Multi-objective process design in multi-purpose batch plants using a Tabu Search optimization algorithm

Abstract

Chemical batch processes are typically used for the production of speciality chemicals and pharmaceuticals. Due to the still growing importance of this type of processing, design methods are required that take into account the special requirements and constraints in the corresponding production facilities. We developed a method that optimizes the design of a single chemical process to be implemented in an existing multi-purpose batch plant, in which a well-defined set of equipment units is available for realizing this process. In the optimization, three objectives with different priorities are considered. A flexible metaheuristic algorithm, Tabu Search (TS), has been implemented to solve this multi-objective combinatorial non-linear problem. We started from a basic form of TS to determine the effectiveness of this version as well as establish the relative strengths and weaknesses of first level TS strategies. Our investigation includes a thorough examination of algorithm parameters and of implementation issues to identify algorithm settings that can handle the whole class of problems considered. Overall, we concluded that the basic form of TS—using fixed default settings—exhibits highly attractive performance features for solving the problems at hand. Moreover, comparison with a multi-start steepest descent algorithm shows that a basic TS approach conducts a global search more effectively. As illustrated by three case studies, the new method is well suited for identifying optimal designs of a chemical process to be implemented in an existing multi-purpose batch plant. The approach is particularly suited for considering multiple prioritized objectives and for enabling the use of external (e.g. commercial) batch process simulation software as a black-box model for the process evaluations.