VALUE OF INFORMATION-BASED OPTIMAL SENSORS POSITIONING BY SUBSET SIMULATION FOR CONDITION-BASED PSA

Digitization in the ongoing 4th industrial revolution brings additional opportunities of monitoring the degradation of Structures, Systems and Components (SSCs). The analysis of the inspection and monitoring data available can allow specific condition-based updating of the risk measures. In this regard, a first task is the positioning of sensors for condition monitoring of SSCs. For doing that, one can rely on the Value of Information (VoI), that allows quantifying the benefits/losses of acquiring information by taking measurements on a SSC in a given position: the more beneficial, the larger the VoI value. In this research, an efficient non-greedy optimization method based on Subset Simulation (SS) is introduced to find the optimal VoI-based sensors positioning. SS, a Monte Carlo simulation technique for efficiently estimating the probability of rare events, is innovatively adapted as an optimization method. The methodology is applied on the sensors positioning of a manifold in a Steam Generator (SG) of a Prototype Fast Breeder Reactor (PFBR) that is degrading under creep. Ultrasonic thickness gauges are considered for spatial deterioration monitoring. Optimal sensors positioning by SS is compared with optimal positioning solution obtained by the non-greedy Particle Swarm Optimization (PSO) method. Results show that the innovative SS method is more computationally efficient than the PSO method.