Evolution of the microstructure and multi-objective optimization of the tensile properties of GH3625 superalloy by selective laser melting

Abstract

Selective laser melting (SLM) is one of the most widely used metal additive manufacturing (AM) technologies and has great potential for forming complex structured parts. As GH3625 superalloy becomes more demanding for complex structured parts, making SLM offers a unique opportunity for manufacturing complex structured superalloys. However, there are relatively few studies on the microstructural evolution of GH3625 superalloys prepared by SLM, and the high tensile strength but the low elongation of superalloy parts prepared by SLM are difficult to meet practical applications. To this end, the effects of different volume energy densities (VEDs) on the microstructure and fracture characteristics of superalloys are investigated in this paper. In addition, a method combining response surface methodology (RSM) with non-dominated ranking genetic algorithm-III (NSGA-III) is proposed to generate optimal process parameters to maximize yield strength, ultimate tensile strength and elongation of superalloys. To demonstrate that the application of the NSGA-III algorithm is outstanding, the NSGA-Ⅱ algorithm is considered for application to this optimization problem and the hypervolume (HV) index is used to objectively evaluate the superiority of these algorithms. The best Pareto front obtained is also verified. The results show that the melt pool size is proportional to the VED in the studied range, and the deepest molten pool is about 74 μm when the VED is 147.62 J/mm3; the smallest cellular grain in the XOY direction is about 0.27 μm when the VED is 66.14 J/mm3. In the optimization of tensile properties, the yield strength, ultimate tensile strength and elongation models constructed by RSM are significant with R2 of 0.991, 0.9621 and 0.9391, in that order. The evaluation index HV indicates that the Pareto front obtained by NSGA-III has a higher priority than the Pareto Front obtained by NSGA-II with HV Finally, the obtained Pareto front was experimentally verified and the maximum relative errors of yield strength, ultimate tensile strength and elongation were only 0.87%, 0.57% and 2.89%, in that order. The proposed optimization method has high acceptability and will have a wider application area.