A major goal of numerical models and computer simulations in materials science is the prediction of the properties as a function of the previous processing or manufacturing of materials. The processing parameters are, however, not state variables of material properties. It is the spatial arrangement of structural elements such as crystal defects, functional domains and chemical constitution that determines the properties of a material – in other words, the microstructure. Simulations of microstructure development provide the state variables for predictions of macroscopic material behavior. In turn, reliable predictions of microstructure evolution require a good understanding and precise modeling or abstraction of the underlying physical mechanisms for microstructure modification. This can be achieved by atomistic simulation of fundamental physical processes and by bottom-up strategies of data transfer to the mesoscopic and macroscopic scales. In this talk, recent advances in the simulation of microstructure evolution will be presented and discussed. The combination of experiments and simulations and the utilization of bottom-up approaches will be emphasized. To this aim, diverse examples -ranging from the atomistic simulation of grain boundary migration to the simulation of recrystallization for crash components- will be presented. Additionally, the benefit of utilizing concepts from data-mining and data-analytics in simulations to extract and analyze useful information that was previously ignored will be highlighted.