Loading...

Accelerating Discovery of MOFs for Hydrogen Storage via Machine Learning in Energy Related Applications

Fallah Kheirandish, Saeed | 2025

0 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 58055 (06)
  4. University: Sharif University of Technology
  5. Department: Chemical and Petroleum Engineering
  6. Advisor(s): Ghotbi, Sirus; Lotfi, Marzyeh; Larimi, Afsaneh; Asgharinezhad, Ali
  7. Abstract:
  8. The quest for sustainable energy solutions has positioned hydrogen as a pivotal alternative fuel due to its clean emission profile, producing only water vapor and heat. However, efficient hydrogen storage remains a formidable challenge due to its low energy density at standard conditions. Metal-Organic Frameworks (MOFs) offer a promising solution due to their high surface area, tunable pore structures, and reversible adsorption properties. This study leverages computational screening and machine learning (ML) methodologies to identify high-capacity MOFs for hydrogen storage. We utilized a large dataset of known and hypothetical MOFs from comprehensive databases and employed advanced neural network models—Feed-Forward Neural Network (FNN) and Pattern Recognition Neural Network (PRNN)—optimized using Equilibrium Optimizer (EO) and Genetic Algorithm (GA). The models were calibrated against Grand Canonical Monte Carlo (GCMC) simulations to predict gravimetric and volumetric hydrogen storage capacities. Results indicated that features such as pore volume (PV) and void fraction (VF) are strongly predictive of hydrogen uptake, aligning with established trends in gas adsorption studies. The FNN model demonstrated superior predictive accuracy for gravimetric capacity, while the PRNN excelled in capturing complex patterns for volumetric capacity. A subset of top-performing MOFs was identified, surpassing benchmarks such as MOF-5 in hydrogen capacity. This approach underscores the synergy between computational screening and ML in advancing the discovery of efficient hydrogen storage materials and informs future experimental prioritization
  9. Keywords:
  10. Hydrogen Storage ; Metal-Organic Framework ; Artificial Neural Network ; Grand Canonical Monte Carlo (GCMC)Simulation ; Database ; Absorptive Capacity ; Machine Learning

 Digital Object List

 Bookmark

...see more