- Conference Article
- 10.1109/rams50514.2026.11424461
Impact of Hydrogen / Natural Gas Blending on Risk of a Pipeline Distribution System and End-Use Industrial Facility
- Jan 26, 2026
- Theresa M Stewart + 3 more +3
SUMMARY & CONCLUSIONSAs part of the effort to decarbonize California’s energy infrastructure, there is ongoing research on the effects of hydrogen on the safety and performance of existing natural gas transmission, distribution, and end-use equipment. This paper presents an approach to system modeling for a distribution line and connected central power station based on a real system being tested for hydrogen compatibility. In total, this system covers about two miles of distribution pipe that connects to a central turbine which is expected to be blended with hydrogen. The system model examines the risk of loss-of-containment as well as the potential reduction or loss of functionality that could result from adverse effects of hydrogen on each component. This model is built using the hybrid causal logic method, which allows Bayesian networks to be mathematically related to fault trees and event sequence diagrams.For pipe segments, the increased risk of fatigue and fracture failures is modeled according to the API 579 Level 2 Assessment for crack acceptability. The initial flaw size is taken as an uncertain variable, allowing the probability of failure to be assessed according to the expected distribution of flaw size. The effect of hydrogen on the fracture resistance of the pipe was evaluated using a regression model based on historical experimental testing of commonly used pipeline steels in hydrogen environments. For discrete components, such as valves, the expected failure modes and the baseline failure rates for each mode are determined by published historical failure data. The change in each failure mode is modeled as a relative change in the underlying physical failure mechanisms that are most likely to cause that failure mode. These mechanisms are modeled using data driven or physics-based models, such as the one developed for the reduction in steel fracture toughness, where available. Where such models are not available, expert opinion is used to estimate the relative change in risk. The combination of differing information sources is accomplished using a Bayesian network where the operating conditions of individual components can be set as evidence to obtain the risk of all failure modes for that component.The distribution lines and central plant are each considered as a separate phase of the model, where a failure in the distribution line will result in a loss of functionality for the central plant.
Read more