• Sato, C.
• Sekiguchi, Y.
• Machado, Jjm
• Kang, C.
• Ji, M.
• Naito, M.

Abstract

Machine learning is extensively used in material research and development, including adhesion technology. However, it requires a large dataset to train the models for optimizing, developing, and designing new adhesives. This study proposes a novel testing machine that enables quick high-throughput measurements of the shear strength of adhesively bonded joints. A small cylindrical butt shear joint (BSJ) specimen placed in a holder was pushed by a metal specimen pusher until failure; during this process, the force and displacement were recorded. This testing machine can be used to quickly conduct the measurement by simply placing the specimen in a holder and pushing it. A comparison of the average shear strength measured by this method and that measured by single-lap shear tests, coupled with stress analysis using finite element simulation suggested that the proposed method can measure the shear strength more accurately, where a higher level of pure shear can be achieved in the adhesive layers with a lower degree of stress concentration and smaller peeling stress at the extremities of the adhesives. This indicates that the shear strength of adhesively bonded joints can be measured quickly using the proposed testing method, thereby facilitating the development of new adhesives using machine learning.

Topics

• impedance spectroscopy
• simulation
• strength
• shear test
• machine learning