Parametrically upscaled model-based predictive platform for fatigue with location-specific microstructural linkages.

Area of Science:

• Materials Science
• Mechanical Engineering
• Computational Mechanics

Background:

• Dwell fatigue in metallic materials, especially titanium alloys like Ti-6Al-4V, presents a significant engineering challenge due to unpredictable crack nucleation under cyclic loading with hold times.
• Microstructural heterogeneity, including texture and anisotropic properties, critically influences fatigue crack evolution and life prediction.

Purpose of the Study:

• To develop a spatiotemporal multiscale computational platform for predicting probabilistic dwell fatigue crack nucleation at the component scale.
• To link location-specific microstructural features to fatigue crack initiation and propagation behavior.

Main Methods:

• Integration of physics-based modeling, machine learning, temporal acceleration, and probabilistic analysis.
• Development of parametrically upscaled constitutive and crack nucleation models (PUCM-PUCNM) for efficient multiscale predictions.
• Experimental validation of the computational platform for nucleation studies.

Main Results:

• The developed platform accurately predicts probabilistic fatigue crack nucleation, considering geometry and microstructure interactions under loading.
• Demonstrated effectiveness of specimen test data-calibrated PUCM-PUCNMs for component-level fatigue predictions.
• Successful exploration of competing effects of geometry and microstructure on fatigue crack evolution.

Conclusions:

• The computational platform shows significant promise for multiscale fatigue predictions in metallic materials.
• The approach enables a deeper understanding of microstructure-dependent fatigue crack nucleation mechanisms.
• This work supports the design of fatigue-resistant structures and materials through accurate predictive modeling.