PSI - Issue 64

F.-B. Cartiaux et al. / Procedia Structural Integrity 64 (2024) 285–292 Cartiaux / Structural Integrity Procedia 00 (2019) 000 – 000

288

4

Thus, the records are organized in a series of “events”, where each event represents the passage of one (heavy) truck on the deck, with a few seconds before and after. Each time the threshold is triggered on any strain sensor of one deck, all the sensors of the deck record for the same duration and are synchronized by the EDAS. Each “event” is considered as an individual entity (in a statistical meaning), described by the 24 timeseries of the sensors of the deck, from which we can extract meaningful features describing the effect of the traffic and the characteristics of the response of the bridge at the time of the event. The challenge then is to translate these features into synthetic health indices by gathering all event features of each deck for large periods of time (days, months etc.). Within the two years of the experiment, as of February 2024, the system remained fully operational with no interruption, except a period of six months from October 2022 to March 2023 due to a power outage of both two EDAS. The 18 months of records yet available represent more than 58,000 events on the upstream deck (built in 1988) and 47,000 events on the downstream deck (built in 2002). The volume of raw data is 22 Gb in compressed datafile format (it would be around ten times larger without compression). All data are sent in real-time by the 4G router of the EDAS to the OSMOS cloud, where they are immediately available for the analysis. The acoustic emission measurements performed by the University of Le Mans are not included in the permanent monitoring system and use a specific interrogator, different from the EDAS. Acoustic emission data was gathered on one single day on July 5, 2022, as a short data acquisition campaign. 4. Combined strain and vibration analysis 4.1. Examples of raw data for one event As an illustration of the above-mentioned raw time series of strain and acceleration measurements, automatically recorded for each truck passage on the decks, graphical outputs are displayed (Fig. 3 and 4) for the most important event in terms of value range of each sensor group on the upstream deck: optical strands or accelerometers.

Fig. 3. Strain measurements on the upstream deck. The pattern on the “SUP” sensors show s numerous axles of a special convoy.

Fig. 4. Acceleration measurements on the upstream deck. The strongest event for vibrations is not the same as for strain range.