Estimating the Probabilistic Size and Shape Distributions of 3D Anomalies From Sectioning Measurements Using the Stereological Unfolding Approach

Abstract

The accuracy of probabilistic risk assessment of rotor disks is strongly dependent on the accurate description of the size and shape distributions of anomalies in alloys. These size-shape distributions of anomalies are often derived from planar sectioning data measurements using stereological unfolding algorithms. Since it is impossible to accurately predict the shape and orientation parameters of a general ellipsoid based on measurements obtained from two-dimensional sectioning data, the anomaly model should be limited to a spheroid. In this study, an unfolding algorithm was implemented and verified that can be used to estimate the probabilistic dimensions and orientations of 3D spheroids based on 2D section data. It is shown that the accuracy of the predicted spheroid model is dependent on the number of sections and the discretization of the mesh used to characterize the data.