· Fuel Cells Hydrogen Shipping

Why a Little Hydrogen Goes a Long Way in LNG Dual-Fuel Engines

Wärtsilä and Bergen Engines are both blending hydrogen into natural gas-fuelled marine engines without hardware changes. The headline blend ratios are quoted by volume — converted to mass, the gap looks very different, and that gap is the whole story.

Two engine builders are running the same experiment from different starting points: take a proven LNG dual-fuel engine, meter in some hydrogen alongside the natural gas, and see how far you can push the blend before something breaks. Wärtsilä and Bergen Engines have both published real test results, and the numbers everyone quotes are by volume — 15%, 25%, 30%. Convert those to mass, which is what actually matters for combustion chemistry and fuel system design, and the headline ratios shrink to single digits. That gap between volume and mass is the quiet mechanical reason hydrogen blending works as well as it does with so little retrofit.

⚡ TL;DR

  • What: Wärtsilä and Bergen Engines are both blending hydrogen into existing LNG/natural gas dual-fuel engines, with no or minimal hardware changes, to cut CO2 and methane slip.
  • Key data: Wärtsilä's "world first" 25 vol% hydrogen blend test (Wärtsilä 50SG, WEC Energy Group, Michigan, Oct 2022) cut GHG emissions by 9.1% per Wärtsilä's own figure. By our calculation, 25% hydrogen by volume is only about 4% hydrogen by mass — hydrogen's molecular weight is roughly 1/8 that of methane.
  • Bergen's trajectory: 15% blend commercialized 2022 → 25% blend fleet-wide, no modification, July 2024 → 30% blend proven at full load → 100% hydrogen firing achieved December 2024, "with only minor modification."
  • Why it matters for slip: Hydrogen has no carbon to leak unburned, so every part of the fuel energy supplied as H2 is a part of the fuel that cannot contribute to methane slip — a direct, mechanical mitigation, not just a displacement effect.
  • Watch for: Wärtsilä's CIMAC 2025 paper on hydrogen blending and methane slip, the H4PERION project's 2030 sea trials on the ferry Aurora Botnia, and whether Bergen's December 2024 100%-hydrogen milestone turns into a sellable product.

The Volume/Mass Gap, and Why It’s the Whole Story

Every hydrogen blending announcement from both engine builders is quoted by volume: Wärtsilä’s “25 vol% hydrogen,” Bergen’s “15% hydrogen blend.” That convention makes sense for gas metering — volumetric (or molar) flow is what a fuel control system actually regulates — but it badly overstates how much hydrogen is actually in the cylinder by mass, because hydrogen is extraordinarily light. A hydrogen molecule (H₂, molar mass 2.016 g/mol) weighs about one-eighth of a methane molecule (CH₄, 16.04 g/mol). Fill a fraction of the fuel’s volume with hydrogen, and you’ve added far less than that fraction’s worth of mass.

The conversion is simple stoichiometry — mass fraction = (volume fraction × MH2) ÷ (volume fraction × MH2 + (1 − volume fraction) × MCH4) — and it changes the picture completely:

Hydrogen, by volume (as quoted) Hydrogen, by mass (our calculation) Where this appears
3% ~0.4% Wärtsilä 34SG, unmodified (Keppel O&M Floating Living Lab, Singapore)
15% ~2.2% Bergen B35:40 / B36:45L6AG, commercialized 2022
17% ~2.5% Wärtsilä 50SG, full (100%) load achievable
25% ~4.0% Wärtsilä 50SG “world first” test; Bergen’s fleet-wide unmodified limit (July 2024)
30% ~5.1% Bergen, full load, industrial-scale (Viscofan, Spain)
40% ~7.7% Bergen, claimed highest tested blend
60% ~15.9% Bergen’s stated target for a piston-upgrade package
100% 100% Bergen’s December 2024 milestone; both companies’ long-term goal

(These figures use pure methane as the natural gas proxy; pipeline-grade LNG is typically 90–98% methane with some ethane and heavier components, which shifts the real numbers by a fraction of a percentage point — not enough to change the order-of-magnitude story.)

The practical consequence: when the trade press reports a “world-first 25% hydrogen blend,” the actual fuel going into the cylinder is roughly 96% natural gas by mass. That’s a meaningfully smaller chemical and thermal disruption to a combustion system than the volume number suggests — which is exactly why both companies have been able to push these blends through largely unmodified engines designed for pure natural gas.

Wärtsilä: A 55 MW Power Plant as the Proving Ground

Wärtsilä’s highest-profile result so far is a power-generation test, not a marine one, but it’s run on the same 50SG engine family used in dual-fuel vessels. In October 2022, an unmodified Wärtsilä 18V50SG engine at WEC Energy Group’s 55 MW A.J. Mihm plant in Michigan ran for three continuous days on a 25 vol% hydrogen blend (our calculation: ~4% by mass), reaching 95% load while continuing to supply power to the grid. A second test point found that a 17 vol% blend (~2.5% by mass) allowed full (100%) load. The Electric Power Research Institute (EPRI) led the performance assessment. Wärtsilä’s own release reports a 9.1% greenhouse gas emissions reduction; trade press figures for the same test range from 9% to 12% depending on reporting basis. NOx stayed within the plant’s permit limits throughout — notable, since hydrogen combustion tends to push NOx up, not down, all else equal.

Wärtsilä has also quoted smaller, truly unmodified blend tolerances for marine dual-fuel gensets: the 34SG engine sets supplied to Keppel Offshore & Marine’s Floating Living Lab testbed off Singapore can run on up to 3 vol% hydrogen (~0.4% by mass) as-is, rising to 25 vol% with modification.

Two further Wärtsilä efforts point at where this goes next:

  • H4PERION (Horizon Europe, led by the University of Vaasa, 16 partners across 7 countries, running to May 2030) tasks Wärtsilä with developing a combustion concept for hydrogen/biomethane blends and a catalyst system specifically to cut methane slip, with sea trials planned aboard the Wasaline ferry Aurora Botnia between Finland and Sweden alongside a full-scale lab engine test. The long-term target is up to 100% hydrogen in open-sea operation.
  • The Hycamite partnership (announced November 2022) is a different approach to the same supply problem: instead of bunkering hydrogen separately, Hycamite TCD Technologies’ methane pyrolysis process would crack a portion of a ship’s own LNG into hydrogen and solid carbon on board, feeding the hydrogen back into a Wärtsilä dual-fuel engine via the existing LNGPac fuel gas supply system. The solid carbon byproduct — potentially graphite or carbon nanotube grade — is easier to store than captured CO2 and could have resale value. A prototype testing unit was targeted for the second half of 2024; we found no published results confirming it ran on schedule.

Wärtsilä also presented a paper at the 2025 CIMAC World Congress in Zurich — “GHG and Methane Slip Reduction: The Role of Hydrogen Blending into Natural Gas” (Ischia, Portin, Delneri) — framing hydrogen blending as a feasible way to cut methane slip specifically in the engine operating region where slip is worst, alongside tuning, control software, and aftertreatment. We were unable to retrieve the full paper directly — the host domain’s TLS certificate doesn’t match its own hostname, blocking a secure fetch — so we’re citing the abstract-level framing reported in conference coverage rather than the paper’s detailed results.

Bergen Engines: A Faster, More Public Ladder

Bergen Engines (Norwegian medium-speed engines, owned by Langley Holdings since a December 2021 acquisition from Rolls-Royce) has climbed the same ladder more publicly and, on the available evidence, faster.

Date Milestone Hydrogen blend Notes
Early 2022 First blend test 15 vol% (~2.2% by mass) B35:40 lean-burn gas engine, no hardware changes, rated output held at all loads
~2023 Industrial field trial 15 vol%, full load Viscofan’s Cáseda plant, Spain — Bergen B36:45L6AG, 5.5 MW combined heat and power; green hydrogen from Electroquímicas de Hernani via Nippon Gases, ramped to 400 Nm³/h; Bergen estimates ~8,000 t CO2/yr potential savings for the site
2 July 2024 Fleet-wide blend increase 25 vol% (~4.0% by mass) Entire natural gas engine range, no modification required
2024 (undated) Full-load claim 30 vol% (~5.1% by mass) Bergen claims to be the first medium-speed engine maker to reach 30% hydrogen at 100% load, proven at the same Spanish industrial site
December 2024 100% hydrogen firing 100% Achieved “with only minor modification,” meeting a target set by Langley Holdings chairman Tony Langley in October 2023

Bergen’s own figures for CO2 avoided (2.15 kg per cubic metre of natural gas displaced) are consistent with straightforward fuel-substitution accounting rather than anything exotic in the combustion process. Its longer-term commercial plan is to offer a 60 vol% hydrogen (~15.9% by mass) capability as a factory-fit upgrade bundled into a scheduled major overhaul that already includes piston replacement — timing the hydrogen-specific parts to a maintenance event that was going to happen anyway, which keeps the incremental cost of the upgrade down.

It’s worth flagging that every figure in Bergen’s public timeline — the 25%, the 30%, the December 2024 “100% hydrogen” result — comes from company and trade-press releases, not an independent third party. We found no outside verification of test duration, emissions measurements, or output stability for the 100% hydrogen milestone specifically.

Why Hydrogen Cuts Methane Slip Mechanically, Not Just by Displacement

Methane slip — unburned methane escaping through the exhaust of a lean-burn gas engine, particularly at low and partial load — is one of the main reasons regulators and class societies haven’t simply waved LNG through as a clean marine fuel despite its CO2 advantage over fuel oil. Hydrogen blending addresses it two ways simultaneously:

  1. Direct substitution. Every unit of fuel energy supplied as hydrogen is energy that wasn’t supplied as methane, so there is correspondingly less methane present to slip past the piston rings or survive the flame front unburned in the first place.
  2. Combustion quality. Hydrogen’s wide flammability range and fast flame speed relative to methane improve flame propagation through the lean, dilute air-fuel mixtures that lean-burn gas engines run to control NOx — the same mixtures that are hardest for a methane-only flame to burn completely, which is exactly where slip is worst. This is the mechanism the CIMAC 2025 paper’s framing points at: hydrogen blending helps most in the engine’s least favorable operating region, rather than uniformly across the load range.

Both companies report lower methane slip alongside lower CO2 in their blend tests, though — as with the CO2 figures — neither has published a slip number in grams per kilowatt-hour that would let us compare their results against each other or against Wärtsilä’s own NextDF methane-slip-reduction technology (a separate, non-hydrogen approach that Wärtsilä says cuts methane emissions on its 46TS-DF engine to under 1.4% of fuel use across the load range).

Why This Matters

For an existing LNG dual-fuel fleet, the appeal of this approach is specifically that it doesn’t ask for a newbuild or a storage-system overhaul. A ship already running LNG dual-fuel propulsion has the cryogenic storage, the fuel gas supply system, and the combustion hardware largely in place; hydrogen blending at the levels demonstrated so far (single digits by mass) asks comparatively little of that existing plant. That’s a meaningfully lower bar than converting to ammonia or pure hydrogen, both of which demand new storage, new safety systems, and in hydrogen’s case a roughly four-times-larger tank for the same energy content.

The honest caveat is scale: even Wärtsilä and Bergen’s best-documented results — 4-5% hydrogen by mass — only get a fraction of the way to a genuinely low-carbon fuel mix, because the carbon-bearing 95%+ of the fuel is still natural gas. The real decarbonization case for this technology depends on how far the blend ratio can climb without requalifying or rebuilding the engine, and on whether the hydrogen itself is green. Bergen’s December 2024 100%-hydrogen result and Wärtsilä’s H4PERION targets both point toward eventually closing that gap — but neither company has yet published a commercial product spec for a blend much above the 25-30 vol% (4-5% mass) range that’s been running since 2024.

Challenges and Open Questions

  • No independently verified slip or CO2 numbers in standard units. Neither company has published methane slip in g/kWh or a CO2 reduction audited by a third party for their hydrogen-blend tests, which makes it hard to compare the two programmes directly or benchmark them against non-hydrogen slip-reduction technology like Wärtsilä’s own NextDF.
  • Bergen’s December 2024 100%-hydrogen milestone is unverified outside company channels. We found no independent report of test duration, load stability, or emissions during that run.
  • The Wärtsilä CIMAC 2025 paper’s full results are currently unreachable. The host (papers2025.cimaccongress.com) serves a TLS certificate that doesn’t match its own hostname, which blocked a secure direct fetch; our summary of it rests on conference coverage and search-indexed fragments, not the full paper.
  • Hycamite onboard hydrogen production’s prototype status is unconfirmed. Wärtsilä and Hycamite targeted a prototype testing unit for the second half of 2024; we found no published update on whether it was built or what it showed.
  • Natural gas isn’t pure methane. Our mass-fraction table uses methane’s molar mass as a proxy; real LNG composition (ethane, propane, and nitrogen content vary by cargo and source) will shift the exact mass percentages slightly, though not the order of magnitude.

Sources