Multi-objective optimization of binder jet additive manufacturing of Co-Cr-Mo using machine learning

Recep Onler, Ahmet Selim Koca, Baris Kirim, Emrecan Soylemez*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

15 Citations (Scopus)

Abstract

Binder jet additive manufacturing is a viable method that can be used with a variety of materials, such as metals and ceramics, and offer higher speed and reduced cost for high-volume production compared to powder bed fusion printing. However, the true potential of binder jet additive manufacturing has yet to be realized due to a significant knowledge gap in the material-process-property relationships. This study presents a robust approach to identifying ideal process parameters for binder jetting of Co-Cr-Mo alloy to enable accurate and reproducible manufacturing of defect-free products with high green density. An experimental investigation of the effect of process parameters on output quality was performed using the design of experiments, followed by the analysis of variance. Further characterization between process parameters and process outputs was revealed using machine learning techniques. The relationship between process parameters and density, dimensions and surface quality of green parts was established using artificial neural networks; qualitative features of green parts were classified by employing the weighted k-nearest neighbors algorithm. Furthermore, a genetic algorithm-based multi-objective optimization was employed to determine optimum process parameters, which predicted the part quality with greater than 90% accuracy. Collectively, the proposed optimization method is capable of identifying favorable process conditions for binder jetting of defect-free Co-Cr-Mo parts, with high dimensional accuracy, surface quality, and green density, and can be adapted to determine unique process settings for different material classes.

Original languageEnglish
Pages (from-to)1091-1108
Number of pages18
JournalInternational Journal of Advanced Manufacturing Technology
Volume119
Issue number1-2
DOIs
Publication statusPublished - Mar 2022

Bibliographical note

Publisher Copyright:
© 2021, The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature.

Funding

This material is based upon the work supported by Istanbul Development Agency under grant TR10/18/YMP/0005. The authors thank Prof. Derya Dispinar and Kerem Dizdar from Istanbul Technical University for conducting pycnometer density measurements, Ahmet Nazim and Adem Sen of Gebze Technical University for their assistance in powder characterization studies, Sentes-BIR company for carbon and oxygen analysis of metal powders, and Eren Tigrel from Sinterjet Company for his assistance throughout the study. The authors would like to acknowledge the valuable discussions with Dr. Emrullah Korkmaz and Dr. Erva Ulu for improving the manuscript. This material is based upon the work supported by Istanbul Development Agency under grant TR10/18/YMP/0005. The authors thank Prof. Derya Dispinar and Kerem Dizdar from Istanbul Technical University for conducting pycnometer density measurements, Ahmet Nazim and Adem Sen of Gebze Technical University for their assistance in powder characterization studies, Sentes-BIR company for carbon and oxygen analysis of metal powders, and Eren Tigrel from Sinterjet Company for his assistance throughout the study. The authors would like to acknowledge the valuable discussions with Dr. Emrullah Korkmaz and Dr. Erva Ulu for improving the manuscript.

FundersFunder number
Istanbul Teknik Üniversitesi
Istanbul Kalkinma AjansiTR10/18/YMP/0005

    Keywords

    • Artificial neural network
    • Binder jetting
    • Co-Cr-Mo alloy
    • Genetic algorithm
    • k-nearest neighbors

    Fingerprint

    Dive into the research topics of 'Multi-objective optimization of binder jet additive manufacturing of Co-Cr-Mo using machine learning'. Together they form a unique fingerprint.

    Cite this