Resource Adequacy Modeling for a Reliable, Decarbonizing Grid
Synapse recently expanded our resource adequacy modeling capabilities by licensing the SERVM model. Please reach out to pknight@synapse-energy.com if you are interested in resource adequacy modeling support in these or other areas.
Growing electric loads and changing resource mixes are raising questions about the future of grid reliability. While capacity expansion combined with production cost modeling is still the cornerstone of resource planning, layering on reliability modeling can provide valuable –and sometimes essential—information on system performance. Resource adequacy models (also referred to as reliability models) help us examine how well a resource portfolio can serve future load across the four seasons, during extreme weather events, and under various generator outage scenarios. We can use these tools to help answer questions like, “Will this set of resources will be able to reliably serve load during a lull in the wind, on a cloudy day?” or, “What are the expected reliability contributions of a battery energy storage system compared to a natural gas combustion turbine?”
The figure below illustrates how resource adequacy models interact with other electric sector planning models in a utility resource planning process.
Source: Modified from Synapse and Berkeley Lab, 2024, Best Practices in Integrated Resource Planning.
Credible resource adequacy modeling requires the development of many inputs:
- Weather-correlated load shapes that account for electrification and energy efficiency: Future loads will have varying degrees of correlations with weather. For example, heat pumps loads will be strongly correlated with temperatures, whereas electric vehicles and data centers will be less strongly correlated with weather. As building electrification increases, systems may shift from summer-peaking to winter-peaking, which changes the types of resources that can most contribute to a reliable grid. At the same time, increasing investment in energy efficiency can help mitigate increasing peaks and lower demand across all hours.
- Renewable energy generation profiles: Future renewable generation potential is modeled based on geographically specific solar and wind generation shapes. These inputs are generally based on historical resource performance and historical weather data.
- Performance data for the existing resource fleet: Data from the existing fleet, such as forced outage rates, ramp time, and black-start periods is used to establish the reliability contributions of conventional resources like coal, gas, and nuclear to a portfolio.
As we consider the reliability of a grid during this time of energy transition, it is important to remember that no resource is a perfect capacity resource. Every technology has its own limitations. For instance, system planners have always had to plan for the impacts of large single (random) contingency failures from coal or gas plants tripping offline. And while renewable energy resource availabilities do depend on the weather, seasonal and daily weather patterns can be predicted with high levels of accuracy. The key takeaway is that a well-balanced and well-planned portfolio is critical to overall system reliability. Pairing renewables with varying durations of battery storage can help smooth out the variation in renewable output and provide firming capabilities. Resource adequacy models can help system planners understand how all these factors come together to impact system performance.
Resource adequacy modeling is critical for a variety of uses, several of which we describe below.
Calculating resource capacity accreditation
Capacity accreditation refers to the quantity of firm capacity that each resource can be depended on to provide during system stress events. Firm capacity accreditation is critical for resource planning, regional transmission planning, capacity market bidding, and a variety of other planning and functions. These values impact the relative economics of each resource and influence the mix of resources that will come online. To determine these accreditation values, grid operators conduct reliability modeling to determine resource-specific contributions to system reliability. Many regional transmission operators are currently reforming their capacity accreditation modeling to reflect changing resource mixes.
“In the weeds” modeling decisions can influence the accreditation value that resources receive. Such decisions include charging/discharging assumptions for batteries, optimization periods for longer-duration storage, firm fuel supply availability for fossil plants, and temporal availability for demand response resources. Good reliability modeling can examine how sensitive capacity accreditation is to these values.
Setting reserve margins
Utility planners and regional grid operators also use resource adequacy modeling to determine the total quantity of firm capacity that they need to build - or will seek to procure in capacity market auctions. Getting this number right ensures customers get a reliable system without paying for unnecessary costs.
If grid operators over-procure capacity, ratepayers can end up paying for unnecessary capacity. For example, a resource adequacy modeling software error recently led MISO to over-procure capacity, leading to $280 million in overpayments in its 2025/2026 capacity auction. However, if grid operators under-procure capacity, such as by underestimating load growth, they run the risk of power interruptions and even blackouts.
Resource adequacy models help to examine a variety of possible load futures and weather conditions and they provide insight into the implications of under- or over-procurement.
Evaluating resource portfolios as part of an Integrated Resource Plan (IRP)
Many vertically integrated utilities are starting to link their resource planning with iterative resource adequacy assessments as a best practice. To do this, they typically run their resource plans (developed using capacity expansion and production cost modeling along with other IRP processes) through resource adequacy models to stress test their preferred portfolios. Utilities may also use resource adequacy models to develop capacity accreditation values, if they are not situated within a broader wholesale market.
Synapse’s recent report on Best Practices in IRPs discusses how this type of iterative capacity expansion and resource adequacy analysis can be used to produce robust, least-cost, reliable resource plans. The figure above (taken from this report) illustrates the interactions between different use cases of a resource adequacy model and other models in a typical IRP process.
Evaluating the need for a resource as part of a Certificate of Public Convenience and Necessity (CPCN) docket
Many vertically integrated utilities are required to seek approval from their public utility commissions to build resources. This approval is known as a CPCN and generally involves establishing need and demonstrating that a resource can meet system reliability requirements. In this way, reliability modeling can support a CPCN application. Close examination of utility reliability modeling assumptions is critical to gain an accurate understanding of a utility’s resource adequacy needs and the extent to which proposed resources—or alternative proposals—can address those needs.
For example, Virginia’s Clean Economy Act (VCEA) requires that utilities meet certain energy efficiency goals before they can build new carbon-emitting generation, with an exception allowing for the construction of new fossil resources if there are system reliability issues. Dominion’s recent CPCN application for the Chesterfield Energy Reliability Center included several reliability modeling analyses to support its claim that it should receive a variance from the VCEA energy efficiency requirement on the basis of reliability concerns. In this case, Synapse’s examination of important reliability modeling decisions revealed key flaws in Dominion’s modeling, and we identified an alternative portfolio—without new gas—that had even higher reliability performance.