Mechanical response assessment of antibacterial PA12/TiO2 3D printed parts: parameters optimization through artificial neural networks modeling

Nectarios Vidakis; Markos Petousis; Nikolaos Mountakis; Emmanuel Maravelakis; Stefanos Zaoutsos; John D Kechagias

doi:10.1007/s00170-022-09376-w

Mechanical response assessment of antibacterial PA12/TiO₂ 3D printed parts: parameters optimization through artificial neural networks modeling

Int J Adv Manuf Technol. 2022;121(1-2):785-803. doi: 10.1007/s00170-022-09376-w. Epub 2022 May 21.

Authors

Nectarios Vidakis¹, Markos Petousis¹, Nikolaos Mountakis¹, Emmanuel Maravelakis², Stefanos Zaoutsos³, John D Kechagias⁴

Affiliations

¹ Mechanical Engineering Department, Hellenic Mediterranean University, Estavromenos, 71410 Heraklion, Crete, Greece.
² Department of Electronic Engineering, Hellenic Mediterranean University, Chania, Greece.
³ Department of Energy Systems, University of Thessaly, 41500 Larissa, Greece.
⁴ Department of Forestry Wood Science and Design, University of Thessaly, 43100 Karditsa, Greece.

Abstract

This study investigates the mechanical response of antibacterial PA12/TiO₂ nanocomposite 3D printed specimens by varying the TiO₂ loading in the filament, raster deposition angle, and nozzle temperature. The prediction of the antibacterial and mechanical performance of such nanocomposites is a challenging field, especially nowadays with the covid-19 pandemic dilemma. The experimental work in this study utilizes a fully factorial design approach to analyze the effect of three parameters on the mechanical response of 3D printed components. Therefore, all combinations of these three parameters were tested, resulting in twenty-seven independent experiments, in which each combination was repeated three times (a total of eighty-one experiments). The antibacterial performance of the fabricated PA12/TiO₂ nanocomposite materials was confirmed, and regression and arithmetic artificial neural network (ANN) models were developed and validated for mechanical response prediction. The analysis of the results showed that an increase in the TiO₂% loading decreased the mechanical responses but increased the antibacterial performance of the nanocomposites. In addition, higher nozzle temperatures and zero deposition angles optimize the mechanical performance of all TiO₂% nanocomposites. Independent experiments evaluated the proposed models with mean absolute percentage errors (MAPE) similar to the ANN models. These findings and the interaction charts show a strong interaction between the studied parameters. Therefore, the authors propose the improvement of predictions by utilizing artificial neural network models and genetic algorithms as future work and the spreading of the experimental area with extra variable parameters and levels.

Keywords: 3D printing; Antibacterial; Artificial neural networks (ANN); Fused filament fabrication (FFF); Polyamide 12 (PA12); Titanium dioxide (TiO2).