Discussion on full network model expansion performance Abhishek - - PowerPoint PPT Presentation

Discussion on full network model expansion performance

Abhishek Hundiwale Senior Market Design Engineering Specialist Market Surveillance Committee Meeting General Session February 19, 2015 ,

Full network model accuracy metric

• Results

– ISO posts the accuracy report by summarizing results

• n a 3-week rolling average:

http://www.caiso.com/Pages/documentsbygroup.aspx ?GroupID=E61CF08C-42B1-4BD9-A9B7- 894FD31EEA39 – More detailed data, hourly by constraint, available to entities that have signed WECC Universal Non- Disclosure Agreement

Page 2

Accuracy metric: Review methodology

• Metric is calculated daily for the ISO as a whole for all

the interties and compares: – Scenario 1: The ISO models external unscheduled flow impacts in the day-ahead – Scenario 2: The ISO does not model external unscheduled flow impacts in the day-ahead

• ISO also analyzes the metric for internal constraints

Page 3

DA External Unscheduled Flow Calculations (Same input as the solved day-ahead case)

Page 4

Power Flow (no ISO NSI, no ISO load, no ISO gen) BAA Forecasted Net Schedule Interchange (no ISO schedules) BAA Demand Forecast (no ISO forecast) DA FNM Topology (Western BAAs including ISO) Planned Outages (Western BAAs including ISO) DA external unscheduled flows per ISO’s inter-tie, per hour (DaExUSFtie,hour) BAAs GDFs and LDFs

Actual External Unscheduled Flow Calculations

Page 5

Power Flow (no ISO NSI, no ISO load, no ISO gen) BAAs Avg Actual Net Schedule Interchange (no ISO schedules) BAAs Avg Actual Load (no ISO load) State Estimator FNM Topology (Western BAAs including ISO) Average Actual external unscheduled flows per ISO’s inter-tie per hour (ActExUSFtie,hour) BAAs GDFs and LDFs

Accuracy Metric Calculations

• Two scenarios:

1) ISO models external unscheduled flow impacts in the day-ahead: |DaExUSFtie,hour - ActExUSFtie, hour| 2) ISO does not model external unscheduled flow impacts in the day-ahead, (DaExUSFtie,hour=0): |0 - ActExUSFtie,hour|

• The metric compares the sum over all ties and hours of

the absolute value of the difference between the day- ahead and actual external unscheduled flows under the two scenarios

• The accuracy metric passes if the magnitude of the

difference with modeling is less than without:

∑|DaExUSFtie,hour - ActExUSFtie,hour| < ∑|0 - ActExUSFtie,hour|  PASS

Page 6

Accuracy Metric (3 – week rolling average) for Interties

Page 7

500,000 1,000,000 1,500,000 2,000,000 2,500,000

Accuracy Metric (3 - week rolling)

Scenario 1 Scenario 2

Page 8

Accuracy Metric (3 – week rolling average) for Internal Constraints for last month

500000 1000000 1500000 2000000 2500000 3000000 3500000 4000000 8-Jan-15 9-Jan-15 10-Jan-15 11-Jan-15 12-Jan-15 13-Jan-15 14-Jan-15 15-Jan-15 16-Jan-15 17-Jan-15 18-Jan-15 19-Jan-15 20-Jan-15 21-Jan-15 22-Jan-15 23-Jan-15 24-Jan-15 25-Jan-15 26-Jan-15 27-Jan-15 28-Jan-15 29-Jan-15 30-Jan-15 31-Jan-15 1-Feb-15 2-Feb-15 3-Feb-15 4-Feb-15 5-Feb-15 6-Feb-15 7-Feb-15 8-Feb-15 9-Feb-15 10-Feb-15 11-Feb-15 12-Feb-15 13-Feb-15 14-Feb-15 15-Feb-15 Scenario 1 Scenario 2

Accuracy Metric: Update

• FNM Metric has passed on all the days except first few

days of FNM implementation

• Some interties in the north such as PACI failed the

individual metric occasionally during first few weeks of implementation, however passing the metric since last month

• The main drivers for the FNM metric:

– Load Forecast in DA for external BAAs – Net Schedule Interchange (NSI) forecast in DA

• Load Forecast: ISO forecasting for 8 external BAAs

Page 9

Next Steps

• Better load forecast for all other BAAs
• NSI forecast in DA: forecast formulation using a similar

day approach

• Analyze the individual interties when they fail the metric

Page 10

Thank you

Page 11