1

WG -K3 …………[FINAL Draft 13.2] 9-16-09 bap

Reducing outage durations throughimproved protection and autorestorationin distribution substations.

Outline:1. Introduction & Evolution of protection

schemes

2. Fault Duration, Reduction Methods, andFunction Utilization of microprocessorrelays

3. Faster Detection

4. Faster remediation through IEDcommunication. How communicationaffects protection.

5. How communication affects restoration &analysis, oscillographic and event records,remote access security.

6. Data utilization for automated and postanalysis.

7. Bibliography & references

8. Appendix A- Reliability Indicators

9. Appendix B- Characteristics of DigitalRelays

…..……………………………………………

WG-K3 Members:Bruce Pickett – ChairTarlochan Sidhu- Vice Chair

Scott AndersonAlex ApostolovBob BentertKirt BoersOscar BoladoPatrick CarrollSimon ChanoArvind ChaudharyFernando CobeloKen CooleyMarion CooperRick CornelisonPaul ElkinAhmed ElneweihiRafael GarciaKelly GardnerTim KernBogdan KasztennyGary MichelGeorge MoskosJim NiemiraTim NissenFrank PlumptreCharles SufanaDon Ware

2

1.0 Introduction and evolution ofprotection schemes

This discussion will primarily address the subject ofreducing outages / durations and autorestorationapplications within the distribution substation.

The early distribution substation had limitedtechnology available to the electric industry. Forexample, the high voltage transmission line wasconnected to a power transformer via a manualfused switch, and the feeder load was connected to adistribution bus. When a fault or overload occurredon the transformer, the only option was to blow thefuse, putting the customers in the dark until aswitchman could be dispatched to the substation .See figure 1.

As time progressed, the fuse protection scheme wasreplaced by electromechanical relays. This provideda more precise protection method but still did notprovide a way to automatically pick up the customerload. One such initial scheme cleared thetransformer fault by having the electromechanicalrelays close a high voltage grounding switch. Thisput a ground on the transmission system, where thetransmission line relays then dropped the linesection that fed the faulted transformer.

As the customer load increased, this resulted in notonly the need for larger transformers, but multipletransformers in a substation. Switch manufacturersalso responded by providing switches for the highvoltage side that were capable of breaking lightfaults on the low side of the power transformer.

As protection technology evolved, movement awayfrom the time overcurrent (TOC) protection scheme

to a differential protection scheme occurred. Thisprovided for defined “zones” that identified faults inthe area of the transformer, bus, or feeder breaker,and also significantly reduced the time required torecognize and clear a fault. Not only did this reducethe time that the fault was on the system, but it alsoreduced the equipment damage due to the fault.One such example is shown in Figure 2.

With multiple transformers, there was now thepotential for more than one source of power for thecustomer in the substation. While the customer isonly hooked up to one source at a time, manualswitching can transfer the feeder load from one busto the other. In the case where automation exists,there are methods to transfer the feeder load fromone bus to another through multiple sources.

As new substations were built, the differentialprotection schemes, coupled with bus tie breakersallowed for the automatic closing of the bus tiebreaker once the “bad” transformer was isolated.This was typically accomplished by monitoringswitch positions and operating auxiliary relays ifthe reclosing logic was satisfied.

A common configuration of a substation maytypically grow to two or more transformers,however, the initial load requirements may onlyrequire one transformer.

Figure 3 is a typical example of one such twotransformer substation. For simplicity, only one

Figure 1 -Fused Protection

Fuse Switch

Transformer

Feeder Load

Manual Disconnect Switch

Figure 2 - Differential Zones

MMM

M M M

ONE DIFFERENTIAL PROTECTIONSCHEME FOR EACH TRANSFROMER

ONE DIFFERENTIALPROTECTION SCHEMEFOR ALL LOWVOLTAGES BUSSES

3

feeder per bus is shown, while in reality, this wouldusually be multiple feeder breakers on each bus.

For many power companies, history has shown usthat almost all bus faults are temporary in nature.Examples of this would be foreign interference dueto animal faults, bird streamers or bird nests,dropped or blown objects, tree branches, etc.While the ideal situation would be to remove allpossibilities of foreign interference, this wouldgenerally be cost prohibitive. Animal guards,fencing, containment walls or enclosed substationswithin a building would be some examples ofpassive preventative measures.

In any case, if a fault occurs, and the source isremoved for a short period of time and thenautomatically restored, the customer can be pickedback up, significantly reducing the outage time.

With the technology available where the faultlocation can be determined to be in the bus zoneand not in the transformer zone, the transformercan be isolated from the fault, wait a predeterminedamount of time, and then reclosed back in.

For example, when a fault occurs on the lowvoltage distribution bus, the relay protection tripsthe motor operated switches, clearing thetransformer, and then the bus tie breaker closeswhich restores power to the customer.

Another topic to consider is that it is bad for thetransformers to be overloaded, as this results indecreased life or failure. You may not want to tryand autorestore the entire station load onto onetransformer if the loading on the two transformersexceeds a certain value. Should this occur, youstand the chance of overloading the remainingtransformer, and it could also be damaged or failwhich would also result in extended outages. Thereare schemes that monitor transformer MVA andtemperature loading. Should it be determined thatthe loading exceeds a predetermined value,

autorestoration is turned off. Once load returns toan acceptable limit then it is automatically enabled.

It should be noted that there is always someequipment risk to automatically restore thesubstation and pick the customers back up, if thefault was permanent. However, autorestoration forsubstation events can have a drastic effect on thereduction of extended customer outages for non-permanent faults.

In new substations, microprocessor based relays arereplacing the old electromechanical relays forseveral reasons. Among those reasons is the addedfeatures and capabilities that are offered with thenewer technology that an Intelligent ElectronicDevice (IED) offers. Retrievable files includesettings, event records, disturbance or oscillographicrecords, fault location, etc. Input/output logic canperform what could be construed as adaptivefunctions to some extent. Adaptive settings canreact to load level changes and adjust settingsaccordingly. This would be advantageous for winterpeak temporary settings. Taking advantage of thesenew functions offers reliability enhancements toremove failure prone auxiliary relays that are usedfor the logic necessary for autorestoration.

By taking advantage of the internal logic functions,wiring on the relay panels can be greatly reduced,and relay communications can take place over fiberoptics cables or Local Area Networks. The relayscan be time synchronized to GPS time. This allowscomparison of all fault records and sequence ofevents files to analyze events that take place.

Many power companies are now measuring orkeeping track of reliability indicators for outagesand interruptions. Further information on thissubject can be seen in Appendix-A.

2. Fault Duration, Reduction Methods, andFunction Utilization of micro-processorrelays

Protection relays are no longer simply a singlefunction device that only mimics theelectromechanical relay function that it replaced. Inaddition to the operate function and output contacts,there are other various features built intomultifunction distribution relays that enhance theprotection. Forms of programmable logic controlhave manifested themselves with names such asinput/output logic, ladder logic, and similar terms.Multiple settings, setting groups, and even adaptive

Figure3- Multiple Transformer Station

Fuse Switch

Transformer No. 1

Feeder Load No. 1

Manual Disconnect Switch

Fuse Switch

Multiple Transformer Station

BUSBreakerNo. 1

Manual DisconnectSwitch

Transformer No. 2

Feeder Load No. 2

4

settings may be offered [1], [2]. Winter/summersettings, load related, or storm related temporaryreclosing sequence settings are just a couple ofexamples currently in use.

2.1 Coordination of multiple overcurrentelements or devices

Using multiple overcurrent elements can reduceoperating time of the relay for certain faults. Whenovercurrent elements are set to coordinate atmaximum fault levels with downstream devices, thecoordinating margin at lower currents is usuallygreater resulting in longer clearing times. Definitetime elements can be set to decrease clearing timesin this mid-current area.

2.2 Double Feeder Fault Coordination Scheme

A double feeder is a term used to describe twoindependent feeders routed on the same polestructures. A double feeder fault is when bothfeeders are involved in a common fault. When thishappens, each feeder tends to experienceapproximately half of the fault current available ateach feeder relay. However, the main bus breaker’srelay will see the entire available fault current. Thiscondition lends itself to the possibility of the mainbreaker timing out and tripping before the feederbreaker and is extremely difficult to coordinate forbetween the main and the feeder breaker relays.This scheme has been devised whereby the mainbreaker utilizes a second 51-2 TOC element. The51-2 element will typically have the same minimumpickup value setting as the 51-1 element in the mainbreaker, but the time setting will be set to operateslightly faster. Once the 51-2 times out, it givespermission to the feeder that is still processing faultcurrent to trip. Therefore, the feeder tripsimmediately before the main breaker trips the entirebus.

2.3 Breaker failure- stuck feeder breakerIn the above, it was described how coordinationbetween the bus breaker and feeder breaker mightbe performed in order to avoid tripping the busbreaker for a feeder fault. On the other hand, wherea feeder breaker is stuck closed, or for whateverreason, fails to trip once its protective relays havecalled for a trip, traditional relaying typically waitsfor the TOC relays on the bus breaker or transformerto backup the feeder breaker and trip that portion ofthe station out for what appears to the transformer asan acute overload condition, when in reality it isbeing subjected to fault current.

In the case where the feeder relay has the capacity tocommunicate to the bus breaker or transformerrelay, then the ability to get a faster trip on thetransformer exists as a local breaker failure type oftrip. This removes the fault much faster thanwaiting on overload protection to pick up, which inturn, reduces the potential damage to all of theinvolved equipment, as well as lessening the I2tcumulative damage.

Local breaker failure protection centers around thefailure of a breaker to trip. It can typically fall into afew categories- (a) The trip coil or coil and linkagehas gone bad; (b) The mechanism has frozen up;(c) The mechanism operates but the interruptersremain closed.

Local breaker failure schemes may assume that theproblem associated with the trip coil failure will tryand retrip the breaker to a second trip coil if soequipped. However, in the case of mechanism orinterrupter problems, then the only recourse is totake a step back and trip the next upstream device.Many, if not most, distribution breakers only containa single trip coil.

Another way of accomplishing breaker failureprotection is to rely on TOC or overload relays.

The new microprocessor relay and it’s ability tocommunicate upstream can signal the upstreamdevice to go ahead and trip if it senses that a tripfailure has occurred and that fault current is stillpresent and a trip output is still being called for.

By the same token, communication can also beutilized to accomplish remote breaker failureprotection, which could be used between thesubstation devices and field sectionalizer devices aswell for isolation of faulted feeder sections andautomatic pickup in the field.

2.4 Fast clearing of bus faults utilizingcommunications between feeder and bus relays

With the use of microprocessor relays for bus andfeeder protection, a, scheme can be utilized to speedup tripping of the bus relay for bus faults withoutovertripping for feeder faults. This is achievedthrough communications between the feeder relayand the bus relay. Communications between thetwo relays could be via hard-wire or through astation LAN.

5

The instantaneous overcurrent element in the feederrelay would be set to detect feeder faults and send a"block" signal to the bus relay. This is done byconnecting a feeder relay output to a feeder busrelay input. This feeder instantaneous element mustbe set higher than the maximum load on the feeder,taking into consideration cold load pickup and anyother operating conditions that could result in higherfeeder load than normal.

For the bus relay, an instantaneous overcurrentelement would be set above maximum bus load,with margin, and above the feeder instantaneouselement which sends the block signal to the busrelay. Via the use of relay logic equations, trippingby this element would be conditioned on lack ofreceipt of a blocking signal from the feeder relay. Adelay of about three cycles should be implementedto allow the blocking signal from the feeder relay tobe "recognized" by the bus relay. This way, a nearinstantaneous tripping is achieved eliminating theneed for an expensive differential protection for thefeeder bus.

This is an example of a scheme requiring thetransfer of protection related data between differentdevices.

This scheme allows for the use of the instantaneousovercurrent unit at the low side of the transformerfeeding the busbar, to provide a fast trip in case of afault in the busbar. To coordinate with theinstantaneous units at the feeders, any pick-up signalfrom the feeder relays must be sent to thetransformer relay to block its instantaneous unit. Seefigure 4, representing a fault in a feeder.

Figure 4. Coordination of instantaneous units

A similar scheme could be implemented betweenthe transformer and the bus protection relays tospeed up tripping for transformer low voltage busfaults (ahead of the feeder bus zone).

One application for fast clearing of bus faultsinvolves sequence protection. In a multipletransformer station with a closed bus tie breaker, bydetecting which side of a bus tie breaker is faultedallows for tripping out only the affected transformerthat the fault is on, as opposed to tripping out bothtransformers.

2.5 Applying negative sequence overcurrentprotection.

The availability of negative sequence overcurrentelements in most modern microprocessor relaysmakes it possible to take advantage of theirinsensitivity to balanced load to achieve fasterclearing of bus phase-phase (and phase-phase-ground with high ground resistance) faults. In orderto achieve this, negative sequence relay settingsshould be applied to both the feeder and the feederbus relays.

On the feeder relay, the negative sequenceovercurrent element would be set to coordinate withthe downstream fuse or recloser without takingbalanced feeder load into consideration Thenegative sequence element setting thus achievedwill be more sensitive than the feeder phaseovercurrent element.

On the feeder bus relay, the negative sequenceelement can be set simply to coordinate with thenegative sequence element on the feeder relay, againwithout consideration for bus balanced loads. Thesettings thus achieved will be much more sensitivethan the bus phase overcurrent element.

One should remember that the negative sequenceelements do not respond to balanced three-phasefaults.

More information on the use of negative sequenceelements and how they could be set to coordinatewith downstream phase elements can be found inreference [3].

2.6 Applying a fuse saving scheme (allowinginstantaneous tripping on first reclose cycle).

Figure 4.

Feeder-1.

5051

5051

5051

50

Feeder- 2. Feeder- 3

Trip

Block50

FaultPoint

6

Where faults tend to be temporary in nature out onthe distribution feeder, quick trip and reclose cyclesmay allow for the de-energization of the faultwithout blowing the feeder lateral fuses. Thisscheme has a few undesired results. It makes it hardto find the fault location if it tends to happen overand over again; all of the digital clocks in thecustomers houses go into the blink mode, whichresult in customer complaints; adds to wear and tearon the breaker for numerous operations. However,

where the feeder relays have fault information, thedata may aid in the location of the event.2.7 Relay logic to allow upstream relay to tripcorrect feeder in case a fault occurs at the same timea feeder relay fails.

In a typical distribution transformer low sideprotective relay scheme, one scheme may use thelow side relay as fast bus protection in combinationwith a feeder relay failure back up scheme.

Figure 5 –Bus Protection Backup For Feeder Protection

Failure

A scheme as depicted in Figure 5 – Bus ProtectionBackup Tripping For Feeder Protection Failure,provides fast selective backup tripping coverage fora feeder relay failure (Note: this is not for a circuitbreaker failure condition).

As shown in the figure, relay alarm fail contacts arerouted to the bus relay, 51B. The purpose of this isto enable alternate feeder protection settings on thebus relay and disable fast bus protection tripping.The bus relay in turn, will direct trip commands toeach feeder breaker via its own feeder relay alarmcontact. In Figure 5, CB1 will be tripped via its51F1 alarm contact (relay failed) whereas CB2 willnot be tripped as the 51F2 alarm contact is open(relay healthy).

The concept is that under normal circumstances thelow side relay is active as fast bus protection, andeach feeder has it's own individual protective relayto clear faults out on each respective feeder. If afault occurs on a feeder, the feeder clears the faultand simultaneously the faulted feeder relay, viahardwired contacts to the low side protective relay,disables the fast bus elements to prevent fasttripping of the bus for a feeder fault.

If a feeder relay fails, it needs to disable the fast busto prevent tripping the bus for faults on a feeder andto set a second level overcurrent element on the lowside relay and backup the failed feeder relay withthe low side relay. This would allow tripping thefeeder with the failed relay through the alarmcontact. The feeders with the healthy relays haveopen alarm contacts and unable to trip the feederbreakers.

2.8 Use of broken conductor protection toalarm or trip for downed distribution conductors forhigh impedance faults.

One of the most difficult types of faults to detect ona distribution system are those involving downedconductors or high impedance faults.

The wire may be simply broken and hanging in midair. The line may be intact but making contact witha tree branch and thus be a high impedance fault.The line may be broken and laying on the ground oron top of asphalt and is thus a high impedance faultto ground. In this case, there is a very small level ofcurrent involved as compared to the overall loadcurrent on the feeder. All of these types ofconditions are virtually impossible to detect viaconventional overcurrent relaying. These aredangerous operating conditions and the public mustbe protected from harm.

There are several methods in use to detect downedconductors and high impedance faults with variouslevels of success. For further details see references[4], [5], [6].

Methodologies to detect downed conductors or highimpedance faults include the use of voltagedetection, current based systems, and complexsystems that make use of pattern recognition andartificial intelligence.

In all cases, the individual utility will have todetermine whether to alarm or trip when theequipment detects a problem. If the option is to trip,then any false trip could put more people at risk than

CB1

CB2

51F1

51B

51F2

RELAY FAIL

RELAY FAILALARM

CONTACTCLOSED

RELAY OKAYALARM

CONTACTOPEN

RELAY FAIL

7

if nothing was done; i.e. traffic lights madeinoperative, home medical equipment de-energized.If the option is to not trip and alarm only, then thereis a risk that the public may come in contact with theline before a crew has arrived at site to investigate.

A voltage based system is basically made up ofundervoltage relays installed at multiple locationson a distribution line. Tripping could occur closerto the faulted location by the use of power circuitreclosers or at the substation by a circuit breaker. Inall cases, the fault interrupting device would need tohave a communication link from the undervoltagerelays. The communication link could be power linecarrier, fiber, cellular, or even radio. The mainadvantage of this type system is that the exactlocation of the problem is known with fairly goodaccuracy due to the use of a large number of thedevices on the feeder. To gain the full benefit ofthis type system, undervoltage relays wouldprobably need to be installed at the end of everylateral and at several locations on the main line ofthe feeder. The great number is also the maindisadvantage; mainly due to the large expense.

There are also current based systems available. Onetechnique is to use the third harmonic current todetect a problem. Obviously there could be settingproblems as the setting would have to be set toignore normal third harmonic current and still besensitive enough. This type scheme would probablybe employed at the substation and thus thedetermination of the fault location is more difficult.

A third available scheme makes use of patternrecognition and artificial intelligence. This currentbased type system is designed to be able to make adistinction as to whether arcing is going on thusindicating that the line may be intact but makingcontact with something like a tree. The relay mustalso determine what the background load current isand if there is a sudden increase or decrease, decideif there is a broken phase or downed conductor.Again, this type scheme would probably beemployed at the substation and thus thedetermination of the fault location is more difficult.

2.9 Restricted earth fault protectionApplying restricted earth fault protection to increasesensitivity and reduce clearing times for transformerground faults could limit transformer damage.Limiting transformer damage could result in fasterrestoration times.

2.10 Cold Load Pickup

One method to eliminate re-tripping when restoringheavily load feeders after an outage could beaccomplished by using a cold load pickup feature.In the older electromechanical schemes, this mostlikely had to be accomplished using mechanicalstepper relays in conjunction with the reclosingrelay and instantaneous elements for phase orground tripping. In the newer microprocessor relayschemes, this most likely can be accomplishedwithin the functionality of the protection relay whenneeded.

Note that this feature may be only a part of thedistribution setting reclosing philosophy. Thisreclosing philosophy may roll up several featuressuch as how many recloses are allowed; how manywithin a certain time window; how fast the breakeris closed back in; whether fast tripping may bedelayed in order for feeder laterial fuses to blow,etc.

2.11 Reduce unnecessary momentaries.

Two instantaneous settings using microprocessorrelays-

The two settings consist of one low set with a 6cycle delay that allows the first reclosing operationand one high set that blocks reclosing for theunderground cable getaway.

The goal of the low set time delayed response is toreduce unnecessary momentaries for close in faultsbeyond fused taps with the use of a the time delayand at the same time provide a 6 cycle fast responseto transient short circuits. For remote transient mainline faults the low set delayed response would stilloperate much faster then the retarded time currentresponse (cycles vs. seconds).

The following setting philosophy is used for thehigh and low set instantaneous settings.

1) The high set instantaneous overcurrent settingwill be set with no time delay and set to the phaseand ground value at the 13.8kV overhead riser pole.The high set instantaneous would block reclosingfor bolted underground faults.2) The low set will be set with a 6 cycle time delayfor the following reasons.

a) Reduce unnecessary momentary operations forhigh magnitude faults by permitting branch fuses toblow before a momentary breaker operation canoccur. Note: fuse saving can not be accomplishedby a breaker for high magnitude faults due to the

8

inherent breaker operating time delay (typically 3 -6 cycles).

b) Reduce the unnecessary operation of the 13.8kVsupply breaker due to inrush current balanced orunbalanced as a result of street tie switching. Note:the relays will also be set such that, when switchingof the 13.8kV breaker will result in a time delaybefore the low set instantaneous protection isenabled. This will further prevent risk ofinadvertent operation of the 13.8kV source breakerdue to inrush current or cold load current.

c) The low set phase instantaneous will be set not tooperate for a three phase bolted fault on the low sideof a pad mounted transformer.

d) The low set phase and ground instantaneousovercurrent relay will be set not to reach throughany pole mounted reclosers. Note: typically the polemounted recloser will already provide fast curveprotection. Coordinating the low set instantaneoussettings with the recloser will help maintainselectivity and limit unnecessary momentaryoperations to a smaller portion of the circuit.

e) The low set phase and ground instantaneousovercurrent relay will be set not to reach throughany 13.8Kv customer stations.

f) Set the low set phase and ground instantaneousovercurrent relays only to be activated once duringeach trip to reclosing to lockout cycle. One fast twoslow operations.

Zone Sequencing to reduce momentaries:

What is Zone Sequencing? Zone Sequencing is theuse of electronic relay logic to eliminate feederbreaker momentaries for faults that are locatedbeyond distribution field reclosing devices. Theobjective of Zone Sequencing is to reduce thenumber of feeder momentaries that occur due tofaults beyond field reclosers. Pilot tests shows thatZone Sequencing is effective in reducing feedermomentaries under certain conditions

What conditions are required for Zone Sequencing?Zone Sequencing can be implemented on feedersthat have microprocessor relays with certain logic, afeeder protection scheme that uses a GroundInstantaneous (GI), a Short Time Ground (GX) anda Long Time Ground (GT) elements and certaintypes of Distribution field recloser(s). Havinghighly accurate and dependable reclose times areessential to the way this particular scheme works.

How does the Zone Sequencing Logic work? Whena fault occurs, the GX element will be removedfrom the relay coordination scheme at the end of thefirst reclose sequence, if it has not tripped within 1.5seconds.

Therefore, if the fault is beyond a field recloser,which has a reclosing time just longer than this, theGX element will be removed from the scheme whilethe recloser is in the open position.

In summary, for all faults beyond a distribution fieldrecloser, the GX element will be removed from thecoordination scheme, eliminating unnecessaryfeeder momentaries associated with the operation ofthe GX element.

What are the benefits of Zone Sequencing?The benefits range from Reducing feeder breakermomentaries (MAIFI), Improve customersatisfaction and Improve longevity of the feederbreaker. It also allows the GX and GI relay to beenabled to protect the Feeder Main section withouttripping the breaker due to faults beyond a fieldrecloser.

2.12 Transformer Thermal Overload ProtectionThermal overload protection can be used to avoiddamaging the transformer due to excessive loading.

Simple overload protection is based upon theamount of load that a transformer can withstandwithout causing unacceptable damage. Alloverloads result in a temperature increase that willcause degradation of the transformer insulationcomponents (oil, fiber boards, paper, etc.), which inturn increases the susceptibility for insulationfailure. The question becomes one of risk andeconomics – overload the transformer to a givenlimit and accept less life, dramatically increase thecooling capability if economically possible, or put ina larger transformer. Sometimes the answer may beto accept the risk and run the transformer to failure.

For the most part, the manufacturer has curves thatestimate the amount of loss in life for a giventemperature rise experienced by the transformer, aswell as heat rise curves for given currents.Conventional alarms are usually set for a settemperature of the oil and of the hot spot, but this isnot predictive but rather reactive to the temperaturethat the transformer has already succumbed to.

If the IED provides the adaptive characteristic ofbeing able to evaluate the changing data and predictthe amount of time before the transformer had to

9

have load reduced, some evasive action might beable to be taken. This could be either manual orautomatic reaction to data or alarms.Further, if load had to be reduced at the station orfeeder level, then preprogrammed actions could betaken. This could be performed at either the stationlevel or the SCADA master level, depending uponwhere the computer resided.

For example, if all of the data was concentrated atthe station level with the appropriate scenario logicin place, then the substation computer could takepreprogrammed actions of closing bus tie breakers,communicating with smart field reclosers, or at lastresort, dropping feeder breaker load by openingpreselected feeder breakers. Obviously, this wouldneed to be a well thought out plan with dynamiclogic to account for varying loads, transformertemperatures, and other variables. These “selectedrolling blackouts” could make use of the studiesinvolved in selecting underfrequency load sheddingsettings in determining loads and essential customerfeeders.

2.13 IEDs can provide logical inputs/outputs

Using internal logical input/outputs available forcontrol and autorestoring can eliminate auxiliaryrelays and the potential failure mode that these havepresented over the years.

Wiring from one relay to another has alwayspresented various restrictions or concerns, not tomention the added costs of wiring. Multiple drycontacts might be required to be wired betweenrelays and panels, which might be located in closeor not-so-close proximity to each other. The use ofauxiliary relays has historically had somemisoperations attributed to various types of failuresof the auxiliary relay itself with very little that couldbe done in terms of early warning of a problem. Onthe other hand, making use of a single input mightbe mapped to several outputs, which in turn couldbe wired or communicated to adjacent or peer IEDs.Some self-checking diagnostics of I/O hardwaremay be available. This further allows timestampingof the events within the Sequence of Events (SOE)function of the IED. This provides a detailed eventlog that is especially helpful in troubleshootingmisoperations.

2.14 Impending FailuresOther functions being implemented may be valuablein preventing faults or problems before theymanifest themselves in failures.

These detections could in theory prevent the outagefrom occurring in the first place. Detections ofspecific identifiable signatures could also aid in postanalysis of what initiated the failures.

Examples that come to mind include loose or noisyconnections, fuse failures, impending arrestor orinsulator failures, capacitor failures, sporadicforeign interference identification. Incipient splicefailure detection in underground cable is alreadyavailable [14], and similar techniques should be ableto be used to detect incipient arrestor failure as wellas incipient capacitor can failures. Bushing failureshave traditionally been predicted through manualtesting. Field tests are currently underway toprovide in-service predictive impendingconditions.[15] [16]

2.15 Partial Discharge Monitoring InTransformers

Partial discharges in transformers are smallelectrical arcs between phase-to-ground or phase-to-phase. In transformers, partial discharges aremainly caused by containments in the oil or partialfailures of the solid insulation around the windings.Contaminants in the oil are indicative of amechanical problem (e.g. pump introducing metalslivers into the oil) or a failure of the sealed tankthat allows air, water, or other contaminants into theoil. Partial failures of the winding insulation areindicative of an incipient fault. In either case,detection of these conditions through monitoring ofpartial discharges can allow early interdictionthrough maintenance to avoid a more severe failurecausing a customer outage.

Several devices are currently available to monitorpartial discharges on in-service transformers. Ingeneral, the principles employed by these devicesfall into two categories: monitoring high frequencyelectrical signals and monitoring acoustical signals.The devices look for signature signals of partialdischarge activity using various algorithms andfiltering techniques. The monitors can bepermanently installed or portable and used as part ofa regular transformer inspection routine.

2.16 Arc-Flash MitigationOne of the concerns facing the field personnel whileworking on the facilities is if an arc occurs shouldan accidential contact take place. One remedy inplace for many years was to put the feeder on “one-time” where the automatic recloser switch getsturned off for the duration of the work on the line.However, this only stops the feeder from reclosing.Of course the drawback to this action is that the

10

customers go in the dark for all faults and do notautomatically reclose as they would normally do.Another technique is to enable additionalinstantaneous relay elements when the recloser isturned off, typically the phase elements. Thesewould be enabled only while the work is in progressas a unwanted trip could also occur as well.Microprocessor relays and their programmingabilities have made this easier to do. For furtherinformation, refer to the IEEE-PSRC WG-K9 reporton this subject [ 7 ], expected to be released in 2009.

3.0 Faster DetectionOne way that the secure operation of themicroprocessor relay can be enhanced is throughhigher sampling rates. When less samples are madeavailable to an algorithm, the process may requiremore blocks of data before it can reach a secureconclusion. This may also manifest itself in fasteroperate times than devices with substantially lesssamples of data. This also allows for actual ratherthan curve fitted oscillographic records. Too low ofa sampling rate can actually miss a fast event.

4.0 Faster Remediation through IEDcommunication. How communication affectsprotection.

4.1: IED Communication

IED communication still sounds like black magic toone that comes from the electromechanical world ofhard-wired contacts and operate coils. Messages getsent over fiber optics or other communication pathswhich get decoded into logic commands. In somecases, data gathering may occur over one network,and control communications over another fasternetwork. Redundancy is another use of this functionto back up traditional protection paths. [8]

For example, one such system would be the use ofcommunications to “tell” the upstream transformerprotection that a feeder breaker has failed to trip fora fault and to go ahead and initiate a trip withoutwaiting for what typically is TOC relays to pickup.This removes the fault from the system morequickly, resulting in less local damage. This alsomay include the identification of the stuck breakerfor the breaker failure action. See section 2.3

Some forms of peer-to-peer communication isavailable that can pass data between devices ordatabases. Peer-to-peer communications areimplemented to continuously broadcast on thenetwork shared information such as counters,voltage or current magnitudes, angles, or other highresolution data between devices. Each device on the

network shares its own data with every other devicesuch as relays, switches, breakers, RTUs, metering,load tap-changers, voltage regulators, etc. Thespeed of communications is dependent on the typeof communication medium used (fiber optic cables,twisted copper wires, etc.) and the data sharedbetween all devices.

Traditionally, protective relays and recloser controldevices in distribution systems were appliedwithout the use of communications between devices.Coordination between upstream and downstreamdevices is attempted by implementing shorter timedelays on the downstream protective devices.

Communications enhances protection andautomation in distribution systems and provideshigh speed substation bus reconfiguration andautomatic load shedding to prevent transformeroverload tripping. Additional benefits such asflexible and improved automation schemes; flexiblecontrol and interlocking schemes; selective andflexible automatic backup schemes offer thepossibility to reduce power outages and permitcustomized autorestoration schemes in distributionsubstations.

Without communication, only a limited number ofTOC devices can be coordinated in series beforeclearing time of the longest-set device becomesobjectionably long. Using intelligent relays andcommunication between devices to provide trippingsupervision, clearing times can be much faster.

Section 2.4 describes a fast bus protection schemebased on communications between the feeder andbus relays.

Within a small geographic area, such as within asubstation between feeder relays and the transformersecondary main circuit breaker, this scheme may beimplemented with hard-wired logic by connectingoutputs from one relay to inputs on another. IED’scould be interconnected with various networkconfigurations.

Over a more dispersed area, relay to relaycommunication may be accomplished usingdedicated fiber-optic transceivers or multiplexers toroute messages over a fiber network. Relays thatsupport IEC-61850 can send and receiveGOOSE/GSSE (Generic Object Oriented SubstationEvent-Generic Substation Event) messages over aLAN (Local Area Network) connecting the variousrelays together [12], [13]. Use of a communicationsnetwork allows the greatest flexibility for futurecircuit reconfiguration.

11

As additional series interrupter devices are added, itis simply a matter of connecting the relaycommunications to the network and configuring themultiplexers or addressing to route messages to andfrom the appropriate relays.

As an illustration of the possibilities offered by thecommunication between devices, here are someexamples of the logic schemes and automaticfunctions that can be implemented.

4.2. Voltage restoration in “H” type substations

The function described below makes use of thesubstation communication system to interchange therequired data and commands between the involveddevices.

In substations of the type represented in figure 6,this function allows for an automatic transfer of theHV line feeding the substation in accordance to thedesired (programmable) operation conditions andthe availability of voltage in the feeding lines.

The function has four different set status: MANUAL(function disabled), Preference A, Preference B orPreference BOTH.

If both voltages VLA & VLB exist, the breakers52LA, 52LB and 52LAB are arranged to feed thestation according to the set preference.

If one of the voltages is missing, the arrangement ischanged so that both MV busbars stay energized.

When the voltage returns, the preferentialarrangement is re-established.

If both voltages are missing, no action is taken.

Due to the fact that only one phase voltage isavailable from the HV lines, this logic iscomplemented by taking into account the voltagemeasured in the three phases of each MV busbar.

A faulted transformer function is also included inthis scheme. In case of a trip of, say, transformer A,if 52LAB is open and there is voltage in Line B, theautomatic function closes breaker 52LB and setsitself to MANUAL.

4.3. Automatic service restoration

The objective of this automatic function is to act asa trained operator would do when the busbar voltagedrops to zero.

The logic performed is described below. See figure7 as a reference. In this figure, the arrows pointingtowards the busbar indicate sources with capabilityto restore the service at the bus.

If VB = 0 , then open all breakersWait until any of the voltages (from a

source with restoration capability) appears. Thenclose the breaker and check if busbar voltage VBstays.

If VB goes to 0 again, open the breaker (ifstill closed), indicate fault in busbar, and stop.

If VB stays OK, then continue therestoration process according to a pre-programmedsequence, always checking VB after closing abreaker.

If, for example, when closing L4 breaker,VB goes to zero, open all breakers, indicate fault inthe line (i.e. L4), and start all over again (withoutincluding L4 in the new process).

4.4. Other schemes

Figure 7– Automatic service restoration.

VB

VT1 VT2 VL1

VL 2 VL4VL 3 VL 5 VL 6 VL 7 VL n

Figure 6 – “H” type substation.

VLA VLB

52 L 52 L

52 LA

52 T52 B1

52 T

VA ab

VA bc

VA Ca

VB ab

VB bc

VB ca

12

Other logic schemes implemented in the substationsthat make use of the substation communicationsystem are:

Capacitor bank automatic connection /disconnection, by measuring reactive power flowand/or following a pre-scheduled calendar.

Load restoration schemes(after underfrequency load shedding).

5.0 How Communication affects restorationand analysis (i.e. oscillographic and eventrecords)

5.1 In past years, data retrieval was confined tofault records from devices such as Digital FaultRecorders (DFRs). However, today’s IEDmicroprocessor relays have the capacity foroscillographic records, event records, and settingrecords that can be accessed from the relay to acomputer.

This access may be local or remote from differentports on the relay. The availability of remote accessallows for record retrieval and analysis by one groupof people, while another is in route to the substationfor investigation and restoration. This also allowsfor the remote analysis to determine if devices in thestation operated properly or if a possiblemisoperation occurred. In many cases the recordsout of distribution protection relays are the only datarecords that are available to analyze a fault andwhether everything operated correctly. Access maybe via the dial-up phone line method, or some typeof Wide Area Network (WAN) or Intra-Net.

In the case of dial-up, rather than have a dedicatedphone line connected to each IED, a more typicalapplication may involve some type of phone linesharing device or port switcher. Dial up lines canpresent remote access security issues should ahacker get into the IED.

In the case of the WAN or Intra-Net, then a moretypical connection may involve a port switcher orrouter with an Ethernet type of setup.

5.2 Remote access security

Local password access tends to limit what could beviewed by all, but changes could be confined to aselect group of authorized users utilizing higherlevel passwords. This was typically implementedfor front panel button access, or possibly through aportable PC connected to a serial port on the relay.In some cases, the concern here was more associatedwith someone making an error, more so than

someone going in and purposely making changesthat would cause problems.

Remote access introduced the problems associatedwith hacker security.

Security is usually addressed through the use ofpasswords, and generally, there may be a passwordfor the first level of connecting to the relay, and asecond or third level to retrieve or make changes insettings. Some systems even involve an automaticdial-back system where the interrogator may have tobe preauthorized and preset in the system so thatonly previously identified phone numbers areallowed access. However, in the more typicalsubstation environment, this is not generally used oreven available.

Recent events have forced utilities to go to the useof what is referred to as “strong passwords” in orderto thwart hackers from getting into the devices andcausing problems. Strong passwords typicallycontain several letters, upper and lower case, severalnumbers, and even some characters. Exceeding anumber of attempts to log in can generally cause therelay to disconnect and lock out from thecommunication. This prevents automated hacking ifso equipped. Security may be generally undertakenat both the corporate level as well as the substationlevel for the network communications, as well asusing hardware and/or software firewalls.

6.0 Data Utilization for automated and postanalysis of faults and events.

6.1 Distance to fault can be automaticallycalculated and displayed. [9], [10]

Many microprocessor feeder relays include a featurecalled ‘distance to fault’, which provides anindication of the location of the fault relative to therelay. This is accomplished using an algorithm withthe measured current and voltage quantities duringthe fault. The distance to fault value is thenavailable for use as part of the fault record and as anoutput of the relay in the form of digital or analogdata. This data can then be passed to the SCADAsystem where the system operator can initiatesectionalizing or direct work crews to the closestupstream switch. This tends to be more accurate ina point-to-point primary system as opposed to atypical distribution feeder type of arrangementcontaining multiple taps and laterals.

Another, less sophisticated method of implementinga distance to fault function in devises that don’t have

13

an internal algorithm or voltage elements is to usethe magnitude of the fault current. The concept isthat the closer the fault is to the source (substation)the higher the magnitude of fault current. Byselecting ranges of current and associating themwith the line length, the operator can get anapproximate location of the fault. The biggestshortcoming of this method is that it cannot accountfor fault impedance. A close-in high impedancefault will have a similar magnitude of fault currentas a distant low impedance fault.

By using this data, it is possible to reduce the outagetime by locating the fault more quickly.

Both of the above techniques have shortcomingswhen the feeder circuit branches or has a source ortied to another hot feeder. In the case of a branch,the relay cannot tell whether the fault is on the maincircuit or the branch. In the case of a source, thefault current provided by the source is not seen oraccounted for by the relay and therefore the distanceto fault calculation is less accurate.

6.2 Automatic Load Transfer on Loss of aSource

For the case of a split distribution bus with a bus tiebreaker that is normally operated in the openposition, an automatic load transfer scheme can beimplemented (refer to Figure 8). Such a schemewould be applicable if the station planning criteriaallows the total station load to be carried by onetransformer. The automatic closing system monitorspre-fault total MVA load for both transformers anddetermines the average MVA load before theinitiating event if the planning criteria does notallow the total station load to be carried by onetransformer.. For conditions that cause the loss ofone of the sources (high side bus faults, high sidebreaker failure, or transformer faults), the bus tiebreaker (52E) can be closed automatically to supplythe other feeders. Initiation of automatic closing isblocked if loading on the remaining transformerwould exceed the published Transformer Ratingsafter Auto-Close.

Automatic closing for the bus tie breaker is typicallycontrolled by a permissive switch and/or bySCADA. The SCADA system also has continuousstatus of the automatic closing scheme. Automaticclosing is supervised by one or more of thefollowing functions

Bus tie breaker is open.The low side breaker associated with the

deenergized transformer is open.The low side breaker associated with the

remaining transformer is closed.Total transformer load was below a

predetermined level for a predetermined time priorto the automatic close initiating event.

The energizing bus is alive.

If required the above functions may be augmentedby a sync-check/voltage-check function..

An enhancement to the automatic closing scheme isto incorporate different seasonal load limits. Forexample, different summer and winter load limitscan be used. Selecting which limit is in effect isaccomplished by a mode switch and/or by SCADA.

6.3 Selective Automatic Sectionalizing

A selective automatic sectionalizing scheme can beimplemented on a configuration where one breakerfeeds two or more individual circuits (sections)through motor operated switches. The protection forall of the circuits is provided by the breaker andassociated relaying. The individual circuits eachhave fault detectors to determine which circuit thefault occurred on.

A typical automatic sectionalizing scheme goesthrough a series of reclosing attempts bysequentially isolating part of the circuit. Thisprocess can take from several tens of seconds toseveral minutes, depending on the number ofsections, the number of reclosing attempts for eachconfiguration, and the reclosing interval.

Figure8 – Transformer load function.

T2

52 B52 E

I2

I1

T1

52A

14

A selective automatic sectionalizing scheme usesthe fault detectors on the individual sections toknow where the fault occurred (See Figure 9 below).The scheme can reduce the number of reclosingattempts and the outage time of the non faultedcircuits by opening the switch of the faulted circuitafter a minimum number of reclosing attempts.

Since this logic is more complicated than typicalreclosing, a programmable device such as a smallPLC is often used to implement the scheme ratherthan an off the shelf reclosing relay.

6.4 Automatic Ring Bus Reclosing

For ring bus configurations, an automatic ringreclosing scheme can be employed to close the ringafter a fault has occurred, restoring the reliability ofthe ring configuration. When a fault occurs, bothbreakers associated with the faulted circuit open.One of the breakers is the primary reclosing breaker,which attempts the reclosing sequence. If the firstbreaker successfully recloses, the second breakerthen automatically recloses to complete the ring bus.

An enhancement to this scheme utilizes a motoroperated disconnect switch on the circuit. If theprimary reclosing sequence is unsuccessful, themotor operated switch is opened and one morereclosing shot is attempted. If the final reclosingattempt is successful, the secondary breaker thenrestores the ring.

Another enhancement to this scheme allows thesecondary breaker to assume the primary reclosingsequence when the primary breaker is out of service(e.g. for maintenance). A contact from the primary

breaker controls (e.g. a maintenance switch in thebreaker or the reclosing cutoff switch) automaticallyapplies the primary reclosing sequence to thesecondary breaker. Since the secondary breaker mayonly have live bus/live line/sync check as its normalreclosing permission, it will not attempt to reclose ifthe primary breaker did not successfully reclosefirst, leaving the faulted circuit out of service. Byautomatically applying the primary reclosingsequence to the secondary breaker, the secondarybreaker can automatically attempt to restore thecircuit.

Similar schemes can be used for breaker and a halfbus configurations. [11]

6.5 Line SectionalizerAutomatic line sectionalizers have long been used inconjunction with reclosers, or reclosing relayscontrolling circuit breakers. The sectionalizercontrol has a fault current detector that arms acounter, and the counter registers a count when thefault is interrupted. After the set number of counts,the sectionalizer opens to isolate the faulted sectionof the circuit. Opening of the sectionalizer occursduring the circuit dead time between recloseattempts by the source side circuit breaker. Thesectionalizer must be set to trip one or more countsbefore the source-side recloser goes to lockout.

Enhancement of the basic recloser / sectionalizerscheme is achieved by applying intelligent controlsand communication to the motor-operated switchesused for sectionalizing (Refer to Figure 10). Bycommunicating switch status, fault detection status,and circuit loading information among themselves,the sectionalizing switches can determine the faultlocation. Faults can be isolated with fewer reclosingshots by the circuit breaker. After the fault isisolated and the fault location determined, theappropriate switches automatically reclose to re-energize the unfaulted line sections. Applying anintelligent switch to the normally-open of a loopedfeeder, or as a connection to a feeder from anothersubstation, allows line sections beyond the faultedsection to be re-energized from the alternate source.

Figure 9–

Selective Sectionalizing Scheme Using Fault Detectors.

Circuit 1

Circuit 2

Circuit 3

Source

M

M

M

52

51 FD

51 FD

51 FD

51FD

15

6.6 Relay Self Diagnostics

All modern microprocessor relays have built in selfdiagnostic logic for self monitoring the health of therelay. Depending on the sophistication of this logic,the relay can monitor most of its internal functions[1, 2]. By combining this self monitoring with otherinternal logic, such as loss of potential and/or loss ofcurrent detection, the user can have a very goodindicator of the health of the relay. A combinationof relay alarm and output contacts can be wired intoannunciator points and passed to SCADA for earlyindication of a relay problem. This early warningcan help reduce the possibility of a relaymisoperation or failure to operate.

6.7 Oscillography and SOE Capabilities ofMicroprocessor-based Relays

6.7.1. Oscillography Recording

As a standard feature, digital relays support somelevel of oscillographic recording. The number ofrecords, record length, sample rate, etc. varies fromrelay to relay. Relay records can aidtroubleshooting and analysis of system events to agreat degree if configured properly. They canprovide additional coverage as compared withdedicated DFRs, which typically may not haveenough points to adequately monitor the distributionequipment and logic levels. The extended coveragemay go as far as having the same secondary signalsrecorded by more than one device. This allowsindependent confirmation of one record versus theothers, and troubleshooting problems internal to therelays, secondary circuits, VTs and CTs.

Typical characteristics and options related todisturbance recording of digital relays are describedin Appendix-B. This covers such things as analogfiltering for ac signals, digital filtering for ac inputs,

de-bounce filtering for dc inputs, sampling rate,programmability, available memory, etc.

6.7.2 Protection-specific recording

Various protection-specific signals could berecorded by protection relays. Examples arecommunications-based digital inputs such as IEC-61850 GOOSE/GSSE or proprietary digitallytransmitted teleprotection signals; derivatives ofinput currents and voltages such as magnitudes,angles, symmetrical components, differentialcurrents; or internal flags created by user-programmable logic or protection and controlelements.

Typically, the above signals are inputs to protectionand control elements of a relay and therefore arevery valuable for analysis. This is because theyreflect the real measurements performed by the relayitself, after the magnetics and usually after thefilters. For example, when extracting a negative-sequence current from a DFR record in software,one uses a generic approach being typically a full-cycle Fourier filtering. This may not reflect theactual filtering of the relay as manufacturers useproprietary algorithms. Recording the actualprotection signals calculated internally is thus anadvantage over the DFRs when looking at what therelay was acting on. However, what the relayrecords and what the system actually experiencedmay be somewhat different depending upon wherethe waveform was recorded relative to filter andmagnetics effects. Saved sample rates may be lowerthan initial sample rates.

Also, some signals, such as communications-basedI/Os and other logic levels, are not easily recordableby standard DFRs.

6.8 Data and data formats

6.8.1 Data format and extra content

Older relays used proprietary or plain ASCII formatto store the records. New solutions apply theCOMTRADE standard which does not fit theapplication perfectly. For example, digital signals orvariable sampling rate records were not originallyintended for recording in COMTRADE.

In addition, extra content may be added to therecords. For example, the point in time when a faultlocation is performed may be marked by a specialcharacter. The active settings could be stored withinthe record to create a complete picture of relayoperation. Fault location may also be merged with

Figure 10– Intelligent Controls & Communication

Source 1MM M

MM M

IED

Control

IED

Control

IED

Control

IED

Control

IED

ControlIED

Control

Source 2

16

the oscillography to create a single comprehensiverecord.

6.8.2. Sequence of Events Recording

Another feature is Sequence of Events (SOE)recording. DC inputs/outputs, internally generatedpickup, dropout, operate and other flags, digitalpoints in the user-programmable logic,communication-based inputs, user-programmablepushbuttons, and other signals can be programmedto log events. Typically a short string describing theevent is stored with the appropriate time stamp.

6.8.3 Programmability

Modern relays support – on a per-point basis – anenable/disable setting for event recording. A multi-function relay processes hundreds of digital flagsthat potentially could be of interest. The user isgiven a freedom to enable some events only in orderto conserve the memory, avoid showers ofunnecessary event logs, and make future analysiseasier. The application, history of the protectedprimary equipment, known problems with thesecondary equipment including the relay itselfwould dictate an optimum set of enabled events.

6.8.4 Time stamping and resolution

The SOE records have limited value if the relay isnot synchronized. Without time synchronization,the SOE record provides relative sequence of eventswithin its contained “system”. They may besuperimposed on other records manually usingprimary power system events recorded in theoscillography as “synchronizing” or “reference”marks (fault inception, breaker operation, faultclearance, etc.).

Older relay designs and low-end digital relays donot support accurate time synchronization. Modernhigh-end relays support time synchronization viaIRIG-B time signals. Typical time accuracy is in therange of a few microseconds.

Other means of synchronization include Ethernet-based algorithms such as Simple Network TransportProtocol (SNTP), which allows synchronization viaTCP/IP with the accuracy of a few milliseconds.

Accuracy of time-stamping should not be confusedwith the resolution. Care must be taken whenanalyzing SOE records created by relays:

DC input signals are typically monitored by aperiodic scan.

DC inputs are subject to “contact de-bouncing”algorithms.

Communication-based inputs, such as IEC-61850 GOOSE/GSSE or proprietary digitalteleprotection signals, may be time-stampedwhen the digital packet starts arriving, when itarrived, or when it got decoded, validated andused for the first time.

Operations of output contacts are typicallytime-stamped when the digital part of the relaysends the command to close or open a givencontact.

Internal flags in the relay such as outputs fromprotection or control elements are asserted incourse of serial calculations. Some flags areasserted at the beginning of the process, whilesome at the end of this process.

Further details on this subject can be seen in theappendix B6.

6.8.5 Formats and memory limitations

There is no single standard format for SOE recordsin the domain of protective relaying. Single eventscan be reported and retrieved via well-definedmechanisms of known SCADA protocols. SOE filesare typically recorded in freely selected ASCIIformat.

Modern relays support thousands of SOE entries,limited only by the size of the memoryconfiguration. Automated analysis softwareprograms would be the next logical progression forfuture development.

6.9 Transformer and Breaker ReliabilityMonitoring Capabilities of Microprocessor-basedRelays

6.9.1 Circuit Breakers

Dedicated breaker monitoring stations allow directand comprehensive monitoring of health and wearof a breaker at the expense of increased overallinstallation cost. Basic breaker monitoringfunctions, however, are possible utilizing signalsalready wired to microprocessor-based relays.Modern relays allow for monitoring of selectedbreaker wear indices and stress factors at no, or lowextra cost.

Breaker arcing current, or accumulated duty isdefined and measured as an integral of the squaredcurrent waveform over the period of time between

17

the poles starting to depart and completeinterruption of the current. This integral isproportional to energy dissipated within the breakerand therefore reflects stress imposed on the breakerduring operation. The per-pole I2T values may bereported on a per operation basis and/oraccumulated by the relay to indicate the total wearor remaining “life” of the monitored breaker.

Breaker time is defined and measured as timeelapsed between the trip signal and the breaker maincontacts coming open. The pole opening is detectedbased on the ac current going to zero. The operatingtime is measured by modern relays with reasonableaccuracy and reported on a per-pole basis, or for thelast pole that opened.

Fault current is typically reported by “fault report”or “fault location” features of digital relays.

Discrepancy between auxiliary contacts and breakercurrent can be measured with some accuracy bycomparing a dropout time of a built-in overcurrentfunction with a time of the auxiliary contacts. Thismonitoring function allows detecting severemechanical problems or issues with the auxiliarycontacts.

Breaker flashover monitoring functions are availableto monitor opened breakers. Corrective actionscould be performed for intermittent flashover thatotherwise will remain undetected.

Counters could be set to count breaker operations.Both the number of operations and the rate-of-change can be polled from the relay in order toestimate breaker wear and pinpoint problems withthe primary equipment.

6.9.2 Transformers

Transformer nursing stations – due to their relativelyhigh cost – may be justified for some transformersonly. Simple transformer monitoring functions areavailable in microprocessor-based relays at no, orlow extra cost.

Hot spot temperature estimation modelsThese are available for safe overload oftransformers. These models are well established anduse mathematical approximation of the heating /cooling processes in order to estimate temperatureof the hottest spot being the major stress factor forthe insulation. These models use current or powermeasurements, one or more direct temperaturemeasurements (ambient, top and bottom oil, etc.),

status of the fans (running or not) and transformerdata to calculate the hot spot in real time. Thecalculated value can be used to alarm, control thefans, shed the load, or even trip the transformer.

Transformer thermal modelsThese are often available to estimate remaining lifeof the protected transformer. Based on the well-established correlation between the temperature anddegradation of the insulating materials, a relative“used” or “remaining” life is calculated (years orpercent). Monitoring the value and rate-of-change ofthe remaining transformer life allows identifyingweakest units within a given population oftransformers. This real time picture is especiallyvaluable when some transformers are intentionallyoverloaded to maintain service while “burning out”quicker than other units. An overload-levelingscheme is conceivable to use up transformer life in away harmonized with long-term expansion andretrofit plans.

Through fault counterA counter can be programmed within a transformerrelay to count through faults. Through faults viatheir electro-dynamic flexing and thermal effectstend to degrade transformers. The accumulated,squared through fault currents reflect the totalexposure of a transformer to dynamic forces andthermal effects. Breaker arcing current feature couldbe programmed to measure this.

Tap changer operationsCounters can be set up to independently countoperations of tap changers. Breaker arcing currentfeature can be set to accumulate the total squaredcurrent being switched by the on-load tap changers.

The above simplistic wear measures prove valuablewhen obtained and logged automatically intoenterprise databases via communication protocols,inter-correlated, and subsequently used on largerpopulations of equipment over extended periods oftime.

Using historical data overlapped with analysis ofsubsequent failures and events allows developingexperimental thresholds for relative equipmenthealth. This allows for the prediction of failureswith reasonable accuracy in order to optimize sparesand schedule retrofits, obtain impartial measures ofrelative capabilities of various types and brands ofthe primary equipment, and move toward data-driven maintenance and testing schedules.

18

Off-the-shelf software is available to set up self-learning reliability models for any type ofequipment using arbitrarily selected wear / stressindices. Breaker and transformer monitoring dataproduced by modern microprocessor-based relaysseem to be well suited for such software packages.

Summary:Incipient detection methods may provide a methodto alarm or de-energize a piece of equipment prior toa primary failure.

High speed fault detection can reduce the faultduration time, and thus limit equipment damage.

Autorestoration techniques can reduce the outagetime to the customers significantly in many cases.

A subsequent working group will be formed toaddress the subject of reducing outages insubstations.

19

Bibliography & references

ReferencesIEEE/PSRC Website: http://www.pes-psrc.org/

1. (section 2.0) Refer to the IEEE/PSRCwebsite, “Understanding Microprocessor BasedTechnology Applied To Relaying, WG-I16 Report,2004

2. (section 2.0) IEEE Tutorial Course,Microprocessor Relays and Protection Systems,88EH0269-1-PWR-1987

3. (section 2.5) For more information on theuse of negative sequence elements and how theycould be set to coordinate with downstream phaseelements, see section 6.1.3 of WG D5 "Guide forProtective Relay Applications to DistributionLines". When approved, this Guide will be called:IEEE PC37.230

4. (section 2.8) - Refer to the IEEE PSRC websiteat:

http://www.pes-src.org/d/D15MSW60.html entitled"High Impedance Fault Detection Technology".

5. (section 2.8) Downed Power Lines: WhyThey Can’t Always Be Detected; A publication ofthe IEEE-PES, 2-22-1989

6. (section 2.8) IEEE Tutorial Course,Detection of Downed Conductors On UtilityDistribution Systems, 90EH0310-3-PWR-1989.

7. (section 2.16) Working Group ReportPSRC Webpage for K9: Protection Considerationsto Mitigate Arc-Flash Hazards.

8. (section 4.1) (Ref: J. C. Appleyard, D. A.Myers, and J. K. Niemira, “Innovations in Use ofMicroprocessor Relays and Controls for ImprovedReliability on the Distribution System”, IEEE/PEST&D Conference and Exposition, Oct. 28 to Nov. 2,2001, Atlanta, GA, USA).

9. (section 6.1-6.6) “Assessing theEffectiveness of Self-Tests and Other MonitoringMeans in Protective Relays,” by J.J. Kumm, E.O.Schweitzer, III, and D. Hou, Proceedings of the 21st

Annual Western Protective Relay Conference,Spokane, Washington, October 1994.

10. section 6.1-6.6 “Philosophies for TestingProtective Relays,” by J.J. Kumm, M.S. Weber,E.O. Schweitzer, III, and D. Hou, 48th Annual

Georgia Tech Protective Relay Conference, Atlanta,Georgia, May 1994

11. (section 6.4), (Ref: Nathan Mitchell,“Automated Switches To the Rescue”, Transmission& Distribution World, October 1, 2000.)

12. (section 4.1), IEEE/PSRC Website,Application of Peer-to-Peer Communication forProtective Relaying Report, also as IEEETransactions-Power Delivery, Vol 17, No 2, April2002.

13. (section 4.1), IEEE, C37.115-2003Standard Test Method For Use In Evaluation ofCommunication Between Intelligent ElectronicDevices In An Integrated Substation Protection,Control, and Data Acquisition System.

14. (section 2.13), “Subcycle overcurrentprotection for self clearing faults due to insulationbreakdown” L.Kojovic & C. Williams (CooperPower Systems detecting incipient splice failures)

15. (section 2.14), B.D.Russell, C.A. Benner,A. Sundaram- TAMU; Feeder Interruptions CausedBy Reocurring Faults on Distribution Feeders-Faults You Don’t Know About“- presented atTAMU 2008 Protective Relay Conference

16. (section 2.14), Bruce Pickett, Mirta Signo-Diaz, Alan Baker, Joe Schaefer, John Meinardi-FPL (Florida Power & Light Co), BogdanKasztenny, Ilia Voloh- General Electric- MultilinAlvin Depew, Joseph Wolete- Potomac ElectricPower Co; “Restrike and Breaker Failure Conditionsfor Circuit Breakers Connection Capacitor Banks”TAMU 2008 Protective Relay Conference andGaTech Protective Relay Conference.

……………………………………………

20

Appendix- A: Reliability Indicators

Below are some of the Reliability Indicators thatare being measured throughout the power utilityindustry:……………………………………………………CI: Customers Interrupted…………………………………………………..CMI: Customer Minutes Interrupted…………………………………………………..SAIDI:System Average Interruption Duration IndexThe average interruption duration

The total time without power for the averagecustomer per year, measured in minutes.(Service Unavailability)

= Sum of customer minutes interrupted

Total number of customers served……………………………………………………..CAIDI (Customer Average Interruption DurationIndex): How long it takes to restore power onaverage for the customers interrupted, measured inminutes

.…………………………………………..

Momentaries: number of times service is lost tocustomers, and restored in less than one minute(this time window may vary between companies)…………………………………………………CAIFI (Customer Average Interruption FrequencyIndex): How often the average customer isinterrupted, measured in times per year.

……………………………………………………..

MAIFI:Momentary Average Interruption Frequency Index

The number of momentary interruptionsexperienced by the average customer per year

= Total no. of customer momentary interruptions

Total no. of customers served(Where no. = number)

………………………………………………….SAIFI:System Average Interruption Frequency Index

Frequency of momentary interruptionsHow often the average customer's lights are out,measured in times per year.

= Total number of customer interruptions

Total number of customers served

SAIFI is measured in units of interruptions percustomer. It is usually measured over the course ofa year, and according to IEEE Standard 1366-1998the median value for North American utilities isapproximately 1.10 interruptions per customer.……………………………………………

Following are excerpts from an article thatappeared in Transmission & Distribution World,Dec 1, 2003 , which goes further into this subject:

The Impact of Regulatory Policy on Reliability

Byline: Cheryl Warren, National Grid USA ServiceCo.

The regulatory purview in the United States hasshifted from stranded assets and generation to thepower grid and distribution reliability. Regulatorsare becoming increasingly concerned with everyissue relating to the delivery of reliable power tocustomers.

Each state has the right to mandate distributionreliability standards and targets, creatingpotentially 50 different regulations throughout theUnited States. Some states have no reliabilityregulation at all. Today, regulators may participatein the National Association of Regulatory UtilityCommissioners (NARUC) where they can shareideas, but no requirement exists to adopt the sameapproach on any issue. Looking 20 years down theroad, it is conceivable that a federal standard couldbe enacted as has been done in other countries. TheU.S. Department of Energy (DOE) has expressed adesire to begin regulating distribution reliability atthe federal level. However, it will take years tochange the status quo because most states areunwilling to forgo this right, and who can blamethem?

The most common metrics used by state regulatorsinclude system index calculations SAIFI, SAIDIand CAIDI. To a lesser extent, CAIFI and MAIFIare used. These indices, as well as the factors thataffect them, are defined in the IEEE Guide onElectric Power Reliability Indices 1366-2003. Inshort, they are engineering metrics that trackfrequency and duration of customer and systeminterruptions. These indices are applied on system,circuit and customer levels for planning andregulatory reporting purposes.

The guide also clarifies some of the othersupporting definitions. Some states are adoptingIEEE 1366 as the basis for their regulation toremove definition variability that often makescomparisons difficult.

21

Appendix- B:

Characteristics of Digital Relays

B-1 Relay A/D and Automatic Checking

The analog input component of a digital relayconsists of the relay connections to the current andvoltage circuits, internal isolation transformers,signal conditioning (filters), multiplexers, and A/Dconverters. This component converts the current andvoltage analog signals to digital signals for themicroprocessor to use. The analog input componentis partially checked by the relay self-test feature [1],[2]. Additional verification of correct operationcomes from:

The relay self-monitoring loss of potentialand/or loss of current functions.

Comparing the metering quantitiescalculated by the relay to anotherindependent intelligent electronic device(IED) (another digital relay or panelmeter).

Analysis of fault records to verify the pre-fault and fault current and voltage data iscorrect.

In an automated control platform, a central controlunit (often a PLC or similar device) collectsinstantaneous metering values (e.g. MWatts andMVARs) from the digital relays and other IEDs andcompares them on a line-by-line, breaker-by-breaker, or bus-by-bus basis. For example, if adistribution line has two digital relays (system 1 andsystem 2 or primary and backup) the two sets ofinstantaneous metering values should be nearlyidentical. If they differ by more than a smalltolerance, then one or the other relay has an analoginput component problem and an alarm is generated.

The protection-specific signals may be recordeddifferently as compared with fast-sampled ac inputs.

First, the magnitudes, angles and other derivativesof the raw waveforms are typically calculated at thedifferent, slower pace compared with the A/Dsampling. For example, a relay may sample at 64 s/cbut calculate the magnitudes 16 times a cycle. Thesesignals when recorded will not change between themoments of their calculation.

Second, the relay logic may be executed at yetdifferent rate, sometimes different for variousprotection functions. For example and InstantaneousOverCurrent element (IOC) may be executed 16times a cycle, while a Time-delayed OverCurrentelement (TOC) element may be executed only 2

times a cycle. Output flags from such internalprotection and control elements when recorded willnot change between the moments of their execution.

Third, some signals may be generatedasynchronously with respect to the main samplingclock. For example, the communication-basedinputs may be activated at any time. Typically, thechange in state when recorded will be aligned withthe moment of the first usage of such input flag.

Analog filtering for a/c signals

Any digital device, both a relay and DFR, mustinclude an anti-aliasing analog filter. The purpose ofanit-aliasing filtering is to eliminate higherfrequencies that would otherwise overlap with thelower portion of the spectrum due to finite samplingrates of digital devices. The cut-off frequency of theanalog filter must not be higher than half thesampling frequency. Protective relays, particularlymodels that use only filtered quantities for theirmain protection and control functions, do notcalculate harmonics, or apply comparatively lowsampling rates, would have their analog filters setcomparatively low. As a result, the signal spectrumbeing effectively recorded becomes limited to fewhundred Hz.

Modern relays sample at 64 to 128 samples/cycle(3,840 and 7,680 Hz at 60 Hz power system,respectively) and have their cut-off filters set wellabove 1kHz yielding a comparatively good spectralcoverage.

Another aspect of analog filtering is the design ofthe filter itself. When high order filters are used,their gain may not be ideally flat in the pass-throughfrequency band. This should be factored in when adetailed harmonic analysis is done using recordsproduced by protective relays.

Typically, the analog filter of digital relays is a low-pass filter allowing the sub-harmonics and dccomponents to go through. However, the frequencyresponse of the input magnetic modules at lowfrequencies may alter the low frequencycomponents (at the level of few Hz). This should beconsidered when analyzing sub-harmonics anddecaying dc components.

B-2 Digital filtering for ac inputs

Digital relays often perform digital pre-filteringprior to applying phasor estimation algorithms suchas the Fourier transform to obtain input quantitiesfor their protection and control algorithms. The

22

primary objective of digital filtering is to filter outlow frequency signals, dc component(-s) inparticular. To do so the filters must include adifferentiating portion that ideally should match theL/R constant of the primary circuit, hence the nameof a “mimic filter”.

As a rule digital relays record raw samples, prior todigital filtering. This not only widens the resultingfrequency spectrum of the recorded waveforms, butalso ensures that the stored information does notdepend on any proprietary digital signal processingalgorithms. Whether the recorded data is raw orfiltered needs to be understood by the user.

B-3 De-bounce filtering for dc inputs

Modern relays allow recording status signals of dcinputs connected to the relay. However, as a rule,protective relays apply user-selectable or hard-codedde-bounce filtering. Typically, a relay would recordthe input status after de-bouncing. This means thatthe record reflects signals validated and used forprotection and control and not signals as they appearacross the input terminals of the relay. Knowing thetime duration of the de-bounce timers one mayapply an approximate correction to obtain theoriginal signals before de-bouncing. Details of thede-bouncing algorithm must be known to performthis correction more accurately.

B-4 Sampling rate

Older relays sample much lower than modernrelays. In some cases, this was as low as 4 samplesper cycle and then utilizing curve fitting to make theoscillographic record look clean. Modern relayssample at 64 or 128 samples per cycle provingrelatively good spectral coverage.

Unlike a typical DFR, a relay may apply a variablesampling rate. In order to increase accuracy ofdigital measurements for protection, control andmetering, microprocessor-based relays track powersystem frequency in order to maintain a constantnumber of samples per cycle. This results in variablespacing between the recorded samples. It is stronglyrecommended to use software that recognizesvariable sampling rate formats to view, analyze andplay back relay oscillography files.

Exceptionally, microprocessor-based relays wouldsample and record at a constant rate and re-samplethe actual samples in software to maintain a constantnumber of samples per cycle for its protection andmetering functions. Also, a constant sampling rate isobtained when the frequency tracking feature isdisabled, or the effective tracking signal is not

applied to the relay at a time of producing therecord.

Typically all the raw ac signals – past the anti-aliasing filter – are recorded.

Some relays sample and work at a fixed sample rateper cycle, allow saving decimated samples in orderto save the memory. Typically, the decimatingfactor is a power of 2. Thus, every sample, everyother sample, every fourth sample, every eightsample, etc. could be recorded as per user setting.

B-5 Programability

Some microprocessor-based relays allow userconfigurability of the oscillography records.Available levels of programmability include:

Triggering condition could be userprogrammable to allow producing records froma number of conditions both internal andexternal to the relay.

Recording rate: all or decimated samples couldbe saved. The decimation factor is often a usersetting.

Content: A number of user-programmabledigital and analog channels may be available.The analog channels may include signalcalculated in real time by the relay such asmagnitudes and angles, positive-sequencequantities, power, differential currents, etc., orinput signals other than ac currents and voltagessuch as transducer inputs. Digital channels mayinclude digital inputs, operands createdinternally such as pickup or operate flags forvarious protection elements, auxiliary flags ofuser programmable logic, etc.

Division between the pre- and post-trigger datais often user-programmable.

Number or duration of records is often userprogrammable in order to maximize therecorded information based on availablememory and anticipated duration of the powersystem events of interest.

Treatment of old records is often user-programmable as well. The choices are tooverwrite automatically or forbid new recordsto protect the old ones.

Clearing the records could be user-programmable as well allowing easy orautomated clearance of old records.

23

B-6 Time stamping and resolution

SOE records have limited value if the relay is notsynchronized. Without time synchronization, theSOE record provides relative sequence of eventsonly that may be superimposed on other recordsmanually using primary power system eventsrecorded in the oscillography as “synchronizing” or“reference” marks (fault inception, breakeroperation, fault clearance, etc.).

Older relay designs and low-end digital relays donot support accurate time synchronization. Modernhigh-end relays support time synchronization viaIRIG-B time signals. Typical time accuracy is in therange of a few microseconds.

Other means of synchronization include Ethernet-based algorithms such as Simple Network TransportProtocol (SNTP), which allows synchronization viaTCP/IP with the accuracy of a few milliseconds.

Accuracy of time-stamping should not be confusedwith the resolution. Care must be taken whenanalyzing SOE records created by relays:

DC input signals are typically monitored by aperiodic scan, not by event-driven trulyinstantaneous hardware. The scan period couldbe as short as few hundreds of microseconds oras long as a few milliseconds. This limits theresolution of time stamping for these group ofsignals.

DC inputs are subject to “contact de-bouncing”algorithms. These validate the input signalsparticularly for critical applications such asBreaker Fail Initiate. The de-bouncingalgorithm yields two events for each change ofthe input. The first event is when the dc voltageacross the terminals changes, while the secondevent occurs when the digital de-bounced flagis asserted for further usage. Some relays allowsetting the de-bounce time to zero, bringing thetwo events together. The de-bounce time mustbe factored-in when analyzing time stamps ofdc inputs recorded by digital relays.

Communication-based inputs such as IEC-61850 GOOSE/GSSE or proprietary digitalteleprotection signals may be time-stampedwhen the digital packet starts arriving, when itarrived, or when it got decoded, validated andused for the first time. The above “events”could span over tens or hundreds ofmicroseconds. Relay design details must beknown in order to make analysis of relay SOErecords accurate with respect to thecommunication-based I/Os.

Operations of output contacts are typicallytime-stamped when the digital part of the relaysends the command to close or open a givencontact. Actual dc voltage or current associatedwith the contact start changing hundreds ofmicroseconds to a few milliseconds later.

Internal flags in the relay such as outputs fromprotection or control elements are asserted incourse of serial calculations. These calculationsmay last few hundreds of microseconds onmodern relays to a few milliseconds in older orlow-end relays. Some flags are asserted at thebeginning of the process, while some at the endof this process. Typically, however, all flags aretime-stamped with the same time – either thebeginning or the end of such “protection pass”.Sequence of internal calculations must beknown in order to make analysis of relay SOErecords accurate with respect to the internalflags.

B-7 Available memory

Modern relays allow recording tens of thousands ofsamples. For example, with a limit of 40,000samples, sampling at 64 samples per cycle andrecording 8 ac channels, one could record for about80 power system cycles total; decimating the recordto 16 samples per cycle, one could make a recordingfor about 320 power cycles, etc.

Available memory is one major differentiatorbetween microprocessor based relays and full-featured DFRs.

Often, the memory could be flexibly managed byconfiguring the content and the recording rate.