Simulation of Inspections on Ship Plates With Random Corrosion Patterns

This paper investigates how the number of thickness measurements and the location of each measurement influence both the correct representation of the level of corrosion and indirectly the correct assessment of the collapse strength of the corroded plates. For this purpose, the corrosion patterns are represented by random fields, which are then discretized using the Expansion Optimal Linear Estimation method. This approach is used to simulate the present practice on thickness measurement patterns and to identify what would be the effect of other strategies of measurement patterns for representing the corrosion in terms of prediction of the average reduction of plate thickness. Finally, the collapse strength of the plates with nonuniform corrosion is calculated by nonlinear finite element analysis and its correlation with the average reduction of plate thickness obtained from the different measurement patterns is assessed.