PSI - Issue 54

Daniel F.O. Braga et al. / Procedia Structural Integrity 54 (2024) 568–574 Daniel F.O. Braga et al. / Structural Integrity Procedia 00 (2023) 000–000

573

6

Fig. 4. UTS with RunOrder

The overall fit of the model was low, with an adjusted R 2 of 44.38 %. As mentioned previously the significant variation in strength, especially the strength drop in run order #7, which coincided with a star point of the CCD, may indicate that the window of the parameter set chosen was too wide, which led to instabilities at the most extreme valued of the DoE. The error due to lack of fit in this model is 40.39 %. The analysis of variance showed that on linear, one square and one 2-way interaction source were signif icant towards UTS. These were, Laser Power with a p -value of 0.096, Laser Power square with a p -value of 0.018 and Scan Velocity × Powder feed with a p -value of 0.054. The resulting regression equation is presented in Eq. 1. UTS=1 . 527+1 . 106 × LaserPower + 1 . 630 × ScanVelocity +123 . 3 × PowderFeed − 0 . 000249 × LaserPower 2 − 0 . 1441 × ScanVelocity × PowderFeed (1) The process parameters set that results in the maximum strength is ≈ 2212.12 W laser power, with 1500 mm/min scan velocity and 5 g/min powder feed (although scan velocity and powder feed values are extrapolations). For this parameter set, the predicted UTS is 1680 MPa, with the an output window with 95% confidence between 1263 MPa and 2098 MPa. A resulting surface plot demonstrating the second order effect of laser power is present in Fig. 5.

Fig. 5. Surface plot of UTS vs laser power and powder feed