It’s going to be a long, hot summer for folks living on the U.S. West and Gulf Coasts.
As early as April, the Enverus Power Market Analytics team has advised its customers to prepare for extreme high temperatures this summer. As we noted last week, western power markets from California to Texas came up against severe heat waves and low wind production, which led to a new June record for power demand in ERCOT last week.
When extreme weather hits and power price volatility goes haywire, energy traders need the most accurate forecasts available to make difficult choices. And as dangerous and deadly as February’s winter freeze and subsequent power outages in Texas were, the high temperatures we’re experiencing early in the summer months are equally unsafe. Secure power supply and accuracy in predicting load will be essential this summer.
Here’s a recap of the impacts to Texans and Californians in last week’s high temperatures:
- From June 14 though June 18, ERCOT and CAISO asked their customers to conserve power to prevent rolling blackouts.
- Historically high temperatures were measured from Houston to San Diego to Palm Springs, California.
- Houston reached 100 degrees Fahrenheit, which was its third earliest time to reach 100 degrees since 1930.
- Palm Springs matched its previous all-time max temperature of 123 degrees Fahrenheit.
The importance of getting load forecasts right throughout extreme weather events is the key driver of Enverus’ innovation in developing the outlooks that guide power traders around the country.
Record-high power demand to match extreme high temperatures
Even as Texans and Californians took on the responsibility of turning up thermostats and conserving power last week, peak loads were breathtaking in both ERCOT and CAISO.
- In Texas, peak loads were tremendous for ERCOT with peak load 104% above last year’s June record peak, or 115% above the average June peak load.
- California peak loads were even more impressive with peak load 120% above last year’s June peak load, or 140% above June averages.
Despite the severe demand and heat conditions, the Enverus peak load forecasts for ERCOT and CAISO SP-15, the grid that serves Southern California, were exceptionally accurate. We’re pleased to share the results from last week’s analysis:
In ERCOT, Enverus averaged a MAPE of 1.41% for June 14 through June 18, beating out the ISO peak MAPE average of 1.59%.
In CAISO SP-15, the results were even more dramatic. Enverus peak MAPEs averaged 6.54% vs. the ISO’s 10.15% — that 3.61 point difference in accuracy can mean the difference between profits and losses in power trading.
Deep learning enhances AI for a better accuracy and lower MAPEs
Last year, we started implementing deep learning into our machine-learning-based power market forecasts. Deep learning can be described as the more advanced cousin of machine learning, with stronger computing power and more granular datasets. The results of this product innovation blew us away and brought immediate results — our MAPEs (mean absolute percentage error) were now significantly lower than that of the Texas power grid operator.
We are proud to continue to help utilities and power traders get through tough spells of grid insecurity. And we continue to look to the future and how we will use our technology to support the management of volatility as renewables’ influence grows.
We know that western markets won’t be easy to manage this year. Having a reliable forecast will always be the key to managing volatility. Sign up below for complimentary access to our daily power market reports to keep ahead of the markets.