Combining synchrotron X-ray diffraction, mechanistic modeling and machine learning for in situ subsurface temperature...

Area of Science:

• Materials Science
• Manufacturing Engineering
• Computational Modeling

Background:

• Laser melting in additive manufacturing creates extreme thermal gradients impacting material microstructure.
• Accurate characterization of these temperature fields is essential for process and material validation.
• Directly measuring subsurface temperatures during laser melting remains a significant challenge.

Purpose of the Study:

• To develop a novel method for extracting subsurface temperature distribution metrics during laser melting.
• To quantify the uncertainty associated with these temperature measurements.
• To provide in situ quantitative temperature evolution data for laser melting processes.

Main Methods:

• Utilized in situ synchrotron X-ray diffraction measurements.
• Developed Gaussian process regression supervised machine-learning surrogate models.
• Trained models with mechanistic heat transfer, fluid flow, and X-ray diffraction simulations.

Main Results:

• Successfully extracted subsurface temperature metrics with uncertainties ranging from 5-15%.
• Applied models to Inconel 625 alloy, confirming maximum solid-phase temperatures reached the solidus.
• Observed increased uncertainty during cooling due to thermomechanical stress.

Conclusions:

• The developed method provides a viable approach for in situ subsurface temperature characterization during laser melting.
• Machine learning surrogate models offer a powerful tool for analyzing complex thermal processes.
• Understanding temperature evolution is critical for controlling microstructure and properties in additive manufacturing.