Fractographic analysis of silicate glasses by computer vision

Significance 

Fractographic approaches are routinely used to assess and determine the cause of failure in engineering structures, particularly in product failure and the practice of forensic engineering or failure analysis. In material science research, fractography is used to develop and evaluate theoretical models of crack growth behavior. Guidelines for the analysis and interpretation of fractographic features in typical brittle materials, i.e. glasses and ceramics, are outlined in ASTM C1678 and generally require locating the fracture origins and estimating the ‘mirror radii’. Recent research has however established that these ASTM guidelines can lead to significant strength estimation errors due to both the subjectivity of the mirror length’s assessment, and inconsistencies in the reported values of mirror constants in the literature. Furthermore, inconsistencies in the use of imaging devices, illumination, magnification etc. can introduce additional errors. In an attempt to address these, Dugnani et al recently showed that in flexural fractures the shape of ‘mirror-mist boundary’ on the fracture surface could be uniquely related to the length of the mirror radius normalized by the sample’s thickness. As such, this suggests that an alternative approach –based on the shape of the ‘mirror mist boundary’– can be used to analyze the fracture strength of amorphous, brittle materials.

In general, ASTM C1678 describes the state-of-the-art’s fractographic techniques to estimate the fracture strength of glasses and ceramics through empirical, strength vs. fracture mirror radius relationships. However, the methodology is subjective and only applicable to a few loading scenarios and relatively pristine fracture surfaces. To address this, researchers from the University of Michigan – Shanghai Jiao Tong University, Joint Institute, China: Lingyue Ma (PhD candidate) and Professor Roberto Dugnani, proposed a new semi-automated, alternative approach to objectively estimate the strength of silicate glasses for ampler loading and geometric scenarios. Their goal was to develop a method for determining the fracture’s strength of brittle, isotropic material. This study is currently published in the Journal of the European Ceramic Society.

In their approach, dimensional arguments were used to establish the relationship between the shape of the mirror-mist boundary and relevant, dimensionless groups. The researchers then introduced a novel computer vision (CV)- based algorithm to objectively determine the shape of the mirror-mist boundary on the fracture surfaces analyzed. Towards the end, fracture strengths of trial samples were estimated using the empirical relationship between dimensionless groups obtained from both experimental testing and an ample literature survey on silicate glasses’ fracture surfaces.

The authors reported the proposed approach required less fractographic experience for the interpretation of the fracture surface features compared to ASTM C1678. Furthermore, it was seen that no information on either the material’s mirror constant, A, or the location of the fracture’s origin were necessary. Overall, the proposed scheme could accurately estimate the strength of specimens beyond the capacity of ASTM C1678, such as in chemically strengthened glasses and fracture surfaces displaying significant damage.

In summary, an extensive collection of fracture strength vs. mirror radii coupled with an objective, computer vision-based algorithm was proposed to estimate the fracture strength of silicate glasses. Unlike in ASTM C1678, the algorithm presented focused on the shape of the mirror-mist boundary rather than the length of the mirror radius at the free surface. In a statement to Advances in Engineering, Professor Roberto Dugnani mentioned that the presented method could be directly applied to classes of materials with similar characteristics and fracture behaviors as the baseline set.

Fractographic analysis of silicate glasses by computer vision - Advances in Engineering

About the author

Lingyue Ma received his B.S. degree in the school of mechanical engineering from Zhejiang University (ZJU), Hangzhou, China, in 2016. He is currently working toward his Ph.D. degree at the University of Michigan-Shanghai Jiao Tong University Joint Institute (UM-SJTU JI). His research focuses on the non-linear fracture behavior of brittle materials.

.

About the author

Professor Dugnani graduated from Stanford University’s Aerospace Engineering Ph.D. program in 2004. Since 2012, Dr. Dugnani is an invited Foreign Expert and Associate Professor in at the University of Michigan – Shanghai Jiao Tong University Joint Institute (UM-SJTU JI). Prior to this appointment, Dr. Dugnani held a position at Exponent Failure Analysis Associates, a leading international scientific and engineering consulting firm. Between 2009 and 2012 Dr. Dugnani held an Adjunct Faculty position at Santa Clara University, CA in both the Civil and the Mechanical Engineering Departments.

Reference

Lingyue Ma, Roberto Dugnani. Fractographic analysis of silicate glasses by computer vision. Journal of the European Ceramic Society; volume 40 (2020) page 3291–3303.

Go To Journal of the European Ceramic Society

Check Also

Fatigue Crack Growth Behavior of WAAM Steel Plates: Experimental Analysis and Comparative Study - Advances in Engineering

Fatigue Crack Growth Behavior of WAAM Steel Plates: Experimental Analysis and Comparative Study