Abstract
Metal-based additive manufacturing (MAM) method with high freedom and special fabricate technology has presented great universality in the aerospace and biomedicine field. However, a wide range of process parameters in the MAM method challenge the experimental study on the formation and evolution of defects. The numerical simulation presents its excellent accuracy and economy in predicting the evolution of multiphysics phenomena and was hence widely applied. In the current review, the available MAM methods with the fundamental phenomena were reviewed. Based on scales, numerical approaches divided into three categories were discussed and focused on their main prediction objectives and strengths or weaknesses of all the scales. To display the prediction results closer to real physical phenomena, advanced multiscale models coupled with various single-scale models are provided. The high prediction accuracy and computational efficiency enable better parameter control and defect avoidance. As a supplement and development to the physical-driven model, the data-driven model provides a new perspective on MAM methods. Based on the database generated by the physical-driven model and experiment, the data-driven models without calibration of input parameters are shown. In addition, this review discussed the development direction of numerical simulation, aiming to provide a reference for technical research in this field.