Infrastructural networks are at the heart of countries’ development worldwide. In this framework, bridges play a crucial role. Ageing represents a growing threat to these structures, and the current policies in matters of monitoring and maintenance do not address it effectively enough. In particular, today’s best practices are not compliant with resource availability. This work begins with an analysis of the problem of bridge monitoring, focusing on system needs and policymaking flaws. The basis for this is a broad literature review and personal experience gained through projects I worked on during my PhD. The findings emerging from this preliminary study motivated the search for new techniques to bridge the gap and support decision-makers in prioritising interventions. Drive-by monitoring is deemed a promising tool for quick and cheap data collection, helpful in performing the first-stage screening of our infrastructural networks. The idea behind this technique is to sense bridge vibrations by putting sensors inside crossing vehicles. This thesis approaches the topic by providing a literature review aimed at presenting what has been done so far and which are the challenges to bringing drive-by monitoring to a real-world application. The result of this analysis has been the choice of a set of operational parameters to study their impact on the performance of the technique. The work is based on a fully data-driven methodology, founded on two experimental case studies designed to address research questions related to the variables of interest. Indirect frequency identification is performed on the first bridge to investigate optimal sensor placement. Also, a wavelet-based algorithm is tested. The second case study analyses the effect of vehicle dynamics, sensor performances, travelling speed, and spatial localisation accuracy by estimating mode shapes. To understand the impact of big data, both case studies explore the relationship between number of trips and accuracy of the results. The positive outcomes of the research on the vehicle dynamics effect, tested through a comparison between cars and e-scooters, combined with the promising contribution of big data, guided the last case study. The latter proposes a scenario of crowdsourced data from shared micromobility to scan urban bridges. The analysis is based on a real data set of usage, made of nine million e-scooter trips in twelve US cities. The results suggest the feasibility of this application.
Le reti infrastrutturali sono al centro dello sviluppo dei paesi in tutto il mondo. In questo quadro, i ponti giocano un ruolo cruciale. L'invecchiamento rappresenta una minaccia crescente per queste strutture e le attuali politiche in materia di monitoraggio e manutenzione non lo affrontano in modo sufficientemente efficace. In particolare, le best practice odierne non sono conformi alla disponibilità delle risorse. Questo lavoro inizia con un'analisi del problema del monitoraggio dei ponti, concentrandosi sulle esigenze del sistema e sui difetti delle politiche. La base per questo è un'ampia revisione della letteratura e l'esperienza personale acquisita attraverso i progetti su cui ho lavorato durante il mio dottorato di ricerca. I risultati emersi da questo studio preliminare hanno motivato la ricerca di nuove tecniche per colmare il divario e supportare i decisori nella definizione delle priorità degli interventi. Il monitoraggio drive-by è considerato uno strumento promettente per raccolta dati rapida ed economica, utile per eseguire lo screening di prima fase delle nostre reti infrastrutturali. L'idea alla base di questa tecnica è percepire il ponte vibrazioni inserendo sensori all'interno dei veicoli che attraversano. Questa tesi affronta l'argomento fornendo una revisione della letteratura volta a presentare ciò che è stato fatto finora e quali sono le sfide per portare il monitoraggio drive-by in un'applicazione del mondo reale. Il risultato di questa analisi è stata la scelta di una serie di parametri operativi per studiarne l'impatto sulle prestazioni della tecnica. Il lavoro si basa su una metodologia completamente data-driven, fondata su due casi di studio sperimentali progettati per affrontare questioni di ricerca relative alle variabili di interesse. L'identificazione della frequenza indiretta viene eseguita sul primo ponte per studiare il posizionamento ottimale del sensore. Inoltre, viene testato un algoritmo basato su wavelet. Il secondo caso di studio analizza l'effetto della dinamica del veicolo, delle prestazioni dei sensori, della velocità di viaggio e dell'accuratezza della localizzazione spaziale stimando le forme modali. Per comprendere l'impatto dei big data, entrambi i casi di studio esplorano la relazione tra il numero di viaggi e l'accuratezza dei risultati. Gli esiti positivi della ricerca sull'effetto dinamica del veicolo, testati attraverso un confronto tra auto ed e-scooter, uniti al promettente contributo dei big data, hanno guidato l'ultimo case study. Quest'ultimo propone uno scenario di dati crowdsourcing dalla micromobilità condivisa per scansionare i ponti urbani. L'analisi si basa su un set di dati reali di utilizzo, composto da nove milioni di viaggi in scooter elettrico in dodici città degli Stati Uniti. I risultati suggeriscono la fattibilità di questa applicazione.
A study on drive-by monitoring, a technique to support decision-making on bridge management
BENEDETTI, LORENZO
2022/2023
Abstract
Infrastructural networks are at the heart of countries’ development worldwide. In this framework, bridges play a crucial role. Ageing represents a growing threat to these structures, and the current policies in matters of monitoring and maintenance do not address it effectively enough. In particular, today’s best practices are not compliant with resource availability. This work begins with an analysis of the problem of bridge monitoring, focusing on system needs and policymaking flaws. The basis for this is a broad literature review and personal experience gained through projects I worked on during my PhD. The findings emerging from this preliminary study motivated the search for new techniques to bridge the gap and support decision-makers in prioritising interventions. Drive-by monitoring is deemed a promising tool for quick and cheap data collection, helpful in performing the first-stage screening of our infrastructural networks. The idea behind this technique is to sense bridge vibrations by putting sensors inside crossing vehicles. This thesis approaches the topic by providing a literature review aimed at presenting what has been done so far and which are the challenges to bringing drive-by monitoring to a real-world application. The result of this analysis has been the choice of a set of operational parameters to study their impact on the performance of the technique. The work is based on a fully data-driven methodology, founded on two experimental case studies designed to address research questions related to the variables of interest. Indirect frequency identification is performed on the first bridge to investigate optimal sensor placement. Also, a wavelet-based algorithm is tested. The second case study analyses the effect of vehicle dynamics, sensor performances, travelling speed, and spatial localisation accuracy by estimating mode shapes. To understand the impact of big data, both case studies explore the relationship between number of trips and accuracy of the results. The positive outcomes of the research on the vehicle dynamics effect, tested through a comparison between cars and e-scooters, combined with the promising contribution of big data, guided the last case study. The latter proposes a scenario of crowdsourced data from shared micromobility to scan urban bridges. The analysis is based on a real data set of usage, made of nine million e-scooter trips in twelve US cities. The results suggest the feasibility of this application.| File | Dimensione | Formato | |
|---|---|---|---|
|
A study on drive-by monitoring, a technique to support decision-making on bridge management.pdf
solo utenti autorizzati dal 24/12/2023
Dimensione
20.93 MB
Formato
Adobe PDF
|
20.93 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/10589/195590