Contributed by Sean Kelly, CEO, Amperon
On September 6, 2023, ERCOT came close to rolling blackouts for the first time since winter storm Uri hit Texas in 2021. Rather than severe winter weather, this time it was caused by the emergence of several late-summer variables. The growing risk highlights one reason why IOU’s and retailers are investing in net-demand forecasting systems. But co-ops and municipal utilities, which often have tighter budgets, need to know whether those systems justify investments as well.
On September 6, electricity demand peaked as people returned from work in the evening and began using the appliances and devices in their homes. But what was unusual about this day was that the temperature stayed very hot — around 100°F — until after dark, and the wind didn’t pick up as it typically does around this time.
On average, Texas gets more than 7% of its electricity from solar power, according to ERCOT figures. On the long, hot summer days of June, July, and August, solar power is available to offset peak demand in the early evening hours. But by September, the sun goes down earlier, and the grid relies more heavily on wind power (23% of the average energy mix in ERCOT). So, if the wind doesn’t pick up, and air conditioners across the region are working extra hard to combat the heat, it generates the kind of risk that triggered Energy Emergency Alerts on September 6.
This particular risk is unique to the “shoulder” months at the beginning and end of summer in networks like ERCOT, CAISO, and MISO, which are significantly increasing renewable generation while also cutting coal-fired power. During these months, net demand forecasting is particularly important for predicting how the narrowed margins of supply and demand will affect wholesale power pricing and demand charges — and perhaps require demand response (DR) events.
But there’s more to it
If advanced demand forecasting systems were only valuable during these shoulder months, they likely wouldn’t justify the investment for any utility. But weather patterns are becoming more volatile year-round, which means utilities need to be alerted of events that may affect market prices. The City of Denton, for example, lost $207 million in four days during Uri, because they weren’t hedged against the enormous price increases that took place. By comparison, a demand forecasting system that helps avoid that kind of hit is easily justified. AI has driven rapid improvements in weather forecasting, but public utilities should know that not all demand forecasting systems integrate comprehensive weather data. The output of demand forecasting models is only as good as the data inputs, so due diligence is required.
Volatile weather isn’t the only variable that’s making demand forecasting too difficult for conventional methods. Historically, it has been performed by human analysts using complex spreadsheets or SQL databases. They used regression and time series analysis to predict future energy demand and complemented it with time-consuming, manual analysis of customer usage and broad-scale weather patterns. In some rural coop territories with highly stable load, this approach may still be feasible. But solar power, electric vehicles, and electric heating are shifting demand patterns in most parts of the country.
Solar, for example, flips historic demand expectations upside-down, on hot sunny days — but not hot, cloudy days. Seasonal expectations are also flipping due to the trend towards electric heating. The New York ISO expects the shift to electric heating will drive cumulative winter demand higher than summer demand somewhere around the year 2040. In addition, electric vehicles could soon increase peak load on typical residential feeder lines by 30%, according to McKinsey calculations. But no one really knows yet what times of the day those EV-driven peaks will occur.
The complexity of technology- and weather-driven variables is analyzed in real-time by advanced demand forecasting systems to produce accurate short- and long-term forecasts that human analysts simply can’t achieve on their own. This is why EY predicted in 2020 that demand forecasting would be one of two areas in which AI will be a “game changer” for the utility industry. Large IOUs often have demand forecasting systems that run within their costly back-office enterprise systems, but public utilities can access the same information via cloud-based software-as-a-service (SaaS) offerings for a monthly subscription fee.
Planning data for public power
Finally, there’s one more way in which demand forecasting can be particularly useful to co-ops, municipals, and community choice aggregations (CCA).
Public utilities often don’t have access to robust planning data, if they don’t have the analytical tools that come bundled with those large enterprise systems IOUs and retailers use. For public utilities that are actively growing, the lack of good data makes it difficult to develop accurate capital spending plans and justify them to stakeholders.
Municipal utilities in the U.S. added nearly 2 million new customers from the previous reporting year, according to the American Public Power Association’s (APPA) 2023 statistical report, and coops added roughly half a million customers. Utilities experiencing this kind of growth need reliable data for planning.
By adding an analytical layer for historical meter counts, demand forecasting models can provide public utilities with short-, medium-, and long-term forecasts based not only on changing patterns of weather and customer behavior, but also on projected growth of the customer base. A municipal utility can then have a timeline for the addition of the next 1,000 meters, for example, and understand how its revenue and load position will be affected and how it should be hedged.
The demand models can generate results for different scenarios as needed to produce presentations for planning and oversight purposes. Customized models can reveal how a new manufacturing facility or agricultural operation will impact demand, for example. Or the system can be used to right-size an investment in solar, wind, or energy storage projects for which public power can now receive reimbursement following the passage of the Inflation Reduction Act in 2022. Accurate demand forecasting and validated reporting will also be necessary for projects supported by USDA Rural Development Grants.
Advanced demand forecasting isn’t necessary for public utilities that aren’t experiencing changing demand patterns or market conditions. For those that are, however, it can be a valuable tool.
About the Author
Sean Kelly is a veteran in the energy markets and Amperon’s CEO. Since 2005 Sean has been active in the evolving energy trading markets, working for Tenaska, Lehman, EDF, and E.On, as well as several family offices and proprietary trading firms. While at EDF, he led the transition of two nuclear plants (Nine Mile and Ginna) into the NYISO market, then led the buildout of E.On’s North American trading desk.
Joined by a team of deep domain experts in energy trading and market risk management, he co-founded Bridge Energy Consulting (sold to Ablireo Energy in 2019), providing energy management services and solutions to clients looking for new ways to operate effectively in a fast-changing market. In 2017 he co-founded Amperon as a way of combining leading-edge thinking in energy forecasting with cutting-edge analytical technology.
