Doctoral Dissertations

Orcid ID

https://orcid.org/0000-0002-7081-3870

Date of Award

12-2024

Degree Type

Dissertation

Degree Name

Doctor of Philosophy

Major

Data Science and Engineering

Major Professor

Russell Zaretzki

Committee Members

Vincent Paquit, Hairong Qi, Mingzhou Jin

Abstract

Before metal powder bed fusion additive manufacturing may be deployed on a wide scale, the reliability and repeatability of the process must be improved. Currently additively manufactured components suffer from mechanical anisotropy and often-poor performance due to the presence heterogeneities in microstructure and defects throughout the component. Work must be undertaken to help determine how the process itself contributes to the evolution of these heterogeneous structures. This is true for both electron beam and laser powder bed fusion technologies. The dynamism of the process is largely driven by the scan path that the heat source traverses to fuse feedstock into solid material - introducing nuanced thermal cycling and histories to different regions. The complexity of the process, time and space scales under consideration, and the increasing availability of data necessitate development of additional process-structure analyses to complement those in existence already. Two distinct tasks will be performed to explore localized heterogeneities and their connection to the powder bed process utilizing scan path data. The first task will investigate how the scan path itself may be manipulated in order to induce desired location specific microcstructure in electron beam powder bed fusion utilizing genetic algorithm and concurrent greedy line search optimization algorithms. The second task will leverage scan path data to localize temporal process data. Specifically, the scan path data will be utilized to spatially map temporal data about gas flow and oxygen content of laser powder bed fusion builds. The data, in its new format, will then be used to predict the local prevalence of soot which has been shown to lead to mechanical anisotropy.

Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS