA project Nr. 36085
​Swedish Energy Agency, Elforsk and ABB
Denna sidan finns endast på svenska

Sidansvarig Publicerad: to 10 apr 2014.