Date of Graduation

1967

Document Type

Dissertation

Degree Type

PhD

Committee Chair

C.Y. Wen

Committee Co-Chair

W. Squire

Committee Member

H.P. Simons

Committee Member

F.X. Hiergeist

Committee Member

A.F. Galli

Abstract

The effects of parameter variation and uncertainty on optimal process design have been studied. Equations necessary for determining the sensitivity of optimal system performance to parameter variation are derived. Design criteria have been developed for optimal design of systems that are either sensitive to parameter variation or involving parameter uncertainty. The sensitivity equations for systems described by algebraic, differential, and difference equations are derived based on the Lagrange multiplier method, Pontryagin's maximum principle, and the discrete version of the maximum principle, respectively. Such sensitivity equations are linear and can be solved for sensitivity coefficients. Two methods are proposed for optimal design of parameter-sensitive systems. Design and operating variables of a system are determined by reducing sensitivity such that the resulting system is close to the optimum and less sensitive. Modified algorithms of the maximum principle for problems with sensitivity constraints are presented. Two design criteria are proposed for optimal design of systems involving parameter uncertainty. One may be used to obtain an appropriate decision which will keep the deviation of the objective from the optimal behavior within a certain tolerance. The other assures a minimum average normalized deviation of the objective from the optimum over the range of uncertainty. Examples in optimal reactor design are given to demonstrate the applicability of the proposed methods.

Download

Share

COinS