The safety, durability and sustainability of modern buildings are essential in the construction industry, especially as structures become increasingly complex and must meet stringent standards. This thesis investigates advanced quality control and monitoring techniques, focusing on surface quality control and structural health monitoring (SHM), to improve the durability and reliability of building structures. Addressing the limitations of traditional inspection methods, the research introduces innovative methodologies based on artificial intelligence (AI), machine learning and sensor-based systems. The first part of the research focuses on surface quality control, emphasising the importance of assessing visible defects such as cracks, pitting, honeycombing and exposed reinforcement bars. These surface imperfections, while often considered aesthetic, can act as precursors to structural problems, allowing moisture and environmental factors to infiltrate, leading to corrosion and deeper structural deterioration. Traditional manual inspections, although widely practised, are labour-intensive, subjective and prone to inefficiency due to a shortage of qualified personnel. To overcome these challenges, this thesis proposes an automated system that combines image acquisition and processing with artificial intelligence-driven algorithms for defect detection and quantification. To train the neural networks, a customised dataset of concrete surface defects was created, which ensures high accuracy in identifying anomalies. The methodology takes measurement uncertainties into account, addressing potential defect recognition errors caused by noise or different background characteristics. The developed system is portable, enabling hands-on inspections in the field, and integrates with digital platforms to facilitate real-time monitoring during the building's life cycle. The research highlights how early detection and quantification of surface issues can mitigate long-term risks. The second part of the research focuses on SHM, with a special focus on curtain walls, which are increasingly common in modern high-rise buildings. These façades, often made of glass or lightweight materials, are crucial for the structural integrity and energy efficiency of a building. However, they are vulnerable to stresses such as weather loads and accidental impacts, requiring continuous monitoring to maintain safety and performance. This thesis introduces a new SHM methodology that integrates data from multiple sensors, including strain gauges and accelerometers, to acquire real-time information on structural behaviour. By correlating dynamic parameters with physical phenomena, the research identifies the main causes of potential failures, such as bolt loosening, thermal stresses or external impacts. Machine learning models were used to analyse some of the data acquired by the sensors, enabling early detection of anomalies and providing useful information for maintenance. The study also addresses compliance with international standards, such as EN 13830, ensuring that the monitoring solutions are in line with regulatory requirements. Experiments conducted on a weathered curtain wall demonstrated the effectiveness of the proposed SHM framework in identifying and predicting failure modes such as shock and vibration-induced stress. This research makes several key contributions to the field of construction monitoring. It develops an integrated system for surface defect detection, combining high-resolution images and AI-guided analysis, and extends the state-of-the-art by quantifying uncertainties in defect detection. Furthermore, it proposes a comprehensive SHM framework for curtain walls, exploiting multi-sensor data fusion and machine learning to improve predictive maintenance.
La sicurezza, la durata e la sostenibilità degli edifici moderni sono essenziali nel settore delle costruzioni, soprattutto perché le strutture diventano sempre più complesse e devono soddisfare standard rigorosi. Questa tesi esamina tecniche avanzate di controllo e monitoraggio della qualità, concentrandosi sul controllo della qualità superficiale e sul monitoraggio dello stato strutturale (SHM), per migliorare la durata e l'affidabilità delle strutture degli edifici. Affrontando i limiti dei metodi di ispezione tradizionali, la ricerca introduce metodologie innovative basate sul l'intelligenza artificiale (IA), l'apprendimento automatico e i sistemi basati su sensori. La prima parte della ricerca è incentrata sul controllo della qualità superficiale, sottolineando l'importanza di valutare i difetti visibili quali crepe, corrosioni, alveoli e barre di rinforzo esposte. Queste imperfezioni superficiali, sebbene spesso considerate estetiche, possono fungere da precursori di problemi strutturali, permettendo al l'umidità e ai fattori ambientali di infiltrarsi, portando alla corrosione e a un più profondo deterioramento strutturale. Le ispezioni manuali tradizionali, sebbene ampiamente praticate, sono ad alta intensità di manodopera, soggettive e soggette a inefficienza a causa della carenza di personale qualificato. Per superare queste sfide, questa tesi propone un sistema automatizzato che combina l'acquisizione e l'elaborazione delle immagini con algoritmi basati sull'intelligenza artificiale per il rilevamento e la quantificazione dei difetti. Per addestrare le reti neurali, è stata creata una serie di dati personalizzati sui difetti superficiali del calcestruzzo, che garantisce un'elevata precisione nel l'identificazione delle anomalie. La metodologia tiene conto delle incertezze di misurazione, affrontando potenziali errori di riconoscimento dei difetti causati dal rumore o da diverse caratteristiche di fondo. Il sistema sviluppato è portatile, consente ispezioni pratiche sul campo e si integra con piattaforme digitali per facilitare il monitoraggio in tempo reale durante il ciclo di vita del l'edificio. La ricerca evidenzia come l'individuazione precoce e la quantificazione dei problemi di superficie possano ridurre i rischi a lungo termine. La seconda parte della ricerca si concentra sulla SHM, con particolare attenzione alle facciate continue, che sono sempre più comuni negli edifici moderni. Queste facciate, spesso realizzate in vetro o materiali leggeri, sono fondamentali per l'integrità strutturale e l'efficienza energetica di un edificio. Tuttavia, sono vulnerabili a sollecitazioni quali carichi atmosferici e impatti accidentali, che richiedono un monitoraggio continuo per mantenere la sicurezza e le prestazioni. Questa tesi introduce una nuova metodologia SHM che integra i dati provenienti da più sensori, tra cui estensimetri e accelerometri, per acquisire informazioni in tempo reale sul comportamento strutturale. Correlando i parametri dinamici con i fenomeni fisici, la ricerca identifica le principali cause di potenziali guasti, quali l'allentamento dei bulloni, le sollecitazioni termiche o gli impatti esterni. Sono stati utilizzati modelli di apprendimento automatico per analizzare alcuni dei dati acquisiti dai sensori, consentendo la rilevazione precoce delle anomalie e fornendo informazioni utili per la manutenzione. Lo studio prende in esame anche la conformità alle norme internazionali, quali EN 13830, per garantire che le soluzioni di monitoraggio siano conformi ai requisiti normativi. Gli esperimenti condotti su una parete-cortina alterata hanno dimostrato l'efficacia della struttura SHM proposta nel l'identificazione e nella previsione dei modi di guasto, quali shock e stress indotti dalle vibrazioni. Questa ricerca apporta diversi contributi chiave al settore del monitoraggio della costruzione. Sviluppa un sistema integrato per il rilevamento dei difetti superficiali, che combina immagini ad alta risoluzione e analisi guidata da IA, ed estende lo stato del l'arte quantificando le incertezze nel rilevamento dei difetti. Inoltre, propone un quadro SHM completo per le facciate continue, sfruttando la fusione di dati multi-sensore e l'apprendimento automatico per migliorare la manutenzione predittiva.
Design and application of methodologies for the quality control of buildings during the construction phase
CALCAGNI, MARIA TERESA
2025
Abstract
The safety, durability and sustainability of modern buildings are essential in the construction industry, especially as structures become increasingly complex and must meet stringent standards. This thesis investigates advanced quality control and monitoring techniques, focusing on surface quality control and structural health monitoring (SHM), to improve the durability and reliability of building structures. Addressing the limitations of traditional inspection methods, the research introduces innovative methodologies based on artificial intelligence (AI), machine learning and sensor-based systems. The first part of the research focuses on surface quality control, emphasising the importance of assessing visible defects such as cracks, pitting, honeycombing and exposed reinforcement bars. These surface imperfections, while often considered aesthetic, can act as precursors to structural problems, allowing moisture and environmental factors to infiltrate, leading to corrosion and deeper structural deterioration. Traditional manual inspections, although widely practised, are labour-intensive, subjective and prone to inefficiency due to a shortage of qualified personnel. To overcome these challenges, this thesis proposes an automated system that combines image acquisition and processing with artificial intelligence-driven algorithms for defect detection and quantification. To train the neural networks, a customised dataset of concrete surface defects was created, which ensures high accuracy in identifying anomalies. The methodology takes measurement uncertainties into account, addressing potential defect recognition errors caused by noise or different background characteristics. The developed system is portable, enabling hands-on inspections in the field, and integrates with digital platforms to facilitate real-time monitoring during the building's life cycle. The research highlights how early detection and quantification of surface issues can mitigate long-term risks. The second part of the research focuses on SHM, with a special focus on curtain walls, which are increasingly common in modern high-rise buildings. These façades, often made of glass or lightweight materials, are crucial for the structural integrity and energy efficiency of a building. However, they are vulnerable to stresses such as weather loads and accidental impacts, requiring continuous monitoring to maintain safety and performance. This thesis introduces a new SHM methodology that integrates data from multiple sensors, including strain gauges and accelerometers, to acquire real-time information on structural behaviour. By correlating dynamic parameters with physical phenomena, the research identifies the main causes of potential failures, such as bolt loosening, thermal stresses or external impacts. Machine learning models were used to analyse some of the data acquired by the sensors, enabling early detection of anomalies and providing useful information for maintenance. The study also addresses compliance with international standards, such as EN 13830, ensuring that the monitoring solutions are in line with regulatory requirements. Experiments conducted on a weathered curtain wall demonstrated the effectiveness of the proposed SHM framework in identifying and predicting failure modes such as shock and vibration-induced stress. This research makes several key contributions to the field of construction monitoring. It develops an integrated system for surface defect detection, combining high-resolution images and AI-guided analysis, and extends the state-of-the-art by quantifying uncertainties in defect detection. Furthermore, it proposes a comprehensive SHM framework for curtain walls, exploiting multi-sensor data fusion and machine learning to improve predictive maintenance.File | Dimensione | Formato | |
---|---|---|---|
tesi_calcagni.pdf
embargo fino al 31/12/2027
Dimensione
6.78 MB
Formato
Adobe PDF
|
6.78 MB | Adobe PDF |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14242/212963
URN:NBN:IT:UNIVPM-212963