Process Engineering Consultant EOS of North America, Inc.
Qualification of additively manufactured (AM) components for mission-critical aerospace, defense, and healthcare applications is constrained by costly, destructive post-process evaluation. Current protocols, reliant on limited witness coupons, are slow, expensive, and insufficient to capture spatial variability in material properties caused by local thermal history deviations. This bottleneck hinders the adoption of AM for serial production.
To address this, we introduce an AI-driven framework for “qualification by analysis,” leveraging EOS’s EOSTATE multi-modal sensors—Optical Tomography (OT), Melt Pool Monitoring (MPM), and Powder Bed (PB) imaging—to generate a spatially registered digital record of the entire PBF-LB build. From this dataset, statistically significant process indicators were engineered and used to train a deep learning model linking in-situ process signatures to final performance of Ti-6Al-4V specimens.
The model was rigorously validated against tensile tests, achieving strong predictive accuracy for yield strength (R² = 0.83) and ultimate tensile strength (R² = 0.74). By reducing reliance on destructive testing by up to 50%, this work establishes a validated pathway for uncertainty-aware, high-throughput AM qualification. The approach provides a foundation for accelerated development cycles and robust, part-specific certification, and is extensible across materials and AM platforms.
Learning Objectives:
Describe the limitations of traditional qualification methods for additively manufactured (AM) components in mission-critical applications, including the challenges posed by destructive testing and spatial variability in material properties.
Explain the principles of an AI-driven “qualification by analysis” framework, including how multi-modal EOSTATE sensor data can be used to generate spatially registered digital process records.
Evaluate the impact of deep learning–based process-performance models on reducing destructive testing requirements and accelerating qualification and certification workflows for AM parts.
