Italy faces a high seismic risk, which makes the assessment of structural vulnerability an urgent priority. This is particularly critical for reinforced concrete (RC) buildings, which represent a large share of the urban stock. Many of these structures remain fragile, not only due to age but also because of geometric irregularities and construction practices that often fell short of modern standards, despite advances in codes and regulations. This doctoral thesis presents a two-stage workflow for assessing the seismic vulnerability of RC buildings at the urban scale. The approach combines mechanics-based analysis with data-driven prediction, generating results that can be directly transferred into geographic information systems (GIS) for vulnerability mapping. In the first stage, geometric and qualitative attributes—such as plan dimensions, number of floors, vertical regularity, age, and maintenance—are gathered from public sources (Google Maps, Quantum geographic information system (QGIS)). Each building is modeled as a virtual twin (VT) and analyzed through finite element limit analysis (FELA), following Melan’s lower- bound theorem, to obtain the collapse coefficient (CC) as an indicator of mechanical capacity. This measure is then combined with qualitative modifiers to derive a vulnerability score V , which is compared with established macro-seismic models (e.g., the European Macro-seismic Scale 1998 [EMS-98], AEDES) expressed in terms of seismic intensity I and ductility index μ∆. The consistency of this comparison confirms the reliability and practical value of the mechanics-based method. The second stage addresses the computational cost, supervised machine learning (ML) models were trained on 106 fully analyzed buildings and then applied to the full RC stock (347 buildings), achieving accuracy comparable to mechanics-based results, while requiring much less computational time. Overall, the proposed method offers a practical and scalable framework for assessing seismic vulnerability and planning urban resilience. By combining detailed mechanical insight with fast predictive tools, it offers civil protection authorities, engineers, and local administrations a reliable basis for decision-making.
L’Italia presenta un’elevata pericolosità sismica e, di conseguenza, la valutazione della vulnerabilità strutturale dovrebbe costituire una priorità, considerando la predominanza di edifici in calcestruzzo armato (CA), che rappresentano una quota significativa del patrimonio edilizio nelle aree urbane. A causa della vetustà, delle irregolarità geometriche e della scarsa qualità costruttiva, numerose strutture in CA risultano ancora vulnerabili nonostante i miglioramenti introdotti nella classificazione sismica e nella normativa edilizia nel corso del tempo. La presente tesi di dottorato propone un approccio ibrido articolato in due fasi per la valutazione della vulnerabilità sismica e la microzonazione degli edifici in CA. La metodologia sviluppata integra il calcolo meccanico con la previsione basata su dati, allo scopo di garantire affidabilità e scalabilità nella stima della vulnerabilità a livello urbano. Nella prima fase, i dati geometrici e qualitativi specifici di ciascun edificio — come le dimensioni in pianta, il numero di piani, la regolarità in elevazione, l’età e lo stato di manutenzione — vengono estratti tramite strumenti accessibili come Google Maps e QGIS. Ogni struttura è successivamente modellata in ANSYS e analizzata in MATLAB secondo il teorema del limite inferiore di Melan, ottenendo un Coefficiente di Collasso (CC) rappresentativo della sua capacità meccanica. Tale coefficiente viene modificato mediante fattori qualitativi per ottenere un Indice di Vulnerabilità (V), la cui validazione avviene attraverso il confronto con modelli macrosismici consolidati, utilizzando come parametri di ingresso l’intensità sismica e l’indice di duttilità. L’accordo ottenuto conferma l’affidabilità dell’approccio basato su calcoli meccanici. Nella seconda fase, i modelli di apprendimento supervisionato sono addestrati su 106 edifici analizzati integralmente e applicati all’intero stock in CA (347 edifici), con accuratezza paragonabile ai risultati meccanici e tempi di calcolo ridotti. Il metodo proposto fornisce uno strumento innovativo, efficiente e adattabile per la valutazione del rischio sismico e la pianificazione della resilienza urbana. Colmando il divario tra simulazioni dettagliate e applicazioni su larga scala, esso offre un supporto concreto per le autorità di protezione civile, gli ingegneri e i funzionari comunali.
Proposal for micro-zoning seismic risk map protocol for safety evaluation of structures [Proposta di protocollo per la mappa di rischio sismico di microzonazione per la valutazione della sicurezza delle strutture]
IMANI, HABIB
2026
Abstract
Italy faces a high seismic risk, which makes the assessment of structural vulnerability an urgent priority. This is particularly critical for reinforced concrete (RC) buildings, which represent a large share of the urban stock. Many of these structures remain fragile, not only due to age but also because of geometric irregularities and construction practices that often fell short of modern standards, despite advances in codes and regulations. This doctoral thesis presents a two-stage workflow for assessing the seismic vulnerability of RC buildings at the urban scale. The approach combines mechanics-based analysis with data-driven prediction, generating results that can be directly transferred into geographic information systems (GIS) for vulnerability mapping. In the first stage, geometric and qualitative attributes—such as plan dimensions, number of floors, vertical regularity, age, and maintenance—are gathered from public sources (Google Maps, Quantum geographic information system (QGIS)). Each building is modeled as a virtual twin (VT) and analyzed through finite element limit analysis (FELA), following Melan’s lower- bound theorem, to obtain the collapse coefficient (CC) as an indicator of mechanical capacity. This measure is then combined with qualitative modifiers to derive a vulnerability score V , which is compared with established macro-seismic models (e.g., the European Macro-seismic Scale 1998 [EMS-98], AEDES) expressed in terms of seismic intensity I and ductility index μ∆. The consistency of this comparison confirms the reliability and practical value of the mechanics-based method. The second stage addresses the computational cost, supervised machine learning (ML) models were trained on 106 fully analyzed buildings and then applied to the full RC stock (347 buildings), achieving accuracy comparable to mechanics-based results, while requiring much less computational time. Overall, the proposed method offers a practical and scalable framework for assessing seismic vulnerability and planning urban resilience. By combining detailed mechanical insight with fast predictive tools, it offers civil protection authorities, engineers, and local administrations a reliable basis for decision-making.| File | Dimensione | Formato | |
|---|---|---|---|
|
Imani_PhD_Thesis_FINAL_for_Reviewers.pdf
accesso aperto
Licenza:
Tutti i diritti riservati
Dimensione
15.2 MB
Formato
Adobe PDF
|
15.2 MB | Adobe PDF | Visualizza/Apri |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14242/360633
URN:NBN:IT:UNICT-360633