2020/05/11 by Christian Gebhardt, Gebhardt, Christian, Torsten Trimborn +9
Engineering · Materials Science · #Computational Engineering #FOS: Computer and information sciences #FOS: Physical sciences #Fatigue and fracture mechanics #Finance #Hydrogen embrittlement and corrosion behaviors in metals #Materials Science (cond-mat.mtrl-sci) #Mechanical stress and fatigue analysis #and Science (cs.CE)
paper · pdf · doi:10.48550/arxiv.2005.06615
openalex publication_date 2020/05/11 · openalex created_date 2022/07/26 · openalex updated_date 2026/07/28
The heterogeneous microstructure in metallic components results in locally\nvarying fatigue strength. Metal fatigue strongly depends on size and shape of\nnon-metallic inclusions and pores, commonly referred to as "defects". Nodular\ncast iron (NCI) contains graphite inclusions (nodules) whose shape and\nfrequency influence the fatigue strength. Fatigue strength can be simulated by\nmicromechanical finite element models. The drawback of these models are the\nlarge computational costs. Therefore, we employ a data-driven machine learning\nmethodology. More precisely, we utilize the simplified residual neural network\n(SimResNet) which was recently introduced (Herty et al., Kinetic Theory for\nResidual Neural Networks, 2020) to predict fatigue strength from metallographic\ndata. For the training, we use fatigue data which is simulated with a\nmicromechanical model and the shakedown theorem. The micromechanical models are\nderived directly from micrographs of nodular cast iron, respectively. The\napplication of SimResNet shows a good performance to predict fatigue strength\nby local microstructures of nodular cast iron. We show several test cases. The\nsimplified character of SimResNet enables fast predictions of fatigue by\nmicrostructures, even in comparision to classical residual neural networks.\n