| Summary: | Despite the flexibility offered by additive manufacturing (AM), various process factors ranging from design to material can lead to inconsistent repeatability and quality issues from build-to-build. Currently, AM process monitoring mainly focuses on gaining an understanding of physical phenomena involved in a single quality indicator, ignoring the correlation between heterogeneous characteristics that can provide insight useful for accurate analysis. In this study, a multimodal Gaussian process methodology is designed to simultaneously quantify interactions between AM design parameters at play (e.g., orientation, thickness, height, and contour spacing) and heterogeneous quality characteristics (e.g., distortion, porosity, and roughness) when fabricating thin walls using AM. A real-world study is performed on Ti6Al4V thin wall structures fabricated with laser powder bed fusion, multiple design parameters are varied to evaluate the capability of the proposed method in quantifying such interactions. Experimental results using x-ray computed tomography show that the proposed multimodal Gaussian process provides an improvement of 21.7% compared to the single response model with respect to the root mean square error. The proposed methodology shows great potential to quantify heterogeneous characteristics of quality according to the process factors in a variety of AM builds, such as lattice and honeycomb structures.
|
|---|