| Summary: | Thesis (Ph.D.)--Massachusetts Institute of Technology, Dept. of Chemical Engineering, 2001. === Includes bibliographical references. === At the beginning of the 21st century, globalization has resulted in an intensely competitive environment for chemical manufacturers. Amongst their many concerns, chemical firms face ever-increasing pressures to reduce the environmental impacts of chemical manufacture and to reduce the costs of all their activities. Central to both concerns is the development both of reaction pathways for their products and of the manufacturing processes used to make those products. Both development activities are difficult and time-intensive; closer integration of the two may lead to synergies in research, development, and operations. Four issues in integrating process chemistry and process synthesis are explored: I. Classification of alternatives: distinguishing between "reactions" and "processes"; 2. Generation of alternatives: computer-aided generation and initial screening of reaction pathways; 3. Evaluation of alternatives: formulating a mixed-integer linear program (MILP) to choose between manufacturing pathways; and 4. Choosing between alternatives: multi objective decision-making and optimization New frameworks for thinking about process chemistry and pathway design are presented. In addition, this thesis also presents a flow sheet for mixed-acid batch nitration, reaction pathways for making various substituted benzenes, a general MILP formulation for evaluating and comparing chemical manufacturing pathways, and a review and analysis of different approaches to multiobjective decision-making. (UMI abstract ends here) A batch process for mixed-acid nitration of 4-chloroaniline is presented as a case study in process design. This study is used to illustrate how simple statements of chemical reaction are not enough to know the "chemistry" of nitration. It is proposed that "chemistry" be thought of as consisting of "reactions" and "processes". ''Reactions" describe differences in molecules, and include transformations, balanced "net" reactions (overall and local), transforms, and elementary reactions. "Processes" describe the conditions under which reactions occur. They include transformations, synthesis approaches, task recipes, flowsheets, and laboratory procedures. In the course of developing manufacturing processes, different kinds of reactions and processes must be considered. The twin issues of molecular representation and initial pathway screening are critical distinguishing factors in several different approaches to reaction pathway synthesis. Both representation and screening depend on the representation of "reactions" as transforms, balanced reactions, or elementary reactions. The bounding conditions on the analysis are also shown to affect strongly what results are obtained, via studies of phenol, para-hydroxybenzoic acid, and various chloro-aminobenzonitriles. Reaction pathway analysis of the chloro-aminobenzonitriles results in a set of pathways comprised of a small set of chemical transforms. These pathways are modeled explicitly as a network of manufacturing processes using unit process models as tasks in a state-task network. Constraints are derived from mass balances, process capacities, and topological features of the network. Nonlinearities from power-law equations and from the product of continuous and integer variables are resolved using piecewise linearization. Net present value is used as the objective function. Examples of the (single objective) optimization under different prices for intermediates and wastes illustrate the usefulness of the approach. Although these pathways are formulated and analyzed for benzonitriles, similar networks can be constructed for any chemical for which the appropriate process models exist. Economic and environmental objectives for the benzonitrile network are evaluated together using a multiobjective problem formulation. Different approaches to multiobjective decision making are reviewed and shown to be variants on two methods, the epsilon-constraint approach and the omega-weighting approach. The benzonitrile network is analyzed using both approaches. The epsilon-constraint approach is found to give a full Pareto curve as the solution, whereas the omega-weighting approach gives a discrete set of points that fall on the Pareto curve from the epsilon-constraint approach. Four different operating strategies are obtained from two binary decisions to market intermediates or not and to purify all useful byproducts or not; results from bi-objective optimizations carried out under these strategies are analyzed. Again, although the case study itself pertains only to benzonitriles, the methods of bi-objective analysis are general. === by Gene C. Lin. === Ph.D.
|
|---|
