Damage identification of railway bridges through temporal autoregressive modelling under train load

• Autores: Stefano Anastasia
• Directores de la Tesis: Salvador Ivorra Chorro (dir. tes.), Vincenzo Gattulli (dir. tes.)
• Lectura: En la Universitat d'Alacant / Universidad de Alicante ( España ) en 2024
• Idioma: inglés
• Número de páginas: 170
• Tribunal Calificador de la Tesis: Francisco J. Pallarés Rubio (presid.), Giada Gasparini (secret.), Filippo Ubertini (voc.)
• Programa de doctorado: Programa de Doctorado en Ingeniería de Materiales, Estructuras y Terreno: Construcción Sostenible por la Universidad de Alicante
• Menciones:
  • Internacional
• Calificación: Sobresaliente Cum Laude
• Enlaces
  • Tesis en acceso abierto en: RUA
• Resumen

  • Over the past decade, structures and infrastructures across Europe have become the focus of attention of researchers and engineers, mainly due to catastrophic events that, in some cases, have caused the collapse of these infrastructures. This growing concern has led to the need to design and develop structural health monitoring systems, known as SHM, that are reliable and capable of detecting damage both locally and globally, focusing mainly on existing bridges. Detecting damage in structures and infrastructures is a complex challenge, given the wide range of environmental and operational conditions to which these structures are exposed during their life cycle. This difficulty is particularly evident in the case of bridge structures. Therefore, the design, construction and maintenance of such bridges must strictly conform to criteria that carefully consider environmental and operational conditions, without neglecting the fundamental dynamic aspects of the structure. To address this challenge, the research work carried out during the PhD period aimed to develop a new damage identification approach that exploits continuously extracted time series of autoregressive (AR) coefficients from deformation data with moving train loads as highly sensitive damage characteristics. Through a statistical pattern recognition algorithm involving data clustering and quality control charts, the proposed approach offers a set of sensor-level damage indicators with damage detection, quantification and localisation capabilities. The effectiveness of the developed approach is evaluated through two case studies, involving a theoretically simply supported girder (preliminary study) and a railway bridge in real-world operation. In the preliminary study in particular as well as in the proposed real-world case, it exploits the implementation of autoregressive (AR) time series analysis for continuous damage detection. The continuously extracted AR coefficients are used as damage-sensitive features. To automate the AR model order definition, the methodology implements a model selection approach based on Bayesian information criterion (BIC), Akaike information criterion (AIC) and mean square error (MSE). In this exploratory investigation, the suitability and effectiveness of deformation measurements with respect to acceleration-based systems are evaluated through the simply supported Euler-Bernoulli beam under moving loads. The problem of moving loads in terms of vertical accelerations and normal deformations is solved through closed-form modal decomposition. The numerical results presented and the discussion highlight the effectiveness of the proposed approach, laying the foundation for its implementation on real-world instrumented bridges. The real case study aimed at identifying structural damage is part of a research project dedicated to the continuous monitoring of a real bridge in operation, namely the Mascarat Viaduct in Alicante (Spain). The viaduct, built at the beginning of the 20th century, is part of the railway network of the Valencian community FGV (Ferrocarriles de la Generalitat Valenciana) belonging to TRAM line 9 in the province of Alicante. The viaduct is equipped with a dense network of deformation sensors and accelerometers that monitor the structural response. The monitoring system was designed and built to receive data both when the line is in operation and when there is no vehicular passage, when the line is interrupted