Optimisation of selective laser melting parameters for the Ni-based superalloy IN-738 LC using Doehlert’s design

Abstract

Purpose The Ni-based superalloy IN-738 LC is known to be susceptible to porosity and different types of cracking during the build-up process and, thus, challenging to manufacture using selective laser melting (SLM). Determining a feasible set of operating parameters for SLM of nickel-based superalloys involves new approach to experimental design based on the Doehlert method that assists in determining an optimal (feasible) set of operating parameters for SLM of IN-738 LC powder alloy. Design/methodology/approach The SLM parameters are evaluated in terms of their effectiveness in obtaining the microstructure with a porosity content of <0.5 per cent and without micro-cracking. The experimental approach is exemplified with the Doehlert matrix response variable, relative density, by comparing Archimedes method with microstructural assessments of pores and cracks from image analysis. The effect of heat treatment (HT) and hot isostatic pressing (HIP) on the microstructure of the SLMed IN-738 LC powder alloy has been examined and the consequential tensile response characterised. Findings By using optimised process parameters (low heat input, medium scanning speed and small hatching distance) which provides medium energy density, samples of IN-738 LC with a macroscopic porosity <0.5 per cent and free of micro-cracks can be manufactured by SLM. The results indicate that HIP of SLMed material did not lead to a noticeable effect on mechanical properties compared to HT of SLMed material suggesting that the level of both porosity and crack density might be already below the detection limit for the mere heat-treated material. Originality/value SLM processing parameters (power, scan speed, hatching distance) for IN-738 LC were successfully optimised after only 14 experiments using Doehlert design. Two independent methods, Archimedes method and image analysis, were used in this study to assess relative density of SLM-produced samples with sets of processing parameters showing coherency in prediction with predicted response by Doehlert design.