Probe maps for the exploration of localized metastable states and prediction of peak load-carrying capacity of axially compressed cylindrical shells

Abstract

A Machine Learning (ML) based analysis is conducted to find the most influential probing parameter in determining the buckling initiation location by assessing Probe Force–Probe Displacement characteristic curves. This is followed by the construction of Probe maps, which systematically depict the variation of this parameter with the help of numerical simulations incorporating Measured Geometric Imperfections (MGI). The probe maps hence constructed are found to be effective in identifying the critical locations where a single dimple is likely to initiate the buckling, enabling the complete exploration of the critical single dimple states of the structure. The experiments by probing the structure at the critical locations identified using probe maps show accurate prediction of the peak load-carrying capacity. The approach presented in the study may be extrapolated to characterize imperfection-sensitive structures in general.