PSI - Issue 46

David Liović et al. / Procedia Structural Integrity 46 (2023) 42 – 48 D. Liovi ć et al. / Structural Integrity Procedia 00 (2019) 000–000

45

4

Based on 216 measurements of the average surface roughness on tensile test specimens, and by applying the Kolmogorov-Smirnov ( p > 0.15) and Shapiro-Wilik ( p = 0.014) normality test, it was found that null hypothesis can be rejected since p < α for Shapiro-Wilik test. Since the data variability of two observed groups is different (Table 2.), a t-test for independent samples (Welch's test) was applied, which confirmed that there is a statistically significant difference in the mean values of the average surface roughness between vertical surfaces of cubic and tensile test specimens ( p < 0.001). In addition, using the Levene’s test, it was found that the difference in the standard deviation of the observed groups is also statistically significant ( p = 0.001). Despite statistically significant differences in the mean values of average surface roughness between cubic and tensile test specimens, from practical and technological point of view those values are rather similar (Fig. 2. a). The reason for the rather large data scatter when measuring average surface roughness is the aperiodic nature of the vertical surfaces of the test specimens (Fig. 2. b).

Table 2. Used SLM process parameters.

Average surface roughness ( R a )

Valid N Mean Median

Min μm

Max μm

Lower Quartile Upper Quartile Std. Dev.

-

μm

μm

μm

μm

μm

Cubic specimens Tensile test specimens

180 216

6,675 6,906

6,652 6,893

5,303 8,555 6,035 8,247

6,179 6,532

7,074 7,265

0,609 0,473

Fig. 2. (a) Box plot of average surface roughness data for cubic and tensile test specimens; (b) Typical side surface of specimen.

Similar average surface roughness values are stated in the work of Pal et al. (2019) and depending on the applied energy density they can range from 4.9 to 6.78 μm. However Mierzejewska et al. (2019) achieved an average surface roughness in the range of 8.32 to 20.62 μm for heat treated test specimens, using the same technology, but different process parameters and powder granulation. Thus, the use of different powder granulations and SLM process parameters can result in different surface roughness of the SLM-ed components. 3.2. Top surface - average surface roughness When measuring the average surface roughness on the upper surfaces of the cubic specimens, it was found that for all combinations of process parameters, the average surface roughness in any case reaches its minimum values in the direction parallel to the scanning direction. When observing the average surface roughness results measured in both directions on the upper surfaces of nine cubic specimens, it can be noticed that the surface roughness reaches its highest values at 1500 mm/s scanning speed for all laser power used (Fig. 3). Besides, standard deviation of the average surface roughness data reaches its minimum value at the scanning speed of 1000 mm/s for all laser power used as part of this study. In any case, the average surface roughness of the upper surfaces of the cubic specimens is lower than the average surface roughness of the side surfaces, as expected. Namely, there are many sintered particles on the side surfaces, resulting in an increase in average surface roughness values.