Comments on Nov 19 Modeling workshop

1. Please provide your organization’s feedback on the changes being considered to the inputs and assumptions.

The Alliance for Retail Energy Markets (“AReM”) would like to clarify its past recommendation regarding the use of resource adequacy (“RA”) compliance filings as an input to the modeling. Slide 12 of the November 19 workshop slide deck indicates AReM supports a new modeling run based on the capacity shown in year-ahead RA compliance filings. This misinterprets AReM’s recommendation. AReM instead was seeking to have CAISO benchmark the assumptions it used to fill in the 28% of load that provided no survey response against the RA compliance filings. If the assumptions CAISO used were inaccurate, the recommendation was not to perform an additional modeling run, but simply focus on the “All RA eligible” scenario for short-term modeling because that scenario does not rely on the load-serving entity (“LSE”) survey. 

The slides also state that CAISO is discussing with the California Public Utilities Commission (“CPUC”) and California Energy Commission the potential for a shared dataset. AReM also supports this, especially if this eliminates or reduces the need for an LSE survey.

In terms of changing inputs and assumptions, as discussed in previous comments, AReM supports alignment between CAISO and CPUC modeling efforts. We encourage CAISO and the CPUC to work together productively on a common set of inputs and assumptions wherever possible. AReM also supports the PLEXOS versus SERVM benchmarking efforts discussed at the workshop as part of the effort to align reliability modeling. AReM recommends these efforts be documented in a report to help stakeholders understand the implications of the benchmarking and what may be done to improve modeling alignment between agencies.

2. Please provide your organization’s input on preliminary mid-term and long-term results.

One notable finding was that CAISO’s mid-term results show a larger 2026 capacity surplus (refer to slide 57) than the 1,500 MW surplus presented by the CPUC’s RA team in Rulemaking 23-10-011. There were some reasons briefly discussed at the workshop to explain this difference related to load modeling. AReM seeks greater clarity on the differences and supports efforts to benchmark the two modeling efforts against each other to improve alignment. Simply publishing two different high-level findings would be confusing and should be avoided.

AReM would also like to understand how long-term results from CAISO’s modeling vary from CPUC’s staff’s expected reliability modeling of the transmission planning process (“TPP”) portfolio slated for adoption in the CPUC’s Integrated Resource Planning (“IRP”) rulemaking (R.20-05-003).

3. Please provide your organization’s feedback on the capacity accreditation methods and PRM approaches presented today.

The discussion regarding capacity accreditation and planning reserve margin (“PRM”), though informative, was also somewhat alarming. Part of the discussion involved an ideological dispute with Zachary Ming from E3 advocating for marginal effective load carrying capability (“ELCC”) methods whereas other stakeholders, especially Doug Karpa from Peninsula Clean Energy, speaking in favor of slice-of-day (“SOD”) rules adopted by the CPUC for its RA program. As discussed in the workshop, different methods have pros and cons. AReM was under the impression that the development of the SOD framework, which was a multi-year process at the CPUC in which CAISO was a stakeholder, was the appropriate forum for an ideological dispute in which experts could opine on what solution was best. Now that the CPUC made its decision and explicitly rejected ELCC methods in implementing SOD, AReM is concerned that any continuing ideological discussion of resource counting rules will only serve to undermine SOD, increase uncertainty in the market, and harm AReM members’ and other LSEs’ efforts to procure resources to meet state policy goals.

AReM has no firm position on the ideological dispute between ELCC and SOD methods, as it recognizes the pros and cons of different approaches. What AReM seeks is regulatory certainty to facilitate transactions in the market. To that end, if CAISO is even considering never adopting SOD methods for its RA program in harmony with the CPUC, at least for CPUC-jurisdictional LSEs, it should state so now and provide its rationale for doing so after so much time has already been expended on the SOD framework. Although AReM is confounded by the prospect of abandoning SOD after so much work has been expended on it and after LSEs are already procuring to meet these standards, if the CPUC and CAISO adopted separate RA frameworks with different resource counting rules, that would be an even worse outcome, as it would require LSEs to do extra work to seek a resource portfolio that met different and potentially conflicting standards for no logical reason. This would only serve to increase costs, and hence consumer rates, which is completely unreasonable when the conflicting standards are only adopted to appeal to different market experts with different ideological opinions.

AReM also rejects the view that somehow marginal ELCC and SOD approaches are different but compatible. First, as was discussed at the workshop, marginal ELCCs can vary based on load shape. Under the SOD framework, each LSE is responsible for meeting its own load shape. This is a critical issue for electric service providers (“ESPs”) that serve commercial load that tends to peak midday, which is not the case for the entire system. No one is contemplating developing separate ELCCs for each LSE’s load shape. So that puts the two approaches in conflict for ESPs, including AReM members. Second, under SOD rules, the energy charging requirement for batteries is explicit, whereas for ELCC methods it is implicit to the resource mix assumed in developing the ELCCs. So just procuring to ELCC standards may not result in a portfolio with adequate charging capacity under SOD rules.

Ultimately, AReM simply seeks intellectual honesty from the CPUC and CAISO on the issue of resource counting rules. If there are ideological or technical concerns about SOD that would cause the CPUC and CAISO not to use the SOD framework for all reliability resource counting impacting CPUC-jurisdictional LSEs, these concerns should be stated so that stakeholders can respond appropriately. If that is not the case, then AReM recommends CAISO clarify its intent when considering marginal ELCC methods. If the intent is only to allow them for other local regulatory authorities, then AReM will not offer any further comment as CAISO’s decision making on this issue will not impact ESPs.

4. Please provide any feedback not already captured.

AReM has no additional feedback at this time.

1. Please provide your organization’s feedback on the changes being considered to the inputs and assumptions.

The California Community Choice Association (CalCCA) appreciates the opportunity to comment on the November 19, 2024, modeling workshop. The California Independent System Operator (CAISO) proposes to make improvements to its inputs and assumptions for modeling hydro and forced outages. First, improving the methodology for modeling hydro by randomly drawing from 25 years of historical hydro years is worthwhile, as the CAISO demonstrated through sensitivity simulations that hydro assumptions have significant impacts on loss-of-load expectation (LOLE) events. Second, the CAISO indicates it will continue to update its forced outage rates and scrub its historical Outage Management System data. CalCCA supports the CAISO undertaking this effort, as it will be necessary to model forced outage rates accurately and use them for an eventual UCAP counting methodology. The CAISO’s forced outage rates are lower than the California Public Utilities Commission’s (CPUC’s) and GADS’s. For example, the CAISO forced outage rate for combustion turbines in 4.5 percent compared to 6.2 percent and 12 percent for the CPUC and GADS, respectively. These differences could have significant impacts on the modeling results, particularly for resources that are likely needed during reliability events.

CalCCA appreciates the CAISO’s analysis of the correlation between load, solar, and wind in the 500 samples used in the reliability simulations. Comparison between the range of correlation coefficients reported by the CAISO and the correlation coefficients from the CPUC’s 23 weather years[1] shows some discrepancies, as demonstrated in Figure 1. The most important difference is that the CPUC weather data shows multiple weather years with load-solar correlation coefficients that are below the range of 500 sampled years in the CAISO dataset. A lower correlation coefficient suggests a lower contribution of solar to reliability. The CAISO and the CPUC should evaluate this difference and determine if it is due to differences in the solar profiles or differences in the load profiles. The CAISO and the CPUC should also decide if their datasets should be adjusted to generate a wider range of load and solar correlations across samples.

Another difference is the solar-wind correlation coefficients: the small range in the CAISO dataset is more negatively correlated than in the CPUC dataset and varies much more narrowly than the solar-wind correlation in historical observations.[2]  CAISO, in coordination with the CPUC, should evaluate whether adjustments to the sampling process are warranted based on these differences. These differences suggest that the CAISO’s data sets, the CPUC’s data sets, or both could require an update to ensure accuracy. The CAISO and the CPUC should coordinate to investigate these differences and align on data sources based upon their findings. 

Figure 1. Discrepancies between Load, Solar, and Wind correlation coefficients between the CPUC’s 23 weather years (dots), historical metered solar and wind (diamonds), and the range of correlation coefficients across all CAISO modeled samples (shaded area).

[1]              Downloaded from https://www.cpuc.ca.gov/industries-and-topics/electrical-energy/electric-power-procurement/long-term-procurement-planning/2024-26-irp-cycle-events-and-materials/system-reliability-modeling-datasets-2024.  Load data is the CAISO Baseline load.  Solar and Wind are aggregate of CAISO generators for the planning year of 2026.

[2]              Historical metered data is from the CAISO production and curtailment data (https://www.caiso.com/library/production-curtailments-data).  Curtailment data was used to estimate the pre-curtailment wind and solar correlations.  We do not compare the metered wind and solar data to the metered CAISO load because the metered load embeds the behind-the-meter solar generation, producing different correlation coefficients than would be calculated with the consumption profiles.

2. Please provide your organization’s input on preliminary mid-term and long-term results.

CalCCA appreciates the efforts of the CAISO and Astrapé to model mid-term and long-term reliability. The results demonstrate surplus capacity in 2026 through 2034. After removing surplus capacity to surface 1-in-10 LOLE, reliability events are concentrated in hours HE 18/19 and then again in HE 22. These separate reliability events appear to indicate a capacity need in the early evening hours when there is insufficient capacity to available to serve load and an energy need in the later evening hours when storage resources reach their energy limitations. One implication of these results is that loss-of-load events associated with depleted storage in HE 22 may be mitigated by actions that delay the discharge of storage, such as additional generation or reduction in demand in the hours immediately preceding HE 22.

The results suggest that “critical hours” should not be assumed as exclusively loss-of-load hours or exclusively gross peak hours, as described by E3. There are multiple ways critical hours can be accounted for within an RA program to ensure there is enough capacity and energy to meet reliability targets. The CPUC’s slice-of-day program, for example, uses hourly capacity requirements and a storage charging sufficiency requirement. As the CAISO assesses near-term, mid-term, and long-term reliability needs in coordination with LRAs, it will become increasingly important to consider “critical hours” beyond peak hours to ensure reliability under a highly renewable and energy storage system.

3. Please provide your organization’s feedback on the capacity accreditation methods and PRM approaches presented today.

CalCCA appreciates the presentations from NP Energy, Astrapé Consulting, E3, and the CPUC Energy Division. The presentations provided a variety of different approaches for valuing RA capacity. As stated in CalCCA’s December 5, 2024, comments to the Issue Paper, the CAISO should provide opportunities for all LRAs to adopt the same resource counting methodologies and accompanying PRMs and availability incentives. The CAISO, in coordination with LRAs, should seek to count resources in a manner that puts all technology types on a level playing field by accurately reflecting their capabilities in their NQC values in both the year-ahead and month-ahead timeframe. They should also demonstrate that proposed changes to NQCs are accompanied by revisions to the PRM.  

4. Please provide any feedback not already captured.

CalCCA has no additional feedback at this time.  

1. Please provide your organization’s feedback on the changes being considered to the inputs and assumptions.

CAISO presentation indicates sensitivity of hydro generation modeling inputs (low, average, high hydro years) impacting Loss of Load Expectation (LOLE) studies. CAISO plans to use average hydro year for modeling inputs. With regard to modeling hydro on an average hydro year as well as annual or seasonal only RA showings, CDWR has following comment:

Forecast of pumping load and associated hydro generation that is  part of the California Department of Water Resources (CDWR) water delivery system operation is primarily dependent on the forecast of hydrology, water demand, and operational constraints. The California water year hydrologic forecast[1] starts in December and continues monthly with an official declaration of water year hydrologic forecast on May 1^st of each water year (e.g. the 2024 water year is 2023 October 1 through 2024 September 30) for the Sacramento Valley and San Joaquin River Valley water year hydrologic classification. SWP operations forecasting mimics the hydrologic forecasting process with monthly updates. Forecast of CDWR pumping load and hydro resources done after May 1^st tend to be closer to the actual operation compared to the prior month’s forecasts with a gradual improvement in the forecast each month starting December. However, CDWR must file annual RA plans on October of the prior year that includes demand and hydro supply resources forecasts for the net qualifying capacity(NQC) for the upcoming RA compliance year. Due to uncertainty in hydrology forecast from October prior year and the May 1^st next year (RA compliance year) and the following months, there could be significant differences of forecasts between the annual and monthly plan timeframe undermining true reflection of contribution to real time grid reliability both in terms or demand and the supply. As a solution, entities like CDWR with pumping load and hydro generation, monthly update of loads and resources must be allowed as it is done today. Doing so will enable updating demand as well as supply to reflect in the monthly plans and associated RA must offer requirements. Using monthly showings will better reflect the hydrology forecast, water pumping demand, and hydro generation in short-term RA planning, which will enhance the objective of RA.

It is to be noted that:

a) average hydro year (example:2018 used by CAISO) refers to the Below Normal (BN) water year hydrologic classification if the water year classification is based on the California hydrologic forecast.

b) a water year covers the period October 1 through September 30; the water year 2024 starts on October 1, 2023, and ends on September 30, 2024. In contrast, RA compliance year starts on January 1 and ends on December 31^st each year. It should be clarified whether hydro capacity monthly NQC values in the modeling correspond to months of the water year type or the RA compliance year months.

c) water year hydrologic classification is done by two regions: Sacramento valley and San Joaquin valley and classification could be different for the two regions for the same year; resources belonging to each area may be associated with a different hydrologic forecast.

d) The California water year hydrologic forecast and water year type classification may impact the capability estimates of any hydro generation (within California) facility.  This implies that hydro generation capability updates within a year could be beneficial and applicable to all hydro generating resources for a more realistic availability.

Following are the comments / questions with regard to Slide 14:

a) (Hydro modeling and Weather Year alignment in the stochastic profile)-. If an average hydro year is assumed for short term LOLE studies, how does the weather year alignment take place in terms of timing (e.g., CA water year hydrologic forecast update schedule)? For example, with the official classification of water year type forecast on May 1^st, will the stochastic modeling be updated for alignment with the forecasted water year hydrology for that year?

b) Loss of Load and reserve sharing (also in slide 26): Can’t non-spin reserve help avoid Loss of Load as the award to non-spin results in the real time energy (dispatch to generate or drop load)? If it does, then why not include such reserves and resources providing such products in the model in the study (and count them as RA resources)?

[1] Department of Water Resources / California Cooperative Snow Surveys / Sacramento Valley & San Joaquin Valley Water Year Hydrologic Classification: Year types are set by first of month forecasts beginning in December and monthly thereafter.  Final determination is based on the May 1^st forecast.

https://cdec.water.ca.gov/reportapp/javareports?name=WSIHIST

https://cdec.water.ca.gov/reportapp/javareports?name=WSI  (for 2024)

2. Please provide your organization’s input on preliminary mid-term and long-term results.

No further comment.

3. Please provide your organization’s feedback on the capacity accreditation methods and PRM approaches presented today.

For the best option on accreditation among the three (LOLE hours focused, risk hours focused, ELCC based), further analysis may determine which option would be the best for reliability.

PRM (monthly, seasonal, annual): further data analysis on these options may provide some clue.

4. Please provide any feedback not already captured.

No further comments.

1. Please provide your organization’s feedback on the changes being considered to the inputs and assumptions.

As discussed in its prior comments, CESA requests CAISO formalize all of its specific input parameters and assumptions into a comprehensive document.

2. Please provide your organization’s input on preliminary mid-term and long-term results.

It was clear from the working group discussions on 11/19 that there is not currently and will not be an energy sufficiency issue on the CAISO system.  In describing the comprehensive loss of load expectation analyses, covering from 2026 through 2034, presenters stated that the cause of the expected unserved energy was due to a lack of capacity, not a lack of energy to supply energy storage resources (i.e. more resource and storage capacity needed in the loss of load hours, not more excess energy needed earlier than the loss of load hours). CESA recommends that CAISO verifies and includes this conclusion in its final report on the mid- and long-term loss of load expectation study.

3. Please provide your organization’s feedback on the capacity accreditation methods and PRM approaches presented today.

No comments at this time.

4. Please provide any feedback not already captured.

No comments at this time.

1. Please provide your organization’s feedback on the changes being considered to the inputs and assumptions.

Microsoft appreciates the efforts made by the California Independent System Operator (CAISO) to provide an overview of its modeling inputs and assumptions. We have three main suggestions:

• First, update all CAISO Loss of Load Expectation (LOLE) models based on the 2024 Integrated Energy Policy Report (IEPR). This would enable better anticipation of the growing demands on the CAISO grid. As demand on the grid continues to rise, leveraging updated forecasts will be fundamental in aligning long-term resource planning and transmission planning needs with cost management objectives.
• Second, continue updating how storage is treated in the LOLE models. Given the grid transformation from gas to renewables and storage, we recognize the importance of advanced storage modeling techniques. Updating the models themselves to versions that can better optimize storage, as suggested during the November 19 meeting, as well as incorporating cycle limits through the utilization of the CPUC Master Resource Database (MRD) would lead to more precise energy storage modeling. Additionally, we echo the recommendation made during the meeting to adjust the optimal dispatch of storage within PLEXOS—either by modeling potential outages or employing other mechanisms—would better reflect the real-world conditions of energy dispatch. This refinement is crucial for maintaining model accuracy, as it allows for accurate resource allocation by acknowledging the imperfect nature of storage dispatch, particularly during peak demand periods or conditions nearing scarcity.
• Third, align outage inputs more closely with actual operational data by leveraging current Operational Management System (OMS) statistics. This methodology ensures that the data used for reliability planning is not only current but also reflective of ongoing conditions. 

2. Please provide your organization’s input on preliminary mid-term and long-term results.

Microsoft supports the on-going collaboration between the CAISO and Astrapé Consulting to conduct comprehensive mid- and long-term modeling. Such modeling is pivotal in establishing a robust framework to ensure grid reliability by giving the CAISO key insights into its Balancing Areas expected reliability standard. Without this LOLE modeling, the CAISO will have no visibility into whether in aggregate the CAISO is meeting a 1-in-10 LOLE standard.   

The application of 1-in-10 LOLE modeling is instrumental, as it provides a benchmark for understanding and mitigating rare, but impactful, supply disruptions. The 1-in-10 standard ensures that the system is designed to handle reliability challenges, thus preparing the grid for exceptional demand spikes or unforeseen contingencies. Microsoft strongly supports integrating the reliability standard in the CAISO retirement processes and believes a 1-in-10 LOLE or equivalent standard is the right balance between reliability while supporting cost-effective strategies to manage and sustain a resilient energy ecosystem.

Mid- and long-term modeling are essential, as they guide strategic decisions regarding resource retirements and helps the CAISO and its sister agencies understand how well ongoing procurement of new resources aligns with projected grid needs. By accurately predicting future energy demands and resource availabilities, these models help prevent the premature retirement of resources that are critical to maintaining grid reliability. This foresight allows for efficient allocation and phasing out of gas assets, ensuring that the energy infrastructure can adapt seamlessly to both current demands and future developments without incurring unnecessary scarcity costs or compromising reliability.

3. Please provide your organization’s feedback on the capacity accreditation methods and PRM approaches presented today.

Microsoft supports the utilization of Unforced Capacity (UCAP) in the California Public Utilities Commission (CPUC) processes as a robust methodology for accurately accounting for resource reliability. By focusing on UCAP, resource accounting can better reflect the true availability and performance of gas and storage resources within the slice-of-day approach, enhancing the precision of capacity planning and ensuring a stable energy supply. 

Microsoft supports the CAISO advancing this initiative alongside the CPUC; however, Microsoft also supports consideration of the Perfect Capacity (PCAP) approach as described by Astrapé within the CAISO default RA program. Given the significant difference between the slice-of-day CPUC paradigm and the peak and/or net-peak CAISO paradigm, Microsoft does not see a benefit from aligning counting rules between the agencies and believes the PCAP approach is worth considering based on the presentations. 

4. Please provide any feedback not already captured.

Microsoft appreciates the CAISO’s modeling efforts and looks forward to participating in future working groups. 

1. Please provide your organization’s feedback on the changes being considered to the inputs and assumptions.

SDG&E generally supports the changes being proposed and recommends that CAISO monitor wind and solar NQCs. These values changed significantly between 2024 and 2025 due to new counting rules. Given these recent changes, CAISO should monitor future wind and solar performance as compared to the historical data these profiles are based on to ensure that they reflect actual performance.

2. Please provide your organization’s input on preliminary mid-term and long-term results.

Results show the highest loss of load probability hours continue to occur in summer months, especially in September, and in HE22. This means that LSEs may need to consider adjustments to time of use peak hours. These results also raise the question of whether or not increased battery capacity, especially long duration storage, will mitigate loss of load probability risk. If CAISO’s modeling appropriately captures most of the expected new batteries on the grid, LSEs and CAISO will need to closely monitor these loss of load probability hours.

3. Please provide your organization’s feedback on the capacity accreditation methods and PRM approaches presented today.

Natural gas and other dispatchable resources should be valued using the UCAP methodology to account for thermal derates and outage patterns under both Slice-of-Day (SOD) and CAISO monthly peak RA counting methods. This is especially important as the state’s natural gas fleet continues to age and may continue to be affected by derates and outages.

Given that SOD valuation looks at all 24-hours, exceedance is the appropriate way to value intermittent renewable resources (e.g., wind and solar). For non-SOD reliability, such as CAISO’s monthly peak load RA consideration, intermittent renewable resources can be valued through ELCC-based NQC or exceedance. SDG&E recommends exceedance to maintain consistency with the CPUC’s SOD approach. If using an ELCC-based NQC, average ELCC should be used instead of marginal since CAISO is looking to value the entire portfolio of resources, and the ELCC values should come from a LOLE study. If using exceedance, these resources should be valued according to the associated peak load hour. For monthly peak load and peak load hour consideration, including for the PRM calculation, it is appropriate for CAISO to consider only the gross load peak if including intermittent renewables. If using a net load peak, the contribution of these resources is already removed from the load profile, so considering any value for wind or solar (exceedance or ELCC-based NQC) would double count the contribution of these resources to grid reliability.

Accrediting all resource types at their respective ELCC values would be more difficult in practice, given that the CPUC is using SOD. SDG&E believes that natural gas and other dispatchable resources should be valued using UCAP and that intermittent renewable resources should be valued using exceedance. This would help minimize differences between CPUC and CAISO methodologies and would simplify the PRM calculation as resource performance would already be baked into the capacity valuation. Notably, the CPUC recently approved its RA Track 2 Proposed Decision which, among other things, encourages Energy Division to coordinate with CAISO on UCAP development.

4. Please provide any feedback not already captured.

 No additional comments.

1. Please provide your organization’s feedback on the changes being considered to the inputs and assumptions.

Six Cities’ Response:  The Six Cities request further explanation with respect to the proposed modeling treatment of (a) different categories of capacity reserves (e.g., Strategic Reliability Reserves, other types of capacity reserves subject to dispatch restrictions); and (b) anticipated Demand Response Resources.  Specifically, the Six Cities request that the CAISO provide a description of whether reserved capacity and Demand Response resources are or are not reflected in the modeling process, if they are reflected, how they are quantified or estimated, and, for each of those types of potential resources, why the proposed treatment is reasonable from a policy perspective. 

2. Please provide your organization’s input on preliminary mid-term and long-term results.

Six Cities’ Response:  The Six Cities have no comments at this time on the preliminary mid-term and long-term modeling results, except to note the inherent and unavoidable uncertainties embedded in the inputs and assumptions for those analyses.

3. Please provide your organization’s feedback on the capacity accreditation methods and PRM approaches presented today.

Six Cities’ Response:  The Six Cities support further consideration and evaluation of seasonal PRM approaches for both gross peak and net peak needs and accreditation standards and performance incentives aligned with critical risk hours as described generally in slides 108-120 of the presentation for the November 19, 2024 workshop. 

4. Please provide any feedback not already captured.

Six Cities’ Response:  The Six Cities reiterate their support for preparation and presentation of a backcast analysis of LOLE based on historical load and resource data for 2023 or 2024.  The Six Cities strongly agree with the views expressed by multiple stakeholders that such an analysis would provide a useful baseline for calibrating and evaluating forward looking analyses.  The resources reflected in the historical analysis should include all contracted RA resources, not just the resources included in monthly RA showings.

1. Please provide your organization’s feedback on the changes being considered to the inputs and assumptions.

• Using randomly drawn hydro years is an improvement over using one “average” year.
• SCE appreciates the extra information on outages, but more discussion is required. The presentation states that CAISO will update forced outage rates but doesn’t indicate what the updated rates will be or how the historical data is being scrubbed.
• The NQC information provided on slide 27 are interesting, but suggest additional questions or analysis that would be useful:
  • In addition to showing the NQC RA values for 2024 and 2025 can the CAISO provide the nameplate MWs of the resources in each year and resource type and the implied PRM from the shown resources?
  • Can the original information provided be split between CPUC and non-CPUC entities?

2. Please provide your organization’s input on preliminary mid-term and long-term results.

• Benchmarking between PLEXOS and SERVM
  • SCE continues to analyze the differences noted in hydro modeling, storage optimization, generator outages and imports.
• Heat Maps
  • SCE finds it interesting that LOLE hours include HE 22, especially in September, but often not HE21 or HE 20.
  • SCE is still analyzing what can be learned from the LOLE values as compared with the values for Average EUE, Max Shortfall and longest event. The LOLE are clustered around 0.1 but there is significant variation in the other three measures. Can CAISO provide additional information about what is driving these differences and if they are the results of the stochastic processes, what sort of combinations of the various stochastic variables lead to these extreme values?

3. Please provide your organization’s feedback on the capacity accreditation methods and PRM approaches presented today.

• SCE believes that the Slice of Day (SoD) methodology is preferable to ELCCs, especially monthly ELCCs, because the contributions of different resources are easier to see. In a grid like California’s where there are likely to be large changes in the amounts of solar and storage resources going forward it seems likely that the ELCCs of resources will change possibly dramatically over time as the amount of the resource and/or other resources and the mix of resources change. For example, if all LSEs attempt to meet their RA requirements by procuring solar resources they may find they are very short in subsequent years when the ELCCs change to reflect the new mix of resources.
• SCE appreciates the discussion that many of the Eastern RTOs are moving to a marginal ELCC, but the California grid is unique in the level of renewables and storage resources and the level of fully dispatchable resources. ELCCs will likely not change as fast in grids with less renewables and storage and more dispatchable resources.
• SCE also wishes to emphasize that the counting methods and the PRM must reflect the RA mechanism being used. CAISO and all market participants must avoid using a PRM and counting rules developed for SoD framework to gauge RA on a peak hour basis.

4. Please provide any feedback not already captured.

1. Please provide your organization’s feedback on the changes being considered to the inputs and assumptions.

The Western Power Trading Forum (WPTF) appreciates the CAISO providing a comprehensive overview of the inputs and assumptions. As noted in our comments during the workshop, we support improved storage modeling that includes modeling cycle limits using the CPUC Master Resource Database (MRD) information. We also support discounting storage in PLEXOS either through outages or another mechanism to reflect the imperfect dispatch of storage. The CAISO has mechanisms in place that intentionally hold or strongly incentivize storage to meet its DA schedule rather than RT grid conditions. This prevents optimal dispatch of the storage during scarce and near-scarce conditions.

WPTF also supports the CAISO updating the outage inputs to be more aligned with reality by using current OMS data and removing the strategic reserve from the observations. Additionally, this data should be further refined as the program evolves to only include forced outages that are likely to continue. For example, if the CAISO implements a policy that encourages replacement of planned outages rather than the conversion to forced outages, then the CAISO should consider removing outages that were converted from planned to forced in the data. 

Finally, WPTF strongly supports updating all results using the 2024 IEPR to better capture increasingly high CAISO load.

2. Please provide your organization’s input on preliminary mid-term and long-term results.

WPTF supports the CAISO using Astrapé Consulting to perform the mid- and long-term modeling and appreciated the detailed explanation of how the LOLE study can be used to create the default planning reserve margin and counting rules. We look forward to reviewing additional results.   

3. Please provide your organization’s feedback on the capacity accreditation methods and PRM approaches presented today.

WPTF supports the exploration of using PCAP for all resources as a capacity accreditation method. This methodology would lead to the most fungible RA MW and if used in any of the outage pool concepts, allow for the most reliable substitution of resources because all MW would provide roughly equivalent amounts of reliability to the CAISO grid. Although the CAISO has committed to assisting the CPUC with their UCAP counting methodology, there is no need to align natural gas and storage counting rules when the rest of the program are so substantively different.

4. Please provide any feedback not already captured.

It is WPTF’s understanding that the CAISO’s LOLE study captures hourly ramping needs that are currently within the Flexible RA requirement. We support the CAISO using the LOLE results to justify the complete removal of the flexible RA product in its entirety.