Mechanical behavior analysis of additively manufactured parts using the Taguchi method and artificial neural networks

Purpose Acrylonitrile butadiene styrene is an important material in 3D printing due to its strength, durability, heat resistance and cost-effectiveness. These properties make it suitable for various applications, from functional prototypes to end-use products. This study aims to model and predict the mechanical properties of acrylonitrile butadiene styrene parts produced using the fused deposition modeling process. Design/methodology/approach The experiment was carefully designed to determine the optimal print parameters, including layer thickness, nozzle temperature and infill density. Tensile tests were performed on all printed samples following industry standards to gauge the mechanical properties such as elastic modulus, ultimate tensile strength, yield strength and breakpoint. Taguchi optimization and variable analysis were used to explore the relationship between mechanical properties and print parameters. Furthermore, an artificial neural network (ANN) regression model was implemented to predict mechanical properties based on varying print conditions. Findings The results demonstrated that layer thickness has the most significant influence on mechanical properties when compared to other print conditions. The optimization approaches indicated a clear relationship between the selected print parameters and the material’s mechanical response. For acrylonitrile butadiene styrene material, the optimal print settings were determined to be a 0.25 mm layer thickness, a 270 °C nozzle temperature and a 30 % infill density. Moreover, the ANN model notably excelled in predicting the yield strength of the material with greater accuracy than other mechanical properties. Originality/value Comparing the accuracy and capabilities of the Taguchi and ANN models in analyzing mechanical properties, it was found that both models closely matched the experimental data. However, the ANN model showed superior accuracy in predicting tensile outcomes. In conclusion, while the ANN model offers higher predictive accuracy for tensile results, both Taguchi and ANN methods are effective in modeling the mechanical properties of 3D-printed acrylonitrile butadiene styrene materials.