Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)


Chemical Engineering


An efficient methodology for using commercial flowsheeting programs with advanced mathematical programming algorithms has been developed for the optimization of operating plants. The methodology was demonstrated and validated using ChemShare Corporation's DESIGN/2000 simulation of the Freeport Chemical Company's plant for sulfuric acid manufacture and three nonlinear programming techniques; successive linear programming, successive quadratic programming and the generalized reduced gradient method. The application of this methodology begins with the development of a feasible base case simulation. Partial derivatives of the economic model and constraint equations are computed using fully converged simulations. This information is used to formulate an optimization problem which can be solved with the NLP algorithms giving improved values of the economic model. A line search is constructed through the point found from the nonlinear programming algorithm to find the best feasible point to repeat the procedure. The procedure is repeated using the ChemShare simulation program and the NLP code until convergence criteria are met. The algorithms were applied on an iterative basis such that the engineer maintained an active role throughout the course of the optimization. This method was applied to three flowsheeting problems; a plant scale contact sulfuric acid process model, a packed bed reactor design model, and an adiabatic flash problem. The results of the optimization studies indicate that the three algorithms investigated, SLP, SQP and GRG, located the optimum of the economic model and satisfied the constraints at similar levels of efficiency. This efficiency is equivalent to those observed by other workers using similar simulation models. One important conclusion from this work is that the method of application, which includes the selection of feasible versus infeasible path, line search parameters, and the means for obtaining derivative data, is more significant than the nature of the optimization algorithm. Future research into flowsheeting and optimization should attempt to determine, establish and standardize efficient and reliable methods for the evaluation of the function and derivative information necessary for optimization. The importance of scaling to the efficiency and success of SQP methods indicates that future research in this area should seek to establish procedures for determining optimal scaling sets.