Industries / Mining and materials / Chemicals and petrochemicals / Hydrogen production
Storage
Keeping hydrogen until it is needed: squeezed into steel or carbon-fiber tanks, chilled into a liquid, or pumped into salt caverns and old gas fields underground.
5
Milestones achieved, from 1898 to 2015.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1898
Liquid hydrogen in a flask
James Dewar became the first person to make liquid hydrogen, in 1898, according to the Royal Institution in London, where he worked. He kept such very cold liquids in a vacuum flask he had made there in 1892: two glass flasks, one inside the other, sealed at the neck with a partial vacuum between them, which keeps the contents cold and slows evaporation. Evaporation, called boil-off, is still the main loss in storing liquid hydrogen (see the 1960s entry).
Credit
James Dewar, Royal Institution
1960s
Launch-pad spheres and boil-off
For the Apollo launches, NASA had a large spherical liquid hydrogen tank built at Launch Complex 39B at Kennedy Space Center in Florida. NASA's own supply-chain analysis of the 30-year Space Shuttle program found that about half the liquid hydrogen it bought was lost, through boil-off and other causes. A new sphere at the same pad, 83 feet (25 m) across and holding about 4,730 cubic meters, was completed in 2022 with a heat exchanger inside, so that a refrigerator can later be connected to stop boil-off. In an earlier NASA test on a smaller tank, every dollar spent on electricity for refrigeration saved $7 worth of liquid hydrogen.
LessonLiquid hydrogen warms and escapes, so a store pays either in lost product or in refrigeration.
Credit
NASA Kennedy Space Center; both tanks were built by CB&I (formerly Chicago Bridge & Iron)
1971
Hydrogen in salt caverns
A salt cavern is a space dissolved out of an underground bed of rock salt with water; salt is tight enough to hold gas under pressure. England's Environment Agency says the first salt caverns to store hydrogen were built at Teesside in northeast England in 1971, held hydrogen for chemical works for about 50 years, and have recently been emptied of hydrogen and filled with brine. The chemical company INEOS Inovyn, planning a larger store in Cheshire, wrote in 2019 that Teesside kept the gas at constant pressure by letting brine fill part of the cavern as hydrogen was drawn off, which greatly limited how fast hydrogen could flow.
LessonBrine-balanced caverns work but flow slowly, so new projects plan caverns whose pressure rises and falls instead.
assets.publishing.service.gov.ukassets.publishing.service.gov.uk
Credit
the Teesside chemical industry; the sources read do not name the operator
1983
Texas salt domes
ConocoPhillips began storing hydrogen in 1983 in a cavern in the Clemens salt dome under Brazoria County, Texas, with a capacity of 92 GWh, according to Engie data reported by S&P Global Platts in 2022; a 2021 presentation by the engineering firm RESPEC dates the Clemens store to 1986. Praxair has stored hydrogen at Moss Bluff in Liberty County since 2007, and in 2017 Air Liquide commissioned a cavern in the Spindletop dome, roughly a mile deep, holding more than 120 GWh, enough to back up a large steam methane reformer for 30 days (S&P Global's figures). RESPEC lists the three Texas sites as mainly used for industry, chiefly refining.
Credit
ConocoPhillips (listed by RESPEC under Chevron Phillips Chemical), Praxair and Air Liquide
2015
Carbon-fiber tanks costed
Fuel cell cars carry hydrogen as a gas at 700 bar, about 700 times the pressure of air at sea level, in Type IV tanks: a plastic liner wrapped in carbon-fiber composite. The US Department of Energy calls 350 and 700 bar tanks the near-term technology for cars and says they are sold in low volumes. A DOE record, peer reviewed by Ford, the tank maker Hexagon Lincoln and three national laboratories, found that a 700 bar system holding 5.6 kg of usable hydrogen stored 1.40 kWh per kg of system and 0.81 kWh per liter, short of DOE's 2020 targets of 1.8 and 1.3, at a projected $14.8 per kWh if 500,000 were made a year. Carbon fiber is the largest cost.
LessonCompressed tanks work in production cars but remain bulky and costly.
Credit
DOE Fuel Cell Technologies Office, with analysis by Argonne National Laboratory and Strategic Analysis
Watch for
Hydrogen in old gas fields
RAG Austria says it stored about 500,000 cubic meters of pure hydrogen in a former natural gas reservoir at Rubensdorf/Gampern, Austria, and recovered it "in high quality" for its own combined heat and power plant. In August 2026 it said a peer-reviewed paper in the International Journal of Hydrogen Energy, with co-authors from Axiom and BOKU University in Vienna, documents two storage cycles in porous sandstone. Neither release gives figures for how much hydrogen came back or how pure it was.
Achieved whenmeasured recovery and purity figures are published and checked by a party outside the project, or the store goes into commercial use.
Claimed by
RAG Austria, May 2025 and August 2026
Salt caverns at Etzel
STORAG Etzel and Gasunie say they injected about 90 tonnes of hydrogen, roughly 1 million standard cubic meters, into two converted salt caverns at Etzel near Wilhelmshaven, Germany, over a few weeks. Next they plan several cycles of injecting and withdrawing it and trials of ways to clean the hydrogen that comes back out.
Achieved whenresults of the cycles, including how much hydrogen is recovered and how pure it is, are published and confirmed by an outside party such as a regulator or a customer.
Claimed by
STORAG Etzel and Gasunie, March 2026
Open
01
Boil-off, measured not assumed
An open table of measured boil-off from real liquid hydrogen tanks, from launch-pad spheres and plant tanks to station tanks, road tankers and ship tanks, taken from test reports and papers. Each row gives the tank's volume, insulation, fill level, age and condition, the measured loss in percent per day, and the design or modeled figure for the same tank, with a link to the page it came from. Fit a simple open model of loss against tank size and insulation, and publish how far real tanks sit from it.
The briefCredit · unclaimed02
Realistic cavern potential elsewhere
A reproducible estimate of how much hydrogen could be stored in salt caverns in a second region, such as the German North Sea coast or the US Gulf Coast, after removing ground ruled out by towns, protected areas and existing industry, using the method a 2026 study applied to England's east coast. Start by reproducing that study's own figure.
The briefCredit · unclaimed
Open 01 · the brief
Boil-off, measured not assumed
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
An open table of measured boil-off from real liquid hydrogen tanks, from launch-pad spheres and plant tanks to station tanks, road tankers and ship tanks, taken from test reports and papers. Each row gives the tank's volume, insulation, fill level, age and condition, the measured loss in percent per day, and the design or modeled figure for the same tank, with a link to the page it came from. Fit a simple open model of loss against tank size and insulation, and publish how far real tanks sit from it.
Why it is still open
Published boil-off figures are mostly design values or model results. A NASA report shows how far a real tank can drift: one of Kennedy's 3,218 m³ spheres was designed to lose about 1.14 m³ of liquid a day but lost about 3.79 m³ a day because of a large gap in its perlite insulation, and dropped to about 0.81 m³ a day after the gap was filled. Searches found single-tank measurements like this one and model studies, but no table that compares measured and design losses across many tanks.
Built with what exists
NASA's Technical Reports Server, DOE Hydrogen Program records, cryogenics conference papers and operator reports; a document-reading AI model to draft rows for a person to check (see the Note "Seven more flagged"); Python; a CSV file on GitHub or Zenodo.
First step
Enter the Kennedy sphere's design, worst and repaired boil-off figures from the NASA report as the first three rows, then search NASA's report server for other measured tank losses.
Done looks like
At least 40 tanks from at least ten independent sources, each with a measured daily loss and a link, a second person re-checking a random 20% of rows, and the fitted model's error published for tanks it was not fitted on.
Open 02 · the brief
Realistic cavern potential elsewhere
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A reproducible estimate of how much hydrogen could be stored in salt caverns in a second region, such as the German North Sea coast or the US Gulf Coast, after removing ground ruled out by towns, protected areas and existing industry, using the method a 2026 study applied to England's east coast. Start by reproducing that study's own figure.
Why it is still open
Estimates for the same stretch of coast differ widely: 22 to 48 TWh in an Arup report against 747 to 1,465 TWh in a 2022 study, as the Environment Agency notes. Garvey and colleagues (Arup, the British Geological Survey and the University of Edinburgh) cut earlier estimates for England's east coast by about 95%, to more than 22 TWh, and wrote that the method "can be applied to other key regions in the UK and internationally."
Built with what exists
Public salt maps from national geological surveys; OpenStreetMap land use; protected-area boundaries such as Natura 2000; Python with geopandas, or QGIS.
First step
Rebuild the study's exclusion layers for East Yorkshire from open data and compare the resulting area with the paper's.
Done looks like
Garvey and colleagues' east-coast figure reproduced to within 10%, then a second region's figure published with code, input maps and every exclusion rule, and reviewed by one geologist outside the project.