PSI - Issue 28

Giovanni Pio Pucillo et al. / Procedia Structural Integrity 28 (2020) 1998–2012 GP Pucillo et al. – Part I / Structural Integrity Procedia 00 (2019) 000 – 000

2008

11

3.1.4. Data comparison between strain gauges measurements around different holes and extrapolation of the strain trend as a function of the distance from hole edge The comparison between residual strains measured by strain gauges installed around different holes but placed at the same positions is shown in Fig. 11. Residual strains are reported as a function of the distance from the hole edge and for different angular locations, with the aim to verify the actual repeatability of the cold expansion process, and at the same time to extrapolate the hoop residual strain distribution along the directions of interest. Fig. 11-a shows the comparison between hoop residual strains along the direction at 0°. For all the holes, the residual strain decreases with the distance from hole edge and, because of the negligible differences between the cold expansion percentages (Section 3), strains are, as expected, almost the same. The percentage difference, calculated with respect to the smallest strain value of each set of experimental points referred to the same position, was also calculated, and is shown in the right vertical axis of the diagrams: a maximum value of 7.5% was found. In Fig. 11-b the comparison between radial residual strains at 180° is shown: the percentage difference is 23%. As mentioned in Section 2.1 the strain trend along the directions at +45° and +135° was expected to be the same, therefore the hoop residual strain distribution was determined considering together the strain gauges installed at +45° and +135° around holes #2 and #6 (Fig. 11-c). Strains acquired at 9 mm from hole edge are the same (percentage variation is equal to 1%); at 11 mm, instead, a higher value of the percentage difference was calculated (60%). However, excluding the latter case, the trend is consistent, and the low percentage difference between the measurements is excellent, which gives a good level of confidence with the acquired data set, above all in view of the FE validation of Part II (Pucillo et al. 2020).

hole #6

hole #5

hole #2

percentage variation

0 1000 2000 3000 4000 5000 6000 7000 8000

0 1 2 3 4 5 6 7 8

7000

70

-8000 -6000 -4000 -2000 0

10 15 20 25 30 35 40

y

y

Θ = 180

6000

60

Θ = 0

x

x

5000

50

4000

40

Strain [ μ m/m]

Strain [ μ m/m]

Strain [ μ m/m]

Θ = +45

Θ = +135

3000

30

-16000 -14000 -12000 -10000

y

2000

20

x

Percentage variation

Percentage variation

Percentage variation

1000

10

0 5

a

b

c

0

0

0

20 60 Distance from hole edge [mm] 40

0

2 8 Distance from hole edge [mm] 4 6

0

5

10

15

Distance from hole edge [mm]

23

Fig. 11. Hoop (a, c) and radial (b) residual strains measured by strain gauges after 2.0% of CE. Percent difference between ER measurements made on different holes and at the same location, and dependency of the residual strains on the distance of the hole edge.

3.2. 2D-DIC data Analysis of images was performed using Vic-2D software (Correlated Solutions Inc., Columbia, SC, USA). The first image was acquired before the mandrel was inserted into the hole and was set as the reference image. Correlation was run between the reference image and the last one of the test data. One of the benefits of using a full-field optical technique, like image correlation, is the easy of strain separation, something that in other techniques, such as photoelasticity, can be quite difficult to perform. The contour plots of normal ( ε xx and ε yy ) and shear ( ε xy ) residual strains near the various holes are reported in Fig. 12. For the hole #2 and hole #6, the images were processed by setting a subset size of 17 pixels 2 , a step size of 5 pixels, and a strain filter window of 15. For the hole #5, instead, images were processed by setting a subset size of 43 pixels 2 , a step size of 7 pixels, and a strain filter window of 15.