PSI - Issue 64

Szymon Grzesiak et al. / Procedia Structural Integrity 64 (2024) 269–276 S. Grzesiak, C. de Sousa, M. Pahn / Structural Integrity Procedia 00 (2019) 000 – 000

273

5

w calc (z) = F exp 12 ∙ E · · z I ∙ ( 3 4 · l 2 −z 2 ) for z≤ 1 2 · l with E·I = ∑ E i ∙ I x,i − 1

(1)

Table 3. Results of bending tests.

Test series

Max. force F exp (kN)

Acting shear force V exp (kN)

Max. moment at mid-span M exp (kNm)

Max. deflection w calc (mm)

B1 B2 B3

7.46 6.46 6.42

3.72 3.23 3.21

0.75 0.65 0.64

6.43 5.56 5.53

3.2. Image processing A direct manual segmentation and characterization of the 3D CT dataset generated for the scanned samples was not possible, as the CT scans were too large to allow for detection and registration of the different internal structures in a reasonable amount of time. Therefore, image processing algorithms had to be used to distinguish between material properties. A CT scan (3D 16-bit images) of the concrete microstructure was used to perform this analysis. In this test the whole CT dataset had a size of 1250 × 384 × 3024 voxels. The voxel size was equal to 125.844 μm. The image analysis carried out in this work was performed with the MAVI software developed by the ITWM of the Fraunhofer Institute in Kaiserslautern (Fraunhofer ITWM (2022)). The first step on the applied image processing approach was to reduce the size of the raw data. This was achieved through a general cropping process, which erases the unnecessary voxels corresponding to air around the samples. This was followed by a 6 ° rotation to correct the position of the samples within the 3D data. A more detailed cropping followed, by which the air edges were cut even closer to the samples so that practically only the reinforced concrete samples remain. Step-by-step image analysis of this procedure is showed in Figure 3.

crop

2·F

F

2·F

crop & rotation & crop

binarization & labelling

2·F

Steel rebar

After segmentation

FRP strip Crack

Fig. 3. Step-by-step image analysis pipeline applied to the 3D images, visualized using a 2D slice parallel to the XY-plane.

The segmentation process is based on the gray values of the obtained CT imaging data, where the minimum value 0 means relates to darker voxels corresponding to materials with lower density, as for example air voids. The maximum gray value corresponds to the brighter objects that are visible in the data (value 2 9 -1 for 8-bits and value 2 16 -1 for 16 bit), associated to materials with the highest density, such as steel reinforcement (see light grey objects in Figure 3).