Introduction: National Grid's technical specifications require MI cable to be tested to the internationally accepted CIGRE test procedures. As the operating voltages of DC cables increase cable manufacturers are progressively taking the view that the CIGRE test voltages are too severe and unless the test voltage is reduced (particularly during the cooling phase of heat cycling) there is an unacceptably high risk of the cable failing the type test.

In order to achieve type registration of these cables it will be necessary for National Grid to consider relaxing the test voltage. There is no published basis to justify this reduction and it is difficult to assess the risk of accepting cable systems which cannot meet the CIGRE requirements.

There is a possible mitigation strategy based on applying condition monitoring techniques during type testing so that the test is not reliant on simple withstand criteria. When a MI HVDC cable fails the heat cycle type test it is likely to be the result of accumulated Partila Discharge (PD) damage. Hence PD monitoring appears to be the most appropriate option to investigate.

PD detection in DC systems is significantly more difficult than in AC systems because (i) the discharge repetition rate if far lower and (ii) there is no alternating voltage to which the discharge activity is synchronised. It is therefore difficult to distinguish between PD activity and random background noise.

Recent work at Southampton on PD from AC cable systems indicates that clustering algorithms can be used to distinguish between PD from different sources. It appears feasible to use this technique during DC testing to distinguish between PD from the cable and that from the terminations or external noise sources. The technique relies on analysing the PD signals to measuring the energy content in a number of time and frequency windows. The multidimensional results are converted to a pseudo 3-dimensional data set for easier visualisation and automatic classification.

In addition to developing a procedure to detect and classify DC PD signals the work will emphasise the need for the technique to be suitable for implementation during DC cable type tests. This requires that PD testing can be done safely in an industrial laboratory without impacting on the smooth running of the type test.

Objectives: To investigate and develop a method for monitoring partial discharge (PD) activity in mass impregnated (MI) HVDC cable. The outputs will enhance National Grid's understanding of high power HVDC cable and facilitate the development of improved Technical Specifications. The test method developed should be sufficiently effective and efficient to allow its deployment within the constraints of a commercial Type Test programme.