Applcatons on openpdc platform at Washngton State Unversty Chuanln Zhao Ebrahm Rezae Man V. Venkatasubramanan Washngton State Unversty Pullman WA
WSU projects OMS - Oscllaton Montorng System Stand-alone system for oscllaton detecton and analyss usng wde-area PMUs VSMS - Voltage Stablty Montorng System Stand-alone system for voltage stablty stress ndcator usng wde-area PMUs GrdSm Large-scale real-tme power grd smulator Powertech Labs new product epmu
Oscllaton Montorng System IEEE C37.118 Real-tme PMU data stream Gateway Protocol OpenPDC OMS SQL server OMS results Gateway Protocol OMS acton adapter bult nto OpenPDC 64 bt verson 2.1. 3
Two Types of Data Two Types of Engnes Ambent Data Rngdown Data 4
Start OMS Flowchart Read data from PDC Event? Yes Dampng Montor Engne No FDD analyss for ambent data Prony analyss for postdsturbance data Event Analyss Engne Movng wndow crosscheck Movng wndow crosscheck Poorly damped mode detected? Yes Alarm Controller trgger No 5
Complementary Engnes Event Analyss Engne Four algorthms: Prony, Matrx Pencl and Hankel Total Least Square, ERA. Amed at events resultng n sudden changes n dampng Dampng Montor Engne Ambent nose based. Contnuous. Two algorthms: Fast Frequency Doman Decomposton, Fast Stochastc Subspace Identfcaton Provdes early warnng on poorly damped modes 6
7 The response after small dsturbances can be expressed n the sum of exponental terms Transfer functon Impulse response Samplng at constant perod n s R s u s y s G 1 ) ( ) ( ) ( b, c R where and are rght egenvector and left egenvector n j t R t y 1 ) exp( ) ( n k R z k y 1 ) ( ) exp( t z where Mathematcal Model for Rngdown Data
Mathematcal Model for Ambent Data Power system s n fact a hgh-order nonlnear tme-nvarant system However, n normal operatng state, power system can be modeled as an LTI system for a short perod of tme Random load varatons modeled as whte nose Power System modeled as LTI system Measured outputs, e.g. voltage magntudes 8
Frequency Doman Decomposton Power System Whte nose System System 9
Procedure summary of FDD Form sgnal group from each PMU Power spectrum estmaton by Mult-Taper Method Apply SVD on the power spectrum Apply nverse FFT on largest sngular values Extract the pole frequency and dampng rato from the exponental form by rngdown analyss 10
Event Montor Engne OMS Engnes Automated Prony type analyss of oscllatory rngdown responses Ten seconds of PMU data analyzed every one second Dampng Montor Engne Automated analyss of ambent nose data Four mnutes of PMU data analyzed every ten seconds 11
Results from Two Engnes Voltage Angle at Cumberland Substaton Angle (Degree) 15 14.5 14 13.5 13 12.5 Ambent Nose Analyss Event Analyss 810 820 830 840 850 860 870 Tme (s) Prony analyss: 1.25 Hz Local Mode @ 1.5% dampng rato FDD Analyss: 1.224 Hz Local Mode @ 1.17% dampng rato Defectve card found n Power System Stablzer and fxed. 12
Mode Shape Estmaton 1 0.8 0.6 Mode Shape Identfed by Prony Analyss at 1.2553 Hz 1 0.8 0.6 Mode Shape Identfed by FDD at 1.224 Hz Imagnary Part 0.4 0.2 0-0.2-0.4 Imagnary Part 0.4 TVA_COLN_MAIN TVA_CUMB_BUS10.2 TVA_CUMB_BUS2 TVA_HEND_BUS1 TVA_LOWN_161K 0 TVA_LOWN_500K TVA_MARS_CUMB -0.2 TVA_RIDG_BUS1 TVA_RIDG_BUS2-0.4 TVA_COLN_M TVA_CUMB_B TVA_CUMB_B TVA_HEND_B TVA_LOWN_1 TVA_LOWN_5 TVA_MARS_C TVA_RIDG_BU TVA_RIDG_BU -0.6-0.6-0.8-0.8-1 -1-0.8-0.6-0.4-0.2 0 0.2 0.4 0.6 0.8 1 Real Part Prony Mode Shape -1-1 -0.8-0.6-0.4-0.2 0 0.2 0.4 0.6 0.8 1 Real Part FDD Mode Shape Mode shape helps pnpont the root cause of oscllatons. 13
Oscllaton Montorng System IEEE C37.118 Real-tme PMU data stream Gateway Protocol OpenPDC OMS SQL server OMS results Gateway protocol OMS acton adapter bult nto OpenPDC 64 bt verson 2.1. Tested at Entergy, TVA, Idaho Power, and WECC. 14
openpdc-based Archtecture openpdc (Server Sde) Clent Sde Dampng Analyss Adapter Real-tme Vsualzaton PMU s Input Adapter GEP Publsher GEP Subscrber openhstoran Event Analyss Adapter Hstory Vsualzaton Server-sde subsystem: two Acton Adapters. Dampng Adapter analyze ambent data to provde early-warnng on poor-damped modes Event Adapter ams at events resultng n sudden changes n dampng Clent-sde subsystem Real-tme vsualzaton receves updates from server, and presents them to user. stores the receved result nto local openhstoran Hstory vsualzaton retreve results from local openhstoran database show the oscllaton trend of the system n a specfc perod. 15
Montor usng ambent data Dampng Analyss Adapter Slde Wndow Concentrator Wndow FFDD Algorthm Result Decomposer Analyss Result Measure ment Measure ment Measure ment GEP Publsher 1. s are concentrated nto frame 2. s are nserted nto a slde wndow. 3. A wndow of frames s fed nto FFDD Algorthm to detect Oscllaton 4. Detecton result are decomposed nto a batch of s. 5. Result s wll be publshed usng GEP 16
Phase I. Statstcs Phase Montor usng event data Event Analyss Adapter (Statstcs Phase) GEP Publsher Concentrator Wndow Event Detector accumulates a long perod of s. calculate some statstcs on ths long perod of data use these statstcs as the thresholds to detect future event Ths phase wll repeat perodcally to update those event thresholds 17
Montor usng event data Detect Phase Event Analyss Adapter (Detecton Phase) Slde Wndow Concentrator Event Detector Wndow Wndow wth Event HTLS Prony MatrxPencl ERA Analyss Result Consstency Checker Result Decomposer Measure ment Measure ment Measure ment GEP Publsher 1. s are concentrated and fed nto slde wndow. 2. A wndow wll go to Event Detector. 3. If contnuous events are detected, further analyss starts. 4. Wndow wth real event are analyzed by Prony, Matrx Pencl, HTLS and ERA n parallel 5. Analyss results need to go through Consstency Check. 6. Consstent result wll be decomposed nto s and publshed usng GEP 18
Real-tme vsualzaton PMU s Current Modes Mode shapes Modes Trend Curve Modes Scatterplot Power Spectrum Densty 19
Hstory vsualzaton 20
OpenPDC at WSU OpenPDC and openhstoran used extensvely n several projects OpenPDC based PMU applcatons beng nstalled at Entergy, TVA, and WECC Suggestons, Debuggng, and WSU code contrbuton (openpdclte) Confg tools, Vsualzaton tools Commercalzaton