Prediction of hydrogen production process by glycerol steam reforming using machine learning models

Elham OmidbakhshAmiri,
Prediction of hydrogen production process by glycerol steam reforming using machine learning models,
International Journal of Hydrogen Energy,
Volume 175,
2025,
151461,
ISSN 0360-3199,
https://doi.org/10.1016/j.ijhydene.2025.151461.
(https://www.sciencedirect.com/science/article/pii/S0360319925044635)
Abstract: Hydrogen is a green fuel with low emission and high efficiency. So, recent researches focused on the hydrogen production and optimization of its process. Different feeds for hydrogen production are considered. Renewable raw materials with their benefits are used as suitable feed. Glycerol, as a byproduct of biodiesel production, is one of these feeds. In this work, hydrogen production by glycerol is studied using machine learning models (support vector regression (SVR) and Gaussian process regression (GPR)) to prediction the weighted global warming potentials (WGWP) of the process and H2 production rate per glycerol feed flow rate. In comparison between the results of GPR and SVR models, results show that GPR-matern model presented the least Mean absolute error and Root Mean Squared Error and highest R-Squared for prediction of normalized WGWP and H2 rate. The temperature of reforming reactor is the most effective parameter on H2 rate, while, the glycerol molar flowrate is most important parameter on WGWP.
Keywords: Glycerol; Hydrogen; HYSYS; Machine learning; Global warming potentials