Analyzing quasi-static fracture of notched magnesium ZK60 using notch fracture toughness and support vector machine

• Published in: Theoretical and applied fracture mechanics 2022-10, Vol.121, p.103463, Article 103463
• Main Authors: Alshaiji, Abdullah, Albinmousa, Jafar, Peron, Mirco, AlMangour, Bandar, Ali, Usman
• Format: Article
• Language: English
• Subjects: Fracture; Magnesium; Notch; Notch Stress Intensity Factor; Support Vector Machine; Theory of Critical Distance
• ISSN: 0167-8442; 1872-7638
• DOI: 10.1016/j.tafmec.2022.103463

•Quasi-static fracture of ZK60-T5 magnesium extrusion was investigated.•NSIF-TCD was used to analyze the fracture of eleven U- and V-notched specimens.•Machine learning model (support vector machine) was used to analyze the fracture. Magnesium alloys have drawn the attention of many industries due to their outstanding properties and lightweight. ZK60 is one of the most commonly used commercial magnesium alloy mainly due to its high strength and corrosion resistance. Such properties make magnesium alloys attractive materials for loading bearing components in various applications such as automotive, aerospace, and biomedical industries. However, notches such as holes, edges, grooves, and other geometrical irregularities are inevitable as they are integrated into structural components. Moreover, notches have a huge impact on the structural integrity due to the stress concentration regions around the notch tip geometry that can cause catastrophic failures. Therefore, an accurate prediction of the fracture behavior is necessary to avoid such failures. This work aims to investigate the ability of a combined notch stress intensity factor and the theory of critical distance method to predict the notch fracture toughness of notched ZK60 magnesium specimens. A support vector machine was also implemented to predict the notch fracture toughness and results from the support vector machine model were compared to that obtained from the combined notch stress intensity factor and theory of critical distance model. Eleven different U- and V-notched geometries were examined. Results with an accuracy between 2.8% and 22.5% were obtained using the combined notch stress intensity factor and theory of critical distance model while an accuracy between 1.8% and 27.6% were obtained using the support vector machine model.