Earth Day 2020: How ISO-NE forecasts behind-the-meter solar

Weather is the biggest factor driving demand for electricity and, among grid operators, it’s long been said that “the load forecast is only as good as the weather forecast.”

That truism has become even truer as New Englanders have added an increasing amount of weather-dependent solar electric systems to their rooftops. With more than 180,000 such systems scattered around New England, an accurate forecast of cloud cover over the region’s 72,000 square miles is a critical element of an accurate, reliable load forecast. An accurate forecast is an important part of making sure the region has an adequate supply of electricity, while also ensuring that too much energy isn’t generated and transmitted across the grid.

Solar power changing historical load patterns

Almost all of the solar installed in New England is connected to the distribution system, rather than the transmission system managed by ISO New England, so the output of these systems is not directly “visible” to ISO system operators. While ISO system operators don’t know how much electricity each of those systems, known as behind-the-meter photovoltaics (BTM PV), is producing, their combined impact on load is clear. When these rooftop systems are cranking out electricity for the homes or businesses on which they are installed, demand for power from the regional grid drops. When the sun goes down, the clouds roll in, or snow covers the panels, those houses and businesses turn to the regional grid for electricity, and demand pops back up.

Computer models revamped to incorporate burgeoning BTM PV variables

As solar power grew in New England, the software used by ISO forecasters absorbed the fact that demand was dipping on sunny days. However, the “why” of it—that solar panels were serving local load—was not data that was available to the software.

To accommodate these changing variables, the ISO’s load forecasters began manually applying the estimated impact of BTM PV resources on the daily forecast. But estimating the hourly load impact of BTM PV became progressively more challenging as more solar was installed.

With a longer-term solution needed, the ISO launched an initiative to integrate the effects of BTM PV into the computer models producing the daily operational load forecast. The new load-forecast models have been in place since March 2019.

To fill in the dips in the daily load curves when solar panels are serving local load, a vendor provides the ISO with the historical output of 10,000 rooftop systems scattered around New England in five-minute increments. The data from that sampling of BTM PV systems is assumed to represent the actual performance of all BTM PV in each town, under the actual weather conditions that existed in each town. The ISO then upscales the data from the smaller sample to approximate the output of all 180,000 BTM PV systems, arriving at the total regional PV output.

With this information, the demand that was met by BTM PV can be added back into the historical load curves. The reconstituted load curves show how much electricity New England was using every hour, whether it was coming from rooftop PV or the regional grid.

With an accounting of what actual electricity use (served by both the grid and BTM PV) was during different weather conditions, the ISO’s forecasting models have an accurate starting point. This is an important step because the demand for power is out there, whether it’s being served by the regional grid or by rooftop panels. When BTM PV is not available, the regional grid must be ready to meet that demand.

The ISO then developed new software to turn a weather vendor’s forecast of irradiance—the strength of the sun’s rays— for each city and town in New England into expected production, in megawatts, from the region’s 180,000 BTM PV systems. This forecasted production is then fed into the ISO’s load forecasting software to produce an even more accurate forecast of grid demand.

Since launching these new forecast models, the ISO’s forecast accuracy has improved, though challenges persist when actual cloud coverage differs from the forecast. As it always has been, the load forecast is only as good as the weather forecast.

Note: Societal changes in response to the COVID-19 pandemic have led to challenges in forecasting consumer demand for electricity in recent weeks. See this Q&A for more information.

For more information on how ISO New England is working to ensure a reliable power system and competitive wholesale electricity markets that will enable the clean-energy transition, read the 2020 Regional Electricity Outlook.