An Intelligent Integrated Method for Reliability Estimation

Uncertainty management of complex dynamic systems is required to design safer and more damage-tolerant systems. One important step in the uncertainty management is estimating the underlying risk. We proposed a novel risk assessment procedure to estimate the underlying risk proposed considering all major sources of uncertainty and nonlinearity. It is based on the multiple deterministic analyses-based concept to address uncertainty related issues in the formulation, currently a major research trend in the profession. A novel risk evaluation concept using a surrogate metamodel Kriging technique was proposed. By doing so, the implicit performance functions (PFs) are expressed explicitly using the significantly improved Kriging-based surrogate modeling technique. The proposed method consists of the response surface (RS) concept significantly modified for the reliability analyses and several advanced factorial schemes producing compounding beneficial effects. The method was clarified with the help of an informative example by estimating risks for serviceability and strength PFs of a large jacket-type OFS. The results were verified using thousands of Monte Carlo simulations (MCSs). This method used about 200 deterministic analyses and compromising less than 10% accuracy compared to MCS with 10,000 deterministic analyses. We believe that the proposed method can be an alternative to the basic MCS technique and a novel risk evaluation procedure.

Please refer to the publication tab for more details. The study was supported by the National Science Foundation (NSF) under Grant No. CMMI-1403844..