New model harnesses machine learning to efficiently predict hydrogen adsorption in shale gas formations, will save significant time in planning underground hydrogen storage
Hydrogen is a promising clean fuel alternative, but its storage poses problems for large-scale use. There is potential for underground hydrogen storage, but researchers need to be able to predict how much it is possible to store in various locations. To make this process quicker and more accurate, a team of researchers including Khalifa University’s Dr. Shams Kalam, Postdoctoral Fellow, and Dr. Muhammad Arif, Assistant Professor, developed a mathematical equation to predict hydrogen adsorption in shale. Drs. Kalam and Arif collaborated with researchers from King Fahd University of Petroleum and Minerals, Saudi Arabia; and University Teknologi Petronas, Malaysia.
Their results were published in the International Journal of Coal Geology, a top 1% journal for stratigraphy.
“Hydrogen has seen remarkable interest lately from the global energy community as a clean fuel,” Dr. Kalam says. “It is an abundant and renewable energy carrier, addresses future low-carbon requirements, reduces dependence on hydrocarbons, and provides both environmental and strategic advantages. Its use requires storage, and we can store hydrogen underground.”
One of the key strategies in harnessing hydrogen’s potential is underground storage in sedimentary formations — in depleted hydrocarbon reservoirs, aquifers, and even decommissioned wellbores. However, challenges persist. Hydrogen is highly compressible and volatile, demanding large storage volumes and raising concerns about leakage and safety.
The researchers point out that while coal bed methane and shale gas reservoirs have traditionally been exploited for natural gas production, they can also be used to store hydrogen. Shale in particular offers high adsorption rates even at low temperatures, and therefore could store large volumes of hydrogen via adsorption trapping in a safe manner.
Almost 32 percent of global natural gas reserves are shale, which is composed of layers of sedimentary rock and organic material known as kerogen. They could be the ideal storage places for large volumes of hydrogen (though the limitations do exist in terms of injectivity of hydrogen in these formations), but predicting hydrogen adsorption in these formations is a complex and time-consuming task, involving detailed laboratory experiments and molecular simulations.
The research team’s new data-driven model uses gradient boosting regression and data from previous studies to predict hydrogen adsorption on kerogen in shales. The model is informed by various parameters like pressure, temperature, and kerogen density, and offers a quick and accurate estimation of hydrogen adsorption in a potential site. The model was compared with other machine learning methods and proved to be the most accurate, especially for different types of shale.
“Machine learning has been extensively applied to develop a model for the prediction of different processes and mechanisms, but less attention has been given to modeling the adsorption of hydrogen in shale for an easy and accurate estimation,” Dr. Kalam says.
While this model is a significant advancement, it’s important to note that it’s just one piece of the larger puzzle. Other factors like diffusivity, permeability, and geo-mechanical characteristics also play a crucial role in the effective underground storage of hydrogen. Future research in these areas is essential to fully unlock the potential of hydrogen storage in shale formations.
7 February 2024