Artificial Intelligence based Estimation of Optimum Cut point Temperature of a Crude Distillation Unit under Uncertainty

Abstract

In a petroleum refinery, a crude distillation unit (CDU) is a major unit operation that separates the crude oil into its components. The CDU separates the crude oil into components, including light gas streams of naphtha, kerosene, diesel, atmospheric gas oil, and residue. Cutpoint temperatures dictate the quality and quantity of these components, hence need for a complete and fitting model for cut point temperature management originates. The cutpoint temperature majorly relies on crude oil process conditions and composition. The CDU consumes the most extensive amounts of energy. Consequently, significant energy-saving opportunities arise. In this study, Pakistani Crudes from the Zamzama and Kunnar field was used, which was separated into their components via a CDU. This study also examines the effects of uncertainty in process conditions, such as flow rates, pressure, temperature, and feed compositions. A thermodynamic approach was used to carry out the exergy analysis. A hybrid approach based on the Taguchi method and genetic algorithm is used to generate optimum cutpoint temperature for different variants of feed composition and process conditions, generating 228 data sets. An artificial neural networks (ANN) model is developed using these datasets to predict the optimum cutpoint temperature by eliminating the need for the hybrid approach. For the ANN predicted cutpoints up to 38.93% decrease in energy required per kilo barrel of diesel and 8.2% increase in diesel production was observed for Zamzama and up to 18.87% decrease in energy required per kilo barrel of diesel and 33.96% increase in diesel production was observed for Kunnar. In addition to minimizing the E/V values and maximizing flow rates, exergy efficiency was found to be increased by 163.5% and 120% for the Zamzama and Kunnar crude respectively.