Optimization strategies for reactive distillation, retrofit synthesis, and production and distribution planning

This dissertation is concerned with the development of mathematical programming modeling and solution techniques to address problems in process synthesis and planning. Chapter 1 reviews the focus areas of process synthesis and planning. Mathematical programming areas employed in this work are reviewed as well. Chapter 2 presents a high level model to address optimization of the retrofit design of process networks. Process flowsheets and retrofit modifications for increasing capacity, yield and energy recovery are represented through a multiperiod Generalized Disjunctive Programming (GDP) model reformulated into a mixed-integer linear program (MILP) using the convex hull formulation. The formulation is applied to two process network examples which include one for the retrofit of a plant producing acetaldehyde, acrylonitrile, cumene, phenol and acetone. Chapter 3 proposes a GDP formulation for the optimal design of reactive distillation columns. Solution of the model yields optimal feed tray locations, number of trays, reaction zones, and operating and design parameters. A modified logic-based outer-approximation algorithm is used to solve the GDP model. Two reactive distillation systems are optimized that include the metathesis reaction of 2-pentene and the production of ethylene glycol. Chapter 4 describes a multiperiod nonlinear programming (NLP) model for site-wide production planning in multiproduct plant facilities. The formulation incorporates empirical nonlinear process models and optimizes monthly production and inventory levels of all products within the site. The scope of the model is illustrated through real-world industrial examples. In Chapter 5, the models developed in Chapter 4 are embedded into a multisite distribution network. Production and global distribution of several continuous multiproduct plant facilities are optimized in the proposed multiperiod NLP model. Spatial and temporal Lagrangean decomposition schemes are proposed to solve these large-scale multiperiod optimization problems. Several examples are presented including a network of four production sites and five global markets. Finally, Chapter 6 summarizes the major contributions of the dissertation and discusses future work directions.