Loss prevention and maintenance modelling for hydrogen-based industry

Hydrogen is increasingly recognized as a key enabler for the decarbonization of energy-intensive industries, particularly where direct electrification is technically limited. However, its deployment introduces specific safety challenges related to its physicochemical properties, interaction with materials, and the limited operational experience of emerging technologies. This PhD thesis develops safety-oriented methodologies to support the reliable and scalable integration of hydrogen systems in industrial contexts, considering safety as a decision-shaping factor across design, operation, and infrastructure planning. Adopting a multi-domain perspective, the research integrates inherent safety indicators into the economic and environmental assessment of hydrogen supply alternatives at component, plant, and district levels. The results show that safety considerations can significantly influence the evaluation of decarbonization strategies, revealing trade-offs that remain hidden in purely techno-economic or environmental analyses. This perspective is extended to industrial hydrogen valleys, where spatial layout and infrastructure sharing are shown to strongly affect safety performance. The thesis also examines hydrogen undesired events through historical accident databases, highlighting that accidents rarely arise from isolated technical failures, but from combinations of technical, organizational, and maintenance-related factors. Maintenance emerges as a critical systemic element, acting both as a preventive safety barrier and, when inadequately managed, as a potential initiating factor. Building on these insights, the thesis proposes a framework linking monitoring, risk assessment, and maintenance planning, supporting risk-based maintenance strategies adapted to hydrogen-specific degradation mechanisms and non-conventional equipment. Particular attention is given to electrolyzer systems through dedicated failure analysis and the definition of Loss of Containment scenarios.