Cluster Detection of Non‐Metallic Inclusions in 42CrMo4 Steel. Issue 11 (8th July 2018)

• Record Type: Journal Article
• Authors: Seleznev, Mikhail; Wong, Ka Yiu; Stoyan, Dietrich; Weidner, Anja; Biermann, Horst
• Is Part Of: Steel research international. Volume 89:Issue 11(2018)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2018-07-08
• Subjects: fatigue -- fine granular area (FGA) -- hierarchical clustering -- non‐metallic inclusions (NMI) -- spatial point processes; Steel -- Periodicals; Steel -- Metallurgy -- Periodicals; 669.142
• DOI: 10.1002/srin.201800216
• Languages: English
• ISSNs: 1611-3683

Abstract

Abstract : Recent progress in steel casting shows significant reduction of non‐metallic inclusions (NMIs). Unfortunately, total NMI volume reduction due to refining can surprisingly lead to NMI size enlargement or cluster formation. This effect decreases the fatigue endurance limit of steel, which is sensitive not only to the presence of large NMIs, but also to clusters of small inclusions, which are usually overlooked in standard metallography inspection. In the present paper, the authors aim to investigate clustering properties of NMIs in a dataset of 42CrMo4 steel, which are observed on fracture surfaces after fatigue tests and from inclusion patterns in polished cross‐sections after metallographic inspection. The authors use both data types for estimating critical distances as main cluster size parameters and find excellent agreement between the results of both methods. The clusters are determined by a classical statistical method of cluster analysis, called "agglomerative hierarchical clustering". The authors show by means of methods of spatial point pattern analysis that their method yields clusters that are more plausible than clusters resulting from the well‐known Murakami "thumbnail rule". Abstract : A method of cluster detection of non‐metallic inclusions in steels and alloys is presented to be useful for metallography inspection. The main cluster size parameter is the critical distance value, which is estimated to be around 10 µm both by fracture surface analysis after high‐cycle fatigue testing and hierarchical agglomerative clustering of a polished cross section.