PSI - Issue 64

Kun Feng et al. / Procedia Structural Integrity 64 (2024) 596–603 Kun Feng et al. / Structural Integrity Procedia 00 (2019) 000 – 000 To effectively evaluate the bridge's structural integrity and pinpoint the location of any damage, this paper employs the Squared Error, , between the ODS curves of the healthy and damaged bridge as a damage indicator. This damage indicator is defined as follows: = | − ℎ | 2 (1) where the ℎ and denote the ODS curves for the healthy and damaged bridge, respectively, derived from measurements taken by a single passing vehicle. To mitigate the variability introduced by using data from just one vehicle, this study employs a fleet of 1000 bus vehicles to calculate the average ODS curves, thereby enhancing the reliability of the results. 4. Numerical Results and Discussions In this study, a fleet with 1000 bus vehicles is utilized, crossing a 30 m approaching road, following a 15 m bridge, at a constant velocity of 5 m/s, with a very smooth profile and constant temperature. This moving speed, equivalent to 18 km/h, aligns with the characteristics of the UK bus cycle (McGrath, Blades et al. 2022), where the average speed ranges between 10 km/h and 31.6 km/h. Furthermore, the analysis is based on a single assumption regarding the number of seated passengers, resulting in the operational weight of the bus vehicles varying from 14437 kg to 19560 kg. Fig. 5 displays the average ODS curve for 1000 bus vehicles, depicted as a solid red line. The distribution of these 1000 ODS curves is emphasized and encapsulated within a light red shaded area. In the following analyses, this average or mean ODS will be leveraged for damage detection and localization using drive-by fleet monitoring. 601 6

Fig. 5. ODS for a fleet with 1000 bus vehicles.

Fig. 6 showcases the simulated average ODS curves for damage scenarios located at 1/3 of the span, or 5 m, featuring three varying levels of damage severity: 10%, 20%, and 30% crack damage, respectively. Notably, significant variances are observable in Fig. 6(a) when comparing the average ODS curves of damaged bridges with that of a healthy one, although pinpointing the exact damage location remains challenging. By employing the damage indicator, as shown in Fig. 6(b), a distinct peak can be observed in the magnitude of the squared errors. For instance, in the 30% crack damage scenario, a pronounced peak is identified at 5.425 m from the left span, as depicted in Fig. 6(b), corresponding to a relative error of 8.5%. However, the proposed damage localization algorithm is not that good for the 10% and 20% crack damage scenarios, where the estimated damage locations are 6.255 m and 6.305 m from the left span, respectively, yielding relative errors of 25.1% and 26.1%.