Original Article: https://www.newcivilengineer.com/sponsored/arcadis-uses-artificial-intelligence-to-spot-bridge-defects-17-10-2023/

How do we use data to make a step change towards better bridge performance?

In 1994, a section of the upper truss of the Seongsu Bridge in Seoul, South Korea collapsed, killing 32 people. Five years ago, the Ponte Morandi bridge in Genoa collapsed, with 43 fatalities. These incidents highlight a recurring problem that is in part tied to climate change-related extreme weather events. But the common thread appears to be a lack of effective maintenance, which always depends on quality asset data.

Better data improves asset and risk management. Ageing UK infrastructure, originally built post-World War II, now faces increased usage and underinvestment, leading to a rising number of substandard bridges. In 2019, 4.4% of local council bridges were classified as substandard, and the maintenance backlog was £6.7bn.

In 2023, tight budgets are straining infrastructure in need of substantial investment. Delaying capital investment will only raise future costs for maintenance, rehabilitation and new construction. Arcadis is collaborating with Niricson to pioneer Advanced Bridge Inspections in the UK, revolutionising the way bridge condition data is collected and analysed.

Why advanced bridge inspections matter

Bridges are critical and complex structures within the UK’s transportation networks.
Ensuring their safety and performance is challenging and prone to human error due to inconsistent inspections. Typically, inspections rely on visual assessments and tools like tapping hammers, leading to variations in methods and subjective interpretations. This also involves working at height and exposing inspectors to traffic hazards, making it hard to track bridge conditions and deterioration over time.

However, recent advances in artificial intelligence (AI), machine learning, robotics, computer vision and Niricson’s patented acoustic technology present the possibility of digitising the inspection process to inform maintenance, so it is targeted, repeatable and more efficient than ever before.

The possible solution?

The solution employs an autonomous robotic device like a drone to collect three layers of data. First, optical images detect and quantify surface defects like cracks and spalls. Second, an infrared sensor identifies subsurface issues such as delamination and moisture ingress up to 50mm from the top concrete surface. Lastly, Niricson’s unique acoustic sensor can detect and quantify delamination up to 200mm in concrete.

All this multi-layered data is collected by Dronic proprietary data collection technology using these sensors. It is then processed by Autospex software which employs deep learning to automatically detect and quantify defects. The software generates a defects map showing the location and extent of surface and subsurface deficiencies.

Arcadis engineers use this information to consistently and comprehensively assess bridge health.

This data feeds into risk and asset management systems, enabling informed decisions for the long term safety and durability of the assets.

The benefits

Having these detailed data sets makes it possible to generate a baseline condition assessment of bridge structures which can be tracked over time. An easier, quicker
and more comprehensive reporting system, it also gives much greater foresight in terms of predicting cost as well as materials and resources. This is of huge benefit to asset owners, governments and ultimately the taxpayer.

With Niricson’s Autospex, an AI-enhanced predictive analytics platform, Arcadis offers a holistic Bridge Health Solution, allowing all stakeholders to collaborate over digital defect maps for the first time.

Through the remote inspection and monitoring of concrete bridges, this service provides support to the whole lifecycle of the asset working to extend the lifespan, saving on capital expenditures, rebuilds and CO2 emissions. But most importantly, it ensures risks to the public and asset owner are drastically reduced.