Optimisation of Vacuum Distillation Units in Oil Refineries Using Surrogate Models

Abstract

To ensure the optimum performance of downstream processes in a petrochemical refinery, the operation of a vacuum distillation unit (VDU) is critical. It needs to satisfy the yield and quality requirements of the downstream process. Otherwise, it will result in a loss of profitability in the refinery. Hence, it is important to optimise the operation of the VDU to ensure optimum performance. Traditionally, VDU is operated within the design envelope, and its operation condition is fine-tuned based on the operator’s experience. However, such action does not guarantee the optimum performance of the entire refinery as it only considers the operation of VDU without understanding the effects towards downstream processes. Therefore, this work presents a framework to optimise VDU operations with consideration of the downstream processes. The framework consists of process simulation, surrogate modelling, and multi-objective optimisation. The developed framework aims to determine trade-offs between high vacuum gas oil (HVGO) yield and total annualised cost (TAC) of a refinery that considers the needs of downstream operations. In this work, crude oil blending ratio, furnace outlet temperature, flash zone temperature, column top pressure, column bottom pressure, stripping steam flowrate, HVGO pump-around flowrate, and light vacuum gas oil (LVGO) pump-around flowrate of the VDU are to be optimised. Based on the optimised result, the heavy-light crude blend achieves higher HVGO yield and lower TAC, and the optimised results were validated with the simulation results via Aspen HYSYS. The proposed methodology was proven to have accurate estimations of the VDU operation in the process simulation environment. Moreover, the optimised results can provide insight into the optimal process conditions of VDU for the refiners. With this insight, effective operating strategies can be developed to overcome the limitations present in real VDU operations.