Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
1
UT Pan Am
Austin
McDonald Observatory
120V wall outlet
120V wall outlet
• Funding provided by EPRI
• Equipment provided by Schweitzer Engineering Labs
• Start-up funds provided by Governor’s Emerging Technology Fund through CCET
• 69kV grid monitor provided by Austin Energy
• Independent. Not affiliated with ERCOT or any utility except Austin Energy
• Will soon have one or more wind farm members
400 miles
300 miles
Wind generation is mostly in West Texas. 7000 MW peak so far this year. Many hours in the spring have 15-20% wind MW, and some have > 20%. Most windy days have wind curtailments.
Observations From the Texas Synchrophasor Network
Monitoring 120V Wall Outlet Voltages and Communicating Through the Public Internet at 30 Readings per Second
Since Spring 2009
Austin
Majority of load is in the green rectangles, i.e. “Central ERCOT”Wind Country
Presentation prepared by Prof. Mack Grady and Graduate Research Assistant Moses Kai, U.T. Austin, for the “Breaking News” Session at the IEEE-PES Annual Meeting, Minneapolis, Monday afternoon, July 26, 2010.
120V wall outlet at U.T. Austin, and 69kV 3Φ monitor at nearby Austin Energy Harris Sub
2
Bus 2DFW
Bus 4SA
Bus 5VAL
Bus 6COAL
Bus 7WIND
300 km
250 km
150 km300 km
400 km
500 km400 km
Bus 3UT
Bus 1HOU
Positive-sequence line constants for each 345kV circuit:
• R = 0.06 Ω/km per conductor
• L = 1 µH/m
• C = 12 pF/m
• Rating = 800 A per conductor
• From the L, XL (500 km) = 188.5 Ω
• For 345 kV, 100 MVA base, Zbase = 1190 Ω
• For each 500 km circuit, XL = 0.158 pu
• Thus, four parallel circuits have XL = 0.040pu, and six parallel circuits have XL = 0.026 pu
Relatively small phase angle variation in central ERCOT
McDonald Observatory
PMU
UT Austin PMU
Bus 2DFWBus 2DFW
Bus 4SA
Bus 4SA
Bus 5VAL
Bus 5VAL
Bus 6COALBus 6COAL
Bus 7WINDBus 7WIND
300 km
250 km
150 km300 km
400 km
500 km400 km
Bus 3UT
Bus 3UT
Bus 1HOUBus 1HOU
Positive-sequence line constants for each 345kV circuit:
• R = 0.06 Ω/km per conductor
• L = 1 µH/m
• C = 12 pF/m
• Rating = 800 A per conductor
• From the L, XL (500 km) = 188.5 Ω
• For 345 kV, 100 MVA base, Zbase = 1190 Ω
• For each 500 km circuit, XL = 0.158 pu
• Thus, four parallel circuits have XL = 0.040pu, and six parallel circuits have XL = 0.026 pu
Relatively small phase angle variation in central ERCOT
McDonald Observatory
PMU
UT Austin PMU
U.T. Pan Am
Synchrophasor Homework Network for ERCOT
3
Thur Fri
West Texas voltage phase angle swings nearly 100º and back with respect to U.T. Austin in about 24 hours
Wind generation and West Texas phase angle can go through large daily swings
Synchrophasor Detective –EVERY day has surprises and clues
McDonald Observatory Voltage Phase Angle with respect to U.T. AustinMarch 25, 26, 2010
-60
-40
-20
0
20
40
60
24 36 48 60 7236 Hours Beginning 12am, March 24, 2010
Deg
rees
Wind Generation in ERCOT - MWMarch 25, 26, 2010
0
1000
2000
3000
4000
5000
6000
24 36 48 60 72
MW
48
4
ERCOT Log. June 23, 2010, 15:20 15:49. Deployed 1150 MW of Responsive Reserve Service from actual adjusted responsive reserve of 3281 MW with an obligation of 2300 MW when frequency dropped to 59.716 Hz due to XXXX and XXXX and XXXX tripped with 1213 MW, issued fleet up of 550 MW. ERCOT load 58446 MW.
Event 1. Large Unit Trip. Freq. Drops to 59.71 Hz. Afternoon of June 23, 2010.
GMT
2-Minute Window of All Four Frequency Readings, 1800 Samples per MinuteThe four
frequency plots are practically identical
5
0.52% drop
0.45% drop
1.78% drop
1.07% drop
U.T. Austin120V wall outlet
Austin Energy Harris Sub 69kV 3Φ positive seq.
McDonald Observatory 120V wall outlet
U.T. Pan American 120V wall outlet
Vrms plot
Vrms plot
Vrms plot
Vrms plot
Event 1. Large Unit Trip, June 23, 2010, cont.
2-Minute Windows of All Four VrmsReadings, 1800 Samples per Minute
120V wall outlet voltages are clearly more noisy than 69kV grid, but nevertheless 120V wall outlets together with Schweitzer relays provide reliable phase angle measurements
6
1-Minute Windows of Phase Angle Ringdowns, 1800 Samples per MinuteEvent 1. Large Unit Trip, June 23, 2010, cont.
• The data stream in from 300 and 400 miles away and are then time synchronized.
• Can you see any difference in the ringdowns observed by the 120V wall outlet on campus and the 69kV grid monitor near campus? (The correct answer is NO!)
• Essentially the only difference is the fixed 90º transformer-related offset plus a 0.3ºpower flow shift through the substation transformer.
• What conclusions can be made about the location of the tripped generator?
Ringdowns Observed atU.T. Austin 120V Wall Outlet
Ringdowns Observed at Austin Energy’s Nearby Harris Sub 69kV (3Φ Positive Sequence)
U.T. Pan Am U.T. Pan Am
McDonald Observatory McDonald Observatory
7
Frequency at the Four PMU Locations
U.T. Pan Am Phase Angle w.r.tU.T. Austin
• Insignificant frequency event – too small to be logged.
• But very significant response in South Texas voltage phase angle.
• Illustrates the sensitivity of synchrophasors.
• Shows that South Texas was importing P because the phase angle dropped.
1-Minute Windows
Event 2. Small Unit Trip in South Texas. Morning of July 18, 2010.
8
ERCOT Log. Frequency dropped to 59.920 Hz due to wind generation dropped 1400 MW after off-set was calculated. Issued fleet up of 400 MW due to out of Up Regulation. ERCOT load 35153 MW.
Event 3. Wind MW Backdown Invoked. Early Morning of July 15, 2010. Thevenin Impedance Calculation.
90-Minute WindowsTotal Generation, MW
0
10000
20000
30000
40000
4.0 4.5 5.0 5.5Hour (CDT)
Wind Generation, MW
0
1000
2000
3000
4000
5000
4.0 4.5 5.0 5.5Hour (CDT)
McDonald Voltage Phase Angle wrt U.T. Austin, Degrees
-10
0
10
20
30
4.0 4.5 5.0 5.5Hour (CDT)
9
Bus 7WINDBus 7WIND
500 km
Local conventional Pload,conv − Pgen,conv
Pexport
Thevenin Impedance jXTH
Central ERCOT
Vangle = 0
Vangle = δ
Event 3. Wind MW Backdown. July 15, 2010. Thevenin Impedance Calculation, cont.
Use the Excel Solver with the 90 minutes of readings tominimize least-squared error and determine Xth
10
Event 3. Wind MW Backdown. July 15, 2010. Thevenin Impedance Calculation, cont.
The knowns (all are given in one-minute average): • Wind generation MW • Wind generation phase angle δ with respect to central ERCOT (i.e., U.T. Austin) • ERCOT total generation
The unknowns
• jXTH • Local conventional load minus generation in West Texas wind country • Pexport
The equations
• Ignoring losses, power balance requires that
( ) )sin(,, δTH
ERCOTwindconvgenconvloadwind X
VVPPP
•=−−
Other assumptions • Having access to no other information, assume that local net conventional ( convgenload PP ,− )
varies with ERCOT total generation according to BPAPP GenTotalERCOTconvgenconvload +•=− ,,
Excel Solver Setup
• Define )(
11puXX
VVC
THTH
ERCOTwind •=
•=
• Instruct Excel Solver to vary coefficients A, B, and C to minimize the sum of squared error
( )[ ]∑=
−−−N
nconvgenconvloadwind CPPP
1
2,, )sin(δ ,
n = minutes, for either the entire interval
The balance equation: want left-hand side ≈ right-hand side for all 90 readings
Three unknowns
11
West Texas Power Export Estimates Pexp1 and Pexp2, MW
-1000
0
1000
2000
3000
0 10 20 30 40 50 60 70 80 90
Minute Beginning 0400 CDT, July 15, 2010
90-Minute Window
Event 3. Wind MW Backdown. July 15, 2010. Thevenin Impedance Calculation, cont.
Squared area between the two curves is minimized when
A = -0.0471
B = 356
Xth = 0.0211 pu on 345kV, 100MVA base
12
• A unit trip causes a drop of 0.06 Hz, which was not significant enough to be noted on ERCOT’s Daily Grid Operation Report.
• However, the phase angle ringdown at McDonald Observatory gives a clear indication of the damped resonant frequency and normalized damping coefficient between West Texas wind country and central ERCOT.
Event 4. Small Unit Trip. Early Afternoon of July 16, 2010. Ringdown Analysis.
2-Minute Window
13
2-Minute Window
Event 4. Small Unit Trip. July 16, 2010. Ringdown Analysis, cont.
14
McDonald Observatory Voltage Phase Angle w.r.t U.T. Austin 120V and Harris 69kV
-14
-13
-12
-11
-10
1600 1700 1800 1900 2000 2100 2200
Sample (30 samples per second)
Deg
rees w.r.t U.T. Austin
w.r.t Harris 69kV
10 seconds
Event 4. Small Unit Trip. July 16, 2010. Ringdown Analysis, cont.
• The fixed net multiple of 30 degree phase shift between U.T. Austin 120V and Harris 69kV has been removed. The variable but steady 0.1 degree power flow phase shift through the substation transformer has also been removed.
• Can you see any significant difference between the waveform seen by the 120V wall outlet (in black) and the waveform seen by the 69kV, 3Φ grid monitor (in red)? (The correct answer is NO!)
Voltage Ringdown at McDonald Observatory Observed at the Following Two Locations in Austin: a 120V Wall Outlet on Campus, and the Harris 69kV Substation that Feeds the Campus
15
U.T. Pan Am Voltage Phase Angle w.r.t U.T. Austin 120V and Harris 69kV
-40
-39
-38
-37
-36
-35
-34
-33
-32
1600 1700 1800 1900 2000 2100 2200
Sample (30 samples per second)
Deg
rees w.r.t U.T. Austin
w.r.t Harris 69kV
10 seconds
• The fixed net multiple of 30 degree phase shift between U.T. Austin 120V and Harris 69kV has been removed. The variable but steady 0.1 degree power flow phase shift through the substation transformer has also been removed.
• Can you see any significant difference between the waveform seen by the 120V wall outlet (in black) and the waveform seen by the 69kV, 3Φ grid monitor (in red)? (The correct answer is NO!)
Event 4. Small Unit Trip. July 16, 2010. Ringdown Analysis, cont.
Voltage Ringdown at U.T. Pan Am Observed at the Following Two Locations in Austin: a 120V Wall Outlet on Campus, and the Harris 69kV Substation that Feeds the Campus
16
Estimated Normalized Damping Ratio ξ = 0.088
McDonald Voltage Phase Angle wrt AustinMeasured Response is Dotted Line,
Ideal 2nd-Order Approximation is Solid Line
-14
-13
-12
-11
-10
0 1 2 3 4 5 6 7
Seconds
1.84sec, 0.54 Hz, 3.39 rad/sec
Event 4. Small Unit Trip. July 16, 2010. Ringdown Analysis, cont.
17
Low Wind
High Wind
McDonald Observatory w.r.t. UT Austin Phase Angle Oscillations
Low Wind Generation Period, Feb. 27, 2010, 1- 3 pm
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0.00.10.10.20.20.30.3
Normalized Damping Ratio
Dam
ped
Res
onan
t Fre
quen
cy
McDonald Observatory w.r.t. UT Austin Phase Angle OscillationsHigh Wind Generation Period, Feb. 28, 2010, 1- 3 pm
0.0
0.5
1.0
1.5
2.0
2.5
3.0
00.050.10.150.20.250.3
Normalized Damping RatioD
ampe
d R
eson
ant F
requ
ency
More Observations
From Last Spring: Modal analysis of ambient oscillations show that high wind causes tighter clustering of damped resonant frequency points
unstable
unstable0.3 0.25 0.2 0.15 0.1 0.05
18
Ringing Frequency vs Wind Generation (% of Total), Sep 2009 ‐ Feb 2010
0
0.5
1
1.5
2
2.5
3
0 2 4 6 8 10 12 14 16 18 20
Wind Generation (% of Total)
Freq
uency (Hz)
Ringdown Analysis of More Than 100 Unit Trips Yields No Clear Relationship Between Wind MW and Damped Resonant Frequency
More Observations, cont.
19
Damping Ratio vs Wind Generation (% of Total), Sep 2009 ‐ Feb 2010
0
0.2
0.4
0.6
0.8
1
1.2
0 2 4 6 8 10 12 14 16 18 20
Wind Generation (% of Total)
Normalized
Dam
ping
Ratio
Ringdown Analysis of More Than 100 Unit Trips Yields No Clear Relationship Between Wind MW and Normalized Damping Ratio
More Observations, cont.
20
What Have We Learned Thus Far?1. After knowing “are the lights on everywhere,” voltage phase angles
across a grid (i.e., synchrophasors) are arguably the next most important grid diagnostic measurement.
2. While voltage synchrophasors are valuable in steady-state analysis, they are far more valuable in observing the dynamic performance of a grid, e.g. damped resonant frequencies and normalized damping coefficients that indicate grid stress and stability.
3. Because voltage synchrophasors are very sensitive to grid disturbances, they can be thought of as the “EKG” of power grids. They quickly point out abnormalities not easily seen in conventional voltage, frequency, and power measurements. Grid “stress tests”come frequently, each time a generator trips off line.
4. If Steinmetz had known about GPS time stamping in the early 1900’s, synchrophasors would have been in common use for the past 50-60 years.
21
What Have We Learned Thus Far, continued ?
5. Excluding redundancy requirements, ten or so strategically-placed synchophasor units in a grid the size of ERCOT are adequate.
6. 120V wall outlets on distribution feeders have proven themselves to provide essentially the same results as transmission voltage measurements.
7. The three reasons that 120V wall outlets are suitable for synchrophasor purposes are that
• Grid oscillations are in the 0.5 to 2 Hz range and readily pass through transformers of all sizes,
• Schweitzer relays effectively filter out distribution noise,
• With 30 points streaming in each second from remote monitoring points, occasional dropouts due to deep voltage sags or internettraffic cause no problems.
22
Texas Synchrophasor Network
Thanks to
• Schweitzer Engineering Laboratories – for all the equipment and technical support that we have and will need
• EPRI – for past, present, and future funding of graduate students and faculty summer support
• Startup money in 2008 from the Texas Governor’s Emerging Technology Fund through CCET
• Austin Energy – for installing the 69kV grid monitor, and providing advice on system operating and protection
Dr. Mack Grady, Mr. Moses Kai, IEEE-PES Breaking News, Minneapolis, July 26, 2010
23
Biographies
William Mack Grady is a Professor of Electrical & Computer Engineering at U.T. Austin and a Fellow of IEEE for “Contributions to the Analysis and Control of Power System Harmonics and Power Quality.” He received his BSEE from U.T. Arlington, and MSEE and PhD from Purdue University. His research topics are power quality, grid studies, synchrophasors, and integration of renewable energy into the grid. His research sponsors are EPRI, Schweitzer Engineering Labs, Austin Energy, DOE, and NREL.
Moses Kai is a Graduate Research Assistant at U.T. Austin, working in the area of synchrophasors and impact of large-scale wind generation on ERCOT. He received his BSEE and MSEE from U.T. Austin in 2007 and 2009, respectively, and is now working toward his PhD. He also serves as a Teaching Assistant for the Power Electronics Lab course, which has an enrollment of more than 130 students per year. He is a student member of IEEE and a U.S. citizen.