Quality control of cast iron: extreme value statistics applied to CT measurements

In this presentation, we will summarize the main results of a research activity aiming at predicting the size of the defect at the origin of fatigue failure in large spheroidal cast iron presses. The endurance limit of the material is largely influenced by the ineluctable presence of defects related to the casting process, such as voids and degenerated graphite. The aim of this activity was to determine a reliable and robust way to control the material in production and to reject the parts which do not meet the company's quality requirements. Defect measurements have been performed with 2D and 3D techniques (i.e. metallographic analyses and computed tomography) and different maxima sampling strategies have been applied to the data, to determine a reliable and robust distribution of the largest defects. The maximum defect in the component volume was estimated adopting the concepts of statistics of extremes. The goodness of the results has finally been validated considering the largest defect detected in the most stressed component volume.