Studies in constrained optimization of chemical process problems

The application of three constrained, nonlinear optimization methods (method of multipliers, complex method and adaptive random search method) to six chemical process problems is described. Based on the numerical results, salient characteristics of the problems and the specific advantages of the methods are discussed. The multiplier method was implemented with numerical approximation to derivatives; the results show that both the increment (or perturbation) and the accuracy of floating-point numbers are important in determining the effect of numerical approximations on the performance of the method.