Prediction of the mechanical properties for 3D printed rapid prototypes based on artificial neural network
More details
Hide details
1
Production Engineering and Metallurgy Deptartament University of Technology Baghdad, Iraq
Adv. Sci. Technol. Res. J. 2025; 19(3):96-107
KEYWORDS
TOPICS
ABSTRACT
With development of Additive Manufacturing (AM) especially the 3D printing technology, make it the widespread application for 3D prototypes in industrials, engineering jawless, biomedical and others filed. In the present work focused on the 3D printing problems that associated to selecting proper printing parameters. Based on the experimental and ANN the effect of printing speed, printing temperature, layer height, and number of top shells on the produced mechanical properties of the 3D prototypes. Ultimate tensile strength, yield strength, and modulus of elasticity have been studied as the main mechanical properties.
Design of experiment for specimens using a MINITAB software has been achieved based on Taguchi method based on the sixteen specimens with four levels values of printing parameters. CAD CAM software (Solid work) used to create 3D model of the testing specimens with the specific dimensions based on the ASTM E8M. ANICUBIC 3D printing machine used to fabricate the specimens under the studied 3D printing parameters. ANN has been used to validate the obtained experimental and DOE.
The obtained results showed that increasing the printing temperature up to 220oC, and high number of top shells arriving to 4 shells will increase the ultimate tensile strength, yield strength, and modules of elasticity. While decreasing the printing speed lower than 100m/sec. and decreasing layer height lower than 0.3mm will produce a gaining in the mentioned mechanical properties.
Comparison results of the experimental work and the predicted results obtained from suggested model of ANN provide the more compatibility between these values, the regression of the ANN observed that the learning of the network is proper and can be application to predict the Ultimate Tensile Stress, Yield Stress, and Modulus of Elasticity, where the validation, training, test and all of data are about (0.95592-1).