1 / 90

R.12-03-014: Energy Division Workshop – Operating Flexibility Modeling

R.12-03-014: Energy Division Workshop – Operating Flexibility Modeling. Nathaniel Skinner Senior Analyst, Generation & Transmission Planning California Public Utilities Commission September 19, 2012. WebEx

kaz
Download Presentation

R.12-03-014: Energy Division Workshop – Operating Flexibility Modeling

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. R.12-03-014: Energy Division Workshop – Operating Flexibility Modeling Nathaniel Skinner Senior Analyst, Generation & Transmission Planning California Public Utilities Commission September 19, 2012

  2. WebEx Meeting Number: 741 769 312Meeting Password: LTPPhttps://van.webex.com/van/j.php?ED=189577152&UID=491292852&PW=NNGQ4MGM0MTBk&RT=MiM0 Remote Access Call in #: Passcode: 866-758-1675 3481442 Note: *6 to mute/unmute Upon entry to the call, please place yourself on mute, and remain on mute unless you are asking a question

  3. Agenda

  4. Workshop Purpose • Review past (2010 LTPP) Operating Flexibility modeling approaches • Examine proposed modeling approaches for the 2012 LTPP • Begin framing discussion for meeting any needs identified for end of 2013 decision

  5. Roadmap

  6. Operating Flexibility Analysis for R.12-03-014 Mark Rothleder, Executive Director, Market Analysis and Development Shucheng Liu, Principal Market Developer CPUC, Workshop September 19, 2012

  7. Description of Past Method and Model

  8. Study process quantifies operational requirements and evaluates fleets ability to meet operating requirements. Statistical Analysis/ model DevelopProfiles Production simulation Variable Resource Wind / Solar and Load Profiles Flexibility Requirements (Regulation, Balancing) Shortages Infrastructure Needs Costs, Emissions Import/Export Capacity Factor Renewable Portfolios

  9. Study Methodology • Variability and uncertainty of renewable resources and load largely determine system regulation and load following requirements • Currently, load is the dominant uncontrollable variable • 33% RPS introduces two additional uncontrollable variables that impact load-following and regulation requirements • Wind, and • Solar • System regulation and load following requirements depend on three factors: • Forecast quality: Load, Wind and Solar forecast errors • Interaction between load, wind and solar: net variability • Market timeline: how fast the market re-commit and re-dispatch the controllable resources

  10. CAISO Scheduling Process MW Generation Requirement Hour Ahead Schedule And Load Following Regulation Hour Ahead Adjustment Load Following Day Ahead Hour Ahead Schedule Schedule t Operating Hour

  11. Calculating hourly load-following requirement • Load Following is defined as the difference between the 5-min forecasted net load and the hour ahead forecasted net load • Determine the 95th percentile of maximum load-following requirement

  12. Calculating hourly regulation requirement • Regulation is defined as the difference between the net load (load– wind– solar) and the 5 minute forecasted net load (load–wind–solar) • Determine the 95th percentile maximum regulation requirement

  13. Flow chart for calculating load-following requirements

  14. Flow chart for calculating regulation requirements

  15. Review of activities between 2010 LTPP settlement and today

  16. Additional sensitivity and analysis performed since July 2011 • PRM Analysis Deep Dive analysis of PRM • All-Gas reserve margin overstated PRM • All-Gas case PRM~21% instead of 41% • Step 1 Sensitivity • Assessed and bounded impacts of forecast errors • Assessed drivers of flexible ramp • 5 minute simulation • Similar but slightly reduced violations observed • Regional modeling and coordination • Improved modeling of the GHG improves the regional flows • If not import constrained regional coordination can improve access to flexible resources

  17. Operational criteria within the context of NERC/WECC standards

  18. Balancing Authority ACE limit (BAAL) – System control opposes frequency deviations • BAAL is designed to replace CPS2 • BAAL relaxes area regulation needs • ACE is allowed to be outside BAAL for up to 30 minutes

  19. Control Performance Standard Scores (CPS1) Scores January 2009 through April 2012 CPS 1 Scores – January 2009 through April 2012 Began operating to BAAL

  20. The assessment of a balancing authority control performance is based on three components • Control Performance Standard (CPS1) - measures the control performance of a BA's by comparing how well its ACE performs in conjunction with the frequency error of the Interconnection • Balancing authority Ace Limit (BAAL) - is a real-time measure of Area Control Area and system frequency which cannot exceed predefined limits for more than 30-minutes • Disturbance Control Standard (DCS) - is the responsibility of the BA following a disturbance to recover its ACE to zero if its ACE just prior to the disturbance was greater than zero or to its pre-disturbance level if ACE was less than zero - within 15 minutes • Control Performance Rating Pass is when CPS1 ≥ 100%; BAALLimit ≤ 30 minutes & DCS = 100%

  21. A Stochastic Model for Analyzing Ramping Capacity Sufficiency

  22. A stochastic model is needed to assess the probability of upward ramping capacity sufficiency. • A deterministic production simulation case adopts only one of the many possible combinations of input assumptions • A stochastic model can evaluate various input combinations based on probability distributions of the stochastic input variables • Monte Carlo simulation determines insufficient ramping capacity probability • It complements the deterministic production simulation

  23. Available ramping capacity depends on the balance of supply and demand. Supply curve is constructed based on variable cost of each generation unit

  24. Uncertainties in supply and demand affect availability of ramping capacity.

  25. Available ramping capacity of each generation unit is determined based on the following factors: • Maximum and minimum capacity • Unit availability (due to forced and maintenance outages) • Dispatch level • Ramp rate • Ramp time allowed (10 or 20 minutes)

  26. Ramping capacity shortage may occur due to variations in both availability and requirement.

  27. This stochastic model considers uncertainties in some of the key inputs, including: • Load forecast • Inter-hour load ramp • Requirements for regulation-up and load following-up • Generation by wind, solar, and hydro resources • Availability of generation units (due to forced and maintenance outages)

  28. A model is developed for a time period in which all hours have similar conditions. • No unit commitment • No chronologic constraint (such as min run time and min down time, etc.) • Independent with identical probability distribution functions for each hour in the period • Insufficient ramping capacity probability for each hour determined through Monte Carlo simulations • Insufficient ramping capacity probability for the whole year calculated based on Binomial distribution

  29. Probability distributions are developed based on data from the Plexos production simulation model. • Hourly load forecast • Hourly regulation and load following-up requirement • Hourly wind, solar, and hydro generation • Uniform distribution functions based on forced and maintenance outage rates of each generation unit

  30. Inter-hour load ramp is calculated based on hourly load forecast. • Upward direction only • A new stochastic variable • Met by 60-min ramping capability • A part of load

  31. These are examples of probability distribution functions of stochastic variables. Exploring the probability to have load higher than 1-in-2 forecast 1-in-2 forecast 1-in-2 forecast 1-in-10 forecast

  32. Examples of probability distribution functions of stochastic variables. (cont.)

  33. Examples of probability distribution functions of stochastic variables. (cont.)

  34. Correlations among the stochastic variables are enforced. This is an example of the correlation matrix

  35. Generation units in the stochastic model have the following characteristics from the Plexos model. • From input data • Maximum and minimum capacity • Ramp rate • Forced outage and maintenance outage rates • From simulation results • Average generation cost (to determine an initial dispatch order)

  36. Generation unit availability is stochastically determined in each iteration of the Monte Carlo simulations. • Forced and maintenance outages are determined independently for each generation unit • Each of the outages is determined based on the unit’s outage rate and a draw using a uniform distribution function • A maintenance outage allocation factor is used to represent seasonal pattern of maintenance schedules • A unit is unavailable when any one of the outages occurs

  37. Contributions of a generation unit to energy and ramping capacity are subject to: • 10-min upward ramping capacity constraint • 20-min upward ramping capacity constraint • 60-min upward ramping capacity constraint • Maximum capacity constraint

  38. Total contributions by all generation units should meet energy and ramping capacity requirements. • 10-min upward ancillary service requirement • 20-min upward ramping capacity requirement • 60-min upward ramping • Energy balance

  39. The model seeks a least-cost solution to meet energy and all ramping capacity requirements. • Generation units are dispatched economically to meet load first • Remaining qualified ramping capacity is used to meet upward ancillary service, load following, and inter-hour load ramp requirements • Energy dispatch and ramping capacity contributions are co-optimized when there is insufficient ramping capacity initially

  40. Monte Carlo simulation determines insufficient ramping capacity probability. • Monte Carlo simulation is conducted using this stochastic model • Insufficient ramping capacity results are presented in a probability distribution format • The key results are the probability to have ramping capacity shortage each hour and the probability distribution of the volume of the shortages

  41. This example has a 0.8% probability to have 20-min ramping capacity shortage each hour. Results for the Super-Peak period.

  42. The highest 20-min ramping capacity shortage is 4,661 MW in this example. Results for the Super-Peak period.

  43. The probability to have 10-min ramping capacity shortage each hour is 0.1%. Results for the Super-Peak period.

  44. Monte Carlo simulation results for all periods are summarized as follows:

  45. Cumulative probabilities of ramping capacity shortage are calculated using Binomial distribution. It is the probability to have at least i hours with ramping capacity shortage in year 2020.

  46. Expected number of hours with ramping capacity shortage in 2020 are calculated based on the probabilities. It is insufficient ramping capacity expectation.

  47. What we learned from this approach: • It does • Use probability distributions to capture uncertainties in key input factors • Implement ramping constraints and flexibility requirements • Present insufficient ramping capacity events in probabilistic format • It does not • Decide unit commitment • Impose chronological constraints • It can be improved to • Use multi-year synchronized historical data to capture more variations of input stochastic variables

  48. A Study to Support Meeting Assembly Bill (AB) 1318

  49. Plexos simulations are conducted about performance of local capacity requirement resources in 2020. • Base model: High-Load scenario in 2011 LTPP study • Reduced min run time of some demand response resources from 4 to 1 hour • LCR resources • Added 3,173 MW local capacity requirement (LCR) resources • A sensitivity case • Reduced capacity of event-based demand response resources

  50. LCR resources are added to SCE and SDG&E zones. • 3,173 MW LCR resources based on the ISO OTC study • Los Angeles Basin: 2,370MW • Big Creek Ventura areas: 430MW • San Diego: 373MW* • LCR resources added as a combination of CCGT and GT units • SCE: 2 x 500 MW CCGT units • SCE: 18 x 100 MW LMS100 GT units • SDG&E: 1 x 373 MW CCGT unit * It assumes that San Diego proposed generation is included already. So the total need in San Diego should be 373 MW plus Pio Pico = 300MW Quail Brush = 100MW Escondido Energy Center = 45 MW

More Related