Surface finishing by laser re-melting applied to robotized laser metal deposition

Abstract

Low surface quality of as-deposited laser metal deposition (LMD) parts is one of the main drawbacks of this additive manufacturing technology that prevents direct use without the implementation of costly and time consuming post-processes. An in-process surface finishing operation without the need of auxiliary equipment is highly appealing to overcome such issues. For this reason, laser re-melting performed with the same laser equipment used for the deposition process is a promising technology capable of redistributing the deposited material onto a smoother surface. Combined with a robotic manipulation system used for both the LMD and laser re-melting phases, complex geometries can potentially be produced in near net-shape conditions. An important issue concerning development of this process development regards the use of standardized roughness and waviness parameters, which can fail to address the texture specific to the LMD process. Indeed, the laser re-melting process can produce new texture formation under non-optimal conditions. Hence, an analysis of the surface topography in the frequency domain along with the standard surface roughness and waviness parameters would be more appropriate. In this work, laser re-melting is applied to AISI 316L thin walled structures right after the deposition process. An experimental campaign aimed at finding the process parameters and re-melting strategy capable of improving the surface quality was carried out. Areal surface profile measurements were used for characterizing the effect of the process parameters. The surface power spectrum on the S-F surface was calculated to point out the presence of periodical components in the surface structure and a transfer function calculation was performed to define the surface quality improvement in the frequency domain compared to the LMD as-deposited surface. The acquired surface areal data was assessed at two different spatial frequency regions corresponding to the S-L and L-F surfaces. The arithmetic average of the filtered surface areal data was used to calculate Sa, SF and Sa, LF analogously to the average roughness and waviness parameters of linear measurements respectively. Results showed, in the optimized conditions, a reduction of up to 79% in Sa, SF corresponding to the roughness related frequency range. On the other hand, the Sa, LF parameter corresponding to the waviness realted frequency range was reduced up to 58% due to the low effectiveness of laser re-melting in reducing the amplitude of the lowest frequency components. The overall improvement in terms of process capability was evaluated as mean ± 3 times the sample standard deviation values (µ±3σ). The average surface roughness Sa, SF could be reduced from 10.35±0.42 µm to 1.92±0.11 µm and average surface waviness Sa, LF from 9.57±0.48 µm to 4.04±0.20 µm.