Evaluation of multiaxial fatigue life prediction approach for adhesively bonded hollow cylinder butt-joints

• Published in: International journal of fatigue 2022-03, Vol.156, p.106692, Article 106692
• Main Authors: Zhang, Jun, Li, Hui, Yu Li, Hai
• Format: Article
• Language: English
• Subjects: Adhesive bonding; Adhesively bonded joints; Bonded joints; Butt joints; Cycle time; Cylinders; Energy dissipation; Fatigue failure; Fatigue life; Life prediction; Low cycle fatigue; Materials fatigue; Multiaxial fatigue; Neural network; Neural networks; Prediction models
• ISSN: 0142-1123; 1879-3452
• DOI: 10.1016/j.ijfatigue.2021.106692

[Display omitted] •Stress-based model can predict the fatigue life of the adhesively bonded joints.•Energy-based model accurately predicted the fatigue life of the bonded joints.•Neural network method is more accurately predict the fatigue life of the bonded joints.•Comparison showed that the neural network method is more simple and accurate. The low-cycle fatigue experiments of the adhesively bonded hollow cylinder butt-joints were conducted in previous work. The effects of loading path, stress amplitude, mean stress and cycle time on the fatigue life were discussed briefly. The stress-based and energy-based fatigue life prediction models were developed to predict the fatigue life of the adhesively bonded joints, respectively. The effects of loading paths and loading conditions on the fatigue life were considered in the prediction models. The prediction results demonstrated that the stress-based model presented a good prediction by reasonably selecting the loading path coefficients and model constants. Furthermore, the energy-based fatigue life prediction model also achieved an accurate prediction on the multiaxial fatigue life through considering the dissipation energy as fatigue failure parameter. In addition, the neural network based method was adopted to predict the multiaxial fatigue life of the adhesively bonded joints. It was shown that the neural network based method obtained a more accurate prediction for the multiaxial fatigue life of the adhesively bonded joints compared with the previous two prediction models. Moreover, the evaluation of the applicability and accuracy of these three prediction methods was carried out. It indicated that the neural network based method had the simple prediction process and it was more accurate than that of the two models for the multiaxial fatigue life prediction of adhesively bonded joints.