How the Government Can Avoid Subsidizing Carbon-Intensive Hydrogen
The Bipartisan Infrastructure Law and the Inflation Reduction Act allocated billions of dollars toward “clean” hydrogen. Now federal agencies need to specify what counts as clean.
One way to make hydrogen is electrolysis: using electricity to split water into hydrogen and oxygen. Whether electrolysis results in clean hydrogen depends on where the electricity comes from (e.g., wind vs. natural gas). Last week, Policy Integrity submitted two sets of comments on how to ensure that the federal government doesn’t accidentally subsidize fossil-fuel-powered electrolysis. More specifically, the comments addressed how to measure the carbon intensity of electricity that is drawn from the electric grid during hydrogen production.
Policy Integrity collaborated with WattTime — a nonprofit with expertise in calculating the carbon intensity of grid electricity — to submit comments to the Department of Energy (DOE). DOE is starting to implement the regional clean hydrogen hubs established in the Bipartisan Infrastructure Law, which means channeling $8 billion toward the development of a network of clean-hydrogen production, processing, delivery, storage, and use. By law, the program must demonstrably aid the achievement of the so-called “clean-hydrogen-production standard.” This standard is a carbon-intensity target that will be selected by DOE. DOE’s draft guidance on the standard proposes a lifecycle target of 4.0 kg CO2e/kg H2 from well to gate, which means that upstream electricity emissions would count toward the carbon intensity of hydrogen production.
Policy Integrity and WattTime commented on this draft guidance, emphasizing two key points about how DOE can accurately quantify the carbon intensity of hydrogen produced with grid electricity. By adopting the policies discussed in the comments, DOE could ensure that it funds projects that demonstrably aid the achievement of the clean-hydrogen-production standard, rather than exacerbating climate change by subsidizing carbon-intensive hydrogen.
A Marginal-Emissions Approach Is Essential
To help select the proposed clean-hydrogen-production standard, DOE used a model called the Greenhouse Gases, Regulated Emissions, and Energy Use in Technologies Model (GREET). But this model cannot adequately assess the emissions intensity of specific projects. The problem with GREET is that it uses an annual-average approach for relatively large regions. In other words, GREET calculates the carbon intensity of electricity for large swaths of the country by averaging the carbon intensity of all the electricity generated there for an entire year. While this approach has some uses, when you want to know the carbon intensity of the electricity powering a new electrolyzer, it makes more sense to analyze the carbon intensity of the generator that increased its output to match the electrolyzer’s demand for electricity. This superior methodology is known as the marginal-emissions approach because it spotlights the marginal generator that actually ramped up its generation.
The stakes of this methodological issue are clear if you understand the following bedrock principle of grid operation: Grid operators generally dispatch generation resources according to their marginal costs. So, the first resources that a grid operator will rely on to meet demand are those that generate cheap electricity after they’ve been built, like solar, wind, hydro, and nuclear. Only when the output of these resources isn’t enough to satisfy demand, will the grid operator call on resources with higher marginal costs, like coal and oil.
Thus, an annual-average approach would obscure the true emissions intensity of using grid electricity to power an electrolyzer. If an electrolyzer were located, for example, in the Pacific Northwest — where hydroelectric generation is abundant and thus the average carbon-intensity of the grid is low — an annual-average approach might indicate that using grid electricity causes few emissions because so much of the energy produced there comes from hydropower. But the real effect of new load from an electrolyzer may be significantly different. Hydropower has a low marginal cost, so, when demand outstrips the electricity from hydropower, the additional load from the electrolyzer would likely be met by additional generation at a coal or natural gas plant.
Our comments to DOE underscored not only the necessity of applying a marginal-emissions approach, but also the need to perform this analysis with temporal granularity. The two figures below were generated by WattTime, and they depict how marginal-emissions rates oscillate throughout the day in the California Independent System Operator and the Southwest Power Pool. Thus, DOE should embrace a marginal-emissions approach that reflects the hourly and sub-hourly changes in the carbon intensity of the marginal resource on the grid.
And — because there are many different grid operators and because transmission constraints exist within the footprints of individual grid operators — the comments also emphasized that the marginal plant varies with location. The following figure, also created by WattTime, depicts the spatial variation in marginal-emissions rates at a representative moment in time. DOE will miscalculate the carbon intensity of hydrogen produced with grid electricity unless the agency properly identifies the actual marginal plant in light of grid-operation boundaries and transmission constraints.
If DOE were to calculate the carbon intensity of hydrogen using marginal-emissions rates with temporal and spatial granularity, the agency would incentivize hydrogen producers to locate in regions with curtailment of renewable resources (i.e., periods when renewable generation exceeds demand for electricity) and to produce during those periods of curtailment. During curtailment, the additional carbon emissions from the additional use of grid electricity is zero because the marginal generator is renewable. Further, because curtailed energy would have otherwise gone to waste, a granular marginal-emissions approach would help ensure an efficient use of resources.
Our comments also highlighted the feasibility of a marginal-emissions approach. Marginal-emissions data (or marginal-fuel data, which are closely related) are available from organizations like WattTime and from ISOs/RTOs like PJM Interconnection, ISO New England, Southwest Power Pool, and the Midcontinent Independent System Operator. Additionally, pursuant to the Bipartisan Infrastructure Law, the Energy Information Administration is in the process of releasing real-time or near-real-time marginal-emissions data.
RECs and Other Market Instruments Shouldn’t Undermine Clean-Hydrogen Standards
If a hydrogen producer purchases renewable energy credits (RECs) from a renewable generator, should this be allowed to reduce the effective carbon intensity of the grid electricity it uses? Alternatively, if a hydrogen producer enters into a power-purchase agreement (PPA) with a renewable generator, should it be able to assert that the power used for electrolysis is therefore clean?
In either scenario, DOE should require the electrolyzer to show that the clean energy associated with a given market instrument was additional to the grid, as opposed to electricity that was always going to be generated and used by some other consumer. The important question is whether the clean generation would have occurred regardless of the REC payments from the electrolyzer or the PPA. Additionality is a slippery concept, but if it isn’t satisfied, an instrument like a REC or PPA might represent the mere reshuffling of the allocation of electricity on paper, not a genuine offset.
If additionality has been satisfied, further carbon-accounting principles should apply to determine how the instrument affects the net carbon intensity of hydrogen produced from grid electricity. For RECs and virtual/financial PPAs, DOE should first calculate the true carbon intensity of the grid electricity used for electrolysis using the marginal-emissions approach discussed above. Then the same marginal-emissions methodology should be used to quantify the avoided emissions associated with the instrument. In mathematical terms, the avoided-emissions value of that instrument would be the product of (a) the amount of renewable generation and (b) the emissions intensity of the marginal resource when/where the renewable generation occurred. If a clean generator produced when the marginal generator in its region was a fossil-fuel resource, then the clean generator displaced emissions by causing the fossil-fuel plant to ramp down. In contrast, there are no avoided emissions if the clean generator produced when the marginal generator was also renewable.
Different principles apply to physical PPAs. Assuming additionality has been met, clean power that is physically delivered and used by the electrolyzer (pursuant to a physical PPA) within a single region at the time of hydrogen production would have an emissions intensity of zero. But if a clean generator cannot itself source all the power contracted for under a physical PPA, the carbon intensity of the electricity procured from third parties would depend on the resources called upon to fill the deficit. For additional energy purchased on the wholesale market, the carbon intensity would be that of the marginal plant for the region at the moment of generation.
The Treasury Department Should Make Similar Decisions on Clean Hydrogen
In addition to these comments to DOE, Policy Integrity submitted a second set of comments to the Department of Treasury (Treasury) and the Internal Revenue Service (IRS) that also addressed how to measure the carbon intensity of hydrogen produced using grid electricity. Treasury and IRS are administering the provision of the Inflation Reduction Act that provides tax credits for producing hydrogen depending on the carbon intensity of the production process, including upstream electricity emissions. Treasury and IRS requested information about how to measure the carbon intensity of hydrogen production, including how to treat instruments like RECs and PPAs. Policy Integrity’s comments to Treasury and IRS referred the agencies to our aforementioned comments to DOE on these same issues.
The Treasury/IRS comments also addressed an issue that is unique to the clean-hydrogen-production tax credits. In contrast to the hydrogen-hub provision of the Bipartisan Infrastructure Law, the clean-hydrogen-production-tax-credit provision of the Inflation Reduction Act specifically requires the emissions intensity of hydrogen production to include only those emissions determined by (a) GREET or (b) a successor model determined by the Secretary of Treasury. So, the comments advised Treasury and IRS to coordinate with DOE to promptly update GREET to adhere to a marginal-emissions approach, or to create a successor model that incorporates that approach.
