PSI - Issue 78

Merani Margherita Gabriella Bruna et al. / Procedia Structural Integrity 78 (2026) 785–792

786

1. Introduction The ability to rapidly and accurately assess the impacts of an earthquake in urban areas is fundamental to effective emergency management. Decision-makers require timely and reliable information to coordinate search and rescue teams, evaluate the status of critical infrastructure, and allocate resources efficiently (Comfort, 2005). To meet this demand, sophisticated platforms have been developed to provide rapid, large-scale estimates of outcomes, from ground shaking intensity maps (e.g., ShakeMap) to projections of economic and social losses (e.g., HAZUS, OpenQuake) (Wald et al., 1999; Kircher et al., 2006; Silva et al., 2014). Studies have consistently shown that the accuracy of these scenarios improves significantly when detailed local data is used (Merani et al., 2025). Damage predictions incorporating site-specific soil conditions and detailed information about building types often differ substantially from those based on generic, national-scale data. This highlights a critical point: a damage scenario's quality is directly tied to the granularity and accuracy of its input data. Despite these advancements, a core limitation persists since current approaches mainly rely on models — such as Ground Motion Prediction Equations (GMPEs) and fragility curves — to forecast outcomes where direct observations are unavailable. However, these models are subjected to significant uncertainties, which can be broadly categorized as epistemic (stemming from our incomplete knowledge and simplified model assumptions) and aleatory. This can result in considerable discrepancies between predicted damage and reality on the ground, leading to the inefficient allocation of critical resources or, conversely, underestimating the risk in severely affected areas. This work builds upon a line of research that leverages dense sensor networks to enhance post-earthquake damage assessment. Some studies focus on intensifying ground arrays to produce more accurate, real-time urban shakemaps (e.g., Costanzo et al., 2021; Patanè et al., 2022), while others deploy sensors directly on structures to monitor their dynamic response, a practice established at a national scale by networks like the Italian Seismic Observatory of Structures (OSS) (Dolce et al., 2017) and applied in various research contexts (e.g., D’Alessandro et al., 2019) . Our work contributes to this line of research by proposing a paradigm shift to address this gap by moving from a purely predictive framework to an observation-constrained framework. This evolution is made possible by integrating real time data from structural monitoring networks. The widespread availability of low-cost, high-fidelity sensors now makes it feasible to deploy dense urban monitoring networks. We outline a methodology that uses this data to achieve two critical enhancements. First, it refines the seismic hazard assessment by using direct ground motion recordings and the identified dynamic properties of buildings to calculate more accurate, site-and-structure-specific shaking intensities. Second, and equally important, it refines the assessment of vulnerability. By identifying a building's fundamental period (T₁), we can move beyond generic fragility models based on Peak Ground Acceleration (PGA) and instead use more physically meaningful, period-specific fragility curves based on Spectral Acceleration (Sa(T₁)). The goal is to create "living" damage scenarios that evolve as real -world data becomes available, moving closer to the concept of an urban-scale digital twin for disaster response. 2. Methodological framework: from predictive to monitoring-driven scenarios The proposed methodology enhances the established two-step process for generating damage scenarios by systematically enriching each stage with real-time monitoring data, as illustrated in Figure 1.

Figure 1: Flowchart of the observation-constrained framework