Governor Response Modeling
Presented at WECC MVWG Meeting
May 2018
Dmitry Kosterev
BPA
1
History
• July 2 and August 10 1996 events became WSCC (WECC) defining moments, just as August 14 2003 was NERC defining moment
• Dynamic model failed to reproduce the actual system oscillation during August 10, 1996 outage (on right)
• Validation studies required many model adjustments, including blocking thermal governors to reproduce system frequency drop, COI response, and to some extent oscillations
• WECC staff developed “OCSGOV.p” EPCL to block thermal governors in transient stability simulations when its active power output above 90% of the turbine capacity
2
History
Late 1990s:
• WECC started studying 2-unit outage as a credible contingency
• 2 Palo Verde outage became a limiting contingency for COI, voltage-
stability limited due to governor powerflow pick-up
• BPA expanded and deployed synchrophasor measurements
• BPA started baselining governor response, and found that COI pick up to
be about 50 to 55% of generation lost in the south
• WECC base cases at that time assumed that all generators are frequency
responsive, and COI pick-up was only 35%
• BPA and CAISO staff started blocking governors in post-transient
governor powerflow studies by providing a list of “blocked” generators.
The list was “tuned” to produce at least 50% COI pick-up for Palo Verde
outages
• Utilities continued to use “OCSGOV.P” EPCL to block thermal governors
in dynamic simulations
3
Dynamic Response
(from 0 to up to 60 sec)
angular stability, oscillation
damping, frequency
response
Post-Transient Response
(a snapshot at 1 to 5 min),
post-transient voltage stability,
transmission thermal loading
Study Timeframes
4
History
Early 2000s:
• WECC initiated efforts to establish a Frequency Responsive Reserve (FRR) standard
– California energy crisis, brown-outs because of insufficient reserves
– BPA wanted to achieve equitable frequency response distribution across the Western Interconnection
• FRR studies were planned to determine the required amount
The rest of FRR story
• WECC had several attempts to develop its own FRR standard, but came up short every time
• FERC directed NERC to develop a standard addressing frequency response in 2010
• NERC had Frequency Response Initiative work that led to the expedited development, approval and implementation of NERC BAL-003-1 Frequency Response Standard in 2013
5
History
Early 2000s:
• WECC conducted governor response tests on May 18 2001
– AGC was disabled by BAs in the West
– Hoover and Grand Coulee generators were tripped
– Model validation studies once again showed deficiencies in the
frequency response modeling
• This led to the development of Thermal Governor Modeling
Recommendation in WECC:
• Les Pereira, John Undrill, Dmitry Kosterev, Donald Davies and Shawn
Patterson, “A New Thermal Governor Modeling in WECC,” IEEE
Transactions on Power Systems, vol.18, no.2, pp.819-829, May 2003.
https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=1198319
6
Thermal Governor Modeling
• Generator Owners used historic SCADA
data to classify generators as (a)
frequency responsive, (b) on-load
control, (c) baseloaded
• Load control model “lcfb1” was
developed, as well as a new thermal
governor model “ggov1”, which was a
better representation of gas-turbine
• Software programs added
“baseload_flag” to generator records
• Workshops were held to assist generator
owners with data preparation
• Dynamic model validation studies were
performed for multiple event
• The governor blocking list was updated
for post-transient studies7
Frequency Response Types
Frequency
Responsive
Load Control
Baseloaded
8
Model Validation Studies for June 14 2004 Event
Model accuracy improved
9
Blue = Actual
Red = Simulations
Dynamics vs. Post-Transient
Dynamics Post-Transient
Time Frame 0 up to 60 seconds a snapshot of system response
at 1 to 5 minutes
Frequency Responsive
Generator (BL=0)
Governor model is active Governor response is allocated
in proportion to generator’s
Pmax
Generator under Load
Control (BL=0)
Governor and load
controller models are active
No governor response
Frequency
unresponsive
generator (BL=1 or 2)
1= Governor is limited in
upward direction
2 = Governor is limited in
upward and downward
directions
No governor response
BL = “baseload” flag in powerflow base case10
Baseload Flag of 1 or 2
11
What is the difference ?
• Flag “1” or “2” will not respond to frequency drop below nominal
• Flag “1” will respond to frequency increase above nominal
• Flag “2” will not respond to frequency increase
Why is this important ?
• Blocking thermal governors affects damping on inter-area oscillations
• A contingency of BC-Alberta separation (with high flows from BC to Alb) will result in system frequency increase and will start power oscillations, such as occurred on August 4, 2000
WECC MVWG developed a report in 2012 recommending using baseload flag of “2” for interconnection-level studies
https://www.wecc.biz/Reliability/WECC%20MVWG%20Thermal%20Governor%20Model%20Revision%202012-06-20.pdf
Fast Forward to Today
• Western Interconnection has experienced addition of significant amount of wind and solar generation, and some amount of gas-fired generation
• Loss of institutional knowledge – retirements and promotions
Post-transient governor powerflow
• BPA continues baselining system frequency response
• Actual COI pick-up is now about 45% because of the system changes on the east side
• Simulated COI pick-up with off-the shelf case is less that 30% in post-transient governor powerflow
– About 12 GW of wind and solar generators are modeled as frequency-responsive
– Program conversion issues
– HRSG in combined cycle plants are often represented as frequency-responsive
• BPA “tunes” base cases for historic governor response performance
12
Fast forward to Today
Dynamic Simulations
• West-wide System Model and breaker-node modeling in GE PSLF enabled much more frequent verification of operating models
• WECC staff developed better tools to conduct periodic verification of planning models
• System provides plenty of events for model validation studies
• Studies show reasonable correspondence between simulations and reality
13
May 31, 2013 event
https://www.wecc.biz/Reliability/May%2030%202013%20Model%20V
alidation%20Report.pdf
May 16, 2014 event
https://www.wecc.biz/Reliability/Model%20Validation%20Report%
20for%20May%2016%202014%20RAS%20Event.pdf
August 18, 2016 event
https://www.nerc.com/pa/rrm/ea/1200_MW_Fault_Induced_Solar
_Photovoltaic_Resource_/1200_MW_Fault_Induced_Solar_Photov
oltaic_Resource_Interruption_Final.pdf
Moving Forward
Let me propose to form a Sub-Task Team under PPMVTF to
develop processes and tools to ensure we have sufficiently
adequate governor response representation in dynamic
simulations and post-transient voltage stability
MVWG had some good ideas:
- Revisions to thermal governor modeling proposed back in
2012
- Model validation studies for July 4 2012, May 31 2013, May 16
and 26 2014 large generation drop events
Let’s make sure we will follow up this time….
14