| Current simulation systems for computer-aided chemical process design are based mainly on the sequential modular approach. However, this approach is not suitable for controlled process simulation and process optimization calculations. Recently, several studies have shown that the simultaneous modular approach is a very promising alternative. These studies, however, are fairly limited in scope. In this study, a general purpose simulator, SIMMOD, was developed to provide a critical evaluation of the simultaneous modular approach for process simulation, controlled simulation, and optimization.; Chen and Stadtherr have demonstrated the effectiveness of a modification of Powell's hybrid method for solving nonlinear equations. In this study, we use this technique to solve process simulation and controlled simulation problems, and we also extend this technique to solve large sparse systems of nonlinear equations. The results of five benchmark problems show that when an accurate Jacobian is used, SIMMOD usually solves an simulation problem in less than five iterations. Of the three options for evaluating an initial Jacobian, we found the diagonal-block perturbation technique of Mahelec et al. to be unreliable, the direct difference approximation scheme of Perkins to be too expensive, and a full-block perturbation technique to be the most suitable. When the initial Jacobian is calculated by full-block perturbation, SIMMOD is much more reliable and efficient than the conventional sequential modular approach for process simulation problems.; For the optimization problems, we use the Han-Powell method to solve the nonlinear more programming problems. Several modifications of the algorithm are introduced to make the method efficient and reliable. The results on four benchmark problems show that SIMMOD is very efficient. As an example, the optimization of an ammonia synthesis process only takes about 6 seconds of CPU time on a CDC Cyber 175 computer. |
