Numerical Simulation And Sintering Zone Prediction Of Direct Laser Sintering Of Multi-component Metal Powders

Direct Metal Laser Sintering (DMLS) of multi-component metal powders is an emerging Rapid Prototyping and Manufacturing (RPM) technology which can be used to produce three-dimensional metal parts directly from a CAD model by the selective laser sintering (SLS) of successive multi-component layers of metallic or pre-alloyed powders. DMLS is becoming a significant trend of Direct Manufacture (DM) for the unique advantage in direct manufacturing of metal parts and eliminating the expensively time-consuming pre- and post-processing steps, compared to indirect laser sintering or other conventional processes. However, in contrast to nonmetal material, DMLS of metal powders needs much higher powder laser, is liable to balling and deforming, and is strongly affected by material properties, temperature and stress distribution whose effects on sintering process is becoming one of the research focuses. Heat transference plays an important role in powder densification in DMLS. Unfortunately, it is hard to precisely measure the transient distributions of temperature, density and stress since they are characterized by fast-changing and great gradient, and the process scheme and suitable process parameters are mostly determined by costly energy- and time-consuming experimental trial presently, so research interests on DMLS mechanism by the use of numerical simulation has grown significantly in recent years. By using numerical method to simulate physical process including transient temperature, density and thermal stresses fields, the influence of concerning material and process parameters on sintering process can be analyzed and sintering zone can be predicted at the given parameters, which is valuable to dramatically reduce the times of the costly experiments, to save experimental cost, to optimize the fabrication process, to improve the sintering parts performance and to eliminate balling in DMLS.The research in this dissertation focuses on modeling of nonlinear and transient powder-to-solid transition of material thermal conductivity during DMLS process, analyzing transient temperature, density, and thermal stresses distribution in DMLS, indirectly verifying temperature distribution by experiments, and predicting sintering depth and sintering width by using Finite Element Method (FEM) and Artificial Neural Network (ANN). This research is of great theoretical and engineering significance. The main characteristics and creative ideas of the dissertation are as follows:1. The nonlinear transient model of the thermal conductivity of metals for powder-to-solid transition is proposed. The change of material thermal conductivity in powder-to-solid transition can not be ignored for the heat transference by the loose powder is much lower than by the nearly full-density solid body, for example, thermal conductivity of Cu is about 100 times bigger than that of Cu powder. The thermal conductivity model, in the range from solidus to liquidus temperature for powder-to-solid transition, is established according to liquid phase sintering (LPS) theory. In addition, the thermal conductivity model of the whole temperature scope is developed in sections. This model addresses the continuously changing in powder-to-solid transition of material thermal conductivity, and more fully considers the significant effects of wide solidus-liquidus temperature scope and great changing thermal conductivity during powder-solid transit of pre-alloyed metal, which lays a foundation for developing a precision model.2. A three-dimensional finite element thermal model including the effect of powder-to-solid transition is established. The effect of laser-induced thermal shock on powder bed is considered by using an amendatory coefficient, and solid-liquid latent heat effect is taken into account with the latent heat option. The model uses solid material properties in liquid-phrase zone, transitional ones in sintering and sintered zone, and powder ones in other zones of powder bed respectively. A method to transform element material properties between load steps based on temperature history and region of powder bed is put forward to take account of nonlinear transient material behavior for powder-to-solid transition in DMLS.3. Based on established Finite Element Analysis (FEA) model, simulations of temperature and density fields are carried out with the use of multi-componentCu-based metal powder. The relationship between various material and processing parameters and the sintering quality is discussed to provide a guideline to optimize process of DMLS. Some conclusions related to the results of the simulations and the comparisons between various material and process parameters are reached: (1) When adopting the method to transform element material properties between load steps based on temperature history and region of powder bed, the temperature gradient around the sintering zone boundary is much higher than that within the sintering zone, and heat affected zone (HAZ) is confined in sintering zone while HAZ expands the whole powder bed without adopting the method, which is consistent with the experimental results with the use of this method. (2) When using substrate, the maximum temperature in powder bed becomes lower and sintered zone is reduced, which mainly is caused by the higher consistent thermal conductivity of substrate. Measures of employing moderate thermal conductivity substrate and preheating substrate should be taken lest input-heat mainly transfers to substrate. (3) Short scan length benefits the sintering performance. "Dead corner" of HAZ hardly exists due to adequate heat transfer with short length of scan vector while "Dead corner" lies in some zones with longer length. This is probably due to shorter heat cycle period and less fluctuant energy using shorter scan length. (4) Raising laser power, slowing scan-speed and reducing scan space inevitably increase energy input into powder bed which leads to better sintering parts performance. Higher thermal conductivity can improve sintering quality in that stepped temperature distribution at the bottom of sintering zone is disadvantage to sintering part with lower thermal conductivity. Greater heat capacity is good for better accuracy of sintering parts in that the sizes of HAZ and sintering zone are smaller using greater heat capacity.4. A stress FEA model is developed with the employment of thermal elastic-plastic FEA in DMLS and the related simulations are conducted to get knowledge about the stresses distribution. The model encompasses the effects of the temperature- and element-dependent mechanical properties of material, and an analysis of indirect coupled thermal-stress is performed in which a sequentially method is used, namely, the stress analysis based on the results of temperature analysis is conducted until the temperature analysis is finished. The obtained results show that the stresses mainly locate the very small region of laser radiation, andcompressive stress occurs at top of sintering region while tensile stress exits around the boundary of sintering region.5. The indirect experimental verification method of the FEA thermal model is proposed in which the isotherm distribution is estimated by using sintering depth and sintering width. After laser is turned off, the region whose temperature is higher than liquidus temperature can be obtained by measurement of sintering width and sintering depth, and is compared with correspond region got by numerical simulation to verify the accuracy of the FEA model. Experiments are carried out by using multi-component Cu-based metal powder. Comparisons between experimental and predicted results at 9 process parameters show that percent error of sintering depth is ranging from -10.7% to 20.0% with mean percent error 7.8%, and percent error of sintering width ranging from -8.2% to 27.3% with mean percent error 16.7%, which verifies indirectly the accuracy of FEA thermal model. Moreover, the numerical simulation results of temperature field can be used to predict the sintering depth and sintering width of DMLS based on FEM, which benefits selecting material parameters, and optimizing process parameters and processing scheme of DMLS.6. Sintering depth and sintering width in DMLS is also predicted based on ANN. A two-layer supervised neural network with 7-node single hide-layer and Sigmoid transfer function is established, and non-linear mapping from 'process parameters space' to 'sintering region' is carried out. Comparisons between experimental and predicted results show that percent error of sintering depth is ranging from -11.8% to 19.6% with mean percent error 8.3%, and percent error of sintering width ranging from -18.5% to 17.2% with mean percent error 12.2%, which is valuable to guide to select suitable process parameters before experiments.