PSI - Issue 78

Mariano Di Domenico et al. / Procedia Structural Integrity 78 (2026) 1237–1244

1240

3. Methodology The goal of this study is twofold: to estimate (i) the volume of debris generated by the out-of-plane collapse of masonry infill walls in RC frames and (ii) the expected ejection distance of debris on the ground. Both aspects are essential for assessing the probability of road interruption in front of buildings after an earthquake. To reach these objectives, bidirectional nonlinear time-history analyses were performed using the Multiple Stripe Analysis (MSA) method. Ten intensity levels ("stripes") were considered, corresponding to return periods from 10 to 100,000 years. For each stripe, 20 pairs of ground motion records were selected based on the pseudo-spectral acceleration Sa at T = 1 s, derived from the PSHA for L’Aquila on soft soil (Eurocode 8, type C , CEN 2004). A single conditioning period (T = 1 s) was adopted for all buildings to enable direct comparison of results. Details about the record selection can be found in Iervolino et al. (2018). Nonlinear analyses included Rayleigh damping based on mass and initial stiffness, with a 5% damping ratio assigned to the first mode (divided by two) and the fifth mode (second in Y direction). In stripes 9 and 10, Di Domenico et al. (2023) observed that more than 11 of the 20 records caused dynamic instability in the buildings, resulting in collapse. Therefore, these stripes were excluded from the debris analysis. Similarly, any record causing collapse in stripes 6 to 8 was not considered. The analysis was limited to no-collapse conditions. Infill walls were modelled with a pair of no-tension struts. A lumped mass node (66% of the infill mass, based on Ricci et al., 2022) was added at mid-span to represent the out-of-plane behaviour. When the displacement of this node exceeded 0.80 times the wall thickness (Angel et al., 1994), the infill leaf was considered collapsed and removed from the model. The remaining analysis continued with the updated structural configuration. The number of removed leaves was counted after each record and converted into debris volume. To account for the fact that some material remains attached to the frame, only 66% of the total volume of each collapsed leaf was counted as debris. This assumption follows the sinusoidal shape of the first out-of-plane vibration mode, where central zones are ejected and edge zones remain. Only walls with out-of-plane displacements directed outward from the building envelope were considered for calculation of debris volume. The debris volume also accounted for infill openings. During analysis, the velocity of the central node was tracked. The ejection velocity was estimated by interpolating the velocity time history over the last second before collapse, to avoid numerical spikes. Then, the ejection distance was computed using the standard parabolic motion equation, as shown in Fig. 2. Each ejection distance was grouped into 10 cm bins to build a frequency distribution, approximated with a lognormal distribution. The mean of this distribution was taken as the most probable ejection distance δ for that record. Results were analysed separately for infill walls on X-direction façades (falling in Y) and Y-direction façades (falling in X), to assess the influence of openings. Volumes and distances always refer to a single façade per direction.

v OOP

h

δ leaf

Fig. 2. Simplified illustrative sketch of the ejection distance determination for a single leaf.