This thesis addresses the development of efficient methods for synthesis of chemical processes. Synthesis of the chemical process means determining both the structure of the process and its design parameters.
The proposed methods apply the Process graph(P-graph) theory to represent the chemical process synthesis problems and to generate the set of feasible process structures. Generally, process synthesis problems can be represented by MINLP (Mixed Integer Nonlinear Programming) models with suitable superstructure. However there had been no mathematical definition and theory about superstructure until P-graph theory was introduced. Also it is difficult to solve the MINLP models in which many combinatorial properties are involved. This problem can be overcome using P-graph theory in which combinatorial properties can be easily incorporated. In addition, if the structural constraints of synthesis problems are used to determine the set of feasible process structures, then the size of search space can be extremely reduced in the synthesis problem. The main contribution of this work is the minimization of search space of binary variables in chemical process synthesis problems.
The original P-graph theory was developed for general flowsheet synthesis problems, but it can be applied to the specific problems with proper definitions of material and operating unit sets.
In this work, methods for synthesis of two chemical process networks are developed. They are mass exchange network (MEN) and scheduling list synthesis. For each synthesis problem, the process synthesis problem is converted to a P-graph synthesis problem.
To demonstrate the proposed methods, industrial synthesis problems are examined.