Advances in Design, Optimization and Control of Semicontinuous Distillation Processes

Abstract

Semicontinuous distillation is a process intensification technique for purification of ternary mixtures to desired specifications. Where conventional continuous process requires two distillation columns for ternary separation, a semicontinuous process utilizes only one distillation column and integrates it with a simple storage tank (called middle vessel) to perform the same task. Therefore, with a lower total direct cost, a semicontinuous system has a lower total annualized cost (TAC) for low to intermediate production rates compared to the continuous configuration. However, the operating cost of the semicontinuous system is higher than the continuous counterpart. The objective of this thesis is to improve the economics of semicontinuous distillation by reducing the operating cost and the TAC of the process. This study investigates potential enhancements of the process by modifying the design, optimizing the design parameters and implementing more efficient control strategies on the process. In this work, a novel intensification technique for purification of ternary mixtures is proposed. The process can purify three components to desired purities in a single distillation column without the necessity of a middle vessel and is called semicontinuous without middle vessel (SwoMV). The proposed configuration reduces the side stream recycling and consequently lowers the operational and the direct costs of a semicontinuous process. Subsequently, the integration of design and control of semicontinuous processes is studied. A methodology is presented to simultaneously obtain locally-optimal structural and operational parameters of the system to minimize the TAC of the process. A mixed integer dynamic optimization (MIDO) problem is formulated and the outer approximation (OA) and the particle swarm optimization (PSO) methods are used to solve the problem. Finally, for the first time, the implementation of model predictive control (MPC) on the semicontinuous process is studied. The subspace identification method is adopted to identify a linear state-space model. Subsequently, a shrinking horizon MPC is implemented on the system to reduce the operating cost of the process while maintaining the desired product purities by the end of the cycle.