NPRR351 Look-Ahead SCED Update

23
NPRR351 Look-Ahead SCED Update Hailong Hui/Sai Moorty ERCOT METF August 30, 2012

description

NPRR351 Look-Ahead SCED Update. Hailong Hui/Sai Moorty ERCOT METF August 30, 2012. Outlines. RTD Observation and False Price Spikes RTD Known Issues Analysis for False Price Spike at xx:55 and xx:00 Analysis for False Price Spike at xx:05 New Fix for the False Price Spike at xx:05 - PowerPoint PPT Presentation

Transcript of NPRR351 Look-Ahead SCED Update

Page 1: NPRR351 Look-Ahead SCED Update

NPRR351 Look-Ahead SCED Update

Hailong Hui/Sai MoortyERCOT

METFAugust 30, 2012

Page 2: NPRR351 Look-Ahead SCED Update

2

Outlines

• RTD Observation and False Price Spikes• RTD Known Issues• Analysis for False Price Spike at xx:55 and xx:00• Analysis for False Price Spike at xx:05• New Fix for the False Price Spike at xx:05• Case Study for the New Fix for xx:05 Price Spike

METFAugust 30, 2012

Page 3: NPRR351 Look-Ahead SCED Update

3

RTD Observations and False Price Spikes

• ERCOT monitors and analyzes RTD price abnormality• RTD LMP forecast is more accurate for the first 30 minutes of

the study period (currently study period is 55 minutes – configurable)

• Sometimes RTD has forecasted false price spikes. These false price spikes have typically occurred at:– First interval after hour boundary. i.e. xx:05– Last two intervals before the hour boundary. i.e. xx:55 and xx:00

• Other false price spikes have occurred at random intervals where the Load Forecast has significant error.

METFAugust 30, 2012

Page 4: NPRR351 Look-Ahead SCED Update

4

False Price Spike at xx:50, xx:00 and xx:05

METFAugust 30, 2012

Page 5: NPRR351 Look-Ahead SCED Update

5

False Price Spike at Other Intervals

METFAugust 30, 2012

Page 6: NPRR351 Look-Ahead SCED Update

6

RTD Known Issues-Short-Term Load Forecast Error

• Errors in Load Forecast have sometimes contributed to false price spikes .

• Improvements to Load Forecast in the works:– Mid-Term / Short Term Load Forecasting Project currently

underway (October go-live for MTLF in production in non-high availability mode)

• Implement new MTLF models (hourly forecast for the next 7 days)

• Ability to create sub-hourly models (15-minute, 5-minute)

• Ability to blend the above models (example use the 5-minute model for the next several hours, then use a blend of the hourly model and 5-minute model for the next few hours)

• Ability to create new STLF model(s) and evaluate accuracy of Load Forecast. The new model(s) could take price as additional input to predict effect of passive demand response on the Load Forecast.

METFAugust 30, 2012

Page 7: NPRR351 Look-Ahead SCED Update

7

RTD Known Issues-Resource Status Used in Study Period

RTD current implementation uses telemetry resource status for all the intervals within current hour and COP status for all the intervals within next hour.

•Some Resources did not follow their COP schedules in real-time.

• It has been observed that some units may remain online if the real-time price is high, when their COP status indicate them to be offline. These units choose to self-commit in real-time and most of them are SCLE90, SCGT90 and GSREH with fast startup time.

• It has also been observed that some units may stay offline in real-time even though their COP status indicate that they plan to be online.

•RTD simulates a shutdown for a Generation Resource when the current telemetry status is online and the next hour COP status is offline. This shutdown process forces the Generation Resource to be offline in the first 5 minute interval after the hour boundary. This has been observed to be too abrupt especially when the hour boundary is the first interval of the RTD study period.

METFAugust 30, 2012

Page 8: NPRR351 Look-Ahead SCED Update

8

RTD Known Issues-Resource Status Used in Study Period

• Telemetry resource status at time of RTD run may not reflect all the intervals status. e.g., a generator may startup or shutdown during the middle of the RTD study period but the current implementation persists telemetered resource status for all intervals in current hour.

• COP status is on hourly basis and does not provide enough granularity. The startup and shutdown in future intervals based on COP status may not reflect what actually occurs when those future intervals become real-time.

METFAugust 30, 2012

Page 9: NPRR351 Look-Ahead SCED Update

9

Analysis for False Price Spike at xx:55 and xx:00

• This false price spike at xx:55 and xx:00 occurs typically for RTD run at the top of the hour (xx:00, xx:05)

• For RTD runs at xx:00 and xx:05, the RTD study period only covers current hour and RTD uses current telemetry resource status for all the intervals. The telemetry resource status as a proxy for the tail end of the RTD study period may be obsolete when it comes to real-time

• Use IE 08/06/2012 12:00 as an example, RTD run at 11:05 and IE 12:00 is the last interval (interval_id=11).

– Total 26 units telemetered offline at 11:05 but telemetered online at 11:55 with total telemetry MW of about 1000 MW. 17 of them have offline COP status for both current hour and next hour.

– RTD run at 11:05 treated these 26 units as offline for current hour including IE 12:00 but actual SCED run at 11:55 treated these units as online. The RTD run at 11:05 saw less online capacity for IE 12:00 and it caused under generation and price spike for IE 12:00.

METFAugust 30, 2012

Page 10: NPRR351 Look-Ahead SCED Update

10

Analysis for False Price Spike at xx:55 and xx:00

• This price spike normally disappears when the RTD run is more close to the end of hour since the telemetry status at RTD run is more accurate

• Need more analysis and discussion for this issue. Possible options:

– Startup the units in the middle of hour if current hour COP is online

– Startup the units in current hour if next hour COP is online

– Incorporate some PRC type calculation for future intervals to simulate additional Resource commitment

METFAugust 30, 2012

Page 11: NPRR351 Look-Ahead SCED Update

11

Analysis for False Price Spike at xx:05

• Use IE 08/06/2012 19:05 as an example, RTD run at 18:55 where IE 19:05 is the 2nd interval (interval_id=2).

• At 19:00, 27 units telemetry are online while COP status indicate they are offline for HE20:

– 16 units remain online for the entire hour, 7 of the 16 units are QSGR, also 15 of the 16 units are SCLE90 or GSREH type. Note that the current RTD QSGR logic is: If a QSGR current telemetry is online then RTD will assume the QSGR will stay online in next hour even if the next hour COP status is offline.

– 11 units were shutdown in the middle of the hour to follow their COP offline status, 1 QSGR and 7 of the 11 units are SCLE90 or GSREH type.

• According current implementation, RTD abruptly shutdown the 19 non-QSGR units total about 2900MW at IE 19:05. This caused under generation at IE 19:05. Also IE 19:05 is 2nd interval (for RTD run at 18:55) and other online units can not be fully ramped up to meet the power balance due to their ramping constraints.

METFAugust 30, 2012

Page 12: NPRR351 Look-Ahead SCED Update

12

New Fix for False Price Spike at xx:05

• RTD Pre-processing Heuristic Rules changes:– Change COP shutdown profile to make the shutdown process more

gradually, i.e. the unit will be ramp down to LASL first before it can be shut down

– If the RTD study period covers both the current hour and the next hour, then do the following : If unit current telemetry status is online and both current and next hour COP status is offline, then assume that unit is online for all the study intervals within the next hour.

– Add SCLE90,SCGT90 and GSREH resources to the QSGR list and apply the same QSGR logic for them.

• Current RTD QSGR logic is: If a QSGR current telemetry is online then RTD will assume the QSGR will stay online in next hour even if the next hour COP status is offline.

• The New Fix is in the process of being released to the testing environment and will be migrated to Production after testing.

METFAugust 30, 2012

Page 13: NPRR351 Look-Ahead SCED Update

13

Case Study for the New Fix for xx:05 Price Spike

• Rerun some RTD savecases having false price spike at xx:05 with the new fix and compare the new results against the original RTD results

– 08/06/2012 IE 19:05

• RTD_201208061855011 (interval_id=2)

• RTD_201208061810011 (interval_id=11)

– 08/06/2012 IE 21:05

• RTD_201208062055011 (interval_id=2)

• RTD_201208062010011 (interval_id=11)

– 08/09/2012 IE 21:05

• RTD_201208092055011 (interval_id=2)

• RTD_201208092010021 (interval_id=11)

• The study results shows that the false price spike at xx:05 has been reduced to normal prices with the new fix

METFAugust 30, 2012

Page 14: NPRR351 Look-Ahead SCED Update

14

08/06/2012 IE 19:05 and 21:05 (interval_id=2)

METFAugust 30, 2012

Page 15: NPRR351 Look-Ahead SCED Update

15

RTD_201208061855011 Results-New VS Original

METFAugust 30, 2012

IntervalEnding

IntervalID

RTDGTBD

UnderGenOrig

System_LambdaOrig

LZ_NOrig

LZ_SOrig

LZ_HOrig

LZ_WOrig

UnderGenNew

System_LambdaNew

LZ_NNew

LZ_SNew

LZ_HNew

LZ_WNew

19:00 1 59982 0.00 37.50 35.47 34.66 33.22 86.89 0.00 37.50 35.47 34.66 33.22 86.89

19:05 2 59938 523.91 4501.00 4508.37 4498.66 4484.84 4610.19 0.00 38.06 35.97 35.23 33.84 87.25

19:10 3 59715 40.00 2149.00 2146.55 2154.78 2138.04 2275.79 0.00 37.98 35.89 35.16 33.77 87.18

19:15 4 59488 30.00 765.88 763.44 771.67 754.93 892.68 0.00 37.72 35.61 34.90 33.53 86.78

19:20 5 59257 30.00 774.09 771.64 779.87 763.14 900.88 0.00 37.54 35.61 34.72 33.13 87.65

19:25 6 59021 24.56 456.40 453.96 462.19 445.45 583.20 0.00 37.22 35.24 34.39 32.88 86.94

19:30 7 58782 5.00 264.29 265.36 267.15 251.46 384.77 0.00 36.85 34.82 34.02 32.56 86.22

19:35 8 58538 0.00 73.65 82.74 78.47 60.11 142.02 0.00 36.51 34.45 33.70 32.26 85.68

19:40 9 58290 0.00 65.66 75.23 66.88 52.99 133.02 0.00 36.24 34.16 33.44 32.02 85.23

19:45 10 58037 0.00 55.37 62.54 56.18 45.96 104.90 0.00 35.30 33.55 32.96 31.73 76.59

19:50 11 57781 0.00 54.59 61.69 55.37 45.28 103.48 0.00 33.64 32.71 32.32 31.43 58.31

Page 16: NPRR351 Look-Ahead SCED Update

16

RTD_201208062055011 Results-New VS Original

METFAugust 30, 2012

IntervalEnding

IntervalID

RTDGTBD

UnderGenOrig

System_LambdaOrig

LZ_NOrig

LZ_SOrig

LZ_HOrig

LZ_WOrig

UnderGenNew

System_LambdaNew

LZ_NNew

LZ_SNew

LZ_HNew

LZ_WNew

21:00 1 56713 0.00 41.70 44.14 40.53 37.18 69.05 0.00 41.70 44.14 40.53 37.18 69.05

21:05 2 56408 200.00 4397.37 4418.12 4386.24 4381.67 4410.23 0.00 43.33 46.67 42.29 38.14 71.96

21:10 3 56180 30.00 782.60 811.70 770.38 751.67 890.61 0.00 37.66 38.46 36.27 34.27 63.62

21:15 4 55940 30.00 691.64 721.88 680.45 661.81 783.68 0.00 34.77 34.25 33.20 32.31 59.32

21:20 5 55687 10.00 307.73 332.49 296.23 278.17 429.33 0.00 34.03 32.97 32.38 31.99 57.65

21:25 6 55423 0.00 63.22 71.48 62.27 54.12 100.34 0.00 33.53 32.36 31.88 31.56 57.17

21:30 7 55146 0.00 42.25 44.09 40.98 38.35 67.08 0.00 32.96 31.63 31.29 31.09 56.56

21:35 8 54857 0.00 36.69 35.91 35.04 34.60 59.25 0.00 32.56 31.16 30.88 30.72 56.25

21:40 9 54556 0.00 35.43 34.40 33.76 33.46 58.19 0.00 33.06 31.72 31.39 31.21 56.55

21:45 10 54242 0.00 34.43 33.21 32.75 32.56 57.36 0.00 32.65 31.26 30.98 30.81 56.30

21:50 11 53917 0.00 33.93 32.66 32.24 32.06 57.06 0.00 32.19 30.76 30.51 30.35 56.03

Page 17: NPRR351 Look-Ahead SCED Update

17

08/06/2012 IE 19:05 and 21:05 (interval_id=11)

METFAugust 30, 2012

Page 18: NPRR351 Look-Ahead SCED Update

18

RTD_201208061810011 Results-New VS Original

METFAugust 30, 2012

IntervalEnding

IntervalID

RTDGTBD

UnderGenOrig

System_LambdaOrig

LZ_NOrig

LZ_SOrig

LZ_HOrig

LZ_WOrig

UnderGenNew

System_LambdaNew

LZ_NNew

LZ_SNew

LZ_HNew

LZ_WNew

18:15 1 61401 0.00 38.80 36.85 36.02 34.53 87.53 0.00 38.80 36.85 36.02 34.53 87.53

18:20 2 61627 0.00 39.65 37.82 36.88 35.22 89.25 0.00 39.65 37.83 36.89 35.23 89.24

18:25 3 61488 0.00 39.33 37.44 36.56 34.99 88.47 0.00 39.28 37.39 36.51 34.95 88.41

18:30 4 61339 0.00 39.05 37.10 36.28 34.80 87.78 0.00 38.99 37.04 36.22 34.74 87.74

18:35 5 61181 0.00 38.90 36.90 36.12 34.69 87.41 0.00 38.81 36.85 36.03 34.57 87.47

18:40 6 61014 0.00 38.71 36.72 35.93 34.50 87.21 0.00 38.56 36.57 35.77 34.35 87.01

18:45 7 60837 0.00 38.42 36.39 35.64 34.27 86.64 0.00 38.24 36.20 35.46 34.10 86.37

18:50 8 60652 0.00 38.10 35.91 35.37 34.08 85.98 0.00 37.88 35.74 35.11 33.84 85.62

18:55 9 60458 0.00 37.85 35.61 35.15 33.88 85.61 0.00 37.56 35.39 34.78 33.56 85.05

19:00 10 60254 0.00 42.17 41.12 39.53 36.73 97.33 0.00 37.30 35.13 34.51 33.31 84.67

19:05 11 60091 203.97 4501.00 4491.66 4501.99 4483.55 4720.93 0.00 38.62 36.84 35.87 34.12 88.67

Page 19: NPRR351 Look-Ahead SCED Update

19

RTD_201208062010011 Results-New VS Original

METFAugust 30, 2012

IntervalEnding

IntervalID

RTDGTBD

UnderGenOrig

System_LambdaOrig

LZ_NOrig

LZ_SOrig

LZ_HOrig

LZ_WOrig

UnderGenNew

System_LambdaNew

LZ_NNew

LZ_SNew

LZ_HNew

LZ_WNew

20:15 1 57342 0.00 42.07 42.58 39.17 33.34 112.55 0.00 42.07 42.58 39.17 33.34 112.55

20:20 2 57474 0.00 45.57 47.37 42.74 35.55 123.39 0.00 45.60 47.36 42.77 35.54 123.37

20:25 3 57423 0.00 45.01 46.65 42.19 35.22 121.77 0.00 44.51 45.95 41.67 34.89 120.17

20:30 4 57365 0.00 44.40 45.77 41.57 35.03 119.26 0.00 43.30 44.28 40.46 34.40 115.74

20:35 5 57298 0.00 41.91 42.28 39.09 34.18 109.53 0.00 41.20 41.30 38.36 33.73 107.30

20:40 6 57223 0.00 40.43 40.32 37.59 33.31 105.03 0.00 40.20 39.91 37.36 33.31 103.65

20:45 7 57141 0.00 39.94 39.60 37.10 33.12 103.04 0.00 39.68 39.14 36.83 33.10 101.54

20:50 8 57050 0.00 39.84 39.54 36.99 32.95 103.26 0.00 37.67 35.94 34.78 32.60 92.00

20:55 9 56950 0.00 37.86 36.26 34.97 32.63 92.99 0.00 37.26 35.37 34.37 32.41 90.54

21:00 10 56843 0.00 46.53 48.14 43.74 37.22 122.01 0.00 37.42 35.46 34.52 32.63 90.33

21:05 11 56727 583.32 4501.00 4522.28 4479.98 4461.71 4724.23 0.00 65.44 67.43 62.67 55.64 143.37

Page 20: NPRR351 Look-Ahead SCED Update

20

08/09/2012 IE 21:05 (interval_id=2)

METFAugust 30, 2012

Page 21: NPRR351 Look-Ahead SCED Update

21

RTD_201208092055011 Results-New VS Original

METFAugust 30, 2012

IntervalEnding

IntervalID

RTDGTBD

UnderGenOrig

System_LambdaOrig

LZ_NOrig

LZ_SOrig

LZ_HOrig

LZ_WOrig

UnderGenNew

System_LambdaNew

LZ_NNew

LZ_SNew

LZ_HNew

LZ_WNew

21:00 1 58789 0.00 40.00 37.06 37.41 36.73 87.00 0.00 40.00 37.06 37.41 36.73 87.00

21:05 2 58620 1089.92 4501.00 4501.00 4501.00 4501.00 4501.00 0.00 39.04 37.84 38.04 37.38 61.11

21:10 3 58409 722.36 4501.00 4501.00 4501.00 4501.00 4501.00 0.00 37.43 35.96 36.12 35.93 59.59

21:15 4 58189 424.24 4501.00 4501.00 4501.00 4501.00 4501.00 0.00 36.95 35.43 35.58 35.48 59.20

21:20 5 57958 200.00 4500.00 4500.00 4500.00 4500.00 4500.00 0.00 36.63 35.09 35.25 35.15 59.02

21:25 6 57718 86.28 3000.00 3000.00 3000.00 3000.00 3000.00 0.00 36.07 34.52 34.68 34.58 58.66

21:30 7 57468 40.00 992.90 992.53 992.57 992.55 998.23 0.00 35.79 34.24 34.39 34.29 58.48

21:35 8 57207 30.00 887.39 884.26 884.57 884.37 933.14 0.00 35.06 33.49 33.65 33.55 58.06

21:40 9 56937 30.00 695.71 692.58 692.89 692.69 741.46 0.00 36.20 34.66 34.81 34.72 58.69

21:45 10 56657 9.03 300.00 302.47 304.10 289.07 383.22 0.00 35.34 33.78 33.94 33.84 58.21

21:50 11 56368 0.00 38.65 37.19 37.33 37.24 60.08 0.00 34.73 33.15 33.30 33.20 57.86

Page 22: NPRR351 Look-Ahead SCED Update

22

08/09/2012 IE 21:05 (interval_id=11)

METFAugust 30, 2012

Page 23: NPRR351 Look-Ahead SCED Update

23

RTD_201208092010021 Results-New VS Original

METFAugust 30, 2012

IntervalEnding

IntervalID

RTDGTBD

UnderGenOrig

System_LambdaOrig

LZ_NOrig

LZ_SOrig

LZ_HOrig

LZ_WOrig

UnderGenNew

System_LambdaNew

LZ_NNew

LZ_SNew

LZ_HNew

LZ_WNew

20:15 1 59377 0.00 38.55 35.51 35.85 35.34 85.06 0.00 38.55 35.51 35.85 35.34 85.06

20:20 2 59819 0.00 42.58 39.75 40.13 39.08 90.46 0.00 42.40 39.57 39.95 38.91 90.25

20:25 3 59801 0.00 42.57 39.74 40.12 39.08 90.44 0.00 42.05 39.20 39.57 38.58 89.79

20:30 4 59761 0.00 42.56 39.73 40.10 39.07 90.41 0.00 41.26 38.37 38.74 37.85 88.74

20:35 5 59699 0.00 42.50 39.66 40.03 39.01 90.30 0.00 40.71 37.79 38.15 37.34 88.00

20:40 6 59614 0.00 42.38 39.53 39.90 38.91 90.13 0.00 40.08 37.13 37.47 36.76 87.12

20:45 7 59508 0.00 41.76 38.88 39.25 38.34 89.30 0.00 39.36 36.36 36.70 36.10 86.12

20:50 8 59379 0.00 41.10 38.19 38.54 37.73 88.40 0.00 38.94 35.91 36.25 35.71 85.52

20:55 9 59229 0.00 40.46 37.52 37.87 37.13 87.57 0.00 38.44 35.37 35.70 35.27 84.73

21:00 10 59056 0.00 48.33 45.22 45.53 45.23 94.31 0.00 38.18 35.09 35.41 35.05 84.31

21:05 11 58860 1337.39 4501.00 4501.00 4501.00 4501.00 4501.00 0.00 53.68 52.62 52.73 52.66 69.01