Page 106 - MSAM-2-3

Materials Science in Additive Manufacturing Validation of a novel ML model for AM-PSP

process category (e.g., DED) in prediction accuracy. Conflict of interest
(iii) On the other hand, cross-training datasets across AM
process categories offer a wider training boundary The authors report no conflicts of interest.
and testing capability for AM Ti-6Al-4V materials Author contributions
with superior prediction accuracy. In addition, the
workflow introduced in this research illustrates the Conceptualization: Xi Gong, Guha Manogharan
robustness of ML-based PSP models with desired Formal analysis: Xi Gong, Guha Manogharan
computational efficiency that is well suited for broader
application in AM research (e.g., prediction of wear Investigation: Xi Gong
behavior and corrosion resistance) Methodology: Xi Gong, Guha Manogharan
(iv) Development of fully functional and computationally Resources: Guha Manogharan
efficient PSP linkages of the AM Ti-6Al-4V to
investigate metal AM materials properties and Validation: Xi Gong
material response, that is, machining behavior. Visualization: All authors
Previous work in this field is based on traditional
PSP linkages that connect limited metal AM process Writing – original draft: Xi Gong, Guha Manogharan
parameters to minimal microstructure information, Writing – review and editing: All authors
and few mechanical properties cannot predict material
response as demonstrated in this study. Ethics approval and consent to participate
(v) The ML-based PSP linkage from this study was validated Not applicable.
to accurately extract multiple structure information
from metal AM parts under different manufacturing Consent for publication
procedures and accurately predict the machining
behavior during post-processing (>99% RMSE). Not applicable.
(vi) The workflow established in this research was shown Availability of data
to be robust across multiple AM surfaces (AM process,
build orientation, and post-AM heat treatment) Data collected and analyzed in this work is available from
under a single framework time with extremely high the authors upon request.
prediction accuracy. For instance, over 1800 SVEs
were involved in data analysis, and it was apparent that References
the workflow is robust and quite flexible to harness a 1. Conner JW, Manogharan BP, Martof GP, et al., 2014,
variety of datasets, material systems, and material Making sense of 3-D printing: Creating a map of additive
response, that is, machining behavior, in this research. manufacturing products and services. Addit Manuf,
1: 64–76.
Although the PSP linkage demonstrated here showed
high prediction accuracy (>99%), ongoing advancements https://doi.org/10.1016/j.addma.2014.08.005
in the AM industry will provide new AM technologies, and 2. Dutta B, (Sam) Froes FH, 2017, The additive manufacturing
continuing to maintain these PSP linkage models will be (AM) of titanium alloys. Metal Powder Rep, 72: 96–106.
essential. In the future, this PSP linkage can be expanded https://doi.org/10.1016/j.mprp.2016.12.062
for other materials and can begin to consider additional
machining responses such as tool wear behavior, surface 3. Arthur NK, 2019, Laser based manufacturing of ti6al4v:
finish, and tolerance analysis. A comparison of lens and selective laser melting. Mater Sci
Forum, 950: 44–49.
Acknowledgments https://doi.org/10.4028/www.scientific.net/MSF.950.44
The authors would like to acknowledge the support from 4. Frazier WE, 2014, Metal additive manufacturing: A review.
Norsk Titanium, Dr. Saurabh Basu, Kazi Shahed, and J Mater Eng Perform, 23: 1917–1928.
Dongrui Zeng. https://doi.org/10.1007/s11665-014-0958-z

Funding 5. Baufeld B, Brandl E, Van Der Biest O, 2011, Wire based
additive layer manufacturing: Comparison of microstructure
Partial support for this study was provided by and mechanical properties of Ti-6Al-4V components
NIST AMTech (Grant no. 70NANB15H070), and fabricated by laser-beam deposition and shaped metal
the Manufacturing PA program. deposition. J Mater Process Technol, 211: 1146–1158.

Volume 2 Issue 3 (2023) 14 https://doi.org/10.36922/msam.0999