Probabilistic fatigue life modelling of additively manufactured aluminum alloys via Bayesian inference
Über dieses Buch
Metal additive manufacturing (AM) of aluminum alloys has shown great potential in high-performance lightweight manufacturing industries, where the undesirable fatigue failure remains a critical challenge for reliable design. This paper focuses on probabilistic fatigue life prediction of AM aluminum alloys by considering the inherent variations in fatigue process and the uncertainties in limited fatigue data. A Bayesian inference-based uncertainty quantification framework for fatigue life prediction is proposed in this paper. Fatigue experiments are conducted with AlSi10Mg alloy specimens fabricated via Laser Powder Bed Fusion (LPBF) under two building direction (BD), generating the original experimental dataset. Four candidate models—the Coffin-Manson-Basquin, Morrow, Ostergren, and Kohout-Věchet models—are adopted for fatigue life prediction. Using the valid but scarce fatigue data, the corresponding uncertain model parameters are inferred via a sampling-based Bayesian updating technique. The results demonstrate that the 95% confidence intervals of the fatigue life models updated via Bayesian inference effectively encompass most of the experimental data.
