In recent years, the life-cycle structural assessment of vulnerable aging components within transportation networks is gaining prominence in the context of optimal management by public authorities and private agencies of critical lifelines. To this purpose, the scientific community is currently focused on developing refined analytical models for structural reliability and seismic vulnerability and risk assessment of aging RC bridges and viaducts. Most advanced predictive models of the structural response should be able to integrate uncertainties associated with material mechanical properties, external loads and environmental exposure scenarios in which the investigated structure operates. Structural analyses often require computational cost that may unfeasible, especially when the mechanical response involves non-linearities in the formulation of structural members. Simulation methods are one of the most used tools for structural reliability analysis. A fervent research activity sows the seeds of new advanced techniques aiming to reduce the computational effort and to accurately estimate failure probabilities with a limited number of generated samples to be analyzed. This thesis discusses the numerical features of Monte Carlo methods and variance-reduction techniques for life-cycle reliability of deteriorating structural systems. Advantages and drawbacks of the investigated algorithms in the life-cycle context are examined with the aid of an archetype structural reliability problem based on limit analysis of a statically indeterminate structure. Among the proposed numerical techniques, further insight is given to Importance Sampling (IS) in the context of structural reliability based on static non-linear pushover analysis of a multi-span highway bridge with three reinforced concrete piers subjected to chloride-induced corrosion. Modeling sensitivity analyses are conducted with the aim of quantifying the influence of different hypotheses on the structural response. The potentialities of the Stationary Proposal Importance Sampling (SP-IS) methodology are investigated in terms of trade-off between computational effort and estimate accuracy of time-variant reliability indices for different limit states.
Negli ultimi anni, l’analisi a ciclo di vita di elementi vulnerabili all’interno di reti di trasporto critiche sta acquisendo importanza sempre maggiore nella gestione ottimale delle infrastrutture gestite da enti pubblici e privati. In questo contesto, la comunità scientifica si sta focalizzando sullo sviluppo di modelli analitici sempre più raffinati per la valutazione dell’affidabilità strutturale e della vulnerabilità e del rischio sismico di ponti e viadotti in calcestruzzo armato soggetti a degrado. Modelli predittivi avanzati sono chiamati ad integrare nella valutazione della risposta strutturale anche le incertezze associate alle proprietà meccaniche dei materiali, ai carichi agenti sulla struttura e agli scenari di esposizione ambientale. Le analisi numeriche su tali modelli richiedono oneri computazionali spesso insostenibili, in particolare nel caso in cui gli elementi strutturali siano caratterizzati da comportamento non-lineare. Recenti attività di ricerca aspirano a concepire tecniche di simulazione avanzate per la stima ottimale della probabilità di fallimento limitando il numero di analisi strutturali su un campione circoscritto di modelli di calcolo. In questa tesi, si presentano le peculiarità numeriche dei metodi di Monte Carlo e delle tecniche di riduzione della varianza per l’analisi di affidabilità a ciclo di vita di sistemi strutturali soggetti a degrado. Vantaggi e svantaggi degli algoritmi analizzati nel contesto dell’analisi a ciclo di vita vengono discussi sulla base di un tipico problema di analisi di affidabilità strutturale relativo all’analisi limite di un sistema iperstatico. Tra le tecniche numeriche proposte, viene dedicato un approfondimento alla tecnica dell’Importance Sampling (IS) nel contesto dell’affidabilità strutturale mediante analisi statica non-lineare di un ponte autostradale con tre pile in calcestruzzo armato soggette a corrosione da cloruri. A valle di diverse analisi di sensitività per valutare l’influenza delle ipotesi di modellazione delle pile del viadotto, vengono messe in luce le potenzialità della metodologia di Importance Sampling con distribuzione di campionamento stazionaria (Stationary Proposal Importance Sampling, SP-IS). In particolare, le applicazioni sviluppate permettono di approfondire il compromesso tra onere computazionale e accuratezza di stima di indici di affidabilità in riferimento a diversi stati limite.
Advanced simulation methods for life-cycle seismic assessment of RC bridges
Padovani, Francesco
2020/2021
Abstract
In recent years, the life-cycle structural assessment of vulnerable aging components within transportation networks is gaining prominence in the context of optimal management by public authorities and private agencies of critical lifelines. To this purpose, the scientific community is currently focused on developing refined analytical models for structural reliability and seismic vulnerability and risk assessment of aging RC bridges and viaducts. Most advanced predictive models of the structural response should be able to integrate uncertainties associated with material mechanical properties, external loads and environmental exposure scenarios in which the investigated structure operates. Structural analyses often require computational cost that may unfeasible, especially when the mechanical response involves non-linearities in the formulation of structural members. Simulation methods are one of the most used tools for structural reliability analysis. A fervent research activity sows the seeds of new advanced techniques aiming to reduce the computational effort and to accurately estimate failure probabilities with a limited number of generated samples to be analyzed. This thesis discusses the numerical features of Monte Carlo methods and variance-reduction techniques for life-cycle reliability of deteriorating structural systems. Advantages and drawbacks of the investigated algorithms in the life-cycle context are examined with the aid of an archetype structural reliability problem based on limit analysis of a statically indeterminate structure. Among the proposed numerical techniques, further insight is given to Importance Sampling (IS) in the context of structural reliability based on static non-linear pushover analysis of a multi-span highway bridge with three reinforced concrete piers subjected to chloride-induced corrosion. Modeling sensitivity analyses are conducted with the aim of quantifying the influence of different hypotheses on the structural response. The potentialities of the Stationary Proposal Importance Sampling (SP-IS) methodology are investigated in terms of trade-off between computational effort and estimate accuracy of time-variant reliability indices for different limit states.File | Dimensione | Formato | |
---|---|---|---|
2022_04_Padovani.pdf
non accessibile
Dimensione
9.2 MB
Formato
Adobe PDF
|
9.2 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/187512