Optimization of hybrid dynamic/steady-state processes using process integration

Abstract

Much research in the area of process integration has focused on steady-state processes. However, there are many process units that are inherently unsteady-state or perform best when operated in an unsteady-state manner. Unsteady-state units are vital to chemical processes but are unable to be included in current process optimization methods. Previous methods to optimize processes containing unsteady-state units place restrictions or constraints on their use. This optimization still does not give the best system design because the solution found will only be the best out of the available options which likely excludes the true optimal design. To remedy this, a methodology was created to incorporate unsteady-state process units into process optimization analysis. This methodology is as general as possible. Unlike many existing unsteadystate optimization methods, it determines all three main components of process design: the network configuration, sizes of units, and operation schedule. This generality ensures that the truly optimal process design will be found. Three problems were solved to illustrate the solution methodology. First, a general mass exchange network was optimized. The optimization formulation resulted in a mixed-integer nonlinear program, and linearization techniques were used to find the global solution. A property interception network was also optimized, the first work done using property integration for systems with unsteady-state behavior. Finally, an industrial semi-batch water purification system was optimized. This problem showed how process integration could be used to optimize a hybrid system and gain insights into the process under many different operating conditions.