Weather, EE Influence Load Forecasting

Exploring-PJM-IconFor PJM, forecasting how much electricity customers will use years into the future involves the right data, hundreds of simulations using the data and thousands of employee hours.

It’s looking at the crazy weather of recent years – whether that’s the Polar Vortex, derechos or El Nino – and realizing that it’s more relevant to use data from last 20 years than the last 40.

PJM uses a number of models to estimate daily peak load for each PJM zone (defined by the area served by each transmission owner), the zone’s contribution to the daily grid peak and monthly net energy for load. The variables include the time of year, day of week, weather, economic drivers and end-use trends (such as spikes or declines in residential or business use).

Since the 2015 report, PJM significantly revised its load forecast model, which reduced projected peaks forElectrical-Storm several key benchmarks by 3.5 percent or more.

How PJM looked at weather played a big part in changes. PJM improved the accuracy of the forecast by shortening the forecast model to 20 years.

“We discovered that more recent weather is generally hotter, resulting in higher summer peak loads, and that pattern is likely to persist into the future,” said Tom Falin, manager – Resource Adequacy Planning.

The other significant change to the forecast is recognizing the impact of energy efficient electrical devices.  Distributed solar generation is now reflected in the forecast; it is important in forecasting summer peak demand and its impact is likely to only increase.

Compared to the 2015 Load Report, the 2016 PJM summer peak forecast shows the following changes for three years of interest:

The next delivery year – 2016 -5,781 MW -3.7 percent
The next RPM auction year – 2019 -5,660 MW -3.5 percent
The next RTEP study year – 2021 -8,406 MW -5.1 percent

To learn more about load forecasting, see the PJM training module, Load Forecasting and Weather. The complete PJM load forecasting process is covered in Manual 19: Load Forecasting and Analysis and a recently published whitepaper.