Prediction of phase equilibria of HIx system using artificial neural network: Experimental verification

Thermochemical sulfur–iodine (SI) cycle is one of the promising technologies investigated for hydrogen production using solar and nuclear energy. The development and validation of a reliable thermodynamic model for the HIx mixture (HI–H2O–I2) encountered in the SI cycle have been identified as a central research issue to provide estimations on the HIx section energy demand.In this contribution, we develop an artificial neural network (ANN) model to predict the real time phase equilibrium behavior. For the binary HI–H2O system, the ANN model is constructed for a pressure up to 84 bar, while for the ternary HI–H2O–I2 system, the model describes the equilibrium behavior for a pressure up to 53 bar. The proposed models show their potential with a maximum relative deviation (RD) of about 2.5% and a root mean square percentage error (RMSPE) of within 0.9% for binary, and a maximum RD of 3.6% along with an RMSPE of 0.64% for ternary systems.

► This work aims at developing an empirical phase equilibria model for the HIx system. ► The system is predicted and validated by the artificial neural network model. ► For further improvement, more experimental efforts are needed.