PJM is facing significant change over the next two decades as large load growth, generation additions and retirements, transmission development, fuel prices, and policy reshape the market.
During Yes Energy’s recent PJM Power Market Outlook: Where Prices Are Headed and Why webinar, we explored how those changes could affect long-term pricing, congestion, and emissions across the region. The outlook shows significant change ahead for PJM: annual load increases 72% by 2046, while average real-time LMP rises 82% to $99/MWh in real 2024 dollars. Aggregate real-time congestion costs are also projected to rise 92% to $6.4 billion.
The discussion prompted a number of questions about what is driving those results and how the forecast represents an evolving PJM system. Below, we answer some of the most common questions from the webinar.
1. How does the forecast account for future transmission upgrades, including planned 765 kV lines?
The forecast uses Multiregional Modeling Working Group (MMWG) summer planning power flow cases that reflect how PJM's transmission topology is expected to evolve over time, including future-year cases for 2026, 2029, and 2034. Approved transmission buildouts reflected in those planning cases are incorporated into the forecast, while speculative projects are not.
The capacity expansion model itself does not currently project new transmission buildout.
2. If transmission upgrades are included, why does congestion still increase on major PJM interfaces as load grows?
Transmission expansion helps relieve congestion in the near term. In the base-case forecast, total PJM congestion costs decline somewhat between 2027 and 2030 as transmission upgrades and new resources come online. Beyond 2030, however, load growth accelerates significantly, particularly in Dominion, and the transmission system has a harder time keeping pace.
AP South is a key example. Dominion load grows substantially over the forecast horizon, while transfer capability across AP South does not increase at the same rate. As a result, the interface binds more frequently and becomes the largest contributor to cumulative congestion costs in the PJM forecast, at approximately $38 billion across the forecast horizon.
3. How do you forecast nodal congestion over a 20-year horizon?
The forecast uses a fully nodal production cost model with security-constrained unit commitment (SCUC) and security-constrained economic dispatch (SCED) designed to replicate how the physical power system operates. The model incorporates transmission topology and interface limits alongside changing load, generation, fuel, policy, and other assumptions.
This allows the forecast to identify binding transmission constraints and quantify their impact on individual nodes. Forecast outputs include congestion as well as constraint decomposition, providing visibility into which transmission constraints are contributing to the congestion component of an LMP.
4. Why do prices and congestion differ so significantly across areas such as ComEd and Dominion?
The transmission system limits how freely power can move across PJM. In the webinar outlook, congestion drives 95% of the price spread between ComEd and Dominion, with AP South and Bedington–Black Oak responsible for much of that difference.
As concentrated data center load grows in areas such as Dominion, that growth places greater pressure on the interfaces needed to move power toward that demand. As those constraints bind more frequently, pricing outcomes increasingly diverge across PJM.
5. How is data center load modeled? Does the forecast rely on interconnection queues, signed contracts, project announcements, or a top-down forecast?
Data center load is one of the most consequential inputs in any long-term power market forecast. Our approach starts with PJM's own official load forecast, which already reflects a mix of interconnection queue activity, signed contracts, and project announcements across the footprint. From there, we don't assume every megawatt of announced or projected demand shows up exactly on schedule. In our base-case outlook, we phase in a portion of that projected load over time — full realization in the near term, tapering gradually in later years — to reflect the reality that some announced projects get delayed, scaled back, or never built.
In a more aggressive growth scenario, we assume the full amount of projected data center load is realized. Either way, the impact is significant: data center demand grows roughly sixfold over our 20-year outlook and is heavily concentrated in a handful of zones, which is exactly why it plays such an outsized role in the congestion and pricing trends we see there.
The importance of the assumption is clear in the PJM outlook. Data center load increases from approximately 86 TWh in 2027 to 512 TWh by 2046 and is the predominant driver of the 72% increase in total PJM annual load. Because that growth is geographically concentrated, it also has a significant impact on future congestion and nodal pricing.
6. How do you model future wind development, and do you account for changes in wind resource availability?
The forecast considers technical resource potential alongside the economic and policy factors that influence how much wind capacity is likely to be built — offshore wind lease areas, technical potential for onshore wind, changing capacity value, and state renewable policies and offshore wind targets.
To be clear on one distinction: we don't build in a long-term climate-driven trend in wind resource quality — for example, an assumption that wind speeds get systematically stronger or weaker over the next 20 years. Instead, we use a consistent, historical weather year as the basis for how wind and solar generate power hour by hour, which keeps that part of the forecast grounded in observed conditions rather than a speculative climate trend. What does change year to year and scenario to scenario is how much new wind capacity gets built — driven by technical potential, policy targets, and economics, not by an assumption that the wind itself is changing.
7. What natural gas price assumptions are used in the long-term forecast?
The forecast combines multiple external sources with Yes Energy's own modeling to develop long-term natural gas assumptions. Gas prices are modeled at the hub level, including expected changes in basis between Henry Hub and regional gas hubs.
The model also captures daily gas price variation rather than relying solely on monthly or annual averages, helping account for seasonal and daily volatility across the grid. Natural gas assumptions vary across the three forecast scenarios, allowing users to evaluate how different long-term fuel price environments may affect market outcomes.
8. How are battery storage resources modeled, and can GridSite be used to evaluate project economics?
Battery resources are represented with characteristics including storage duration and round-trip efficiency. Storage is optimized in the day-ahead market based on nodal price signals, which determine modeled charging and discharging behavior. Those decisions are then carried into the real-time simulation.
GridSite provides forecast outputs including plant generation, costs, revenues, and margins that can provide additional context for evaluating potential project economics.
9. How much storage capacity is assumed in the PJM forecast?
Storage plays a growing role in our PJM outlook. Battery storage capacity grows more than tenfold over the forecast horizon, expanding from roughly 1.5-2 GW today to somewhere between 18-20 GW by 2046, depending on the scenario. Pumped storage hydro, PJM's other major storage resource, stays essentially flat over the same period — the existing fleet remains in place, but we don't assume significant new pumped storage gets built. A meaningful share of the battery growth is tied to state clean energy policies that specifically call for storage buildout, including targets in Michigan, New Jersey, Virginia, and parts of Appalachian Power's territory.
10. What assumptions are used for generator offers in the SCUC and SCED?
The forecast uses calculated unit-level marginal costs rather than participant bid curves. These marginal costs reflect the fuel, emissions, and operating costs associated with each unit.
11. How are day-ahead and real-time markets represented in the forecast?
The model runs both day-ahead and real-time cycles. Unit commitment decisions are made in the day-ahead cycle and carried into the real-time model. The real-time model then responds to forced outages and potential changes in other inputs.
Additional cycles, such as reliability unit commitment, may also be run as needed, although those intermediate results are not reported.
12. Does the long-term forecast get adjusted daily to match actual real-time market results? How are generator outages represented?
No. GridSite's long-term forecast is designed to provide a fundamental view of how the system may evolve over time, rather than to predict near-term prices. The model does not make post-processing adjustments to force forecast prices to match current real-time market values.
Planned maintenance outages are scheduled across the modeling horizon based on typical unit outage patterns. Forced outages are also represented and can appear in the real-time cycle, requiring the modeled system to respond accordingly.
Forecast assumptions and key market inputs are reviewed quarterly as market conditions evolve.
13. How are reserves and scarcity pricing represented in the forecast?
Reserves and scarcity conditions are represented through the physical operating requirements of the system rather than a separate administrative pricing formula. The model requires the system to hold enough regulation, spinning, non-spinning, and supplemental reserves to maintain reliability, and it specifies which resources can provide each type — for example, battery storage can supply the full range of reserve products, while nuclear, wind, and solar generally don't participate. When the system is tight, prices reflect that tightness because the model is solving for the cost of meeting reserve requirements alongside energy demand, rather than applying a fixed penalty price.
14. What power flow methodology is used to produce the forecast?
The forecast combines nodal capacity expansion planning with nodal production cost modeling. For PJM, MMWG planning power flow cases are used to represent the evolving transmission topology, while the production cost model performs security-constrained unit commitment and economic dispatch with nodal fidelity.
This fundamental approach simulates how generation and transmission interact rather than extrapolating future nodal prices from historical price relationships.
15. How does the forecast account for uncertainty such as offshore wind delays, fuel price changes, or interconnection queue backlogs?
The forecast provides three scenarios rather than treating the future as a single deterministic outcome. Scenarios vary assumptions around factors including the timing and delivery of load and new resources, policy, fuel prices, capital costs, and other market drivers.
Forecast assumptions are market-specific and reviewed quarterly as conditions change.
16. Are the scenarios simply "high gas" and "low gas" cases?
No. The scenarios represent broader, internally consistent views of how the market could evolve rather than changing a single variable.
The PJM forecast includes an Equilibrium scenario, which serves as the base case; a Low Regulation scenario representing a more thermal-dominant future; and an Infrastructure Expansion scenario representing a cleaner-energy trajectory.
Differences between the scenarios reflect assumptions across policy, load growth, timing and delivery of resources, generation buildout, fuel prices, capital costs, and other market drivers.
The larger takeaway from the forecast is that PJM's long-term outlook cannot be reduced to a single driver.
Load growth, particularly from data centers, puts significant pressure on the system in the base-case outlook. But where that load appears, what generation is built to serve it, which transmission upgrades come online, how fuel and policy costs evolve, and how quickly the system can respond all influence the ultimate market outcome.
The webinar's congestion analysis illustrates why those interactions matter. AP South emerges as PJM's largest cumulative congestion driver in the forecast as concentrated Dominion load growth increases pressure on a transfer interface that does not expand at the same pace. At an individual node, that same system-level trend can translate into materially different congestion and project economics depending on which side of the constraint the node sits.
For developers, utilities, and large energy buyers making decisions that need to hold up for years, a future price alone is not enough. Understanding the assumptions, physical constraints, and market mechanisms behind that price can help teams evaluate risk, compare possible outcomes, and defend the decisions they make.
Want to explore the PJM outlook in more detail? Watch the PJM Power Market Outlook: Where Prices Are Headed and Why webinar to see how load growth, congestion, emissions, and other market drivers are expected to evolve across the region.
Have questions about how changing PJM market conditions could affect a project, procurement decision, or portfolio? Talk to a Yes Energy expert.