Quantitative X-ray tomography of advanced polymer composites

Advanced composite materials (ACMs) typically contain two or more constituents, such as matrix, fibers, inclusions and pores, with different physical and chemical characteristics. When combined, they produce a material with unique properties in terms of weight, strength, stiffness, or corrosion resistance. To inspect and study their 3D internal structure in a non-destructive way, the ACMs are imaged using X- ray computed tomography, in which a 3D dataset is reconstructed from the X-ray radiographs. The 3D dataset is subsequently further processed and analyzed in multiple sequential steps. This conventional workflow, however, suffers from inaccurate modeling and error propagation, which severely limits the accuracy with which ACM parameters of interest can be estimated. In this project, we will develop a paradigm shifting approach in which the quantification of ACM parameters is substantially improved. This will be realized by a novel workflow 1) accounting for possible deformation of the ACM during scanning and thereby reducing image reconstruction artefacts; 2) accurately modelling all constituents of the ACM (matrix, pores, inclusions and fibers); 3) directly estimating the ACM model parameters from the X-ray radiographs and thereby preventing error propagation by providing a feedback mechanism; 4) analyzing the workflows input parameter space with respect to sensitivity and stability of output parameters / characteristics of interest. Such a framework is up to now unprecedented. If successful, our framework will to provide substantially more accurate characterizations of internal structures of the ACMs in comparison to conventional workflows.

When analyzing advanced polymer composite components, it is important to know the distribution of interesting features (e.g., fibers, pores) and their properties (length, orientation, diameter, etc.) , in order to understand how the component will behave in its targeted application. In this project, methods and techniques were developed for facilitating the characterization of features as well as their properties and distributions along with the respective data processing and analysis. The first step material experts perform when analyzing advanced polymer composite components is to run a fiber characterization algorithm - these algorithms analyze volumetric datasets of a component, as e.g., generated by X-ray computed tomography, and return for each feature properties as mentioned above. However, as there is no characterization algorithm suitable for every purpose yet, these require an adaptation to the analyzed material and component type. For this reason, algorithms aiming at significantly simplifying and improving the characterization of features in comparison to previous algorithms were developed. Previous algorithms would apply a pipeline of several image analysis steps to arrive at a final characterization of the properties of interesting features. Each single step in this pipeline potentially introduces errors. As part of this project, algorithms were developed, where the computed feature characteristics are matched back against the raw input data, and the final characterization happens by iteratively refining the characteristics so that they match the raw input data as good as possible. Furthermore, we developed methods for the visual analysis of such algorithms: In a first step, we explored methods to visualize the uncertainty that occurs in some of the steps of the analysis pipeline. In addition, we developed methods that are able to compare two or more different characterizations, such as they would result either from different steps in the iterative refinement process, or also more generally, from different fiber characterization methods applied to the same dataset. This enables algorithm developers as well as users of such feature characterization algorithms to analyze how well different algorithms perform. It can also be used to deduce which parameters of the algorithm need to be tuned in what way to enhance the result in terms of the target application. Our methods are applicable for a diverse range of features and objects such as straight and curved fibers as well as pores. We also added ways to monitor how sensitive the analyzed characterization algorithms are with regards to subtle changes to their parameters - the user can see whether the output changes little or a lot, when a certain parameter value is changed slightly.