Oscillation Monitoring System - Damping Monitor -

Transcription

1 Washington State University Oscillation Monitoring System - Damping Monitor - Mani V. Venkatasubramanian Washington State University 1

2 OMS Flowchart Start Read data from PDC Event? Yes No Damping Monitor Engine FDD analysis for ambient data Prony analysis for postdisturbance data Event Analysis Engine Moving window crosscheck Moving window crosscheck Poorly damped mode detected? Yes Alarm Controller trigger No

3 Voltage (V) Results from Two Engines 5.15 x 10 5 Bus Voltage Magnitude at Cumberland Event Analysis Ambient Noise Analysis 1.2 Hz at +1.8% damping. Local Mode Hz at +1.5% damping. Local Mode Time (s) Nov. 29th 2007 TVA event

4 Complementary Engines Event Analysis Engine (EAE) Multiple algorithms Prony, Matrix Pencil, HTLS, ERA, MFRA, METRA. Aimed at events resulting in sudden changes in damping Damping Monitor Engine (DME) Ambient noise based. Continuous. Provides early warning on poorly damped modes. Several algorithms Fast Frequency Domain Decomposition (FFDD), DFDO, Recursive Adaptive Stochastic Subspace Identification (RASSI).

5 PSD Singular Values from WECC data

6 Mathematical Model for Ambient Data Power system is in fact a high-order nonlinear timeinvariant system However, in normal operating state, power system can be modeled as an LTI system for a short period of time Random load variations modeled as white noise Power System modeled as LTI system Measured outputs, e.g. voltage magnitudes

7 Frequency Domain Decomposition Form signal groups from many PMUs Power spectrum estimation by Multi-Taper Method Apply SVD on the power spectrum estimates Apply inverse FFT on largest singular values Extract the mode frequency and damping ratio by ringdown analysis Singular vectors give mode shapes

8 5.5 Hz mode Actual Current Magnitude seen in PMU (From openhistorian) OMS FDD 5 Hz mode energy level captures the change.

9 Entergy 5 Hz mode Mode frequency changes during some days

10 5.45 Hz mode shape Oscillation detection part of the story. Analysis is rest of the story.

11 5 Hz mode shape different source

12 Stochastic Subspace Identification There is no known input In power systems: random fluctuations in loads x Ax w k 1 k k y Cx v k k k

13 SSI Key Steps Extended observability matrix from PMU data System order => Estimates of A and C Estimate modes and mode shapes SSI-data and SSI-covariance: two different formulations of observability matrix SSI-covariance: Can be made recursive (RSSI) Recursive Adaptive Stochastic Subspace Identification (RASSI)

14 RASSI Adaptive Switching

15 RASSI step test response

16 August 10, 1996 blackout data

17 RASSI Advantages Multi-dimensional recursive algorithm Fast and responsive. Efficient. Modes and mode shapes Can estimate forced oscillations and system modes simultaneously Challenges System order Spurious modes not system modes Data quality issues on recursive methods

18 Apparent Power (MVA) Apparent Power (MVA) 2013 Western system case Time (sec) FDD Damping Ratio 0.5% Time (sec) FDD Damping Ratio 12%

19 FDD Mode Shapes PMU 1 PMU 2 PMU 3 PMU 4 PMU Gen Tie-line a FDD Damping Ratio 0.5% b FDD Damping Ratio 12%

20 Apparent Power (MVA) Apparent Power (MVA) 145 SSI Mode Shapes Time (sec) PMU Time (sec) PMU 2 PMU 3 PMU 4 PMU Gen Tie-line Tie-line a 300 SSI Damping Ratio 0% b SSI Damping Ratio 14% c 300 SSI Damping Ratio 14%

21 Resonance Issues Forced oscillation at a frequency near system mode frequency Resonance can occur when system mode is poorly damped and forced oscillation at a location where the system mode is active Especially problematic for inter-area modes

22 Kundur system simulation Interarea mode damping ratio = 2% 10 MW forced oscillation at any generator results in 400 MW oscillation on tie-lines

23 Forced oscillations at hydro plants Can occur on a routine basis when hydro units come in and out of service 10 to 15 minutes spent in the rough zone Low frequency forced MW oscillations from Francis turbine vortex phenomena Vulnerability for the western system if the MW oscillation frequency near system mode frequency and if the system mode poorly damped Vulnerable locations should be identified and monitored