Optimal operation of parallel distillation systems with multiple product grades: An industrial case study

Abstract

In the fine chemical industry, customers often demand different grades with different purity specifications. To achieve the best performance, the production tasks should be assigned to different distillation columns at the most suitable operating conditions and time periods. In this paper, an optimal scheduling method is presented through an industrial case study with multiple products and parallel distillation columns. Rigorous nonlinear models are built for each distillation column and validated with plant data, based on which, a reduced-order model is obtained with data of optimal operating points at various conditions. The reduced-order model is then incorporated into a mode-based discrete-time mixed integer linear program (MILP) scheduling model, where transitions between different operating modes are specified based on plant data. The MILP-based scheduling is applied to a real-word industrial case study to demonstrate its computational efficiency and effectiveness in improving economic performance with comparison to two heuristic scheduling methods.