Hydrogen blending in natural gas systems has emerged as an important decarbonization strategy, but introduces potential new risks to the components in these systems. Much research has gone into quantifying these effects for metallic materials such as those used in pipelines, but significant gaps remain in data and modeling for polymeric materials.
This work models the effects of hydrogen on polymers and elastomers commonly used in seals and gaskets on natural gas components. A Bayesian network is used to define relationships between operating conditions, material properties, and failure mechanisms. Using a Bayesian network enables the use of a combination of deterministic models, regression from data, and expert opinion to quantify relationships depending on the availability of data and knowledge.