Statistical accuracy of fractographic estimation in silicate glasses with design of experiments and pairwise T-tests. (October 2020)

• Record Type: Journal Article
• Authors: Ma, Lingyue; Djurdjanovic, Dragan; Dugnani, Roberto
• Is Part Of: Engineering failure analysis. Volume 116(2020)
• Publication Date: 2020-10
• Subjects: Fractography -- Silicate glass -- Mirror constant -- Design of experiment -- T-test; System failures (Engineering) -- Periodicals; Fracture mechanics -- Periodicals; Reliability (Engineering) -- Periodicals; Pannes -- Périodiques; Rupture, Mécanique de la -- Périodiques; Fiabilité -- Périodiques; Fracture mechanics; Reliability (Engineering); System failures (Engineering); Periodicals; Electronic journals; 620.112
• DOI: 10.1016/j.engfailanal.2020.104699
• Languages: English
• ISSNs: 1350-6307

Abstract

Highlights: The effect of subjectivity in measuring mirror constant per ASTMC1678 is discussed. Optical (bright/dark/DIC), SEM imaging, and magnification are considered. Statistic tools like design of experiment and T -tests are used in the assessments. Quantitative statistics are compared to the literature's anecdotal descriptions. Abstract: In brittle materials, the material-related "mirror constant" correlates the length of the mirror radius measured on the fractured surfaces with the fracture strength. The mirror constant is traditionally estimated visually following ASTM Standard C1678 guidelines. Some bias is inevitably introduced due to different imaging tools/methods, magnifications, and observers' interpretation of fractographic features. Quantitative characterization of these uncertainties is not well understood and remains a challenge. In this study, thirty-one aluminosilicate glass plates were fractured by 4-points bending and twelve observers independently measured the mirror constant using various imaging methods and magnifications. A systematic study to determine the source of bias was conducted with design of experiment analyses and pairwise T -tests for statistical interpretations. It was found that the uncertainty introduced by observers was 1.5 times the one introduced by the imaging methods considered, and 3 times the effect of magnifications for optical microscopy.