GNGTS 2024 - Atti del 42° Convegno Nazionale

Session 2.2 GNGTS 2024 brick masonry. In Tab. 4, the building classes identified for Montenegro are associated with the VMs of Serbia (Blagojević et al. 2023), Slovenia (Babič et al. 2021) and Italy (Rosti et al. 2021). Slovenia and Italy VMs were chosen because they are well-recognized in the seismic community and have undergone validation over the years. On the other hand, the Serbia VM is relatively recent, but is the result of a comprehensive effort involving also structural engineers from the Balkans with extensive experience in the local built environment. Thus, to exploit the common features of these three VMs, we created a heuristic VM called EXPLORA. Specifically, the fragility curves of EXPLORA are represented as cumulative distribution functions (CDFs) of lognormal distributions. These distributions show the probability of exceeding a certain DS at various IMs. The EXPLORA VM is designed in a way that its fragility curves are a linear combination of the VMs from (Blagojević et al. 2023), (Babič et al. 2021) and (Rosti et al. 2021), in terms of mean and standard deviation, as follows: Θ = w S θ S + w Sr θ Sr + w It θ It + w Sl θ Sl , (1) Β = w S β S + w Sr β Sr + w It β It + w Sl β Sl , (2) where Θ and Β are the heuristic mean and standard deviation of the EXPLORA VM, while θ and β are the mean and standard deviation of the employed VMs: SERA ( S ), Serbia ( Sr ), Italy ( It ) and Slovenia ( Sl ). The weights w are calculated such that: w S + w Sr + w It + w Sl = 1 . (3) In Eqs. (1) and (2), the means θ and standard deviations β are derived by weighting the various VMs’ fragility curves with predetermined weights λ . These weights are indicated in Tab. 4 and are applied similarly to Eqs. (1) and (2) for each building class and VM. This step is taken to consider the potential influence of flexible and rigid diaphragms, as well as the variability in building heights ( L or MH ), which are unfortunately missing in the damage data. Next, to leverage the damage information, the weights w in Eqs. (1) and (2) are fine-tuned by minimizing the error between the final weighted solution and a linear regression from the damage data pertaining DS2. Currently, this error minimization is done by means of a trial-and-error process, but a more advanced procedure is planned for future stages of this research. The weights w computed for the EXPLORA VM, which characterizes the seismic vulnerability of Montenegro, are detailed in Tab. 5. Building class w S w Sr w It w Sl URM-St 0.30 0.15 0.10 0.45 URM-Br 0.20 0.05 0.70 0.05 CFM 0.35 0.03 0.55 0.07 Tab. 5 – Weights w of each building class for the EXPLORA VM defined in Eqs. (1) and (2). The ultimate fragility curves for each building class of Tab. 4, derived from the EXPLORA approach outlined in Eqs. (1) and (2), are depicted in Fig. 2 and offer a visual comparison with the aforementioned damage data.