Fatigue Analysis of Printed PLA using Neural Networks

The implementation of additive manufacturing as a disruptive process for the development of components has generated the need to know manufacturing parameters to obtain mechanical properties similar to or better than those of conventional processes. Thus, the improvement and optimisation of materials used have aroused interest in converting Polylactic acid (PLA) 3d printed parts from prototyping to functional components. In this regard, they must withstand cyclical loads. To do so, conventional methods must consider all variables used in the manufacturing process. Herein, we propose the use of neural networks to perform mechanical fatigue analysis of the printed components of PLA. The mechanical behaviour of the printed components was expressed for the neural network as parameters based on quasistatic tests, and the output of the network was the expected fatigue life in cycles. The initial quasistatic behaviour of the 3d printed parts was linearly elastic. However, a viscoelastic behaviour developed over time. The discharge time between charging cycles influenced the cumulative damage process. According to the experimental results and their correlations, the fatigue life of printed components can be predicted using neural networks by achieving an average error of 1.59%. © (2024), (International Association of Engineers). All Rights Reserved.