PSI - Issue 77

Francisco Afonso et al. / Procedia Structural Integrity 77 (2026) 575–583 F. Afonso et al. / Structural Integrity Procedia 00 (2026) 000–000

577

3

FEA Finite element analysis SFM Structure from motion MP Megapixels ROI Region of interest

2. Experimental setup

A stereo camera system and a terrestrial laser scanner (TLS) were used to monitor one wall of a transformer tank. The stereo setup consists of two cameras angled relative to each other and focused on the wall, enabling three dimensional measurements using DIC. The target typically consists of a painted speckle pattern, where small relative motions of the speckles caused by surface deformation allow precise displacement measurements. For this setup, two Allied Vision Alvium U-508c cameras (2464x2056 pixel resolution) were used with two 25 mm LINOS MeVis-C lenses. One LED light source (Wallimex Niova 150-F Pro) was used to increase contrast between the tank wall and the targets. Since the wall paint could not be altered, painting a speckle pattern was not possible. Thus, two alternatives were tested: speckle patterns printed on adhesive paper and applied to the wall, and cylindrical magnets mounted on the metallic wall in a layout resembling a speckle pattern. The main concern with these approaches is potential relative motion between the targets and the wall, which would introduce measurement errors. However, given the nature of the test, only small deformations are expected, reducing displacement and loads on the targets. To mitigate relative motion between targets and the wall, the adhesive paper, which is not elastic, was cut into smaller pieces. Due to uncertainty in available space to implement the DIC setup, di ff erent speckle sizes were evaluated to assess the measurement quality of these approaches. The region of interest (ROI) was defined as the central area of the wall, where the largest deformation is expected. The tank wall features a vertical reinforcement bar along its central area, spanning its entire height, and a panel on its central right area; both features locally increase sti ff ness. The ROI, wall features and targets are shown in Figure 1.

(a)Di ff erent target types: magnets with 8 and 12 mm (highlighted in red), adhe sives with 2 mm black speckle patterns (highlighted in yellow) and with 5 mm black speckle patterns (highlighted in blue) printed on a white background.

(b) Transformer wall representation with the ROI high lighted in red.

Fig. 1: Target types and transformer representation with region of interest highlighted.

As shown in Figure 1a, the following targets were implemented: cylindrical magnets with 8 mm and 12 mm diameters (red); adhesives with 2 mm and 5 mm speckle patterns (yellow and blue, respectively). In addition, three non-adhesive papers with speckle patterns were a ffi xed to the wall by magnets on each of the paper’s four corners. These speckle patterns were also tested but produced no readable results and are therefore not highlighted.