Introduction
The main scope of this project is to enhance the existing Digital Twin (DT) of the IB5 Staffordshire Bridge with new features. These include new data-driven and hybrid algorithms for the continuing monitoring of the structural integrity of the bridge along with the production of a DT Control Centre (DTCC). The latter is devised so that it can also serve as a tool for demonstration and dissemination of the benefits and added-value of enhancing new and existing civil infrastructure with sensors and digital technology.
Explaining the science
Vibration-based damage detection methodologies in civil engineering structures have concentrated historically on ambient vibration or static response. Nonetheless, early damage is frequently non-observable as it is connected to small-scale and localised structural changes which affect higher-order natural frequencies, which are extremely difficult to measure reliably, based on ambient vibrations.
Railway bridges can act as natural laboratories to study structural performance due to large and recurrent train-induced dynamic responses. This creates several unique conditions including:
- Large dynamic responses are induced several times per day, which heighten damage-sensitive structural features above those ones controlled by environmental conditions, such as temperature.
- Repeated loading imposed by specific trains results in a signature response of the bridge.
- Myriad of measured structural responses are accumulated due to the frequent passage of trains.
All these distinctive features create conducive circumstances to use statistical data analyses and machine learning (ML) techniques, to detect early damage due to small structural changes and modal properties. The main obstacle to unleashing the power of such novel techniques has been the lack of permanent and reliable deployment of sensors and remote data acquisition systems, capable of monitoring continuously so that the influx of quality data is steady and guaranteed. Such obstacle no longer exists for the IB5 Bridge, thanks to the robust long-term monitoring technologies (i.e., sensors, data collection hardware, wireless communications, etc) and the present DT. This project will develop a new data-centric feature for unsupervised early damage detection and later embedded into the existing DT.
Project aims
New Feature I: Unsupervised Damage Detection
An algorithm and visualisation dashboard for an online system of early damage detection based on an unsupervised learning method.
New Feature II: Control Room for Condition Monitoring
The existing DT has 4 main dashboards with multiple online features ranging from a bridge weigh-in-motion system to structural utilisation ratios of the main steel girders. A centralised platform where all outputs of the DT coalesce is still missing. To rectify this, a control room for condition monitoring will be implemented. The objective of such system is to simplify the interpretation of results via key performance indicators (KPI) in a unified platform to provide clear, concise and actionable data. Thus, to provide the stakeholders with useful real-time information for the condition monitoring of the IB5 Bridge.
Applications
Our ever-expanding transport networks are vital for the normal functioning of modern society, where infrastructure such as bridges and tunnels play a major role. In this context, research using structural monitoring aims to reduce the costs of inspection and maintenance, as well as to extend the operational life of aging infrastructure. Although the field of Structural Health Monitoring (SHM) has seen progress in recent years due to the development of new sensing technologies and the myriad of data these produce, there are many challenges that are imposed by the nature of the problem at stake:
- Every building or asset is different, limiting the portability of monitoring approaches across these, therefore demanding time-consuming bespoke solutions each time a new SHM option is explored.
- Lack of a baseline of the structural condition in new assets, where SHM solutions are rarely included at the design stage, and in particular, in existing structures which pre-date modern structural health monitoring approaches.
- Damage scenarios are not observable, for inducing damage to a real structure is not acceptable, hence these scenarios need to be recreated in computer-aided simulations, such as Finite Element (FE) models and DTs
With the objective of having one of the most typical bridges in the UK Network Rail grid serving as a testbed for research, the Cambridge Centre for Smart Infrastructure and Construction (CSIC) and Network Rail, in a joint effort to monitor a newly built bridge in 2017. The IB5 Bridge is a Network Rail E-type steel half-through bridge with a single skew span of 26.84 m that carries two new rail lines across the West Coast Main Line near the city of Crewe in the UK. A unique feature of this monitoring project is that the sensors were embedded into the structure during the construction of the bridge, offering a structural health baseline since the start of operations. More recently, the instrumentation system of the bridge was expanded with new sensing technologies and provided with a permanent in-situ bespoke data collection system, enabling real-time monitoring of the structural response of the bridge due to live trains. In parallel, the production of a DT of the bridge started, following a novel framework.
Collaborators
