Multiaxial fatigue life prediction of polychloroprene rubber (CR) reinforced with tungsten nano-particles based on semi-empirical and machine learning models. (April 2021)

• Record Type: Journal Article
• Authors: Choi, Joeun; Quagliato, Luca; Lee, Seungro; Shin, Junghoon; Kim, Naksoo
• Is Part Of: International journal of fatigue. Volume 145(2021)
• Publication Date: 2021-04
• Subjects: Fatigue life prediction -- Multiaxial stress state -- Limiting dome height experiment -- Deep neural network -- Machine learning; Materials -- Fatigue -- Periodicals; Materials -- Fatigue; Periodicals; 620.1122
• DOI: 10.1016/j.ijfatigue.2020.106136
• Languages: English
• ISSNs: 0142-1123

Abstract

Graphical abstract: Highlights: Multiaxial fatigue life research of the of tungsten-filled polychloroprene rubber. Experiment utilizing notched specimens and limiting dome test experiments. Semi-empirical fatigue model development considering anisotropy and triaxiality. Additional fatigue life estimation considering machine learning-based models. Reliable estimation of the fatigue life for both developed approaches. Abstract: In this paper, multiaxial fatigue experiments on a hyperelastic rubber-like material made of polychloroprene rubber (CR) reinforced with tungsten nano-particles have been carried out on notched specimens and hourglass specimens, utilized for limiting dome height fatigue tests. Based on the uniaxial (Choi et al., 2020) and multiaxial fatigue experiments, a semi-empirical ε-N fatigue model is proposed, allows accounting for both material anisotropy and complex stress states, showing an average error of 20.7%. Furthermore, six machine learning models have been employed for the fatigue life prediction and shown that the Deep Neural Network is the most accurate, with an average error equal to 14.3%.