Schematic view of a typical Smart Cable Guard setup with a monitored cable section spanning multiple RMUs installed in medium-voltage/low-voltage substations.

On-line Fault Detection to Prevent Network Failure

June 18, 2018
Application of on-line fault and partial-discharge activity location helps to improve the reliability of medium-voltage underground cable networks.

Medium-voltage underground cable systems are subject to outages, but distribution network operators can improve system reliability in terms of restoration time if the location of faults on the network are located precisely and immediately when they occur. Preferably, improved system reliability could be realized if a fault is predicted before an outage and thereby avoided.

One of the largest distribution network operators (DNOs) in the Netherlands, Alliander has a medium-voltage network approximately 38,000 km (23,613 miles) in length. Although the majority operates at 10 kV, some sections of 20-kV cable and even 6-kV as well as 3-kV cables are scheduled for replacement. Keen to improve the fault performance of its medium-voltage underground cable network, the utility focused on faults attributable to partial discharge (PD) and undertook a comparison of the main features of several commercially available on-line PD detection systems. Alliander selected the Smart Cable Guard (SCG) system, designed to locate on-line PD activity.

Alliander has been using the SCG system for several years, which has enabled the utility to replace an asset — often a cable joint — when suspicious PD activity or intermittent faults are detected, before a breakdown and circuit outage occur.

Smart Cable Guard

An underground cable fault is defined as a failure, or breakdown, when the cable insulation is bridged by a short circuit. However, the magnitude of the short-circuit current does not always result in a detectable short circuit. It depends on the network grounding/earthing whether the generated short-circuit current will be large enough to trip the protection equipment.

Initially, the SCG was a proven and effective monitoring instrument for on-line diagnosis of medium-voltage underground cables for locating defects while the cable remained in service. These defects are weak spots that generate PD but have not yet resulted in a cable failure. In 2014, a new feature was incorporated in the design of the SCG to enable the detection system also to be an on-line cable fault locator. Hence, the SCG on-line detection system is now a valuable dual-purpose unit for on-line medium-voltage cable monitoring.

One of the unique features of the SCG system, it does not measure 50-Hz or 60-Hz short-circuit currents to measure and locate a fault. Instead, it detects the traveling waves that occur in the initial microseconds following a fault by way of high-frequency (100 kHz to 10 MHz) current sensors installed at both ends of the cable circuit, which are accurately time synchronized. This offers various advantages aside from detection — and localization — of normal faults, followed by switching actions. The system also detects and localizes other types of faults:

• Faults in a network with an isolated neutral where the fault current is far below the detection level of the protection equipment

• Faults with very short durations (a few milliseconds), after which they disappear, often referred to as intermittent faults. Such self-healing faults typically happen in fluid-filled oil and mass-filled joints in cable networks with an isolated neutral or impedance grounding. Sometimes this can happen many months before a full breakdown occurs, on which the protection equipment can operate.

The fact any fault is located within an accuracy of 1% of the cable length between the sensors enables Alliander to identify the affected cable section, often even the defective component.

The SCG has two inductive sensors placed around the earth leads at each end of the underground cable circuit. Through pulse injection, accurate time synchronization is achieved, which results in accurate localization and good noise reduction. The sensors can be installed on cable sections that are several kilometers in length.

Because it consists of portable equipment, the SCG system enables both temporary and permanent monitoring. Alliander usually installs sensors for a four-month period, although — on some occasions — they can remain in-situ on a circuit permanently. The equipment can be installed without a shutdown in locations where there is safe access to the cable earth, but a shutdown is required on some types of medium-voltage switchgear.

Fault-Location Performance

Alliander’s medium-voltage network is designed as rings but operates radially. Each medium-voltage/low-voltage distribution substation is equipped with an medium-voltage ring-main unit (RMU). Operation as a ring with a midpoint open point provides an alternative supply to all customers in the event of a fault once the faulty medium-voltage cable section is isolated. Faults on the medium-voltage network often occur at joint positions, interrupting supplies to many customers. Hence, isolation of the faulty section and repair is given high priority to minimize the impact on the system average interruption duration index (SAIDI) and system average interruption frequency index (SAIFI).

The SCG system can reduce the impact on these indices significantly because of identification and localization of both PD activity and (intermittent) faults in paper-insulated lead-covered (PILC) and cross-linked polyethylene (XLPE)-insulated cables and cable joints. These phenomena often precede a permanent fault.

SCG also can be used on interconnected medium-voltage networks, but this affects the maximum effective length of the cable between two sensors. Signal dampening in PILC cable is different from that in XLPE cables, and this also varies between different types of PILC and XLPE cables as well as the number of intersections in the cable circuit. To enable these variations, Alliander considers cable lengths for SCG applications of 5.5 km (3.4 miles) for PILC cable and 13 km (8 miles) for XLPE cable as the maximum lengths for successful PD activity detection. The fault-location functionality also works at longer circuit lengths.

Alliander began to install SCG systems and, over a period of 22 months, installed 67 units on some 240 km (149 miles) of 20-kV and 10-kV underground cable. On average, each circuit was 3.7 km (2.3 miles) in length, often including more than one cable section and with RMUs installed in distribution substations between the two sensors.

During this period, 19 intermittent faults occurred on the network sections monitored by SCG, of which 15 were detected correctly. The cause for the four missed detections was identified and incorporated into the operating instructions to prevent a recurrence.

In one case, the SCG was installed after other equipment detected but could not locate an intermittent fault. When, after 250 days of monitoring, there were still no detections or cable failures, the SCG system was removed and used elsewhere. These faults were assigned five different fault categories (from A to E), based on the occurrence and time frame of intermittent faults preceding breakdown or repair.

Case Studies

The following outlines two specific case studies using the SCG system.

At an intermittent fault location, intermittent faults were occurring in a 4-km (2.5-mile)-long, 10-kV PILC cable (fault category D):

• The cable section was isolated manually and tested (18-kV DC withstand test).

• The cable broke down at 2869 m (9413 ft).

• SCG predicted a location at 2872 m (9423 ft), a 0.07% location error.

• The cause of the fault was a small stone penetrating the lead sheath of the cable.

At a PD location, PDs occurred in a 13-km-long, 20-kV XLPE-insulated cable:

• There were several wind farms connected to this radial feeder.

• A sharp rise occurred in PD intensity at 4 km, increasing from <5000 pC to 10,000 pC in a few days.

• The cable joint was pinpointed and removed before a circuit outage.

• After replacement, there were no more discharges (apart from randomly distributed wide-band noise caused by the wind farms).

• The fault cause was an overheating connector that cracked the XLPE insulation.

Results Analysis

In total, 20 potential fault outages were examined of which results for four potential faults were indeterminate. It is assumed there will be fewer instances over time, and it is anticipated the next-generation SCG introduced recently will eradicate this problem.

In general, in most cases, multiple intermittent faults previously unknown were detected prior to breakdown or repair. The time frame for the intermittent faults can vary from a few seconds to hours and days. Of the 16 remaining fault cases, 11 cases (69%) resulted in an immediate breakdown or were preceded by intermittent faults in such a short time frame Alliander was unable to prevent the fault outage. In these cases, the SCG system helped to support the network restoration by providing accurate fault location information.

In the remaining five cases, the fault could have been prevented. In addition, because of SCG’s other feature — locating PD from weak spots — Alliander could identify nine such weak spots, which were replaced, preventing failures.

Consequently, the application of SCG provides the opportunity to reduce customer minutes lost. It is estimated with 64 SCG systems — guarding 240 km (149 mile) installed on sections of the medium-voltage underground cable network — the average potential reduction over a period of 22 months was approximately 6150 customer minutes lost per SCG annually.

The accuracy of the SCG system fault location was consistently high, with an average location error smaller than 1%. A similar value is known to be applicable for SCG’s PD locator. Based on this experience, Alliander considers the SCG system a reliable and valuable tool to help reduce SAIDI and SAIFI. For the year ending 2017, it is estimated the installed SCG units resulted in a reduction of more than 1.4 million customer minutes.

Economically, preventing failures by acting on PD activity as well as intermittent faults and identifying the position of the breakdown quickly is extremely valuable. In case of PD- and intermittent-fault-related replacements, the fact the component has not destructively failed yet also facilitates determination of the exact root cause. This is valuable feedback that can be used in the training of technicians engaged on underground cable laying and jointing as well as for manufacturers to improve cable joint design.

Fault Passage Indicators

Fault passage indicators (FPI) are installed on underground cable networks normally in distribution substations that, by indicating the passage of fault current, can assist field engineers in identifying the faulty cable section on the underground cable network. Alliander has decided the SCG can replace FPIs because they only identify the area of a breakdown, whereas SCG directly determines the fault location. Also, SCG is responsive to intermittent faults and PD activity, neither of which can be detected by FPIs.

SCG can be used in any underground cable network, independent of the network grounding (for example, direct grounding, impedance grounding and isolated neutral). This means the SCG works well in a network with an isolated neutral, whereas an FPI is not normally activated because of the reduced fault current level.

Alliander considers SCG an important on-line monitoring system that can monitor many different parameters simultaneously and is a future-proof concept. Based on its operational experience, Alliander decided to extend the number of SCG systems available from 42 in 2015 to 142 in 2016, with a total of 400 units expected to be in service by September 2018. Long-term, the utility’s objective is to increase the number of units in service to several thousand.

Effective Tool

Alliander is satisfied with SCG, as it has been a highly effective tool for on-line detection and localization of faults as well as PDs in medium-voltage power cables. These phenomena are located with an inaccuracy less than 1% of the total cable length in almost all cases. On average, each SCG system prevented about 6150 customer minutes lost per year.

Preventive repairs also give the opportunity to see the cause of degradation, helping to develop countermeasures. SCG can be used in all types of medium-voltage cable networks and effectively replace the need for FPIs; hence, Alliander plans to implement more SCG systems in the future.


The authors wish to thank Stefan Lamboo at Alliander for his support and contribution to this article. ♦

Denny Harmsen is a senior innovation consultant working for Alliander. During his 22-year career working on asset strategy and innovation, he has developed intimate knowledge of the energy infrastructure and can recognize the value of proven technology, and then translate and embed it within Alliander. He believes ingenuity lies within simplicity.

Frank van Minnen is a consultant of electrical power systems working for Alliander. Currently, he is deeply involved in several core innovation projects combining technical expertise with a comprehensive knowledge of the energy infrastructure. With a broad interest ranging from components to analytics and field work to strategic topics, Minnen has a hands-on approach and likes to get things done.

Paul Wagenaars studied electrical engineering at the Eindhoven University of Technology in the Netherlands. In 2004, he received his master’s degree and his Ph.D. in 2010 from the Eindhoven University for his work on the integration of partial-discharge monitoring systems in medium-voltage cable networks. Wagenaars joined KEMA — now Det Norske Veritas in Norway and Germanischer Lloyd in Germany (DNV GL) — in 2010 where he has continued to work on the development and application of on-line cable monitoring systems called Smart Cable Guard. Currently, he is product manager of Smart Cable Guard and a senior engineer in power cables and diagnostics at DNV GL.

Voice your opinion!

To join the conversation, and become an exclusive member of T&D World, create an account today!