PSI - Issue 34

Zoé Jardon et al. / Procedia Structural Integrity 34 (2021) 32–38 Zoé Jardon/ Structural Integrity Procedia 00 (2019) 000 – 000

36

5

increases for increasing deposition heights and for lower capillary diameters. To avoid this phenomenon, the machine acceleration (1g) should remain inferior to the heat dissipation or cooling rate around the capillary, that is strongly dependent of the capillary diameter. Capillary diameters of 2mm seem to come close to this lower limit. Note also that drilling small diameter holes can become challenging over longer drilling depths. Considering these observations, the capillary diameter is chosen as mm and the drilling operations are realized with an uncoated 71221-3 HSS-E Stock drill (DIN338) of 3mm diameter. It is advised to pre-drill a guidance hole, to properly guide the final drill. The hardness of 316L DED samples is significantly higher than for conventional 316L. Iterations has been therefore required to find proper milling (?) and drilling settings (peck drilling cycle: spindle speed = 1200rpm, feed rate = 15mm/min, 2mm/step, total drill depth = 15mm).

Fig. 4 : Pre-tests for capillary diameter choice: DED trajectories (left), Printed result (right) .

Further, 30 samples with different parameter combination have been printed and their print process and resulting geometry analyzed. Fig. 5 presents a build-plate with samples as an example, together with high-speed in-situ monitoring images acquired during the print process. The laser power, powder mass flow rate, layer thickness and overlap have been varied within the ranges = [360 − 750] W, = [ 0.4 − 0.8] mm, % = [36 − 60]% , , ̇ = [5 − 8] g/min. The vertical build-up of a fatigue sample appeared to be challenging for high build-ups. The hybrid aspect (reprint on a previously printed n -layer sample) makes it even more essential to respect acceptable geometrical tolerances. Coaxial images monitored in-situ allowed to tune the laser control by checking the quality of the melt pool (liquid state) thanks to pixel intensity values. Rounding of the sample corners is observed for higher powder mass flow rates and higher laser powers and always associated with dripping. The rounding systematically induces a convex shape of the sample top surface. When the top surface is flat, no rounding of the corners is observed and the dripping phenomenon is limited. This is considered as being an optimal geometry. Control of the laser power seems also to be a need in order to respect the target geometric features and to not have excessive rounding of the corners (that induces drips on the sample walls). The laser power is therefore adjusted layer per layer and on the contours. The starting laser power is = 360 , it is reduced with 10% per layer and end at = 600 . The contour power is always 10% lower than the infill. The adjustment is done over 6 layers and further the laser power is kept constant at the end value till the first sequence of operations is realized. For the repetition on top of the milled -layer sample, the starting laser power is lowered, since the build-up is created on top of -layers instead of the starting build plate. Overall, , , % and ̇ showed to be closely related parameters for which an optimal choice is essential to obtain acceptable geometries. From the parametric study, it is concluded that and % have the largest impact on the sample geometry. The optimal parameter combination appeared to be = 0.6mm , o % = 48% and ̇ = 3.9 g/min. The corresponding spacing between the deposited tracks equals 0.83 for the chosen overlap.

Fig. 5 : Build plate with printed samples (left), In-situ monitoring of print process (middle: full sample, right : sample with capillary).