Process optimisation in laser metal deposition

Abstract

Laser metal deposition (LMD) has become a compelling technique for fabricating high-entropy alloys (HEAs). However, its adoption on a commercial scale is limited, hindered by process-related defects and difficulty in controlling numerous interacting parameters. This review examines defect formation, process parameter-property relationships, and optimisation approaches for LMD-fabricated HEAs. The findings establish that the quality of LMD part is governed by about 39 process variables, including 13 key process parameters, out of which laser power, scan speed, powder feed rate, beam diameter, layer thickness, overlap fraction, and shield gas flow rate exercise dominant effects on the melt-pool stability, microstructure and mechanical properties. Cracking, lack of fusion, gas porosity, surface-connected porosity, and gas porosity constitute the main defects, with lack of fusion reported when scan speed deviates from the optimum range by more than 15%. Evidence from reviewed studies indicates that increasing laser power from 600 to 800 W enhances hardness from 200 to 600 HV in AlCoCrFeCu HEA, and from 500 to 850 HV in AlTiCrCoNi HEA. Optimised process windows for fabricating defect-free CoCrFeMnNi are 400-600 W and 10-30 mm/s. Nevertheless, excessive layer thickness of about 300 µm results in reduced elongation from 11.45% to 2.0%. Advanced optimisation techniques enhanced the reliability of prediction: Taguchi-grey relational analysis obtained 0.95% deviation, response surface methodology produced 5.27%-10%, and machine learning approaches achieved R2 values between 0.97-1.00, including phase-prediction precision of 98.5%. In general, hybrid machine learning, physics-based optimisation, and digital twin frameworks present the most compelling way for attaining defect-free, high-performance LMD-produced HEAs.