Development of a Minimal Information Data - modelling approach for the monitoring of short and medium span bridges

Abstract

The research presented in this thesis aims to develop a novel Minimal Information Data‐ modelling (MID) approach which can monitor the structural behaviour of short and medium span bridges. A significant number of bridges in many countries are approaching or have exceeded their design life. For this reason, structural health monitoring (SHM) has been widely studied in the last two decades. However, the industrial adoption of innovative technology for monitoring is broadly low, except for long‐span bridges. Long‐span bridges have been the focus of numerous studies and generally involve large and expensive monitoring systems. The cost of the systems, and the limited number of field trials to validate their capabilities, are the main reasons for the lack of widespread adoption. The management of short and medium bridges, which make up 95% of our network, is currently achieved using traditional visual inspections. The focus of this research is to identify the level of useful information that can be gained by using minimal instrumentation that is economically feasible to install on short and medium span bridges. MID is a frequency‐based method and accurate frequency extraction from acceleration data is vital. The widely utilised frequency extraction methods work best when the acceleration data has a high signal to noise ratio. When using a sparse number of low‐cost accelerometers, a lower signal to noise ratio is expected. A novel method was developed to overcome this, optimising the quality of the extracted natural frequency. The broad principles of data modelling have already been established; however, each data set is subtly different and requires a customised workflow to achieve the best results. Using data collected from a 36 m steel girder bridge, a novel data modelling workflow suitable for sparsely instrumented bridges was developed (MID). MID could detect an abnormal frequency shift of as little as 0.01 Hz, representing a localised stiffness loss to one girder of 25%. This detection capability is equal to, or exceeds, many studies presented in the bridge SHM literature. Notably, this data also contained the confounding effects of the environment present in real‐world data. The workflow was further tested and refined on four additional bridges of varying spans and construction types to ensure a similar level of performance across various types of bridges. While the performance of the data models varied across each of the different bridges, the average abnormal frequency shift that could be detected was 0.021 Hz and the damage detection capability was broadly similar to the first bridge. As the vast majority of bridges in infrastructure networks are short and medium span, this low‐cost approach to SHM could significantly affect the management of bridges, if the significant challenges of low‐cost SHM can be overcome. The average cost of the SHM systems used in this study was £300; this cost is unlikely to be prohibitive in a bridge management budget.

Date of Award	Jul 2022
Original language	English
Awarding Institution	Queen's University Belfast
Sponsors	Northern Ireland Department for the Economy
Supervisor	Patrick McGetrick (Supervisor), Desmond Robinson (Supervisor) & David Hester (Supervisor)