PSI - Issue 84

Lorenzo Brezzi et al. / Procedia Structural Integrity 84 (2026) 1159–1166

1164

over a wider area and reaches higher maximum thicknesses, locally approaching 4 m. The standard deviation maps indicate that uncertainty is particularly pronounced in proximity to the structure, where relatively small variations in rheological parameters significantly influence the final deposit thickness. A more detailed comparison can be carried out by referring to two representative locations, labelled A and B in Figure 4, corresponding respectively to the final accumulation zone upstream of the viaduct (A) and to the final deposit at the infrastructure location (B). In the single-landslide scenario, the mean final deposit thickness at point A is 2.2 ± 1.3 m, increasing to 3.7 ± 1.8 m in the double-landslide scenario. This marked increase indicates that the presence of multiple source areas promotes a larger accumulation of material upstream of the viaduct in the final deposit configuration. At point B, corresponding to the viaduct location, the mean final thickness increases from 2.7 ± 0.95 m in the single-landslide scenario to 3.2 ± 1.2 m in the double-landslide scenario. Compared to point A, the increase in mean values is more limited, while the associated variability becomes more pronounced, suggesting a more heterogeneous redistribution of material in proximity to the structure. Overall, these results show that multiple-source scenarios primarily affect the magnitude and organisation of the final deposit upstream of the infrastructure, while leading to more variable accumulation patterns at the viaduct location.

Fig. 5. Average values of the resultant velocities, with focus on the viaduct area, considering all the 200 simulations and the two scenarios.

Beyond final deposits, the temporal evolution of landslide propagation provides additional insight into the potential severity of interaction scenarios. Fig. 5 illustrates the spatial distribution of resultant velocities at the instant of maximum interaction with the viaduct, corresponding to the time at which the product of flow thickness and velocity reaches its maximum at points defining the contact with the infrastructure. This instant represents the most critical dynamic conditions in terms of potential impact. In both scenarios, peak velocity values in the vicinity of the viaduct are of the same order of magnitude, although their spatial location differs. In the single-landslide scenario, the maximum velocity is attained at point A (Fig. 5), where the mean resultant velocity reaches 8.4 ± 3.5 m/s. In the double-landslide scenario, the position of the velocity peak shifts approximately 15 m eastward, with the maximum located at point B and characterised by a mean value of 8.6 ± 3.0 m/s. Despite this spatial shift, the similarity of peak velocity values indicates that the presence of a second landslide source does not significantly increase the maximum flow velocity at the infrastructure level. Flow thicknesses associated with the peak velocity locations are also comparable in the two scenarios. At point A, the mean thickness at the instant of maximum interaction is 1.9 ± 0.2 m, while at point B it increases slightly to 2.0 ± 0.5 m. As a result, the increase in the product of flow thickness and velocity between the two scenarios is limited in terms of local peak values. A more substantial difference emerges when considering the spatial extent of the interaction zone. In the single-landslide scenario, the contact between the flowing mass and the infrastructure is confined to a section of approximately 25 m along the viaduct. In contrast, the double-landslide scenario results in a significantly