Wu, Wen
(2024)
Physics-based guided wave structural health monitoring, and its integration in asset management modelling.
PhD thesis, University of Nottingham.
Abstract
Structural health monitoring (SHM) plays a vital role in ensuring the integrity of engineering infrastructures. Its primary objective is the implementation of a systematic approach to detect and identify potential damage or structural issues within these assets. By incorporating the identified information about asset condition from a monitoring system, engineers can utilise asset management models to manage maintenance activities in order to lower operating and maintenance costs and increase the life of these assets. The main aim of this thesis is to develop efficient damage identification methods using ultrasonic guided waves and integrate them with asset management modelling approaches.
Firstly, a physics-based Bayesian framework using a guided waves interaction model for damage identification of infinite plate is proposed. A semi-analytical approach based on the lowest order plate theories is adopted to obtain the scattering features for damage geometries. The proposed framework is able to identify the geometry of a partly through-thickness circular hole in plate-like structures and reconstruct scattering field. Compared with a traditional finite element model and similar methods, this approach results in an efficient inversion procedure for damage size identification.
A Bayesian framework combining a Kriging surrogate model for damage identification in joints is also hereby developed. A wave and finite element method (WFE) is adopted to numerically predict the scattering coefficients corresponding to different size defects in joints. Furthermore, the Kriging surrogate model is integrated with the WFE to generate a database containing measured scattering properties and to replace the WFE as a forward model within probabilistic inference, which enhances the computational efficiency of the proposed scheme. Both damage identification frameworks are validated using numerical, as well as experimental case studies.
Next, a general framework is developed to evaluate the SHM reliability that takes into account sensor failures. The framework involves modelling the degradation process of sensor networks using Petri nets (PNs) and calculating the expected information gain of the sensor network based on the information theory. The proposed framework is able to estimate performance, including monitoring accuracy and uncertainty, of the monitoring system throughout its operational time, which is rigorously validated through both numerical simulations and experimental tests using an ultrasonic monitoring system. Importantly, this approach can be potentially extended to assess the reliability of various monitoring systems, particularly those vulnerable to sensor failures.
A leading edge (LE) erosion growth prediction model utilizing Bayesian updating and drone-inspected data is also developed. A probabilistic physical model is presented to estimate incubation period and mass loss rate of LE erosion. Due to the high sensitivity of the physical model, a Bayesian parameter updating scheme is presented to infer the material property and its uncertainty by using drone-inspected data. The updated physical model is then used to predict LE erosion growth.
Finally, a wind turbine blade asset management PN model is proposed, including degradation, inspection, condition monitoring, and maintenance processes. The model can forecast the future blade condition for a given asset management strategy, taking into account detailed industry guidelines. Besides, it investigates the impact of the the monitoring system reliability on the asset management modelling results. The simulation results illustrate the degree of uncertainty introduced into the monitoring results by the reliability of the monitoring system and, consequently, the extent to which this factor influences the maintenance strategies.
Overall, the thesis proposes four aspects of novelty. Firstly, it for the first time presents a dedicated physics-based Bayesian framework for extracting damage characteristics from ultrasound measurements in plates and joints (doi: 10.1002/stc.2728; 10.1016/j.ultras.2022.106773). Secondly, it proposes a comprehensive methodology for assessing the reliability of SHM systems throughout their operational lifespan, focusing on localization and sizing aspects. Thirdly, it explores a novel method that combines physics models and inspection data for predicting defect growth. Lastly, it introduces an asset management PN model tailored for wind turbine blades, integrating the SHM process and its reliability.
Actions (Archive Staff Only)
 |
Edit View |
Tools
Tools
![[thumbnail of Corrected version after PhD viva]](https://eprints.nottingham.ac.uk/style/images/fileicons/application_pdf.png "Corrected version after PhD viva")