Effects of Defects

The starting point is an experimental analysis of the defects in existing materials. Suitable imaging methods are used for the respective question. Volume defects are usually detected using computed tomography. However, other non-destructive measurement methods such as ultrasonic testing, thermal wave analysis, or terahertz wave analysis can also be used. Very small volume defects are analyzed using a scanning electron microscope, for which fracture surfaces or target preparation methods are used. Surface defects are detected with special microscopes (laser scanning, confocal, or atomic force microscope) or computed tomography, depending on the required resolution.

The image data is then further processed using in-house programs. In the case of complex structures, defect classification is carried out first. This is based on machine learning. The identified defects are then transferred to finite element (FE) models and subjected to uni- or multi-axial loads. The FE models contain the relevant material data, which are determined through measurements, calculated using microstructure-property simulation or taken from material databases. The stress concentration caused by the defects is recorded for numerous defects of a defect type.

Subsequently, the components are subjected to the loads expected in operation in further FE simulations. The stress concentrations calculated in the previous step are inserted into the component models with their calculated frequency using a probabilistic method, so that their effects on the probability of failure under load can be determined. Calibration or validation of the models is done by measuring the fracture strength and its distribution on standard samples.