Methodology for simultaneous optimization with reliability: nuclear PWR example

Approaches to combining optimization and reliability usually treat the two as separate problems, determining the reliability by integer programming. The purpose of the paper is to describe a method for including reliability in the usual process optimization problem. The algorithm applied to this mixed integer problem is the adaptive random search, which can search both discrete and continuous variables simultaneously. This method is applied to optimization of the primary cooling system of a PWR, which has twelve independent variables, three of which are integers directly affecting system reliability. Reliability data were obtained from USAEC-WASH 1400 report. Optimization results are presented for various levels of reliability.