|About this Abstract
||7th World Congress on Integrated Computational Materials Engineering (ICME 2023)
||Statistical Learning Approaches for Predicting Pore Formation from In-situ Characterization for Additive Manufacturing of SS 316L Using Laser Powder Bed Fusion
||Erika I. Barcelos, Nathaniel K. Tomczak, Jayvic Cristian Jimenez, Sameera Nalin Venka, Kristen J. Hernandez, Raymond Wieser, John J. Lewandowski, Laura Bruckman, Roger French
|On-Site Speaker (Planned)
||Erika I. Barcelos
Designing material systems with reproducible desirable properties remains a challenge for the fabrication of metal additive parts. The overall performance of a material can be heavily influenced by both direct and indirect stressors.
In this work, spatiotemporal models were leveraged to model the spatial and temporal coherence between deposits of both single and multi-track additively manufactured stainless steel (SS 316L) samples using laser powder bed fusion (LPBF). Initially, interactions and potential additive effects between variables were established using exploratory data analysis. After, statistical modeling was used to generate insights and model the data. Network structural equation modeling (netSEM), p-values, and correlations were employed to perform variable selection followed by modeling using generalized linear models (GLM) and general additive models (GAM). These techniques successfully modeled the data which suggests that predicting pore formation for LPBF deposits is possible using statistical learning approaches, even when a limited number of points are available.