Topographic analysis and Weibull regression for prediction of strain localisation in an automotive aluminium alloy

We fit a two-parameter Weibull regression model by maximum likelihood estimation (MLE) to filtered surface roughness data. These data were acquired with scanning confocal laser microscopy performed on aluminium alloy AA 5754-O surfaces that were subjected to a range of plastic strain intensities in three different in-plane strain modes. Noting that one of the two Weibull regression parameters is, to a good approximation, invariant with strain intensity for a given strain mode, the authors find that the variation of the second parameter with strain intensity conforms to a simple quadratic function. These functions may then be used to generate accurate, statistically significant, single parameter predictors of both strain intensity and strain mode up to and including the onset of critical strain localisation and/or failure.