Encoding Stability into Laser Powder Bed Fusion Monitoring Using Temporal Features and Pore Density Modelling

Area of Science:

• Additive Manufacturing
• Materials Science
• Computational Modeling

Background:

• Melt pool instability in laser powder bed fusion (LPBF) leads to pores and reduced structural integrity in 3D printed parts.
• Current camera-based monitoring systems offer limited, indirect insights into melt pool stability.

Purpose of the Study:

• To improve melt pool stability in LPBF by explicitly encoding stability into monitoring systems.
• To develop a novel approach using temporal features and pore density modeling for enhanced LPBF process control.

Main Methods:

• Introduced temporal features (variances of melt pool area, intensity) to quantify printing stability.
• Developed a neural network model linking video features to pore densities from CT scans.
• Implemented and tested the monitoring system on 316L stainless steel prints.

Main Results:

• Achieved improved correlation (up to 42%) between predicted and true pore densities.
• Demonstrated the effectiveness of explicit stability quantification in LPBF monitoring.
• Reduced the need for online printer interventions by focusing on porosity avoidance.

Conclusions:

• Explicitly encoding temporal features and pore density modeling significantly enhances LPBF monitoring.
• The proposed method offers a more accurate and direct approach to ensuring part quality in 3D printing.
• This advancement contributes to more reliable and robust additive manufacturing processes.