Piotr Cichecki - Energy market specialist
Failures on MV PILC cables, many of which have been in service for forty or more years, are a significant headache for DNOs. Drawing on material from a long-term study carried out jointly by Warsaw University of Technology and Megger, this article looks at how partial diagnostic analysis can help to address this important issue.
Before discussing the role of partial discharge (PD) analysis in assessing the condition of MV cables, it’s useful to review exactly what PD is. The concept is, in fact, very simple: partial discharge is a small electric spark or discharge that occurs at an insulation defect, but does not completely bridge the insulation. The defect may take the form of a cavity within the insulation, it may be along the interface between insulating materials (typically within accessories) or it may be along surfaces, for example in terminations or potheads.
The characteristics of the PD depend on many factors, including the type, size and location of the defect, the type of insulating material, the applied voltage and the temperature. The characteristics also vary with time. The damage caused by PD can range from negligible to severe, the latter causing complete insulation failure over a time period anywhere between a few days and a few years. A common example of PD is corona on overhead HV transmission lines in damp weather, which is the source of the buzzing noise often heard under HV transmission towers.
PD within solid insulation typically occurs when a spark jumps a gas-filled void in the insulation, producing a small current within the conductors. By using time-domain reflectometry (TDR) techniques and taking into account the cable characteristics that affect pulse propagation, a PD analysis test set can determine the location of the discharges and produce a PD map that plots discharge intensity against location.
Two major sources of PD problems in PILC cables are dry insulation, typically as a result of high thermal loading or bad impregnation, and water ingress due to deterioration of the lead sheath. Both issues lead to the formation of carbonised tracks in the paper insulation, and these can be present over long lengths of cable – examples covering a few metres or more are by no means uncommon. PD problems are also found in the accessories used in PILC cable systems, often resulting from sharp edges being left on the conductor connector.
While on-line PD analysis of MV power cables is possible, this approach inevitably has limitations and is necessarily a far less sensitive way of detecting defects than off-line testing. The remainder of this article will, therefore, focus on off-line PD analysis, for which several techniques are available. The most important of these are:
• Testing at 50/60 Hz power frequency
Although this could be considered to be the most desirable and revealing technique, as it accurately analyses the behaviour of the cable under conditions close to those it will experience in service, it is usually not practical because of the enormous size and cost of the power source needed to energise all but the shortest of cables.
• Testing with damped AC (DAC) voltage
This is the most widely used method worldwide, since it uses easily transportable and affordable test equipment, and it works at frequencies comparable to power frequency.
• Testing at VLF (0.1 Hz) with a sinusoidal test voltage
This technique is also widely used, but the PD results obtained with sine waves at this low frequency may be very different from those that would be obtained at power frequency. The test may not, therefore, accurately reflect the in-service behaviour of the cable.
• Testing at VLF (0.1 Hz) with a cosine-rectangular (CR) test voltage
This is a relatively new technique. The key feature is that the 0.1 Hz CR waveform used has rise and fall times closely comparable with the rise and fall times of a 50/60 Hz sine wave. PD measurements made during these transitions are, therefore, close to the results that would be obtained at power frequency.
Table 1 – Test methods for MV cables
Table 1 summarises the test techniques that are available for MV cables. It is worth commenting that, based on a Megger data bank holding test results reported by a wide range of cable operators since 2005, 88% of MV cables passed withstand/hi-pot test with 0.1 Hz VLF voltage sources, where these tests were not monitored for partial discharge. However, around 5% of these cables failed within six months of apparently successful commissioning, with the most common problem being poor workmanship in relation to cable accessories such as joints and terminations.
Returning to the use of DAC voltage for PD testing, the studies carried out by Megger in conjunction with Warsaw University of Technology have shown that this is an excellent method for reliably identifying partial discharges using a non-destructive test voltage. Another important benefit is that this is a well-proven method that yields results that can be directly compared with the cable’s performance at 50/60 Hz, allowing reliable decisions to be made about whether it needs attention.
With respect to the VLF CR waveform, a major benefit is that PD testing can be carried out concurrently with a standard VLF withstand test in line with IEC/IEEE 400.2/3. This saves time because, if no PD is detected during the test, no separate PD analysis is needed. Direct indication of the quality of workmanship on the cable is also provided during the withstand test. Once again, the PD data is comparable with the cable’s performance at 50/60 Hz, and so supports reliable decision-making. Finally, VLF testing at 0.1 Hz satisfies the requirements of IEC 60502 – 2.
It is important to note that the DAC and VLF CR techniques for PD analysis both have specific advantages. Fortunately, purchasers of test equipment no longer have to decide between them as the latest cable test sets, such as the Megger TDS NT, support both techniques, allowing users to choose the method most suited to the cable being tested.
That both techniques are equally good at detecting faults and revealing their locations is illustrated by Figure 1, which shows the results of tests carried out on one phase of a 240 m long PILC cable in Warsaw, using both DAC and VLF CR techniques. As the figure clearly illustrates, there is little difference between the results obtained.
Figure 1: Cable investigated with two voltage sources
In reality, results are not always as clear-cut as those obtained on the Warsaw cable, as can be seen in Figure 2. However, advanced filtering and signal processing techniques can make even results like these much easier to interpret. Figure 3, which shows filtering applied to the same data used in Figure 2, shows just how effective it can be.
Figure 2: PD results without filtering
Figure 3: PD results with filtering
We have seen that PD characteristics such as PD magnitude, location and occurrence are similar whether the test voltage source is DAC or VLF CR. We have also seen that the latest test systems, which offer both of these voltage sources and incorporate powerful and effective filtering, make fast and reliable PD analysis a reality. This means that DNOs operating networks that include old PILC cables now have a convenient and cost effective way of pre-empting cable faults and, therefore, enjoying greatly enhanced peace of mind!