Assessment of the estimation methods for the normal boiling point of liquid organic hydrogen carriers

Abstract

Liquid organic hydrogen carrier (LOHC) presents a promising solution for hydrogen storage. However, its development has been impeded by the absence of normal boiling point (NBP) data for hydrogenation products, especially saturated alicyclic compounds. This gap spurs the search for a reliable and convenient estimation method to aid in the screening of new mediums and the design of chemical processes. Thanks to the intrinsic simplicity, 17 group contribution models were selected and evaluated to identify the optimal ones for different types of alicyclic compounds. The best-performing model was subsequently refined. Furthermore, two models tailored for predicting the NBP of saturated alicyclic compounds were proposed using classical approaches and artificial neural networks. The results indicate that the GCN and GM models yield lower errors in estimating the NBP of alicyclic hydrocarbons, while the GIC2 model performs better for heterocyclic compounds. The developed models are more accurate compared to the existing models. Notably, the neural-network-based model exhibits excellent performance. These models are expected to provide useful insights for large-scale screening of LOHCs for hydrogen storage.