Optimal design under uncertainty of bearing arrangements

This study presents a methodology for an optimal design under uncertainty of a real bearing arrangement. The aim is to find the optimal bench bearing arrangement preload/clearance that maximizes the life of the bearing arrangement considering several uncertain geometric and operating parameters. The paper also presents a new systemic method used to determine the bearing loads required for the life calculations of the bearing arrangements. This method is suitable to be used within an Evolutionary Algorithm due to the fact that it is fast and reliable. The optimal design of the considered bearing arrangement-from the maximum bearing life standpoint-is achieved using a new and efficient stochastic optimization tool obtained by combining the features of Cuckoo Search algorithm and the ones of the Knowledge Gradient policy. Also, a specific approach for optimization under uncertainty using three stages is proposed in this paper. The obtained results are validated by Monte Carlo Sampling.