Splitting water into hydrogen and oxygen requires a minimum of 39.4 kWh of electrical input per kilogramme of hydrogen at 100% thermodynamic efficiency — the enthalpy of formation of water, ΔH = 285.8 kJ/mol. No commercial electrolyser stack operates at that theoretical ceiling. Proton Exchange Membrane (PEM) and alkaline systems running at industrial current densities sit at 65% to 80% cell efficiency once ohmic resistance across the membrane, activation overpotential at the catalyst layer, and parasitic loads from balance-of-plant equipment (water circulation pumps, gas driers, power electronics) are accounted for. That is a minimum 20% to 30% parasitic energy loss at the molecular level, incurred before a single kilogramme of hydrogen has left the stack.
The loss compounds downstream. Compressing hydrogen from atmospheric pressure to the 350–700 bar required for vehicular or grid-injection storage consumes a further 10% to 15% of the gas's own energy content, owing to hydrogen's low Joule-Thomson coefficient and correspondingly poor compression thermodynamics relative to methane. Reconverting stored hydrogen back to electricity — via a combined-cycle turbine or, worse, a low-efficiency reciprocating engine — reintroduces a second conversion penalty on top of the first. The arithmetic is unforgiving: electrolysis (75% average), compression and storage (88%), and reconversion (55% to 60% for best-in-class turbines) multiply out to a net round-trip efficiency (RTE) of 30% to 45% for a full generate-store-reconvert cycle.
Set that against the counterfactual use of the same input electricity. 1 MWh of renewable generation routed through an air-source or ground-source heat pump yields 2.5 to 4 MWh of usable heat, a coefficient of performance of 250% to 400%, because the heat pump is moving thermal energy rather than manufacturing a chemical energy carrier from first principles. 1 MWh routed into a battery-electric drivetrain delivers 85% to 90% of that energy as motive force after charge, discharge, and inverter losses. Route the same 1 MWh through electrolysis, compression, and combustion or fuel-cell reconversion, and 550 kWh to 700 kWh of the original input is dissipated as waste heat and mechanical inefficiency before it does useful work. Every megawatt-hour of UK offshore wind or nuclear capacity diverted into a hydrogen pathway for an application with a viable direct-electrification alternative is, on this arithmetic, a 55% to 70% destruction of that generation asset's output.
The Industrial Hydrogen Vector Matrix
Grid-scaling policy therefore hinges on a single filter: does the target application have a direct-electrification substitute? Where one exists, routing electrons through electrolysis is a subsidy for inefficiency. Where no substitute exists — because the application requires hydrogen as a chemical reducing agent or because battery energy density cannot close the gap over the required range — the RTE penalty is the unavoidable cost of decarbonising an otherwise unreachable sector.
Target Application Sector | Proposed Technological Vector | Net Round-Trip Efficiency (RTE) | Economic & Infrastructural Viability Grade |
|---|---|---|---|
Domestic Space Heating | Blending hydrogen into existing gas distribution grids for boilers | 40% to 45% | Unviable — requires 3x to 4x more upstream renewable generation capacity than direct electrification via heat pumps |
Heavy Freight & Transport | Hydrogen Fuel Cell Electric Vehicles (FCEVs) | 35% to 40% | Low viability — outcompeted by the rapid advancement of megawatt-scale DC battery charging networks |
Primary Steelmaking Metallurgy | Direct Reduced Iron (DRI) shaft furnaces replacing blast furnaces | N/A — chemical reducing agent, not an energy carrier | Critical / highly viable — the only scalable chemical pathway to eradicate coal combustion in primary iron metallurgy |
Aviation & Maritime Transport | Synthetic e-fuels and liquid ammonia synthesis | 15% to 25% (highly complex synthesis) | Necessary / moderate viability — technically unavoidable given the volumetric energy density constraints of battery chemistries over intercontinental range |
Resource Constraints: Water Feedstocks and Grid Connection Friction
- The demineralised water penalty. Electrolysis demands ultra-pure, demineralised water as an atomic feedstock — trace mineral contamination degrades membrane and catalyst life within a PEM stack in a matter of weeks. Synthesising 1 tonne of green hydrogen requires roughly 9 tonnes of pure water, which in practice means 15 to 20 tonnes of raw feedwater once reverse-osmosis rejection rates are factored in. At gigawatt-scale deployment, that introduces heavy regional water-infrastructure stress and a non-trivial auxiliary energy overhead for the desalination and demineralisation loops feeding the electrolyser hall.
- Embrittlement and storage friction. Hydrogen's molecular radius is roughly a third that of methane, and its small size allows atomic hydrogen to diffuse into the crystal lattice of high-tensile carbon steel, causing hydrogen embrittlement that reduces fatigue life and fracture toughness. The UK's existing methane transmission network cannot carry high-concentration hydrogen without re-lining with polymer composites or a wholesale re-lay using hydrogen-rated steel grades — a capital programme measured in tens of billions of pounds, not a blending exercise.
- Spatial grid misalignment. Gigawatt-scale electrolyser farms need to sit directly adjacent to offshore wind landfall nodes — the Scottish and North Sea coastlines — to avoid crippling Transmission Network Use of System (TNUoS) charges levied on generation drawing from the transmission network at a distance from demand. That creates a structural geographic mismatch: the cheapest renewable electrons for electrolysis land hundreds of miles from the Midlands and South Wales industrial clusters that actually need the hydrogen, reintroducing transmission and compression losses that erode the RTE case further still.
These constraints bear directly on how primary steel producers should sequence their own decarbonisation roadmaps. A hydrogen-fed DRI plant sidesteps the RTE penalty because there is no substitute chemical reducing agent for virgin ore reduction — but it inherits the full weight of the water, embrittlement, and grid-siting costs set out above. A scrap-fed Electric Arc Furnace (EAF) avoids all three, at the cost of a different engineering problem entirely: the violent, cyclical MVA draw of arc melting and the local power-quality volatility it imposes on the transmission network. Our companion analysis, Electrifying British Steel: The EAF Transition and the UK Grid's Peak Load Problem, sets out that peak-load case in detail — read alongside this piece, the two form the full infrastructural cost-benefit picture a UK steelmaker needs before committing capital to either pathway.