PSI - Issue 78

Andrea Miano et al. / Procedia Structural Integrity 78 (2026) 1903–1910

1908

the second approach suffers from a general low density of MPS per tile. Indeed, about 35% of tiles are empty, while only 30% of the tiles are populated with at least 4 MPs. Nevertheless, while the first approach provides more robust statistics due to the higher MP density per tile, it may fail to capture and highlight displacement patterns that exhibit spatial variability along the deck, particularly in the transverse direction. This limitation can hinder the identification of differential motion phenomena that may be critical for structural assessment. An example of such a case is shown for a viaduct in Figure 4b. The observed viaduct exhibits different values of average velocity along the LOS between the right side (mountain side) and the left side (valley side) of the deck. These differential displacements – worth further investigation, especially through higher-resolution monitoring data – cannot be detected when using a coarser discretization scheme. Further investigation is needed to define threshold values for the standard deviation capable of capturing such deformation phenomena. This would allow for the classification of structures affected by these behaviors as requiring more detailed assessments, when performing a mass-appraisal territorial scale analysis.

30x30 m – Undivided cross-section grid

20x20 m Subdivided cross-section grid

a) c) Fig. 4 Comparison of tested procedure for bridge footprint discretization: a) number of MPs per tile; b) average velocity [mm/year]; c) a,b Results of the analysis: Histograms of LOS velocity values for the 2018–2022 and 2019–2023 time windows; c,d Estimated vertical displacements; e,f Acceleration values derived from differences in LOS velocities between the two time windows for Asc044, Asc117, and Dsc095. To assess the potential of EGMS data for large-scale infrastructure monitoring, a sensitivity analysis is performed focusing on a road network comprising 19 bridges, that vary in length and structural typology. For each bridge, a linear arrangement of 30×30 m cells is defined, centred along the bridge axis and the average LOS velocity of PS within each cell is computed. This results in a total of 49 cells across all bridges. In this study, only LOS data are considered. Two sensitivity analyses are conducted. The first investigates the influence of different EGMS datasets. In the EGMS archive, multiple satellite orbits often cover the same area; for the case study, three orbits are considered: Ascending 044, Ascending 117, and Descending 095. The analysis explores how the distribution of LOS velocity varies across these datasets. The second analysis evaluates the temporal consistency of the velocity estimates by comparing two distinct time windows: 2018–2022 and 2019–2023. EGMS is updated annually, allowing us to track changes in velocity over time. Velocity values are calculated as the slope of the best-fitting line through the displacement time series, while acceleration is estimated as the difference between the velocities computed in the two time-windows. Table 1 summarises the number of populated cells for each orbit and period and differences in datasets. Table 1 Number of populated cells per EGMS dataset and time window. Time-frame Asc044 Asc117 Dsc095 2018-2022 32 40 33 2019-2023 34 36 31 2018-2022/2019-2023 30 35 29 b)