PJM Proposes Refinements to Long-Term Load Forecasting

New method produces more accurate forecast, accounts for growing impact of electric vehicles, allows for improved energy efficiency and slowing industrial output.


To improve overall accuracy, reflect current energy usage trends and account for the increasing adoption of electric vehicles, PJM has introduced proposed adjustments to methods for calculating the annual long-term load forecast.

Proposed refinements are expected to trim the current load forecast. The proposed downward adjustment mirrors PJM’s expected load with similar, neighboring economic regions served by grid operators in New York and New England.

PJM’s summer peak load forecast is expected to decline 1.5 – 2 percent, versus the current forecast in coming years, to as much as an 8 percent decline by 2034.

Presentations during the Load Analysis Subcommittee meeting Aug. 6 represented PJM’s first conversation with stakeholders to seek feedback on proposed changes. That process, in coordination with consideration by the Planning Committee in September, will culminate this December with the traditional release of the 2020 Load Forecast Report.

Primary changes proposed for PJM’s load forecast model are expected to reduce the load forecast from a slight growth outlook to, for the first time, a slight reduction in both peak demand and overall electricity use. The downward adjustments in demand would be offset, in part, by expected growth in electric vehicle usage over the 15-year forecast horizon.

“We are concerned that the current model structure was overstating load,” PJM Senior Analyst Andrew Gledhill told stakeholders of the new load forecasting methods which have been in development since last year. “We typically think of it [the load forecast] as an evolving thing…. a better model than what we are currently using.”

Accounting for Plug-in Electric Vehicles

Proposed changes also explicitly account for, for the first time, plug-in electric vehicles (PEVs). While still a work in progress, this analysis derives estimates of PEV load at peak from a broad range of data sources, including the number of registered electric vehicles by state, electric vehicle sales forecasts, an estimate of PEV annual energy at 4,500 kWh per vehicle and more. While PEV sales currently account for fewer than two percent of light-duty vehicles sold, that figure is expected to grow to up to six percent of passenger vehicles by 2030.

Better Modeling to Reflect Current Energy Trends

Specific changes incorporate more information on energy use by sector. This accounts for how economic activity factors into expected peak load and energy, better aligning non-weather-sensitive and weather-sensitive load with their drivers.

Gledhill discussed how the load forecast is modeled today and the proposed changes. Non-weather-sensitive load makes up the bulk of demand in summer or winter. Correctly forecasting this variable is critical to the forecast of weather-sensitive load, Gledhill said. Long-term energy usage patterns, as a result, are more fully accounted for, including the continuing decline in industrial energy intensity. Proposed model changes result in a final load forecast that factors in trends in both weather- and non-weather-sensitive load.

In general, non-weather-sensitive load currently flatlines year to year, a reality not reflected in existing model projections for future years, Gledhill said. In reality, these steady load results reflect increasing energy efficiency and the less energy-intensive industrial mix in PJM’s footprint, he said.

With weather, behind-the-meter solar generation, electric vehicles and all factors accounted for, the PJM summer peak load forecast is expected to decline by 1.5 – 2 percent, versus the current forecast in coming years, to as much as an 8 percent decline by 2034.

Long-Term Forecast on Track for December Release; Mid-Year Updates to Resume 2020

As these changes are being considered in the stakeholder process, PJM is on track to release its usual long-term load forecast in December. However, as the model continues to evolve, PJM will not post its usual mid-year load forecast update in 2019. Instead, PJM will resume mid-year load forecast updates in the summer of 2020 to provide market participants insight into upcoming auctions.

“PJM is not doing a mid-year forecast, given the changes to the forecast this fall,” PJM Senior Lead Economic Analyst John Reynolds said. “We felt it would be misleading.”

PJM requests information from electric distribution companies on any significant load changes to be considered for incorporation in the 2020 load forecast by September 1. As stakeholders consider new models for peak and energy estimates in their zones and provide feedback, it is anticipated that the Load Analysis Subcommittee may reconvene before the Planning Committee meeting in September, Reynolds added.

“This proposed model has gone through intensive internal scrutiny,” Reynolds said.

“I implore stakeholders to look at the data and provide feedback to PJM,” Gledhill added.

Stakeholders are encouraged to send feedback to PJM’s Planning Department at load_analysis_team@pjm.com.