(343e) Advancing the State of Hydrogen Systems Quantitative Risk Assessment: Design and Requirements of a Hydrogen Component Reliability Database (HyCReD)

Safety and reliability have long been recognized as key issues for the development, commercialization, and implementation of technologies or infrastructure. Reliability engineering and risk assessment play a key role in proactively addressing safety – before problems happen – and help us learn from problems if they happen. A safe and sustainable transition to expanded hydrogen usage and new applications requires that the risks associated with it be proactively and rigorously investigated, quantified, and mitigated – at the design and early deployment stage when it’s cheapest and easiest to engineer-in these features. With hydrogen technologies expected to play a key role in the near future to enable decarbonization of several sectors including energy storage and transportation, the time to address these issues is now. For hydrogen fueling stations, quantitative risk assessment (QRA) is an important tool for enabling their safe deployment and has been increasingly embedded in the permitting process. However, most QRAs to date have used data from non-hydrogen specific data sources. QRAs built on hydrogen-specific component and system reliability data will provide more robust insights, but these data are lacking, thereby hindering the development of necessary safety codes and standards. With 57 operating stations in the United States as well as increasing deployments globally, hydrogen-specific reliability data can be collected and used to update QRA tools.

QRA and reliability data can refer to various types of data depending on the specifics of the analysis being performed. A review of risk and reliability analysis for hydrogen storage and delivery by Moradi and Groth (2019) outlined 8 types of data used in hydrogen QRA [1]. This covers the data needed to perform all steps of a QRA from hazard and exposure identification to frequency and consequence analysis. In another work, West et al.[2] reviewed four existing hydrogen safety data collection tools: H2Tools, Hydrogen Incident and Accident Database (HIAD), and the National Renewable Energy Laboratory’s (NREL) Composite Data Products (CDPs). This review found that several elements are systematically missing from current hydrogen data collection practices, including collecting system-level data (component configuration and population), frequencies and component failure data. These elements are necessary to generate failure rates for components of a system, which are a fundamental aspect of QRA and reliability engineering.

To address this gap, a multi-year effort involving a number of research activities is being undertaken at the University of Maryland and NREL with the goal of developing a Hydrogen Component Reliability Database (HyCReD). Based on a review of 10 reliability engineering textbooks and database documentation and guidelines from analogous industries, we developed a set of 23 requirements covering the database characteristics and types of data (static and event) to be collected. The requirements were then used to start an iterative process of identifying and defining the data elements and fields to be part of the database structure. We reviewed the data fields recommended by the references consulted for the developing the requirements and incorporated additional information from two handbooks on reliability data collection in the oil & gas and chemical industries[3], [4]. These references were augmented by monthly discussions with a team of four experts with a total of 70 years of expertise in the field of reliability engineering and hydrogen technologies, and feedback from presentations at academic and industrial conferences as well as multiple internal presentations to stakeholders and potential data providers.

Based on these requirements, we define the structure of the HyCRED database and 25 data elements to be collected, spanning system description, failure, shutdown, or near-miss events, and maintenance events. The data elements are then defined according to international standards used in the safety and reliability practice and potential choice lists are provided for each field. System information refers to the data fields that characterize the facility from which the report originated, its operational environment, and the context of the event being reported. The equipment hierarchy elements are to be completed using drop-down lists from the generic hydrogen station component hierarchy developed by West (2021) [5]. Data elements describing the reported events include date/time, failure mode, mechanism, root cause(s), and severity. Additional event data elements include hydrogen release, release size, accumulation, and ignition are included to support calculating hydrogen-specific leak rates and ignition probabilities in the future.

Since HyCReD is being piloted for hydrogen fueling stations, a generic component hierarchy for these stations based on a representative set of six publicly-available, geographically-varied station designs was developed by West (2021) [5]. This approach combined functional decomposition with service-based decomposition to capture the function of the system as well as all of the necessary components for hydrogen reliability data. Based on the piping and instrumentation diagrams (P&IDs) reviewed, the generic component hierarchy developed detailed 6 subsystems and 21 functional groups with potential components under each functional group. This blended decomposition allows operators, maintenance personnel and reliability engineers to understand component failure in the context of system functions and operating conditions. This is critical to perform process-based risk assessment with a full understanding of system behavior and interactions between system functions, components, and system response as a whole.

Furthermore, a failure mode taxonomy was constructed from an analysis of seven publicly available failure modes and effects analyses (FMEAs) and failure mode sources. While no publicly available FMEAs for hydrogen fueling stations were found, FMEAs for similar systems such as hydrogen fuel cell vehicles and adjacent industries such as oil and gas were consulted. The resulting 33 failure modes and their definitions are initially validated through discussion with a team of five reliability engineering experts and through feedback from presentations as discussed earlier in this section.

To demonstrate the viability of the proposed database, we sought to populate it based on available narrative descriptions of events that occurred in hydrogen fueling stations within the last 10 years. Five events were selected from H2Tools Lessons Learned, the Hydrogen Incident and Accident Database (HIAD), and an incident where we had direct access to the data provider. We determined that almost all of the system data could be readily obtained from the event descriptions. However, it was clear that information on component nominal working pressure, population and time in service (determined from installation date) were not generally reported in these sources. Further, we demonstrated that the component hierarchy was useful in specifying the subsystem, functional group and component that failed for each event. Developing component hierarchies for other hydrogen technologies (e.g., electrolyzers, vehicles, power generation systems) will be key to expanding HyCReD to these systems.

Through the HyCReD structure, we present a major step towards achieving the goal of having hydrogen component and system reliability data to support QRA, safety analysis, reliability analysis, and maintenance planning. The results generated from HyCReD can provide participants and stakeholders with tangible generating insights that enable prioritizing maintenance activities and targeting components with the highest failure probabilities, thereby eliminating costly downtime. This leads to improved station uptime, higher number of fills, and ultimately greater economic returns. Further, these insights lead to improved station designs for the future and enable targeted component research and development activities improving the most failure-prone components. For the wider hydrogen community, HyCReD will generate results that enable risk assessment models that can be used to help establish safety codes and standards. A proactive treatment of safety and reliability can help industry avoid overregulation and facilitate the wider adoption of profitable hydrogen technologies.