| About this Abstract |
| Meeting |
MS&T23: Materials Science & Technology
|
| Symposium
|
Additive Manufacturing Modeling, Simulation, and Machine Learning: Microstructure, Mechanics, and Process
|
| Presentation Title |
Multi-Model Monte Carlo Simulations of Mechanical Behavior of Additively Manufactured Metals |
| Author(s) |
Joshua D. Pribe, Patrick E. Leser, Saikumar R. Yeratapally, George Weber, Brodan Richter, Andrew R. Kitahara, Edward H. Glaessgen |
| On-Site Speaker (Planned) |
Joshua D. Pribe |
| Abstract Scope |
Complex microstructures and process-induced defects cause significant uncertainty in the mechanical behavior of additively manufactured (AM) metals. Quantifying uncertainty with repeated experiments is costly, particularly for structural applications where mechanical allowables must be established. Computational materials simulations can reduce these costs but with different challenges: uncertainty propagation using Monte Carlo (MC) simulation is intractable for expensive high-fidelity models, while inexpensive low-fidelity models reduce accuracy and introduce bias. In this work, multi-model MC methods are used to estimate the mean and extreme values of macroscale and micromechanical quantities of interest (QoIs) for AM microstructures. Multi-model MC methods leverage correlations among available models to generate unbiased estimates of the QoIs with reduced computational cost. Efficiency gains are demonstrated relative to conventional MC simulation. The results show the utility of multi-model MC methods for propagating uncertainty in computational materials simulations of AM. |