AI-based prediction and interpretation of column head temperature in a continuous distillation: a case study on methylcyclohexane separation

Abstract

Abstract This study explores the application of artificial intelligence (AI) to predict the column head temperature in a continuous distillation process designed to separate methylcyclohexane from a binary mixture containing 23 % methylcyclohexane by mass. Several AI models were developed and evaluated, using key operational parameters such as heating power, reflux ratio, feed rate, pressure drop, and boiler temperature as input features. The column head temperature served as the target variable, representing the performance of the distillation system. Among the models tested, the Decision Tree Regressor achieved the best performance, with a mean absolute error (MAE) of 0.0090, a root mean square error (RMSE) of 0.0258, and a coefficient of determination R 2 of 0.9572. To enhance model interpretability, SHapley Additive exPlanations (SHAP) analysis was applied, revealing that reflux ratio and boiler temperature are the most influential variables. These results demonstrate the model’s effectiveness in predicting the normal operation of an automated continuous distillation column. Furthermore, the model shows potential for real-time implementation, offering a promising approach for online monitoring and fault detection in industrial distillation processes.