Development of an Intelligent Framework for Selecting Underground Hydrogen Storage Sites

According to a report by the Public Relations Office of the University of Tehran, this research was conducted by Fatemeh Hashemi, a master's student at the University of Tehran, under the supervision and guidance of Dr. Behnam Sedaee, a faculty member of the School of Chemical Engineering at the College of Engineering, University of Tehran, and in collaboration with Dr. Mohammadreza Rasaei, an associate professor at the College of Engineering, University of Tehran. The research was carried out within the framework of the activities of the University of Tehran's Underground Hydrogen Storage Research Core (UT-UHS) and presents an integrated framework for identifying and ranking suitable options for underground hydrogen storage in depleted gas reservoirs.

Dr. Sedaee, the head of the research core, said regarding this project: "The Underground Hydrogen Storage Research Core of the University of Tehran (UT-UHS), at the College of Engineering, as one of the world's leading research units in the field of hydrogen, systematically studies underground hydrogen storage. The main objective of this core is to develop technical knowledge for the safe and cost-effective storage of hydrogen in depleted oil and gas reservoirs, saline aquifers, and salt caverns."

Stating that underground hydrogen storage is considered one of the important options for long-term energy storage and for supporting the development of renewable energy sources, Dr. Sedaee remarked: "Hydrogen storage in depleted gas reservoirs, due to its available infrastructure and suitable technical and economic characteristics, is regarded as one of the practical options for large-scale hydrogen storage. However, selecting a suitable site from among a large number of potential reservoirs is a complex and multi-dimensional problem that must simultaneously take technical and economic criteria into account."

The head of the research team further emphasized: "One of the main challenges in the site selection process is the high cost and time required for numerical simulations. On the other hand, some multi-criteria decision-making methods rely on expert opinion-based weighting, which can introduce subjective biases into the ranking process. In this research, an attempt has been made to address these two limitations by combining artificial intelligence models with a data-driven multi-criteria decision-making method."

He added: "In the first stage, using numerical simulations and the Latin hypercube sampling method, a comprehensive database of various scenarios was created. Then, five machine learning algorithms were evaluated for predicting reservoir performance indicators, including hydrogen recovery factor, hydrogen purity, and economic index. Subsequently, using the Shannon entropy method, the weights of the criteria were determined in a data-driven manner, and finally, the TOPSIS method was employed to rank the storage options."

Fatemeh Hashemi, the student researcher of this project, also said: "The results showed that the LightGBM model was able to predict all three main performance indicators—including recovery factor, hydrogen purity, and net present value (NPV)—with very high accuracy, and its coefficient of determination for these indicators was obtained at more than 0.98. This level of accuracy makes it possible to use AI-based surrogate models for screening a large number of scenarios."

Regarding the significance of these findings, she added: "One of the important advantages of the proposed framework is the significant reduction in the time required for the screening process. According to the research results, the proposed framework is capable of providing the final ranking of options within a few hours, whereas approaches based on full reservoir simulation and decisions relying on expert judgment may require several weeks."

The associate professor of the University of Tehran further remarked: "This research demonstrated that the proposed framework, by transforming complex simulation results into a transparent and data-driven process, can enable the screening and ranking of a large number of depleted gas reservoirs prior to conducting detailed studies. From this perspective, the use of this method can help identify more suitable options, reduce financial risk, and prevent the allocation of capital to unsuitable options in the early stages of developing underground hydrogen storage projects."

In conclusion, Dr. Sedaee said: "The combination of artificial intelligence with data-driven multi-criteria decision-making methods can provide a transparent, reproducible, and scalable framework for selecting underground hydrogen storage sites. By increasing the speed of the screening process and reducing dependence on subjective judgments, this approach has the capacity to be used as part of the evaluation process for industrial-scale underground hydrogen storage projects."

It is worth noting that the achievements of this team, resulting from collaboration with reputable international universities including the University of Calgary in Canada and the University of Arkansas in the United States, have so far been published in the form of more than ten articles in prestigious international journals. Furthermore, one of the most important hydrogen projects in the country was carried out by this core with the support of the Iranian Gas Engineering and Development Company.

https://www.sciencedirect.com/science/article/pii/S2949891026004252