Prediction and Sensitivity Analysis of Bubble Dissolution Time in 3D Selective Laser Sintering Using Ensemble...

Area of Science:

• Materials Science
• Manufacturing Engineering
• Computational Science

Background:

• Gas bubbles are inherent defects in selective laser sintering (SLS) parts.
• Accurate prediction of bubble shrinkage dynamics is critical for SLS process optimization.
• Understanding bubble dissolution is key to improving the quality of fabricated components.

Purpose of the Study:

• To develop and evaluate artificial intelligence (AI) models for predicting bubble dissolution time in SLS.
• To compare the performance of Ensemble Bagged Trees (EDT Bagged) and Ensemble Boosted Trees (EDT Boosted) models.
• To identify key input parameters influencing bubble dissolution time.

Main Methods:

• Generated a metadata set of 68,644 data points using a numerical tool.
• Constructed two AI models based on the Decision Trees algorithm: EDT Bagged and EDT Boosted.
• Evaluated model performance using Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and R-squared (R²).
• Performed sensitivity analysis using the Monte Carlo approach.

Main Results:

• The EDT Bagged model outperformed the EDT Boosted model in predicting bubble dissolution time.
• Sensitivity analysis revealed diffusion coefficient, initial concentration, and initial bubble size as the most influential input parameters.
• The AI models effectively predicted bubble dissolution time using parameters like bubble size, diffusion coefficient, and chamber pressure.

Conclusions:

• AI models, particularly EDT Bagged, offer a viable method for rapid prediction of bubble dissolution time in SLS.
• This predictive capability can aid in optimizing SLS process parameters.
• The findings contribute to enhancing the quality and reliability of parts produced via selective laser sintering.