discussion Amber Motley Manager, Short Term Forecasting Market - - PowerPoint PPT Presentation

▶
discussion
SMART_READER_LITE
LIVE PREVIEW

discussion Amber Motley Manager, Short Term Forecasting Market - - PowerPoint PPT Presentation

Feb 28, 2024 100 likes •459 views

Flexible ramping product requirements and load forecast discussion Amber Motley Manager, Short Term Forecasting Market Surveillance Committee Meeting General Session June 7, 2018 ISO Public ISO Public Outline Flexible Ramp Product

slide-1
SLIDE 1

ISO Public ISO Public

Flexible ramping product requirements and load forecast discussion

Amber Motley Manager, Short Term Forecasting Market Surveillance Committee Meeting General Session June 7, 2018

slide-2
SLIDE 2

ISO Public

Outline

  • Flexible Ramp Product Update
  • Renewable Forecast: Persistence Market Model

Background and Performance

  • Market Forecast vs. Raw ALFS Forecast

Page 2

slide-3
SLIDE 3

ISO Public

FLEXIBLE RAMP PRODUCT UPDATE

Page 3

slide-4
SLIDE 4

ISO Public

Summary of Recent Updates for the Flexible Ramping Requirement

  • Item 1:

– Uncertainty Requirement Calculation – Summary of Change:

  • Prior to 2/21/2018 the BARR tool was calculating

the uncertainty using B2-A2 instead of B2-A1. – Presented at February 20, 2018 MPPF

  • http://www.caiso.com/Documents/AgendaandPres

entation-MarketPerfomanceandPlanningForum- Feb202018.pdf – Fix Deployed 2/20/2018 for operating date 2/21/2018

Page 4

slide-5
SLIDE 5

ISO Public

Expected Flexible Ramping Uncertainty Requirement Five Minute Real-Time Dispatch (RTD) Calculation

Page 5

RTD Net Load Forecast Error is difference between the binding interval net load forecast and the prior market run first advisory net load forecast Uncertainty = B2-A1

slide-6
SLIDE 6

ISO Public

Prior to Fix: Flexible Ramping Uncertainty Requirement Five Minute Real-Time Dispatch (RTD) Calculation

Page 6

RTD Net Load Forecast Error is difference between the binding interval net load forecast and the second run first advisory net load forecast

slide-7
SLIDE 7

ISO Public

RTD Prior to Fix: Average Flexible Ramp Product Cleared Awards for EIM_Area – January 23, 2018

Page 7

slide-8
SLIDE 8

ISO Public

RTD After Fix: Expected Average Flexible Ramp Product Cleared Awards for EIM_Area – January 23, 2018

Page 8

slide-9
SLIDE 9

ISO Public

Summary of Recent Updates for the Flexible Ramping Requirement Continued

  • Item 2:

– Renewable Resources Time Interval – Summary of Change:

  • Prior to 3/23/2018 the BARR tool was using

starting interval instead of ending interval in the calculation. – Fix Deployed 3/22/2018 for operating date 3/23/2018

Page 9

slide-10
SLIDE 10

ISO Public

Summary of Recent Updates for the Flexible Ramping Requirement Continued

  • Item 3:

– Treatment of RTPD time frames in the uncertainty calculation (averaging vs. no averaging) – Summary of Change:

  • BARR was previously using one interval within the

RTPD time frame instead of performing an average of the 3-5 minute intervals for the renewable resources inputs into the net load

  • calculation. Following the change the 3-5 minute

intervals were averaged for the renewable resources. – Fix Deployed 3/30/2018 for operating date 3/31/2018

Page 10

slide-11
SLIDE 11

ISO Public

Summary of Recent Updates for the Flexible Ramping Requirement Continued

  • Item 4: FRP Requirement Threshold Documentation

– PRRs Created for Business Practice Manual Changes:

  • Energy Imbalance Market

– Resource Sufficiency Evaluation » Section 11.3.2

  • Market Operations

– Flexible Ramping Product » Section 7.1.3

Page 11 You can follow these BPM changes at the following links: https://bpmcm.caiso.com/Pages/default.aspx https://bpmcm.caiso.com/Pages/ViewPRR.aspx?PRRID=1051&IsDlg=0

slide-12
SLIDE 12

ISO Public

PRR 1051 – Flexible ramping clarification

Page 12

  • Reason for revision

– This is to clarify the flexible ramping requirements for the new EIM entities joining the Energy Imbalance Market.

  • Language Proposed

– CAISO shall set the histogram values described in Section 7.1.3 of the Market BPM to ensure the flexible ramp requirements stay within a reasonable level for a transitional period following implementation. This histogram value will be used until the ISO is able to collect sufficient production-quality data to accurately calculate the flexible requirements based on the historical information gathered from Production. These initial thresholds may be adjusted according to each balancing authority area’s conditions including factors and data observed during market simulation and parallel operations. These thresholds will allow the Flexible Ramping Requirements to stay within a reasonable band during the transitional period until an accurate histogram can be calculated from Production data for the balancing authority area.

  • Initial comments

– No comments submitted

  • Initial comment period expired

– April 18, 2018

  • Next step

– Post ISO recommendation

slide-13
SLIDE 13

ISO Public

PRR 1053 – Flexible ramping clarification

Page 13

  • Reason for revision

– This revision is to further clarify the flexible ramping product requirement thresholds.

  • Language Proposed

“The ISO shall use a rolling 40 day average, with a separate histogram for weekends and holidays, to evaluate the historical advisory RTUC imbalance energy requirement error pattern for each RTUC hour. The ISO will also evaluate if hours with similar ramping patterns could be combined to increase the sample size used in the historical analysis. The ISO expects that the estimate of uncertainty will improve over time. Therefore, the actual method of calculating the demand curve will be included in the business practice manual versus including these details in the tariff. Additionally, because the requirements are based on historical information, the requirements determined through this process may be representative of future forecast uncertainty and may at times also produce extreme outlier values. To ensure the CAISO does not set extreme requirements, the CAISO enforces thresholds that are determined based on the 98% percentile of the historical uncertainty

  • calculations. The CAISO will evaluate these thresholds every quarter, or as needed with changing weather conditions.

To the extent permissible, the CAISO will provide EIM entities a week’s notice prior to making any changes to the thresholds.”

  • Initial comments

– No comments submitted

  • Initial comment period expired

– May 15, 2018

  • Next step

– Post ISO recommendation

slide-14
SLIDE 14

ISO Public

RENEWABLE PERSISTENCE MARKET METHODOLOGY UPDATE

Page 14

slide-15
SLIDE 15

ISO Public

Why Is the Persistence Method needed?

Page 15 Site Data Collected FSP Creates Forecast ALFS Process Complete Binding Interval

13:45-13:50 13:51-13:53 13:54-13:57 13:57

Mkt Runs

14:05-14:10

5-10 minutes up to 3 minutes 3 minutes 7.5 minutes

PI Data submitted to FSP Forecast to ALFS Data to Market Site Data Collected Binding Interval

13:56 13:57

Mkt Runs

14:05-14:10

7.5 minutes

Data to Market

Persistence Method:

  • More recent telemetry

is used in forecast

  • 6+ minutes are

eliminated from lag

Forecast calculated in market, eliminating ALFS & processing time needed outside of CAISO Current:

Why Is the Persistence Method needed?

slide-16
SLIDE 16

ISO Public

Impact of Shortened Processing Time for Wind

Page 16

slide-17
SLIDE 17

ISO Public

Results: Contour Persistence Works for Solar

Page 17

  • 2000

2000 4000 6000 8000 10000 6:15 6:35 6:55 7:15 7:35 7:55 8:15 8:35 8:55 9:15 9:35 9:55 10:15 10:35 10:55 11:15 11:35 11:55 12:15 12:35 12:55 13:15 13:35 13:55 14:15 14:35 14:55 15:15 15:35 15:55 16:15 16:35 16:55 17:15 17:35 17:55 18:15 18:35 18:55 19:15 19:35 19:55

Solar

Persistence External Actual supp

slide-18
SLIDE 18

ISO Public

Results: Contour Persistence Method when Heavy Supplemental Dispatches are Present

Page 18

  • 2000

2000 4000 6000 8000 10000 6:15 6:35 6:55 7:15 7:35 7:55 8:15 8:35 8:55 9:15 9:35 9:55 10:15 10:35 10:55 11:15 11:35 11:55 12:15 12:35 12:55 13:15 13:35 13:55 14:15 14:35 14:55 15:15 15:35 15:55 16:15 16:35 16:55 17:15 17:50 18:10 18:30 18:50 19:10 19:30 19:50 20:10

Solar

Persistence External Actual supp

slide-19
SLIDE 19

ISO Public

Results: RTD Wind Mean Absolute Percent Error (MAPE)

Page 19

0.00% 1.00% 2.00% 3.00% 4.00% 5.00% 6.00% 7.00% 8.00% 9.00% 10.00% 0:15 1:25 2:35 3:45 4:55 6:05 7:15 8:25 9:35 10:45 11:55 13:05 14:15 15:25 16:35 17:45 18:55 20:05 21:15 22:25 23:35

MAPE (% Error) Time

MAPE Comparison: Resource

Persistence External FSP

0.00% 0.50% 1.00% 1.50% 2.00% 2.50% 3.00% 3.50% 4.00% 0:15 1:20 2:25 3:30 4:35 5:40 6:45 7:50 8:55 10:00 11:05 12:10 13:15 14:20 15:25 16:30 17:35 18:40 19:45 20:50 21:55 23:00 23:50

MAPE (% Error) Time

MAPE Comparison: System

Persistence External FSP

slide-20
SLIDE 20

ISO Public

Results: RTD Solar Mean Absolute Percent Error (MAPE)

Page 20

0.00% 1.00% 2.00% 3.00% 4.00% 5.00% 6.00% 7.00% 8.00% 9.00% 10.00% 11.00% 6:15 7:00 7:45 8:30 9:15 10:00 10:45 11:30 12:15 13:00 13:45 14:30 15:15 16:00 16:45 17:30 18:15 19:00 19:45

MAPE (% Error) Time

MAPE Comparison: Resource

Persistence External FSP

0.00% 0.50% 1.00% 1.50% 2.00% 2.50% 3.00% 3.50% 4.00% 6:15 7:00 7:45 8:30 9:15 10:00 10:45 11:30 12:15 13:00 13:45 14:30 15:15 16:00 16:45 17:30 18:15 19:00 19:45

MAPE (% Error) Time

MAPE Comparison: System

Persistence External FSP

slide-21
SLIDE 21

ISO Public

MARKET FORECAST VS ALFS FORECAST

slide-22
SLIDE 22

ISO Public

Background Information

  • What is the current role that exists within the CAISO

Operations Functions for updating the load forecast within the Market Optimization?

  • Why did this functionality originally get developed?
  • In this presentation we will look at the differences of the

pure raw ALFS Load Forecast vs the Market Optimization

Note: analysis was done with data from 1/1/2017 through 4/30/2018

Page 22

slide-23
SLIDE 23

ISO Public

RTD Forecast Accuracy Trends

slide-24
SLIDE 24

ISO Public

RTD Time Evolution of Accuracy Trends

Page 24

slide-25
SLIDE 25

ISO Public

RTPD Forecast Accuracy Trends

slide-26
SLIDE 26

ISO Public

RTPD Time Evolution of Accuracy Trends

Page 26

slide-27
SLIDE 27

ISO Public

Load Conformance patterns do not show yet a meaningful difference with the changes introduced

Page 27

slide-28
SLIDE 28

ISO Public

A large set of load conformance instances do not

  • ffset the effect of the manual forecast updates

Page 28

15-minute market 5-minute market

slide-29
SLIDE 29

ISO Public

Summary of Findings

  • ALFS Forecast performs better than Market Submitted
  • n Average.
  • Market Submitted forecast is performing better in some

intervals; such as during the morning turn of the load, and at times during the evening peak.

  • Market Submitted forecast has less movement between

RTD and RTPD.

Page 29

slide-30
SLIDE 30

ISO Public

Next Steps

  • Continue to work with Operations to automatically push

the ALFS load forecast updates during periods of

  • bvious better performance and limited system reliability

impact.

  • Continue to work to improve the ALFS load forecast in all

intervals including key real time intervals as well as improving the model accuracy for the morning turn.

Page 30

slide-31
SLIDE 31

ISO Public

APPENDIX

Page 31

slide-32
SLIDE 32

ISO Public

How the VER Persistence Market Method Works

Page 32

  • Simple Persistence:

F(t) = A(t-lag)

  • Contour Persistence Model:

F(t) = A(t-lag)*R(t)/R(t-lag) R(t) is expected full fuel (full sun) reference curve

slide-33
SLIDE 33

ISO Public

Complexities of Load Forecasting

  • “May Gray” Marine Layer Example

Page 33

slide-34
SLIDE 34

ISO Public

RTPD Distribution of MW Trends

Page 34

slide-35
SLIDE 35

ISO Public

RTD Distribution of MW Trends

Page 35