Structural health monitoring gives assurance that the 54 year-old Forth Road Bridge is in fine form, while machine-learning will soon be predicting where any weaknesses are developing.
Gigabytes of data are pouring off Scotland’s Forth Road Bridge (FRB) every day. The information is automatically collected, processed and analysed to help maintenance contractor Amey manage the A9000 trunk road across the Firth of Forth.
The data comes from 500 sensors installed on the suspension bridge. They monitor, among other things, temperature, wind speed, vehicle load and displacement, giving insight into its structural health and whether the road is safe for the public to use (New Civil Engineer July 2017).
The first set of these was initially required to monitor the behaviour of the bridge following the discovery of a fractured steel member, known as the truss-end link, in 2015. Initially, temporary splints were applied to the link, and seven similar ones. This was then followed by permanent structural repair to support the new truss-end post off brackets fixed to the towers (New Civil Engineer December 2016). The problem had been that a pin had seized up, causing large forces to pass through the original “link” member, causing the fracture. Access steelwork for the last of the truss-end arrangements was due to be completed in November as this issue went to press.
Instrumentation and sensors on the eight new truss-end posts at tower locations (see diagram) measure stress, bearing load, bearing wear and truss-end movement. They demonstrate that this solution is working as it should be. Sensors have also been placed on four similar details near the bridge’s side towers as well as on other key members to give the contractor a full picture of the bridge’s condition.
“Structural health monitoring (SHM) reassures us, [bridge owner] Transport Scotland and Scottish ministers that the bridge is safe and helps us identify problems before they happen,” says Amey major bridges director Ewan Angus. “If defects develop, then we are alerted and we have time to take action.” So far, he is delighted to report that there is, well, nothing major to report.
Discovery of the fractured truss-end link in 2015, coupled with knowledge in 2004 that the bridge’s main suspension cables were corroding, led to the construction of the replacement bridge – the Queensferry Crossing which was completed in 2017. This now carries the 100,000 vehicles a day which were using the Forth Road Bridge, leaving the older structure to carry buses, pedestrians and cyclists.
Forth rd bridge
In the meantime, the care and attention the Forth Road Bridge continues to receive – cable corrosion is being controlled with dehumidification – means that it can support the full weight of traffic if an incident causes the Queensferry Crossing to be closed (see box).
Wit less traffic using the Forth Road Bridge, it is easier to carry out essential repairs, such as replacing its expansion joints, without too much disruption.
Essentially, two pieces of software make up the FRB’s SHM system – an inspection defects and repair management database called Pearl and a data gathering and analytical tool called Mercury. Amey developed both pieces of software, with significant investment from client Transport Scotland.
To date, £3.5M has been spent on the SHM system and a number of improvements have been made in the last two years. The ability to filter key information to generate reports is saving time and is making better use of the information stored on Pearl.
“I picture it like a Rubik’s Cube of information where you can slice through it in different ways to get to what you want,” says Angus. A mobile app connected to Pearl also automatically generates inspection reports and uploads them onto the database. Photographs of Amey staff carrying out this activity put into context just how difficult the field work is due to location, height and exposure.
Pearl is also being used on the Queensferry Crossing and will be rolled out to two other cable stayed bridges – the Erskine bridge north west of Glasgow and the Kessock bridge in Inverness – with others to follow later.
“Having information from more bridges on Pearl means that cross-bridge reporting is now possible, so Transport Scotland can, for example, call up the deterioration of cables across all four bridges or the condition of expansion joints,” explains Angus.
These four “major bridges” are part of the 5% of structures owned by Transport Scotland that require special attention.
“For good data and asset management, we wanted to develop a major bridges database. Although there aren’t many major bridges, they have large value forward maintenance plans,” says Transport Scotland roads directorate chief bridge engineer Hazel McDonald.
Forth rd graph
“Typically, in any financial year, they consume more than 50% of the structures maintenance budget.”
Mercury brings together data from a number of different SHM suppliers on the Forth Road Bridge and analyses it.
It is then able to translate these readings into useful characteristics, which can be compared with threshold values set up by Amey. Alerts based on a traffic light system are sent to designated people, also advising on the agreed course of action, such as closing the bridge to traffic.
So, explains Angus, when Storm Ali struck on 19 September this year, sensors detected that a bearing on one of the new truss-end brackets had “jumped 200mm north”. Corresponding readings from anemometers recorded a maximum wind speed of 134km/h and, around this time, a change in axial load of 600kN was picked up by another sensor in one of the diagonal members of a side tower.
The read-outs also confirmed that these components returned to a “rest” state immediately after the storm subsided. Meanwhile, in another location, a rocker bearing supporting a truss-end at a side tower location continued to articulate as intended during the storm, so that no bending was transferred to the adjacent member.
We’re looking for a needle in a haystack, but with machine learning we will find that needle
All this feedback assures Transport Scotland that the Forth Road Bridge is in good health, even when the road has to be closed to traffic due to high winds.
In the last two months, machine learning has been added to Mercury, which will enable it to use historical data to identify trends and behaviours not yet anticipated by the contractor.
The software will be able to predict dangers ahead. So, data collected from Storm Ali combined with weather data for an oncoming storm will eventually inform Amey if any safety measures or repairs must be done in advance. So far, no alarming trends have been discovered – it takes time to collect data to make meaningful predictions – but the capability will lead to better maintenance planning and even alert levels refinement.
“We’re looking for a needle in a haystack,” admits Angus. “We’ll never find it on our own, because we can’t possibly see everything that is happening to the bridge. But with machine-learning, we will find that needle.”
Angus’ analogy also applies to the newly built Queensferry Crossing where the “haystack” is effectively being built from the moment machine-learning comes online. Arguably, a new bridge should be without the worries of a 1960s one, but Angus stresses that the importance of the Forth corridor to the economy is what is also at stake – worth some £1M a day.
“We’re not expecting to find anything catastrophic on Queensferry,” he asserts, but with some moving parts and high winds always a possibility, there is the potential for failure somewhere down the line. SHM, as at the Forth Road Bridge, will also involve visual inspections carried out by engineers.
“The future is to continue the use of technology to have better insight into the future,” he says, adding that a virtual reality (VR) model of the Forth Road Bridge has already been created.
Amey is looking at how engineers can inspect the bridge while wearing headsets that allow a VR model with vital data to be viewed on a flip-down visor.
It is also exploring the benefits of 360° camera footage captured on site, which can be scrutinised later in the comfort of a VR viewing room.
The FRB’s system is currently world-leading and many bridge owners and maintainers around the world have visited the bridge to learn about how its life is being extended.
“I think all bridge owners are facing the same issues in maximising the operational capacity of their ageing stock while minimising disruption and expenditure,” explains McDonald.
“As the structures are ageing, access for major maintenance works is becoming more difficult and budgets don’t always permit optimum and timeous interventions, so the benefits of instrumentation for these structures will increase.”
Two thousand sensors have been installed on Queensferry Crossing to monitor its structural health, costing around £10M.
Although the cable-stayed bridge and M90 motorway across the Forth Estuary was opened to the public in August 2017, there is still outstanding work that must be completed before it is handed over to Transport Scotland and subsequently onto Amey to maintain. This is expected to take place at the end of 2018. Until then, Amey has been unable to collect and process data from Queensferry’s sensors, which will use Pearl and Mercury in the same way as on the Forth Road Bridge for its structural health monitoring.
Future of Bridges | Monitoring the Forth Road Bridge