Green hydrogen has a tools problem. The physics of splitting water with solar power is well understood, the electrochemistry of electrolyzers is textbook material, and the finance of multi-hundred-million-dollar projects follows standard discounted cash flow rules. Yet when a project developer in Morocco, Chile or Namibia wants to know whether a proposed plant will actually work — hour by hour, through the night, across the seasons — the software landscape fragments. One tool models photovoltaic yield, another prices hydrogen, a third handles microgrid dispatch, and none of them, in a single workflow, connects the choice of a GPS coordinate to the thickness of a steel pressure vessel wall. A team of Moroccan and French researchers now reports a platform built to close that gap.
The platform, called H2CHAIN and described in the journal Results in Engineering, is a browser-based planning environment that walks a user from site screening to a full probability distribution of production costs without a single line of code. Its authors, led by M. Nachtane, position it against a crowded field of established tools — NREL’s H2FAST and H2Integrate, the SAM renewable yield model, HOMER Pro, GeoH2, HyRAM+ — each of which is mature within its own narrow scope. What none of them offers, the team argues, is the full sequence in one place: geospatial screening, hourly solar and wind generation, electrolyzer performance, mechanically rigorous storage sizing, and chain-level economics, all gated by an enforced energy balance that rejects impossible configurations before a single cost figure is computed.
The technical core is deliberately conservative. Photovoltaic output is modeled from first-principles cell physics, tracking how cell temperature shifts with irradiance and ambient conditions, then correcting the maximum-power-point current and voltage accordingly. Wind turbines follow a piecewise power curve between cut-in, rated and cut-out speeds. The electrolyzer module implements the Ulleberg parameterization of alkaline water electrolysis, in which cell voltage depends on current density and operating temperature through dimensionally consistent coefficient groups. For the reference machine — 24 parallel stacks of 376 cells each, running at 80 degrees Celsius and 3,500 amperes per square meter — the model yields a cell voltage of 1.829 volts and a system-level energy consumption of 52.0 kilowatt-hours per kilogram of hydrogen, once a 98 percent rectifier and a 2.6 percent balance-of-plant overhead are folded in. That translates to a system efficiency of 75.8 percent on the higher heating value, a figure the authors state explicitly to avoid the HHV-versus-LHV ambiguity that plagues the literature.
Storage is where H2CHAIN departs most visibly from cost-correlation-based competitors. Rather than assuming a price per kilogram of stored hydrogen, the platform sizes pressure vessels from Lamé thick-wall elasticity theory. This matters because at the 700-bar pressures typical of hydrogen service, the thin-wall membrane equations that most tools quietly rely on fall outside their validity: the platform’s own comparison shows the membrane relation underpredicts bore hoop stress by more than 10 percent once the radius-to-thickness ratio drops below about 5.4, and by 16 percent at a ratio of 3.5. The software checks the ratio automatically, switches to the thick-wall solution when required, and evaluates both the proof condition at 1.5 times service pressure and the burst condition at 2.25 times, following the safety factors of ISO 11439 and ASME Section VIII Division 3. The resulting steel mass drives the storage capital cost — and, as the case study shows, that choice can move a project’s economics by seven percent.
To demonstrate the workflow end to end, the team sized a hypothetical 100-tonne-per-day green methanol plant at Dakhla on Morocco’s southern Atlantic coast, a site with an annual mean horizontal irradiance of about 6 kilowatt-hours per square meter per day and a prominent role in the national hydrogen program. Tilting the array onto horizontal single-axis trackers lifts the average plane-of-array irradiance to 7.99 kilowatt-hours per square meter per day, a tracking gain of 32 percent. Feeding the synthesis loop’s appetite for hydrogen — 20 tonnes per day once realistic excess factors over stoichiometry are applied — the platform scales the array to 181 megawatts-peak and the alkaline electrolyzer to 151 megawatts nameplate. Because the plant runs on photovoltaics alone, the electrolyzer turns over at just 27.9 percent of the year, roughly 2,446 full-load hours, producing on average 19.5 tonnes of hydrogen per day.
The intermittency is bridged by a fleet of 95 Type I steel pressure vessels at 250 bar, holding 20.3 tonnes of deliverable hydrogen — about 24 hours of backup at the reactor’s consumption rate. Under the platform’s conservative default, which applies both the 2.25 burst multiplier and a separate material safety factor, the governing wall thickness comes to 123.9 millimeters at a bore hoop stress of 306.7 megapascals. The authors are unusually candid about the alternative: under a code-consistent reading in which the 2.25 ratio is itself the margin on ultimate strength, the same vessels need only 51.8-millimeter walls, the storage bank sheds more than half its steel, and the levelized cost of methanol falls from 2,057 to 1,915 dollars per tonne. Any reported vessel design, they conclude, must declare its safety-factor interpretation — because the two readings differ by more than a factor of two in steel mass.
The balance sheets cascade from there. The synthesis loop consumes 51,496 tonnes of carbon dioxide annually, delivered as a concentrated stream at 35 dollars per tonne, and produces 35,515 tonnes of methanol at 97.3 percent availability. Electrolysis demands about 71,000 cubic meters of demineralized water per year, which in water-scarce Dakhla means seawater reverse osmosis: 157,842 cubic meters of intake, 86,813 cubic meters of brine discharge, and a modest 0.55 gigawatt-hours of desalination electricity. Oxygen co-produced by electrolysis — 56,369 tonnes per year — goes uncredited, and the reaction water that could theoretically recycle back to the electrolyzer is likewise ignored, keeping the accounting conservative.
The headline economics are sobering. The installed capital of the whole chain reaches 546.6 million dollars after an 18 percent engineering and contingency allowance, with the photovoltaic plant at 144.7 million and the electrolyzer at 189.9 million dominating the equipment list. Annualized over 25 years at an 8 percent real cost of capital, the project costs 73 million dollars per year to own and operate, of which 70 percent is capital recovery — the signature of any system whose marginal energy input, sunlight, is free. The resulting levelized cost of hydrogen is 8.80 dollars per kilogram and the levelized cost of methanol 2,057 dollars per tonne, against fossil methanol production costs of 100 to 250 dollars per tonne. The platform does not sugar-coat this: the configuration is feasible on the hourly hydrogen balance but uncompetitive, with a gap at the base case of roughly 1,700 dollars per tonne.
What distinguishes the study is how that uncertainty is framed. A 40,000-sample Monte Carlo ensemble, seeded for reproducibility, propagates six parameters — electrolyzer and photovoltaic capital costs, cost of capital, fixed operations, specific consumption, and capacity factor — through the full sizing chain rather than a linear approximation around the base case. The distributions come out right-skewed: the median hydrogen cost of 9.04 dollars per kilogram sits 2.7 percent above the deterministic figure, no sample falls below 5.04 dollars, and the maximum reaches 14.80 dollars when an expensive cost of capital coincides with expensive electrolyzers. Critically, when the authors re-ran the ensemble with photovoltaic and electrolyzer capital costs fully rank-correlated — acknowledging that equipment prices respond to common supply-chain drivers — the median barely moved but the uncertainty interval widened appreciably, from 7.51–10.78 to 7.20–11.09 dollars per kilogram. Assuming independence, they note, understates the risk precisely when the two largest capital items move together.
The authors are equally forthright about what the platform is not. The results are verified — checked against closed-form solutions, dimensional homogeneity and mass and energy balance closure — but not validated against operating plant data, and every output should be read as a planning-level estimate rather than engineering guidance. Fatigue, hydrogen embrittlement, permeation and code certification lie outside the vessel model; part-load electrolyzer behavior, grid tariffs, batteries and kinetic reactor modeling are absent; and the tool evaluates user-specified configurations sequentially rather than searching for an optimum. A conversational assistant built on large language models and retrieval-augmented generation helps non-specialists navigate the interface, but it is deliberately decoupled from the numerical core and touches no computed result. Within those limits, H2CHAIN makes a straightforward promise to developers, policymakers and financiers: before capital is committed, the full chain — from a set of coordinates to a wall thickness to a probability distribution of cost — can be stress-tested in a browser, with the energy balance enforced every hour of the year.
Subject of Research: Development of an integrated web platform for energy-balance-constrained techno-economic planning of green hydrogen and power-to-methanol value chains
Article Title: H2CHAIN: Integrated web platform for energy-balance-constrained planning of green hydrogen value chains
Article References: Nachtane, M., Zaari, A., Adeli, K., Qarssis, Y., & Faik, A. (2026). H2CHAIN: Integrated web platform for energy-balance-constrained planning of green hydrogen value chains. Results in Engineering, 32, Article 113417. https://doi.org/10.1016/j.rineng.2026.113417
Image Credits: AI Generated
DOI: 10.1016/j.rineng.2026.113417
Keywords: green hydrogen, power-to-X, electrolysis, photovoltaics, hydrogen storage, pressure vessels, levelized cost of hydrogen, green methanol, techno-economic analysis, Monte Carlo uncertainty, Dakhla Morocco, energy modeling
