#Hydrogen and Hydrogen-Derived Fuels

The production and use of hydrogen for both non-energy applications and potential use as a fuel are modeled explicitly in US-REGEN. Currently, hydrogen is produced and used within the chemical and petroleum refining industries as a feedstock for industrial processes (i.e., not for energy). In the chemical industry, its primary use is as intermediate input to ammonia production for fertilizer, and in petroleum refining, it is used to reduce the sulfur content of liquid fuels in a process known as hydrotreating. Nearly all current production of hydrogen in the U.S. is from natural gas via steam methane reforming. Although only a very small fraction of hydrogen is used for energy today, mainly in nascent hydrogen fuel cell vehicle (HFCV) applications, it could potentially be used more broadly in the economy as an end-use fuel or input to the production of other fuels. Like electricity, hydrogen is an energy carrier requiring conversion from a primary energy source. In decarbonization scenarios, it is potentially valuable as a zero-carbon alternative to conventional non-electric end-use fuels. However, its production, storage, and delivery entail additional costs and energy losses, and its potential deployment must be evaluated against other low-carbon alternatives.

In the model’s base year of 2020, the total scale of hydrogen supply and demand, including hydrogen produced as an intermediate step in the manufacture of ammonia (via the Haber-Bosch process), is around 10 MtH₂, or roughly 1.1 quad Btu in energy terms, all produce from natural using steam methane reforming (SMR).^[1] Hydrogen consumption for petroleum refining is round 0.5 quad Btu (based on EIA data), while hydrogen inputs to ammonia production are around 0.3 quad Btu (based on ammonia production data from the USGS commodity survey and the model’s representation of the Haber-Bosch process energy and feedstock requirements). The remaining demand is assumed to be non-energy use as an industrial gas for other chemical products such as methanol. Non-energy demand for hydrogen is projected in future scenarios based on the scale of these respective activities: petroleum refining evolves based on the demand for refined products (endogenous to the model scenario), while demand for ammonia as fertilizer and other chemicals is projected based on exogenous growth trends for these industries. The model also includes the potential for price-responsive improvements in the efficiency of industrial use of hydrogen, which is endogenous to the model scenario.

Potential energy uses for hydrogen include both direct end-use applications and intermediate or indirect uses. End-use applications include fuel cell vehicles in on-road and non-road segments in transportation and industry, process heat in industry, and space and water heating in buildings. Intermediate uses include blending with natural gas in existing gas infrastructure, power generation, and inputs to synthetic fuel production (potentially including the production of ammonia for use as a fuel). The deployment of hydrogen technologies in each of these applications depends on the trade-offs with competing alternatives. Blending of hydrogen in gas pipelines is assumed to be technically feasible without significant modifications to pipeline infrastructure or end-use equipment up to 20% by volume, which translates to roughly 7% on an energy basis.^[2] In all future projection scenarios, there is some hydrogen use as a fuel, increasing the scale of hydrogen activity. Figure 27 summarizes hydrogen supply and demand in 2050 across scenarios and compared to today.

In the Reference scenario, there is some new hydrogen demand from the deployment of fuel cells in non-road vehicles in agriculture, construction, and mining (represented elsewhere as non-manufacturing industrial energy use). Based on assumed costs and efficiencies, as well as assumed additional costs of electrification in these non-road segments relative to comparable on-road vehicles, hydrogen fuel cell vehicles (HFCVs) are competitive with both internal combustion engine vehicles (ICEVs) and battery electric vehicles (BEVs). There is also some deployment of HFCVs in medium- and heavy-duty (MD/HD) on-road segments, but this is more limited due to more favorable costs of BEVs and charging infrastructure. Most production remains based on conventional SMR (without carbon capture), although some new SMR capacity with carbon capture and storage is deployed.^[3] SMR with CCS deployment is driven by existing state-level decarbonization policies included in the Reference scenario, for example, in California and New York. Average production costs are similar to today at around $1/kg.

In the Net-Zero All Options scenario, competition from biofuels means there is relatively little incremental demand (relative to the Reference scenario) for hydrogen used directly as an end-use fuel. However, the net-zero target for emissions makes hydrogen blending into the gas network cost-effective, up to the 7% limit. Because economy-wide gas demand remains relatively high in this scenario, this blend share represents a large increase in hydrogen demand. Most hydrogen production shifts to gas with CCS, with about a third supplied from biomass gasification with CCS (which contributes a negative emissions flow).

In the Net-Zero Higher Fuel Cost scenario, biofuels and gas are more expensive, which increases direct demand for hydrogen but decreases demand for pipeline gas; thus, the quantity of hydrogen blended into the natural gas system is lower in absolute terms as well. These two effects offset each other, resulting in similar levels of hydrogen demand to the Net-Zero All Options case. Higher gas supply costs result in increased market share for electrolysis, even as the total volume of hydrogen is similar. The model includes several electrolysis technologies, including a polymer electrolyte membrane (PEM) technology and a high-temperature solid oxide electrolysis technology that has a higher capital cost and requires a thermal input but operates at higher electrical efficiency. Because of the difference in cost structure, both technologies are deployed, with PEM electrolysis operating at lower capacity factors (and output aligned with renewables) and high-temperature electrolysis operating at higher capacity factors and paired with nuclear for the thermal input.

In the Net-Zero Limited Options scenario, the absence of CCS means that negative emissions are much more limited in scale (only natural climate solutions can contribute), so positive emissions must be nearly eliminated. This leads to increased electrification and efficiency, as well as stronger demand for hydrogen used directly as an end-use fuel compared to the other net-zero cases. Because demand for pipeline gas is much lower, the scale of hydrogen blended into the natural gas system is much smaller in absolute terms. However, there is now increased demand for hydrogen as an input to synthesis of ammonia^[4] and other fuels, such as synthetic jet fuel or natural gas produced by combining hydrogen with captured carbon. These fuels have high realized levelized costs in this scenario but enter the market because of the need to minimize fossil use. Because CCS is unavailable, all hydrogen production is from electrolysis. There is also a small amount of hydrogen used for power generation, which amounts to a storage pathway as all hydrogen is originally produced from electricity. While the energy flow is small, the hydrogen generation capacity is larger in relative terms, i.e., a small amount of hydrogen is used to provide critical balancing of intermittent renewables (see Figure 22 and Figure 25). The scale of deployment of electrolysis in this scenario is much larger than hydrogen production capacity in the other scenarios because it is operated at a lower capacity factor (based on the model’s optimization, which balances capital costs with the incremental costs of electricity supply). Figure 28 shows hydrogen production capacity across scenarios. Total nominal electrolysis capacity in the Net-Zero Limited Options scenario is around 650,000 tH₂/day, which translates to around 1,200 GW-e in terms of electric input. In this scenario, the average annual capacity factor for PEM electrolysis was around 36% (with a range across regions of 24%–45%). In comparison, high-temperature electrolysis was dispatched more regularly with a capacity factor of around 77%.

#Hydrogen Infrastructure

With the increased use of hydrogen in a broader range of applications comes the need for additional infrastructure for storage and delivery. Using hydrogen directly as a fuel in end-uses (such as fuel cell vehicles or industrial process heat) requires either distributed scale production or delivery of centrally produced hydrogen via new or re-purposed infrastructure designed for pure hydrogen. These scenarios evaluate the decision to adopt hydrogen at the end-use based on the availability of a pipeline delivery network with an assumed levelized cost that varies by sector (see us-regen-docs.epri.comopen in new window). The assumed costs are based on both bottom-up estimates of hydrogen pipeline costs as well as a top-down comparison with the levelized costs of natural gas delivery infrastructure. In general, the levelized costs of hydrogen delivery by pipeline are assumed to be roughly twice the levelized cost of natural gas delivery for the same type of application, based on its lower energy density and requirements for higher pressure and more expensive materials. While the model evaluates hydrogen delivery infrastructure costs on a levelized basis, in reality, these costs would take the form of significant up-front investments. The levelized costs assumed here would only be realized with a sufficient scale of deployment. Total direct delivery of hydrogen to end-use sectors (primarily industry and transportation) ranges from roughly 1 quad Btu in the Reference scenario to nearly 3 quad Btu in the Net-Zero Limited Options scenario by 2050. The corresponding annualized delivery infrastructure costs ranges from roughly $8 billion to $24 billion, implying total up-front investments roughly an order of magnitude higher.

For applications where hydrogen is used as an intermediate input to other energy conversions, the delivery infrastructure requirements may be less significant due to opportunities to co-locate linked activities and/or locate facilities near transmission trunk lines. For the purposes of the scenarios in this study, the delivery costs for hydrogen used for power generation and synthesis of ammonia and other fuels are assumed to be negligible. A more realistic interpretation would be that the costs of supporting hydrogen delivery infrastructure for large-scale energy sector hydrogen use should be included as an additional component of the capital costs for the hydrogen-using plant. However, these additional costs may be small relative to the cost of the plant itself.

Hydrogen storage infrastructure investments are also included explicitly in the modeling. Storage requirements are particularly salient with respect to the production of hydrogen from electrolysis, which is intermittent and seasonal, following the output profiles of wind and solar resources (see Figure 25). This is true even when the production of hydrogen is primarily to meet end-use demands, as in these scenarios. The profile of demand over a year for industrial and transportation applications is assumed to be more or less constant,^[5] so bulk storage is needed to reconcile seasonal variation in electrolysis output. Hydrogen storage may also be needed in scenarios where hydrogen is produced from natural gas or bioenergy, but production, in this case, would also be more or less constant over the year, so that supply and demand mismatches on a weekly or daily basis would be much smaller. The total deployment of bulk hydrogen storage across scenarios is shown in Figure 28. Storage requirements in the scenarios with production from electrolysis are many times higher than current hydrogen storage and projected capacity in the scenarios with non-electric production. Because daily and weekly production can significantly exceed demand, storage injection capacity needs to be almost half of production capacity in the Net-Zero Limited Options scenario, with a reservoir size of around 23 days (based on nominal injection capacity). Note that the cost assumptions for hydrogen storage are based on underground salt cavern formations, which may not be present in all regions. Further research will develop cost assumptions and availability constraints for underground storage in other types of geology.