PSI - Issue 61

Jose Beltra Mira et al. / Procedia Structural Integrity 61 (2024) 156–163 Author name / Structural Integrity Procedia 00 (2024) 000–000

161

6

driven by their ligament area (the area fractured during the test) and not by the core material properties. Therefore, the results gathered from the statistical analysis were found to be confirmed by this additional analysis. As with the statistical analysis, the results suggest that there is no significant impact on performance relative to the sample size, at least within the tested size range.

Table 6. Sample break area and intervals.

Broken Area (mm 2 )

Size

Interval Percent (%)

W = 30mm, B = 6mm, a = 15mm 90 W = 40mm, B = 8mm, a = 20mm 160 W = 50mm, B = 10mm, a = 25mm 250

—-

43.75 56.25

(a) Versus Yield Stress

(b) Versus Elastic Modulus

(c) Versus Elongation

1000 1200 1400 1600

1000 1200 1400 1600

1000 1200 1400 1600

ABS W = 30 mm PC W = 30 mm PLA W = 30 mm ABS W = 40 mm PC W = 40 mm PLA W = 40 mm ABS W = 50 mm PC W = 50 mm PLA W = 50 mm

0 200 400 600 800

0 200 400 600 800

0 200 400 600 800

Fracture Force (N)

Fracture Force (N)

Fracture Force (N)

22

32

42

52

62

1.5

2

2.5

3

3.5

1.5

2

2.5

3

3.5

Yield Stress (MPa)

Elastic Modulus (GPa)

Elongation (%)

Fig. 6. Observed mean fracture force for each material and sample size relative to (a) material yield stress, (b) elastic modulus, and (c) elongation.

4.3. Sensitivity Analysis

As a final check and evaluation of the specimen size impact, two more specimens of each material were made using theW = 30mm, B = 6 mm dimension, but with B = 3 mm. This is half the original thickness and the thinnest CT sample that the authors was able to test successfully without sample buckling. Ideally, a B < 3 mm should be used to capture plane stress more e ff ectively but the samples have been observed to buckle at thinner dimensions. This sample dimensions clearly do not fit the requirements for ASTM D5045 (and therefore are not necessarily valid for testing) but are useful as a sensitivity analysis. Calculating the K Q for these thinner samples and comparing them to the original (Figure 7a) continues to show that there is no major impact from sample thickness as expected from plane stress-plane strain theory shown in Figure 7b.

Table 7. Statistical results for the sensitivity analysis.

Factor

Test

H-value p-value Conclusion

n

Material

Kruskal-Wallis 15 15.34

< 0.001 Statistically significant impact 0.515 Not statistically significant impact

Sample Thickness B Kruskal-Wallis 15 0.42

Table 7 gives the results of a statistical analysis for this sensitivity analysis, where the same procedure was followed for the analysis given previously. As with the previous analysis, the sample thickness was not found to have any statistically significant impact on the results, suggesting that observed di ff erences in the data is due to noise from AM-driven printing defects as observed in the main study. The only material with some major di ff erence is PC, but it is in the opposite direction as expected if there was a sample size e ff ect. This results could well be the result of a more severe printing defect in one of the samples, given that FFF-processed PC is known to occasionally demonstrate unpredictable behavior when fracturing [11, 20–25].