Acrobat

September 18, 0430 hrs following switching of Cumberland-Marshall 500 kV line
(Jville - Gleeson out)
Cumberland Fossil Plant
TVA's largest coal plant, completed in 1973
Plant can output 2,700 MW
Two cross-compound units- four 722 MVA generators
Unit 1 - original Unitrol B analog exciter, SCRX model
Unit 2 - exciter upgraded to Unitrol P in 2003, ESST1A
model first applied in Sept 2006.
Instability
Plant operated under Operating Guide
Previous event Nov 4 2000, + 600 MW oscillation, following
switching of Cumberland-Davidson, pre-outages of CumberlandMarshall, Weakley-Shelby, Maury-Davidson, Jville bank. Not
recorded, not reproduced by PSS/E
Event of Monday Sep 18, 2006
Plant at full output (>2,600 MW)
Pre-outage of Jville-Gleason
Oscillation followed switching of Cumberland-Marshall for
maintenance
Operator reclosed line after ~3 minutes
500 kV Transmission System Around Cumberland Fossil Plant
Cumberland
500 kV Line Outages During Sept 18, 2006 Oscillations
Switched line
which triggered
oscillations
WBN out and
Bus split
First record - Dataware system
3 seconds per sample
Why did the plant oscillate?
Typical of an exciter-driven event
Not shown by PSS/E study with recorded settings oscillation only with large change in gain settings (later)
Possibilities:
- exciter setting incorrectly applied
- exciter settings or models incorrect
- circuit error
- inadequate modeling
- other? (evil spirits?)
Powerflows at Cumberland 500 kV Switchyard prior to Sept 18 event.
Davidson FP 500 kV
1350 MW, -106 MVAR
1321 MW, -67 MVAR
Johnsonville FP 500 kV
-217 MW, 41 MVAR
1332 MW, 178 MVAR
1279 MW, 149 MVAR
Marshall 500 kV
~200 MW initial
Magnitude Swing
~680 MW max
Magnitude Swing
PMU record
30 samples per second
First Attempt: Simulation of Sept 18 Cumberland
switching event
<70 MW
Magnitude Swing
This simulation shows no damping problems.
Cumberland with Exciter gain set too high
Unit 1 K= 1000 (from 100) Unit 2 Ka=1000 (from 450)
Oscillates @ ~1.25 Hz
400 MW
Magnitude Swing
This simulation shows that incorrect Exciter settings could have caused the
oscillations. The decision was made to do testing on Cumberland’s exciters.
TVA
Investigation of Cumberland
Generator Data
Investigation of Generator data.
Manufacturer’s data did not provide values for T’q0 and T’’q0
This data was marked “typical data” and did not indicate
where the values came from.
T’q0 and T’’q0 immediately became suspect values. We
needed to determine if these values were out of normal
ranges.
How might we accomplish this?
Investigation of Generator data.
To accomplish this we looked at all similar generators in
MMWG dynamics case
Similar values based on MVA Base, reactances, and direct
axis time constants.
Investigation of Generator data.
Conclusion: T’’q0 appears to be within normal ranges while
T’q0 is below the expected range.
Simulations indicate that raising T’q0 lessens damping at
Cumberland Fossil Plant,Leonardo
but as long as ranges are within
reason the oscillations still damp out for the Sept 18
switching event.
Epilogue
Long-term undetected error in generator model
Instability still not exactly reproducible - criteria issue
PSS installed on both generators of one unit
Limited support from manufacturer