Rapid evaluation of crack growth using scanning electron micrographs a novel trained neurone algorithm to determine crack initiation in a large data set. 

Abstract number
60
Presentation Form
Poster
DOI
10.22443/rms.mmc2023.60
Corresponding Email
[email protected]
Session
Poster Session Three
Authors
Dr Maadhav Kothari (1)
Affiliations
1. Cranfield University
Keywords

Key words: Machine Learning, algorithm development, SEM Microscopy, Metallurgy

Abstract text

This paper presents a novel approach for the rapid assessment of corrosion product (NaCl, KCl, and CMAS) of CMSX-4 to determine crack initiation and growth using machine learning of point to point data utilising a neurone algorythim. To demonstrate this approach, over 600 SEM microscopy images were rapidly evaluated in minutes. The result of this machine learning evaluation was used to determine the extent of crack initiation and growth of the CMSX-4 material. The results of this evaluation revealed that the neurone algorythim was able to accurately predict the crack initiation location and growth rate of the CMSX-4 material with a high degree of accuracy (R2= 0.98). This approach provides a rapid and accurate way to monitor and evaluate the mettalurgy of CMSX-4 in a production environment.

References