Decision support tool based on machine learning for damage control in concrete structures

Bridges play a central role in our modern infrastructure. They facilitate transport and communication, connecting communities and regions. One of the most prevalent types of bridges in our cities is the prestressed concrete bridge, which has the ability to span long distances and support heavy loads. However, many of these bridges, built over 50 years ago, now face serious challenges due to age, environmental impact, and lack of maintenance. To ensure that our bridges remain safe and functional, innovative solutions are needed to detect damage before it leads to disasters.

Concrete bridges are subjected to a range of stresses: repetitive load cycles, weather conditions, corrosion, and material fatigue. One of the most critical components of bridges is the prestressing tendons, which hold the structure together; it is essential that this system functions as intended. The problem is that inspection of these complex structures is often both time-consuming and costly. Traditional methods, such as core drilling and visual inspections, are invasive and rely on subjective evaluations.

A tragic reminder of the importance of diligent maintenance is the collapse of the Morandi Bridge in Genoa, Italy, where inadequate inspection led to devastating consequences. This incident underscores the need for effective inspection methods that can identify structural deficiencies early, before they result in accidents.

This is where machine learning (ML) comes into play. By utilizing advanced algorithms, we can analyze large amounts of data quickly and accurately. This project aims to develop a decision-support tool that uses ML to automate the detection of damage in concrete bridges, particularly voids in the ducts that protect the prestressing tendons. By combining technologies such as ground-penetrating radar and ultrasound with machine learning, the tool can detect patterns and anomalies that the human eye might miss. By applying ML, we can not only improve the accuracy of inspections but also save time and money. Early detection of damage reduces the risk of serious problems and enables more effective planning of maintenance work.

A proactive approach to inspection and repair can extend the lifespan of these critical structures and reduce societal costs. The project will involve collecting data from fabricated prototypes and bridges through non-destructive testing. This data will be used to train the ML algorithms, allowing them to learn to recognize signs of damage. For example, mock-up samples with artificially created voids will be used to create a robust database for the algorithms to train on.

A prototype of the tool will be tested in the field on real bridge structures to ensure it functions under practical conditions. The goal is to create a user-friendly solution that can be integrated into existing inspection methods.

By focusing on improving inspection accuracy and reducing the human factor, we aim to revolutionize how we assess the condition of bridges.

It is not only engineers and technicians who will benefit from this innovation. By improving the inspection process, we can protect lives and reduce costs associated with congestion and infrastructure repairs. A more efficient maintenance strategy can also contribute to more sustainable resource utilization. By extending the lifespan of existing bridges, we can reduce the need for new construction, saving both money and environmental resources.

By applying advanced technology and machine learning, we have an opportunity to transform how we monitor and maintain our bridges. The developed algorithm for identifying potential damage in ducts will represent a significant advancement in the field of infrastructure. By combining traditional engineering with modern technology, we can not only protect our bridges but also contribute to a safer society.

The digital transformation of the construction and infrastructure sector means that we are taking an innovative step forward. We are moving from reactive methods to proactive strategies, where early detection of damage can lead to prompt and effective repairs. Through this project, we also hope to inspire others in the industry to explore and adopt new technologies to enhance the safety and sustainability of our infrastructure Projects.