Reliability oriented degradation modeling of fused filament fabrication printed acrylonitrile butadiene styrene components under cyclic mechanical loads using explainable AI techniques

The fatigue performance of Fused Filament Fabrication (FFF) printed acrylonitrile butadiene styrene (ABS) is critically governed by process-induced anisotropy, yet the integration of reliability analysis with explainable machine learning remains unexplored. This study presents a reliability-oriented Explainable Artificial Intelligence (XAI) framework to model fatigue degradation in FFF fabricated ABS. An empirical dataset was generated by testing 150 specimens with systematically varied printing parameters. Reliability analysis using a two-parameter Weibull distribution yielded a shape parameter (β) of 2.48, confirming a distinct wear-out failure mode. Among the tested algorithms, Gradient Boosting achieved the highest predictive accuracy (R²= 0.90). Explainability analyses via SHAP and LIME revealed that build orientation and layer height are the dominant factors influencing fatigue life, aligning with physical degradation mechanisms. This framework offers a transparent tool for fatigue life prediction and process optimization in additive manufacturing.