lmprovement of Convergence in the Simulation of Complex Chemical Plants

Abstract

<p> The various criteria and methods suggested in the literature for choosing the optimum sequence of calculation are compared. They are then analysed using the results of runs on three different simulations. Two techniques of convergence are used, and it is found that Geometric Convergence Promotion will improve the rate of convergence compared to Direct Substitution for any feasible sequence. It is also shown that the optimal sequences of calculation are different for the two techniques. </p> <p> It is concluded that the Minimum Cut Streams criterion is the best to use for Direct Substitution, but that the use of the feasible Set criterion and Convergence Promotion will yield better convergence. The iterative improvement of convergence is discussed, and it is concluded that the available time is best spent on improving the starting point and the Convergence Promotion technique rather than on changing the sequence of calculation. </p>