PSI - Issue 62

Fabio Micozzi et al. / Procedia Structural Integrity 62 (2024) 848–855 Author name / Structural Integrity Procedia 00 (2024) 000 – 000

851

4

third acquisition the acquisition frequency was lowered to 30 FPS and  = 100 was kept. The sizes of the ROI were respectively 183×178, 174×170, and 170×167 pixels, the margin between ROI and template were 25 pixels in the first one and 20 pixels in the other two. This resulted in maximum measurable displacements of ±10.46 mm, ±8.31 mm, and ±8.43 mm, as detailed in Micozzi et al. (2023).

Accelerometers

Video camera

Video camera

Accelerometers

a

b

Fig. 2. (a) lateral view of the video camera and accelerometers; (b) front view of the video camera and accelerometers.

3. Results 3.1. Displacement measures

Comparisons in terms of displacements are summarized in Table 1 where the columns report from left to right the considered 10-minute measure windows, the label assigned to the measure (having indicated between parenthesis the adopted sensor), the number of heavy vehicles in the considered window (defined as those inducing displacements larger than 1 mm), the maximum downward displacement at midspan, the differences between vision based and contact measurements of the maximum displacement (value and relative percentage).

Table 1. Comparisons of displacement measures.

Heavy vehicles

Peak displacement (mm)

Difference peak displacement (mm)

Difference peak displacement (%)

Window

Measure

DT1 (displacement transducer)

2.533

Reference

Reference

1

4

FVB1 (video camera)

2.584

0.129

2.00

DT2 (displacement transducer)

3.433

Reference

Reference

2

3

FVB2 (video camera)

3.468

0.035

1.01

DT3 (displacement transducer)

3.778

Reference

Reference

3

24

FVB3 (video camera)

3.844

0.066

1.75

It is observed that the two measuring technologies, despite being very different, provide very close results in terms of the peak displacements along each acquisition window. The maximum difference between the peak displacement from the video camera and those of the contact sensor is 2% (0.129 mm) when 120 FPS and  = 50 are used, reduce to the minimum difference of about 1% (0.035 mm) when 120 FPS and  = 100. Even when 30 FPS is used, the accuracy is very good, with a difference of 1.75% (0.066 mm). The entire time-histories of the acquired displacements are shown in Figures 3, 4, and 5, along with the close-ups at the peak deflection.