Liquid Hydrogen Pumps: What the −253 °C Duty Demands

At −253 °C a liquid hydrogen pump has almost no margin: the fluid sits close to its boiling point, so any heat that reaches it becomes gas, and any gas in the impeller becomes cavitation. Two Chinese tests reported in 2026 put figures on the state of the art. An electric-drive high-pressure pump held an outlet pressure of 90 MPa at a flow above 100 kg/h for four continuous hours at −253 °C, and a submerged transfer pump ran at 25 m³/h and 110 m head at the same temperature. Both tests were witnessed by the Beijing Institute of Aerospace Testing Technology (北京航天试验技术研究所) and a provincial mechanical and electrical testing institute, and both are reported as domestic firsts.

This article sets out where −253 °C sits on the low-temperature scale, which four problems the duty creates, how the two machine routes divide the work, and where the published evidence stops.

Where does −253 °C sit on the low-temperature scale?

It sits in the deep-low-temperature band, well below the range that ordinary refrigeration practice calls cold, and the boundary is stated in the industry's own terms.

The definitions used by a major Chinese cryogenic equipment maker's hydrogen business unit are: low temperature, −26.85 °C to −150.15 °C; deep low temperature, −150.15 °C to −272.85 °C; and ultra-low temperature, below −272.85 °C. On that scale liquid hydrogen at −253 °C is a deep-low-temperature duty, and so is liquid helium service for superconductivity at 4.5 K, which is about −269 °C.

The band matters because the design consequences are not linear in temperature. At the cold end of a deep-low-temperature duty the constraint stops being "does the material stay strong" and becomes "does the material stay ductile, does the seal stay closed, and does the fluid stay liquid".

Why does the fluid, not the pressure, set the design?

Because the pump is moving a liquid that is one small heat leak away from becoming a gas, and the four problems that follow are all consequences of that single fact.

Materials and hydrogen compatibility. The published development list for the 90 MPa machine names the study of materials in contact with hydrogen as the first problem. Hydrogen embrittlement and low-temperature embrittlement are separate mechanisms that occur together on this duty, and the material has to satisfy both. The transfer pump development is described as using special deep-cryogenic alloys, chosen for ductility at temperature rather than for strength alone.

Heat management. The second named problem is precise heat management. The published explanation is direct: if heat management is insufficient, the liquid hydrogen vaporises, which causes cavitation in the pump and can stop it running altogether. The engineering response on the transfer pump was a low-heat-leak structure designed from an analysis of the heat balance during transfer, combined with multiple insulation technologies.

Cavitation and hydraulic design. Vaporisation and cavitation are handled on the hydraulic side by an inducer-impeller integrated three-dimensional dynamic simulation, which the developer states was used to suppress thermal deformation and cavitation damage and to optimise the hydraulic components.

Sealing. The third named problem is low-temperature high-pressure sealing, and the published consequence of getting it wrong is stated just as plainly: if the safety control is insufficient, hydrogen leaks. The transfer pump development claims zero-leak, zero-fault operation under ultra-low temperature and high rotational speed, and describes the goal as complete sealing of the liquid hydrogen.

Two further items complete the list for the high-pressure machine: spool-valve dynamic control, which is a control problem inside a reciprocating machine, and matching the electric drive precisely to the duty. On the efficiency side, the transfer pump development names a special motor coil with ultra-low resistance in a liquid hydrogen environment.

What work do the two machine routes do?

They do different jobs at opposite ends of the pressure range, and the published figures show the split.

The high-pressure machine is a national programme output: it belongs to a 14th Five-Year Plan key research and development project on the development and safety of key equipment for liquid hydrogen refuelling stations. The same test loop carried a regulating valve developed alongside it, which is stated to regulate a wide flow range across a differential from 90 MPa down to 0.1 MPa and is described as the first Chinese valve to pass a liquid hydrogen test at a 90 MPa differential.

The transfer machine is the one that moves bulk liquid: hydrogen liquefaction plants, liquid hydrogen tanker filling and ship loading and unloading are the applications named for it.

The two are not alternatives. A refuelling station needs the high-pressure feed; a liquefaction plant or a tanker loading arm needs the transfer flow. Reading the 25 m³/h figure and the 90 MPa figure as a single capability would be a mistake — one is a flow, the other is a pressure, and they come from different machines.

Who built what, on the record?

Both machines come from the same developer, Hangyang Group, and the surrounding hydrogen position is published as design capability rather than as operating results.

The company states that it has completed process-package design for hydrogen liquefaction units of 0.5 to 10 tonnes per day, and that it holds a 1,200 L/h helium-refrigeration-cycle liquefaction technology reserve, with liquid hydrogen purity above 99.999% and para-hydrogen content above 95%. Adjacent cryogenic products are also published: a 40-foot liquid helium tank container stated to achieve static lossless storage for 80 days at −269 °C, and a 100 m³ horizontal vacuum-insulated liquid hydrogen storage tank whose boil-off rate is stated to be better than the industry standard. A 4.5 K cryogenic system for superconducting applications was reported as a successful bid in early 2026.

The distinction worth keeping is between a process package, a technology reserve and a delivered operating record. The first two are design statements. Only the two pump tests are measured results, and they are measured over hours.

Where does this route stop?

At four places, all of them visible in the published record.

Both results are hours-long tests, not service records. Four continuous hours at a test institute establishes that a machine can reach the condition and hold it. It does not establish a maintenance interval. No mean time between failures, no start-stop cycle count and no operating hours after the test are published for either machine.

"Largest publicly verified" is a statement about evidence, not about machines. The 25 m³/h transfer flow is described as the largest in China among publicly verified liquid hydrogen transfer equipment. That is the correct way to read it: the claim is about what has been publicly demonstrated, not about what has been built.

The valve claim is a single-item claim. The 90 MPa differential regulating valve is described as the first Chinese valve to pass a liquid hydrogen test at that differential. A first-of-its-kind statement rests on the test it names, and the test is one item in a programme.

The supporting numbers are design figures. The 0.5–10 t/day range, the 1,200 L/h technology reserve, the stated purity and para-hydrogen content, and the storage performance figures are company design and product statements. They describe capability, not measured service. Where a figure is a claim rather than a measurement, this article has said so.

What does the trend add up to?

Three movements are visible at once, and they point the same way. The pressure end moved to 90 MPa on an electric drive, which removes the need for a hydraulic drive circuit and the equipment that goes with it. The flow end moved to a 25 m³/h submerged machine, which is the size a liquefaction plant or a tanker loading arm needs. And the named problem list moved to the parts that cannot be bought: hydrogen-facing materials, heat management, low-temperature sealing and the valve that has to close across a 90 MPa differential.

What has not moved is the evidence base. Every measured figure in this article comes from a test campaign of four hours, witnessed by named institutes, in a programme that is still running.

What this article does not cover

Note that this article does not select a pump, size a machine, calculate net positive suction head, specify a seal, a material or an insulation system, or give prices and delivery times. It describes the low-temperature band, the four design problems as the developers state them, the two machine routes and the measured results that are published. It carries no compliance determination.

Where the next constraint sits

The next constraint on a liquid hydrogen route is not on the pump at all — it is on the fluid, because the same thermodynamic behaviour that makes the pumping hard also makes the storage and transfer hard. The temperature question more generally, and how a material's low-temperature limit is read against an ambient rating, is set out in Chemical Compatibility Chart for Pumps: What the Grade and the Temperature Actually Mean.

The mirror-image problem — a fluid whose density changes steeply with pressure and temperature rather than one sitting at its boiling point — is worked in CO₂ Injection Pumps: Why the Clearance Volume Decides the Efficiency.

Sources