Data centers that promise to reduce electricity use when the grid is strained could pay other, more flexible users to make some of those reductions instead, under a market proposed by Cornell University researchers. Simulations suggest the approach could lower the cost of meeting required cutbacks, but it has not been put into operation.
The researchers tested the proposed Curtailment Credit Market on three modeled grids, including a simplified version of New York's, according to a Cornell Chronicle account of the study. The paper was published Sept. 22 in Applied Energy, with doctoral student Richard Mahuze as lead author.
How the proposed market would work
Some large energy users can connect to the grid sooner through a non-firm agreement, accepting a requirement to cut their demand at certain times. That obligation can be costly for a data center whose customers expect nearly uninterrupted service, said Max Zhang, the Cornell engineering professor who oversaw the project.
Under the researchers' design, a grid operator would signal a need to reduce demand as transmission lines become constrained. A data center facing a cutback could buy a credit from another eligible user willing to reduce or shift its electricity use for payment. Trades would be approved only if they met the grid's physical constraints and delivered the required reduction.
Zhang offered a hypothetical 100-megawatt cutback: A data center could buy credits from users able to absorb some of that obligation at lower cost. Potential sellers include factories, large businesses, virtual power plants and aggregators that coordinate batteries, solar generation, electric vehicles or building controls. The proposal would not simply let the data center ignore its agreement; the reduction would have to occur elsewhere on the network.
Such flexibility has a price. Zhang compared it to an airline paying passengers to give up seats until enough volunteer: Energy users that can move or reduce demand might do so if the payment exceeds the inconvenience or cost. The proposed market is designed to find willing sellers while ensuring power lines can handle each trade.
What the simulations found — and did not prove
Across three simulated test grids, the researchers found that trading could avoid at least a third of the economic damage associated with power reductions. That is a modeled result, not a measured saving on New York customers' bills. The New York case used a simplified grid model, and the paper did not establish that the market would deliver the same result under live operations.
The researchers say using existing lines more fully could reduce pressure for upgrades and help contain energy costs. They also see a possible air-quality benefit if a data center can buy a credit rather than turn to a backup diesel generator. Those potential effects depend on how a real program is designed, who participates and what resources actually supply the reduction.
The proposal comes as large electricity users face long waits for grid connections and regulators examine how they should be served. In June, the Federal Energy Regulatory Commission directed six regional grid operators, including New York's, to justify or revise their connection rules for data centers and other large loads. That federal action did not adopt Cornell's proposed credit market.
The research grew from discussions with industry representatives about incentives and network limits. Cornell Atkinson Center for Sustainability is supporting the team's work with the Environmental Defense Fund on a mechanism for fair and efficient compensation. Zhang said the published study shows feasibility in simulations; the next challenge is making the market implementable.
Ali Amadeh and Charlotte Gressel also contributed to the study. Support came from the David and Patricia Atkinson Foundation and the Indonesia Endowment Fund for Education.
FingerLakes1.com is the region’s leading all-digital news publication. The company was founded in 1998 and has been keeping residents informed for more than two decades. Have a lead? Send it to news@fingerlakes1.com.
