The microstructure of modern steel grades tends to become increasingly complex. The need for more complex requirement profiles leads to microstructural features that affect the mechanical behavior of the material. These complex features such as the grain morphologies or morphologies of secondary phases cannot be considered with classical material models. Therefore, material models that represent the single crystal behavior have become more and more famous in recent years. These crystal plasticity models can be used together with geometrically reconstructed microstructure models to account for the complexity of the microstructure. Yet, many features that can be found in microstructure measurements cannot be reconstructed with the most common microstructure generation tools. Thus, our group developed a generator tool called DRAGen (Discrete Rve Automation and Generation) which offers multiple different features that can be found in steel microstructures. Also, an application example for the numerical estimation of fatigue life is presented.

DRAGen is based on pure python code, making it a very modular tool with high flexibility. For the input data, we also provide a neural network that can generate an infinite amount of grains for a trained material. The DRAGen output is currently available for Abaqus, Damask, and Moose. Further extensions for other simulation software such as Micress are planned.

For the numerical determination of the fatigue life using DRAGen RVEs a crystal plasticity model is used which is implemented in an Abaqus Umat Subroutine. The study's results show that the model is sensitive to geometrical features, proving that DRAGen'sflexibility for different geometrical features is a very powerful tool.