STUDIES IN THE OPTIMAL SYNTHESIS OF CHEMICAL PROCESSING AND ENERGY SYSTEMS.

摘要

A general synthesis strategy based on mixed-integer linear programming (MILP) is proposed in this thesis. The first stage of the strategy is to develop a superstructure which includes a large number of alternative plant units and configurations capable of performing the necessary processing tasks for the given problem. The second stage of the synthesis strategy is to model the proposed superstructure as a mixed-integer linear program, and then solve it in order to determine the optimal structure and operating conditions of the processing system. The application of this synthesis strategy is demonstrated on the three major components of a total processing system: Chemical Plant, Utility System and Heat Recovery Network.; The synthesis of arbitrary chemical plants has been formulated as a MILP model which can perform simultaneously structural and parameter optimization for obtaining the optimal design. Two different cases are considered for the synthesis of utility systems. In the first case the synthesis of utility systems providing a set of time independent (constant) utility demands is formulated and solved as a MILP model. In the second case, a multiperiod MILP model is developed for the synthesis of flexible utility systems, that satisfy a time dependent (multiperiod) utility demand set. Finally, several formulations of the transshipment model are proposed for the synthesis of efficient heat recovery networks that exhibit minimum utility cost and the least number of heat exchanger units.; The most important aspect of all the synthesis models presented in this dissertation is their compatibility. In particular, the MILP synthesis models for the chemical plant and the utility plant, as well as the transshipment model of the heat recovery network can be added in order to provide the common mathematical frame required for the optimal synthesis of total integrated systems. Several numerical examples are presented to evaluate the performance of the proposed MILP synthesis models. The computational effort required to obtain the optimal system structure and operating conditions is small to modest even for large size problems. Thus, it is shown that MILP is a powerful and efficient tool for a variety of important synthesis problems.