Doctoral Dissertations

Keywords and Phrases

Mathematical Modeling; Metal Additive Manufacturing; Model Validation

Abstract

"This research aims to present a methodology for optimizing input datasets to increase the accuracy of mathematical models and accelerate their application to engineering problems. To accomplish this goal, this work focused on the application of mathematical models to metal additive manufacturing (AM), specifically the thermal history of laser directed energy deposition (DED) of aluminum alloys. The initial steps of this body of work were to develop a mathematical model that is capable of simulating the metal AM process and applying it to the laser DED of aluminum. It was validated using the well characterized material Ti-64 and shown to have an error of 3% when predicting the width of the melt track and 20%, or less than 2 resolution steps, when predicting the depth of the melt track. Upon validation, the input parameter dataset which had the most impact on the thermal history was determined using a sensitivity analysis design of experiment (DOE), these properties were the absorption of the laser at 607◦ C and 649◦ C, the thermal conductivity at 649◦ C, thermal conductivity at 1281◦ C, and the specific heat at 460◦ C. Upon down selection of the input parameter to increase search algorithm efficiency, a Nelder-Mead search algorithm was applied to the simulation which developed an optimized input dataset. This dataset was able to increase the accuracy of the simulation from the original dataset by over 500%, increasing the accuracy from over 600% for a generic aluminum alloy to 9.1%. It was found that the values of the laser absorption at the liquidus temperature and the specific heat at 733◦ C, for the optimized dataset were triple that of the generic dataset. Conversely, at 922◦ C, the generic dataset was triple that of the optimized dataset values. The thermal conductivity of the optimized dataset was about double that of the generic dataset at 1491◦ C. Lastly, the laser diameter rudely estimated via experimentation was nearly double that of the optimized input dataset. This methodology of model development, critical parameter selection, and the application of a search algorithm is applicable across mathematical models and disciplines" -- Abstract, p. iv

Advisor(s)

Liou, Frank W.

Committee Member(s)

Chandrashekhara, K.
Midha, A. (Ashok)
Kinzel, Edward C.
Newkirk, Joseph William

Department(s)

Mechanical and Aerospace Engineering

Degree Name

Ph. D. in Mechanical Engineering

Publisher

Missouri University of Science and Technology

Publication Date

Summer 2024

Pagination

xii, 114 pages

Note about bibliography

Includes_bibliographical_references_(pages 34, 54, 76, 104 & 111-113)

Rights

©2024 Aaron Flood , All Rights Reserved

Document Type

Dissertation - Open Access

File Type

text

Language

English

Thesis Number

T 12382

Electronic OCLC #

1460027108

Download

Share

 
COinS