Overview
ISO/ASTM TR 52958:2026 defines a comprehensive workflow for in-situ flaw detection in metal additive manufacturing using powder bed fusion-laser based (PBF-LB) techniques. This technical report centers on the use of coaxial photodiode sensors for real-time monitoring during PBF, applying statistical and clustering machine learning algorithms to detect lack of fusion flaws in manufactured parts. The workflow supports setting detection thresholds and configuring cluster quantities for machine learning, leveraging intentionally seeded flaws for calibration and validating results with computed tomography (CT) scans. The document also addresses hardware considerations, including multi-laser setups and sensor limitations.
Key Topics
- In-situ Coaxial Photodiode Monitoring:
Utilizes a coaxially aligned photodiode to capture process signals in real time, enabling ongoing detection of process anomalies associated with lack of fusion during PBF.
- Flaw Detection Algorithms:
- Statistical Methods: Employing moving average-based thresholding (Absolute Limits, Short Term Fluctuations) to highlight deviations indicating lack of fusion.
- Machine Learning: Application of clustering algorithms such as self-organizing maps (SOM) and K-means for unsupervised anomaly grouping.
- Calibration with Seeded Flaws:
Strategic inclusion of intentionally seeded flaws in test coupons to systematically tune and validate flaw detection algorithms.
- Data Validation:
Use of CT scan data for post-build verification, with a voxel-based approach to directly compare algorithmic detection against physical flaw presence.
- Hardware and Process Considerations:
Discussion of sensor frequency and resolution constraints, effects of process parameter variations, and challenges in multi-laser machines.
Applications
- Quality Assurance in Metal Additive Manufacturing:
Facilitates the early detection of lack of fusion flaws, which are critical to part performance and integrity in industries such as aerospace, healthcare, and automotive.
- Process Optimization:
Enables iterative adjustment of manufacturing parameters and fine-tuning of detection thresholds using real build data, driving continual process improvement.
- Research and Development:
Provides a standardized approach for benchmarking and comparing flaw detection methods, supporting advanced studies in process monitoring and materials science.
- Equipment Evaluation:
Assists in assessing the efficacy and limitations of monitoring hardware (especially photodiodes) in industrial PBF platforms, informing future sensor selection and integration.
Related Standards
- ISO/ASTM 52900:
General principles and vocabulary for additive manufacturing, serving as a foundational reference for terminology within ISO/ASTM TR 52958:2026.
- ISO/ASTM 52901:
Standard specifying requirements for additive manufacturing process qualification, which may intersect with in-situ monitoring validation.
- ASTM E3166:
Standard terminology for additive manufacturing, including definitions for lack of fusion and related terms cited in this report.
- Computed Tomography Standards:
Standards governing non-destructive testing (NDT) using CT, relevant for validating flaw detection workflows.
By integrating coaxial photodiode-based in-situ monitoring with advanced statistical and machine learning algorithms, ISO/ASTM TR 52958:2026 provides manufacturers with a robust, standardized methodology to enhance quality control and defect detection in metal PBF-LB processes. This approach supports traceable, data-driven decision making, ultimately advancing the reliability and industrialization of metal additive manufacturing.