Comprehensive Pre-Processing, Process and Quality Data from Multi-Part PBF-LB/M Manufacturing with Job- and Part-Level Annotations

This dataset presents a comprehensive collection of data from a multi-part build manufactured with Laser Powder Bed Fusion of Metals (PBF-LB/M). It combines job-level process information with part-specific inspections and tests, offering researchers the opportunity to study relationships across the entire process chain. This dataset demonstrates the complexity of working with heterogeneous industrial data sources and provides a realistic challenge for data integration, analysis, and interpretation. The data were generated at the Digital Additive Production (DAP), RWTH Aachen University, where the build was produced and most inspections were conducted. Additional computed tomography data were contributed by the Laboratory for Machine Tools and Production Engineering (WZL), RWTH Aachen University, complementing the dataset with volumetric insights into internal part quality. The build contains 17 distinct geometries, which were selected to cover a broad range of use cases. They are organized into three functional clusters: Mechanical cluster: cubes and tensile specimens according to DIN 50125, used for hardness, density, and tensile testing. Metrology cluster: standardized test specimens according to DIN 52902 for evaluating dimensional accuracy, surface quality, roundness, and related metrics. Distortion cluster: cantilever specimens designed to study warpage, residual stresses, and distortion effects under defined thermal conditions. Each cluster has a distinct purpose, which is reflected in the type of data collected during quality control. For example, mechanical samples are linked to tensile curves and hardness measurements, while DIN 52902 geometries are assessed by optical scans, tactile measurements, and computed tomography. This diversity creates a dataset that covers multiple perspectives within a single manufacturing job. All parts were produced on an EOS M290 system using stainless steel with the material designation X2CrNiMo17-12-2 (EN 1.4404 / AISI 316L). The build was carried out with standard EOS parameter sets at a layer thickness of 30 µm. The exposure strategies included separate parameterizations for infill, upskin, downskin, contour, and edge regions. Detailed settings such as laser power, scan speed, hatch distances, and contour strategies are documented in the dataset (process_parameters.csv and build job file). The dataset is provided as eight compressed .7z archives. After extraction, they recreate the full directory hierarchy with two top-level folders: Job_Level_Data/: build-wide information such as CAD/CAM files, process parameters, job files, in-situ monitoring data, and build plate inspections. Part_Specific_Data/: subfolders for each geometry with CAD references, inspection results, and test data. The accompanying README.md describes the resulting folder tree in detail, including naming conventions and guidance on navigating the different data modalities. The dataset comprises a wide variety of file types, including: CAD and geometry: STEP, STL Process definitions and logs: CSV, EOS build job (.openjz) Monitoring and imaging data: PNG, TIF, JPG Inspection and measurement results: CSV, PDF, XLSX, TXT Reports and scans: PDF, GOM Ginspect (.ginspect) Some file types can only be opened with commercial software, for example EOSPRINT for EOS build job files or GOM/ZEISS software for optical measurement data. Where possible, open formats are provided in parallel, such as CSV exports accompanying measurement reports. The dataset enables diverse applications, such as: Linking process parameters with part-specific inspection data. Developing and validating machine learning models for defect detection and property prediction. Performing simulation studies and benchmarking them against experimental measurements. Exploring data integration challenges across the process chain, supporting research on digital shadows and interoperability in manufacturing. By combining pre-processing, process, and quality data within a single build, the dataset provides a unique opportunity to work with interconnected, heterogeneous information. It can serve as a foundation for both methodological developments and domain-specific investigations in the field of metal additive manufacturing. Funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany’s Excellence Strategy – EXC-2023 Internet of Production – 390621612