A version of this analysis was submitted as a public comment to the Environmental Protection Agency.
In April of this year, the Environmental Protection Agency released a draft proposal for regulating carbon emissions from the U.S. electric power sector. Our analysis of the EPA’s Regulatory Impact Analysis (RIA) shows that, with the Inflation Reduction Act passed, the agency has already baked in a dramatic fall in nationwide emissions without any further action from the EPA—as long as the nation’s transmission network, long stifled by bureaucratic and permitting obstacles, can grow fast enough to accommodate growth in clean energy capacity.
In reality, the new EPA rule yields relatively small additional emissions cuts, and the rule in turn depends on large cost reductions in emissions control technologies like carbon capture and hydrogen co-firing that are hardly assured. In other words, the EPA itself provides compelling evidence that deregulating clean energy deployment will deliver far more emissions reductions than directly regulating emissions themselves.
As others have observed, the agency justifies new standards on fossil thermal power plants based on the assumed cost-effectiveness and widespread availability of both post-combustion carbon capture technologies and hydrogen co-firing with natural gas combustion by the early-to-mid 2030s. These two technologies have yet to achieve significant scale anywhere in the world today. If these “inside-the-fenceline” carbon mitigation options do not prove viable by the 2030s, analysts generally expect utilities to comply with the new EPA rules via “outside-the-fenceline” actions, like shutting down coal and gas plants outright and building low-carbon capacity to replace them.
But as our analysis finds, the vast majority of future nationwide emissions reductions modeled by the EPA already occur in the agency’s “post-IRA business-as-usual” reference scenario.
Specifically, the EPA’s reference scenario achieves an over 1,000 million-ton decline in annual power-sector CO2 emissions between 2022 and 2035, while the proposed rule would reduce emissions by just an additional 36 million tons of CO2 (Mt CO2) per year in the year 2035. As other modeling has shown, the large reductions built into the EPA’s reference scenario themselves depend on the administration’s assumptions about clean energy deployment and expansion of the U.S. electricity transmission network. The Princeton REPEAT model shows that unconstrained transmission growth would reduce emissions by approximately 300 Mt CO2/yr more than a scenario in which transmission is limited to historic rates of deployment–a difference that is an order of magnitude larger in scale than the EPA’s proposed rule.
By the same token, if the EPA’s assumed rates of clean energy deployment do not materialize in line with the reference scenario’s expectations, the power plant regulations could impose significantly higher economic costs. Meanwhile, it is likely that the RIA underestimates the overall costs of those regulations, given that the proposal assumes cost improvements in carbon removal technologies that are likely unrealistic, along with implausible costs and technical feasibility for hydrogen co-firing in thermal power plants.
Advocates lobbying to make the EPA’s draft power plant rule even more stringent would do more to advance decarbonization by refocusing their attention on reforms to the nation’s transmission siting and broader permitting rules, cost reductions, and commercialization in carbon removal technologies, and targeting incentives for hydrogen demand in higher-value sectors.
Direct carbon emissions across the nation’s power sectortotaled 1,680 Mt CO2 in 2022. Compared to this present-day figure, the EPA’s reference case for their proposed power plant rule expects power-sector emissions to drop to 608 Mt CO2 by 2035. The EPA then expects the proposed regulations to cut 2035 power-sector emissions marginally further to 572 million tons CO2, a reduction of little more than 36 million tons CO2.
Almost all the emissions reductions forecast by the EPA come from clean energy and transmission deployment already leveraged in the models’ reference scenario, with the rule itself motivating relatively little added mitigation. Most of this expected progress thus stems from the more than tripling of non-hydroelectric renewable generation the EPA forecasts, from ~650 TWh/yr in 2022 to 2,180 TWh/yr in 2035 in the reference scenario. As the figure below shows, this growth of renewables and the accompanying decline in coal generation—and the underlying modeling assumptions driving both shifts—dwarf the effect of the EPA’s proposed rule.
The EPA’s assumptions about future national electric power transmission are particularly critical. The EPA model deploys as much transmission as necessary “to solve for the optimal mix of generation and transmission additions to meet capacity and energy needs.” By contrast, the Princeton REPEAT Project modeled more realistic scenarios, in which transmission expansion is constrained at a variety of levels, including the historical rate of ~1% per year. In the 1% per year scenario, net emissions in 2035 decline to ~900 million tons CO2, about twice the emissions in the team’s unconstrained scenario and about 300 million tons CO2 higher than the EPA’s reference scenario.
The EPA reference emission levels (608 Mt CO2/yr in 2035) most closely resemble that of REPEAT’s more ambitious scenario (~600 Mt CO2/yr in 2035), a case that assumes expansion of the transmission network by ~1.5%/yr. This represents around a 50% improvement in the recent historical annual rate of transmission growth. Meeting or exceeding this rate of transmission deployment will likely require reforms to FERC and to the broader infrastructure siting and permitting regime in the United States.
Other assumptions in the EPA’s reference case may also require closer scrutiny. For instance, the EPA’s modeling exogenously assumed the future delivered cost of clean hydrogen to be $0.5-$1 per kilogram. This includes existing hydrogen subsidies, thereby further assuming that most of the benefits of the IRA’s hydrogen production tax credit pass on to the end user, in this case a methane-fired power plant modified to burn some fraction of clean hydrogen. Given that current low-carbon hydrogen costs over $15/kg today, and given that the IRA’s hydrogen tax credits may begin to expire for some producers at around the same time that the EPA’s proposed rules start to take effect in the mid-2030s, one wonders whether a power plant operator could still cost-effectively consider hydrogen co-firing under an alternative scenario where hydrogen costs have yet to decline to such low levels.
But even if cheap, green hydrogen is widely available by the beginning of the 2030s, using it to co-fire natural gas plants would be to misuse an otherwise valuable resource. It is not at all clear that hydrogen blending for power generation is technologically feasible at the scale modeled by the EPA. Even very low fractions (>5%) of hydrogen co-blending in existing gas pipelines can raise the risk of steel embrittlement, hydrogen and methane leakage, and compromised integrity of transmission pipelines and gas storage tank infrastructure, according to a recent analysis prepared by University of California, Riverside researchers for the California Public Utilities Commission. Even if 30% co-blending of hydrogen in existing gas infrastructure proves feasible without major modifications, current infrastructure may not be able to safely support subsequent increases in hydrogen co-firing to 96% by 2038 as envisioned by the current proposed rule.
The volumes of hydrogen envisioned in the RIA, if truly sourced from a low-carbon supply chain, could serve far more valuable decarbonization uses elsewhere in the economy, such as in fuel cell-powered heavy trucks, fertilizer, steel, or chemicals production. Such hard-to-decarbonize sectors may possess few clean alternatives other than hydrogen, in contrast to gas-fired electricity generation which possesses a wide variety of competing clean technologies.
As the figure below shows, RIA also assumes rather heroic cost figures for CCS retrofits on coal and gas plants, with per-ton costs of capture one-half to two-thirds lower than what has been documented at the few power plant CCS projects in North America to date. The data sources used to calculate CCS costs rely largely on literature estimates, including only a very limited pool of “actual as-spent costs for CO2 capture projects,” which, given the industry’s nascency, provide only a limited sample for forecasting future costs. Most of the large cost reductions assumed for coal and NGCC carbon capture retrofits come from preliminary estimates from “feasibility and FEED studies.” The EPA essentially assumes technology learning will occur prior to technology deployment, which in the EPA’s model does not begin before the 2030s.
If transmission and renewables deployment do not keep pace with the EPA’s assumptions, then significantly more unmitigated coal and natural gas capacity could remain online in 2035, thus falling subject to regulations forcing them to adopt as-yet unproven carbon capture and/or hydrogen co-firing technologies. Perhaps these technologies will become much more widely available and affordable in a decade. Perhaps Congressional and state-level permitting reform will accelerate transmission and clean energy deployment at a pace commensurate with the EPA’s reference scenario.
But regulators and climate advocates should recognize these uncertainties as central, not tangential, to the EPA’s proposed rule and to broader efforts to rein in power-sector emissions. Successful national decarbonization will depend far more upon building transmission, clean energy projects, and supply chain infrastructure at vast scales, than upon stricter regulations targeting power plant emissions.