When Matt White was a first-year assistant professor more than two decades ago, a new field in economics was in its infancy: market design. Today, he is one of the foremost practitioners in the field, overseeing a $10 billion suite of auction-based electricity markets as the chief economist for ISO New England.
Using three real-world examples from the grid he helps oversee, White outlined the state of renewable energy integration by regional grid operators in a presentation, “Market Design for a Greening Grid,” on Feb. 28 at the Saieh Hall for Economics. The presentation was an installment of EPIC’s Seminar Series, which hosts experts from around the world to explore different facets of the energy challenge.
ISO New England is one of several regional grid systems in the U.S. that coordinate wholesale electricity markets and delivery to customers ranging from the local utility to large energy users such as factories or airports. Like many power system operators, ISO New England has faced opportunities and challenges associated with the recent increase in production from renewable energy sources such as wind and solar.
“Sometimes, what works beautifully in economic theory, you learn—sometimes a little late—doesn’t quite work that way in practice,” White said.
Wind and solar power’s primary technological issue is intermittency—they only generate power when the wind is blowing or the sun is shining. Whereas the maximum output available from traditional power sources such as coal, nuclear and gas could be easily forecasted a day in advance, wind and solar can be highly volatile. That throws a wrench into the dynamic, real-time congestion pricing of whole electricity markets, White explained.
The challenge in maximizing market efficiency is two-fold: to forecast renewables’ output as accurately as possible (prediction) and to incorporate that forecast information into a market’s other suppliers’ decisions (coordination).
Take wind, for example. In European markets, grid operators offer a day-ahead prediction for next-day electricity delivery, along with several sequential day-of markets that occur in advance of real-time markets. That system allows wind operators to more accurately adjust their production to match actual delivered output, which addresses both prediction and coordination concerns.
U.S. markets, on the other hand, do not have sequential day-of markets. So, if wind over-performs its previous day’s forecast, costly back-up generation that was started early (like coal or gas) become unneeded and lose out to wind in the real-time market, distorting the wholesale market. If wind under-performs forecasts that means only very- expensive, fast-starting generation units can meet consumer demand, raising electricity prices.
To help address the intermittence issues, the federal government in 2012 required ISOs to conduct centralized wind forecasting, in which wind farms confidentially supply their “local” wind data feeds. White said those forecasts are generally “reasonably accurate,” coming within 12 percent of actual output, on average.
As for coordination between supplies, ISO New England publicly posts aggregate wind forecasts and lets sellers adjust their prices and generation mix decisions. California ISO, meanwhile, has proposed to act as a “market maker” that would buy back-up generation reserves (gas, nuclear, coal, etc.) from non-wind sources in the day-ahead market.
After presenting both the European and American wind power systems, White posed a central question to the audience: which is more efficient? The answer: only time will tell, and White said he hoped for some empirical research on the question.
White then moved onto prices and coordination for solar photovoltaic (PV). He explained that electricity prices must coordinate sellers’ real-time production closely. Before solar, updating prices every five minutes sufficed.
“But this is New England,” White said. “We have cloudy or full-cloudy days 70 percent of the time. This is not unusual.”
In other words, solar output is variable over a much shorter period of time—down to the second. That means markets have to coordinate output from other producers, such as certain hydro-electric and gas-turbine generators and grid-scale lithium-ion batteries (commonly referred to as energy storage), to offset that variation.
Coordinating between producers in this scenario requires the backup generation sources to ramp up rapidly. Assuming a grid battery responds two-times faster than a fast-generation gas-turbine, the battery would provide twice the output, regardless of cost—a mechanism that doesn’t currently exist, White said.
“It drops like ‘that,’” said White with a snap of his finger, referring to large drops in solar production. “There is no market design that discovers a market price for this service today.”
White closed the presentation with a look into the future. One alternative to the backup generation solar system would be “Prices to Devices.” In this scenario, thousands of electric vehicles (EVs) plugged in during the day could serve as the “battery” for the grid, filling in when solar production dips – then pumping electricity back to the car in quick cycles. EV owners might even be paid for plugging in their vehicles, as they’re providing a service to the grid.
But, as with any new technology, using EVs as the grid battery would present challenges.
“There are some hard tradeoffs there,” White said. “You can’t do it solely on willingness to pay, you want to do it fast. What’s value of speed of response? It’s not obvious there’s a mechanism that can properly balance the tradeoff between response speed and cost.”
It will be up to market designers like White to figure that out.