Faults on distribution feeders are a fact of life. Compared to transmission networks, distribution networks are very large and must extend to every customer. The result is that some distribution feeders may be very long in sparsely populated rural areas and frequently must be constructed in adverse environments, especially through wooded areas where tree branches and wildlife may frequently come in contact with the distribution wires, causing both temporary and permanent outages.

In today’s environment, there is increased pressure from regulators and customers to reduce the number and duration of outages. The use of performance-based rates with financial penalties for poor performance is a commercial incentive for utilities to better manage outages attributable to system faults.

The essential characteristics of improved fault management are:

• Quicker detection that an outage has occurred
• Accurately determining the location of the fault
• Isolation of the faulted section of a feeder
• Re-energizing the un-faulted sections of the feeder outside the isolation zone, upstream and/or downstream of the faulted section.

Fault Detection
On traditional distribution networks without Supervisory Control And Data Acquisition (SCADA) at the substation utilities had to wait for customer calls reporting outages to be aware that an outage has occurred. With SCADA Remote Terminal Units (RTUs) at the substation the SCADA system is immediately aware of faults that cause both temporary and permanent breaker trips. However, faults that are cleared by fuses can still go undetected until customers call to report outages.

An interesting possibility – one not too widely used – is for the SCADA system to detect sudden drops in feeder amps larger than normal load variations and predict the presence of partial feeder outages due to faults when the SCADA system is monitoring feeder currents at the substation. Moreover, when the DMS (Distribution Management System) has a real-time model of the distribution feeder with some form of state estimation, the DMS can use the magnitude and phase of the drop in amps to trace down the feeder and thus, determine possible locations where a fuse may have blown and or an open-conductor fault has occurred. This information can also be combined with customer call data to get more specificity about the suspected location of a blown fuse or other open circuit fault.

Fault Location Methods
Fault passage indicators being tripped by the passage of a high phase current with a threshold above the maximum expected normal feeder load have traditionally determined fault locations. In the past, such fault detectors were purely local, and field crews had to travel to each fault indicator site to see if it was set or not. However, with telemetered fault indicators, the status of each fault detector can be determined almost immediately by a SCADA/DMS system and can be used by the DMS – again in conjunction with the DMS network topology model – to determine the section of the feeder in which the fault has occurred.

However, very long feeders with only a few widely-spaced, telemetered fault passage detectors, may require that field crews drive considerable distances, continuously patrolling the line, to locate the actual fault location.

Modern digital relays have the ability to calculate the fault impedance based on measured phase voltage and current waveforms captured during the fault before the breaker opens.

This additional impedance information, as well as the phase of the fault, can be mapped against the feeder impedance model to estimate the actual location of the fault. Unfortunately, when a feeder has a tree-like structure with multiple branches, several theoretically possible fault locations – all having the same network impedance – may exist.

Another source of error when estimating the fault location from fault impedance measurements is that the impedance of the fault itself is not known. If the fault has significant impedance, the estimated fault location (assuming zero fault impedance), may be
further away from the substation than the true location. Nonetheless, this information is still very useful since the estimated fault location distance from the substation is an upper boundary, and dispatchers can know that the true fault location is closer to the substation. Since the fault impedance tends to be purely resistive, another alternative is to do the calculations considering only the reactive component of the impedance values.

Switching actions to isolate the fault and restore power to un-faulted sections
One of the most important drivers for feeder automation is to permit fairly rapid switching actions to isolate the fault and restore power to un-faulted sections occurring upstream and/or downstream of the faulted section.

Most regulatory bodies consider momentary outages lasting less than a defined maximum period of time (typically of the order of one minute)  to be less serious than sustained outages. Therefore, it is highly desirable that the fault isolation and restoration switching be comp­leted in less than the threshold for momentary outages.

Various types of systems are available to provide automated detection and location of feeder faults, followed by automated switching actions to isolate the faulted section and to restore power to un-faulted sections of the feeder, as follows:

• Distributed fault management using multiple intelligent local controllers
• Centralized intelligent fault mana­gement system

Using a distributed system with multiple intelligent local controllers
In this type of system, multiple feeder switches are equipped with intelligent local controllers that are able to monitor electrical variables at that switch as well as open and close the switch autonomously. From a fault management perspective, the main variables monitored by the intelligent switch controller are the detection of fault currents larger than normal load currents and loss of voltage due to an upstream breaker or re-closer trip.

It is also desirable that the local controllers collect and store historical load data for each switch that can be used later to estimate the total load of an un-faulted downstream feeder section for determining the feasibility of the load transfer to a neighboring feeder. Another key requirement is that the intelligent controller needs to have a basic topological model of the feeder, including the relative location of other switches having intelligent controllers.

Finally, in order to perform the fault location analysis and reasoning about feasible switching actions, the local controllers need communications facilities to communicate with one another and determine which switches saw (or did not see) the fault current.

The advantage of using local intelligent switch controllers is that small systems for fault management can be deployed relatively quickly and inexpensively. For example, it is possible to deploy a scheme for a single feeder using only two intelligent controllers; one at a normally-closed switch at the midpoint of a feeder and one at a normally open tie switch downstream of the first controller. This allows a utility to deploy a small pilot project on a few high priority feeders with a relatively small investment. Then, after gaining experience with a small deployment, the utility can gradually expand the system to automate additional feeders.

The main disadvantage of using local intelligent controllers is that the system usually functions properly only under normal network operations. That is, whenever the feeders are in an abnormal condition, the local controllers must be turned off or reprogrammed, which is not usually a trivial task. Also, as the number of deployed intelligent controllers increases, the data maintenance and management of the decentralized scheme becomes increasingly burdensome.

Using a single centralized fault management system
Centralized fault management schemes are usually imple­mented as a subsystem within a general purpose SCADA and Distribution Management System. In such a system feeder switches can be made remotely controllable by the addition of relatively simple motorized switch controllers and a low cost small RTU equipped with communications facilities to communicate with the centralized DMS. As a minimum, the RTU must be able to detect the passage of fault currents, report this to the DMS and subsequently process and apply switch open/close commands received from the DMS.

There are a number of advantages of using a centralized scheme, as follows:

• Among the most important advantages is that the DMS typically has a comprehensive real-time view of all network conditions – including all of the planned and unplanned outages – as well as all abnormal network topologies. Thus, when a fault occurs, the centralized fault management application can automatically take into account any currently abnormal network conditions when analyzing fault locations. This is especially key when determining and evaluating possible load transfers to restore power to un-faulted sections downstream of the identified feeder fault location. For this reason, the centralized fault management scheme is generally deemed more robust and reliable than a de-centralized scheme
• A centralized system is able to analyze and recommend more complex switching scenarios, including second-order load transfers, to increase the capacity of neighboring feeders to pick up load from the faulted feeder.
• In a centralized scheme implemented on a DMS, the switching actions determined by the fault management software – in some cases including multiple possible alternatives – can optionally be first presented to a network operator for approval before the switching is performed. (This is generally not possible on a distributed system.)
• The centralized system can also be used on non-telemetered feeders, once field crews have provided indications of a feeder fault, to determine possible isolation and restoration switching scenarios, using manually operated switches, in cases where the time to repair the fault will be significant.
• To a considerable extent, the algo­rithmic logic used to determine the switching actions to isolate and restore faults in a centralized system is quite similar to the logic required to determine pre-planned switching actions to realize planned outages on a distribution network. As such, the fault management software can also be used to generate planned switching sheets.
• Because all of the fault management parameter data and algorithms are stored locally on one centralized system, the general management of this data is much easier than with a distributed system.

A centralized fault management system can, when desired, be enhanced to perform more precise fault location using fault impedance data available from modern digital relays. Compared to a distributed system, perhaps the biggest disadvantage of a centralized fault management system is that it is generally less practical to deploy for managing faults on a small number of feeders.

Finally, since centralized systems are usually part of a comprehensive SCADA/DMS system, these systems are generally deployed with a view to managing a complete distribution network, and the capital cost is typically significantly larger than that required to deploy a small, distributed system.

Conclusion
This article has focused on the basics of fault management on distribution feeders. However, a relatively wide range of fault management schemes are available from multiple suppliers that can provide assistance with decreasing the duration of outages experienced by utility customers. Both distributed systems (using local intelligent switch controllers) and centralized systems are available, each of which has advantages and disadvantages that must be weighed by the user to determine which approach is best for a given application.

About the Author
Roy Hoffman is currently DMS Product Manager for SNC-Lavalin Energy Control Systems. He has over 25 years experience in SCADA, EMS and DMS systems. Dr. Hoffman is interested in the application software of SCADA, Energy Management and especially Distribution Management Systems as applied to the real-time operation and control of electric power networks. He is a member of several committees of the IEEE Power Engineering Society related to his area of expertise as well as the Canadian Standards Committee for IEC TC 57 – Power System Control and Associated Communications and is a member of the IEC TC57 Working Group 14.