Water Requirements for Green Hydrogen: Electrolysis Purification, Desalination, Wastewater Reuse & ZLD

By Green Gas Turbines Team · Published January 4, 2026 · 16 min read

Why Water is a First-Order Constraint for Green Hydrogen (Not an Afterthought)

Every electrolyzer is ultimately a water plant with a power block attached. The chemistry is simple:

2H2O → 2H2 + O2

But the operational reality is messy. “Tap water in, hydrogen out” fails the moment you meet real-world contaminants—especially silica, hardness, chlorides, and organics. Operators end up running a continuous purification + polishing process (RO, resin beds, and increasingly EDI/CEDI) to protect the stack, keep efficiency stable, and avoid warranty disputes.

This article focuses on practical project questions engineers and EPCs get stuck on:

The Two Numbers Everyone Quotes (and the One That Actually Matters)

1) The stoichiometric minimum: ~9 L per kg H2

By basic chemistry, producing 1 kg of H2 via electrolysis consumes about 9 liters of H2O (≈ 9 kg of water).1–3

2) The system-level reality: often ~15–25 L per kg H2

Real plants lose water in purification reject streams and, more importantly, in cooling systems and auxiliaries. Multiple analyses converge on a practical range of ~15–25 L/kg H2 as a common “project-level” number (site design matters).1,2,4

Design tip: If your project pro forma only budgets 9 L/kg, it’s missing purification losses and thermal management. That can break your water permits long before you hit nameplate MW.

Water Quality: Alkaline vs PEM Specs (Conductivity is the Gatekeeper)

Electrolyzers do not tolerate “mostly clean” water. OEMs specify tight water quality envelopes, and operating outside them can reduce stack life and jeopardize warranty support.5,6 The fastest proxy measurement on site is conductivity (how many ions remain in the water).

Typical conductivity targets (rule-of-thumb)

Electrolyzer type Typical feed water quality target Why it matters
Alkaline Often < 5 µS/cm (project examples vary; some guidance targets < 1 µS/cm).5,7,8 Higher ionic content increases scaling/corrosion risk and can destabilize long-term performance.
PEM Frequently < 0.1 µS/cm (ultrapure / reagent-grade territory).6,7 PEM stacks are highly sensitive to ionic contamination; impurities can poison expensive catalyst-coated components.

Where ASTM D1193 and ISO 3696 fit

When EPCs say “ASTM Type I/II water,” they are referencing reagent water specifications like ASTM D1193 (reagent water types) and ISO 3696 (laboratory water grades). These standards are not written specifically for electrolyzers—but they provide a common language for “how clean is clean.”9,10

The “Polishing” Headache: RO Is Not Enough (Silica Is the Silent Killer)

Most sites use a multi-stage train:

  1. Pre-treatment: screens/filters, often multimedia + cartridge filtration; dechlorination where required
  2. Reverse osmosis (RO): removes most dissolved salts
  3. Final polishing: either mixed-bed ion exchange or EDI/CEDI to reach ultra-low conductivity and low silica

The operations pain is the last step. If silica or ions “slip” through, you can end up with deposits that reduce efficiency and shorten stack life. Water-polishing vendors explicitly call out EDI as a route to achieve both low conductivity and low silica after RO.11

Mixed bed ion exchange vs EDI/CEDI

Both can produce high-purity water, but the economics and operating model are different:

2026 trend: For large-scale green hydrogen projects, EDI/CEDI is increasingly favored because it fits continuous operation and reduces on-site chemical handling—especially when the plant is remote or labor-constrained.12,13

Water Sourcing Models (What Actually Works at Project Scale)

1) Industrial/municipal surface water + demin train

In industrial clusters, the “lowest drama” option is often a stable surface/municipal source, followed by a robust demin train. For example, Shell’s Holland Hydrogen 1 project documentation describes sourcing water from a surface water body (Lake Briel) and producing large volumes of demineralized water for electrolysis.14 Vendors like Pure Water Group have publicly stated they are supplying ultrapure water systems for the project, underscoring that the UPW skid is not a side detail—it is core plant equipment.15

2) Desalination for desert-scale electrolysis (and the ZLD permitting pivot)

In arid regions, desalination is often unavoidable. NEOM’s water strategy has been closely watched because it pairs massive new demand (including energy and hydrogen ecosystems) with large desalination buildout and brine-management ambitions. NEOM, ENOWA, and partners publicly signed an MoU to develop a “selective desalination” approach tied to a circular economy model and zero liquid discharge (ZLD) aspirations—aimed at avoiding direct brine effluent discharge and enabling minerals recovery.16

Industry-awareness note: A later update reported that a high-profile NEOM desalination JDA associated with that concept was not carried forward as originally planned.17 That doesn’t negate the trend—it highlights it: brine management is becoming a first-class permitting issue, and developers are being pushed toward high-recovery, brine valorization, and ZLD-style solutions.

3) Wastewater reuse and industrial reclaimed water

Reclaimed water is increasingly attractive where freshwater permits are tight. The challenge is not “can we treat it?”—it’s variability (TOC swings, trace organics, and seasonal shifts) and building the monitoring discipline to protect the stack. In practice, many projects use reclaimed water as a feed source but still run a full RO + polishing train to hit PEM-grade conductivity targets.

4) Air-to-water (early-stage, but strategically important)

Air-to-water is still niche, but it matters because it attacks the hardest siting constraint: making hydrogen in deserts without pulling down aquifers. Australia’s Desert Bloom Hydrogen concept drew attention for piloting atmospheric water harvesting (Aqua Aerem) in the Northern Territory. Public tracking indicates the project later stalled (“archived/put on ice”), but it remains a concrete example of how developers are experimenting with water sourcing beyond pipes and aquifers.18–20

Desalination Energy: Debunking the “Double Energy Penalty” Myth

Seawater reverse osmosis (SWRO) has a typical specific energy consumption on the order of 2.5–4.0 kWh/m3 (including real-world plant loads).21 That sounds large—until you convert it into “per kg H2” terms:

Even if you use a “system water” number of 20–25 L/kg H2, you’re still at roughly 0.05–0.10 kWh/kg. Compared to ~50 kWh/kg electricity consumption for electrolysis (order-of-magnitude), desalination is typically <0.2% of the energy budget—negligible in LCOH terms.1,21

What desalination really changes: not energy cost, but permitting (intakes, brine discharge) and capex footprint (RO trains, pretreatment, brine management/ZLD).

Recycling & Water Balance: Designing the Hydrogen Plant as a Closed-Loop System

The practical water footprint is dominated by what you choose to throw away:

Where recycling is realistic

Brine Disposal and ZLD: What Permitting is Starting to Demand

Brine is not “just salty water.” It can carry chemical residuals from pretreatment and high salinity that harms marine ecosystems if poorly managed. UNEP and peer-reviewed literature describe brine discharge as a growing environmental concern, driving interest in minimization and ZLD approaches.22–24

What ZLD actually means

Zero Liquid Discharge (ZLD) aims to recover water and convert remaining brine into solid salts for disposal or valorization (e.g., industrial salts), rather than discharging a liquid concentrate. Recent academic and industry reviews describe ZLD as a key brine-management pathway with growing relevance as desalination expands.25

Project Checklist: Water System Decisions That Make or Break Bankability

  1. Start with a water balance: model 9–10 L/kg stoichiometric and add purification reject + cooling losses to get a credible range (often 15–25 L/kg).1,2
  2. Lock the conductivity target by electrolyzer type: alkaline can be looser than PEM, but confirm OEM limits and warranty terms.5–7
  3. Pick your polishing philosophy: mixed-bed IX (chemical/regeneration logistics) vs EDI/CEDI (continuous, chemical-free).12,13
  4. Plan for silica control: ensure the polishing stage is designed to maintain low silica, not just low conductivity.11
  5. Choose a cooling approach early: cooling often dominates consumptive water use; it’s not a late-stage mechanical detail.
  6. Desalination? Budget for permitting and brine: the energy cost is small; the environmental and regulatory design work is not.21–25
  7. Instrumentation & QA: continuous conductivity monitoring and robust sampling protocols prevent “silent drift” that damages stacks.

Frequently Asked Questions

What is the water quality standard for PEM electrolyzers?

PEM electrolyzers typically require ultrapure / reagent-grade water with extremely low ionic content—often targeted at < 0.1 µS/cm conductivity in project references. Standards like ASTM D1193 and ISO 3696 are commonly used to define reagent-grade water quality classes, even though OEMs provide the final acceptance specification for warranty purposes.6,9,10

Can seawater be used directly in an electrolyzer?

Not in commercial plants today. The salts and minerals in seawater would rapidly corrode components and foul membranes/electrodes. Commercial projects desalinate first (typically RO), then polish to ultralow conductivity before the stack. “Direct seawater electrolysis” remains an emerging research pathway rather than a bankable industrial design.

How much water is actually consumed to produce 1 kg of hydrogen?

The chemical minimum is about 9 liters per kg H2.1–3 In practice, total water demand often rises to ~15–25 liters per kg once purification losses and cooling are included, depending on cooling strategy and water recovery design.1,2,4

What is the difference between Mixed Bed Deionization and EDI for hydrogen plants?

Mixed-bed ion exchange achieves high purity but requires resin regeneration logistics (often involving chemicals or off-site service). EDI/CEDI is a continuous polishing technology widely described as chemical-free and self-regenerating for RO permeate, which aligns well with steady-state hydrogen production and minimizes on-site chemical handling.12,13

Does sourcing water from desalination make green hydrogen too expensive?

Generally no. Modern SWRO energy use is typically 2.5–4.0 kWh/m3.21 Since electrolysis needs ~10 L/kg H2 of ultrapure water, desalination energy is roughly 0.03–0.04 kWh/kg H2 (even lower than 0.2% of a ~50 kWh/kg electrolysis energy budget). The larger “cost” is usually capex footprint and environmental permitting around brine management, not electricity consumption.

Further Reading & References