Work

IoT-Enabled Predictive Maintenance for Hydrogen Infrastructure: A Design-Science Framework for Safety, Reliability and Operational Decision Support

Year: 2026
Type: article
Abstract: Hydrogen infrastructure is growing in size and complexity with scale and safety, reliability, leak, compressor degradation, storage integrity and sensor drift/downtime being still significant challenges to dependable operation. The project presented in this paper is developed using a design science and scenario based engineering approach for hydrogen-specific, Internet of Things (IoT) enabled predictive-maintenance framework. The contribution is a conceptual and technical contribution instead of an empirical field implementation, therefore no claim is made with regard to the reduction of downtime, cost or incidents that can be measured in an experiment. It has six layers that are integrated into this framework, namely: assets, sensing, edge processing, secure communications, analytics and maintenance decision support. Also, it implements a clear Hydrogen Asset Health Index (HAHI), a safety-override rule, an asset-failure-sensor-decision matrix, as well as a mapping to the implementation of the hydrogen, functional-safety and industrial-cybersecurity standards. As an application example a medium-scale hydrogen production and refuelling station is considered consisting of an electrolyser, compressor, high pressure storage, dispensers and a safety-detection network. The analysis reveals the possibility of combining hydrogen concentration, electrical measurements, strain, temperature, acoustic emission, vibration and pressure decay with a combination of rule-based logic, physics-based models, machine learning and digital twins for a more timely diagnosis and maintenance based on risk assessment. Those systems are certified and are designated to be used as the primary protection and emergency shutoff device and should not be supplanted by the framework. It is therefore proposed to follow a staged validation roadmap for laboratory testing, fault-injection studies and pilot deployment and multi-site evaluation. The paper characterizes the predictive maintenance as a function of safety and reliability that relies on proven sensing, explainable analytics, cybersecurity, standards and disciplined maintenance practices.
Language: English
Open Access status: hybrid
OpenAlex ID: W7235283756
FWCI:
Cites: 28
Cited by:
Related to: 10