绿氢装备的缺陷清单：一次验收测试事故与电堆的衰减路径Defect List of Green Hydrogen Equipment: An Acceptance Test Accident and the Degradation Path of the Stack

Discussions in the industry about defects in green hydrogen equipment usually stop at the statement that the lifespan needs improvement, while the details that could truly help engineers make judgments are rare. On October 19, 2026, a progress announcement from a UK listed company provided some rare technical details: Clean Power Hydrogen plc disclosed that its 1 MW membrane-free electrolyzer MFFE220 experienced a system ignition failure during the third phase of factory acceptance testing on May 28, 2026, causing severe damage to the equipment. The conclusions listed in the announcement are very valuable; they do not attribute blame to 'the technology itself,' nor do they provide a simple answer. Instead, they list multiple possible factors side by side and explain the physical mechanism of each. This announcement, along with nearly two years of continuous academic research on the degradation mechanism of proton exchange membrane electrolyzers, forms the two most engineering-valuable pieces of the current defect map for green hydrogen equipment.

Accidents during the factory acceptance testing phase: Why this stage carries the highest risk

First, clarify the sequence of the accident. According to the announcement, the company has completed an internal technical investigation of the accident that occurred during the third and final phase of the factory acceptance test on May 28, 2026. Although some technical details are still being verified by external experts outside the field, the Board believes that the progress of the investigation is sufficient to provide an update. The investigation concluded that, so far, several unrelated factors have been identified, which likely collectively contributed to the accident. No single definitive ignition source has been determined, and no evidence has been found to suggest that the accident originated from the company's proprietary membrane-free stacks or supporting separator containers. The four main factors listed are: first, the gas flow velocity and particle impact near valve V705, which could generate enough energy to trigger ignition during a standard controlled venting process; second, deviations in the specifications and configuration of certain check valves, hoses, and pipe elbows.

Third, the quality of suppliers and the environmental characteristics of the outdoor testing facilities have affected the 'cleanliness' required for the overall system, leading to particulate contamination inside the unit; fourth, the applicability and compliance of this new technology design with the relevant guidelines of the European Industrial Gases Association are currently being evaluated as part of the remedial design process. The board believes that the combination of these main factors is highly likely to be the cause of the test failure.

Reading these four items side by side, one finds that the technical logic of the accident is much more complex than a single-point failure. It is particularly noteworthy that the accident occurred during the controlled discharge phase rather than during normal operation—controlled discharge itself is a routine action designed as a safety procedure, and it is precisely during this action that the system needs to safely release gases from within the equipment. If ignition occurs during normal operation, the problem mostly lies in electrochemical aspects or the main process; if it occurs during the discharge phase, it indicates that the triggering conditions come from the flow details at the end of the pipeline and the random impact of particles. Particulate contamination can become an independent factor because it has two characteristics: first, it is ubiquitous, introduced through supplier materials, assembly processes, and outdoor testing sites; second, its energy release depends on impact speed—the higher the gas flow speed, the greater the kinetic energy carried by particle impacts, and theoretically, the more likely the energy required for ignition will be provided.

Local flow velocity anomalies near the valve can further amplify this effect—despite having the same pipe diameter and the same total flow rate, once the flow velocity distribution is disrupted by the installation angle or geometric changes of a particular component, it can create locally a speed far exceeding the average. The remedies the company has undertaken in response also carry engineering implications. The announcement clearly states that the investigation concluded that in order to successfully complete the tests, changes need to be made to the design of certain components, system cleanliness procedures, manufacturing controls, and operational condition monitoring. The company has already initiated remedial research and design work in this regard and plans, based on satisfactory test results and the board’s approval of the final investigation conclusions, to provide the revised design packages to relevant licensors and potential manufacturing partners. It is worth noting that the company has not yet announced a schedule for completing factory acceptance testing. A clear technical judgment is that the cost of exposing defects during the acceptance testing phase is far lower than that of on-site explosions after delivery, but the correction cycle usually spans years. For manufacturers, this means that the speed of establishing the credibility of a technical approach can be substantially slowed by a failure that could otherwise have been attributed to external factors.

After the accident: How a failure changed the manufacturing model

The chain reaction triggered by this accident illustrates the current stage of the industry more than the technical details themselves. According to public reports, after the accident, the company's business strategy underwent a major shift, moving from an internal manufacturing model to a strategy centered on intellectual property. On June 25, 2026, the company announced that it had signed a binding term sheet with Ireland's Lisheen H2 Energy Park Limited (Hidrigin), involving a secured convertible loan note of £750,500, conditional upon meeting certain conditions and signing the final documents. Consistent with the previously announced strategy of shifting to asset-light technology development and licensing business, the completion of electrolyzer manufacturing would no longer be carried out internally by the company; therefore, a broader strategic partnership and manufacturing and technology development agreement is currently under negotiation with Hidrigin.

After the financing was oversubscribed and completed, both parties agreed not to proceed with the convertible loan notes. The company has refunded the previous interim payment of £740,500 received under the original MFE220 sales contract. The convertible loan notes were not issued, and no related guarantees were granted. Hidrigin's technology license is being expanded, and the agreement has reached a consensus on extending the rights for self-use in the United States.

The engineering implications of this set of actions are more noteworthy than the business implications. When a tech company shifts manufacturing from its own assets to licensing and joint ventures, it indicates that the company acknowledges shortcomings in its process control and cleanliness management capabilities, while transferring the risks to partners who have manufacturing capabilities. For the entire industry, this points to a commonly underestimated fact: the feasibility of an electrolyzer technical route and the ability of the manufacturing system to reproduce it reliably are two independent assessments. A new architecture that works in the laboratory does not mean it can maintain consistency on the production line; likewise, a company that already has large-scale manufacturing capabilities may still reveal weaknesses during acceptance due to a lack of cleanliness and testing systems. The value of manufacturer assessment systems and third-party certification lies in addressing the latter issue at the design stage.

Defect paths inside the stack: how hydrogen permeation eats into the stack's lifespan.

Returning from the device layer to the electrochemical layer, the defect pathways are equally clear. In proton exchange membrane electrolyzers, hydrogen permeation from the cathode to the anode is considered a key degradation pathway. A study published on August 14, 2026, in ACS Energy Letters provided mechanistic confirmation: the study identified hydrogen permeation from the cathode to the anode as the critical reason for the insufficient stability of low-loaded iridium catalysts at high current densities. Temperature-programmed reduction experiments confirmed that iridium dioxide is reduced in an 80°C hydrogen environment, highlighting the vulnerability of iridium dioxide-based catalysts when exposed to hydrogen. To mitigate this effect, the study introduced palladium as an anode additive, enabling it to act as a hydrogen oxidation catalyst to suppress the reduction of iridium dioxide. Verification via inductively coupled plasma atomic emission spectroscopy and in situ X-ray absorption spectroscopy showed that under conditions of oxygen-covered palladium surfaces and at the oxygen evolution reaction potential, iridium dissolution is significantly suppressed under 80°C hydrogen flow.

The reason why permeation is difficult to eliminate lies in its direct conflict with cost reduction. Membrane thickness is the most effective physical means to inhibit permeation; however, in order to reduce ohmic loss and pursue higher energy efficiency, the membrane is continuously being thinned, which significantly enhances the ability of hydrogen molecules to penetrate the membrane through concentration gradients and diffusion. Publicly available industry technical analyses, citing collaborative research between Cummins and the academic community, indicate that under typical industrial operating conditions, the "hydrogen permeation flux" usually remains at a level equivalent to 0.1 to 0.5 mA per square centimeter in current density; although this number seems small, the consequences are multi-dimensional. Hydrogen that permeates to the anode is oxidized on the anode catalyst surface to form a mixed potential; this process does not produce oxygen but consumes electricity, directly lowering Faraday efficiency. While efficiency is usually close to 100% in the early stages of system operation, as the membrane ages and develops pinholes or cracks, efficiency may drop below 95%.

More importantly, hydrogen and oxygen react on the catalyst surface to produce hydrogen peroxide and its radicals, and these active species are the culprits causing the cleavage of the perfluorinated sulfonic acid main chain and the detachment of sulfonic groups; this type of chemical degradation is usually more difficult to repair through reverse currents and other operations than electrochemical degradation. Structure determines permeability: quantitative conclusions provided by two 2026 papers.

A study published in ACS Electrochemistry in 2026 quantified this pathway at the structural level. The study established a multi-component reaction-transport framework that considers microstructure, separately analyzing the hydrogen transport pathways in the dissolved phase and the gas phase, and mechanistically linking electrode architecture with safety and performance issues driven by permeation. There are two key conclusions to remember: first, operability is jointly dominated by the cathode catalyst layer and the anode porous transport layer, with the former determining the proportion of hydrogen lost through permeation during generation, and the latter determining the proportion of hydrogen escaping on the anode side; second, within the considered design window, increasing the platinum-carbon content in the cathode catalyst layer can reduce the permeation flux by up to 23%, while increasing the porosity of the anode porous transport layer can reduce the oxygen volume fraction in the hydrogen at the anode tail gas by 0.6 percentage points. The study further integrates these responses into a mapping of cathode pressure versus current density, overlaying safety and efficiency thresholds, and defines two unified indicators to measure the scale and quality of the operational window.

The significance of these numbers lies in their ability to transform a vague issue of hydrogen permeation into variables that can be designed. The platinum-carbon content in the cathode catalyst layer and the porosity of the anode porous transport layer are not fixed parameters; they are engineering variables that can be adjusted at the level of material formulation and structural design. Increasing the amount of platinum-carbon raises the cost of the cathode, so the real engineering decision is to find a three-dimensional balance among permeation flux, precious metal usage, and Faradaic efficiency, rather than merely aiming for the lowest iridium loading. Another layer of constraint comes from safety: the reaction of hydrogen-oxygen mixtures on a platinum catalyst is dangerous, and the mixture becomes explosive at 4%, so relevant standards require that the mixture ratio be kept below 2%. Both increased temperature and higher current density will increase permeation, which means these two operating parameters must be managed together and cannot be independently optimized.

Three general conclusions from the defect list

Placing equipment accidents together with electrochemical research allows us to extract three conclusions that also apply to the domestic green hydrogen equipment industry. First, the distribution pattern of defects is such that the closer to the end of the system, the more likely they are to be overlooked. Specifications of valves and pipe elbows, system cleanliness, and the flow organization at emission points—these aspects, which seem to be part of auxiliary systems, actually play a major role in accident attribution, while the stack itself is often ruled out as a suspect—not because the stack is unimportant, but because the failure modes of auxiliary systems are more complex and more dependent on specific assembly. Second, the acceptance testing phase is the most cost-effective stage to expose defects; if an accident occurs at the factory rather than on site, it is mostly a matter of luck. Incorporating particulate cleanliness, valve selection and configuration, and the flow organization under emission conditions into mandatory design-stage checks is more economical than attempting interception at the end of the production line.

Third, the degradation path often conflicts with the cost-reduction path. Thinning the membrane increases permeability, permeability consumes electrical energy and damages the catalyst layer, and reducing the amount of catalyst decreases tolerance to hydrogen exposure. This chain means that the gains from optimizing a single point will be offset by degradation at the other end of the chain. Therefore, real engineering breakthroughs can only come from a simultaneously closed-loop system design, rather than record-setting parameters in a single component. For the domestic industry, the practical implication of this judgment is: tracking laboratory indicators along one route is of limited significance; what is more worth tracking are the consistency data on the mass production line, failure statistics, and progress in third-party certification. These three sets of data are the true basis for determining whether a technology route can scale.

The references of this article are sourced from public announcements of Clean Power Hydrogen plc, public reports from RenewablesNow and Global Hydrogen Network; research papers published in ACS Electrochemistry and ACS Energy Letters, as well as documents indexed by the U.S. Department of Energy's Office of Scientific and Technical Information; publicly available technical lectures and industry technical analysis materials.