An improved strategy for solving large-scale process flowsheeting problems which may be nonlinear, nonconvex, and usually have a large set of equality constraints is presented. The strategy is composed of two stages: simulation and optimization. In the first stage, a modularly organized "Equation-Oriented" approach is used to solve the simulation problem. The strategy in this stage is intended to serve as the problem preparation for the second stage. In the second stage, the strategy emphasizes computational reliability and efficiency; an augmented Lagrangian approach, that seems to be the most natural decomposition method for the structure of the process flowsheeting problem, is selected. This approach provides the feasibility of a future implemention on a parallel processing environment. The above strategy makes an evolutionary approach applicable to perform the design calculations and to provide diagnostic information for badly-posed problems. A Multiple Effect Evaporator System operated in a Pulp and Paper plant and five examples taken from published articles are used as testing examples to show the feasibility of the presented strategy.
展开▼
