Using Waveshape Analysis for Fault Location in Distribution Systems

By Mario Tremblay, IREQ; Michel Demers and Georges Simard, Hydro-Québec; and Mark McGranaghan and Jinsang Kim, EPRI

Locating faults quickly and accurately is important in distribution system reliability, and distribution companies continuously look for improved approaches to locate faults.

A number of fault-location approaches exist. Some companies use faulted circuit indicators to make it easier for crews to find faults when they patrol lines. These indicators are available with communications interfaces that allow them to be integrated with automation and outage management systems.

Another approach involves the use of monitoring information from either the substation or along the feeder. This is important because the application of monitoring devices (PQ monitors, digital fault recorders, digital relays, reclosers and even advanced meters) is more common.

The Electric Power Research Institute (EPRI), Hydro-Québec and other utilities are evaluating two approaches for using available monitoring data to locate faults:

1) An approach using substation voltage and current waveforms associated with the fault, calculating the impedance between the substation and fault and then locating the fault using distribution circuit models. This technique is called reactance-to-fault (RTF) because it is used only with reactance so the results are not biased by the fault resistance.
2) An approach using distributed monitors on the feeder that provide the voltage waveforms during the fault (these could be at three-phase customer locations). The voltage waveforms are used with the circuit topology and impedance data from a circuit model to locate the fault. This is called the voltage drop fault location (VDFL) technique.

Progress Energy Carolinas has used substation monitoring for years to locate faults successfully. The approach was then implemented as part of a power quality monitoring system (the EPRI/Electrotek PQView system) in cooperation with Con Edison. It is being fully used at Con Edison to locate faults in overhead and underground circuits and subsequently has been demonstrated at San Diego Gas & Electric and United Illuminating.

Additional trials are underway with Wisconsin Public Service and Detroit Edison.

Hydro-Québec is testing the VDFL technique on an automated fault location system prototype named MILE, covering seven distribution feeders. One feeder was selected for the joint project and a monitoring device was installed at the substation supplying the feeder for the evaluation of the RTF technique in parallel to the VDFL approach.

Additionally, Long Island Power Authority is testing both approaches on two feeder circuits.

Method 1: Rtf

Fault location based on substation fault currents began at Progress Energy Carolina using steady-state trend data and fault events on all feeders using a remote-terminal unit (RTU) that can sample at a rate of 16 samples per cycle.

The utility’s fault-location system started as a spreadsheet that assumed a constant conductor size for a given circuit. When the fault current and type of fault was entered, the spreadsheet would estimate a distance.

Success using that system led to development of more automated systems to locate faults, including an interface to its distribution models to identify all possible locations for the fault and then interfacing with its outage management system to display the possible locations.

The RTF system builds on Progress Energy’s basic approach. The distance from the monitoring location to the faulted section of circuit can be estimated by using the voltages and currents during a fault.

The equation is simple, just Ohms Law:

Click here to enlarge image

where,
V = voltage during the fault, V
I = current during the fault, A
Zl = line impedance, ohms per length unit
d = distance to the fault, length unit such as miles

With complex values entered for the voltages, impedances and currents, the distance estimate should come out as a complex number. A simplification of this approach uses the reactance(X) to the fault as:

Click here to enlarge image

The advantage of using the reactance is avoiding the arc impedance, which is mainly resistive.

Con Edison initially developed the RTF approach and has developed a complete system around it (see Figure 1). The operator can access the fault location information available as part of a complete Web-based interface that also incorporates relay target information and information from line crews. This has provided fault location to within one to three manholes of the actual fault location for more than 70 percent of the fault events.

Click here to enlarge image

The system has also been tested at San Diego Gas & Electric with excellent results for overhead system applications and is being implemented at other utilities. Challenges include integration with the distribution models to identify possible circuit locations for the fault and then integration with geographic information systems and outage management systems to display the results.

Method 2: VDFL

VDFL uses voltage waveforms associated with fault conditions (voltage sag waveforms) measured at distributed locations on the medium-voltage feeder. To reduce installation cost, monitors can be installed on the low-voltage side of distribution transformers and could even be part of advanced metering systems.

The VDFL requires at least two monitoring sites along the feeder to locate the fault, but the full potential and accuracy of the technique is obtained with approximately four monitoring sites. When monitoring several feeders connected to the same substation transformer (busbar), the number of sites could be reduced by one. The Hydro-Québec study indicates a ratio of 4.5 monitoring sites per feeder without considering the possibility of matching feeders connected to the same busbar.

The VDFL algorithm can be compared to an interpolation. Two measurements are used to evaluate the fault current considering the line impedance and a third one to identify the faulty laterals and the possible fault locations. The substation impedance is needed if only two monitors are used. Figure 2 shows a feeder with three monitors and use of the voltage phasors from the sag voltages to determine the fault location.

Click here to enlarge image

The distributed monitoring approach allows the faulted lateral to be identified. This is superior to substation monitoring that can identify only a range of possible locations that are the same electrical distance from the substation.

The voltage at the fault location is also deduced from the algorithm. Because the voltage magnitude is proportional to the electric arc’s length, this can provide information about the cause of the fault and the component affected. Arcing over a distance greater than a meter, for instance, could indicate vegetation-related faults. The Hydro-Québec application (MILE) uses meteorological data recorded at the time of the fault to refine the analysis. Finding a defective insulator or a repeating vegetation contact can allow the problem to be corrected before such issues can cause a customer outage. The system is implemented with a Web-based interface (see Figure 3) to allow convenient access for all technical personnel.

Click here to enlarge image

null

Summary of Results

The research project assesses the fault location performance on specific feeder circuits at Hydro-Québec and Long Island Power Authority. The Hydro-Québec feeder being evaluated includes 211 miles of conductor and feeds 1,760 customers. Six monitors have been installed, one at the substation and five at distributed locations on the feeder. A distributed generation resource (DGR) of 2 megavolt ampere (MVA) is connected to one of the feeder laterals. This section of feeder was scheduled for a transfer to another feeder in 2008, but the transfer was postponed to late 2010. In response, a study on how the delay impacts this DGR on fault location techniques was added to the project.

From Oct. 29, 2008, to May 1, 2009 (six months), waveforms for 26 fault events were recorded. Twenty were correlated to seven outage locations. The difference between the number of measurements and outages comes from multiple operations of circuit breakers during faults.

The accuracy for the VDFL approach is expressed in percentage of main feeder length (absolute distance error divided by the main feeder length). The accuracy for the RTF approach is expressed in percentage of main feeder impedance (absolute impedance error divided by the main feeder impedance depending of type of fault). This method allows simple translation of the error to an equivalent distance error. The main feeder length is 22 miles and 1 percent of error for both techniques represents 0.22 mile.

The mean of the absolute error for VDFL is 2.4 percent, which represents 0.53 mile of deviation, and the presence of distributed generation did not affect results. The mean of the absolute error for RTF is 3.9 percent, representing 0.86 mile of deviation. These are preliminary results and further assessment is under way for feeders at Hydro-Québec and Long Island Power Authority.

Participants can draw a few conclusions from preliminary results as more information develops from the ongoing work:

1. The RTF method is easier to implement because it requires only monitoring at the substation where many utilities already have monitoring and communications.
2. The RTF method does not, however, identify the specific lateral that is faulted. This method can be enhanced by using faulted circuit indicators or by integrating with outage management systems or other information that can help identify the specific lateral that is faulted.
3. The VDFL has the advantage of providing more specific information about the fault location and is somewhat more accurate based on preliminary results.
4. Distributed PQ monitors in VDFL also provide measurements of voltage level, voltage unbalance, transients and more at several locations along the feeder. Those measurements are useful for controlling supply quality and detecting problems such as inadequate load transfer, defective voltage regulators and blown capacitor bank fuses.
5. The VDFL approach also can take advantage of advanced metering infrastructure (AMI) at three phase customers and of advance distribution automation infrastructure where communication is available.

A combination of these two approaches might provide greater accuracy and will be evaluated in the ongoing assessment.

Mario Tremblay is a researcher at IREQ (Hydro-Québec’s research laboratory). He may be reached at tremblay.mario@ireq.ca.

Michel Demers and Georges Simard are with Hydro-Québec Distribution. They may be reached at demers.michel.f@hydro.qc.ca and simard.georges@hydro.qc.ca.

Mark McGranaghan and Jinsang Kim are with the Electric Power Research Institute (EPRI) and may be reached at mmcgranaghan@epri.com and jinsangkim@epri.com.

Previous articlePOWERGRID_INTERNATIONAL Volume 14 Issue 8
Next articleElectric distribution co-ops report strong performance during downturn
The Clarion Energy Content Team is made up of editors from various publications, including POWERGRID International, Power Engineering, Renewable Energy World, Hydro Review, Smart Energy International, and Power Engineering International. Contact the content lead for this publication at Jennifer.Runyon@ClarionEvents.com.

No posts to display