Integration of Synthesis and Operational Design of Batch Processes

This thesis describes the development of a methodology that targets the synthesis problem for batch processes and additionally handles the operational design for batch reaction, distillation and crystallization. The lack of available synthesis approaches has led to the use of simple and conservative routes that have proved to work in the laboratory, but can significantly profit from optimization even at the synthesis level. Rule-based algorithms are developed that work at two communicating levels. Given the reaction sets that lead from raw materials to product and the purity and yield specifications for the latter, the upper level algorithm identifies and places the necessary separation tasks in the generated batch route. The most feasible separation technique is also identified for the separation task by exploiting the relationship between physio-chemical properties and separation process principles. The developed algorithm provides also initial estimates for the intermediate objectives of each task in the batch route. At the lower level, three algorithms are developed with the intention of providing a feasible and near optimum batch operations model for several batch processes. Motivation is found in the recognition of the fact that even when rigorous optimization is performed, the resources needed to determine the optimum batch operations model depend strongly on the initial estimate. Therefore, there is a need for a simple and fast way to provide feasible, near optimum solutions that could eventually be used for further optimization. Based on thermodynamic insights and knowledge gained from available simple models, the developed algorithms generate a sequence of operations that satisfy all path and terminal constraints, while trying to do so in minimum time. For batch reaction, the algorithm attempts to maximize the selectivity of desired reactions over competing reactions, while remaining physically feasible. Selectivity is promoted through changes in operating temperature, as determined from kinetically and/or thermodynamically derived knowledge of the temperature sensitivity of competing reactions. For batch distillation, reflux ratio profiles are determined taking into consideration that operating at the largest driving force makes the separation easier and faster. The driving force approach is also used to identify when a reflux ratio becomes infeasible, determining in that way the end of the corresponding operation. For batch crystallization, phase diagrams are used to identify the feasibility of the precipitation of a specified solid, the necessary batch operation and the operating conditions that will achieve the maximum amount of precipitating solid. The lower level algorithms generate batch operations models that satisfy the intermediate objectives for the task, set by the synthesis algorithm. Once the algorithms are applied, information of the existing state of the mixture is returned to the upper level synthesis algorithm. After that more detailed calculations can be made for the remaining tasks in the batch route. Various computational tools are necessary for applying the algorithms developed and verifying the batch operation models generated. The ICAS computational package is used in this thesis and attention is drawn to the particular features of the ICAS package tools employed in the various steps of the algorithms. The algorithms developed have been tested in a series of application examples, where the batch operation models were generated for various batch processes. The application of the algorithms on an integrated example for both the synthesis and operational design of batch processes have also been performed.