You are here

A Year in Review of Behind the Meter Solar: How is Behind the Meter Solar Generation Driving Price Volatility in New England?

solar panels

In this blog series, we explore the regional power market impacts of behind the meter solar generation (BTM). In this post, we examine the ISO New England (ISO-NE) power market region, where we are starting to see clear patterns of price volatility driven by incorrect expectations of BTM. Stay tuned for the next posts when we identify new CAISO patterns of congestion triggered by BTM and analyze the outlook of BTM market impacts in NYISO.


This past winter, behind the meter solar capacity in New England (NE) was reported by the ISO at an all-time high of 2.4 GW, nearly 10 percent of annual forecasted peak demand. For the first time, unexpected uses of behind the meter solar generation (BTM) have caused negative real-time (RT) pricing across the entire NE footprint, introducing massive opportunities for traders to capitalize on this new market volatility. On May 1, 2018, we saw that BTM now has the potential to cause RT prices to drop past -$150, as demonstrated in the figures below. Actual demand dropped 700 MW between 1pm and 2pm ET, when the market expected an increase of 260 MW in demand. This caused RT prices to plummet.

The market impacts of BTM are simple economics: increased BTM decreases the demand for wholesale power, which in turn suppresses prices. So why is BTM confounding the market?

MA demand

Background Information

The challenge with resources for BTM is that they appear as a lack of demand, not as a traditional resource. Looking at a typical ISO-NE shoulder-season demand curve (below) from April 30, 2015, there is a slight midmorning peak and a dip in the afternoon before an evening peak. Only two years later, for a similar weather day on April 13, 2017, the demand curve has changed significantly and displays a drop-in demand during daylight hours. Many factors can contribute to decreasing demand, so it is hard to pinpoint the magnitude of actual BTM and isolate price volatility driven by it. Estimates for BTM are available, but models are limited by a variety of factors such as availability of surface irradiance measurements as well as assumptions of panel performance impacts like soiling and shading.

In ISO-NE, the ISO forecasts demand, clears virtual demand at MA Hub, and incorporates virtual demand with the forecast into demand cleared in the Day-Ahead (DA) market. MA Hub is a “virtual” zone because it does not have physical generation or demand, so every megawatt (MW) is virtual demand or generation at that location. It is considered a metric for market expectations and energy costs in New England. The demand cleared is a financial commitment for the purchase of wholesale power – or the market expectations of demand the next day. When virtual demand is negative it will bring demand cleared lower than the ISO forecast, and vice versa, with the aim of better estimating actual demand. Significant differences between demand cleared and actual demand cause price volatility and signal that the market is missing information.

Actual Demand

Summary of Findings

Genscape and Locus Energy have combined their proprietary energy monitoring networks and market insight to provide measurements of BTM and analysis of market impacts. While the market is struggling to understand the magnitude of BTM – creating significant differences between demand cleared and actual demand - Genscape analysts are leveraging Locus measured BTM data to compare historical BTM and pricing volatility, and improve forecasts of volatility. When the magnitude of BTM misses expectations, it consistently creates price volatility. To further clarify, when market expectations of demand are too high and BTM comes in high, RT prices are suppressed; whereas when market expectations of demand are too low and BTM also comes in low, RT prices increase.

Analysis Methods

Using measured generation data collected by Locus monitors, Genscape calculates demand zone level BTM by scaling the data with reported installed capacity values. For the two similar weather days shown below, the BTM added to actual demand for the 2017 day illustrates how the BTM matches the magnitude and timeframe of the drop in demand.

Actual Demand + BTM

Analyzing BTM alongside market conditions for last year, several patterns emerge.

“On days when solar outperformed expectations, ISONE saw demand come in dramatically lower than expected and a positive spread between Day Ahead and Real-Time prices (DART)."

For example, on June 30, 2017, load came in at 90 percent of what the ISO forecasted. At 1pm ET, there was almost 1 GW of BTM (nearly 8 percent of load). At that time, there was a $22 difference between DART prices and the percentage of BTM in the demand curve was more significant than any other factors decreasing demand, such as temperature difference. The actual demand for this day began to diverge from ISO forecasted demand and demand cleared around 8am remaining significantly lower until 4pm. This timeframe matches when BTM was greatest.

Cleared Demand

Again, on July 21, 2017, BTM suppressed RT prices. This was a day with massive demand that would have reached over 23,100 MW without BTM. The system was so tight that oil units were ready to come online if needed with a spinning reserve of $0.44/MW. Typically, spinning reserves are $0 most of the day in ISO-NE and anything higher than that is a signal that generators need to be ready to produce in 10 minutes. Despite this signal, RT prices were kept significantly lower while BTM generation was strongest, as seen in the charts below. At 2pm, the DA price was $72.42 and with the penetration of BTM at 6.36 percent of demand, we saw a $41 DART spread.

Day Ahead and Real Time Price Differences

For both examples, virtual demand was negative throughout daylight hours, indicating that the market expected strong BTM, but did not have enough information to understand the full magnitude of impact it could have and successfully mitigate the risk.

“As for days when solar underperformed, ISO-NE saw demand come in higher than expected along with negative DART spreads.”

On May 29, 2017 (Memorial Day), actual demand reached nearly 950 MW higher than ISO forecasted demand. The demand curves for that day (below) show demand cleared even lower than the ISO forecast signaling that the market expected BTM. However, as illustrated in the generation curve for that day (below), Genscape BTM data  showed that it only reached a maximum 20 percent of capacity at 1pm while it had reached 58 percent on the previous day, May 28, and reached 33 percent on May 30. As a result of unexpected cloud cover, BTM fell below expectations causing negative DART from 11am to 5pm on an otherwise bearish day.

Actual and Cleared Demand

June 27, 2017, is an example of localized storms affecting BTM differently across the region and driving volatility. The region overall saw clouds in the morning with afternoon storms. From 11am to 12pm, solar spiked about 260 MW and DART spreads flipped from -$7.00 to +$3.00. RT prices remained bearish until BTM started ramping back down. Across those hours, virtual demand was negative with an expectation of higher BTM. The morning underperformance paired with the dramatic increase in BTM starting at 11am caused this unexpected DART spread flip midday. This change in cloud cover (shown below) was especially pronounced in southern Massachusetts, where it was overcast at 10:45am and cleared significantly from 11am to 12pm. BTM in this region outperformed other NE Demand Zones that afternoon.

Mass Hub Day Ahead and Real Time

Visible satellite

Outlook & Conclusions

ISO-NE is starting to include BTM forecasts in their DA demand cleared, however, consistently less demand is cleared to hedge against unpredictable spikes in generation. There is an opportunity here to capitalize on expected impacts of BTM at the hourly level. Increasing capacity means estimating this magnitude is only going to become more impactful. BTM installed capacity growth rates have increased steeply over the last five years as seen in the survey results from ISO-NE and show no signs of slowing.

Massachusetts already leads the region in greatest installed capacity and aims to add 1,600 MW through the Solar Massachusetts Renewable Target Program alone. With such aggressive goals being set at the state level, the driving force of BTM in the market is expected to strengthen and expand in the near future as shown in the ISO-NE cumulative capacity forecast below.

Final 2017 PV Forecasat.png

Historical Installed PV Capacity Survey Results

Genscape is preparing for these changes in the market by analyzing trends in BTM volatility and corresponding RT price impacts. Exclusive insight into the magnitude of BTM from Locus data combined with Genscape analysis of market conditions provide unparalleled insight into the increased pricing volatility we are already beginning to see in 2018. Contact us to learn more about increasingly negative RT prices due to BTM in 2018 and forecasts for future volatility.

Similar weather days are ranked by day of the week, holiday flag, and weather parameters including forecasted and observed temperature, dew point, and cloud cover

This entry was posted in: