Flattening energy consumption curves

Power companies have long implemented short-term electricity interruptions as a strategy to meet customer demand during peak usage periods, such as during a heat wave, through direct load control contracts (DLCC). But according to Johns Hopkins Carey Business School Assistant Professor Ali Fattahi, the quest to manage demand spikes has gotten more complex — and pressing — in recent years due to factors ranging from global warming to the war in Ukraine.

In research described in two recent papers, Fattahi and his colleagues have developed a mathematical model that, they conclude, will improve upon existing DLCCs by allowing power companies to maximize the benefits of these contracts.

“By improving methods of estimating power demand and better allocating electricity consumption reductions among customers, our model would result in a 37.7% savings for utilities,” says Fattahi, whose team’s work on flattening energy consumption curves has appeared in Operations Research and has been highlighted in UCLA Anderson Review.

‘Smoothing out’ demand

For power companies, the task of estimating and managing electricity demand has been complicated by the rise in renewable energy sources such as wind and solar power, which are partially replacing old power plants. But output from renewable energy sources varies widely according to the season and time of day, and there currently isn’t enough battery capacity to store renewable energy at scale, explains Fattahi.

At the same time, extreme weather patterns — notably heat waves — are boosting the demand for power to historical levels, while the war in Ukraine has interrupted oil and some gas flow from Russia to Europe, which has lessened capacity for electricity generation in Europe.

“All of these factors contribute to the fact that nearly 20% of the total costs for electrical utilities come from the top 100 highest-priced hours in a year,” notes Fattahi, citing a study published in Popular Mechanics.

He explains that utilities attempt to “smooth out” consumption by offering discounts to households or businesses in return for limiting their energy usage during peak hours. A key weakness with existing approaches, Fattahi says, is that companies build in too much of a cushion.

“The current practice is very conservative. They want to be on the safe side — to be prepared for that summer heat wave that might never happen — so power companies aren’t using all of the interruption events they are paying for. This approach is suboptimal,” he says.

Forward looking and customer-friendly

In developing a more optimal model, Fattahi and colleagues had to factor in a variety of constraints since utilities can only reduce customer power use for a limited number of hours and days, and for a limited duration, each year.

“When we are establishing a contract at the start of the year, first we need to be forward-looking and develop a good forecasting model to utilize energy resources optimally,” says Fattahi. “Secondly, we need to assign interruption events to participating customers in a balanced way,” so that some customers aren’t impacted more than others.

“It turns out to be a very complicated math problem,” he says.

Using historical data from the California Independent System Operator, which is the umbrella organization of the utility firms in California, the researchers analyzed historical energy consumption profiles from 2009 to 2018 and then grouped these profiles into day types. Their resulting model, they note, is more flexible than existing models, allocating interruption events based on the forecasted demand on a daily or even an hourly level. With that information in hand, the team assigned energy reductions to individual customers.

They found that it was most efficient and customer-friendly for power companies to institute less frequent energy reductions of at least two hours. But they were also sensitive to ensuring customer satisfaction on extreme-condition days.

With the dynamic model in hand, Fattahi and colleagues ran tests on data pulled from the California utilities from 2016 through 2018. Their improved forecasting of demand, coupled with better management of interruption events, resulted in decreased energy production costs for utilities.

“We found this method is more cost-effective, reducing peak energy demand and saving money for power companies,” says Fattahi.

The team has published their model in an open platform and it is already garnering interest from utility company leaders, Fattahi says.

Incentivizing and managing power consumption for EVs

Looking ahead, Fattahi is now focusing on a related challenge in energy demand management, involving electric vehicles (EVs).

“EVs are the new air conditioners,” he says. “When air conditioners were first introduced, they created a whole new load of energy use to manage. Electric vehicles are even more difficult to manage because they consume a significant amount of energy and are charged at highly variable times.”

Fattahi is currently focusing on developing efficient models for managing demand of residential charging.

“Many EV owners, myself included, install home charging stations. It’s very convenient to charge the car overnight and wake up to a full charge,” he says. But as EV sales take off, the expected uptick in overnight charging will place a huge burden on the power grid, he notes.

“While we aren’t there yet, in a few years we will need to have programs in place that provide incentives to EV owners to change their charging time to off-peak hours, or to slow the speed of their charge,” he says. “So this work is very timely.”