ENHANCING HYDROGEN ENERGY STORAGE AND TRANSPORTATION SYSTEMS: PREDICTIVE MODELING OF SEALING MATERIALS THROUGH EX-SITU EXPERIMENTAL ANALYSIS

Abstract

As the global energy trend shifts towards cleaner alternatives, hydrogen arises as a prospective candidate due to its high energy density and zero carbon emissions. However, significant challenges persist, particularly in the durability and reliability of materials used for hydrogen storage and sealing. This study explores the challenges and opportunities associated with hydrogen production, incorporating CO2 capture technology and the interaction of materials to understand the correlation between materials and hydrogen exposure. This dissertation presents a literature review and experimental analysis focused on hydrogen-induced aging in polymeric sealing materials, specifically nitrile butadiene rubber (NBR), ethylene propylene diene monomer (EPDM), and silicon. The research reviewed hydrogen-induced aging mechanisms, including thermal, chemical, mechanical, and thermo-mechanical aging, which leads to the loss of integrity of sealing materials. A series of controlled experiments were conducted to simulate real-world conditions, with the aim of assessing the effects of high-pressure hydrogen exposure on the physical, mechanical and electrical properties of NBR, EPDM and silicon. Findings from the experimental results indicate a complex interaction among swelling, hardness reduction, and tensile strength degradation in sealing materials under hydrogen exposure. Tensile strength degradation was strongly associated with crack formation. At the same time, Silicon exhibited the lowest swelling due to its Si-O-Si backbone but showed significant reductions in tensile strength and hardness and high crack and cavity counts. EPDM displayed moderate swelling and hardness loss, coupled with strong leak resistance and minimal morphological degradation, due to its saturated backbone and low hydrogen solubility. Despite the higher hydrogen solubility and free volume, NBR retained hardness better than Silicon but exhibited moderate morphological degradation that weakened the leak resistance over time. NBR’s absolute resistivity was found to remain much lower than that of Silicon and EPDM with hydrogen exposure. Both EPDM and NBR maintained comparable sealing performance up to 1500 psi, however, EPDM’s resilience against further degradation under prolonged pressures suggests it as the optimal elastomeric material for consistent sealing under high-pressure hydrogen system. Several machine learning approaches were applied to predict key sealing properties of materials exposed to hydrogen. This study employed multiple models like Multivariate Linear Regression (MLR), Generalized Additive Model (GAM), and Random Forest to capture linear and non-linear dependencies between pressure and sealing properties. It was observed that MLR performed better handling linear relationships, especially for tensile strength and hardness, while GAM performed better in capturing non-linear resistivity behavior. A CNN model is trained on SEM images to capture morphological changes, showing the lowest MAE for Silicon. The final recommendation is a combined MLR-CNN model, which balances numerical and morphological predictions for optimal material property assessment under hydrogen exposure, suggesting EPDM for high-pressure applications. This research is expected to contribute to the life cycle estimation and customization of elastomeric materials for hydrogen infrastructure. That will assist in accurately improving the scheduling of e the predictive maintenance of the seals used in the hydrogen industry. Thus, this research will help the industry to meet ISO 15869 and the ASME B31.12 through the developed predictive models for seal failure and performance in hydrogen pipelines. It provides a foundation for future research in sealing technology, aimed at enhancing the long-term safety and performance of hydrogen storage and transportation systems.