RAN Reliability Requirements RASC June 10, 2020 RASC010, RASC011, - - PowerPoint PPT Presentation

RAN Reliability Requirements

RASC June 10, 2020 RASC010, RASC011, RASC012

Purpose & Key Takeaways

Key Takeaways:

• How we define system reliability needs must

change, recognizing an evolving risk profile

• Analysis indicates shift of risk patterns
• utside typical summer peak load periods

and growing flexibility needs with the changing resource mix

• Resource adequacy approaches vary across

regions including pros and cons of different metrics Purpose: Present draft framing, discuss preliminary analysis results, review industry trends and benchmark

MISO’s product development process moves from problem definition to buildable solutions

3

Explore Decide Do

Problem Definition & Education Exploration

• f Options

Proposal & Business Rules Software, BPM updates & Training Deploy to production Review

• utcomes

Defines important design elements and derives options via higher fidelity modeling. Selects an option and details specific design.

We are here

Defines needs to address through additional steps. Can iterate as move through the process.

3

Problem Definition

• Today’s analysis does not sufficiently capture risk across the year
• The analysis should better reflect patterns across the year and not just

summer

• The analysis should better reflect the magnitude of risks
• Today’s approach to setting requirements does not sufficiently

mitigate risk

• The approach to setting targets should leverage enhanced risk calculation
• The approach to setting targets should sufficiently mitigate risk under

current and future portfolios

• System risk profiles will continue to change with evolution of the

resource mix

• Risks could emerge in time periods other than summer peak
• The margin between available resources and total obligations will be more

impacted by uncertainty and variability

4

Operating margins impact reliability

• The margin between supply resources and obligations is an

indicator of how close the system is to emergency or loss of load.*

• It is influenced by a number of factors, some of which are

highly variable and uncertain

• Outages
• Intermittent generation
• Net scheduled interchange

5

Margin = Available non-intermittent generation + intermittent generation + RDT limit + Net Scheduled Interchange + Load Resources (BTMG + LMR + EDR) - Load - Operating Reserve

RDT = Regional Dispatch Transfer Limit | BTMG = Behind the Meter Generation LMR = Load Modifying Resources | EDR = Emergency Demand Response * Emergencies include alerts through to load shed

Time Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

A historical look indicates that periods of tight

• perating margin aligns with low efficiency / high risk

6

2018

None Moderate High

Systemwide Historical Margins

Month   Hour

Risks exist outside

• f summer months

Severe Moderate Mild Margin Risk Occurrence Occurrences are defined as Max Gen, Alert, or RSG greater than three times average Severity is color-coded by margin size

Adjustment to modeling assumptions can better capture risk across the year

7

Margin = Available non-intermittent generation + intermittent generation + RDT limit + Load Resources (BTMG + LMR + EDR) + Net Scheduled Interchange - Load - Cleared Operating Reserve

Assumption Current Approach Trial Analysis Assumptions

• Intermittent

resource capacity Flat capacity throughout the year based on summer performance. Using 8760 profiles corresponding to 2018 weather year.

• Non-firm

external support Adjustment to the PRM based on imports during summer peak. Using monthly average NSI from the last 3 years, assuming a perfect unit.

• Forced outage

rates (FOR) Modelled as a single average forced outage rate for the entire year. Modeled with adders / subtractors at different date-hour based on temperature correlation model using 3- years of historical data.

• Planned outages

Optimized to avoid outages during peak summer load periods. Scheduled using a 90% optimality (“best behavior”) assumption.

Adjustments to summer-Focused LOLE Modeling Assumptions

Trial inputs were modified to better reflect system risk across the year

8 2 4 6 8 10 12 14 16 1 2 3 4 5 6 7 8 9 10 11 12 GW Month 2 4 6 8 10 12 14 16 1 2 3 4 5 6 7 8 9 10 11 12 GW Month 2 4 6 8 10 12 14 16 1 2 3 4 5 6 7 8 9 10 11 12 GW Month 2 4 6 8 10 12 14 16 1 2 3 4 5 6 7 8 9 10 11 12 GW Month

Planned Maintenance Forced Outages Intermittent Production Non-Firm External Support

Current Trial

Draft analysis of 2018

Time Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

Modifying Loss of Load Expectation inputs to reflect seasonality can better reflect risk throughout the year

9

2018 EUE

Trial Method Current Method

Risks shift outside of summer months

Loss of Load Expectation*

No risk Moderate Risk High Risk

Time Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

Month   Hour

Modeled 2018

*0.1 LOLE Target

The team also ran trials on future sample future scenarios

• The futures reflect many dynamics in the region
• Changing fuel costs, clean energy commitments, aging fleet, electrification
• The analysis leveraged draft MTEPs as a starting point

10

• Less coal retirement
• Reflects member plans*

** Does not include very recent change in MTEP F3 that adjusts load growth from 60% to 50%.

MTEP 2019 AFC 2033

(760TWh)

• Higher solar
• Middle load growth
• Gas in lieu of coal
• High load

**

* As announced plans submitted to commissions

Future III, 2030**

(954 TWh)

Future 1 2040

(825 TWh)

Growth in intermittent resources could increase variability and overall impact of uncertainty

11

+ Standard deviation ++ Reflects changes between Day Ahead and Real Time, including forecast error

Scenario Input Size Average (Range) (GW) Variability+ (GW) Current Wind

6 (0 to 16) 3

Solar

NA NA

FutureI-2040 Wind

13 (0.3 to 31 ) 7

Solar

14 (0 to 52 ) 16

FutureIII-2030 Wind

37 (0.8 to 85) 20

Solar

2 (0 to 6) 2

MTEP19AFC 2033 Wind

18 (0.4 to 43) 10

Solar

6 (0 to 23) 7

Total Wind

23 (0.3 85) 17

Solar

7 (0 to 52) 11

11

Future resource evolution and total capacity shape pattern of need & non-summer peak

• nly focus is still relevant

12

Time Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

Future 1 2040

No risk Moderate Risk High Risk

0.1 Loss of Load Expectation

Time Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

Month   Hour

Proxy Operator Experience*

Risks shift outside of summer months & traditional peak hours

* 0.6 LOLE Target

Draft results

Alternative measures may help differentiate system impacts

Interval LOLE (day) LOLP (%) EUE (MWh) EDNS (MW) Annual 0.10 0.03 1,765 700 Interval LOLE (day) LOLP (%) EUE (MWh) EDNS (MW) Annual 0.10 0.03 3,021 1,550 Interval LOLE (day) LOLP (%) EUE (MWh) EDNS (MW) Annual 0.10 0.03 3,550 2,400

MTEP 2019 AFC 2033

**

Future III, 2030 Future 1 2040

Portfolios with similar LOLE result in a wide range of EUE values

LOLE = Loss of Load Expectation EUE = Expected Unserved Energy

13

EDNS = Expected Demand Not Supplied

Draft results

Stakeholder Feedback Request

• MISO is requesting feedback by June 24, 2020 on
• The proposed problem definition
• Priority adjustments to inputs for higher fidelity analysis in

the evaluation of options

• Issue Tracking ID#: RASC010, RASC011, RASC012
• Feedback requests and responses are managed

through the Feedback Tool on the MISO website: https://www.misoenergy.org/stakeholder- engagement/stakeholder-feedback/

14 |

Alternative RA Metrics, and RA Construct Elements

June 10, 2020

PRESENTED TO PREPARED BY

MISO Resource Adequacy Subcommittee Sam Newell Michael Hagerty Hannes Pfeifenberger Walter Graf

Agenda

• RA Metric Benchmark
• RA Construct Elements Benchmark

Brattle.com | 2

RA METRIC BENCHMARK

Resource Adequacy Metrics

Brattle.com | 3

Metric Description Pros Cons Examples

Loss-of-Load Probability (LOLP)

Probability of demand exceeding available resources at least once within a year. Units: % chance of >= 1 event per year Easy to calculate and understand Does not consider duration or size of an unserved load event Northwest Power and Conservation Council: 5% LOLP

Loss-of-Load Events (LOLE)

Expected number of events per year in which demand is not served. One event in ten years translates to 0.1 LOLE per year. Units: Events per year Easy to calculate and understand Used by most U.S. systems Does not consider duration or size of an unserved load event Most U.S. Systems: 1 loss-of-load event per decade or 0.1 event per year

Loss-of-Load Hours (LOLH)

Expected number of hours per year in which demand is not served. One day in ten years translates to 2.4 LOLH per year. Units: Hours per year Considers the loss of load duration Used by NERC Does not consider size of an unserved load event SPP: 2.4 LOLH per year (equal to 1 day in 10 years)

Normalized Expected Unserved Energy (EUE)

Expected MWh of load that will not be served as a result of demand exceeding available

• supply. Can be normalized as % of load.

Units: % of expected annual load Considers both the duration and depth

• f supply shortages

Used by NERC Requires more sophisticated statistical methodologies Alberta: Max annual EUE of 800 MWh Australia NEM: Max of 0.002% normalized EUE

RA METRIC BENCHMARK

Application of Resource Adequacy Metrics across RTOs

Brattle.com | 4

Under NERC Standard BAL-502-RF-03, MISO must calculate the planning reserve margin necessary to achieve 0.1 LOLE (but may have flexibility in how it sets requirements) RTOs across the U.S. implement 0.1 LOLE differently

RTO 1-in-10 Standard Definition Event Type

MISO 0.1 loss-of-load events per year

Firm load shed after all operating reserves and DR deployed

NYISO 0.1 loss-of-load events per year

Firm load shed after 10 min and 30 min operating reserves and voltage reduction deployed

ISO-NE 0.1 loss-of-load events per year

Firm load shed after voltage reduction and DR deployed, but 200 MW

• f operating reserves maintained

PJM 0.1 days with loss-of-load per year

Firm load shed after interruptible load and 30 min reserves deployed, but before voltage reduction or 10 min reserves deployed

SPP 2.4 loss-of-load hours per year

Not explicitly defined

Source: Pfeifenberger, et al., Resource Adequacy Requirements: Reliability and Economic Implications, Prepared for FERC, September 2013.

RA METRIC BENCHMARK

In its 2016 assessment, NERC chose two metrics to represent a consistent measure across different areas: Expected Unserved Energy (EUE)

• Measure of the system’s capability to continuously serve all loads at all delivery points while

satisfying all planning criteria

• Energy-centric and analyzes all hours of a particular year
• Summation of the expected number of MWh of load in a given year that will not be served as a

result of demand exceeding available capacity Loss of Load Hours (LOLH)

• The number of hours during a given time period where system demand will exceed generating

capacity, which accounts for duration of events but not magnitude NERC has included estimates of these two metrics in its reliability assessment reports since then

NERC Assessment of RA Metrics

Brattle.com | 5

Source: NERC, Probabilistic Assessment Improvement Plan: Summary and Recommendations Report, December 2015. Available at: https://www.nerc.com/comm/PC/Reliability%20Assessment%20Subcommittee%20RAS%202013/ProbA%20%20Summary%20and%20Recommendations%20final%20Dec%2017.pdf

RA METRIC BENCHMARK

We recommend that MISO consider adopting EUE if:

• The benefits of planning for a

consistent level of load shed each year…

• Outweighs the challenges of

switching away from LOLE

Tradeoffs of Resource Adequacy Metrics

Brattle.com | 6

Metric Pros Cons

Loss-of-Load Metrics (LOLE, LOLP, LOLH)

• Easy to calculate and understand
• Widely used across U.S. RTOs
• LOLH accounts for duration
• LOLE & LOLP do not account for

duration or size of events

• Reliability level changes with

system size. Does not allow for direct comparison among jurisdictions

• Measurement/interpretation are

not aligned across markets

• Has not been updated for changes

in electricity industry

EUE Metric

• Measures both the duration and

magnitude of load shed events due to inadequate supply

• If normalized, reliability levels are

not effected by growth in system

• size. Can also be used to compare

across systems of different sizes.

• Used by NERC in their assessments
• Not commonly used in U.S. RTOs
• Would need to determine desired

target level based on LOLE/EUE studies

Agenda

• RA Metric Benchmark
• RA Construct Elements Benchmark

Brattle.com | 7

RA CONSTRUCT ELEMENTS BENCHMARK

MISO is considering reforms to address three different types of shortages

MISO Reliability Challenges

Brattle.com | 8

We primarily focus in these slides on the Resource Adequacy construct to ensure sufficient installed capacity in MISO

Types of Shortages Installed capacity

insufficient to meet demand year-round

Available capacity

insufficient during shoulder months due to excess outages Hourly (8760) RA simulations Hourly (8760) RA simulations

Primary Analytical Tool Committed capacity

insufficient to meet real-time flexibility needs DA-to-RT (DART) simulations Yes, and the RA construct can account for sub- annual needs and capabilities (to inform investments) Primarily an outage coordination issue, but can be informed by identified RA needs and have implications for resource accreditation

Is it a Resource Adequacy Issue?

Primarily an operations and E&AS markets issue to better use installed capacity; add RA requirement

• nly if insufficient installed flexible capacity

RA CONSTRUCT ELEMENTS BENCHMARK

MISO analyses of historical data suggest reliability risks are shifting away from just summer peak So what is next? What are the key elements of its design to consider when evaluating potential changes to the MISO RA construct?

Resource Adequacy Construct Elements

Brattle.com | 9

Future Reliability Risks Resource Adequacy Requirements Resource Accreditation

Key elements fall into three categories:

RA CONSTRUCT ELEMENTS BENCHMARK

Future Reliability Risks

Brattle.com | 10

Tools

What tools can assess future reliability risks?

 Primarily, SERVM and GE-MARS, combined with future scenarios

Patterns of Reliability Risks

Are there times outside summer peak with reliability risks?

 Southern Company and TVA observe risks in the summer & winter  PJM and ISO-NE identified winter fuel security and availability risks  AESO identified tight “supply cushion” hours year-round, many in summer despite load being highest in winter

Metrics

What are the right metrics to quantify those risks?

 Most U.S. system operators use LOLE  Alberta, Australia, European systems use EUE  NERC uses EUE and LOLH in its assessments Future Reliability Risks

Note: Examples from Alberta and Ontario refer to their proposed market designs that have since been delayed or cancelled

RA CONSTRUCT ELEMENTS BENCHMARK

Resource Adequacy Requirements

Brattle.com | 11

Requirement Periods

Should there be multiple periods or a single annual period addressing year- round risks?

 ISO-NE, PJM, and Alberta set a single annual requirement to address year-round risks  Ontario, Southern Company, and TVA set seasonal requirements  CAISO and NYISO set monthly requirements

System-Wide Requirement

How to determine the system-wide or zonal requirement for each period?

 Most markets continue to set a RM based on peak load hours  Alberta proposed setting its RM based on tightest supply hours  Ontario proposed design considers the relative costs of reducing LOLE in each season for setting seasonal requirements

Local Requirements

How to translate the system-wide or zonal requirement to each LSE?

 CPUC requires LSEs to meet the same 15% RM each month with separate local/zonal requirements  SPP requires its LREs to meet 12% RM during summer peak

Additional RA Requirements

Are additional RA products needed?

 California added a Flexible RA (installed capacity) requirement Resource Adequacy Requirements

Note: Examples from Alberta and Ontario refer to their proposed market designs that have since been delayed or cancelled

Basis for Accreditation

How to determine resource availability during each requirement period? Should it account for planned outages?

 CPUC uses ELCC for both wind and solar; UCAP for rest  NYISO sets solar & wind values based on average output during peak load hours (e.g., 2-6 pm in June – August for summer)  Alberta proposed UCAP as average output during 200 tightest supply-cushion hours, irrespective of planned or forced outages

Participation Requirements

Will different requirements be allowed for different resources across periods? What obligations to place on resource availability during shortage events?

 PJM used to allow Summer DR (only available in the summer)  Ontario proposes to allow for Seasonal or Annual resources  CPUC and NYISO set monthly/seasonal values for solar and wind  Singapore proposes to allow daytime-only participation for DR

Penalties & Incentives

How to assess performance and set penalties and incentives during events?

 Performance incentives/penalties assessed in ISO-NE and PJM based on availability during shortage events  Alberta proposed to assess based on shortage events and tight supply hours

RA CONSTRUCT ELEMENTS BENCHMARK

Resource Accreditation

Brattle.com | 12

Resource Accreditation

Note: Examples from Alberta and Ontario refer to their proposed market designs that have since been delayed or cancelled