In my last column I took Dominion’s Integrated Resource Plan (semi-) seriously, giving the utility the benefit of the doubt in its projections for data center growth and the alleged need for more fossil fuels to keep up with the power demands of that ravenous industry. But as I also noted, Dominion doesn’t deserve to be taken seriously with this document.
Under Virginia law, an Integrated Resource Plan (IRP) is supposed to explain how the utility expects to meet demand reliably and at low cost within the constraints of the law. In this IRP, however, Dominion asked a different question: how to make Gov. Glenn Youngkin happy by keeping fossil fuels dominant regardless of both law and cost.
Coming up with a favorable answer required Dominion to ignore the Virginia Clean Economy Act (VCEA), create arbitrary limits on solar deployment, use wishful thinking instead of facts and, for good measure, do basic math wrong. They also assume the Youngkin administration will succeed in pulling Virginia out of the carbon-cutting Regional Greenhouse Gas Initiative (RGGI), a move that is being challenged in court. This makes Dominion’s IRP, with its plan to double carbon emissions, a sort of evil twin to a plan from any RGGI state.
Sure enough, the governor loves it. Elsewhere, however, this evil twin found a cold reception. Experts retained by environmental, consumer and industry groups all agree that this IRP should be chucked in the trash bin and Dominion told to start over.
It’s worth taking a look at the testimony from these groups to understand where Dominion went so badly wrong, and what a better plan might look like.
It’s all about the data centers
Northern Virginia data centers are the driving force behind Dominion’s plans to burn more coal and gas, but there is some disagreement whether their growth will be absolutely off-the-charts crazy, or merely eye-poppingly huge. Dominion’s IRP projects the data center industry’s power use in its territory will quadruple over the next 15 years, rising from 2,767 megawatts (MW) in 2022 to more than 11,000 MW in 2038. At that point it would represent close to 40% of Dominion’s load.
Experts testifying in the IRP case believe data center demand won’t reach the dizzying heights Dominion projects. They question whether Virginia communities will accept so much new data center development, given the pushback already evident in localities like Prince William and Fauquier Counties. Their thinking is essentially that if things can’t go on this way, they probably won’t.
They also suggest that some of the demand Dominion expects may also be reflected in the plans of other utilities serving Northern Virginia’s Data Center Alley, leading to double-counting. A load forecast published by grid operator PJM shows that Northern Virginia Electric Cooperative (NOVEC) projects its data center demand to rise from about 400 MW in 2022 to 4,000 by 2028 and 8,000 by 2034. Two other cooperatives project a combined 3,000 MW of data center development in the same time period.
Aside from data centers, demand for electricity in Dominion territory is flat or declining over the next fifteen years; presumably this is due to the increased efficiency of homes and businesses offsetting the increased demand from electric vehicles and building electrification.
If you were already experiencing vertigo over Dominion’s data center numbers, and you are now hearing for the first time that Dominion’s numbers represent only half of the total projected data center load coming to Virginia, maybe you won’t find it all that cheering that some of that added demand could be illusory. Still, it is a reasonable point. If you reduce demand by a few thousand megawatts here and there, pretty soon you might not “need” a new gas plant.
As an aside, one Virginia utility stands out for apparently not expecting much in the way of new data centers in its territory. Appalachian Power, which serves customers in Southwest Virginia, West Virginia and Tennessee, has experienced declining demand for years, and rural Virginia leaders would dearly love to see data centers come there. Yet the PJM forecast shows Appalachian Power’s parent company, American Electric Power (AEP), told grid operators it expected only about another 200 MW on top of the 500 MW of data center load it serves across its entire 11-state territory. AEP seems to regard this as quite a lot, which, when compared with what is happening in Northern Virginia, seems rather sweet.
I can’t help but wonder how Gov. Youngkin managed to make a deal with Amazon to bring $35 billion worth of new data centers to Virginia without securing a guarantee that many of the facilities would be located in a part of the state that actually has surplus energy capacity and desperately needs new economic development. Southwest Virginia voters may have put Youngkin in office; you’d think he’d be looking out for them.
Instead of real investment today, Youngkin promises the area a little nuclear plant a decade from now, when and if small modular reactors (SMRs) prove viable. Local leaders must be muttering, “Gee, thanks.”
And this leads to another point raised by several of the experts in the IRP case: All of the load growth Dominion projects is due to a single industry in Northern Virginia; elsewhere, demand is decreasing. Yet, instead of crafting a solution specific to the industry and region experiencing runaway growth, Dominion proposes to build a fossil fuel plant 140 miles away in Chesterfield and a nuclear plant 300 miles away in another utility’s territory.
Gregory Abbott, a former SCC Deputy Director, offers a particularly withering assessment of the IRP in his testimony on behalf of environmental advocacy group Appalachian Voices. Dominion’s computer model, he says, “is proposing supply-side solutions that are not focused on solving the actual problem, are likely unnecessary, and driving costs higher than they should be.”
Although Dominion insists this IRP represents just a “snapshot in time,” Abbott says that’s misleading: The IRP “sets the stage for multi-billion-dollar investments that Dominion’s customers will pay for decades to come. If a future snapshot in time changes, based on new public policy goals or market dynamics, ratepayers are stuck with paying for these sunk costs.”
Garbage in, garbage out
Abbott and others also note that unlike factories or other high consumers of energy, data center operators can shift some functions to other data centers elsewhere for short periods of time. Dominion could save money and reduce the need for new investments by capitalizing on this capability to develop demand-response programs tailored specifically to this industry.
Instead, Dominion treated the surge in demand as if it were statewide and spread across all its customers; then it used a computer model to figure out how to meet the soaring demand.
An expert for the Sierra Club, Devi Glick of Synapse Energy Economics, noted several problems with Dominion’s approach. Among them: the company told its computer model that it couldn’t select energy efficiency as a resource; it had to include gas combustion turbines in 2028; it had to adhere to artificial limits on solar, wind and battery storage; and it had to assume prices for solar that were “substantially higher than industry projections.” Dominion also did not instruct the model to account for proposed (and since finalized) new federal pollution limits that will raise the cost of burning fossil fuels, and miscalculated — by a billion dollars — the penalties associated with failure to meet the VCEA’s renewable energy requirements.
As they say, garbage in, garbage out. The model did what it was told, and produced plans that limited solar and battery storage, called for new gas combustion turbines and/or SMRs, and kept uneconomic coal plants running past their previously-planned retirement dates. Accordingly, none of the modeled scenarios complied with Virginia law and all would be unnecessarily expensive for customers.
Synapse ran its own computer model that kept most of Dominion’s load and cost assumptions but corrected for the company’s errors and artificial constraints. The results, not surprisingly, show that building more solar and storage and retiring coal plants earlier than Dominion wants to will lower carbon emissions and “reduce costs for Dominion’s ratepayers by between $4.1 and $9.0 billion over the 25-year study period.”
When Synapse then tweaked the model to reflect the new federal pollution rules and prices for solar and battery storage in line with industry projections, the results saw solar and battery investments soaring, while the “need” for firm capacity such as a new gas plant disappeared altogether during the planning period.
Clean energy, vindicated
So finally, we begin to see a path forward founded on real data and not constrained by political expediency. With none of its plans meeting the basic requirements of Virginia law, Dominion should be ordered to go back to the drawing board. The company should reexamine its data center load projections and design a demand-response program tailored to that industry. Then it should re-run its computer model with energy efficiency allowed as a resource, with no artificial constraints on battery storage and renewable energy,with federal and state compliance costs associated with fossil fuels fully included and with cost estimates for solar and storage consistent with industry norms.
The General Assembly has a role to play, too, in ensuring the data center industry does not shift costs onto other customers and cause Virginia to fall short of its carbon reduction goals. Data centers should be required to meet energy efficiency targets and to secure an increasing percentage of renewable energy on their own as a condition of obtaining generous state tax subsidies. Likewise, the State Corporation Commission should be required to ensure that data centers pay for the transmission upgrades they need. Finally, the General Assembly should pass the data center study bill adopted by the Senate this year before being killed in the House.
Finally, it’s clear that the computer models will select as much low-cost solar as they are allowed to, so the General Assembly should make it easier to build solar projects at competitive rates. They can do this by further opening the market to third-party developers, who are currently constrained by an interpretation of the VCEA that caps their share of the solar Dominion procures at 35%.
The SCC will hold a hearing on the IRP beginning September 18.
An earlier version of this article appeared in the Virginia Mercury on September 5, 2023.