Abstract
Modern steel grades with increasingly complex microstructure morphologies demand advanced modeling methods. Especially in terms of geometrical microstructure modeling these steels tend to be very challenging. In this study a framework is presented that tackles the modelling of modern microstructures using a new generator for representative volume elements called DRAGen (Discrete Rve Automation and Generation), and a strategy for the numerical investigation of the influence of the surface roughness and residual stresses on the fatigue life and endurance is proposed. The proposed framework consists mainly of three parts: microstructural characterization, generation of statistical representative volume elements and the investigation of micromechanical behavior under extrinsic loading conditions. The microstructural characterization is supported by neural networks while the generation algorithms of the representative volume elements uses the information generated by those neural networks. The mechanical behavior of these representative volume elements is simulated with Abaqus. Due to the usage of a neural network DRAGen can produce very realistic microstructure models following the statistical behavior of the real world material with a high level of detail. It is also possible to introduce features such as surface roughness and residual stresses into the simulations of DRAGen’s representative volume elements. Therefore, the fatigue life estimations performed with these representative volume elements enable a better understanding of the influences that single microstructural features have on fatigue life. This understanding will help to generate a good basis for numerical optimization of the structural integrity of safety-related components.