B1 - New measurement technique and use cases in the inspection of partial discharges of circuits with insulated cable in the Spanish TSO
Authors
Ricardo GÓMEZ, Ricardo REINOSO, Gonzalo DONOSO, Elena NOGUEROLES - REE, Spain
Summary
This article aims to show the advances in a new technique of partial discharge measurement that the Spanish TSO has been focusing on in recent years.
Partial discharge measurement (PDM) is a reliable tool for determining the state of insulation in high-voltage equipment, especially in insulated cable systems, as well as for preventing faults in these installations.
Although PDM can offer very valuable information on the state of the insulation, the management and analysis of the data from the measurement equipment is not easy. It is necessary to filter the signal and a correct interpretation of the partial discharges (PD) parameters to identify if there is PD in the installation and, if so, if the PD corresponds to corona, external surface discharge, for example, outside the termination, or internal cavity (the dangerous one for the insulation).
This TSO carried out laboratory tests that demonstrated that partial discharges can be measured during existing overvoltages in the grid, whether due to fast transients such as lightning or slow transients such as manoeuvres. This laboratory research led to a change in the method of measuring short-duration partial discharges.
This paper continues the work performed until 2022 showing new advances, mainly from the test in the field. During the last two years, tests have been carried out in real installations to verify what was seen in the laboratory. Several cases of the maintenance procedures applied in real electrical installations where this new technique was used are shown. The advantages of this new measuring technique have been confirmed in the use cases: the measurements are observed more clearly and, therefore, the diagnosis can be more accurate or even, in some cases, an earlier failure forecast can be achieved.
The PD measuring equipment needs to measure exactly during the time that the overvoltage occurs and therefore must be synchronized with the start signal of the overvoltage. The TSO is asking to the PD equipment providers to implement this functionality in the continuous monitoring PD equipment.
In addition, the TSO is working on the continuous monitoring of the entire insulated cable network. The continuous monitoring of partial discharges allows for greater control and management of maintenance based on the risk and criticality of the installations. Based on the data analysed, decisions can be made to prioritise maintenance tasks and thus reduce the downtime of the circuits, in addition to anticipating any faults that may occur.
During the last years, new partial discharge monitoring equipment has been installed, starting from 2014 (the first monitored circuit). In 2023, there are 100 monitored circuits and, by the end of 2024, around 100 more circuits are expected to be monitored. Several challenges have been faced in order to complete the monitoring system installation:
- Analysis of quality information obtained from punctual PD measuring vs continuous PD measuring.
- Limits of monitoring systems regarding to power supply of the equipment.
- Study of location of PD sensors to optimize the sensitivity.
- Limits of communication systems, data management and cybersecurity conditions to fulfil.
As a great number of circuits are planned to be monitored in the next years, the integration of automatic diagnosis (provided by an artificial intelligence-based software) is necessary to manage all the monitoring data.
Keywords
Partial Discharge, Monitoring, Measuring, Cable, Overvoltage, Preventive Maintenance, Automatic Diagnosis1. Introduction
One of the main difficulties in maintaining cable systems, underground and submarine, is that checking the health status of the facilities is not always easy or possible. Indirect parameters must be measured to estimate the remaining useful life or predict whether a circuit is likely to fail in the short or medium term. Examples of parameters that can be controlled are the temperature throughout the entire circuit or at some points, the current measured in the covers, mechanical vibration or partial discharges detected in the main insulation of cables and accessories. An important decision for an electrical company, in order to optimize the resources dedicated to preventive maintenance, is to choose which parameters should be measured and what is the optimal frequency to carry out the different inspections. The choice is usually made between one off measurements or continuous monitoring. This work presents a new method for measuring partial discharges during periods of time in which an overvoltage occurs in the network. This new method, which was tested in the laboratory in 2021 [1], has been verified in the field in several facilities throughout 2022 and 2023. After the results obtained, an analysis has been made comparing the traditional vs new technique. In addition, the progress made regarding the monitoring of partial discharges, the current inspection strategy and the challenges foreseen for the future are shown.
2. Partial discharge measuring in cable systems maintenance
PD measurements are widely considered as one of the best tools to know the state of the insulation of the cable systems and, therefore, try to prevent failures in the circuits before they occur, that is one of the objectives of every maintenance strategy.
The experience of managing the cable systems of the Spanish TSO (with more than 2.000 km of cable installation in service) and the analysis of the PD measuring policy have led to a similar conclusion: online PD measurements are, at this moment, the most reliable tool to know the state of the insulation of the cable and accessories [1].
However, PD measurements are not always easy or even possible to perform, so technical and economical evaluation of the benefits should be done before undertaking the investment in inspections, equipment or installation of PD measuring systems. Management and analysis of the data coming from the measurement equipment has to be included in this evaluation, because the valuable information coming from the measures is not directly obtained.
Signal filtering and correct interpretation of the PD parameters must be done to identify if PD exist in the installation and, if so, if the PD corresponds to corona, external surface discharge, e.g., on the outside of the termination, or internal cavity (the dangerous one).
2.1. PD Monitoring strategy
The traditional maintenance inspection model in the Spanish TSO included periodic inspections such as thermography, cover testing, partial discharge measurements, etc. in all underground cable circuits. Specifically, partial discharges were measured in all elements of the underground cables, but only by performing short-term periodic measurements, that is, measurements carried out once every five years and lasting approximately 30 minutes. The company's maintenance policy is to replace the affected accessory as soon as an internal defect is identified in the PD measurements, as seen in Figure 1. This policy and the PD threshold values were made according to the experience from more than ten years with these punctual measures.
Figure 1 - Procedure for action regarding the PD measure of the Spanish TSO
After analyzing the measurements carried out over several years, it was concluded that the results obtained from the specific PD measurements had low effectiveness because few failures were detected. An analysis process was carried out to study the effectiveness of the periodic inspection of the traditional model. Consequently, some changes were introduced in the partial discharge inspection strategy, such as continuous monitoring of facilities and the introduction of a new method of measuring partial discharges during network overvoltages [1].
2.2. Continuous monitoring of installations
The following aspects should be considered in order to optimise resources by continuous monitoring of the PD [1]:
- The quality of the PD inspection is increased by performing continuous monitoring or temporary monitoring instead of periodic measurements of short duration.
- Power supply: The connection to the substation power grid was made, when available, via a standard AC outlet. Where necessary, different stand-alone systems were implemented, e.g. photovoltaic power panels and batteries were installed at the terminations on the transition towers. In junction chambers, the absence of power supply is often a limitation for the installation of partial discharge monitoring systems, although there are new technologies with passive or pseudo-passive sensors that may solve this problem. The consequence of this limitation in measurement is that only some parts of the circuit length can be measured. Sensors must be installed in a way that optimises the measured length and costs.
- Placement of the equipment: complete engineering projects have had to be carried out to find the best location and connection option for each installation within a substation.
- Communication: Connection to the internal telecommunications network with internal cybersecurity criteria was necessary due to the amount of data generated by the PD equipment and to allow scalability of the parameter monitoring.
Automatic analysis and interpretation of PD measurements using artificial intelligence should be promoted if the number of sensors is high, to allow scalability of the PD monitoring circuits. In this sense, the Spanish TSO (that has more than 100 circuits continuously monitored in 2023 and more circuits planned for the next years) has promoted and integrated a commercial software tool, based on artificial intelligence, that analyses continuously all the monitored circuits and generates alerts in case an internal cavity is detected [3].
In addition, partial discharge measurements in HVDC circuits are not sufficiently developed.
3. New measurement technique
3.1. New technique for periodic in-service PD measurements in very high voltage systems (laboratory phase)
As mentioned above, periodic PD measurements, performed at short time intervals (<30 min), have shown low efficiency in very high voltage cable systems. One of the reasons is that in networks above 100 kV at nominal voltage there is normally no detectable PD activity of the internal defect type (cavity or internal surface). This circumstance forces to reconsider how to perform periodic measurements of short duration. Recent research [2] has shown that partial discharges can be measured during existing network overvoltages, either due to fast transients such as lightning or slow transients such as switching. Based on this research, a new partial discharge measuring instrument has been implemented.
The new PD measuring instrument acquires the PD pulse train generated only by transient overvoltages in the network. Research has shown that the PD pulse train lasts for several cycles, so the PD measuring instrument must be immune to interference caused by the surge itself, but at the same time it must be able to acquire all PD pulses that occur several network periods later (e.g. 1 second). The phase resolved PD (PRPD) pulses synchronised with the network sine wave will allow the construction of a PRPD pattern similar to conventional PRPD patterns recorded in continuous PRPD measurements.
To demonstrate the capability of this new method, HV laboratory tests were performed.
The detailed description of these tests was shown in [1]. The results proved that PRPD patterns obtained from a short time interval (around 30 min), could be compared to cumulative PRPD patterns caused by multiple switching surges (e.g. 7 or 10), being easier to recognise as a “cavity type” fault using the new technique, as it is shown in Figure 2:
Figure 2 - a) Phase-resolved PD pattern generated by the test cell at 6 kV (short time interval), b) Cumulative PRPD of 7 transient switching type overvoltages
3.2. Transient overvoltage on circuits in service
To transfer this new measurement technique from the laboratory to real installations (circuits in service), the first thing to check is whether the circuit is going to have any switching and whether this switching may produce any overvoltage that can be used to detect partial discharges. In all the measures carried out, manoeuvres due to the normal operation of the network were taken advantage of.
An initial approach could lead to think that the switching manoeuvres should be done near the point when voltage sine waveform is zero, so overvoltages, that can accelerate the aging of the insulated cables, are avoided. In Figure 3, the transient voltage in a circuit after switching is performed following this criterion is shown (some small overvoltages appear even when trying to avoid them, 5.1 % higher in this diagram from a real case).
Figure 3 - Voltage wave form if switching is performed at the point when voltage crosses zero
However, in a real installation, voltage and current are desynchronised, in such a way that switching manoeuvres performed near the point when voltage sine waveform is zero could, in some cases, generate currents that damage other parts of the installation (for example, damage to circuit breaker if current waveform lasts too much time to cross zero, like in zero-missing phenomenon). In Figure 4, it is shown the transient voltage in a circuit after switching is performed in a point different to voltage zero (overvoltages appear, 34 % higher in this diagram from a real case).
Figure 4 - Voltage wave form if switching is performed at the point when current crosses zero
The higher is the overvoltage, the bigger is the damage to the cable isolation and the PD will be more likely to be produced.
3.3. New technique for periodic measurements of in-service PD in very high voltage systems (field testing phase)
During the research process, continuing with the mentioned laboratory work, numerous measurement tests have been carried out on real installations in the Spanish TSO network during 2022 and 2023. A summary of the results is shown in Table 1.
| Performed tests | Tests with different patterns | Tests with similar patterns | |||
|---|---|---|---|---|---|
66 | 3 | 63 | |||
Electronic noise | Internal PD | External Surface PD | Corona | ||
45 | 2 | 12 | 4 | ||
To date, 66 tests have been performed measuring from the moment of energisation (switching overvoltage). 3 of them detected patterns during the overvoltage time that had no continuity during the permanent regime (1 out of 3 is the laboratory test). In the rest of the tests, no significant difference was found in the pattern detected during the switching overvoltage (measurements during the first 600 ms) and the pattern detected during the long duration measurement. It should be noted that, from the 63 tests, 45 of them only detected electronic noise coming from the substation equipment, 2 of them were internal insulation discharges, 12 patterns were identified as external surface at the terminations and 4 as corona.
Following, 6 of the 63 tests mentioned above (where no significant difference could be distinguished between the measurements made during the transient and the later minutes) are shown, from Case a) to Case f).
After these 6 cases, the 2 field tests with differences between the PD measured during the transient time and the permanent regime are described, Case 1) and Case 2).
Case a)
Internal discharge in cable joint of a 66 kV circuit:
Figure 5 - a) Pattern generated in a manoeuvring surge during 600 ms, b) Long time measure pattern
Case b)
Internal discharge in a 66 kV circuit (not located):
Figure 6 - a) Pattern generated in a manoeuvring surge during 600 ms, b) Long time measure pattern
Case c)
Surface discharge in a termination of a 220 kV circuit:
Figure 7 - a) Pattern generated in a manoeuvring surge during 600 ms, b) Long time measure pattern
Case d)
Surface discharge in a termination of a 220 kV circuit:
Figure 8 - a) Pattern generated in a manoeuvring surge during 600 ms, b) Long time measure pattern
Case e)
Corona discharge in a termination of a 220 kV circuit:
Figure 9 - a) Pattern generated in a manoeuvring surge during 600 ms, b) Long time measure pattern
Case f)
Corona discharge in a termination of a 220 kV circuit:
Figure 10 - a) Pattern generated in a manoeuvring surge during 600 ms, b) Long time measure pattern
The following two tests are described with more detail because they are the only tests in which the patterns detected during the transient time were different from the patterns measured during long time periods.
Case 1:
In this case, measurements were taken at a position with GIS type terminals in a shielded substation with SF6 insulation. This position had been monitored (continuously) for months without detecting any partial discharge associated with an internal defect in the insulation. The pattern measured in this way gave noise values as shown in Figure 11. Subsequently, measurements were made during several switching transients in the network and different results were obtained from the previous ones, as can also be seen in Figure 11.
Figure 11 - a) PRPD measured during continuous monitoring (without overvoltages), b) PRPD measured during several switching transient overvoltages, c) Pattern once filtered
Once these PD data measured during several transients have been acquired, the criticality of these transients must be analysed.
The pulse train captured during the first 360 ms after the switching overvoltage produced can be seen in Figure 12, as well as the specific case of one of the pulses with a duration of approximately 1 ms.
Figure 12 - a) Pulse train for 360 ms after the switching overvoltage, b) pulse <1 ms,
(c) Frequency response of a particular pulse
Once the pattern has been analysed, it is ruled out that the PD is related to an internal insulation cavity and it is considered that it could be related to some arcing or sparking activity caused by a bad contact. The next task will be to locate the defect contact point.
Case 2:
On this occasion punctual inspection measurements were performed on the GIS terminals of a 220 kV substation taking advantage of the sequence of operation manoeuvres that were to be executed due to maintenance tasks.
The pulse train after the transient switching overvoltage and the pattern generated can be seen in Figure 13.
Figure 13 - a) Pulse train during 1200 ms after the manoeuvring surge, b) Pattern generated in a manoeuvring surge c) Cumulative generated pattern after several manoeuvres
After analysing the patterns generated, it is concluded that they are related to a defect in the main insulation of one of the GIS terminations.
Once the termination has been replaced and analysed, the damage to the internal insulation is verified and confirmed as internal cavity.
3.4. Analysis of new technique vs traditional technique measures
From the analysis of the results, comparing PD measured with new and traditional techniques in every case, 100% of the patterns detected and diagnosed as some kind of PD were observed with the new technique during the first milliseconds after the occurrence. Therefore, the new technique provided zero false negative measures.
Secondly, and more important, it has been seen that almost 5% of the PD patterns could only be measured using the new technique within 600 ms after the occurrence of an overvoltage (If you can’t measure these cases, you will not be able to identify wether or not they belong to internal defects). In these cases, the traditional measuring technique presented false negatives.
Further tests should be done to fine tune the current analysis.
4. Conclusion
The ability to make measurements during network overvoltages should be implemented in the PD measuring equipment since, thanks to this new measurement technique, possibilities of observing discharges in an incipient state increase.
After analysing maintenance inspections for several years, PD has been identified as a relevant parameter of useful information and a significant change has been made in the strategy of maintenance PD inspections by the Spanish TSO. The current strategy includes the selection of critical underground circuits and the installation of PD monitoring systems, as well as measurement during maintenance work requiring manoeuvres.
It has been proven in the field, in real installations, that the measurement of PD synchronised with transient overvoltages due to manoeuvres is a very effective method for the early detection of insulation faults in high voltage networks. In some cases, the new technique provides PD patterns that are not detected using the traditional technique, while the opposite has not happened.
The benefits obtained include improvements in maintenance costs, safety, quality of the data used to estimate the health of the assets and availability of the installations, due to the reduction of repair times and the possibility of anticipating faults before they occur.
References
- R. GÓMEZ, R. REINOSO, G. DONOSO, E. NOGUEROLES - “REE´s commitment to partial discharge monitoring in its underground cable network”. (CIGRE Paris Session 2022, ID-10868)
- Jiayang WU - “Effects of Transients on High Voltage Cable Insulation” (Thesis Dissertation, Delft University of Technology 28th February 2020)
- A. SÁNCHEZ, F. GARNACHO, J. ORTEGO, F. MARTÍN, R. REINOSO, R. GÓMEZ, A. VIVAS, A. RAMÍREZ, A. KHAMLICHI, C. VERA, J. DI DECO, S. GONZÁLEZ, A. MUNICIO, E. SANTOLARIA -“Requirements for Artificial Intelligence Platform addressed to Automatic Assessment of Insulation Condition of Indoor and Outdoor Installations through Partial Discharge Monitoring” (CIGRE Paris Session 2022, ID-11049)
