Oscillation Monitoring Venkatasubramanian_pserc_project_tele-seminar_slides_oct2008

Oscillation Monitoring Venkatasubramanian_pserc_project_tele-seminar_slides_oct2008

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: PSERC Oscillation Monitoring System Mani V. Venkatasubramanian Washington State University PSERC S29 Project Tele-seminar October 28, 2008 1 PSERC Overview of S29 project • S19 project from June 2002 to May 2005 • S29 project from June 2006 to July 2008 • Detection, Prevention and Mitigation of Cascading Events • Three tasks: • Task I: Detection: Mladen, Texas A&M Advanced warning • Task II: Prevention: Mani, Wash. State Wide-area monitoring and controls • Task III: Mitigation: Vijay, Iowa State/ASU Adaptive islanding S29 Focus on Prototype Implementations • 2 PSERC Task II: Mitigation • Implementation of Wide-area Small-signal Stability controller, Wash. State University • Two subtasks: Reliable Oscillation detection: • Multi-input Prony, Matrix Pencil and HTLS algorithms • Rules for real-time analysis of data • Noise? Linear versus nonlinear? Switching events? Real-time design of damping controls: • Which control to trigger? What design? • Not part of the prototype testing 3 PSERC Problem Overview • Low frequency electromechanical oscillations Local or inter-area oscillations Insufficient damping Example - Aug 10, 1996 WECC blackout • Detection and control of small • signal stability problem in power systems Research supported by PSERC, TVA, Entergy, BPA and EPG (CERTS) 4 PSERC • WSU: Task II Project Team • Guoping Liu, Qiang Zhang, Jaime Quintero, Mani V. Venkatasubramanian • TVA: Ritchie Carroll, Gary Kobet, Lisa Beard • Entergy: Floyd Galvan, Sujit Mandal, Sharma Kolluri • BPA: Bill Mittelstadt, Dmitry Kosterev • EPG: Manu Parashar 5 PSERC Oscillation Monitoring System (OMS) • Goal of Oscillation Monitoring System (OMS) Early detection of poorly damped oscillations as they appear Trigger warning or control signals OMS is made possible by Wide Area PMU Measurements Growing numbers of PMUs across the power grid Fast algorithms available for online measurements Rule based automatic analysis of PMU measurements Prototype implementation at TVA LOAD P SVC SVC LOAD • PDC P P SVC LOAD LOAD 6 PSERC OMS Flowchart Damping Monitor Engine Event Analysis Engine 7 PSERC TVA Cumberland event Four minutes of oscillations 8 Source: Gary Kobet/TVA PSERC TVA Cumberland Event • Recent oscillatory event at TVA: Oscillations at Cumberland plant 9/18/2006 PMU recordings enabled the analysis Local 1.2 Hz mode changed from +1.5% damping to –0.2% damping and back to +1.5% damping during the event PSS installed at the plant subsequently PMU based real-time alarm coded by TVA into TVA PDC as back-up measure – uses standard deviation thresholds – plant operators to reduce MW output when alarm received. 9 PSERC Oscillation Monitoring System • Software Engines built into TVA PDC • Real-time streaming data input to the engines • Fast detection of poorly damped oscillatory modes: mode frequency, damping and mode shape • Multiple algorithms integrated by expert system like rules • Focus on Redundancy and Reliability 10 PSERC OMS Engines • Event Analysis Engine Automated Prony type analysis of oscillatory ringdown responses Five seconds of PMU data analyzed every one second • Damping Monitor Engine Automated analysis of ambient noise data Three minutes of PMU data analyzed every ten seconds Provisional Patent application filed by WSU 11 PSERC Results from Two Engines x 10 5.15 5 Bus Voltage Magnitude at Cumberland 5.14 5.13 5.12 Ambient Noise Analysis 1.2 Hz at +1.8% damping. Local Mode. 5.11 5.1 Event Analysis 1.2 Hz at +1.5% damping. Local Mode. 820 840 860 880 900 920 940 5.09 Time (s) Nov. 29th 2007 TVA event 12 PSERC Mode Shape – Local Mode Mode Shape Identified by FDD at 1.224 Hz 1 0.8 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 -1 Im aginary Part Other PMUs Cumberland TVA_COLN_MAIN TVA_CUMB_BUS1 TVA_CUMB_BUS2 TVA_HEND_BUS1 TVA_LOWN_161K TVA_LOWN_500K TVA_MARS_CUMB TVA_RIDG_BUS1 TVA_RIDG_BUS2 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 Real Part Cumberland oscillating against rest of system 13 PSERC Basics of Prony Analysis & x = A x + Bu y = C x + Du yi (t ) = ∑ cij e j −ζ j ωnj t cos(ω dj t − ϕ j ) + ∑ cij e j −a jt • Assumptions: Linear Time-Invariant System, Distinct Eigenvalues, Step changes in input, … • Any output is a linear combination of fundamental modal responses • Well-suited for Prony type curve fitting methods. Estimate oscillatory frequency, damping ratio and mode shape. • Estimates should be consistent: • Moving time-windows (Linearity of responses) • Different groupings of outputs (Superposition) 14 PSERC Power System Prony Analysis • Nonlinear Large Scale System • In theory, Prony Analysis works well for analyzing “Small-disturbance responses” • Nonlinearity dominant just after large disturbances • Switching of lines and cap banks in the middle of analysis windows • Noise effect on results if disturbance “fades away” • How to get reliable estimation automatically? 15 PSERC Rules for Real-time Prony Analysis MALN_Malin_Bus_Voltage_VMag vs time 550 548 546 544 542 540 538 536 534 60 80 100 120 Time (s) 140 160 180 M L _ a _ u _ o g _ Mg A N M lin B s V lta e V a Three types of Consistency Crosscheck rules Different Curve-fitting Methods (Redundancy) Different Signal Groups (Superposition) Moving Window Analysis (Linearity of Reponses) 16 PSERC Event Analysis Engine Local PMU Analysis Inter-area Mode Analysis 17 PSERC Local PMU Analysis • • • • Signals from one PMU used at a time Parallel implementation of multiple PMU analysis Parallel implementation for multiple algorithms Check for consistency using rules: Crosscheck results from Prony, Matrix Pencil and HTLS Crosscheck results among moving time-windows 18 PSERC Interarea Mode Analysis • Identify interarea modes and related PMUs from • • local analysis Grouping of signals from relevant multiple PMUs Check for consistency using three sets of rules: Crosscheck results from Prony, Matrix Pencil and HTLS Crosscheck results among moving time-windows Crosscheck results from different groupings Parallel implementation for different interarea 19 modes • PSERC Event Analysis – Inter-area Example • WECC Aug. 4, 2000 • Alberta system • separated at 19:56 GMT 0.27 Hz oscillation is poorly damped MALN_Malin_Bus_Voltage_VMag vs time 550 548 MALN_Malin_Bus_Voltage_VMag 546 544 542 540 538 536 534 60 80 100 120 Time (s) 140 160 180 20 PSERC Case Study 1 – Local PMU Analysis Frequency estimates 0.5 Frequenc y (Hz ) 0.4 0.3 0.2 0.1 60 70 80 90 100 110 Time (s) Damping ratio estimates 120 130 Frequency estimates 0.5 Frequency (Hz ) Prony MatrixPencil HTLS 0.4 0.3 0.2 0.1 60 70 80 90 100 110 Time (s) Damping ratio estimates Prony MatrixPencil HTLS 120 130 Dam ping ratio 0.2 0.1 0 -0.1 -0.2 60 70 80 90 100 Time (s) 110 Dam ping ratio Prony MatrixPencil HTLS 0.2 0.1 0 -0.1 -0.2 60 70 80 90 100 Time (s) 110 Prony MatrixPencil HTLS 120 130 120 130 Grand Coulee Malin 21 PSERC Inter-Area Oscillation Mode Consistent local estimates vs time DVER 8 VNCT 7 KEEL 6 JDAY 5 MALN 4 GC50 3 COLS 2 BE50 1 Consistent estimate at 106 sec Oscillation frequency = 0.286 Hz Mean damping ratio = +2.77% 60 70 80 90 100 Time (s) 110 120 130 22 PSERC Event Monitor – Local Mode x 10 2.975 5 PMU 2 PMU 8 PMU 7 Consistent local estimates vs time 2.97 PMU 6 Voltage at PMU 2 (V) 2.965 PMU 5 2.96 PMU 4 2.955 PMU 3 PMU 2 PMU 1 2.95 2.945 320 325 330 335 340 345 Time (s) 350 355 360 365 335 340 345 Time (s) 350 355 360 • Consistent estimate at +9 sec • Frequency = 1.1785 Hz. Damping at 0.04% 23 PSERC Case Study 3 – Growing Oscillations x 10 5.275 5.27 5.265 Voltage at PMU 1 (V) 5.26 5.255 5.25 5.245 Damping ratio 5.24 5.235 5.23 5.225 720 740 760 780 800 Time (s) 820 840 860 880 -0.2 740 750 760 770 780 Time (s) 790 800 810 820 0.4 0.2 0 Frequency (Hz) 0.9 0.8 0.7 0.6 0.5 740 750 760 780 790 Time (s) Damping ratio estimates 770 800 810 820 5 PMU 1 Frequency estimates Prony MatrixPencil HTLS Prony MatrixPencil HTLS • Consistent estimate at 809 sec; Local mode. • Frequency = 0.6930 Hz. Damping ratio = -0. 12% 24 PSERC Damping Monitor Engine • Frequency Domain Decomposition (FDD) algorithm proposed for off-line analysis in other areas • Extended for real-time PMU analysis by Guoping • Can detect damping ratio as well as mode shape of poorly damped oscillatory modes using short spans of ambient PMU data • Mode shape information critical for correctly identifying problematic mode towards control actions • Excellent results for modes with damping ratio up to +10% 25 PSERC FDD for Ambient Analysis • FFT and Power Spectra from multiple signals • Clean up spectra using Singular Value Decomposition procedure near dominant modes • Prony type damping analysis after Inverse FFT around dominant modes • ISCAS 2008 paper • Simultaneous extraction of mode damping and mode shape from ambient data 26 PSERC Example of Damping Monitor Engine Active Power from Malin to Round Mountain #1 1700 1600 Ross-Lexington Line trips Keeler-Allston Line trips Active Power (MW) 1500 1400 1300 1200 1100 1000 300 400 500 600 700 800 Time (s) 27 PSERC Before Keeler-Allston Trip Frequency (Hz) Frequency Estimates 1.5 1 0.5 0 150 200 250 300 350 400 Damping Ratio (%) Time (s) Damping Ratio Estimates 8 6 4 2 0 150 200 250 300 350 400 Time (s) Dominant mode is McNary local mode 28 PSERC Before Keeler-Allston trip Frequency (Hz) Frequency Estimates 1.5 1 0.5 0 150 200 250 300 350 400 Damping Ratio (%) Time (s) Damping Ratio Estimates 8 6 4 2 0 150 200 250 300 350 400 Time (s) Second Dominant mode is COI interarea mode 29 PSERC After Allston-Keeler Trip Frequency (Hz) Frequency Estimates 1.5 1 0.5 0 600 620 640 660 680 700 720 Damping Ratio (%) Time (s) Damping Ratio Estimates 8 6 4 2 0 600 620 640 660 680 700 720 Time (s) Dominant mode is COI interarea mode 30 PSERC After Keeler-Allston Trip Frequency (Hz) Frequency Estimates 1.5 1 0.5 0 600 620 640 660 680 700 720 Damping Ratio (%) Time (s) Damping Ratio Estimates 8 6 4 2 0 600 620 640 660 680 700 720 Time (s) 31 Second Dominant mode is McNary local mode PSERC Complementary Engines • Event Analysis Engine Three algorithms: Prony, Matrix Pencil and Hankel Total Least Square. Crosscheck Rules. Aimed at events resulting in sudden changes in damping • Damping Monitor Engine Ambient noise based. Continuous. Frequency Domain Decomposition Algorithm. (Prony). Crosscheck Rules. Provides early warning on poorly damped modes 32 PSERC Example of results for TVA Damping history of 1.2 Hz mode Sept. 18, 2006 Dec. 16, 2006 Nov. 29, 2007 Feb. 5, 2008 Event Analysis +1.7% +7.2% +1.5% +4.0% Damping Monitor +1.7% No data +1.8% +3.0% PSS Status No PSS (2U) PSS installed (1U) PSS offline (2U) PSS offline (1U) PSS status and effectiveness from the damping level of the local mode. PSS not effective for two units in service. PSS hardware problem detected and fixed (June 2008). 33 PSERC Eastern System Interarea Mode • Interarea mode frequency varies between 0.4 Hz to 0.5 Hz depending on season. • Damping Monitor (ambient noise) showed the mode to be poorly damped around +3% to +5% seasonally. • 0.47 Hz Interarea mode clearly visible in Event Analysis of Feb. 26th 2008 Florida blackout event. • Mode involves many eastern control areas • Frequency ~ 0.47 Hz, damping ~ +7%, on Feb. 26th 2008 • Likely not related to the blackout. Mode damping at +7% is comparable to the interarea modes in the western system. 34 PSERC Possible Control Actions • Operator Actions Reduce Critical Tie-line Transfers Switch Damping Enhancement Controls at Specific Thyristor Devices – SVC, HVDC • Automatic Control Actions Switching of Damping Controls: Series Capacitors, Shunt Capacitors, Thyristor Devices Generation Rescheduling 35 PSERC SVC Damping Control • Stressed Operating Condition • Near-by tie line active power-flow used as control input • Sending end => Phase Lag Compensator • Receiving end => Phase Lead Compensator SVC P SVC 36 PSERC HVDC Damping Control • Stressed Operating Condition • Phase Angle Difference used as Control Input • Phase Lead Compensator for HVDC Modulation • Effective Improvement in Damping of Interarea Mode for Diverse Levels of AC Power Transfer PAC PDC 37 PSERC OMS Summary • Successful implementation of real-time code into TVA PDC • Advanced signal processing algorithms for oscillation analysis of events and ambient noise • Automatic detection of poorly damped electromechanical modes and their mode shape • Operator alerts, Operator alarms, Control actions, … • Provides early warning on emerging oscillatory problems • Can validate effectiveness and status of PSS at generators when PMU near generator 38 PSERC Future Work • Testing and tuning at TVA • Conversion of OMS code to 64 bit architecture • New dedicated eight processor machine with 32 GB dynamic memory at TVA • OMS Engines for Eastern Grid? • Operator Alerts and Alarms? Operator Actions? • Implementation and testing at BPA and California ISO in collaboration with EPG, Operator Actions? • Implementation at Entergy 39 ...
View Full Document

This note was uploaded on 01/29/2011 for the course ENGR 52 taught by Professor Mcmillan during the Spring '10 term at Baylor Med.

Ask a homework question - tutors are online