Developing an online composition prediction for an HI–I2–H2O system using deep neural network

• Published in: Chemical engineering science 2024-11, Vol.299, p.120479, Article 120479
• Main Authors: Tanaka, Nobuyuki, Takegami, Hiroaki, Noguchi, Hiroki, Kamiji, Yu, Myagmarjav, Odtsetseg, Ono, Masato, Sugimoto, Chihiro
• Format: Article
• Language: English
• Subjects: Composition prediction; Deep neural network; Hydrogen production; Iodine–Sulfur process; Operating condition control
• ISSN: 0009-2509
• DOI: 10.1016/j.ces.2024.120479

•New composition prediction method for HI–I2–H2O system was proposed.•A deep neural network (DNN) model using measurable properties was applied.•Composition prediction using trend data showed good correlation with measurements.•DNN model analysis provided valuable information for effective composition control. We developed a deep neural network (DNN) method to predict the composition of the HI–I2–H2O system in the iodine–sulfur (IS) process of thermochemical water-splitting hydrogen production using measurable properties. Unlike conventional titration analysis, this approach allows for a quick understanding of fluid composition, providing essential information for controlling operating conditions. This study focused on the HI–I2–H2O three-component system within the IS process. Gibbs’ phase rule was used to construct the DNN model using online measurement parameters, such as temperature, pressure, and density, as input conditions. The model was trained with experimental data, and the structural parameters were tuned. Composition prediction using actual trend data correlated well with titration analysis measurements. The local interpretable model-agnostic explanations method was incorporated to acquire insights into the significance of input parameters for compositions from the DNN model, providing valuable information on crucial parameters for effective composition control.