Bayesian Reliability Modeling and Maintenance Recommendations for Cryogenic Hydrogen Pump Cold Ends

Abstract

Downtime and reliability issues are a threat to hydrogen fueling infrastructure and the potential for a carbon-neutral future. One of the leading causes of equipment-related station downtime at liquid hydrogen stations is failure of the cryogenic reciprocating pump cold ends responsible for pressurizing liquid hydrogen out of the storage tank. Reliability modeling using component data can produce robust results that inform component maintenance, design, and operation. With limited cold end data and a need for more rigorous maintenance recommendations, Bayesian methods are a strong candidate. In this work, we define a Bayesian updating methodology for cryogenic reciprocating pump cold end reliability modeling, produce time to failure (TTF) distributions, and recommend preventative maintenance times for each failure mode. The methodology paves the way for continually improved results as new field and test data become available. We demonstrate the methodology for the four most frequently occurring cold end failure modes: abnormal high vapor return, fail to operate, abnormal low output, and external hydrogen leak. Results show how to adapt understanding of equipment performance using Bayesian updating, which will allow for more adaptive and nuanced component decision making as new data are gathered. We provide maintenance recommendations for the cold end failure modes based on the B10, B25, and B50 life and show that the time at which preventative maintenance should occur varies depending on the failure mode. The methodology developed in this work can be adapted for additional hydrogen fueling station components to continually improve the reliability and maintainability of hydrogen fueling infrastructure.