High-Pressure Hydrogen Compression: The Pressure Ladder, the Sealing Breakthrough, and Where the Grades Now Stand

Hydrogen compression is a ladder, and the rung decides the machine. Production and blending sit at a few megapascals; long-distance pipeline duty at 6.3 MPa; tube-trailer filling around 20 to 30 MPa; and a 70 MPa refuelling dispenser means a compressor discharging near 87.5 MPa. No single machine family covers that range, and the two ends of it are where the technology moved recently — a water-lubricated seal that removes oil and barrier gas from the machine, and Chinese diaphragm compressors appraised at 90, 250 and 320 MPa.

The ladder, rung by rung

What matters about the ladder is that the duty changes character as it climbs. Up to about 10 MPa the gas is still a gas, the pressure ratio per stage is modest, and the machine choice is familiar from natural gas service. Above that, the compression ratio climbs, the discharge temperature becomes the binding constraint, and the containment requirement — no oil, no leakage, no contamination — starts to decide the design rather than being one line on a datasheet.

The refuelling rung is where the numbers are hardest. A Chinese high-pressure diaphragm compressor developed for a 70 MPa refuelling station is specified at 12.5 MPa inlet, not less than 87.5 MPa discharge, not less than 200 Nm³/h, and not more than 42 kW shaft power — a pressure ratio of seven in one machine, on a gas with a molecular weight of 2.02 (2).

Why this duty is a reliability problem before it is a pressure problem

The industry data on hydrogen compression is unusually blunt, because publicly funded station reliability programmes had to collect it. In the US Department of Energy's station reliability work, the compressor is first among station components for downtime and first for maintenance event count, and accounts for about one third of station maintenance hours (3).

The test record behind that finding is public. A commercially available diaphragm machine was run on a fuelling duty for 753 operating hours over 81 start-stop cycles, compressing from about 100 psi suction to 3,000–6,000 psi discharge. It moved 1,808 kg of hydrogen at 3.7 kg/hr and 3.54 kWh/kg average system efficiency, and recorded a mean time between failures of 49 days. The dominant failure mechanism was the seal: five seal failures, four in the compressor head and one on the second-stage discharge check valve (4).

Two design consequences follow, and both show up in the hardware described later in this article. A machine on this duty is started and stopped far more often than a process compressor, so its duty pattern is part of its failure record. And the parts that fail are the non-metallic ones — so their life, and the temperature they run at, becomes the specification that matters.

The standard has already tightened around hydrogen

The mechanical standard for reciprocating machines has moved the same way, which is a useful signal that the reliability problem is recognised rather than local.

The standard's own note gives the physical reason hydrogen is singled out: non-lubricated hydrogen services generally run at higher discharge temperatures than lubricated ones, because of internal leakage, and because hydrogen can release heat when it expands — behaviour unlike that of the heavier gases in this category. The same standard's speed clause runs in the same direction: in general, the rotating speed and piston speed of non-lubricated machines should be less than in equivalent lubricated ones. Going oil-free in hydrogen costs speed, and speed is what the non-metallic parts trade against (5).

The sealing breakthrough: taking the oil and the barrier gas out

The barrier to running a centrifugal machine on hydrogen has never been aerodynamics. It is the seal system: a conventional wet seal needs a circulating oil system whose degassing vents gas, and the oil is both a contamination risk and an emissions source. A dry gas seal removes the oil, but still needs a barrier gas — usually nitrogen — and that gas then has to be separated from the hydrogen downstream.

In late 2025 Flowserve qualified a seal that removes both. Aquapac uses ultra-pure water as the sealing fluid, which eliminates the oil system and the inert-gas barrier system together, and with them the process contamination route (6). The published specification and test record are specific:

• Peripheral speed above 100 m/s for sealing integrity, with stable operation confirmed at speeds up to 12,500 rpm;
• Operating temperature up to 180 °C, standard application window up to 30 bar maximum allowable working pressure, shaft sizes up to 300 mm;
• Designed to API 692 and API 682;
• Validated across continuous and variable-speed operation, temperature and pressure upsets, dry turning during assembly, slow roll, and emergency trip events — the test matrix is a fuelling-and-pipeline duty profile rather than a steady-state bench test;
• The design work included a data-analytics framework over the test data, using multivariate correlation and regression modelling to identify the parameters driving performance, with machine-learning seal-performance models named as the next step (7).

The boundary is as important as the breakthrough. A standard application window of 30 bar puts this seal at the lower-middle of the ladder — pipeline and recycle duty, not the 87.5 MPa refuelling rung. It is a machine-side enabler: it lets a centrifugal compressor enter oil-free hydrogen service where an oil system or a nitrogen barrier would otherwise have been required. It does not change who builds the high-pressure end of the ladder, and it does not by itself mean a centrifugal machine can go anywhere on it.

The high-pressure end, from the Chinese supply side

The high-pressure rung has been supplied by a small number of international brands. The Chinese progress in the last two years is best read as a sequence of grades, each with its own published evidence.

90 MPa. A diaphragm compressor developed as the key output of an Inner Mongolia science and technology programme for 70 MPa refuelling station equipment passed project acceptance as the first domestic 90 MPa diaphragm compressor. At 12.5 MPa inlet it delivers not less than 87.5 MPa and not less than 200 Nm³/h within 42 kW shaft power, and completed 500 cumulative hours of fault-free operation including 50 start-stops. Its gas valves, oil-spill valve and metal diaphragms exceeded 1,000 hours of fault-free operation on their own. Four sub-assemblies were developed for the duty: a 90 MPa composite high-pressure diaphragm head, high-pressure hydrogen gas valves, an impact-resistant anti-jamming high-pressure oil-spill valve, and a high-pressure profiled closed-loop metal sealing ring. The unit was appraised by the China General Machinery Industry Association as reaching an international advanced level, and the associated 70 MPa containerised refuelling unit was applied at the 2022 Beijing Winter Olympics and entered the National Energy Administration's first-of-its-kind major technical equipment programme (2).

250 MPa and 320 MPa. The 250 MPa ultra-high-pressure diaphragm hydrogen compressor programme, run as first undertaking unit, was awarded second prize in the 2025 Machinery Industry Science and Technology Award, with the award notice describing it as breaking a thirty-year foreign technology monopoly in this class (8). The same maker's 90 MPa, 250 MPa and 320 MPa diaphragm hydrogen compressor grades have passed product appraisal organised by the association, and its reciprocating and diaphragm lines now run past thirty product series covering hydrogen production, filling, refuelling, integrated production-and-refuelling, off-gas recovery and hydrogen chemical service (9).

Condition monitoring. The same 90 MPa programme produced what is described as the first domestic health management system for refuelling-station compressors — full-lifecycle real-time condition monitoring, fault diagnosis and failure warning (2). That is the reliability finding from the station data turned into an instrument, and it is the direction high-pressure hydrogen compression is moving in: not a machine that lasts longer between failures, but a machine that reports a developing failure early enough to act.

The three grades do not carry the same evidence, and the difference matters. The 90 MPa machine has a published operating point and a published endurance record: 12.5 MPa inlet, ≥ 87.5 MPa discharge, ≥ 200 Nm³/h, ≤ 42 kW, 500 hours including 50 start-stops, with four named sub-assemblies developed for it. The 250 MPa and 320 MPa grades are reported at the level of a programme award and of association product appraisal, without an equivalent published operating record. Both are real milestones in the domestic record; treating them as equally demonstrated would overstate the position.

What none of this has solved

Three things stay open, and the trend is easier to read with them in view. The reliability record has not been rewritten. The 49-day mean time between failures comes from the 2016 test campaign and remains the published reference point; condition monitoring reports a developing failure earlier, it does not remove the seal as the failure mechanism. The oil-free centrifugal opening has a pressure limit, and the limit is set by the seal's application window rather than by impeller aerodynamics — 30 bar in the published standard window, which is pipeline and recycle duty. And the high-pressure grades are not equally documented: only the 90 MPa machine carries a published operating point and endurance record, while 250 MPa and 320 MPa are reported at award and product-appraisal level. The direction of travel is consistent across all three; the distance covered is not the same at every rung.

What the trend looks like when the three parts are put together

The middle of the ladder is being opened to oil-free centrifugal machines by sealing technology, and the limit of that opening is the seal's pressure window rather than the aerodynamics. The high-pressure end is climbing grades — 90 MPa, then 250 MPa and 320 MPa appraisal — with the sub-assemblies (diaphragm head, gas valves, oil-spill valve, metal sealing ring) named as the parts that had to be solved. And the standard has stopped leaving the machine to the project: for hydrogen-rich and non-lubricated service it now writes 120 °C, which is a statement about which materials can be used at all.

What this article does not cover

Note that this article does not select a compressor, size a diaphragm or a seal, calculate stage count, power or efficiency, or give prices or delivery times. It describes the pressure ladder of hydrogen compression, the standard provisions that apply to it, and the specific technical developments reported by the organisations named.

Related on this site

• Reciprocating or Diaphragm for High-Pressure Hydrogen?
• Diaphragm vs Piston vs Screw Compressors: A Comparison Table

Sources