Most failures in component operation occur due to cyclic loads. Validation has been performed under quasistatic loads, but the fatigue life of components under dynamic loads should be predicted to prevent failures during component service life. Fatigue is a damage accumulation process where loads degrade the material, depending on the characteristics and number of repetitions of the load. Studies on the mechanical fatigue of 3D-printed Onyx are limited. In this paper, the strength of 3D-printed Onyx components under dynamic conditions (repetitive loads) is estimated. Fatigue life prediction is influenced by manufacturing processes, material properties, and applied loads, which can cause scatter in the results due to the interplay of these factors. By utilizing synthetic parameters derived from mechanical properties, the accuracy of fatigue life predictions has been improved significantly, from 23.13% to 98.33%. Additive manufacturing is flexible, but this flexibility generates scatter in the mechanical properties of produced components. This work also proposes the use of synthetic data with a neural network to improve the fatigue life prediction of printed Onyx subjected to tension–tension loads. Experimental uniaxial loads were used to characterize the mechanical behavior of printed specimens. The experimental data were used to evaluate the numerical predictions obtained through finite element analysis using commercial software and an artificial neural network. The results showed that the use of synthetic data helped improve fatigue life prediction.
Jimenez-Martinez et al. (2026) studied this question.