A framework for application of genetic algorithm to model-based design of reactive distillation process

In the present study, we investigate a framework for the application of genetic algorithm (GA) methods to reactive distillation (RD) process design. First, a computational method for hybrid simulation is proposed by linking a commercial process simulator with an in-house GA-based optimizer in order to accelerate the search for process designs showing high performance. When developing the hybrid simulation system, detailed methods for the GA operations of crossover, mutation, and selection were investigated to achieve a combination that maximizes the effectiveness of the simulation framework. Next, we investigate the designs obtained from the simulations for a case study, an RD process for acetyl acetate. The proposed GA method is shown to provide not only the lowest total annual cost (TAC) design, but also various alternative design solutions. Since the generation of alternative designs lets us consider many other factors, e.g., the distillate composition and the overall conversion of the product, we believe that the application of techniques such as multi-niche crowding (MNC) GA could be effective for searching simultaneously structural and operational conditions of multifunctional processes.