Existing structures are affected by material degradation and aging, and part of these structures are considered strategic due to their role in emergency management following seismic events. Their potential failure may therefore compromise public safety and infrastructure reliability, emphasizing the need for comprehensive safety assessment strategies to enhance structure safety. Within this context, seismic risk assessment represents a key component of the safety evaluation process of existing structures. Hence, this thesis focuses on the dynamic response of existing reinforced concrete structures under seismic actions and develops methodologies and tools to support their safety assessment. The research first addresses vibration-based structural health monitoring, proposing an optimization procedure for the deployment of accelerometers aimed at improving the reliability of modal identification. Subsequently, the study extends to the seismic risk assessment of an existing structure, integrating detailed numerical modeling with nonlinear analyses and performance verification to identify structural vulnerabilities and define retrofit strategies that enhance seismic performance and extend service life. In Chapter 1, the research context, motivations, and objectives of the study are introduced. In Chapter 2, a frequency-based optimization of sensor network (OSN) procedure is developed and validated. Such a procedure relies on the effective independence (EFI) method for optimization of sensor locations. To minimize the number of sensors, an additional optimization step based on a modal frequency metric is adopted. This step includes a numerical simulation of an ambient vibration test performed by exciting the finite element (FE) model of the structure with white Gaussian noise. The simulated acceleration responses are processed using the frequency domain decomposition (FDD) technique. The resulting procedure is applied to an existing precast reinforced concrete bridge using freely available ambient vibration data, which serve both for operational modal analysis and experimental validation of the optimized sensor network. Results show that, for the analyzed structure, the optimized sensor network, comprising four accelerometers, identifies the same natural frequencies obtained with a larger benchmark network of 30 accelerometers. Furthermore, among alternative four-sensor layouts with different sensor locations, the optimized network yields the lowest frequency error, confirming the effectiveness of the proposed OSN with minimal instrumentation. In Chapter 3, the performance of three different optimal sensor placement (OSP) methods, i.e., the modal kinetic energy (MKE), the EFI and the information entropy (IE) methods, is evaluated within the OSN procedure using the same case study bridge presented in Chapter 2. MKE, EFI and IE are compared by applying each method separately within the OSN framework. Moreover, the FDD technique is replaced with the stochastic subspace identification with unweighted principal components (SSI-UPC) to process the outcomes of the numerical simulation of ambient vibration test. SSI-UPC enables the identification of higher-frequency modes, allowing to increase the target mode number for the optimization. The comparison indicates that the results of the frequency-based OSN procedure slightly depends on the adopted OSP method. In Chapter 4, a modified OSN framework is proposed and assessed. The objective is the accurate reconstruction of mode shapes, measured by the modal assurance criterion (MAC), which replaces modal frequencies as the metric for minimizing the number of sensors. As in Chapter 3, optimization performance is assessed by applying the OSN procedure separately with each OSP methods (MKE, EFI, IE) and comparing the outcomes in terms of resulting optimized sensor layouts. A retrofitted, four-story reinforced concrete building exhibiting spatial irregularity is used as case study structure. Comparative results demonstrate that both EFI and IE strategies achieve high accuracy in mode shape reconstruction using a sparse sensor network comprising six sensors, while the MKE approach requires a larger number of sensors to reach a similar result (approximately 50 sensors). All analyses are based on numerical simulations and offer a useful basis for the design of actual dynamic tests and experimental validation. In Chapter 5, alternative performance metrics that can be included in the OSN framework for determining the minimum number of sensors (e.g., the singular value decomposition ratio, the Fisher information-based metrics, the information entropy-based evaluation criteria, the relative dispersion index) are first presented, together with a multi-objective optimization approach. Then, a procedure to optimize the sensor network by simultaneously accounting for the frequency-based and the mode shape-based performance metrics presented in the previous chapters is carried out on both case study structures (i.e., the existing precast reinforced concrete bridge and the irregular reinforced concrete building). Applications, performed by executing the OSN procedure separately with the EFI, MKE, and IE methods, show that the EFI and the IE outperform the MKE method. EFI and IE achieved accurate frequency and mode shape identification with few sensors and proved robust to sparse layouts. MKE does not provide an optimized sensor layout that satisfies both the optimization objectives. Furthermore, for the two other OSP methods, the mode shape-based metric remains stable above the accuracy threshold as the number of sensors varies. Therefore, for the examined cases, the multi-objective formulation reduces to a single-objective problem governed by the frequency-based metric. The results demonstrated consistent behavior of EFI and IE strategies and the OSN metrics across the two considered structurally different systems. In Chapter 6, the seismic risk assessment of a Niagara-type overpass built in Sicily (southern Italy) is presented. This structure is selected as representative of a series of structurally similar overpasses in the region. Geometry and material properties of substructure elements are derived from the original design drawings and an experimental campaign on 18 typologically similar overpasses, including the analyzed one. These data inform the FE model of the studied structure, which includes concentrated plastic hinges for structural nonlinearity and soil-structure interaction. Risk assessment outcomes of the as-built overpass reveal seismic capacity deficiencies. Hence, a retrofit strategy is assessed comprising: transverse seismic restrainers at abutments, shock transmitter devices at the free half-joint and strengthening of abutment columns and transverse beams using fiber reinforced polymer strips. The seismic risk assessment of the retrofitted overpass highlights that interventions improve the structural seismic performance, with a reduction of about two orders of magnitude from the as-built condition in terms of annual failure rate, thus reducing the seismic risk. In Chapter 7, the main findings and contributions of the research are summarized, outlining outlooks for future developments.

Le strutture esistenti sono soggette a degrado dei materiali e processi di invecchiamento, e parte di esse è considerata strategica per il ruolo che svolge nella gestione delle emergenze a seguito di eventi sismici. Il loro potenziale collasso può, pertanto, compromettere la sicurezza pubblica e l’affidabilità delle infrastrutture, sottolineando la necessità di strategie di valutazione della sicurezza che siano complete e mirate a migliorare la sicurezza strutturale. In questo contesto, la valutazione del rischio sismico rappresenta un elemento fondamentale del processo di verifica della sicurezza delle strutture esistenti. La presente tesi si concentra quindi sulla risposta dinamica delle strutture esistenti in calcestruzzo armato soggette ad azioni sismiche, sviluppando metodologie e strumenti a supporto della valutazione della sicurezza. La ricerca affronta innanzitutto il tema del monitoraggio strutturale dinamico basato sulle vibrazioni, proponendo una procedura di ottimizzazione per il posizionamento degli accelerometri, volta a migliorare l’affidabilità dell’identificazione modale. Successivamente, lo studio si estende alla valutazione del rischio sismico di una struttura esistente, integrando una modellazione numerica dettagliata con analisi non lineari e verifiche prestazionali, al fine di individuare le vulnerabilità strutturali e definire strategie di rinforzo capaci di migliorare la risposta sismica e prolungare la vita utile della struttura. Nel Capitolo 1 vengono introdotti il contesto della ricerca, le motivazioni e gli obiettivi dello studio. Nel Capitolo 2 viene sviluppata e validata una procedura di optimization of sensor network (OSN) basata sulle frequenze modali. Tale procedura si fonda sul metodo effective independence (EFI) per l’ottimizzazione delle posizioni dei sensori. Per ridurre il numero di sensori necessari, viene introdotto un ulteriore passaggio di ottimizzazione basato su una metrica di errore in frequenza modale. Questo passaggio include la simulazione numerica di una prova di vibrazione ambientale, ottenuta eccitando il modello agli elementi finiti (EF) della struttura con rumore bianco gaussiano. Le accelerazioni simulate vengono poi elaborate mediante la tecnica frequency domain decomposition (FDD). La procedura risultante è applicata a un ponte esistente in calcestruzzo armato prefabbricato, utilizzando dati di vibrazione ambientale liberamente disponibili, impiegati sia per l’analisi modale operativa che per la validazione sperimentale della rete di sensori ottimizzata. I risultati mostrano che, per la struttura analizzata, la rete ottimizzata composta da quattro accelerometri consente di identificare le stesse frequenze naturali ottenute con una rete di riferimento di 30 sensori. Inoltre, tra diverse configurazioni alternative composte da quattro sensori, quella ottimizzata fornisce l’errore in frequenza minimo, confermando l’efficacia della procedura proposta con un numero ridotto di sensori. Nel Capitolo 3 vengono valutate le prestazioni di tre diversi metodi di optimal sensor placement (OSP), ovvero il metodo modal kinetic energy (MKE), il metodo EFI e il metodo information entropy (IE), applicati nell’ambito della procedura OSN al medesimo caso studio del Capitolo 2. I tre metodi sono applicati separatamente all’interno dello stesso framework di ottimizzazione. Inoltre, la tecnica FDD viene sostituita con la tecnica stochastic subspace identification with unweighted principal components (SSI-UPC) per l’elaborazione dei risultati della simulazione numerica di prova di vibrazione ambientale. L’uso dell’SSI-UPC consente di identificare modi a frequenza più elevata, ampliando il numero di modi considerati per l’ottimizzazione. Il confronto mostra che i risultati della procedura OSN basata sulle frequenze dipendono solo marginalmente dal metodo OSP adottato. Nel Capitolo 4 viene proposto un framework OSN modificato, che ha come obiettivo la ricostruzione accurata delle forme modali, misurata attraverso il modal assurance criterion (MAC), che sostituisce le frequenze modali come metrica per la riduzione del numero di sensori. Come nel Capitolo 3, le prestazioni della procedura OSN sono valutate applicando separatamente tale procedura con ciascuno dei tre metodi OSP (MKE, EFI, IE) e confrontando i layout di sensori ottimizzati ottenuti. La struttura di riferimento per l’applicazione è un edificio in calcestruzzo armato a quattro piani, oggetto di intervento di rinforzo e caratterizzato da irregolarità spaziali. I risultati del confronto mostrano che le strategie EFI e IE raggiungono un’elevata accuratezza nella ricostruzione delle forme modali utilizzando una rete composta da solamente sei sensori, mentre l’approccio MKE richiede un numero sensibilmente maggiore di sensori (circa 50) per ottenere risultati equivalenti. Tutte le analisi sono basate su simulazioni numeriche e forniscono una base utile per la progettazione di prove dinamiche reali e per la validazione sperimentale. Nel Capitolo 5 sono presentate diverse metriche prestazionali alternative che possono essere integrate nel framework OSN per determinare il numero minimo di sensori (ad esempio, il singular value decomposition ratio, metriche basate sulla matrice di informazione di Fisher, criteri basati sull’entropia informativa, il relative dispersion index), insieme a un approccio di ottimizzazione multi-obiettivo. Viene quindi implementata una procedura di ottimizzazione della rete di sensori che considera simultaneamente le metriche basate sulle frequenze e sulle forme modali presentate nei capitoli precedenti, ed applicata a entrambe le strutture di riferimento (ossia, il ponte esistente in calcestruzzo armato prefabbricato e l’edificio irregolare in calcestruzzo armato). Le applicazioni, eseguite separatamente con i metodi EFI, MKE e IE, mostrano che EFI e IE hanno prestazioni migliori rispetto al metodo MKE, consentendo un’identificazione accurata delle frequenze e delle forme modali con un numero ridotto di sensori e mantenendosi consistenti anche con un numero ridotto di strumenti. Il metodo MKE, invece, non fornisce configurazioni che soddisfino entrambi gli obiettivi di ottimizzazione. Inoltre, per i metodi EFI e IE, la metrica basata sulle forme modali rimane stabilmente al di sopra della soglia di accuratezza al variare del numero di sensori. Pertanto, nei casi esaminati, la formulazione multi-obiettivo si riduce a un problema mono- obiettivo governato dalla metrica basata sulle frequenze. I risultati dimostrano un comportamento coerente delle strategie EFI e IE e delle metriche OSN nei due sistemi strutturalmente differenti considerati. Nel Capitolo 6 viene presentata la valutazione del rischio sismico di un cavalcavia di tipo Niagara situato in Sicilia (Italia meridionale). La struttura è scelta come rappresentativa di una serie di cavalcavia tipologicamente simili presenti nella regione. La geometria e le proprietà meccaniche degli elementi della sottostruttura sono ottenuti dalla documentazione originale di progetto e da una campagna sperimentale condotta su 18 cavalcavia tipologicamente analoghi, inclusa la struttura in esame. Questi dati sono usati per la modellazione agli EF della struttura, che include cerniere plastiche concentrate per la modellazione delle non linearità strutturali e l’interazione terreno-struttura. I risultati della valutazione del rischio del cavalcavia nello stato di fatto evidenziano carenze significative in termini di capacità sismica. Pertanto, è stata definita una strategia di adeguamento sismico che comprende: dispositivi di ritegno trasversale in corrispondenza delle spalle, dispositivi di trasmissione degli urti in corrispondenza del giunto libero, e rinforzo delle colonne e delle travi trasversali delle spalle mediante fasciature in materiali compositi fibrorinforzati. La valutazione del rischio sismico del cavalcavia adeguato sismicamente mostra come gli interventi migliorino significativamente le prestazioni sismiche della struttura, con una riduzione di circa due ordini di grandezza del tasso annuale di fallimento dell’opera rispetto alla sua condizione allo stato di fatto, riducendone così il rischio sismico. Nel Capitolo 7 sono sintetizzati i principali risultati e contributi della ricerca, delineando anche delle prospettive per sviluppi futuri.

Seismic Behavior of Existing Reinforced Concrete Structures: Optimizing Sensor Placement for Operational Modal Analysis, Numerical Modeling, and Risk Assessment / Giunta, Daniele. - (2026 Apr 27).

Seismic Behavior of Existing Reinforced Concrete Structures: Optimizing Sensor Placement for Operational Modal Analysis, Numerical Modeling, and Risk Assessment

Giunta, Daniele
2026-04-27

Abstract

Existing structures are affected by material degradation and aging, and part of these structures are considered strategic due to their role in emergency management following seismic events. Their potential failure may therefore compromise public safety and infrastructure reliability, emphasizing the need for comprehensive safety assessment strategies to enhance structure safety. Within this context, seismic risk assessment represents a key component of the safety evaluation process of existing structures. Hence, this thesis focuses on the dynamic response of existing reinforced concrete structures under seismic actions and develops methodologies and tools to support their safety assessment. The research first addresses vibration-based structural health monitoring, proposing an optimization procedure for the deployment of accelerometers aimed at improving the reliability of modal identification. Subsequently, the study extends to the seismic risk assessment of an existing structure, integrating detailed numerical modeling with nonlinear analyses and performance verification to identify structural vulnerabilities and define retrofit strategies that enhance seismic performance and extend service life. In Chapter 1, the research context, motivations, and objectives of the study are introduced. In Chapter 2, a frequency-based optimization of sensor network (OSN) procedure is developed and validated. Such a procedure relies on the effective independence (EFI) method for optimization of sensor locations. To minimize the number of sensors, an additional optimization step based on a modal frequency metric is adopted. This step includes a numerical simulation of an ambient vibration test performed by exciting the finite element (FE) model of the structure with white Gaussian noise. The simulated acceleration responses are processed using the frequency domain decomposition (FDD) technique. The resulting procedure is applied to an existing precast reinforced concrete bridge using freely available ambient vibration data, which serve both for operational modal analysis and experimental validation of the optimized sensor network. Results show that, for the analyzed structure, the optimized sensor network, comprising four accelerometers, identifies the same natural frequencies obtained with a larger benchmark network of 30 accelerometers. Furthermore, among alternative four-sensor layouts with different sensor locations, the optimized network yields the lowest frequency error, confirming the effectiveness of the proposed OSN with minimal instrumentation. In Chapter 3, the performance of three different optimal sensor placement (OSP) methods, i.e., the modal kinetic energy (MKE), the EFI and the information entropy (IE) methods, is evaluated within the OSN procedure using the same case study bridge presented in Chapter 2. MKE, EFI and IE are compared by applying each method separately within the OSN framework. Moreover, the FDD technique is replaced with the stochastic subspace identification with unweighted principal components (SSI-UPC) to process the outcomes of the numerical simulation of ambient vibration test. SSI-UPC enables the identification of higher-frequency modes, allowing to increase the target mode number for the optimization. The comparison indicates that the results of the frequency-based OSN procedure slightly depends on the adopted OSP method. In Chapter 4, a modified OSN framework is proposed and assessed. The objective is the accurate reconstruction of mode shapes, measured by the modal assurance criterion (MAC), which replaces modal frequencies as the metric for minimizing the number of sensors. As in Chapter 3, optimization performance is assessed by applying the OSN procedure separately with each OSP methods (MKE, EFI, IE) and comparing the outcomes in terms of resulting optimized sensor layouts. A retrofitted, four-story reinforced concrete building exhibiting spatial irregularity is used as case study structure. Comparative results demonstrate that both EFI and IE strategies achieve high accuracy in mode shape reconstruction using a sparse sensor network comprising six sensors, while the MKE approach requires a larger number of sensors to reach a similar result (approximately 50 sensors). All analyses are based on numerical simulations and offer a useful basis for the design of actual dynamic tests and experimental validation. In Chapter 5, alternative performance metrics that can be included in the OSN framework for determining the minimum number of sensors (e.g., the singular value decomposition ratio, the Fisher information-based metrics, the information entropy-based evaluation criteria, the relative dispersion index) are first presented, together with a multi-objective optimization approach. Then, a procedure to optimize the sensor network by simultaneously accounting for the frequency-based and the mode shape-based performance metrics presented in the previous chapters is carried out on both case study structures (i.e., the existing precast reinforced concrete bridge and the irregular reinforced concrete building). Applications, performed by executing the OSN procedure separately with the EFI, MKE, and IE methods, show that the EFI and the IE outperform the MKE method. EFI and IE achieved accurate frequency and mode shape identification with few sensors and proved robust to sparse layouts. MKE does not provide an optimized sensor layout that satisfies both the optimization objectives. Furthermore, for the two other OSP methods, the mode shape-based metric remains stable above the accuracy threshold as the number of sensors varies. Therefore, for the examined cases, the multi-objective formulation reduces to a single-objective problem governed by the frequency-based metric. The results demonstrated consistent behavior of EFI and IE strategies and the OSN metrics across the two considered structurally different systems. In Chapter 6, the seismic risk assessment of a Niagara-type overpass built in Sicily (southern Italy) is presented. This structure is selected as representative of a series of structurally similar overpasses in the region. Geometry and material properties of substructure elements are derived from the original design drawings and an experimental campaign on 18 typologically similar overpasses, including the analyzed one. These data inform the FE model of the studied structure, which includes concentrated plastic hinges for structural nonlinearity and soil-structure interaction. Risk assessment outcomes of the as-built overpass reveal seismic capacity deficiencies. Hence, a retrofit strategy is assessed comprising: transverse seismic restrainers at abutments, shock transmitter devices at the free half-joint and strengthening of abutment columns and transverse beams using fiber reinforced polymer strips. The seismic risk assessment of the retrofitted overpass highlights that interventions improve the structural seismic performance, with a reduction of about two orders of magnitude from the as-built condition in terms of annual failure rate, thus reducing the seismic risk. In Chapter 7, the main findings and contributions of the research are summarized, outlining outlooks for future developments.
27-apr-2026
Settore ICAR/09 - TECNICA DELLE COSTRUZIONI
Settore CEAR-07/A - Tecnica delle costruzioni
CHIOCCARELLI, EUGENIO
PIETRAFESA, Matilde Mariarosa Consolata
Doctoral Thesis
File in questo prodotto:
File Dimensione Formato  
PhD Thesis_Giunta Daniele_XXXVIII ciclo.pdf

embargo fino al 30/04/2027

Tipologia: Tesi di dottorato
Licenza: Tutti i diritti riservati (All rights reserved)
Dimensione 7.52 MB
Formato Adobe PDF
7.52 MB Adobe PDF   Visualizza/Apri   Richiedi una copia

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/20.500.12318/165146
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact