THE SEQUENTIAL-CLUSTERED METHOD FOR DYNAMIC CHEMICAL-PLANT SIMULATION

We describe the design, development and testing of a prototype simulator to study problems associated with robust and efficient solution of dynamic process problems, particularly for systems with models containing moderately to very stiff ordinary differential equations and associated algebraic equations. A new predictor-corrector integration strategy and modular dynamic simulator architecture allow for simultaneous treatment of equations arising from individual modules (equipment units), clusters of modules, or in the limit, all modules associated with a process. This "sequential-clustered" method allows for sequential and simultaneous modular integration as extreme cases. Testing of the simulator using simple but nontrivial plant models indicates that the clustered integration strategy is often the best choice, with good accuracy, reasonable execution time and moderate storage requirements. © 1990.