Comments on 10/8 call

1. Please provide your organization’s feedback on the draft inputs and assumptions.

American Clean Power-California (“ACP-California”) appreciates the opportunity to comment on the California Independent System Operator (“CAISO”) Draft Inputs and Assumptions (“I&A”) and supports CAISO’s efforts to develop a more robust Resource Adequacy (“RA”) modeling framework.

ACP-California looks forward to developing a further understanding of CAISO’s proposed I&A as well as its broader methodology for stress testing system conditions based on weather inputs to solar, wind, and thermal resources, as well as load. While these input parameters may be reasonable for the near-term completion of the CAISO’s Loss of Load Expectation (“LOLE”) analysis, further vetting is needed to review and assess CAISO’s methodology before it should be used for binding regulatory requirements, and care should be taken to avoid cementing the current methodology for later use in, for instance, the establishment of default counting rules.

Specifically, ACP-California’s understanding is that CAISO’s proposed method differs in several ways from current methods used in the CPUC’s Integrated Resource Plan (“IRP”) and RA modeling ecosystem, utilizing different input data, different simulation methods, and a different approach to stochastics. We have reached out to CAISO for additional information and data access for further review of this approach, and look forward to further review and discussion as this Initiative moves forward.

ACP-California understands CAISO’s current methodology as focusing on testing a range of potential load, solar, and wind combinations (alongside outages) to ensure the simulation captures the full range of potential outcomes[1]. This method uses the CPUC’s Integrated Resource Plan dataset for solar resources and recent historical data for wind resources, which are then perturbed using the “mean walk reversion” methodology parameterized based on historical solar and wind data from the National Renewable Energy Laboratory. We look forward to reviewing the input and simulation data in greater detail as it becomes available.

While this methodology provides a broad range of potential outcomes, ACP-California would like to better understand how accurately and effectively this method reflects weather correlation, specifically how the simulated distribution aligns with probabilities for future weather – for instance, excluding draws which bring together incongruent weather samples across loads and different resources, which may introduce error or dampen the signal of more realistic reliability risk. This approach differs from the CPUC’s weather-correlated approach within SERVM, which focuses on correlated permutations directly leveraging the historical weather dataset: testing simulated load, solar, and wind on aligned weather inputs and increasingly looking to incorporate weather-driven thermal outages to their appropriate probabilities.

ACP-California is interested in further discussions with CAISO and the CPUC regarding the relative tradeoffs of these approaches as the CAISO model is developed further, and encourages CAISO to explore the potential to more directly simulate weather and other correlated risks. Additionally, it is unclear that the CAISO method differentiates solar and wind technologies and regions, which complicates simulation, particular mid- and long-term simulation with an evolving portfolio.

In the long-term, development of a shared dataset for use across proceedings and agencies would be beneficial for both agencies and stakeholders, reducing workloads while focusing review and improvement on a single dataset. ACP-California expects the methods for solar and wind profile development to return in scope to the Commission’s IRP proceeding for the 2025 IRP cycle, as well as Track 3 of the RA proceeding and looks forward to working with the Commission, CAISO, and stakeholders on progressing the accuracy, quality, and alignment of these data and simulation methodologies across modeling venues.

Finally, ACP-California encourages the CAISO to continue its efforts to stress test conditions across a range of water years to reflect potential discrete impacts to system reliability resulting from annual draws with more or less precipitation, impacting both in- and out-of-state hydro fleets. Recognizing the significance of this variable within the RA and IRP frameworks will be important to ensuring reliability in the new era of load growth and constrained regional resources.

[1] I&A Document, 17-18

2. Please provide your organization’s feedback on the preliminary results.

As discussed in Question 1, ACP-California seeks further dialogue and review of the proposed data inputs and simulation methods, which are key inputs to the accuracy and precision of reliability modeling results.

3. Please provide your organization’s input on what types of capacity accreditation methods and PRM approaches should be studied.

ACP-California appreciates CAISO’s efforts to improve methods and rules for resource counting through this process, but urges caution against utilizing current data, methods, or results to inform counting rules without further development and dialogue, as discussed in Question 1.

As an overarching principle, ACP-California continues to encourage alignment between CAISO and CPUC resource accreditation methods. Currently, distinct methods for solar and wind counting are used for the CAISO’s Net Qualifying Capacity (“NQC”) list, the CPUC’s Slice of Day RA Program, and the CPUC’s IRP procurement targets. While some variation is inherent in the differing nature of these programs – single monthly values for NQC, monthly profiles for Slice of Day, and single annual values for IRP – efforts to align data inputs, simulation methods, and conceptual approaches for translation to compliance values can significantly improve alignment between programs.

In developing default counting rules and informing changes to the NQC list, CAISO should consider alignment with CPUC programs. Notably, the CPUC recently issued a Proposed Decision to better align Slice of Day profiles with data and methods used in IRP and previously used for RA Effective Load Carrying Capability (“ELCC”) values in R.23-10-011 (Decision on Track 2 Issues, 10/29/2024, p. 18-19).

With regards to future approaches for calculating the default PRM, ACP-California recommends that 1) CAISO work to better reflect actual showings that will be made by LSEs due to regulatory obligations in future scenario modeling and 2) develop a clearer methodology for the default PRM calculation that aligns with CPUC modeling.

On item 1), modeling performed by CAISO presented in the October workshop does not reflect current regulatory requirements.  CAISO staff did not model the “effective” PRM nor the types of portfolios that LSEs will be showing under Slice of Day.  Not considering either of these factors in Scenarios 1 and 2 produces results that are disconnected with current LSE RA showings. 

For item 2), it remains unclear what methodology CAISO will use to calculate a new default PRM.  Modeling performed in Scenario 3a, where the portfolio was calibrated to a 1 in 10 LOLE, may not be the preferred approach for calculating a default PRM, as similar CPUC analysis has shown.  Simply removing 1810 MW of import RA in the model will lead to inconsistent PRMs across each month, with much higher reserve margins in the winter.  In addition, attempting to apply summer PRMs from Scenario 3a across the winter months will likely lead to a LOLE higher than 0.1.

CAISO’s analysis parallels challenges in the CPUC RA program related to aligning an LOLE metric to monthly obligations which vary significantly. The CPUC intends to revisit a durable long-term approach to PRM calibration in Track 3 of the current RA proceeding. ACP-California encourages CAISO to collaborate with the CPUC to work through these technical and policy questions to align on a shared approach to PRM which provides for alignment between the two agencies.

ACP-California looks forward to working collaboratively with CAISO and stakeholders to discuss potential counting options which fairly and accurately represent resource contributions while aligning, to the degree practical, with CPUC programs.

4. Please provide any additional feedback not already captured.

No additional feedback at this time.

1. Please provide your organization’s feedback on the draft inputs and assumptions.

The Alliance for Retail Energy Markets (“AReM”) thanks CAISO for the opportunity to provide comment on its new reliability modeling efforts. AReM has reviewed the Draft Inputs & Assumptions document dated October 8, 2024 (“I&A Document”) and offers the following comments.

As discussed in past comments, AReM is concerned about differences between modeling assumptions by California Public Utilities Commission (“CPUC”) Staff and CAISO in their respective efforts to model system reliability. Differences could create confusion in interpreting the results. CAISO stated its goal to conform inputs and assumptions to those of CPUC Staff, but differences remain:

• Software: CPUC Staff uses SERVM and CAISO uses PLEXOS for short-term studies.
• Stochastic Modeling Methodology: CAISO creates 500 hourly profiles for its stochastic load and renewable generation modeling and draws load and generation profiles independently; CPUC Staff models 23 weather years based on historical data.
• Hydro Modeling: CAISO models average-year hydro conditions; CPUC Staff models hydro conditions for 23 weather years to include the impact of hydro uncertainty in the loss of load expectation (“LOLE”) results.
• Imports and Topology: CAISO’s “All RA Eligible” scenario includes a net import limit consistent with the CPUC Staff, but models all external generation as one zone instead of the multi-zonal topology used by CPUC Staff; for long-term modeling performed in the Integrated Resource Planning proceeding, CPUC Staff “tunes” external regions to 0.1 LOLE, a practice which should be considered by CAISO for its mid- and long-term modeling.

Given the complexities of stochastic modeling, AReM cannot determine the precise impact these differences will have on the final results of each type of study. AReM recommends benchmarking the different approaches to each other to determine their comparability or further conforming the methodologies for consistency.

AReM also has concerns regarding CAISO’s choice of modeled scenarios. The “All RA eligible” scenario is most comparable to the LOLE modeling by CPUC Staff in the resource adequacy (“RA”) proceeding. It is not clear that the “Showings capped at obligation” scenario adds value because it does not appear to capture the entire showing of CPUC-eligible load-serving entities (“LSE”) under slice-of-day (“SOD”) rules. Based on Figure 4.1 of the I&A Document, CAISO evaluated a minimum obligation for the system stack on an hourly basis and not each individual LSE, even though each individual LSE is required to meet its own load shape. It is also unclear what the benefit of the “Showings based on historical pattern” scenario is, given the major change to SOD rules in 2025. AReM does not expect that showings from CPUC-jurisdictional LSEs will mimic past years. At a minimum, CAISO should consider a scenario that includes all the expected shown capacity from the LSE survey results, which apparently is not included given that both the “Showings capped at obligation” scenario and the “Showings based on historical pattern” scenario have capacity removed per Table 4.2 of the I&A Document.

AReM is also concerned about the accuracy of assumptions CAISO made to supplement for the 28% of load that did not provide a survey response. CAISO should assess its assumptions against the actual year-ahead RA filings to see if the “Showings capped at obligation” and “Showings based on historical pattern” scenario results are meaningful. If they are not, CAISO should consider dropping the LSE survey for the short-term modeling and only modeling the “All RA eligible” scenario which relies on the NQC list and not the LSE survey.

Finally, the I&A Document on page 16 states all battery storage is modeled as 4-hour duration. However, some battery resources in CAISO have lower durations, and over the long-term, the CPUC has ordered procurement of 8-hour+ storage. CAISO should better explain how it models the ability of storage to dispatch longer than four hours at less than Pmax and how it handles the introduction of long-duration storage for mid- and long-term modeling.

2. Please provide your organization’s feedback on the preliminary results.

AReM is concerned that the “Showings capped at obligation” scenario and the “Showings based on historical pattern” scenario are now out-of-date because they are based on LSE survey results and not actual year-ahead filings. LSE survey results and CAISO’s assumptions for those scenarios should be verified against actual RA showings.

AReM may also have further comments on mid- and long-term modeling results when CAISO presents those results.

3. Please provide your organization’s input on what types of capacity accreditation methods and PRM approaches should be studied.

There are differences between CAISO and CPUC resource counting rules, since only the CPUC uses the SOD paradigm. When CAISO presents planning reserve margin (“PRM”) results, it should be clear those results are based on peak-hour capacity accreditation and are not comparable to a SOD-calibrated PRM calculated by CPUC Staff. However, for benchmarking purposes, it would be advisable for both the CPUC Staff and CAISO to calculate a PRM using comparable capacity accreditation.

CAISO should also be clear regarding whether it includes resources on planned outage in its PRM calculation. The “All RA eligible” scenario models planned outages, but the “Showings capped at obligation” and “Showings based on historical pattern” scenarios exclude such outages due to RA market capacity substitution rules. One of the goals of the “Showings capped at obligation” and “Showings based on historical pattern” scenarios is to evaluate the effectiveness of the default and/or implemented PRM(s) in meeting reliability goals. This implies capacity on planned outage should not be in the final default PRM calculation.

4. Please provide any additional feedback not already captured.

AReM has no further feedback at this time.

1. Please provide your organization’s feedback on the draft inputs and assumptions.

The California Community Choice Association (CalCCA) appreciates the California Independent System Operator’s (CAISO’s) efforts to model the reliability of the resource adequacy (RA) fleet for the upcoming year and generally supports the draft inputs and assumptions. The CAISO should make the following improvements, however, to enhance transparency and better reflect the availability of the RA fleet:

• Wind and solar profiles – The CAISO states that it uses stochastic wind and solar profiles based on actual generation as the basis for its model. With the information provided, it is difficult to understand the sampling approach the CAISO uses to derive the profiles. The CAISO should provide increased transparency into its derivation of the profiles by publishing a data file of the profiles for stakeholder review. This would promote confidence in the profiles used by allowing parties to assess the profiles’ correlation with load and perform other checks to ensure the profiles are reasonable.

• Hydro - The CAISO only uses one year, 2018, to set daily minimum and maximum energy limits for dispatchable hydro. 2018 represents an average hydro year, but loss-of-load events in probabilistic modeling tend to be driven by extremes. The CAISO should use multiple hydro years or probabilistic profiles to capture scenarios with above-average and below-average hydro conditions.

Outages - The CAISO’s presentation indicates that forced outages are calculated using technology averages from actual 2006 to 2012 outages. This outage data is outdated, and the CAISO should explore using more current outage data from its Outage Management System to better reflect the availability of technologies providing RA in 2025. As the CAISO evaluates how to update its outage data, it should also consider how to use this data to support the future unforced capacity (UCAP) design. Outage reporting must accurately reflect the reasons for unavailability (maintenance or forced outages) to differentiate which outage types should apply to a UCAP calculation. For example, current outage types used for storage likely do not accurately reflect reasons for unavailability and may result in inaccurate UCAP calculations if they are not clarified.

2. Please provide your organization’s feedback on the preliminary results.

The CAISO’s analysis assessed three scenarios: (1) RA showings capped at the obligation to assess reliability with the minimum amount of capacity needed to meet RA obligations; (2) RA showings based on historical patterns to assess reliability with capacity at levels of historical excess; and (3) all eligible RA resources to assess reliability if all RA eligible resources are shown. The CAISO should use Scenarios 1 and 2 as sensitivities to inform expected levels of reliability based on estimates in advance of RA showings. Because these scenarios use surveys completed in advance of the first binding slice-of-day (SOD) showings for load-serving entities (LSE) in the California Public Utilities Commission’s (CPUC’s) RA program, reliability outcomes from the actual RA showings may differ from those estimated in Scenarios 1 and 2. In the future, the CAISO should consider using actual RA showings for Scenarios 1 and 2 to gauge whether the RA showings meet the 1-in-10 reliability standard. Scenario 3 uses the full portfolio of RA-eligible resources, and therefore should be calibrated to a 1-in-10 loss-of-load event (LOLE), as the CAISO has done in Scenario 3a. Scenario 3 should also be used as the basis for exploring changes to the default counting rules and default planning reserve margin (PRM).

In Scenarios 1 and 2, the CAISO observes risks of LOLE in summer and non-summer months. In Scenario 3, all eligible resources are more than sufficient to meet a 1-in-10 standard and LOLE in the summer months only. The CAISO asks stakeholders if “the PRM should be based on the peak hour.” Concurrently, the CPUC’s RA proceeding is considering PRMs that vary by season. Underpinning the concept of seasonal PRMs is an expectation that the risk of unmet load for a specific level of planning reserves differs across the year. Depending on weather patterns, variability of demand around the 1-in-2 peak demand will differ from month-to-month, along with variability of wind and solar around the level accredited in the SOD, effective load carrying capability, or other accreditation rules. Each month could, therefore, have a unique relationship between the risk of unmet load and the reserve margin. The CAISO should explore these differences to determine if it is necessary and feasible to adopt seasonal default PRMs.

If the CAISO considers whether its default PRMs should vary by season, the primary objective of such a shift should be to minimize ratepayer costs. If modeling indicates a need to increase the PRM to cover LOLE in non-summer months, ratepayer costs can be saved through the adoption of seasonal PRMs. The PRM in non-summer months can increase, reducing LOLE in those months when RA is cheapest, instead of increasing the PRM in all months. As a result, the PRM in the low reserve price months may have to be set to elevated levels such that the LOLE for the low reserve price months is near zero to retain a 0.1 LOLE across the year. Allocating the annual 0.1 LOLE budget to months where procuring additional reserves is the most expensive could result in ratepayer savings and should be considered when the CAISO evaluates annual versus seasonal PRMs.

The CAISO will need to consider other factors to evaluate the advantages and feasibility of a shift from annual to seasonal PRMs. It would need to characterize variability in peak demand around the 1-in-2 peak day for each month and evaluate monthly and seasonal differences in supply variability. The CAISO’s evaluation methodology will also need to ensure that the generation resources needed to meet each month’s PRMs have sufficient opportunity for planned outages. If the PRM is too high in low-demand months, all resources may need to be shown, eliminating opportunities for maintaining the generation fleet and adversely impacting reliability.[1]

[1]            If the PRM results in an inability for resources to take reasonable maintenance outages, such a result should be communicated to LSEs for their Integrated Resource Planning processes to ensure that additional new resources are procured to enable reasonable outages without reliability impacts.

3. Please provide your organization’s input on what types of capacity accreditation methods and PRM approaches should be studied.

The CAISO should study the UCAP capacity accreditation method and align the PRM with the UCAP method. UCAP provides incentives to resources to perform maintenance that supports reliable operation of the resources by attributing forced outage performance metrics to resources’ capacity values. Unit-specific forced outage rates allow LSEs to assess the reliability of specific resources when making contracting decisions rather than spreading forced outage rates through the PRM. It also simplifies the RA program and would allow the CAISO to eliminate its Resource Adequacy Availability Incentive Mechanism tool.

Capacity accreditation and the PRM are inextricably linked. When using UCAP capacity accreditation, the PRM should not include forced outages because they are already accounted for in resources’ net qualifying capacity (NQC). The CAISO should also ensure that any adoption of UCAP: (1) is coordinated with the CPUC, as the CPUC has also been considering transitioning to a UCAP method for several years in its RA proceedings; and (2) includes the definition of new terminology in a way that does not have unintended impacts on existing contracts.[1]

[1]            For example, CAISO’s previous UCAP proposal in the RA Enhancements initiative defines the terms “NQC” and “Deliverable Qualifying Capacity” in a manner that would not disrupt existing contracts while aligning the must-offer obligation and counting rules with the new UCAP method. The CAISO should retain such an approach so that the implementation of UCAP does not result in the renegotiation of existing contracts. CAISO Resource Adequacy Enhancements, Draft Final Proposal – Phase 1 and Sixth Revised Straw Proposal, at 74: https://stakeholdercenter.caiso.com/InitiativeDocuments/DraftFinalProposal-SixthRevisedStrawProposal-ResourceAdequacyEnhancements.pdf.

4. Please provide any additional feedback not already captured.

CalCCA has no additional feedback at this time.

1. Please provide your organization’s feedback on the draft inputs and assumptions.

On the slide #35, CAISO indicates that solar and wind resources are modeled with nameplate capacity which contradicts the slide #61 statement that exceedance values from past eight years are used. As stated, stochastic modeling will have 500 samples for each hour. Using nameplate capacity for all the hours may not reflect the potential available capacity resulting in overcounting on those resources. Apparently, for resources other than solar and wind, modeling for resource adequacy (RA) capacity and associated reliability assessment is based on the NQC values of resources, not based on the nameplate capacity.

2. Please provide your organization’s feedback on the preliminary results.

The result of modeling is a step forward in assessing reliability in short term, near term and the long term.

Slide #66 indicates excess / shortfall  in the portfolio to be calibrated to 0.1 Loss of Load Expectation(LOLE) is measured based on NQC MW. As stated above, it should be clarified whether solar and wind resources’ capacity modeled is based on either: i) Name plate capacity, ii) Exceedance and the exceedance level considered, or iii) NQC values.

CDWR notes the result showing sufficient capacity to meet 0.1 LOLE in the “Scenario 3” where all RA eligible resources are considered.

3. Please provide your organization’s input on what types of capacity accreditation methods and PRM approaches should be studied.

CDWR has no comment at this time.

4. Please provide any additional feedback not already captured.

CAISO indicates that this initiative is not designed to expand CAISO jurisdiction over resource adequacy (RA), is not aimed at setting minimum requirements for all LSEs regardless of jurisdictional status and is not an assessment of individual LRA RA programs. As stated in prior comments, CDWR supports the defined scope in recognition of LRA jurisdictional boundaries.

1. Please provide your organization’s feedback on the draft inputs and assumptions.

Please see the attached Comments from the Department of Market Monitoring.

2. Please provide your organization’s feedback on the preliminary results.

Please see the attached Comments from the Department of Market Monitoring.

3. Please provide your organization’s input on what types of capacity accreditation methods and PRM approaches should be studied.

Please see the attached Comments from the Department of Market Monitoring.

4. Please provide any additional feedback not already captured.

Please see the attached Comments from the Department of Market Monitoring.

1. Please provide your organization’s feedback on the draft inputs and assumptions.

California Public Utility Commission Energy Division Staff (aka ED Staff) provides the following comments on the California Independent System Operator Corporation’s (CAISO) Draft Inputs & Assumptions Track 1: CAISO Resource Adequacy Modeling[1] and discussed at the October 8, 2024, working group meeting.  Herein, ED Staff comments are focused on the year-ahead 2025 study, as the results and assumptions for the mid-term and long-term study will be released later this year.

ED Staff appreciates CAISO’s efforts to enhance its reliability requirements and RA program in light of the evolving resource mix. Through these efforts CAISO staff have consulted and socialized its modeling approaches and results with ED Staff in an effort to stay as aligned as possible with CPUC’s reliability modeling efforts. ED Staff appreciates this close collaboration and looks forward to further collaboration as CAISO refines and enhances its reliability modeling.     

In CAISO’s recent Track 1 study results paper and working group slides, it states that the objective of Track 1 “aims to enhance CAISO’s resource adequacy modeling capabilities by conducting a probabilistic assessment to evaluate the sufficiency of the CAISO BAA’s resource adequacy (RA) portfolio in the year-ahead, medium-term (2-4 years) and long-term (5-10 years) timeframes to meet reliability objectives. In addition, this track will also consider updating CAISO’s default resource counting rules and default planning reserve margin (PRM) to reflect reliability contribution of different resource types in a portfolio that achieves a “one day every 10 years loss-of-load expectation” (“1-in-10 LOLE”) planning target.”

ED Staff is supportive of CAISO updating its default resource counting rules and PRM in its tariff for purposes of using them only when a local regulatory authority (LRA) has not established its own rules. While not explicitly called out in the slides or proposal, ED Staff is not supportive of CAISO using the default values for any other purposes such as backstop procurement or an assessment of LRA’s RA programs.  Staff requests that CAISO make this explicitly clear in future iterations of its proposal.  As noted in prior comments, ED Staff is concerned about a possible disconnect between front-stop decisions (i.e., LRA RA requirements) and backstop decisions.[2]  ED Staff is not supportive of the latter as it would create two different RA programs and paradigms. This would result in two sets of different RA requirements, which could result in double payments, penalties, and unnecessary and/or expensive backstop procurement.

Specific recommendations on draft inputs and assumptions

Forced outage rates

As mentioned on the stakeholder call, the generation forced outage data is from 2006-2012.  Generation units that were online then are now twelve years older and may not have the same level of reliability.  Staff requests that the CASIO update the outdated forced outage data.  ED Staff has recently updated its forced outage rate data used in its reliability modeling and would be happy to coordinate with the CAISO to update this data input.

Planned Resource (resources not yet on-line)

In all three YA (2025) modeling scenarios CAISO utilized load serving entity (LSE) survey data to inform the number of MWs that are expected to be online to meet reliability needs in 2025.  Scenario 3 specifically uses the NQC list plus survey data on expected additions and retirements and is most comparable to the way the CPUC modeled reliability in its recent 2026 RA LOLE analysis.  Staff recommends that CAISO coordinates with the CPUC’s Integrated Resource Plan (IRP) LSE data submission process and timeline in future studies to leverage CPUC jurisdictional LSE data that CPUC is using in its modeling assumptions.  Leveraging this process will also reduce the administrative burden put on LSEs that is largely duplicative of what is being requested in the IRP data filings.   

[1] https://stakeholdercenter.caiso.com/InitiativeDocuments/Track1Modeling-DraftInputsandAssumptions.pdf

[2] CPUC Energy Division Staff (October 20, 2023) Energy Division Comments on CAISO Resource Adequacy

Modeling and Program Design Initiative.

2. Please provide your organization’s feedback on the preliminary results.

ED Staff notes that scenario 3 (all RA eligible resources) results indicate that there is sufficient resource sufficiency in 2025, and is similar to the ED Staff’s modeling results for 2026, to meet the 1 day in 10 years outage standard.

3. Please provide your organization’s input on what types of capacity accreditation methods and PRM approaches should be studied.

ED Staff appreciates the coordination the CAISO is doing with CPUC ED and the California Energy Commission (CEC) Staff on Resource Adequacy and looks forward to continuing partnership.  As noted in CAISO’s presentation at the October 8, 2024, workshop, the CPUC jurisdictional load in the CAISO balancing authority area is about 91%.[1]  Therefore, whenever possible it makes sense for the CAISO and the ED Staff to coordinate to use similar inputs (such as resource counting and load forecast approaches) and assumptions for the annual PRM calculation.[2]   As noted above, to reduce the reporting burden on LSEs and align resource assumptions with the CPUC’s, CAISO  should leverage the CPUC’s IRP LSE data submission process and timeline. 

The CPUC has just begun its implementation of the slice of day (SOD) RA framework which uses a different set of resource accreditation methods and PRM calculation from the CPUC’s prior RA framework.  For resource accreditation for variable energy resources (VERs), SOD uses an exceedance based worst day hourly profile for each month instead of the historical effective load carrying capability (ELCC) method.  While CAISO may not be pursuing a SOD framework change to its program at this time, CAISO should seek to align its accreditation methodology to something that can best represent the VERs in the most constrained hour of the month to ensure that they are not over or undervalued at the time of stressed reliability. 

Finally, having consistent assumptions will help achieve equity across the LRA areas to the extent other LRAs use CAISO’s default PRM.

[1] CAISO, (October 8, 2024) Year ahead Showings Assessment: Review of Modeling and Preliminary Results, slide 24.

[2] The CPUC’s RA is moving toward the use of monthly PRM values that may be different from the traditional single annual PRM value that is used for resource planning.  In addition, the CPUC has moved to the Slice of Day (SOD) RA program for the 2025 compliance year and recognizes that other LRAs may not be able to implement a similar program. 

4. Please provide any additional feedback not already captured.

The recent scoping memo[1] in the CPUC RA proceeding states that “Energy Division will coordinate with CAISO to develop a[n unforced capacity] (UCAP) accreditation methodology for thermal power plants and battery electric storage systems. The Commission will consider Energy Division’s UCAP proposal in Track 3.”  This is another area where alignment between the CAISO and ED Staff would be beneficial. 

[1] https://docs.cpuc.ca.gov/PublishedDocs/Efile/G000/M544/K652/544652400.PDF

1. Please provide your organization’s feedback on the draft inputs and assumptions.

The Public Advocates Office at the California Public Utilities Commission (Cal Advocates) is the independent ratepayer advocate at the California Public Utilities Commission (CPUC).  Our goal is to ensure that California ratepayers have affordable, safe, and reliable utility services while advancing the state’s environmental goals.  Cal Advocates offers the following recommendations on the California Independent System Operator’s (CAISO) draft inputs and assumptions for the Resource Adequacy Modeling and Program Design Initiative:

• The CAISO should develop a shaped PLEXOS[1] import constraint in place of the proposed draft PLEXOS import constraints.  The shape of the import constraint should allow for flows from both specified and unspecified resource adequacy (RA) import resources.  An hourly shaped import constraint would improve upon the draft inputs and assumptions by properly accounting for the hourly shape of generation from imported variable energy resource (VERs), in addition to unspecified RA import levels. 
• The CAISO should revise the set of resource identifiers (IDs) that contribute to the import constraint, to remove: a) any capacity associated with retiring resources, b) resources that serve non-CAISO load-serving entities (LSEs), and c) system resources that function as unspecified imports for modeling purposes.  The removal of these resource IDs will result in a more accurate representation of the specified RA import capacity available to CAISO LSEs.

Background: The CAISO’s draft inputs and assumptions documentation includes two sets of import constraints for the CAISO’s 2025 modeling.  For Scenario 1, where the resource level is capped at LSEs’ RA obligations, and Scenario 2, where the resource level is capped at LSEs’ historical RA showings, the CAISO’s draft import constraints use the average 2019-2024 import levels of qualifying capacity.[2]  For Scenario 3, where the resource level includes all RA-eligible resources, the CAISO’s draft import constraints include a “net import limit” that includes both RA imports and economic imports.  This Scenario 3 draft net import limit is set to 5,500 megawatts (MW) in hours ending 16 through 22 from June through September and to 11,665 MW, or the portion of the Maximum Import Capability (MIC) available to LSEs, for all other hours.[3]  CAISO staff also provided the following total import constraint values by month for each scenario, in an October 28, 2024 email response to a question from Cal Advocates:[4]

CAISO staff provided the following tie-generator category breakdown for Scenarios 1 and 2.  The CAISO described these MW values as “[Qualifying Capacity (QC)] amounts based on historical RA showings”:[5]

Finally, the CAISO provided the following tie-generator category breakdown for Scenario 3 (nameplate MW).  These capacities reflect the summed capacities of all imports with CAISO resource IDs:

Discussion: The CAISO’s draft import constraints are inappropriate inputs for two reasons.  First, the draft import constraints do not match the hourly shape of generation from specified imports, which include a significant and growing proportion of VERs.  For Scenarios 1 and 2, the import constraint allowances for solar and wind resources are capped at “QC amounts based on historical RA showings,” according to the CAISO’s October 28, 2024 email to Cal Advocates.  However, VERs produce shaped generation profiles in which a VER’s generation level may be lower or higher in any given hour, compared to the VER’s QC.  For example, solar resources’ generation can approach their nameplate MW capacity levels in the middle of a summer day, while dropping to zero during nighttime hours.  The use of an import constraint that reflects a fixed import flow from solar resources, at the corresponding QC levels, may artificially reduce solar imports in midday hours.

Similarly, for Scenario 3, the 5,500 MW import constraint for hours ending 16 through 22 in June through September may be lower than the total hourly generation from specified and unspecified RA imports during this same time period.[6]  This import constraint risks the artificial reduction of energy from RA imports.

The second reason that the CAISO’s draft import constraints are inappropriate inputs is that the Scenario 3 import capacities are based on the total set of external CAISO resource IDs, rather than the subset of these resources that serve CAISO load.  The CAISO’s use of the larger set of resource IDs overestimates the specified import capacity that is actually available to CAISO LSEs.[7]  For example, Scenario 3 includes 2,549 nameplate MWs of specified gas imports, whereas the CAISO’s calculation of the LSEs’ 2019-2024 RA showings included a September average of only 469 MWs of specified gas imports (see charts above).  Likewise, Scenario 3 includes 1,875 nameplate MWs of specified hydro imports, whereas the CAISO’s calculation of the LSEs’ 2019-2024 RA showings included a September average of only 419 MWs of specified hydro imports (see charts above). 

The primary cause of the excessive Scenario 3 nameplate levels appears to be the inclusion of some external resources that do not regularly serve CAISO load but nonetheless retain CAISO resource IDs.[8]  Some of these resources may be contracted to serve external load, such as GRIFFI_2_LSPDYN (Griffith Energy).[9]  Others may be so-called “system resources” used for market participation by multiple external resources participating as a single resource.  For modeling purposes, such system resources are akin to unspecified imports and should not be treated as specified imports.  For example, the resource ID “MALIN_5_BPADYN” does not represent a pseudo-tie or a dynamically scheduled VER but rather the system resource that allows Bonneville Power Administration (BPA) to market up to 700 MWs of capacity from the ten hydro facilities of the Federal Columbia River Power System.[10] 

Finally, in all three scenarios, the CAISO includes 477 nameplate MWs of specified coal imports in all months.  This level corresponds to the CAISO LSEs’ shares of the Intermountain Power Plant, which is set to retire in June 2025.[11]  Therefore, the CAISO’s inclusion of this resource in later months will overstate the reliability of the system in those months.

Recommendation: The CAISO should make two changes to address the aforementioned issues with the draft import constraints.  First, the CAISO should review the set of specified import resource IDs and exclude import capacity from any resource IDs whose capacity retires, serves non-CAISO load, or represents a system resource that does not generally provide CAISO RA as indicated by the LSEs’ RA showings.  Next, the CAISO should develop a shaped hourly import constraint that allows for specified RA imports to flow to the CAISO, including imported VERs whose generation varies from their QC values hour-by-hour.[12]  This shaped hourly import constraint should also include an allowance for unspecified RA imports.

[1] PLEXOS is the CAISO's production cost modeling software.

[2] CAISO, Draft Inputs and Assumptions: Track 1: CAISO Resource Adequacy Modeling, October 8, 2024 at 5.  Available at: https://stakeholdercenter.caiso.com/StakeholderInitiatives/Resource-adequacy-modeling-and-program-design.

[3] CAISO, Draft Inputs and Assumptions: Track 1: CAISO Resource Adequacy Modeling, October 8, 2024 at 5.  Available at: https://stakeholdercenter.caiso.com/StakeholderInitiatives/Resource-adequacy-modeling-and-program-design.

[4] In an October 28, 2024 email to Cal Advocates, CAISO staff also explained:

External resources are modeled as a combination of specified imports (those with a CAISO resource ID) and unspecified imports (up to the import constraint for each scenario). Conventional dynamic schedule resources were modeled as individual units, hydro, solar and wind dynamic schedule resources were modeled as aggregated resources, plus unspecified imports.  Specified imports are always dispatched in priority to unspecified imports.  [The] Scenario 1 import limit is adjusted to cap total resources at the 2025 estimated obligation.  [The] Scenario 2 import limit is adjusted to cap total resources at the 2025 estimated obligation plus historical excess showings by month.  [The] Scenario 3 import limit is allowed up to the maximum import capability (MIC) for non-summer months and capped at 5,500 MW for summer months (5500 MW for HE16-22, and 11,665 MW for the other hours).

[5] In an October 28, 2024 email to Cal Advocates, the CAISO explained:

There are two different resources lists for external resources, one for Scenarios 1&2 (QC amounts based on historical RA showings), and one for Scenario 3 (Nameplate amounts).  The tables below show the amounts by fuel.  These are the totals of specified resources that would be dispatched first, followed by non-specified resources up to the limits shown in the table above.

[6] The CAISO’s inputs and assumptions document itself notes that

modeled import capacity on ties and all pseudo-tied and dynamic imports from out of state generators . . . exceeds the net import limit of 5,500 MW enforced in the “All RA eligible” in some summer months as indicated by “Tie-generators” and “Import RA on ties” values in Table 4.1. 

CAISO, Draft Inputs and Assumptions: Track 1: CAISO Resource Adequacy Modeling, October 8, 2024 at 10.  Available at: https://stakeholdercenter.caiso.com/StakeholderInitiatives/Resource-adequacy-modeling-and-program-design.

[7] This issue affects Scenario 3 hours ending 1 through 15, and 23 to 24 in June through September, when the high draft import constraint enables the importing of all resource IDs that the CAISO characterizes as “tie generators.”

[8] Cal Advocates bases this statement on its reconciliation of the resource category volumes provided by the CAISO, as shown in the tables above, with the resource IDs on the “Specified Imports” tab of the CPUC’s Master Resource Database, available at: https://www.cpuc.ca.gov/industries-and-topics/electrical-energy/electric-power-procurement/resource-adequacy-homepage/resource-adequacy-compliance-materials.

[9] Cal Advocates’ review indicates that Griffith Energy is under a long-term contract with Arizona Public Service Company.  See Comments of the Public Advocates Office on Proposed Loss of Load Expectation Inputs and Assumptions (Public Version), April 2, 2024 at 11.  Available at: http://docs.cpuc.ca.gov/SearchRes.aspx?DocFormat=ALL&DocID=529073512.

[10] Bonneville Power Administration, Agreement No. 14TX-15976: Dynamic Transfer Operating Agreement Executed by the United States of America Department of Energy Acting by and through the Bonneville Power Administration and California Independent System Operator Corporation, Exhibit C, Table 2.  Available at: https://www.bpa.gov/-/media/Aep/foia/foia-2022/BPA202201074F.pdf.

[11] The CPUC’s baseline resource list uses June 2025 as the final month for the Intermountain Power Plant, and the Intermountain Power Agency continues to report changes in net position due to June 2025 as the intended retirement month for the coal units.  See Intermountain Power Agency, Financial Statements as of and for the Years June 30, 2024 and 2023, Supplemental Schedule for the Years Ended June 30, 2023 and 2024, and Independent Auditor’s Report at 7.  Available at: https://www.ipautah.com/wp-content/uploads/2024/09/IPA-Issued-Financial-Statements-FY24.pdf.

[12] Cal Advocates would also support an alternative implementation of this recommendation that would involve the exclusion of specified RA imports from the import constraint in combination with the reduction of the import constraint to reflect only unspecified imports (both RA and economic unspecified imports).  During summer peak hours specifically, the CAISO should lower the import constraint to allow unspecified RA imports only.  Cal Advocates suggests a 4,000 MW allowance for unspecified import RA, consistent with the CPUC’s assumptions.  See CPUC, Inputs and Assumptions: 2022-2023 Integrated Resource Planning (IRP), October 2023 at 160.  Available at: https://www.cpuc.ca.gov/-/media/cpuc-website/divisions/energy-division/documents/integrated-resource-plan-and-long-term-procurement-plan-irp-ltpp/2023-irp-cycle-events-and-materials/inputs-assumptions-2022-2023_final_document_10052023.pdf. 

2. Please provide your organization’s feedback on the preliminary results.

 Cal Advocates does not have comments at this time.

3. Please provide your organization’s input on what types of capacity accreditation methods and PRM approaches should be studied.

 Cal Advocates does not have comments at this time.

4. Please provide any additional feedback not already captured.

 Cal Advocates does not have additional feedback at this time.

1. Please provide your organization’s feedback on the draft inputs and assumptions.

CESA is pleased with CAISO’s initial loss of load expectation study efforts. It is clear and encouraging that CAISO staff listened to stakeholders and incorporated the feedback it gathered earlier this year into its design of the study scenarios. Further, the CAISO provided a good framework for documenting the study inputs and assumptions. CESA is also pleased with CAISO’s continued stakeholder engagement on the modeling, its transparency, the robust discussions at the workshops, and CAISO’s general openness to improving its modeling with the help of the vast and diverse knowledge of the stakeholder community. CESA looks forward to continuing to engage with CAISO staff throughout this initiative.

These comments focus primarily on improvements to the energy storage modeling and improvements to the documentation of the inputs and assumptions to ensure that there is a clear and transparent record of the treatment of energy storage resources in the modeling. With over 13 GW of energy storage now on the CAISO system, it is important that the modeling of this resource class is as accurate as possible. The up-front work CAISO must do to define and clarify the documentation on the inputs and assumptions will be extremely beneficial as CAISO iterates on this study methodology from year to year.

CAISO should calculate an accurate effective forced outage rate and maintenance outage rate for energy storage resources. It is important to ensure the outage modeling for such a large resource class is accurate.  CAISO stated that the draft inputs and assumptions used a single outage rate calculated over an extremely limited sample size (rate based on eight resources over a few months in 2020). CAISO has since accumulated much more outage data which can be used to produce both a forced outage rate and a maintenance rate. However, CAISO must be diligent in its treatment of outages when it calculates the outage rates. The CAISO must not incorporate outages used to manage energy storage state-of-charge into its calculated forced outage rates. Forced outage rates for energy storage resources should reflect plant failures but not state-of-charge because the model already accounts for state-of-charge when dispatching energy storage resources.

CAISO should clearly document the methodology it uses to calculate the forced outage rate and maintenance outage rate. Given the various outage types and natures-of-work, it is important for stakeholders to have a transparent description of the methodology CAISO uses to synthesize its outage data into outage rates.

CAISO should clearly document any other parameters it uses in the modeling of forced outages.  The models used to perform loss of load expectation studies are complex and include parameters such as time-to-repair, effective forced outage rate, effective forced outage rate on demand, maintenance rate, maintenance frequency, etc. CAISO should investigate each parameter/setting used in its modeling to make sure it accurately reflects the operation of energy storage in the CAISO balancing authority area. CAISO should also clearly document the settings being used along with its rationale for using such setting. Although not all stakeholders may agree with the settings, they will appreciate the transparency. The transparency will also allow for meaningful discussions to continuously improve the modeling each year.

CAISO should model many energy storage resources with PMAX values that reflect the energy storage resources currently on the system.  It appears that CAISO may be using many resources of equal size (e.g. 100 resources with 100 MW nameplate each) in its modeling.  It would be more accurate to have the size of the resources reflect the size of the energy storage resources currently on the CAISO system. In the stochastic analysis, losing a 100 MW resource to a forced outage will have a very different impact than losing a 400 MW resource. With 13 GW of energy storage resources of varying sizes on the CAISO system and with energy storage resources impacting both the instantaneous ability and the future ability to avoid load loss in the model, this improvement could be impactful on the results.

CAISO should ensure that the modeled energy storage round-trip efficiency reflects the actual round-trip efficiency of its energy storage fleet. It appears that CAISO may be using a single round-trip efficiency of 85% for all energy storage resources in its model. It would be more accurate to have the individual round-trip efficiencies of the modeled resources to match those of the resources on the CAISO system.  If CAISO continues to model a single round-trip efficiency, it should ensure that modeled round-trip efficiency reflects the round-trip efficiency of the resources on the CAISO system by potentially modeling a weighted average round-trip efficiency.

CAISO should accurately model the duration of the energy storage resources on the system. Although it may not provide a material impact on the current results due to most of the energy storage resources on the system having a 4-hour duration today, it is important to establish an accurate modeling of energy storage duration in the study methodology. Longer or shorter duration energy storage resources have a different contribution to reliability than 4-hour duration energy storage resources. As longer duration energy storage resources come online soon, CAISO’s loss of load expectation study model should seamlessly incorporate their contribution to reliability.

CAISO should clarify the energy storage dispatch strategy its model uses when expected unserved energy is inevitable in the inputs & assumptions documentation and discuss the pros and cons of various energy storage dispatch strategies that could be employed when expected unserved energy is inevitable.  As discussed at the workshop on October 8, 2024, the production cost model could dispatch energy storage resources a variety of ways when expected unserved energy is inevitable because the expected unserved energy is of equal quantity and cost in the model under multiple dispatch strategies.  For instance, the energy storage resource may be used in a first-come first-served manner in which the energy storage resource fleet is used in time-series order until completely exhausted.  The model may also be configured to minimize the depth or the duration of the expected unserved energy. These various dispatch strategies may have different implications on the determination of the planning reserve margin or resource accreditation, so it would be helpful for CAISO to report on its considerations in establishing the dispatch strategy.

CAISO should investigate increasing the optimization horizon beyond 48-hours, especially as it applies to its mid-term and long-term modeling efforts. Long duration energy storage resources are currently being procured and built. Multi-day storage resources (up to 100 hours duration) are also being developed and will be procured as long lead time resources. These resources provide reliability value to the CAISO over time horizons that span beyond 48-hours. CAISO’s model will need to incorporate longer optimization horizons to reflect this reliability value in its loss of load expectation results.

2. Please provide your organization’s feedback on the preliminary results.

CESA recognizes that the preliminary results are based on a set of draft inputs and assumptions that are still under development. The accuracy of the assessment will be significantly improved as the CAISO incorporates more comprehensive data and refines the underlying assumptions.

Although the purpose of this study is forward-looking (2025), CAISO should also consider annually executing the methodology for scenarios one and two for the past year, given it has all RA and Supply Plan data. This would provide CAISO a helpful benchmark to measure the accuracy of its forward-looking results year after year.

3. Please provide your organization’s input on what types of capacity accreditation methods and PRM approaches should be studied.

 No comment at this time.

4. Please provide any additional feedback not already captured.

 No comment at this time.

1. Please provide your organization’s feedback on the draft inputs and assumptions.

Middle River Power LLC (“MRP”) appreciates the CAISO creating hypothetical scenarios to identify the potential loss of load expectation (“LOLE”) risk of various resource portfolios.  That said, MRP disagrees with the make-up of the hypothetical portfolio of scenario 1. Instead of adjusting the scenario 1 portfolio based on previous planned outages, MRP believes the scenario 1 portfolio should be made up of resources that achieve the most decarbonization.  Nuclear and storage should be included and natural gas-fired resources should be removed from the portfolio in this scenario.  This scenario would provide stakeholders with a better understanding of the level of reliability that is achieved by this type of portfolio without any changes to the RA program’s PRM, which depends on resource counting.  The current decision to exclude some of the storage and nuclear resources based on historic outage submissions seem more apt for scenario 2 rather than scenario 1, because the capacity lost due to all planned outages is expected to be replaced through substitution.  MRP agrees with the decision to assume planned outages are fully substituted for on a monthly basis.  This decision aligns with the current CAISO policy that capacity under planned outage must be replaced through substituted or the outage will be canceled. 

MRP would like the CAISO to focus on scenario 3a that was not included in the inputs and assumptions document but was included in the presentation.  This scenario is somewhat helpful in potentially identifying the annualized portfolio necessary to maintain a 0.1 LOLE.  However, it is not consistent with any monthly RA program.  As the CAISO is aware, LSEs’ monthly RA obligations are based on the monthly load shapes.  This is why scenario 1 and 2 are “limited” to only the monthly obligations.  Therefore, MRP strongly recommends that CAISO “stress test” its scenario 3a by limiting the RA portfolio to that needed to meet a minimum annual PRM to ensure 0.1 LOLE. 

While the CAISO uses stochastic variables for wind and solar generation, the CAISO models the hydro fleet using an “average hydro year (2018).  Given the significant variation that can occur in daily and seasonal hydro energy – and even daily and seasonal hydro availability (compare, for example, the different between hydro production in 2019 and 2021), MRP believes that the CAISO should consider modeling hydro in a way that better accounts for the potential variation in availability and production across a spectrum of water year conditions.  This could involve “dry” and “wet” year sensitivities, or project the likelihood that the coming year will be wet or dry based on historical patterns (e.g., El Nino vs. La Nina). 

2. Please provide your organization’s feedback on the preliminary results.

MRP finds the preliminary results reinforce the issues that MRP has been highlighting for several years.  First, the current PRMs, along with the current capacity accreditations, are not capable of achieving the 0.1 LOLE standard.  The current LOLE analysis based on hypothetical portfolios show that the CAISO’s default PRM urgently needs to be updated to guide LRAs in designing the requirements for their own RA programs.  Second, the CAISO’s peak demand-based RA program is incompatible with the CPUC’s Slice-of-Day (“SOD”) RA program and cannot determine whether reliability is achieved using the CPUC’s SOD PRM.  The SOD PRM is expected to be lower than a peak demand-based PRM because the SOD program can spread out storage capacity utilization over multiple hours rather than stacking it to meet a single hour requirement.  Utilizing the SOD PRM for the CAISO’s peak demand-based program will lead to an inaccurate and artificial conclusion that there is significant surplus capacity.  Such a finding will limit the CAISO’s ability to procure needed capacity under the current Tariff.

3. Please provide your organization’s input on what types of capacity accreditation methods and PRM approaches should be studied.

MRP supports the CAISO’s efforts to model resource adequacy within its Balancing Authority Area (BAA) to assess BAA reliability as measured by LOLE (or, going forward, by some other mutually agreed-upon adequacy metric).  

First, MRP reiterates that it believes a critical first step in the CAISO’s modeling process should be to assess adequacy reliability using the current year’s actual RA showings, particularly that of 2024 given that December 2024 showings have been submitted at this point.  As the CAISO’s preliminary results (presented at the October 8 meeting) demonstrate, measured reliability varies widely - from a LOLE of 0.782 to a LOLE of 0.024 - based on the portfolio assumed.  This wide range of results strongly suggest that starting with a known portfolio – namely, the current year’s shown RA fleet – will provide a much-needed reference point for the CAISO and its stakeholders to determine that the modeling effort yields reasonable and dependable results before applying the modeling effort to a speculative portfolio (i.e., a portfolio that projects what a future showing, or future RA requirements, will be).  While studies that involve speculative portfolios are not without value, such studies do not provide the “grounding” necessary to validate CAISO’s modeling approach.

Second, MRP respectfully urges the CAISO to move towards a modeling approach that aligns with the CPUC’s SOD framework.  As discussed at the October 8 meeting (in particular, by SCE’s representative), PRMs that are developed from analysis using a peak demand-hour framework and associated capacity accreditation rules with PRMs developed from analysis using the SOD framework and its associated capacity accreditation rules are not compatible and comparing those different PRMs is likely of little value.   

Third, CAISO must stress test its scenario 3a so that it’s applicable to a monthly RA program.  Additional information regarding scenario 3a should be included in the CAISO inputs and assumptions document to clearly detail this scenario. 

Finally, to facilitate the CPUC’s SOD framework going live in 2025, the CAISO must use an appropriate PRM that is applicable to a peak demand-based program to avoid artificial surplus within the CPUC LSEs’ showings. 

4. Please provide any additional feedback not already captured.

Default PRM

The CAISO has stated that it wishes to determine a default PRM to update its tariff.  MRP strongly supports this endeavor.  However, it is unclear which type of forward-looking assessment should be used to set the default PRM.  MRP believes different types of assessments will impact the calculation and results significantly.  MRP recommends further discussion on this topic.

Outage rates

The CAISO must update its forced outage and planned outage assumptions so that the LOLE analysis reflects the current capabilities of the plants.  In January of this year, the CAISO observed that recent forced outage rates have been higher than those assumed in state planning studies.[1]  Yet, the CAISO’s forced outage assumptions used in the LOLE analysis is based on data from 2009 – 2014.[2]  This CAISO must use the most up-to-date outage data available.

[1] CAISO Presentation for January 16, 2024 Resource Adequacy Modeling and Program Design Working Group Meeting at slide 23 (“Key Takeaway: The planned and forced outage rate exceeds assumptions used in state analysis (CAISO 2023 Summer Assessment; CEC 2023 Summer Reliability Outlook”).   MRP notes that the date on the title slide for this presentation is January 16, 2023, but the deck was used for the January 16, 2024 meeting. 

[2] CAISO October 8, 2024 Draft Inputs and Assumptions Track 1: Resource Adequacy Modeling at page 20. 

1. Please provide your organization’s feedback on the draft inputs and assumptions.

SDG&E appreciates the CAISO providing detail and background regarding the RA modeling inputs and assumptions. SDG&E does not have recommendations at this time but requests clarification on one item. It would be helpful if the next iteration of the Draft Inputs & Assumptions document provided the source and derivation details for the Min/Max Flow Ratings (MW) provided in table 6.1.

2. Please provide your organization’s feedback on the preliminary results.

No comment.   

3. Please provide your organization’s input on what types of capacity accreditation methods and PRM approaches should be studied.

Overall, SDG&E recommends that the CAISO and CPUC work together to minimize discrepancies between RA programs. Regarding capacity accreditation, CAISO should use the exceedance methodology to maintain consistency with the CPUC’s Slice-of-Day approach. CAISO should also consider calculating the PRM consistent with the intra-year intervals used by the CPUC. Finally, CAISO’s white paper outlining the interactions between the CPUC and CAISO RA processes is helpful, but it only addresses near-term implementation. SDG&E encourages CAISO to consider providing updated information about mid-term CPUC and CAISO RA interactions as the CPUC’s Slice-of-Day program evolves.

4. Please provide any additional feedback not already captured.

Slide 9 from the October 8 meeting lists “consider development of a UCAP mechanism” as a current objective of Track 1. On October 29, the CPUC issued a draft proposed decision on Track 2 of its RA program, stating:

We note that Energy Division has been working on a UCAP methodology for over a year and CAISO will be initiating a stakeholder process on a UCAP methodology. As such, Energy Division should coordinate with CAISO to develop a UCAP accreditation methodology for thermal power plants and battery electric storage systems for consideration in advance of the 2028 RA compliance year and to submit a revised UCAP proposal in Track 3 of this proceeding… Energy Division should harmonize its UCAP proposal with CAISO, to the extent possible, and coordinate on critical issues… We encourage Energy Division to coordinate with CAISO to develop data acquisition and analysis procedures using alternative public sources, to the extent possible, for a UCAP methodology and to develop a protocol with CAISO to account for missing or outlier data. (PD, pp. 21-23)

SDG&E notes it is unclear at this time how the mismatch between the deadline for Track 3 proposals for the CPUC’s RA program (currently January 17, 2025), and CAISO’s Track 1 schedule, will impact this effort. SDG&E encourages CAISO to coordinate with Energy Division on this matter.  

1. Please provide your organization’s feedback on the draft inputs and assumptions.

SCE appreciates the work that CAISO has put into the analysis. The general conclusion that reliability is improving over time, and for 2025 when including all eligible RA resources shows a capacity surplus of 1,810 MW over the capacity needed to meet 0.1 LOLE confirms what others studies have reported.  However, SCE is concerned with several of the inputs and assumptions and believes that the CAISO and stakeholders should continue to explore possible modifications to the inputs and to discuss their potential impacts on the results.

• The model structure with four regions and one external region with a load/supply/price curve and hurdle rates for imports and exports is one potential issue of concern for SCE. Recent market experience has shown that reliability problems can arise in the CAISO due to weather conditions in the other western areas, and that often these weather conditions are not uniform across the west. This history suggests that there may be benefits for reflecting potential variability in the external region and considering more than one external region.
• The method of selecting the profile for each of the four stochastic variables independently and spreading them evenly across the various zones based on the ratio of their contribution to the base profile should also be examined more closely. There may be correlation between the load and solar and wind profiles – hot days may have increased demand but also generally higher solar outputs. Additionally, there may be regional differences in the various profiles, such as excess heat in the north or south, but not both, or cloud cover limiting solar output in one region or another. Such cases may challenge reliability within the CAISO in different manners than evenly allocating changes across the regions.
• SCE understands the reasons why more recent forced outage data may present problems, but relying on outage data from 12-18 years ago will likely not provide accurate outage estimates. Finding appropriate outages measurements that are not decades old is important, not only for the modeling but for any discussion of UCAP.

2. Please provide your organization’s feedback on the preliminary results.

SCE is unclear about the various scenarios that were used in the analysis. With the changes happening for the 2025 RA year in terms of how certain resources are counted it is not clear what level of resources are included in the “showings capped at obligation” scenario, or how the “historical pattern” in the second scenario was constructed. Given the tightness of the RA markets, it seems that for the last few years all eligible RA resources are included in the RA filings, at least in the summer months, which makes it hard to understand how there can be such large differences in the LOLE values for the different scenarios.

SCE believes that outage risks arising in non-summer months is something that deserves further discussion and analysis. These results confirm what many stakeholders have been expecting to see in the future as natural gas generation is retired and replaced by solar and storage, and more electrification and electric transportation happen. The fact that potential reliability issues are already starting to show up should raise red flags. Combined with the heat maps showing potential reliability issues not just at the peak hour but as late as HE 23 (Scenario 3a July) and in HE8 in January (Scenario 1) strongly suggests that the CAISO should consider an RA mechanism similar to the CPUC’s Slice of Day that doesn’t just focus on the one peak hour per month.

3. Please provide your organization’s input on what types of capacity accreditation methods and PRM approaches should be studied.

There have been large changes in the “technology factors” used to determine the NQCs of solar and wind resources due to the change in the CPUC entities using of Slice of Day and the change from ELCCs to the exceedance profile value during the peak hour. This demonstrates that both the capacity accreditation and the associated PRM need to recognize how the RA program works. As the CAISO works through the modeling and attempts to determine the appropriate default PRM and counting rules it must clearly state what it is assuming about how the RA program is structured.

To help stakeholders understand these issues, SCE urges the CAISO to provide information about how the 2025 RA showings provided by LSEs have changed from last year. The CPUC LSEs are required to show CAISO all resources that they show to the CPUC for RA. It will be important to understand both how the RA showings provided by CPUC LSEs to the CPUC and CAISO differ. It will also be important to understand how the resources provided for RA have changed from the previous year. This should be done for both the CPUC and non-CPUC entities. If non-CPUC entities use the counting rules developed for the CPUC Slice of Day program but without having to demonstrate Slice of Day RA compliance the amount of RA resources may change from the previous year.

4. Please provide any additional feedback not already captured.

SCE requests that CAISO provide a more detailed explanation of why it has decided to use SERVM for mid-term and long-term modeling but will continue to use PLEXES for short-term modeling. SCE looks forward to the results of the CAISO’s benchmarking of the different models for the 2025 year.