U.S. EPA is relying on a complex computer model of the nation's energy networks and markets in declaring that its proposed Clean Power Plan will not lead to grid blackouts and unsupportable consumer costs.
The EPA model, which the agency has used in various forms for 20 years, can require hours to churn through data on U.S. power plants. EPA used it in this case to analyze the consequences of the administration's proposal to cut carbon emissions from power plants by about a third.
Despite its pedigree, the EPA integrated planning model (IPM) is the target of a sharp attack from several states opposed to the plan and is expected to be under fire next year when Republican-led committees challenge the Clean Power Plan (CPP).
"Ohio EPA has serious reservations concerning U.S. EPA's over reliance on the IPM model to predict the proposed rules feasibility, cost to consumers and impact on reliability," the Ohio state environmental agency said in comments to EPA.
The North American Electric Reliability Corp. (NERC), the federally designated monitor of interstate power grid performance, has retained a critic of EPA's model -- the Northern Virginia consulting firm Energy Ventures Analysis (EVA) -- to assess the CPP's potential impact on the grid.
In an analysis of the CPP last month commissioned by coal producer Peabody Energy Corp., EVA said the EPA analysis was compromised by "flawed assumptions" in the four "building blocks" that the agency has advanced as compliance strategies.
"They did it all wrong," Thomas Hewson Jr., an EVA principal who directs the firm's environmental studies, said in an interview. "There are so many problems with EPA's building block assumptions, it boggles the mind."
EVA said the EPA proposal vastly exaggerates the amount of possible energy efficiency savings and potential operating efficiencies in remaining coal plants. EPA overestimates the amount of gas-fired and renewable power generation that will be available to replace retired coal units, the consultancy says.
With realistic figures, the compliance costs for CPP escalate, EVA says. It calculates that current and proposed EPA power plant regulations would increase electricity costs by $177 billion in 2020 compared with 2012.
EPA's analysis puts annual compliance costs at no more than $8.8 billion, while it calculates climate and health benefits of $55 billion to $93 billion in 2030. Estimated benefits were not included in EVA's calculations, nor in Ohio's. "Ohio EPA does not focus on the state objectives related to climate change," the state filing said.
The choice of EVA was challenged at a NERC planning meeting Dec. 10 by Michael Goggin, research director for the American Wind Energy Association, according to an account of the meeting. Goggin expressed "serious concerns" that the EVA study "is being set up in a certain way to find a certain conclusion." Mentioning other EVA studies, he said, "there's a pretty consistent bias in some of the work on being pro-coal, anti-renewables."
Hewson disputed accusations of bias on EVA's part. Asked about Goggin's comments, John Moura, NERC's director of reliability, said NERC had worked for many years with EVA and had no issues with its objectivity.
Models at odds
Hewson said that while EPA's assumptions are critical inputs to the EPA model, the model's structure itself is not detailed enough to identity possible future shortages and bottlenecks in electricity delivery that could come from the CPP.
The computer model EPA employs is its adaptation of a planning tool developed by ICF International nearly 30 years ago and has been steadily refined to serve a long list of government and private-sector clients, the firm says.
Most recently, the ICF planning model was chosen for a study on coordinating the electric power and natural gas delivery sectors to ensure adequate fuel supplies for gas generators. The report was conducted at the request of the Eastern Interconnection States Planning Council, representing state government agencies in the 39 states in the eastern U.S. power grid. The project was funded by the Energy Department and administered by the National Association of Regulatory Utility Commissioners.
Jennifer Macedonia, senior adviser at the Bipartisan Policy Center, said pitting the EPA planning model against EVA's dispatch model may miss a basic point -- they are designed to answer different questions.
The EPA model compiles information about all U.S. power plants broken down into 64 regions of the country, including the type and quality of the fuels they burn, their output, environmental controls, wind availability, fuel transportation networks serving the plants and plants' operating costs. The model includes data on high-voltage transmission capacity to move power between regions to tell planners how much power a region might be able to import if its power resources aren't sufficient.
Using a series of mathematical equations, the model seeks the least expensive way to ensure a region will have the electricity it needs long term, whether by building plants, importing power or conserving it, with an adequate reserve margin of spare capacity.
IPM models have helped utility operators make the "build or buy" decision on whether to put money into a new modern plant or retrofit an old one with new environmental controls. The "integrated" model's complexity comes from its simultaneous examination of outcomes for all the power plants in its database, since the best decision for each plant may affect decisions for its neighbors.
As the multi-year scenario unfolds, some plants close and new ones start up, changing the competitive relationship among the remaining plants, and then the model recalculates the supply-demand mix. This dynamic process is repeated over and over until the model has found the most cost-effective way to supply the necessary amount of electricity in the future target year, statisticians explain.
IPM "does a good job of characterizing the dynamics of power markets -- for example, fuel prices, capacity needs and compliance strategies for environmental regulations," Macedonia said in an interview.
"It certainly has a lot of strengths. Like any model, it's a tool. And so it's just one part of an analysis and shouldn't be viewed as the full picture. Given the uncertainties associated with how states will implement the CPP, as well as the typical uncertainties associated with future technology costs and fuel prices, no model is going to be able to pinpoint the exact costs. But it can inform on the range of likely costs and highlight lowest cost pathways to meet the CO2 reduction goals of the CPP," she said.
Hewson said EVA's analysis employs a commercial Aurora grid model that calculates the power dispatched from every generating unit that is needed to meet the electricity demand at every hour of the year, calling on the most efficient units first, then next best, until power requirements are satisfied. Its plant-by-plant details create too many variables for an integrated planning model to solve, he said.
The Kansas Corporation Commission, in its filing with EPA, found the same faults as Hewson did, saying the EPA model cannot provide a verdict on future grid reliability because it lacks detailed weather data affecting demand and does not map local power transmission bottlenecks and data on individual generating plants or realistic impacts of intermittent wind power supply. "IPM cannot even estimate the cost and reliability impact on the grid of simply significant reducing the use of a coal plant," the Kansas filing said.
What EPA's model meant to do
EPA responds that its model results show that the CPP goals can be met, if state and regional grid officials continue to carry out the grid management decisions they are charged with performing.
"It's important to be clear about what our modeling does and what it wasn't even setting out to do," said an EPA official, who was not authorized to be quoted by name. The scope of the modeling analysis provides a broad, reasonable picture of likely outcomes one or two decades ahead based on different scenarios, he said.
"In all of those instances ... we saw the system coming back and successfully complying with state goals while respecting reserve margins and not violating the bulk power transfer constraints" -- the physical limitations on power shipments from one region to another.
"This doesn't necessarily reach the different type of modeling exercise that the reliability authorities do all the time, that tends to be very granular," the official added. Grid managers need to understand the consequences of a particular coal plant retirement or new transmission expansion, and have to model power flows locally. "There is no way to do that type of modeling at the level we are investigating here," the EPA official said.
"Our expectation is that ... reliability authorities would analyze those implementation details and be able to identify any particular instances where they feel they need to develop workarounds on the local level. We weren't trying to stand in the shoes of local reliability planners and basically do their business for them," the official said.
Macedonia said it is clear that even without CPP, "a significant transition is underway in the power sector and can be expected to continue in the coming years."
"Reliability is largely a timing issue," she added. "There are existing processes in place to make sure these changes won't be felt at the light switch. If there is enough time, the system can adapt and the process works.
"If you really accelerate that over too short a time frame, that's when the risk goes up," she said.