|About this Abstract
||2017 TMS Annual Meeting & Exhibition
||Additive Manufacturing of Metals: Establishing Location-Specific Processing-Microstructure-Property Relationships
||A-30: Machine Learning Approaches to Optimize Additive Manufacturing Parameters for SLM of Inconel 718
||Branden Kappes, Henry Geerlings, Senthamilaruvi Moorthy, Andrew Petersen, Douglas Van Bossuyt, Aaron Stebner
|On-Site Speaker (Planned)
We have tested samples from five SLM build plates, each with 625 cylindrical Inconel 718 samples, and have used these 3125 samples to develop high throughput characterization and analysis procedures. These data serve as the basis for development of machine learning (ML) algorithms, from principal component analysis to ensemble classification and neural networks, that focus on two-way modeling of the process-property and process-structure relationships. Our results show that laser power, speed, spot size, pass overlap, even sample orientation and position on the build plate significantly effect microstructure, particularly porosity, which in turn, has a pronounced effect on the mechanical performance of the sample. We will present on the data collection, processing, validation and distribution framework; on ML performance, accuracy and validation procedures; and conclude with a brief discussion on the extension of this model to other data input streams and materials systems.
||Planned: Supplemental Proceedings volume