Electrolyser water quality requirements: what the datasheets don’t tell you

Every electrolyser datasheet specifies feed water quality. Most of them state a conductivity figure, a pH range, and not much else. What they do not tell you is what happens when you try to actually achieve those figures at your specific site, using your specific water source, at the flow rates your project requires. The gap between the datasheet number and the engineering reality is where projects pick up unbudgeted CAPEX, commissioning delays, and early membrane degradation.

This is not an edge case. Water treatment is underspecified at feasibility in the majority of electrolyser projects we review. Here is a structured guide to what actually needs to go into a water quality assessment.

What the datasheets actually specify

PEM electrolyser vendors typically specify feed water conductivity below 0.1 μS/cm, measured at the electrolyser inlet. Some vendors specify total organic carbon below 0.05 mg/L, silica limits, and chloride thresholds - but these parameters are often buried in the technical annexes to the supply agreement rather than the headline datasheet.

Alkaline systems are less demanding on conductivity - typically below 1 μS/cm at the demineralised water makeup point - but the lye system introduces its own water quality management requirements that are often absent from feasibility cost estimates entirely.

The datasheet figure applies at the electrolyser inlet. What happens between your water source and that inlet is entirely your responsibility, and it is where most of the complexity and cost lives.

Source water matters more than most feasibility studies assume

Mains potable water is the easiest starting point. In most European locations it arrives with conductivity in the range 200–600 μS/cm, with known hardness, controlled microbiological content, and consistent supply pressure. Treating it to electrolyser inlet quality via reverse osmosis and deionisation is well-understood and the equipment is readily available. The cost is real but it is not a surprise.

The problems start when projects are sited in locations where mains supply is constrained, expensive, or insufficient for the required flow rate. In those cases, developers look at groundwater, surface water abstraction, or process water recovery. Each introduces additional treatment complexity:

• Groundwater: Often contains elevated iron, manganese, and dissolved gases that foul membranes if not removed upstream of RO. Iron fouling in particular causes irreversible membrane damage quickly.
• Surface water: Variable suspended solids, biological load, and seasonal changes in ionic composition mean the treatment system must be designed for worst-case inlet conditions, not average conditions. Ultrafiltration ahead of RO is typically required.
• Process water recovery: Where cooling water blowdown or other process streams are considered as a feed source, detailed characterisation is essential. Scaling inhibitors and biocides used in cooling systems can be incompatible with RO membranes.

The treatment train and what it costs

A typical treatment sequence for mains water feeding a PEM electrolyser involves: pre-filtration, activated carbon filtration (to remove chlorine, which destroys PEM membranes), reverse osmosis, and mixed-bed deionisation with conductivity monitoring. For a 10 MW PEM system at 70% capacity factor, water consumption is approximately 6–8 m³/hr at the electrolyser. The treatment system needs to deliver that continuously at specification.

At feasibility, water treatment is frequently estimated as a lump sum in the “utilities” line of the CAPEX summary. In practice, for a 20 MW project, a properly specified treatment system including pre-treatment, RO, DI, monitoring, and chemical dosing adds €400,000–€900,000 to installed CAPEX depending on source water quality. The O&M cost - membrane replacement, resin regeneration or replacement, chemical consumption, and monitoring - needs to be in the LCOH model as a separate line item, typically €0.02–€0.06/kg H₂ depending on source water quality and system scale.

Chlorine: the most common membrane killer

Mains water in most European countries contains 0.1–0.5 mg/L residual chlorine as a disinfectant. PEM membranes - specifically the Nafion membrane used in most commercial systems - are highly sensitive to oxidising agents including chlorine. Even trace concentrations at the inlet cause irreversible membrane degradation over time.

The fix is activated carbon filtration upstream of the RO, combined with online oxidation-reduction potential (ORP) monitoring at the RO inlet with automatic shutdown on high ORP. This is standard practice in well-designed systems but is sometimes omitted in early-stage designs where the water treatment package has been quoted by the electrolyser vendor without site-specific characterisation.

What to do at feasibility stage

• Commission a water quality analysis of your intended source, covering conductivity, hardness, iron, manganese, silica, chlorine, TOC, and microbiological parameters.
• Confirm the available supply flow rate and pressure at site, including any constraints on abstraction volume or mains connection capacity.
• Size the water treatment system using worst-case inlet conditions, not typical conditions.
• Include water treatment CAPEX and O&M as explicit line items in your project model, not as a utilities contingency.
• Check vendor requirements for chloride, silica, and TOC in addition to conductivity - these are often specified in the supply contract rather than the datasheet.

Key takeaways

• Electrolyser datasheets specify feed water conductivity but not the full set of requirements or how to achieve them from real water sources.
• Source water characterisation should happen at feasibility, not during FEED.
• Chlorine removal is critical for PEM systems and is sometimes absent from early-stage designs.
• Water treatment CAPEX and O&M are material cost items that belong in the project model, not in contingency.