Proposals are expected to develop and integrate innovative materials, cell, and stack and balance-of-plant configurations, including connections, intelligent monitoring/control tools that can detect, and reduce or eliminate the risk of hazardous gas crossover, and inherently safer solutions that prevent hydrogen leaks and build-up of critical concentrations in the module. This includes but is not limited to: next-generation membrane materials with reduced gas crossover, hydrogen permeability, and improved mechanical integrity; novel electrode structures that enhance gas separation; architectures that reduce the potential leak points and physically or operationally decoupled hydrogen and oxygen evolution; novel stack and balance-of plant components integrating efficient H-O recombination catalysts. Novel and advanced optical and spectroscopic techniques for real-time, on-line monitoring of hydrogen purity can be proposed as an integral part of the system’s monitoring and control architecture. These tools can significantly reduce the risk of in-situ cell breakdown while simultaneously supporting an increased number of safe start-up/ shutdown cycles. In parallel, failed components should undergo advanced experimental analysis to identify underlying damage mechanisms and material degradation states. These insights will feed into a dedicated numerical tool—coupling finite element modelling, degradation kinetics, and operational data—to simulate, predict, and optimise component performance under varying conditions. This model should support both real-time decision-making and early-stage design improvements to enhance durability and intrinsic safety. Complementary sensing technologies—such as electrochemical and thermal conductivity sensors—may also be integrated to ensure data redundancy and robust fault validation. Sensor data streams should feed into AI/ML-based models for early anomaly detection, predictive maintenance, and optimised system response strategies.

In parallel with materials, components and hardware development, the topic also encourages the advancement of smart sensing and control solutions to ensure safe operation in real-time. These may include AI- or machine learning-based systems, ideally embedded within a digital twin framework that integrates real-time sensor data with numerical models. Such models can simulate and predict system behaviour under varying conditions, enabling early detection of faults such as membrane failure, electrode delamination, or abnormal thermal and pressure events. Spectroscopy-based diagnostics may further enhance this architecture by providing high-resolution insights into critical degradation processes. Long-term degradation modelling should be combined with embedded diagnostics to support predictive maintenance, reduce unplanned downtime, and extend operational lifetimes. Emphasis should be placed on the performance of these tools under challenging dynamic conditions—including intermittent renewable energy supply—to replicate real-world operating environments (TRL5).

Proposal should validate the proposed solutions. Testing should be carried out at the component, cell, and stack level under relevant conditions (e.g. pressure, temperature, power cycling), with clear metrics for safety, performance, durability, and regulatory compliance. The safety improvements provided by the proposed solutions should be evaluated for their beneficial effects on risk management procedures. Targeted prototype scale and cell size should be appropriate for the considered technology and future scale-up.

The proposal should demonstrate at the end of the project the construction and validation on a stack with the following requirements:

• PEMEL: minimum 100 kWel designed to operate at >100 bars of output pressure. The stack should exhibit a minimum operation performance of current densities > 3.0 A/cm² at <1.9 V.
• AEMEL: minimum 50 kWel designed to operate at >50 bars of output pressure. The stack should exhibit a minimum operation performance of current densities > 1.5 A/cm² at <1.85 V.
• AEL: minimum 100 kWel designed to operate at >30 bars of output pressure. The stack should exhibit a minimum operation performance of current densities > 1 A/cm² at <2 V.
• Other emerging low temperature electrolysers: minimum 5 kWel designed to operate at >30 bars of output pressure. The stack should exhibit a minimum operation performance of current densities > 1 A/cm².

Stacks should be validated for performance and safety for a minimum of 1000 h under diverse operating regimes (steady-state, dynamic load-following, frequent start/stop cycles, and off-normal transients), with results reported under harmonised EU protocols (see below).

Additional KPIs may be proposed, in particular for non-conventional architectures (e.g., decoupled designs), provided that key safety and performance KPIs are fulfilled. Wherever possible, testing should adopt or contribute to harmonised EU protocols and pre-normative research efforts. Proposals are encouraged to liaise with standardisation bodies (e.g., CEN, CENELEC, ISO) and relevant regulatory stakeholders to ensure compatibility with emerging safety frameworks and certification pathways. This alignment is critical for ensuring that innovations move beyond the laboratory and into safe, deployable commercial systems.

Projects are also expected to contribute to the definition or refinement of safety-relevant KPIs, beyond traditional efficiency and cost metrics. These may include indicators such as crossover detection sensitivity, response time of safety shut-off systems, operational uptime due to preventive maintenance, leak probabilities, or compliance with forthcoming regulatory thresholds on gas purity and leakage. KPIs should be integrated in a comprehensive safety-by-design evaluation of the proposed solutions both at component and at system level. Where possible, KPIs should align with EU safety standards and be backed by sensor-based data to support reliable validation and comparison across systems.

To address the full complexity of the safety challenge, proposals should adopt a multidisciplinary approach and involve actors across the electrolyser value chain. This may include component manufacturers (membranes, electrodes, sensors), electrolyser OEMs, digital technology providers (AI, modelling, control systems), testing laboratories, and certification or regulatory entities.

Applicants should clearly articulate the added value and innovation of their proposed approach relative to the state-of-the-art . Projects should also reference, complement and build on existing European initiatives (e.g. European Hydrogen Safety Panel) and projects (e.g., REFHYNE, HYScale, DELYCIOUS, INSIDE, , PEACE, HYPRAEL, ADVANCEPEM and projects funded under Topic HORIZON-JTI-CLEANH2-2023-01-01), and demonstrate how they build upon and complement the results of ongoing JU projects. Duplication of effort should be avoided, and synergies with parallel EU or national initiatives should be identified. In particular, while predictive maintenance tools have previously been explored with a focus on performance and lifetime, their integration here plays a critical role in enabling the early detection of safety-relevant failures, thereby reinforcing the complementarity between the two project scopes.

For activities developing test protocols and procedures for the performance and durability assessment of electrolysers proposals should foresee a collaboration mechanism with JRC (see section 2.2.4.3 “Collaboration with JRC”), in order to support EU-wide harmonisation. Test activities should adopt the already published EU harmonised testing protocols to benchmark performance and quantify progress at programme level.