Steve Eckles, El Paso Electric Co.
Providing safe and reliable service is the goal of electric utilities. This goal is compromised when utility equipment fails or foreign objects and wildlife provide an alternative path for electrons to flow. These short circuits (technically referred to as faults) usually cause conductor current to increase dramatically and voltage to sag or dip significantly. To mitigate utility equipment damage, system instability and safety concerns, protective relays are used to quickly detect overcurrent and de-energize faulted circuits for repair.
Unfortunately, not all faults produce elevated current that conventional overcurrent relays can detect and react to appropriately. Occasionally, multi-grounded primary overhead distribution (4 kV to 34.5 kV) lines fall on poorly conductive surfaces that do not generate enough fault current for conventional relays or fuses to detect and clear. If left undetected, these high-impedance faults (HIFs) are obviously hazardous to life, limb and property.
Conventional Fault Detection and Protection
Most overcurrent faults produce several thousand amperes in a conductor that is designed to carry load current a fraction of that. These high-current faults damage equipment and jeopardize normal system operation. Protection engineering goals are to detect and de-energize these faults quickly–many times under a second. Minimum and maximum fault current calculations for electric lines are used to set relays and select fuse sizes that will de-energize faulted circuits quickly.
However, HIFs are much different. They are defined as faults that “do not produce enough fault current to be detectable by conventional overcurrent relays or fuses” (see Figure 1).
Utility relay protection engineers have been concerned about HIFs for decades. From limited data in the 1940s, researchers gravitated toward using 40 ohms as the maximum reasonable resistance between fallen conductors and the earth ground. This value of resistance is still used by many utilities for overcurrent detection.
More extensive research and experience have shown that a universal 40-ohm ground fault resistance of a downed conductor to earth is over simplified and often much too low. Measurements of staged faults reveal numerous high-impedance ground faults in the neighborhood of several hundred ohms, resulting in fault current from nearly 0 to 50 amps. This is in the range of normal load current and not detectable by traditional over current relays or fuses. Every ground fault is different due to the many variables affecting ground fault resistance including conductor voltage and material; how it contacts the ground surface; and, most importantly, the ground surface itself.
A section of line falling on grass will likely have different fault current levels than one falling on concrete. Moreover, soil make-up differs by geographic regions and its moisture may vary by season. Not only is every high-impedance fault different, they can be dynamic also–changing over time. As an energized conductor lies on the ground, it may break down insulating properties and start conducting more. Conversely, the heat from fault current may decrease soil moisture with time thereby reducing fault current magnitude (see Figure 2). The heat often causes silicon in sandy soils to become molten and cool into glass. In other situations, electromagnetic forces may physically move an arcing conductor.
HIF current can literally change from cycle to cycle. Not only can its magnitude fluctuate widely, its waveform can exhibit random behavior also (see Figure 3).
Some Early Attempts at HIF Detection
Nordon Technologies of Sarasota, Fla., made the High Impedance Fault Alarm System (HIFAS) which focused on the HIF’s third harmonic current changes in relation to the faulted phase voltage.
In the 1970s, Pennsylvania Power and Light Co. (PP&L) with Westinghouse Advanced Systems Technology staged fault tests to develop the Ratio Ground Relay (RGR). It used an induction disc with operating and restraining windings.
The Kearny Manufacturing Co. Open Conductor Detection (OCD) system used loss of voltage to detect a broken conductor in the early 1990s. The system sensed voltage at the end of single phase laterals.
Recent HIF Detection Progress
These early attempts for economical HIF detection did not withstand the test of time. The problem remained; so research went on. The advent of advanced digital microprocessor has resulted in three commercially protective relays that are worth investigating.
Ideal HIF detection relays should be quick to detect all HIFs yet be secure–that is, immune to misoperations and false indications. Detection must distinguish between HIF current waveforms and those of typical intermittent “noisy” loads such as motor commutation, inrush current from multiple motor starting, arc furnaces, welding and DC rectification. Such ideal relays do not presently exist; despite this, present HIF relays do a good job of balancing between detecting HIFs and security against false trips and misindications. Moreover, to increase detection confidence of a changing HIF current signature, relays may take additional time to determine whether a likely fault exists.
Even though the techniques each relay uses are different, there are some general similarities that enable the relays to discriminate between low-level faults and load and formulate HIF decisions.
- Adaptation and learning. Besides noisy customer loads (e.g. motor commutation and welding) stated above, utility operation introduces transients from capacitor and load switching. Thus, methods have been developed to have relays monitor and “learn” typical feeder loads and differentiate between ambient feeder noise and anomalies that a HIF may cause. (These methods include artificial intelligence techniques of pattern recognition and classification to determine probable feeder loads.)
- Multiple algorithms and voting schemes. Due to the wide range of HIF current waveform possibilities, modern HIF relays use multiple algorithms for better dependability and security. The diversity of HIF characteristics commonly result in some algorithms indicating a fault while other algorithms don’t. Internal relay logic evaluates algorithm confidence and probabilities to decide whether a HIF exists. In effect, relay algorithms vote for a fault or not–not necessarily with equal weight.
Three Available Commercial Relays
Research at Texas A&M University since the early 1980s in HIF detection resulted in licensing for a commercial product supplied by General Electric. It analyzes harmonic and non-harmonic current components between 30 and 780 Hz and uses nine algorithms from load analysis to arc burst. By detecting load loss, GE’s HIF detection distinguishes between downed conductors and other arcing faults.
ABB teamed up with Lafayette College in Easton, Penn. to develop and test HIF detection algorithms. ABB’s REF 550 relay detects HIFs using three algorithms centered on higher order statistics, wavelets and neural networks.
Schweitzer Engineering Laboratories Inc. offers HIF detection from its Arc Sense technology (AST) as an option in its SEL-451 relay. At the heart of its detection strategy is the sum of difference current (SDI) method which uses the current rate-of-change to detect arcing from downed conductors. SEL uses adaptive tuning and trending and memory for better accuracy.
Relay Tripping versus Alarm
Utilities employing HIF distribution relays must consider whether to have them trip feeders or only respond with an alarm. Public safety includes guarding against false trips, since customers depend on service for critical loads for hospitals and safety infrastructure.
The reasons for tripping the feeder when a HIF has been detected include to:
- Mitigate injuries or death to those who may contact the line. Current as small as 100 milliamps can cause death in humans.
- Limit legal liability by preventing property damage and wildfires. Under certain conditions, some utilities may be held responsible for the costs of fires caused by their lines. The State of California has asked San Diego Gas & Electric Company to pay over $21 million for fighting the Witch Creek and Rice wildfires that its electric lines have reportedly started in October 2007 (see news story on page 8).
Conversely, utilities may not want to de-energize circuits but create an alarm instead after considering the following:
- Low-current HIFs rarely cause appreciable equipment damage.
- Multiple false trips may persuade utilities to disable HIF detection altogether.
- Continuity of service commitments may require additional evidence of a downed conductor besides a relay indication before de-energizing a feeder.
- Whether they have the ability to dispatch utility personnel quickly to patrol distribution feeders.
- HIFs may be more difficult to find on de-energized circuits and may be downstream of fuses.
Other utilities may make HIF alarm only or de-enerization decisions for each individual feeder based on:
- The criticality and type of feeder load served.
- The feeder’s urban or rural characteristics and its route through residential neighborhoods and school zones.
- The likelihood of broken conductors based on feeder age, history and size.
- Feeder voltage–HIF are more common on feeders under 15 kV.
It is expected that utilities first implementing HIF detection will choose to alarm for HIF until the relays prove themselves.
HIF Relay Testing
Utilities will find testing HIF relays more complex than conventional over-current relays because there are no industry standards or commonly accepted test methods. Practicality prevents many utilities from staging faults on live feeders supplying customers. A more economical alternative is laboratory testing that uses recorded waveforms of various staged HIFs superimposed on a typical load current. A library of staged faults should include a broad range of terrain, climates and ground surfaces. Laboratory testing becomes even more complex when phase angles and scaling factors are considered.
As the cost decreases and capability increases for automated meter reading (AMR) and advanced metering infrastructure (AMI), many utilities have implemented pilot programs. A few utilities are progressing to system wide deployment of this relatively new technology. Although down conductor detection and location is not the primary goal of AMR/AMI, its ability to do so is similar to Kearny’s OCD concept of monitored and reporting voltage at the end of distribution system laterals.
Many AMR/AMI revenue meters send out a “last gasp” upon a power outage that can be instrumental in dispatching utility personnel quickly and accurately to investigate the outage. Using such a system allows power to be restored before some customers know they were involved in an outage.
Detecting all HIFs in electrical distribution conductors have proved inconsistent and uneconomical in the past; conversely, ignoring HIFs did not make them go away. Fortunately, HIF detection is now more reliable and affordable than ever before. Some relay algorithms and decision logic have been designed to make a HIF determination within seconds; others may take several minutes.
With a high degree of accuracy, these relays can detect HIFs with a small percentage of false reports. Utilities can set them to de-energize feeders or only activate an alarm–depending on their policies, philosophies or feeder circumstances.
Downed wires may appear inert but can prove deadly–modern HIF detection is another tool to improve public safety.
Eckles has been a distribution engineer for 16 years at El Paso Electric Co. and is a licensed PE in New Mexico and Texas.