| About this Abstract |
| Meeting |
MS&T21: Materials Science & Technology
|
| Symposium
|
Additive Manufacturing Modeling and Simulation: Microstructure, Mechanics, and Process
|
| Presentation Title |
P3-5: Reinforcement Learning Aided Simulations for Determining Process Parameters for Optimizing Microstructure in LPBF Additive Manufacturing Parts |
| Author(s) |
Junwon Seo, Joseph Pauza, Anthony Rollett |
| On-Site Speaker (Planned) |
Junwon Seo |
| Abstract Scope |
3D-printing of alloys via laser powder bed fusion (LPBF) additive manufacturing has led us to a new possibility of manufacturing complex parts for various applications. However, current printing technique generally relies on a set of predefined process parameters for the entire process. In this research, the optimal process parameter function for 3D-printing an optimized microstructure in IN718 is obtained by applying reinforcement learning technique to mesoscale Monte-Carlo grain growth simulation data. The melt pool morphology and scan strategy in the microstructure simulations is varied with respect to time to generate the data to train the algorithm. The algorithm chooses its optimal process parameter for each time step, which in turn leads us to an adequate parameter selection for achieving optimized microstructure in parts. This research suggests a new opportunity for controlling the process parameter during the printing process to obtain desirable microstructural features and properties in printed parts. |