Importance measures and genetic algorithms for designing a risk-informed optimally balanced system

This paper deals with the use of importance measures for the risk-informed optimization of system design and management. An optimization approach is presented in which the information provided by the importance measures is incorporated in the formulation of a multi-objective optimization problem to drive the design towards a solution which, besides being optimal from the points of view of economics and safety, is also ‘balanced’ in the sense that all components have similar importance values. The approach allows identifying design systems without bottlenecks or unnecessarily high-performing components and with test/maintenance activities calibrated according to the components’ importance ranking. The approach is tested at first against a multi-state system design optimization problem in which off-the-shelf components have to be properly allocated. Then, the more realistic problem of risk-informed optimization of the technical specifications of a safety system of a nuclear power plant is addressed.