White-box methodologies for achieving robust correlations in hydrogen storage with metal-organic frameworks.

Hydrogen is recognized as a clean energy replacement for non-renewable fossil fuels, and the utilization of metal-organic frameworks (MOFs) for hydrogen storage has gained considerable interest in recent years. In this study, hydrogen storage in MOFs was estimated using white-box methods, namely group method of data handling (GMDH), genetic programming (GP), and gene expression programming (GEP), which are robust soft-computing methods known for generating innovative correlations. To this end, temperature, pressure, pore volume, and surface area were implemented as input parameters for constructing these robust correlations. After that, the superiority of the established correlations was demonstrated through multiple statistical and graphical error assessment. The results indicated, the GMDH model demonstrates the highest accuracy with root mean square error (RMSE), and mean absolute error (MAE) values of 0.410 and 0.307, respectively. However, the GEP model's accuracy was comparable to that of the GMDH model. In addition, sensitivity assessment showed that the pore volume and the pressure exhibit the strongest linear and non-linear relationships, respectively, with the H storage in MOFs. This was demonstrated by a Pearson correlation coefficient of 0.5 and a Spearman correlation coefficient of 0.56, respectively. Furthermore, temperature had a minimal negative impact on the H storage in MOFs according to Pearson, Spearman, and Kendall coefficients. Finally, to confirm the findings of the GMDH model, the leverage approach was applied, demonstrating that 96% of the data falls within the acceptable region, confirming the statistical reliability of the developed models.