Excitation voltage options for PD investigations on MV cables
Author: Robert Probst, Product Manager
Since the 1980s, Very Low Frequency (VLF) testing of cables, in which the tested cable must withstand an over rated AC voltage for a specified testing time without flashover, has been accepted as a way of significantly improving system reliability by reducing the number and duration of outages. VLF testing was the key technology that allowed utility companies to move away from the widespread reactive maintenance culture, often described as “run to failure”, and instead adopt a proactive preventive maintenance approach. However, VLF testing is a pass/fail method only, and when a cable network is VLF tested regularly, predictive maintenance strategies and condition-based asset management are required to provide further insights into the condition of the cable network.
Today, especially outside North America, diagnostic testing on both medium voltage (MV) distribution and high voltage (HV) transmission cables is well established in most utility companies. Out of a variety of available diagnostic techniques, two have become dominant in the industry: dissipation factor (tan delta) measurements and partial discharge (PD) analysis.
During the life cycle of a cable, these two methods are complementary. Both have advantages and disadvantages depending on whether the user’s aim is identifying and localising defects, evaluating the ageing condition of the main insulation, or condition monitoring. Tan delta testing is a global, integral measurement of dielectric losses, and it implicitly determines the average water tree content in a polymer without having to rely on actual isothermal relaxation current (IRC) measurements. It is unable, however, to identify isolated defects and weak spots. Tan delta diagnostics are, therefore, best suited to evaluating service-aged cables before proceeding with further testing, such as VLF withstand testing.
In contrast, PD analysis is a powerful tool in any phase of the cable’s life cycle. It cannot provide a global assessment of the main insulation, nor can it indicate average water tree content, but it can reveal and pinpoint local defects and weak spots of a high resistance nature, such as voids. Also, surface discharges and internal discharges occur most commonly in layered interfaces (splices and terminations), which means that PD testing is the optimal method for assessing ‘accessories’, that is, the mechanically and electrically altered parts of the cable assembly, such as terminations, joints, and splices. Furthermore, PD analysis is particularly useful for acceptance and commissioning testing. It makes it possible to directly assess the quality of workmanship by revealing defects that would pass a standards-compliant VLF withstand test, but then slowly deteriorate and eventually cause the cable to fail a relatively short time after the circuit was energised.
Key parameters for PD analysis
Partial discharge inception voltage
The Partial Discharge Inception Voltage (PDIV) is the voltage at which PD events start to occur. Many scientific publications have shown that PDIV (the initiation of electrical trees) is a function of test voltage frequency or, in other words, the rate of change of voltage (dU/dt) in the region of the zero crossing during polarity reversal. The inception voltage for surface discharges and internal discharges in layered interfaces, like splices and terminations, is particularly sensitive to the voltage gradient.
Partial discharge extinction voltage
The Partial Discharge Extinction Voltage (PDEV) is the voltage at which an existing PD event or electrical tree extinguishes. There is no known direct correlation between PDIV and PDEV, but the parameters follow a hysteresis pattern. Empirical observations and practical experience indicate that PDEV levels are typically between 0.8 and 0.9 times PDIV and sometimes as low as 0.7 times. It has been found that PD activity extinguishes with very slow dU/dt, which makes it difficult to conduct PD measurements using low frequencies like 0.1 Hz. Only a damped oscillating voltage source (DAC) can properly measure PDIV and PDEV in the same test cycle.
Partial discharge intensity, mapping and localisation
State-of-the-art PD equipment uses software to visualise PD data in a map-like representation of the cable under test. The y-axis indicates the measured and weighted apparent charge, scaled in picocoulomb (pC). The x-axis indicates the physical length of the cable, scaled in meters or feet. In accordance with IEC 60270, the criticality of a weak spot is decided by considering PD intensity at the operating voltage U0. PD mapping visualises:
- PD rate (the number of PD events)
- The intensity level of each PD event
- The location of each PD event
The location of the PD events is determined using Time Domain Reflectometry (TDR) and analysis of the runtime differences between the initial pulse and the reflected pulse.
Partial discharge patterns
Plotting each detected PD event according to its phase angle with respect to the excitation voltage generates Phase-Resolved Partial Discharge (PRPD) patterns. Different PD conditions can be associated with different PRPD patterns, so that PRPD patterns can provide valuable information about the type of weak spot.
Excitation voltages for offline PD testing in the field
With offline PD testing, a separate power source is connected to the cable. This allows the PD test to be performed at any desired voltage which, incidentally, is an advantage over online PD measurements. By varying the voltage from about 0.5 times rated voltage to 1.7 times rated voltage for aged cables, or up to 2 times rated voltage for new cables, it is possible to find out at what voltage PD begins. If PD events don’t occur until well above operating voltage (say 1.5 or 1.7 times operating voltage), the cable should be monitored, but it is not necessarily an immediate concern. However, if PD events begin much closer to operating voltage (say 1.1 or 1.2 times operating voltage), then that cable is of greater concern and should be monitored more frequently. Cables with a PDIV below operating voltage would be of immediate concern and repaired or replaced as necessary. Conversely, cables that do not exhibit PD up to maximum voltage can have a much longer testing interval.
Ideally, cables would be PD tested at 50/60 Hz to simulate what occurs in service. However, to do this as an offline test would require very large power sources. To lower the power requirements, a VLF source is used instead. In cable testing, the term VLF may be referring to a VLF sinusoidal, VLF cosine rectangular (CR), or damped AC (DAC) waveform. VLF sinusoidal and VLF cosine rectangular are continuous waves, while damped AC consists of discreet pulses that may have significant time between pulses. As excitation voltages for a PD test, these VLF sources present noteworthy differences.
0.1 Hz cosine rectangular VLF
The 0.1 Hz Cosine Rectangular (CR) VLF technology was invented and patented in the early 1980s by HDW. It was the first VLF technology introduced to the market and is still the most common of the two standard VLF methods that are used all over the world. The CR waveform, an 0.1 Hz example of which is shown in Figure 1, has two components:
- The dwell stage with a DC-like voltage plateau. This steady state allows the true leakage current to be measured, something which cannot be done with sinusoidal VLF technology.
- The transition (or swing over) stage where fast polarity reversal takes place with the same slope as a 50/60 Hz power frequency sinewave.
CR test sets use an LC resonant circuit with periodic energy transfer between the capacitance of the cable under test and the test set’s internal inductive reactor. Because of the inherent recovery of energy at each half cycle, CR test sets intrinsically have a very high test capacity (the ability to charge/discharge the capacitance of the cable) and a very low power consumption, making them considerably more practical for field testing than sinusoidal test sets.
The industry’s most powerful VLF test sets (at the moment 25 μF, 60 kV RMS) can only be produced by using CR technology. They operate at continuous power levels that are not feasible using sinusoidal VLF technology, as can readily be seen from these calculations.
The apparent power, in VA, needed to charge a capacitor (in this case the capacitance of the cable under test) is given by S = 2π.f.C.U2, where ‘f’ is the frequency of the applied test voltage in Hz, C is the capacitance to be charged in μF and U is the applied test voltage in kV. With a test frequency of 0.1 Hz, the apparent power needed for a capacity of 25 μF at 60 kV RMS is therefore: 2π x 0.1 x 25 x 3600 = 56 520 VA ≈ 56 kVA
In reality, the HV sources of sinusoidal VLF tests always need to supply more apparent power than the required charging power, for example 115 to 130 %, so, in our example, they would have to deliver 64 kVA at a minimum. It is impossible to build a practical test set with this capability. As explained earlier, however, test sets that use CR technology recover most of the power when the polarity reverses – approximately 86 % of it. So, a CR test set to test the same cable would need to deliver only 56 x 0.14 = 8 kVA, which is a much more practical proposition (e.g. 3 x 208 V or 3 x 380 V).
Another characteristic of CR technology is that polarity reversal of the CR waveform happens at a point on the waveform that has a slope comparable with the slope of the 50/60 Hz power frequency voltage. As shown later in this article, the rate of change of voltage in the region of the zero crossing (dU/dt) is a crucial factor in obtaining valid, reliable and repeatable PD results.
CR technology is also a key building block for creating a DAC waveform. The components that make up DAC systems are the same as those in CR systems, and it needs only a minor change to the discharge switch for a CR test set to operate in DAC mode. Because DAC and 0.1 Hz CR VLF technologies are so closely related, combination test sets were introduced to the market many years ago and have proved to be both popular and successful.
Since 0.1 Hz CR VLF produces PD results comparable to DAC and 50/60 Hz power frequency, it is the only cable test technology that can simultaneously provide the benefits of a VLF withstand test and those of a PD diagnostic test. This combination test is designated “monitored withstand” and it offers a comprehensive diagnostic solution that is particularly useful when commissioning new cable installations.
Damped AC (DAC)
Like 0.1 Hz CR VLF, DAC is a recognised waveshape in all relevant international standards such as IEC 60270 and Parts 3 and 4 of the IEEE400 guide. DAC testing, which has been in worldwide use for over ten years, has become the proven standard method for carrying out non-destructive PD field diagnostics on MV and HV cables, since it exposes the main insulation of the cable under test to the minimum possible stress (the voltage exposure time is only around 10 cycles). It uses essentially the same hardware as 0.1 Hz CR VLF test sets, but with modifications to the internal switches.
Figure 2: Typical DAC excitation voltage with PD activity
Figure 2 shows a PD measurement with DAC. The method of creating a damped oscillating voltage is resonance: a DC source charges the cable and the discharge switch closes and remains closed, thereby creating a series resonant circuit. The resonant frequency for the energy transfer between a DAC test set and the cable under test is a function of the inductance of the inductive reactor in the test set, the capacitance of the auxiliary capacitor in the test set, and the capacitance of the cable under test. It is typically in the range 30 to 500 Hz. PD results obtained with DAC have been proved to show very good correlation with results at 50/60 Hz power frequency, and thus are fully comparable. DAC is the only excitation technology capable of properly measuring PDIV and PDEV in the same test cycle.
C. 0.1 Hz sinusoidal VLF
The 0.1 Hz Sine VLF technology was introduced to the market in the mid 1990s. This development was driven by advancements in power electronics and by a growing customer interest in the use of tan delta diagnostics on service-aged cables. This requires a sinusoidal waveshape; tan delta cannot be measured with 0.1 Hz CR VLF. 0.1 Hz Sine VLF test sets are based on an HV AC-DC-AC converter line-up that uses IGBT modules. In contrast to 0.1 Hz CR VLF, the energy to charge the cable cannot be recovered and must be dissipated as heat in a resistor within a quarter cycle time window before each polarity reversal. The power consumption of 0.1 Hz Sine VLF test sets is typically 1.15 - 1.3 times the power necessary to charge the cable under test and the test capacity is, for the most part, 5 to 10 times less than that of comparable 0.1 Hz CR VLF test sets.
With a 0.1 Hz Sine VLF source, the voltage gradient in the region of the zero crossing is approximately 1000 times less than the equivalent gradient for a 50/60 Hz power frequency voltage. Empirical studies on realworld cable systems, as well as scientific research, have shown that a 0.1 Hz sinusoidal testing generally does not provide results comparable to 50/60 Hz power frequency. PDIV – a very important parameter in assessing the criticality and severity of PD activity – is often evaluated incorrectly, typically deviating by between 10 % and 60 % from the actual value and, in extreme cases of surface discharge, by up to 300 %. This has the effect of masking defects and weak spots near the operating voltage, U0. PD intensity and PD mapping may also differ from 50/60 Hz power frequency PD data, which means that condition assessments may be deceptively optimistic because of the apparently low or even non-existent PD activity at U0 that is falsely attributed by this test method to splices and terminations.
This case study provides an example of the differences in PD measurements that may be expected when using each of the three excitation voltage waveforms. Testing was carried out by ENSO (Energie Sachsen Ost), a German utility in the federal state of Saxony, which services approximately 500 000 customers and has revenues of around €1.1 billion per year. The cable under test was a 1335 m (4380 ft) long, service-aged, mixed 12/20 kV cable with 10 splices. It was predominantly composed of XLPE sections (NA2X(F)2Y 3 x 1 x 150), but also had two fairly old sections of paper-mass-impregnated, non-draining, oil-filled cable (NAHKBA 3 x 120). There were, therefore, four transition splices. Figure 3 shows an overview of the cable.
Figure 3: ‘In-software’ representation of the cable under test: 1335 m long, 10 splices, 11 sections, green = XLPE, blue = paper-mass, black = splices
The cable was tested with a fully integrated system installed in the test van that can be seen in Figure 4. The PD-free HV connection cable from the test van ended in SF6-insulated compact switchgear, so that an adapter was used to achieve a PD-free connection to the cable under test. Figure 5 shows the PD data obtained using three excitation voltages (0.1 Hz Sine VLF, DAC, 0.1 Hz CR VLF). The PD mappings in the left column were observed at U0, and those in the right column at the maximum test voltage of 1.7 U0. It can clearly be seen that the results of the 0.1 Hz CR VLF and DAC investigations are comparable. This was expected, as the frequency corresponding to the polarity reversal of the 0.1 Hz CR VLF excitation voltage and the frequency of the DAC excitation voltage are exactly the same (280 Hz, which is within the power frequency band). In addition, PD level and PD intensity only showed small differences between U0 and 1.7 U0. PDIV was exactly the same (5.8 kV) for all three excitation voltages.
Some major differences are, however, apparent between the 0.1 Hz Sine VLF and the DAC/0.1 Hz CR VLF results. At operating voltage, only one weak spot was identified with 0.1 Hz Sine VLF, whereas two weak spots were found with the other excitation voltages. Also, PD intensity was far lower with 0.1 Hz Sine VLF excitation compared with the other wave shapes. Only at elevated voltages was 0.1 Hz Sine VLF able to detect the second weak spot in the cable. The PD rate in the first weak spot hardly increased with higher test voltages for 0.1 Hz Sine VLF, while with 0.1 Hz CR VLF and DAC there was a noticeable rise with increasing voltage. PD intensity level measured with 0.1 Hz Sine VLF was also significantly lower compared to DAC and 0.1 Hz CR VLF, giving the impression of a non-critical situation.
Figure 4: Connection of PD diagnostics system to the cable under test
The conclusion that can be drawn from this case study is that PD tests carried out with 0.1 Hz CR VLF and DAC excitation provide comparable results and are very effective in revealing faults that may, without remedial action, lead to cable failures. The results obtained with 0.1 Hz Sine VLF excitation are significantly different from the results obtained with the other two forms of excitation and are much less effective in revealing weaknesses in the cable.
Summary and conclusion
As we have seen, there are significant performance limitations when performing PD testing with 0.1 Hz Sine VLF excitation and these limitations are not seen with 0.1 Hz CR VLF and DAC excitation. While DAC is the well-established standard excitation voltage for nondestructive, minimally intrusive PD field diagnostics on MV distribution and HV transmission cables, 0.1 Hz CR VLF is the more common and more effective method of VLF withstand testing on MV cables, and it can be used to perform very useful monitored withstand tests during the acceptance and commissioning of new cable installations.
Primarily because it allows reasonably sized equipment to be used for tan delta testing, which cannot be performed with any other type of excitation, 0.1 Hz Sine VLF testing has also become an established technology. Sinusoidal VLF test sets are, however, susceptible to producing incorrect measurements of PDIV and frequently produce representations of PD rate, PD intensity and PD mapping that are not comparable to those obtained by DAC testing or at 50/60 Hz power frequency. This is due to the very slow rate of change of voltage (dU/dt) in the region of the zero crossing. In addition, the monitored withstand tests provided by sinusoidal test sets (0.1 Hz sine VLF + tan delta + PD) are blind to low resistance faults because leakage current cannot be measured. These limitations pose a significant problem for correct condition assessment and monitoring of cables and cable accessories, and therefore for precise and confident decision making.
Figure 5: 1 ENSO – PD mapping comparing three different waveshapes at operating voltage U0 and 1.7 U0. Top: 0.1 Hz Sine VLF; Center: DAC; Bottom: 0.1 Hz CR VLF. X-axes: cable length in meter; Y-axes: PD level in picocoulomb