化工过程系统工程专论-2010.pptVIP

化工过程系统工程专论 硕士研究生先修课程（32学时） 2010-11 In this course -- General concept -- Algorithmic methods for process synthesis Concepts Optimization technique: MINLP Process synthesis Problem representation Modeling Solution strategies Examples 1. Introduction General concept Process synthesis -- central issue Process Optimization -- trade-off Objectives – mono- or multi- Integrations among variables -- coupling Solution of optimization problems -- combinatorial problems Process synthesis -- central issue HEN, Dist. sequence, Reactor networks etc. Process integration Integrations among variables -- coupling Multi-variables Solution of optimization problems -- combinatorial problems 2. Problem representation Solution strategies for tree representation -- Branch-and-Bound Properties: - an intermediate node gives a partial cost of a structure: a lower bound on the cost of the structure - a terminal node gives the cost of a particular structure: an upper bound to the problem Rules: - Depth-first : less storage, more nodes to evaluate - Breadth-first : fewer nodes examined yet more storage needed Models and solution strategies for network representation -- Simultaneous solution Net work representation: - binary variable : yi = - continuous variables x Problem formulation: - Mixed-integer Nonlinear Programming (MINLP) (P) Min f(x, y) h(x, y) = 0 g(x, y) ? 0 x?X, y ?Y For problem (P) LP (Linear Programming ) if the functions f, h and g are linear and y is constant (no longer variable) NLP (Nonlinear Programming ) if any of f, h and g keeps nonlinear and y is constant (no longer variable) MILP (Mixed-integer Linear Programming ) if the functions f, h and g are linear MINLP (Mixed-integer Nonlinear Programming ) nonlinearity of any f, h and g Solution procedures for problem (P) Solution