Estimating the reliability of guided wave SHM systems through modelling

Guided wave structural health monitoring (SHM) presents a promising, cost-effective, and efficient modality for evaluating pipeline integrity. To support its widespread industrial adoption, quantitative reliability estimation in realistic environments is essential. This thesis develops a framework for estimating the reliability of guided wave SHM systems through realistic finite element (FE) modelling, with a strong focus on practical applications. Accurate model-based reliability estimation requires high-fidelity simulation of real-world conditions, including instrumentation installation, initial pipe conditions, and environmental influences. A methodology is first presented for analysing key features of real guided wave signals, particularly random and coherent noise, based on laboratory data. The formation and composition of coherent noise are studied in detail, with attention to transducer performance imbalance. A numerical framework is proposed to generate realistic guided wave signals that incorporate both coherent and random noise. This is validated through simulation and experiments, effectively capturing noise characteristics at both individual and statistical levels. The framework also supports the simulation of defect signals, enabling more accurate Probability of Detection (POD) analysis. Additionally, by modelling transducer imbalance, it facilitates noise reduction in experimental signals and improves the signal-to-noise ratio (SNR). A digital twin-based reliability estimation framework is then proposed to provide timely, system-specific reliability assessments over the SHM system’s lifecycle. It constructs a digital twin based on in-situ measurements and estimates key simulation parameters affected by environmental temperature and structural conditions. Compared to traditional model-assisted reliability methods in Non-Destructive Evaluation (NDE), this approach offers a more accurate and dynamic performance assessment, supporting better-informed maintenance and inspection decisions. Lastly, the influence of geometric imperfections on guided wave signals is investigated. The effects of pipe wall thickness variations on wave behaviour are quantified through high-fidelity modelling, enhancing the integration of initial pipe conditions into digital twins and improving reliability estimation.