Artificial neural network model for process monitoring of crude oil distillation column in a petroleum refinery.

This article focuses on the application of artificial neural networks (ANN) for process design and monitoring of a crude distillation unit (CDU) in a petroleum refinery, specifically using operational data from the Port Harcourt Refining Company. The study employed 230 experimental data sets, incorporating both controllable and uncontrollable input variables, to train and validate ANN models with architectures of 14-1-7 for CDU design and 13-1-6 for the neural network controller, using a logistic sigmoid transfer function. Results demonstrated that the ANN models achieved low mean absolute error (MAE) and mean square error (MSE) values, indicating high prediction accuracy for CDU output parameters such as distilled temperatures and flow rates. The study concludes that ANN is an effective machine learning tool for CDU process monitoring, while recommending future research to integrate energy consumption optimization and cost analysis.