Andrew Barclay principal engineer, Kinectrics
This article is based on a paper which Andrew Barclay, Principal Engineer at HV cable test specialist Kinectrics, presented at a recent Megger cable test seminar.
Insulation tests carried out as part of the commissioning procedure for cables have a simple goal: to maximise the reliability of the newly installed cable circuits by detecting defects before the circuits go into service. Commonly used test techniques seek to achieve this goal by converting latent defects into faults during the test and by detecting the latent defects as they are being converted.
Typically, the conversion of defects into faults start with the initiation of a discharge at the fault location by putting the cable under electrical stress. This leads to electrical tree growth in the cable dielectric – which can be detected and monitored using partial discharge (PD) techniques – and ultimately to breakdown of the cable insulation.
While this procedure is straightforward in principle, selecting the optimum test voltage presents a significant challenge. The voltage must be high enough to overstress defects so that they age to breakdown during the test period and also to ensure that the defects suffer detectable partial discharge. The voltage must, however, not be so high that it significantly degrades healthy circuits – it is essential to convert only those defects that, if left undetected, would limit the life of the circuit. Selecting the most suitable source of test voltage is also challenging. Options include DC, AC on-line at system voltage and frequency, AC off-line from a resonant test set, AC very low frequency (VLF), damped AC (oscillating wave), and impulse voltage. As we shall see, each of these options has its own merits and shortcomings.
Insulation testing with DC applied voltage was once widely used, particularly with paper insulated cables. The equipment needed is relatively small, inexpensive and uncomplicated, but problems were experienced with early XLPE cables that were prone to failing in service soon after testing. For this reason, many end users no longer allow DC testing on XLPE cables, even at low voltages. In any case, because DC testing does not involve polarity reversal, it is less effective than AC methods at detecting faults on XLPE cables.
Some specifications allow commissioning tests to be performed simply by energising the cable at system voltage – in other words, by carrying out an on-line soak test. The major advantage of this approach is that it is quick and inexpensive, as no test equipment is needed, and, of course, it is necessarily carried out at power frequency. There are, however, substantial drawbacks.
If a failure does occur during on-line testing, it will almost certainly be destructive. This not only means that valuable evidence about the nature of the problem will be destroyed, but also that there may be significant hazards for personnel and equipment. In addition, as the cable is tested at its normal working voltage rather than with overvoltage, there is no accelerated ageing of defects. On the plus side, however, on-line testing can be combined with PD investigations.
Off-line AC testing of HV and EHV cables at or near power frequency is stipulated in IEC 60840, IEC 62067 and many other cable testing standards. The recommend voltage test levels range from 1.1U0 to 2U0, but a typical recommendation is testing should be carried out at 1.7U0 for 60 minutes. This method of testing has, on the face of it, much to recommend it.
The elevated voltage usually used provides accelerated ageing of defects and, because the test is performed at or near power frequency, the stress distribution and ageing mechanism are directly related to those the cable system will experience when it is in service. This test method, which typically uses a resonant circuit to energise the cable, is well established and well proven.
The major drawback of off-line AC resonant testing power frequency is that the test equipment is unavoidably large, heavy and expensive because of the amount of reactive power it must supply to continuously energise the cable. The equipment needed to test even a moderately long cable using this method is likely to need an articulated low loader to transport it!
Off-line testing at very low frequency (VLF) – typically either 0.1 Hz or 0.01 Hz – addresses the problem of bulky test equipment. The reactive power needed to charge the cable reduces in direct proportion to frequency, which means that VLF test sets are relatively small and lightweight as well as being much less costly than power frequency test sets. Lower test frequencies do, however, reduce the ageing of defects and tests carried out at 0.01 Hz in particular can take a very long time to yield useful results.
VLF testing at 0.1 Hz is widely used for cable systems operating at voltages from 11 kV to 33 kV and good results have also been obtained from tests on cables operating at higher voltages. VLF testing is especially useful for service-aged cables, and it can be combined with both partial discharge and tan d testing. Another approach to reducing the size and cost of the equipment needed to carry out AC tests on cables is to use oscillating wave – also known as damped AC – testing techniques.
In principle, this involves connecting an inductor in series with the cable under test and then charging the cable from a high voltage DC source. When the cable is sufficiently charged, a high-speed solid-state switch effectively connects the inductor in parallel with the capacitance of the cable to form a resonant circuit. With the correct size of inductor, the result is that damped oscillations are set up in the cable at approximately power frequency, and it is these oscillations that provide the test voltage.
The main benefits of oscillating wave testing are that elevated test voltages can be used, the test frequency approximates to power frequency, and that the equipment is much smaller and lighter than that needed for power frequency AC testing using continuous energisation. Oscillating wave testing can also be usefully combined with partial discharge analysis. On the downside, an oscillating wave tests applies only a very few cycles of test voltage to the cable, so the ageing of defects is minimal.
The relative merits of the competing technologies for supplying the voltage needed to carry out commissioning tests on HV and EHV cables are summarised in Table 1. As the table shows, all of the methods, with the sole exception of DC testing, can be readily complemented by partial discharge (PD) testing. This is very significant, as PD testing provides an opportunity to detect signs of distress from defects before failures occur.
PD testing can be performed with cable on-line or off-line but, in commissioning scenarios, it is most often performed during a high-voltage off-line insulation test. For the best results, PD testing should be carried out at or near power frequency and at elevated voltages to accelerate the ageing of defects.
Most PD monitors need expert handling to get the best from them and considerable experience is also required to interpret the data they produce. It is also worth noting that there are no established standards for PD testing in the field. A good case can, therefore, be made for having the tests carried out by a third-party test service provider with proven experience.
For off-line PD testing, cables must be disconnected and the conductors exposed. In cross-bonded cable systems, sheath voltage limiters (SVLs) must be shorted and, for preference, the cross links should be removed to create a continuous earth sheath. An external power source, as discussed earlier, is needed and the test should be carried out at elevated voltage. With off-line testing it is possible to measure both PDIV (partial discharge inception voltage) and PDEV (partial discharge extinction voltage).
In contrast, on-line PD testing is a non-invasive test carried out system voltage. PDIV and PDEV cannot, however, be measured and it is, of course, only possible to detect those PD sources that are present at the time of testing! In addition, the interpretation of on-line PD test data can be complicated as multiple PD sources from various types of HV equipment are often detected.
Notwithstanding these limitations, on-line PD testing does have one important benefit – it can be performed continuously on in service cable systems over a long period of time, thus making it possible to characterise the dynamic PD behaviour of the system.
As we have seen, a wide variety of techniques are available for testing HV and EHV cable systems. Kinectrics has almost a decade of experience in applying these techniques and, during that time, the company has tested more than 1,715 km of cable operating at 66 kV and up. The tests have covered 307 individual phases, 614 terminations and 2,654 joints and it is in interesting to note that failure rates have remained substantially constant over the decade. The results of tests performed since 2005 are summarised in Table 2.
The testing of HV and EHV cables is, without doubt, a specialist field where expert advice on selecting the most appropriate test method for a particular application, taking into account budgetary and operational constraints, is all but essential, as is the use of dependable test equipment with proven performance. When the right advice is combined with the right test equipment, however, testing can dramatically reduce the risk of power outages and their frequently astronomical consequential costs.