PJM has released its new long-term load forecast, and it predicts estimated electricity demand growth of 1.7% per year for summer peaks, 2% for winter peaks, and 2.4% for net energy over a 10-year planning horizon starting in 2024.
The 2024 summer forecast peak demand, or load, is 151,254 MW, according to the 2024 PJM Load Forecast Report (PDF), with summer peak load increasing to 178,895 MW in 2034 and 193,123 in 2039, an increase of nearly 42,000 MW.
Peak winter load for the 2024 winter is forecast at 134,663 MW for the 2023–2024 winter, going to 164,824 MW in 2034 and 178,241 in 2039, an increase of more than 43,000 MW.
Total annual energy use throughout the PJM footprint is expected to increase nearly 40% by 2039, from 800,000 gigawatt-hours (GWh) to about 1.1 million GWh.
“This forecast reflects the accelerated growth that we discussed with our stakeholders throughout 2023, driven by the electrification of multiple sectors combined with consumer demands for technology,” said Kenneth S. Seiler, Sr. Vice President – Planning. “It also underscores the need to maintain and develop enough generation resources to serve that growing demand.” Through its Ensuring a Reliable Energy Transition initiative and related research, PJM has noted that increased electricity demand, combined with accelerated generator retirements and the slow pace of replacement generation, will challenge reliability in the PJM footprint by 2030 if not addressed.
|Summer Peak (MW)
|Change From 2024 (MW)
|Winter Peak (MW)
|Change from 2023-2024
Rising energy demand in the region PJM serves is increasingly driven by the development of data centers throughout the PJM footprint, combined with the accelerating electrification of transportation and industry.
The 2024 PJM Load Forecast Report was derived using an electric vehicle forecast by S&P Global with projected counts by zone and impacts for passenger and freight vehicles.
The forecast uses historical weather data from 1994 to 2022 as the basis for constructing forecasts. PJM added an extreme cold weather variable that allows the PJM forecast model to better calibrate at colder temperatures in response to Winter Storm Elliott and other extreme weather events.
The 2024 Load Forecast Supplement (PDF) provides an overview of the methods of assembling the forecast and the various factors that contribute to it.
“These are useful numbers for lawmakers to understand as they craft policies for this energy transition we are experiencing,” said Asim Z. Haque, Sr. Vice President – State & Member Services. “As we speed up the development of resources that are coming through our study process in large numbers, we will have to look hard at how to make sure our generation supply matches the increase in demand expressed here.”
PJM has engaged with states and cities to incorporate public policies related to electrification, with extensive discussions at the Long-Term Regional Transmission Planning stakeholder workshops. Some of the policies, such as the goals for electrification of home and building heating systems, can be found in this presentation (PDF) to the PJM Load Analysis Subcommittee in November.
The PJM load forecast is constructed using 24 hourly models for each transmission zone. In each model, load is the dependent variable, considered alongside weather, calendar events, economic data and end-use variables. In the history, PJM starts with metered load and then reconstitutes total load with load-management addbacks, load drops associated with peak-shaving programs, and distributed solar generation estimates.
For the 2024 Load Forecast, PJM contracted with S&P Global to provide an electric vehicle (EV) forecast for the number of light-, medium- and heavy-duty vehicles across our footprint.
The PJM footprint has about 500,000 light-duty EVs in 2024, and S&P Global is forecasting about 23 million light-duty EVs by 2039, a growth rate of just under 30% annually during that period.
PJM has about 25,000 medium- and heavy-duty EVs in 2024, and S&P Global is forecasting about 1.45 million medium- and heavy-duty EVs by 2039, which is again about 30% per year for a 15-year growth rate.
S&P Global considered a number of state- and regional-level key drivers in their EV forecast, including zero-emission vehicle (ZEV) states and internal combustion engine (ICE) bans. They have models for which households they predict will purchase an EV.
PJM annually solicits information from its member electric distribution companies (EDCs) for large load shifts (either positive or negative) that are known to the EDCs but may be unknown to PJM. For the 2024 Load Forecast, these include:
- Expanding data center load and a large Intel computer chip plant outside of Columbus, Ohio, in the AEP service area
- Adjustment for data center load by FirstEnergy in the Allegheny Power System transmission zone, primarily the Quantum Frederick campus in Maryland
- PSE&G adjustment for data centers and electrification of New Jersey ports of Bayonne, Elizabeth and Newark, funded by the Inflation Reduction Act
- Dominion adjustment for data center load in Virginia
East Kentucky Power Cooperative (EKPC) requested a peak-shaving adjustment that began with the 2023 Delivery Year. EKPC provides PJM with weather triggers as well as program response matrixes for their three programs (smart thermostats, air conditioning switches and water heater switches), resulting in forecasted peak reduction of 5 MW to 10 MW.
The Long-Term Forecast Process
This report presents an independent load forecast prepared by PJM staff. The load forecast process considers residential, commercial and industrial sectors, each with its own set of models and inputs, including input variables for end-use saturation and efficiency as well as for economic drivers.
Insights from this process, combined with data on historical weather, are the starting point for determining peak and energy forecasts. PJM staff then makes adjustments based on forecast growth in behind-the-meter solar generation, battery storage and plug-in EVs, and also considers information from electric distribution companies on non-modeled trends, such as data centers.