Onsite production of gigawatt-scale wind- and solar-sourced hydrogen (H2) at industrial locations depends on the ability to store and deliver otherwise-curtailed H2 during times of power shortages. Thousands of tonnes of H2 will require storage in regions where subsurface storage is scarce, which may only be possible using liquid organic H2 carriers. We evaluate aboveground system with a focus on providing technical insights into toluene/methylcyclohexane (TOL/MCH) storage systems in locations suitable for gigawatt-scale wind- and solar-powered electrolyzer systems in the United States. Here we show that the levelized cost of storage, at a national median of US dollar $1.84/kg-H2 is spatially heterogeneous, causing minor impact on the cost of H2 supply in the Midwest, and significant impact in Central California and the Southeast. While TOL/MCH may be the cheapest aboveground bulk storage solution evaluated, upfront capital costs, modest energy efficiency, reliance on critical materials and pre-sulfided catalysts, and greenhouse gas emissions from heating are opportunities for further development.
The distribution in LCOS is centered around a median of $1.84/kg-H2, with a minimum of $1.31/kg-H2 and maximum of $2.68/kg-H2 at the national level (Fig. 3).
national simulations support the distribution under a common process model without independent operational validation
Based on the locations evaluated, significant amounts of H2 storage (1457–19,837 tonnes) are necessary to smooth annual supply from the electrolyzers
annual-profile sizing supports the modeled range without establishing interannual supply reliability
When compared with other large-scale H2 storage technologies (Supplementary Fig. 7), we find the carrier system [TOL/MCH] outcompetes gaseous buried pipes and cryogenic liquid H2, but not geologic solutions such as LRC and salt caverns.
modeled comparisons support the ordering, with detailed cross-technology assumptions unavailable for assessment
At this location [Southwest Texas] about one quarter of H2 production required storage, and the resulting ACEU would be $0.54/kg-H2.
case-study annual accounting supports the added cost under the specified process and financial assumptions
Derived from the full evaluation — not a separate score.
Strengths
The analysis evaluates 50,052 locations and shows how renewable timing changes storage requirements and costs, rather than relying on annual energy totals alone.
↳ Results, Performance for Storing Otherwise-Curtailed H2; Methods, Supply Chain Model
Sizing and financial equations provide an inspectable chain from hydrogen production to LCOS and added end-user cost. The scaling surrogate is reported to differ from detailed calculations by no more than $0.03/kg-H2 within the evaluated locations.
↳ Methods, Eqs. 1–7; Methods, Scaling Equations Development
The Discussion connects natural-gas heating, moderate energy efficiency, catalyst dependence, and capital expenditure to the conditions for industrial use and further development.
↳ Discussion; Methods, MCH/TOL Process Model
Limitations
The sizing and cost estimates depend on reactor kinetics and load-following behavior that have not been demonstrated for the proposed configuration. The Methods identify intrinsic rates and subsequent dynamic simulation as necessary development steps.
↳ Methods, MCH/TOL Process Model; Discussion
Storage sizing is derived from an annual hourly sequence without demonstrated interannual weather robustness. This limits interpretation of the reported capacities as guarantees of reliable supply.
↳ Methods, Supply Chain Model; Results, Storage Design and Key Assumptions
The five wind and six solar capacities imply 30 combinations rather than the stated 25. Table 1's 211,444 tonnes/year of dehydrogenation toluene consumption also conflicts with the reported cost scale, although the scaling equation supports the stated LCOS.
↳ Methods, Supply Chain Model; Table 1; Methods, Eq. 7
The national hourly analysis and explicit process-to-cost equations make this a useful engineering-screening contribution. Its storage requirements and costs remain conditional on uncertain kinetics, assumed transient behavior, and annual renewable profiles rather than demonstrated operation. The Texas emissions values use different stored- and delivered-hydrogen denominators, so their numerical difference alone does not establish a carbon-accounting error, although the reconciliation should be explained. The sweep-count and toluene-consumption entries require correction or clarification without presently overturning the reconstructed storage-cost result.
Nabu’s assessment, alongside the field’s view.
Are you an author of this paper?
Sound3.7
Confidence mediumThe contribution is the integration of hourly renewable profiles, stationary TOL/MCH process design, and national storage-cost screening. Its advance lies in system-level sizing and reusable equations rather than new carrier chemistry or demonstrated operation.
Explicit sizing equations, process balances, and a checked scaling surrogate support the merit score. Unvalidated reactor dynamics and uncertain kinetics constrain absolute estimates, while the sweep-count and toluene-entry findings require reporting clarification.
The paper provides a followable progression from production profiles to storage size and industrial cost. Unexplained reconciliation of stored-versus-delivered emissions, shifting terminology, and optimistic target language reduce precision without establishing the alleged emissions contradiction.
The paper compares relevant storage alternatives and connects energy use, catalyst requirements, capital cost, and emissions to interpretation. It does not establish robustness across weather years or systematically propagate kinetic uncertainty into costs.
Caveats4 of 4 checks
2 reported values do not reconcile, but the key claims do not rest on them.
No reporting or research-conduct inconsistency is established. The non-public basis for the shelf-life assumption limits independent assessment of that assumption rather than demonstrating misconduct or a reporting conflict. One or more identifiers named in the paper's availability or registration statements did not resolve when checked on 2026-10-28T16:20:42.716183+00:00: https://github.com/NREL/HOPP/tree/dev/refactor/greenheart. Recorded as a discrepancy between the paper's claims and the cited sources; not an assessment of the underlying research.
38 references in manuscript 29 of 30 checkable references found in an index 8 have no canonical index record — counted, but not index-checkable 1 reference confirmed by manual review
References not verified automatically are resolved manually within 24 hours.
No retraction or expression of concern found in Crossref (includes Retraction Watch).
Sources: Crossref (incl. Retraction Watch) ✓
Where this paper’s evidence sits on the path from initial observation to real-world use.
A specified process design and financial model move the work beyond a conceptual proposal. Physical operation, transient controls, catalyst scale-up, and heat integration remain to be validated.
AI-generated, human-governed. Something look off? Contact us to request a review.
Keep reading
Other papers Nabu has evaluated blind, starting with the same field.
Thaha, Ghamari, Jothiprakash, et al. · 2025 · Chemical Engineering
Hamed Nilforoshan, Julia Reisler, Erfan Jahanparast, et al. · 2026 · Health sciences
the test-based design resulted in 83.5% fewer symptoms being attributable to COVID-19 between 30 to 119 d after infection (43 attributable symptoms vs. 262 for the unrestricted control design)
TentativeDominikus Brian, Xavier Isidro Pereira Hernández, Hao Yin, et al. · 2026 · Computer and information sciences
Our findings suggest that LLMs can identify methodological errors in reports of randomized clinical trials that human peer review may miss.
Tentative