Probabilistic Assessment of PMU Integrity for Planning of Periodic Maintenance and Testing

Transcription

1 robablstc Assessment of MU Integrty for lannng of erodc Mantenance and Testng Tamara Becejac Student Member IEEE ayman Dehghanan Student Member IEEE and Mladen Kezunovc Fellow IEEE Department of Electrcal and Computer Engneerng Texas A&M Unversty College Staton Texas USA Abstract The standard C a-2014 has specfed the permssble lmts for MU measurement errors under varous statc and dynamc test condtons. Ths paper proposes a statstcal measure to evaluate the probablty of MU performance degradaton wth regards to certan standard requrements. The proposed approach s mplemented usng a feld calbrator system for phasor measurement unts (MUs). Assessment of the test results provdes an addtonal nsght about: (a) whether the expected functonalty and ntegrty of the MUs s mantaned over tme; (b) whch synchrophasor standard requrements are most vulnerable for a gven devce over tme; (c) when the mantenance schedule needs to be expedted on certan MUs based on observed performance degradaton probabltes; and (d) the rss of loss of trustworthness of varous end-use synchrophasor-based applcatons. The applcablty of the suggested technque s verfed through mplementaton on several MUs n a calbraton and testng set-up. Keywords calbraton; perodc mantenance; phasor measurement; probablstc; standard type-tests; synchrophasor. I. INTRODUCTION Snce synchrophasor technology was ntroduced developed and partly ntegrated nto the electrc power systems n early 1980s t has been shown that t offers great advantages n control and montorng of the grd thans to ts hgh-resoluton tme synchronzed measurements [1] [2]. hasor Measurement Unts (MUs) are nstruments that can provde a precse and comprehensve vew of the system dynamcs. They can trac the state of the system n real tme by measurng the voltage and current phasor frequency and rate of change of frequency at a rate of 10 to 120 phasor frames per second [3]. MUs serve as the bacbone for many system-wde applcatons and algorthms n electrc power systems. Some of the applcatons out of many that have been reported are: accurate measurement of frequency and magntude of voltages and currents state estmaton nstablty detecton and evaluaton contngency analyss adaptve relayng system-wde control and montorng etc. [4]. Snce 2005 varous standards for the statc and dynamc performance of the phasor measurement unts (MUs) as well as communcaton requrements for the synchrophasor data transfer have been developed and eventually adopted. Standard IEEE C defnes the acceptable performance of synchrophasor measurements n power systems [5]. In 2014 ths standard was revsed where some tests were removed and some requrements were revsted due the fact that most of the MU devces and Intellgent Electronc Devces (IEDs) wth MU capabltes avalable on the maret at the tme were not meetng the standard [6]. The procedures and requrements for test equpment such as tmng reference sgnal sources calbraton devces and envronmental condtons are specfed n the IEEE Synchrophasor Measurement Test Sute Specfcaton (TSS) document publshed by the IEEE Conformty Assessment rogram (ICA) [7]. TSS provdes a sute of unambguous test plans n accordance wth Smart Grd Interoperablty anel Recommendatons and Interoperablty rocess Reference Manual. Standard IEEE C covers requrements for the MU data transfer n power systems [8] and standard IEEE C provdes gudance for synchronzaton calbraton testng and nstallaton of MUs appled n power system protecton and control [9]. Testng procedures for the hasor Data Concentrators (DCs) are presented n the IEEE C Gude for hasor Data Concentrator (DC) Requrements for ower System rotecton Control and Montorng [10]. Consderable research has been devoted to complance analyss of MUs through calbraton type-tests and algorthm testng [11]-[18] and feld-testng procedures [19]- [22]. General evaluaton of MU dynamc performance n both the lab envronment and under feld operatng condtons s reported n [19]-[22]. By utlzng the standard type test results under both dynamc and statc test condtons ths paper tres to ntroduce an analytcal methodology to probablstcally evaluate the deteroraton of MU performance over tme wth regards to the standard requrements. The proposed probablstc technque can quanttatvely reveal the MU response to varous test performance thresholds specfed by correspondng IEEE standards and also the probablty of specfc falures to meet the standard requrements. The trends revealed by such probablstc metrcs could help n better plannng the mantenance perods and future n-fled testng procedures. The remander of the paper s structured as follows. Secton II elaborates the general structure of the developed MU calbraton test set up as well as ts varous functonaltes. The proposed technque for assessment of synchrophasor devce ntegrty for perodc mantenance and testng s ntroduced n Secton III. Expermental results from the test studes on several MU devces are presented n Secton IV followed by the concludng remars n Secton V.

2 II. MU CALIBRATION AND TESTING TESTBED A. Test Needs and Requrements In addton to standard type-test practces that need to be performed durng acceptance tests performed n the calbraton lab envronment and durng the early commssonng stages of MU deployment testng s also desrable after the MU has been n servce for a longer tme to ensure that the MU functonalty over tme stll comples wth the standard requrements. Testng MUs n the feld needs to be performed usng a portable test devce namely feld calbrator. Tools need to be developed to provde an nfeld test opportunty for the users to (1) mplement the typetests n the feld manly to meet perodc mantenance plans and (2) perform the applcaton testng to ensure the trustworthness of the MU results for end-use synchrophasor applcatons. Focusng on the applcatons perodc testng of the MUs under statc and dynamc test condtons provdes a database contanng valuable nformaton to the users on how the MU measurements are affected over tme n face of devce wear and tear mechansm and envronmental/operatonal condtons n real-world. Such nformaton further helps n detectng possble orgns of the measurement abnormaltes and can be used to develop rs mtgaton plans to constantly ensure the synchrophasor applcaton trustworthness. B. Software and Hardware Implementaon A MU test and calbraton platform used to verfy the conformance of the evaluated MUs under varous statc and dynamc tests accordng to the IEEE standards s developed at Texas A&M Unversty. The software and hardware desgn can be used for MU calbraton and testng n both laboratory and n-fled envronment. A general structure of the MU calbraton lab s shown n Fg. 1 and the actual mplementaton of such testng envronment s llustrated n Fg. 2. As shown n Fg. 1 the MU test system conssts of tmng reference (GS recever) sgnal generator power amplfer and data management and results analytcs tools. Tmng reference provdes GS tmng cloc to the calbraton equpment and to the MU under test so that the entre system s tme-synchronzed and tme-stamped. Test sgnals are generated by the sgnal generator accordng to the test types determned by the IEEE TSS document [7]. Synchrophasor measurements from the devce under test are compared wth reference values calculated usng mathematcal formulatons gven n IEEE C standard then followed by an analyss and documentaton of the test results. The MU test and calbraton platform s mplemented usng Natonal Instrument (NI) hardware as shown n Fg. 2. The entre system conssts of the XI vrtual nstrument system wth embedded Controller NI XIe-8105 a userprogrammable FGA whch s a part of NI XI-7854R multfunctonal reconfgurable I/O module to generate the requred waveforms and an OMICRON CMS 356 power amplfer that generates 3-phase voltage and current sgnals feedng the MU devce under test. As a part of the system software based hasor Data Concentrator (DC) module that receves and parses the data s runnng on the XI system. Measurements from the tested MU are acqured through fast speed Ethernet communcaton ports then analyzed and reports are generated usng the NI LabVIEW software nterface [23]. Reports consst of all data resultng from MU tests whch allows extensve post-analyss of the collected results. Whle the standard tests are conducted n the laboratory envronment they are also desgned to be used n calbrator set-up for testng the MUs n the feld. C. Dynamc and Steady-State Testng of MUs MUs provde nformaton such as voltage and current magntude angle frequency rate of change of frequency etc. Accordng to the IEEE C a standard each devce that s capable of provdng GS synchronzed measurements has to undergo varous steady state and dynamc test scenaros whle beng calbrated. Durng the steady state tests MUs are exposed to varous type-test scenaros where all varables are ept unchanged durng each test and measurements are captured accordng to the standard procedure. As shown n Table I such statc type-tests nclude performance evaluaton of MUs over a range of frequency values voltage/current ampltudes as well as nfluence of harmonc and nter-harmonc nterferences. Fg. 1. Calbraton and testng platform for MUs.

3 TABLE I. MU STEADY STATE AND DYNAMIC STANDARD TYE-TESTS Test Category Steady Tests Dynamc Tests Test Name Input Range Class M Class Magntude Sweepng pu pu Frequency Sweepng ±2Hz ±2Hz or ±5Hz Harmonc Dstorton 1% each harmonc up to 50th 10% each harmonc up to 50th Latency 1000 consecutve messages 1000 consecutve messages Dependng on reportng frequency harmonc Out-of-Band Interference N/A nfltraton from 10Hz to twce nomnal frequency; nterferng sgnal 10% of sgnal magntude Ampltude Modulaton Modulaton frequency from 0.1Hz to 2Hz Modulaton frequency from 0.1Hz to 5Hz Angle Modulaton Modulaton level 0.1 Modulaton level 0.1 Frequency Ramp ±1Hz/s frequency range wthn ±2Hz ±1Hz/s frequency range wthn ±5Hz Magntude Step Change ±10% Step Angle Step Change ±p/18 radans developed usng the normal dstrbuton assumpton wor qute well even when the dstrbuton s not normal. To proceed wth the methodology the mnmum and mum bands are adopted from the IEEE standard for each test sgnal. If one value x falls n the desrable margn then t ndcates a proper functonalty of the MU reflected by that specfc test. Smlarly one new value of x may fall out of the desrable band leadng to the test falure. In general and accordng to the probablty dstrbuton assgned the probablty of a gven test success can be calculated n (1)-(2) respectvely for the steady-state and dynamc condtons. e ST ST f ( ) d (1) e e mn e e DT DT f ( ) d (2) e e mn e Fg. 2. Actual mplementaton of the MU testng platform. Dynamc type-tests nvolve testng MUs wth the modulated sgnals checng ther performance durng the step occurrence n ampltude and angle as well as testng the MU response to the frequency ramp events. As a part of the standard requrements latency of a MU devce has to be measured too. The MU performance conformty under the aforementoned statc and dynamc condtons needs to be ensured durng the lfe-cycle of the MUs and hence perodc mantenance and testng of the MUs can help checng the desred functonalty over tme. III. ROOSED METHODOLOGY A practcal approach to evaluate the performance degradaton of MUs consderng varous steady-state and dynamc test sgnals s devsed. Conductng the standard tests on the MUs n the feld varous MU responses correspondng to each type-test and sets of data n terms of varous error ndcators are captured and recorded. The error ndcators calculated are typcally the Total Vector Error (TVE) Frequency Error (FE) and error n estmaton of Rate of Change of Frequency (RFE). A normal probablty dstrbuton can then be assgned to each test measurement error data as demonstrated n Fg. 3. The proposed approach s however generc enough to be adopted wth dfferent types of probablty dstrbutons as data may dctate n varous applcatons. In many practcal cases the methods Where the followng nomenclature apples: Success probablty of a type-test regardng the e error ndcator e for MU n the system. Error ndcator for test type (.e. TVE FE RFE). e ST Steady-state type-test. DT Dynamc type-test. mn Mnmum threshold for error ndcator for test. e Maxmum threshold for error ndcator for test. e fe ( ) robablty densty functon of an error ndcator for type-test of MU n the system. There are several error ndcators for some MU tests that all need to be n complance wth standard requrements. In order to be able to conclude a probablstc measure for the success/falure of a gven test on a MU equatons (3)-(4) are mn Fg. 3. robablty dstrbuton and the bands assgned to the error ndcators for each MU type-test.

4 proposed to ntegrate such ndcators where applcable nto one metrc: Where 3 ST ST e 1 e TVE FE RFE (3) 3 DT DT e 1 ST DT and e TVE FE RFE (4) are the success probablty of the statc and dynamc tests respectvely consderng all the requred error ndcators wthn the desrable thresholds. Even though MUs may pass the standard tests t may be desrable for the user to now how far the measurement errors correspondng to varous tests are from the desrable standard thresholds. The equaton (5) ndcates a probablstc metrc for the dstance of the reported error mean values from the desrable standard thresholds. Ths metrc s useful n tryng to understand how relable a gven MU s wth regards to a gven test and what adjustments need to be made and how fast. That s t can dfferentate varous MUs that need perodc mantenance and troubleshootng by nowng whch test requrements are more lely to cross the standard requrements over tme. N 1 e e e n (5) n n 1 where e s the probablstc metrc representng the dstance between the mean error value and the mum standard threshold; N s the total number of error observatons for a gven type-test. In order to have an overall evaluaton of the MU performance robustness consderng all the requste statc and dynamc tests a probablstc measure representng the ntegrty of the MU n terms of testng results n face of all the statc and dynamc test condtons s suggested n (6). K K Integrty ST DT (6) 1 1 In whch K and K represent the sets of statc and dynamc tests respectvely. Observatons on the usng the Integrty proposed n-feld tests over tme show how the MU typetest errors are movng wth respect to the desrable thresholds representng the degradaton of the MU measurements as the tme goes on. The trend on such observatons could also help better plan for perodc mantenance over the MU lfecycle as needed. IV. NUMERICAL RESULTS The proposed technque for MU n-feld testng s appled to several MUs from varous manufacturers wth dfferent operatonal characterstcs and settngs. Sample test results on a gven MU are presented n Fg. 4 usng the aforementoned calbraton test set. Varous MUs were exposed to dfferent test sgnals and the performance of each product was thoroughly analyzed. As can be seen n Fg. 4 one of the MU under test has faled most of the tests (e.g. voltage magntude sweep frequency sweep out-of-band nterferng frequency) whle t has passed only one (.e. harmonc dstorton). The suggested probablstc metrcs are calculated for varous statc and dynamc tests for the MU under study. The results are tabulated n Table II. The success probablty ndcators for varous statc and dynamc tests are evaluated n columns 3-5. As can be seen n Table II whle the error values for some standard tests are perfectly wthn the desred thresholds n all the test scenaros (e.g. harmonc dstorton) the success probablty of some other tests s very low (as low as 0.18% for out-of-band test) reflectng the fact that the probablty of ncorrect or not accurate reported measurements s extremely hgh wth unacceptable level of uncertanty. Some condtons that MU under test s seeng (a) voltage magntude sweepng test (b) frequency sweepng test (c) harmonc dstorton test

5 (d) out-of-band dstorton test (e) ampltude modulaton test (f) phase modulaton test (g) frequency ramp test (h) magntude step test () phase angle step test Fg. 4. Test rsults of a gven MU udner varous statc and dynamc test condtons tang nto account wde range of sgnals (.e. frequency equal to 45/65 Hz or voltage magntude equal to 10% of nomnal value) are not very lely to happen n real-world but are mportant to certfy a MU calbraton procedure. Therefore even f the MU under test fals n one of such test condtons t does not mply that the tested devce s 100% unrelable whle exposed to extreme sgnal condtons. Rather than solely a pass/faled status ths ndex hghlghts the success rate of all

6 TABLE II. ROBABILSITIC METRICS FOR MU VULNERABILITY AND INTEGRITY ASSESSMENT Test Category Steady Tests Dynamc Tests Test Name TVE FE ST RFE DT TVE FE RFE Magntude Sweepng N/A 0.67 N/A N/A Frequency Sweepng N/A Harmonc Dstorton N/A N/A Out-of-Band Interference N/A N/A Ampltude Modulaton N/A Angle Modulaton N/A Frequency Ramp N/A Magntude Step Change N/A N/A N/A Angle Step Change N/A N/A N/A the test scenaros wthn a gven type-test and as a result s a relable measure to recognze sustanable problems n a MU devce and dfferentate them from ephemeral condtons that mght have led to a test falure. Tang nto account all varous error values the probabltes reported n columns 6-7 demonstrate the MU ntegrty values wth respect to all the statc and dynamc tests respectvely. The MUs may pass all the statc and dynamc tests but the test results do not reflect the uncertanty level of the test acceptance. In other words the orgnal test results do not reveal how close the errors are to the standard thresholds. The probablstc values calculated n the last three columns of Table II show n the range of 0-1 the dstance of the measurement errors to the desred thresholds for varous tests. The N/A entres n Table II reflect the fact that there s no desrable thresholds reported for a gven error metrc correspondng to several statc and dynamc test n the standards. The closer the dstance values ( e ) are to 1 the more relable the test results are wth respect to the desred thresholds. As the probablstc dstance metrcs decrease the test results are vulnerable to falure n the next test nterval and the perodc mantenance plans need to be expedted; otherwse the MU measurements and consequently the enduse synchrophasor applcatons n power system may not be relable n practce. V. CONCLUSIONS The paper accomplshes the followng: It shows the mportance and necessty of testng the synchrophasor devces n the feld for a more effcent perodc mantenance plannng. robablstc metrcs are proposed to evaluate: a) underlyng vulnerablty of the MU responses under both statc and dynamc tests and b) the ntegrty measure of the MU devces over tme. The proposed method could recognze the sustaned problems n MU under test for an expedted mantenance plan. The suggested analyss tool offers the user test results n face of varous dynamc and statc condtons revealng a more realstc vew on the vulnerablty of MU whch may lead to fne tunng of the perodc mantenance schedules when and where necessary. REFERENCES [1] A. hade Synchronzed hasor Measurements - A Hstorcal Overvew n IEEE ower and Energy Socety (ES) Asa acfc Transm. and Dstr. Conf. and Exhbton vol. 1 pp Oct [2] J. Gr D. Sun and R. Avla-Rosales Wanted: A More Intellgent Grd IEEE ower & Energy pp Aprl [3] M. D. Hadley J. B. McBrde T. W. Edgar L. R. O Nel and J. D. Johnson Securng Wde Area Measurement Systems acfc Northwest Natonal Lab. Techncal Report NNL [4] F. Amnfar et.al. Synchrophasor Measurement Technology n ower Systems: anorama and State-of-the-Art IEEE Access vol. 2 pp [5] IEEE Standard for Synchrophasor Measurements for ower System IEEE Std. C TM [6] IEEE Standard for Synchrophasor Measurements for ower System Amendment 1: Modfcaton of Selected erformance Requrements IEEE Std. C a TM [7] IEEE Synchrophasor Measurement Test Sute Specfcaton--Verson 2 pp.1-43 Sept [8] IEEE Standard for Synchrophasor Data Transfer for ower System IEEE Std. C TM [9] IEEE Gude for Synchronzaton Calbraton Testng and Installaton of hasor Measurement Unts (MUs) for ower System rotecton and Control IEEE Std. C TM [10] IEEE Gude for hasor Data Concentrators Requrements for ower System rotecton Control and Montorng IEEE Std. C [11] M. A. Donolo and V. A. Centeno A fast qualty assessment algorthm for phasor measurements IEEE Trans. ower Del. vol. 20 no. 4 pp Oct [12] U. oglano J. Braun B. Voljc and R. Lapuh Software platform for MU algorthm testng IEEE Trans. Instrum. Meas. vol. 62 no. 6 pp Jun [13] G. Barch D. Mac and D. etr Synchrophasor estmators accuracy: A comparatve analyss IEEE Trans. Instrum. Meas. vol. 62 no. 5 pp May [14] H. Lu T. B and Q. Yang The evaluaton of phasor measurement unts and ther dynamc behavor analyss IEEE Trans. Instrum. Meas. vol. 62 no. 6 pp Jun [15] D. R. Gurusnghe A. D. Rajapase and K. Narendra Testng and enhancement of the dynamc performance of a phasor measurement unt IEEE Trans. ower Del. vol.29 no.4 pp Aug [16] Y.-H. Tang G. N. Stenbaen and A. Goldsten Calbraton of phasor measurement unt at NIST IEEE Trans. Instrum. Meas. vol. 62 no. 6 pp Jun [17] I. Kamwa S. R. Samantaray and G. Joos Complance analyss of MU algorthms and devces for wde-area stablzng control of large power systems IEEE Trans. ower Syst. vol. 28 no. 2 pp May [18] Q. Zhang V. Vttal G. T. Heydt N. Logc and S. Sturgll The ntegrated calbraton of synchronzed phasor measurement data n power transmsson systems IEEE Trans. ower Del. vol. 26 no. 4 pp Oct [19] R. E. Wlson and. S. Sterlna Verfcaton of measured transmsson system phase angles IEEE Trans. ower Del. vol. 11 no. 4 pp Oct [20] R. E. Wlson MUs [phasor measurement unt] IEEE otentals vol. 13 no. 2 pp Apr [21] Z. Huang J. F. Hauer and K. E. Martn Evaluaton of MU dynamc performance n both lab envronments and under feld operatng condtons roc. IEEE ES General Meetng pp.1-6 Florda [22] K.E. Martn T.J. Fars J. F. Hauer Standardzed testng of phasor measurement unts Fault and Dsturbance Analyss Conference Georga Tech Atlanta GA May [23] [Onlne] avalable at: