That the key to the macroscopic material properties lies in the microstructure is particularly true for a wide range of materials, e.g. ceramics or fiber reinforced polymers. Their complex and inhomogeneous microstructure leading to a distinct anisotropy in the macroscopic material behavior. Due to the declining costs of micro-computer tomographs more and more 3D images of the material’s microstructure are available. Nevertheless, for a lot of materials only 2D images e.g. from polished micrograph sections are available. This work describes the modelling of digital twins based on 2D and 3D images and their validation using the software GeoDict.
From 2D polished micrograph sections digital twins of six variants of a cordierite ceramic are modelled. In a first step the models are validated comparing the chord length distribution from the polished micrograph section and the models in 2D. Afterwards the 3D models of the ceramics are validated by the calculation of the permeability and the porosity of the digital twins compared to the measurements. Using the digital twin, the stresses inside the ceramics due to a thermal heating can be simulated and the different variants of the cordierite can be compared.
From a 3D micro-tomography scan a digital twin of a non-crimp fabric made of glass fibers is generated. Therefore, an artificial intelligence is used to identify the single rovings of the fabric. After the extraction of the single rovings, the geometry is analyzed and the digital twin is modelled. First the model is validated comparing parameters like local fiber volume fractions and fiber orientations. In a second step a flow simulation is performed to determine the permeability of the non-crimp fabric. The results of the experiment, the simulation on the CT-Scan, and the simulation on the digital twin are compared.