UNCERTAINTY QUANTIFICATION ON MECHANICAL BEHAVIOR OF CORRODED PLATE WITH STATISTICAL SHAPE MODELING

Corrosion is a process of uncertain nature considering the randomness associated with corrosion initiation and growth, therefore estimating the stochastic behavior of a corroded structure therefore quantifying of system performance uncertainties would be very important for a wide range of engineering systems such as the shipment structure to ensure structural safety and reliability. In the presented study, we have focused on estimating uncertainty on plate mechanical behaviors due to the impact of the corrosion, such as the shape and depth. Considering the limitation of small quantity of corroded plate samples, we firstly regenerate simulated images based on the statistical shape modeling method. Secondly, these simulated images are imported into a multiphysics-based corrosion simulation platform to reconstruct the shape of corroded plate and the mechanical behavior of the plate with different severity levels of corrosions can be obtained through finite element analysis. Thirdly, uncertainty quantification study is then conducted to understand the statistical characteristic of stochastic behavior of corroded plates from simulated and origin image data. The case study results showed that the statistical characteristic of mechnical behavior from both source data are similar, and the statistical shape modeling method could be useful in situations where there is insufficient sample data for uncertainty quantification.