Identifying uncertainty in laser powder bed fusion additive manufacturing models

Area of Science:

• Materials Science and Engineering
• Computational Modeling and Simulation
• Manufacturing Processes

Background:

• Additive manufacturing (AM) models are crucial for design and process planning but often lack precision and accuracy information.
• Model uncertainty is critical for validating AM models, qualifying parts, and managing process variability.
• Laser powder bed fusion (L-PBF) is a key AM technology where model uncertainty significantly impacts outcomes.

Purpose of the Study:

• To discuss the origins and propagation of uncertainty in L-PBF models.
• To present techniques for quantifying uncertainty from various sources.
• To explore methods for reducing prediction uncertainty using online measurements.

Main Methods:

• Identification of four primary uncertainty sources: modeling assumptions, unknown simulation parameters, numerical approximations, and calibration data measurement error.
• Brief presentation of techniques to quantify uncertainty for each identified source.
• Development and application of estimation algorithms to reduce prediction uncertainty with online measurements.

Main Results:

• Quantification of model uncertainty for single-track experiments in L-PBF.
• Simulation-based study of online estimation's effect on overhanging structures.
• Demonstration of uncertainty quantification and reduction techniques using a transient, stochastic thermal model for melt pool width prediction.

Conclusions:

• Understanding and quantifying uncertainty in L-PBF models is essential for reliable AM processes.
• Online measurements and estimation algorithms can effectively diminish prediction uncertainty.
• The presented concepts offer a pathway for improved estimation and control in L-PBF.