Additive manufacturing processes make it possible to produce components optimised for load paths and material requirements. This requires a sound understanding of the relationships between the processes involved, component structures, and component properties, which can be incorporated into the component design process. If this is not the case, additively manufactured components may fail prematurely under load, especially under cyclic stresses. To systematically build knowledge about process-structure-property relationships, data-driven methods are being developed within the AMTwin collaborative project. For this purpose, data from the involved processes, the materials used, and the testing procedures carried out on components made from Ti6Al4V produced by selective laser melting are recorded. A data-based analysis across process boundaries to determine process-structure-property relationships requires systematic management of the recorded research data, including ID management to ensure traceability of the components. Documentation and linking of the data along the process chain are two essential prerequisites.

This article presents the data management necessary for producing analysable datasets and, based on this, conducts detailed data analyses to determine process-structure-property relationships in additively manufactured components. It first shows which steps are necessary in data management and data preprocessing, especially in data integration. The requirements for research data management (RDM) are described, a concept for practical RDM is presented, and the research data infrastructure implemented in AMTwin is demonstrated. The developed RDM concept takes into account the entire data lifecycle, from data collection to the publication of final datasets. The focus is on documenting research data based on an application ontology for additive manufacturing, which establishes a common technical language across laboratory and project boundaries.

After creating analysable datasets, the data are examined to determine process-structure-property relationships and to address further selected questions. Exploratory data analysis methods, such as parallel coordinates plots, are used to answer these questions. Parallel coordinates plots allow correlations across process boundaries to be visualised in an easily understandable way. The results of the data analyses are discussed, the main influencing factors are identified, and conclusions are drawn regarding the optimisation of additively manufactured components. The results presented are part of the AMTwin project, which is funded by the Free State of Saxony.