Book Description

"Selective laser melting (SLM) is a commonly used powder bed fusion (PBF) additive manufacturing (AM) process that fabricates a part through layer-wised method. Due to its ability to build customized and complex parts, SLM process has been broadly studied and applied in both academia and industry. However, rapidly changing thermal cycles and extremely high-temperature gradients in the melt pool induce a periodically changed thermal stress in solidified layers. Different types of manufacturing defects can be induced by this laser melting and layer-wised manufacturing method. These defects are controlled by different process parameters and can be minimized through optimizing these parameters. The high cost is typically the result of experimental trial-and-error methods when they are used to optimize the related process parameters. Therefore, most studies focus on developing numerical methods to estimate transient temperatures and thermal stress distributions in the melt pool and powder bed. The big challenge in the numerical thermo-mechanical analysis of a part during the SLM process is to reduce its high computational cost. The high computational cost origins from the non-linear thermo-elasto-plastic material behavior during the fabrication process, fast laser melting and solidification process, and dynamically changed build domain. Though some numerical methods, such as multiscale method and inherent strain method, have been utilized to model the SLM process, these methods cannot incorporate the influences of many process parameters such as the scanning pattern and scanning speed. In this research, an efficient thermo-mechanical finite element (FE) method aiming to reduce the computational cost is developed to model the SLM process at part level. This simulation scheme is based on an open source FE library named Deal.II, which supports adaptive mesh refinement and parallel computing. High computational cost mainly originates from large cell number and time step number. To reduce the computational cost, the Gaussian line heat source (GLHS) model with a proper time step length was developed to replace the conventional used moving Gaussian point heat source (GPHS) model. In addition, several mesh strategies were developed to reduce the total cell number and timestep number from scanning track level, layer level, and part level, respectively. To achieve a compromise between computational efficiency and solution accuracy, a hybrid of GLHS and GPHS was developed as the input heat flux. The modeling results validated the robustness of the hybrid model. To further improve solution accuracy, temperature-dependent material properties were used and the developed adaptive mesh strategy could always capture the effective heat input by increasing the mesh density around the heat source region. After transient thermal analysis of each step, a thermo-elasto-plastic constitutive model was established to predict the quasi-static mechanical behavior of the material and to calculate the deformation and thermal stress of the deposited layers. A scanning path file was designed to include all necessary process parameters and was used to guide the simulation process track-by-track and layer-by-layer. In summary, the simulation speed is 12 ~ 18 folds faster compared with the conventional simulation scheme. The simulation results were compared with experimental results. The comparison demonstrated that each point in the simulation experienced the same thermo-mechanical cycles as in the experiment. Therefore, the developed simulation scheme in this research can be used to optimize the process parameters, such as scanning pattern, scanning speed, and layer thickness. It also has the potential to be easily extended into other PBF based AM processes"--