Residual deformation and failure are two critical issues in powder bed fusion (PBF) additive manufacturing (AM) of metal products. Residual deformation caused by the non-uniform residual stress distribution dramatically affects the quality of AM product and can result in catastrophic failure in operation. Therefore, the development of an effective numerical approach to predict residual deformation and failure characteristics of AM product is always a major concern in industrial applications.

In this paper, a numerical approach in predicting residual distortion, stress and failure in AM products is presented. The conventional inherent strain method used in welding process is modified to consider the specific characteristic of AM process, such as the influences of reheating and scanning pattern. This approach consists of three simulation steps including a detailed process simulation in small-scale, a onetime static mechanical finite element analysis in part-scale, and a material failure analysis. First, the inherent strains are calculated from a thermo-mechanical process simulation in small-scale, which considers AM process parameters, such as laser power, scanning speed and path. The physical state in deposited materials including powder, liquid and solid states are taken into account in the simulation by specifying the solidus and liquidus temperature and corresponding material properties. Then the inherent strains are applied layer by layer to the part-scale simulation, where the residual distortion and stress can be predicted efficiently. Finally, a Lagrange particle method is utilized to study the failure characteristics of AM products. Numerical examples are studied to investigate the effectiveness and applicability of present approach.

The transient thermal process of a thin-wall beam with CO2 Gas Metal Arc Welding (GMAW) is analyzed by Finite Element Analysis Method (FEA). The thermal input is simplified as transient section body heat sources and loaded as its actual sequence in the analysis. The transient temperature field obtained can represent the basic characteristics of the real welding process and can be used as the foundation of thermal elastic-plastic analysis. Based on the temperature field, thermal elastic-plastic FEA is performed on the thin-wall beam. The distribution and change of the welding deformation, stress and strain are obtained and compared with the experiment results. Also an improvement can be presented on the inherent strain method. Using the inherent strain method, the welding deformation of the thin-wall beam is calculated. The temperature loading method is developed to load the variable inherent strain value expediently. The loading of inherent strain value on spatial welding line that is unparallel to the global coordinate axis is achieved with the application of element coordinate system. Comparison with the experiment results shows that both the thermal-elastic-plastic analysis and inherent strain analysis method can be used to predict the welding deformation effectively, the results calculated by both the thermal-elastic-plastic analysis and inherent strain analysis are close to the test measure results.