About this Abstract |
Meeting |
MS&T24: Materials Science & Technology
|
Symposium
|
Computational Materials for Qualification and Certification
|
Presentation Title |
Efficient Sensitivity and Uncertainty Analysis of a Laser Powder Bed Fusion Thermal Model Built Using HYPAD-FEM |
Author(s) |
Harry Millwater, Juan-Sebastian Rincon-Tabares, Samuel Roberts, Matthew Balcer, Mauricio Aristizabal, Arturo Montoya, David Restrepo |
On-Site Speaker (Planned) |
Harry Millwater |
Abstract Scope |
This study introduces a methodology to perform an efficient sensitivity and uncertainty analysis of a metal laser powder bed fusion additive manufacturing thermal simulation. The methodology implements HYPercomplex-based Automatic Differentiation within traditional FEM formulations to deliver fast and accurate sensitivity solutions. HYPAD-FEM was used to obtain multiple arbitrary-order sensitivities of the thermal history with respect to initial conditions, loading conditions, thermal properties, powder bed properties, and geometric shape. The sensitivities were used to obtain estimates of the mean and variance of the thermal behavior and to perform global sensitivity analysis. The model incorporated material property variations due to temperature, phase change, and solidification. The HYPAD-FEM statistics and global sensitivity results were compared against standard Monte Carlo, random Gaussian processes, and polynomial chaos expansions. Good comparisons were between all methods were obtained; however, HYPAD-FEM required 10-100X reduced computational time. |