PSI - Issue 78

Marco Civera et al. / Procedia Structural Integrity 78 (2026) 1783–1790

1785

analysis and transportation network modelling to quantify the operational capacity of road networks post-disaster. Thus, for direct comparability with the results previously reported in (Miano et al. 2024), the same resilience index – based on hospital-to-hospital connectivity – has been used. However, as the focus is here shifted from the first post event hours to the first days and weeks, the same considerations can be applied, with appropriate modifications, between any two critical nodes, thus, not only limited to patients' mobility from one healthcare facility to another. The post-disaster RN efficiency index is quantified using a probabilistic approach and works as follows: first, the efficiency ( ) of each road segment is calculated as , =1 − , / (3) Where , is the mean travel time on the -th path and is the maximum permissible time to connect critical nodes. Very simply, the mean travel time is computed, for each path, as , = , / , (4) Where , is the mean velocity expected on that i-th path and , is total length. Then, having computed these efficiency values for each segment, the overall efficiency of the undamaged RN ( , ) is derived by aggregating the performance of all accessible roads - more details are available in (Miano et al. 2024) and not reported here due to length limitations. Instead, for each damage scenario, it is possible to run simulations of seismic events, obtaining a mean and standard deviation value for the post-event efficiency as = ∑ , = 1 , = � ∑ � , − , � = 1 2 � 1 2 (5) Where , is the mean of the efficiency at the j -th simulation and the corresponding standard deviation. That way, the simulations consider the impact of disrupted segments and rerouting. Importantly, the Monte Carlo method used for these simulations is, as it is well-known, stochastic; hence, every run will be different, but the validity of the results is given by the convergence to a stable outcome after enough simulations – here and in (Miano et al. 2024), 2000. This concludes the recall of the methodology proposed in (Miano et al. 2024). The main difference between that framework and this study lies in the seismic scenarios. In (Miano et al. 2024), two cases were considered: Sc50 and Sc475. Sc50 included an earthquake with a return period = 50 years, associated with an average between DS1 and DS2. Conversely, Sc475 is considered a major, rare event ( =475 years) but associated with DS3 (Sc475). 3. Updated Methodology: Corrections for Reopening Time To compensate for the binary nature of the first version of the framework (disrupted/non-disrupted), which does not consider the different outcomes of different damage levels in a longer post-event timeframe, and to consistently bridge the intended meaning of ‘road network resilience’ with the currently most accepted definition of ‘structural resilience’ – see, e.g., – (Argyroudis et al. 2020), the corrective factors reported in Table 1 are proposed. As shown, the correction is assumed to be proportional to the difference in terms of hours to reopening, which scales almost exponentially from one DS to the next one. The rationale for the correction factors is to scale at the same rate; e.g., assuming 6 months for DS4 and 2 years for DS5, being the elapsed time four times larger, the resilience index is corrected by a factor of four as well. The variability in the time scale, especially for larger damage levels (DS3 onwards), is very high since key transportation routes may receive expedited repairs compared to less critical ones. However, even considering the fastest possible operations, local or global reconstruction will require years; e.g. the new San Giorgio viaduct, built after the partial collapse of the Polcevera viaduct, was realised at almost record speeds but took almost two years from the event (one after the demolition of the remaining parts). It is important, however, to highlight that these values are here deterministically assumed based on known engineering experience. A more methodical and statistically principled investigation of the expected recovery times will be carried out in future works, based on relevant works such as (Opabola and Galasso 2024) and (Sun et al. 2021).