Carnegie Mellon University

wind turbine

January 11, 2019

Faculty Spotlight: Wind Farm Research Illustrates Why it’s Not Easy Being Green

Finding clean ways to provide a constant source of energy represents the proverbial Holy Grail of environmental policy — and, like the Grail, it’s a goal that has proven elusive. 

Our seemingly insatiable demand for energy has long been at cross purposes with the desire to preserve the environment. Efforts to balance the two have created an interesting dilemma: clean power sources aren’t always constant, and constant sources aren’t always clean. For example, in the U.S., wind energy farms tend to produce more energy at night, when wind is strongest, than during the day, when consumer demand for electricity is highest. To help bridge that gap, a few wind farms have added batteries to store excess energy for when demand peaks. 

But managing such systems, to maximize profit when various market factors and uncertainty come into play, can be very challenging, according to new research from the Tepper School of Business. The research, “Managing Wind-based Electricity Generation in the Presence of Storage and Transmission Capacity” recently published in the journal Production and Operations Management was coauthored by Helen Zhou (MS 2009, Ph.D. 2012); Alan Scheller-Wolf, Richard M. Cyert Professor of Operations Management, Tepper School Senior Associate Dean for Faculty and Research; Nicola Secomandi, Professor of Operations Management and head of the Ph.D. program; and Stephen Smith from Carnegie Mellon’s Robotics Institute.

If electricity prices are always positive, the optimal operating policy is fairly straightforward, notes Scheller-Wolf. This policy simply factors in price, wind speed, and inventory (the amount of energy currently stored in the batteries), and offers thresholds for operating the farm. This structure extends those of known optimal policies for other related energy storage models, points out Secomandi. 

However, and somewhat surprisingly, electricity prices are not always positive. For some traditional electricity producers — notably coal and nuclear power plants, which are less environmentally desirable than renewable energy generators — significantly adjusting the output can be expensive. Thus, they may keep generating and selling power in excess of demand, driving prices into negative territory, to avoid ramping their systems up and down. In addition, the government sometimes subsidizes renewable energy as an incentive. Consequently, wind farm operators can turn a profit by generating energy even if market prices are negative. So, especially if they can’t store wind energy, they keep generating it, exacerbating the negative prices.

It turns out that these negative prices can lead to significant and unforeseen effects at wind farms with storage: When prices are negative, wind farms with batteries can get paid to take potentially dirty power dumped by other generators, such as nuclear and coal plants, store it, and then sell it when demand is high — but in so doing, they use up storage space intended for their own clean power, potentially displacing the very clean energy they were built to provide. Because energy storage and transmission are inefficient, a similar issue arises even if the purchased power is green.  

“It’s kind of hard to beat physics. We need the power during the day, and wind blows at night,” says Scheller-Wolf. “It’s hard to adjust baseline power loads quickly. And if you build batteries to help wind farms, people may repurpose them, because that’s how capitalism works.”

The Tepper School research offers a model for potentially negative energy prices that enables the authors to optimize how much electricity a wind farm should generate, how much it should store, and how much it should sell, as well as how much battery storage it should use to buy power from the grid in order to maximize profits. An adaptation of the optimal policy for the case when electricity prices are always positive works remarkably well also when they can be negative, says Secomandi. Though the model is designed to maximize profit, Scheller-Wolf points out that it could be adjusted to meet other criteria. For example, the operator could impose a constraint that ensures that no more than 10 percent of its battery storage goes to dumped power.

In the future, Scheller-Wolf said he’d like to see negative energy prices disappear. “Inherently, paying people to burn your electricity seems like a bad thing,” he says. “It’s like life: sometimes, the easiest way to make a problem go away is to throw money at it. But it seems like it would be better just to not generate the electricity in the first place. Especially if it comes from dirty sources.”