| About this Abstract |
| Meeting |
2025 TMS Annual Meeting & Exhibition
|
| Symposium
|
Additive Manufacturing Modeling, Simulation and Machine Learning
|
| Presentation Title |
Digital Shadow Model Reference Control for Directed Energy Deposition |
| Author(s) |
James C. Haley, Bruno Turcksin, Callan Herberger, Ashley Gannon, Steve Dewitt |
| On-Site Speaker (Planned) |
James C. Haley |
| Abstract Scope |
In Directed Energy Deposition, complex thermal patterns effect microstructure and residual stress formation. Homogenizing the heat input and permitted cooling times to achieve greater uniformity can be achieved through optimizing simulation; however, simulation alone both requires thorough calibration and cannot react to process disturbances. To address this, a ‘Digital Shadow’ controls schema is demonstrated; a real-time finite element heat transfer model is corrected with spatially mapped in-situ infrared temperature readings via an ensemble Kalman filter. This establishes an improved state estimate of the DED thermal system, which are then used for advanced controls to modify the element-wise thermal history trajectories. Controls feedback can alter laser power, interlayer dwell time, or scan strategy, among other variables. We explore controls methods to induce convergence of the Digital Shadow back to the planned ideal reference simulation state, even in the face of unplanned process disturbances that modify thermal heat flows. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Additive Manufacturing, Modeling and Simulation, Machine Learning |