Automated feature analysis as a tool for non-metallic inclusion analysis in solidified metals

In-depth analysis of the non-metallic inclusion population in solidified metals is essential considering the high cleanliness requirements for metallic components. For this purpose, automated feature analysis (AFA) is a powerful inclusion analysis tool, providing statistical supported information about quantity, elemental composition, morphology and size of inclusions. However, thorough data evaluation is imperative to ensure accurate and reliable results. The paper introduces the key principles of evaluating AFA data, emphasizing its capabilities and constraints by utilizing four large datasets obtained from solidified steel blocks, each containing a minimum of 4000 features. Additionally, methodologies for identifying characteristic inclusion species within a dataset, constructing suitable rulefiles for inclusion classification through statistical analyzes, and interpreting the resulting outcomes are presented. The size distribution of inclusions is analyzed using population density functions, focusing on the absolute inclusion count within specific size ranges. Cluster analysis is presented considering the inclusions‘ positional parameters.