Lye Circulation Pumps in Alkaline Electrolysis: What a 30–120% Load Range Demands

On 18 September 2026 the Department of Energy Conservation and Comprehensive Utilization of China's Ministry of Industry and Information Technology opened a public consultation on draft manufacturing conditions for hydrogen equipment — specifically alkaline water electrolysis equipment. The draft sets the electrolyser's power regulation range at 30% to 120% of rated power, its load-change rate at not less than 20% of rated load per minute, its gas purity across the full load range at not more than 1.5% oxygen in hydrogen and 0.5% hydrogen in oxygen, and its direct-current power consumption drift after 2,000 hours of hydrogen production at not more than 2%, of which the last 1,000 hours must drift by not more than 0.8%. The consultation runs to 17 October 2026. In the same month, the hydrogen production section of a 505 MW wind-and-solar-to-green-ammonia project in Inner Mongolia reported all 48 of its electrolysers at stable rated output after a one-pass start-up.

This article sets out why a load range becomes a pump requirement, where the gas-purity limits are actually met, what the draft sets and at what level, what the standard record says about the pump itself, and where the record stops.

Why does the electrolyser's load range become a pump requirement?

Because the electrolyte has to be circulated for two reasons — to carry product gas away from the electrodes and to carry heat to the exchanger — and both of those duties change with the load.

An alkaline electrolyser splits water in a potassium hydroxide solution. Circulation of that solution is what removes the gas bubbles forming on the electrodes and what transports the heat generated in the stack to a cooler. The industry guidance is explicit that this circulation is achieved either by pumps or by natural circulation arising from temperature and density gradients — so the loop is not automatically a pumped loop, but where it is, the pump sits on the path that both functions depend on. (Source: the AIGA guideline for small-scale hydrogen production.)

The load range is what turns that duty from a point into a range. At 120% of rated power the stack is producing more gas and more heat than at nameplate; at 30% it is producing less than a third of the gas and correspondingly less heat, in the same loop with the same pipework. The draft's ramp requirement — a change of at least 20% of rated load per minute — adds the third dimension: the loop does not merely have to work at two ends of a range, it has to follow a moving duty point, which is what a wind or solar supply imposes.

The row that matters most is the last two. The draft's gas-purity limits and its ramp requirement are set on the electrolyser, but they are met — or missed — in the loop.

Where are the gas purity limits actually met?

In the gas-liquid separation and in the liquid streams that return from it, because that is where the two gas streams are separated from each other as well as from the electrolyte.

The mechanism is described in the industry guidance. Hydrogen is produced at the cathode and oxygen at the anode. Both leave the stack as gas carrying entrained electrolyte, and both are separated in a gas-liquid separator — hydrogen in the hydrogen separator, oxygen in the oxygen separator. The liquid streams that come out of the two separators still contain dissolved gas: hydrogen on the cathode side, oxygen on the anode side. Those streams are typically mixed before being recirculated to the stack, which prevents the electrolyte from being diluted or concentrated — water is consumed at the cathode and produced at the anode — but mixing them means the two gaseous constituents cross-contaminate. Where the two streams are instead kept separate and cross-fed to the opposite side of the cell, the cross-contamination is reduced but not eliminated, because gas still diffuses from one side of the cell to the other through the diaphragm. (Source: the same guideline.)

The same guidance states the operating consequence: to avoid hazardous contaminant concentrations, manufacturers limit the turndown of the electrolyser stack to between 10% and 40% of design load, and the system shuts down on contaminant concentration. Two things follow from that. First, the purity limits in the draft are of the same kind as the limits that already constrain turndown — the loop sets how far down the stack can be taken. Second, the loop's ability to keep circulating at the bottom of the range is part of the purity story, not just a utilities question.

The guidance also gives the temperature band the loop works in: the nominal operating temperature of alkaline water electrolysers is between 60 °C and 90 °C, well below the temperatures at which a caustic solution's effect on carbon steel becomes the governing materials question, and high enough that elastomer and polymer selections sit inside their own temperature bands rather than at ambient.

What does the draft set, and at what level?

Eight indicators are worth recording, and every one of them is a requirement on the electrolyser — not on the circulation pump.

The document's own status is part of the record. It is described in coverage of the consultation as a guiding document for the industry, one that does not carry prior or mandatory force in administrative approval. (Sources: the ministry's own notice, for the document and the consultation window; reproductions of the draft's attachment, for the indicator text.)

What does the standard record say about the pump?

That the circulating lye pump is a named component of the system in China's national standard for water electrolysis hydrogen systems — and that the alkali itself is subject to a separate standard.

The first row is the one that matters for this article's subject. A national standard for the system names the pump that circulates the electrolyte as part of the system's defined equipment, and requires the alkali to meet a chemical reagent standard rather than an industrial-grade specification. Neither standard sets a requirement on the pump's flow, head, material, net positive suction head or turndown. The system standard names the pump; it does not specify it.

The second row supplies the temperature and the pressure envelope the loop works inside, and the fact that the standard carries an electrolyte quality requirement confirms that the loop's chemistry is treated as something to be held within a band rather than as a fixed fluid.

Who has put a plant on the record?

One project, published with the number of electrolysers and the outcome of the start-up — and no machinery detail.

The project. A 505 MW wind-and-solar-to-hydrogen integrated green ammonia project at Otog Banner in Inner Mongolia, built by Shenzhen Energy Group's Beijing-controlled subsidiary, with a hydrogen production section built by East China Engineering Science and Technology Co. (东华工程科技股份有限公司) as the engineering, procurement and construction contractor. (Source: Shenzhen Energy Group's own site.)

The result. Reporting on 27 August 2026 states that all 48 electrolysers in the hydrogen production section reached stable rated output and that the whole process started up successfully on the first attempt, which the report presents as establishing that the hydrogen production station has full, stable hydrogen production capability. The project's own account states that the commissioning from the first electrolyser to all 48 took two months, that the team worked against three named difficulties — adapting to fluctuating wind power, integrating multiple systems, and the safety management of hydrogen production — and that hydrogen purity was continuously stable at specification. (Sources: China National Radio (央广网), 2026-08-27; Shenzhen Energy Group's own site.)

Why the load-following part is the point for this article. A wind-and-solar supply is a fluctuating power source. A stack train that starts up all 48 units and holds them at rated output has demonstrated the full-load end of the range; it does not, on the published record, demonstrate the 30% end or the 20%-per-minute ramp. The project is evidence that the equipment set reached a stable full-load condition on first start-up, which is a real milestone and a different one from a demonstrated load range. (Sources: the same.)

Where does this route stop?

At five places, all of them visible in the published record.

The draft is a consultation document, not an implemented standard. The consultation runs from 18 September to 17 October 2026, and coverage of the document describes it as a guiding document without prior or mandatory force in administrative approval. The indicators above are a proposal's requirements; they are not obligations, and the text may change before it takes any other form. They should not be quoted as if they were in force.

Every indicator is a stack-level or system-level figure. The load range, the ramp rate, the current density and the power consumption figures are requirements on the electrolyser. None of them is a specification for a circulation pump, and the draft as reproduced does not name a pump flow, a head, a material or a suction condition.

No pump-level data appears anywhere in the project record either. The 48-electrolyser accounts state the count and the start-up outcome. They do not state the circulation pump package, the loop flow rate, the pump material specification, the pumping power consumed by the circulation loop, the net positive suction head available at the pump, or the pump's own turndown range.

The loop is not necessarily a pumped loop. The industry guidance states that the electrolyte is circulated for gas-bubble and heat removal either by pumps or by natural circulation from temperature and density gradients. Which is used is a design choice, so the pump's share of the duty is not fixed by the standard or by the draft.

The 2,000-hour figure is not a pump reliability figure. It is a drift limit on the stack's direct-current power consumption after 2,000 hours of hydrogen production at rated current density. A pump failure over the same period would end the operation, but the figure itself says nothing about mean time between failures, seal life or bearing life in this duty.

Put plainly: the published material answers what the stack has to hold across the load range. It does not answer what the circulating loop has to be built to, and it does not answer which machine should be selected for a given loop.

What does the trend add up to?

Three movements are visible, and they sit at different levels of maturity.

The requirement moved from a nameplate condition to a range with a rate attached. The draft does not only say that the stack must run between 30% and 120% of rated power; it also fixes how fast the load may change, at not less than 20% of rated load a minute. A requirement expressed that way is a requirement on everything that has to follow the load — including the loop that carries the bubbles and the heat.

The purity requirement moved to the full load range. The 1.5% and 0.5% cross-contamination limits apply across the whole range, not at a design point. Since cross-contamination is a property of the gas-liquid separation and of the recirculated liquid streams, that shift puts a wider band of the loop's operation inside the acceptance criteria.

The evidence type has not moved. Every figure in this article is a draft requirement, a standard requirement or a project operator's own statement about a start-up. The installed base is described by equipment counts and start-up outcomes, not by loop performance, and there is no published circulation-pump record for an alkaline electrolyser duty anywhere in the material reviewed here.

What this article does not cover

Note that this article does not select a pump, size a machine, calculate flow, head or net positive suction head, specify a material, a seal or a lining, or give prices and delivery times. It describes the draft requirement, the loop's function, the standard record and the boundary of that evidence. It is not a purchase specification and it carries no compliance determination.

Where the next constraint sits

The constraint that follows this one is the other end of the same hydrogen chain, where the temperature problem inverts. Liquid Hydrogen Pumps: What the −253 °C Duty Demands takes a fluid that has to be kept cold rather than merely circulated, and shows what changes when one heat leak is enough to turn the fluid into gas and cost the machine its suction.

The other adjacent constraint is how the loop's own material question gets settled. The electrolyte is a caustic solution, the loop works at a temperature, and the answer for a given part depends on reading the medium's grade and the service temperature together rather than reading the medium alone — Chemical Compatibility Chart for Pumps: What the Grade and the Temperature Actually Mean sets out how that reading is done.

Sources