Residual thermal stress prediction for continuous tool-paths in wire-arc additive manufacturing: a three-level data-driven method

Abstract

Continuous tool-path is often chose to improve the deposition efficiency and surface accuracy of metal additive manufacturing, while it also causes large residual thermal stress, which will result in part deformation and performance degradation. This paper focused on wire-arc additive manufacturing (WAAM) with arbitrary part geometries and continuous tool-paths, and proposed a three-level data-driven method to predict the residual thermal stress filed accurately and rapidly. The first two-level of the proposed method predict the thermal field history of the whole WAAM process. The third level of the proposed method realises the residual thermal stress field prediction of WAAM based on above prediction results. Each level is based on a machine learning method, and their data were obtained based on the finite element method. The prediction accuracy of the proposed method exceeded 92%, and the time cost of one prediction process was only at the second level.

Keywords

• wire-arc additive manufacturing
• thermal field history prediction
• residual thermal stress field prediction
• machine learning