Finding leaks with pressure and AI

It can be challenging to find a water leak within a network of pipes. Colombian-Australian researcher Dr Jessica Bohorquez has developed a system that uses AI to increase the chance of detecting cracks.

It can be challenging to find a water leak within a network of pipes. Colombian-Australian researcher Dr Jessica Bohorquez has developed a system that uses AI to increase the chance of detecting cracks.

Dr Jessica Bohorquez completed her Bachelor’s and Master’s Degrees at the Universidad de Los Andes in Bogota, Colombia. She graduated Summa Cum Laude with her Bachelor’s in Civil Engineering, studying the use of drag-reducing agents in heavy crude oil pumping. Bohorquez completed her Master of Science in Civil Engineering, graduating Cum Laude. She majored in Water Resources and Hydroinformatics, with her thesis titled Optimization in valve operation for minimising transient flow effects in Water Distribution Systems.

Moving from civil engineering to water and pipes

“I didn’t know much about the water industry within civil engineering. I thought civil engineering was about bridges, buildings, and roads. Halfway through my degree, I started taking water-related courses and realised it was a fascinating industry to be involved in,” she said.

Bohorquez got the opportunity to work in a research group focused on water infrastructure early in her undergraduate degree. She spoke to Inside Water about pipeline design, inspection, and modelling – crucial aspects of her research.

“That was the area I eventually focused on for my research,” she said. “I started as an undergraduate research assistant in Colombia and became a lecturer at the university before coming to Australia. I spent a lot of time in the research group, and I will always be thankful to collaborate with them.”

Meeting her supervisors

“I met my supervisors, Professors Angus Simpson and Martin Lambert, at a conference back in 2015,” Bohorquez said. “At the time, I was finishing my master’s and wanted to stay in research. We had a conversation and just clicked. We kept in contact, and I discovered I could get a scholarship to come to Australia. I applied successfully, and I’ve been in Adelaide since 2017.”

Bohorquez studied under Professors Simpson, Lambert and Dr Bradley Alexander at the University of Adelaide from 2017 to 2021, completing her PhD in civil engineering.

Since moving to Australia, Bohorquez developed a more focused interest in pipeline inspection and leaks. Her supervisors have been vital to that. One thing that she was intrigued about was the use of pressure waves to find leaks and bursts. The connection between the three is one of the reasons Bohorquez moved to Australia to continue her research.

“My supervisors have been working on this area for over two decades,” she said. “We discussed a common interest in hydraulic transients, also known as water hammer. It is a type of pressure wave that moves quickly along a pipe. Understanding how this wave propagates is vital, and we have been trying to work out how to use them to inspect pipelines.”

What her research has achieved

The key takeaway from her research is the combination of artificial intelligence with pressure waves to find faults in major water pipelines faster than existing methods. She is utilising the deep learning capabilities of AI to increase the chances of detecting cracks in pipes.

“In a country where water is scarce, there is an urgent need for this technology,” she said. “At the moment, about 15 per cent of treated drinking water is lost through cracks in pipes before it reaches households in some cities in Australia. These cracks are sometimes tiny and almost impossible to locate.”

The best way to picture Bohorquez’s work is to shout into a cave. In the case of her research, the echo is the pressure wave reflecting on different defects along the pipeline.

What Bohorquez has found

“One of the big advantages is that we can see changes in the pressure wave that are indicative of changes in the pipeline,” she said. “For example, we can close a valve along the pipeline quickly to create a wave of pressure. We can find changes in materials, leaks, corrosion, or bursts.”

“We measure the change in pressure at a very high frequency and interpret what those changes mean.”

The big step in Dr Bohorquez’s research is her consideration of creating an algorithm that can read the pressure waves when they reflect along the pipeline. She worked with her supervisors to develop a machine-learning tool to process the information. Bohorquez wanted to see if it was possible to train an Artificial Neural Network (ANN) to the point where it would provide a quicker and more accurate answer.

“Our results so far indicate that once fully trained, the AI can pick a leak within three per cent of its actual location just a few seconds after the pressure wave has passed it,” she said. The future of this technique aims to design sensors connected to an AI system that learns to identify the small and short-lived changes in the behaviour of the wave when it encounters a defect in the pipe wall.

Other techniques in waves

When considering the thought of an echo in a cave, a variation of that technique is also used in Adelaide. While it is not a focus of Bohorquez’s research, she understands the importance of acoustic waves in detecting leaks.

“There is a whole project here in the Adelaide CBD around it. Colleagues have deployed acoustic sensors that do a short recording every night between 2 am and 3 am. With that processing, they can also find leaks in pipelines,” she said.

The difference between the two techniques is the balance of accuracy and distance. Both waves spread throughout the pipes in different ways. That impacts the distance they can travel, as well as their accuracy.

“Acoustic waves tend to propagate over a shorter distance than pressure waves. However, acoustic waves could provide more detail than enhances accuracy. Understanding that balance is important, and we can also use both technologies together,” Bohorquez said.

She gave an example of searching pipelines of 50 to 100 kilometres with pressure waves. That could provide results to the nearest 10 to 20 metres, which allows for better use of acoustic waves in that smaller area. That smaller area could allow for pinning a leak or a burst pipe down to a much shorter distance.

Future developments and challenges

One of the challenges of using waves to find leaks and bursts is understanding the dissipation of the waves throughout a complex pipe network.

“Both acoustic and pressure waves will bounce and reflect off everything they find. In a network, the topology is far more complex. It means that the waves dissipate quickly, so, at the moment, we can lose a lot of information quickly. That’s one of the challenges with the technology now,” she said. “It will be a fantastic technique for really long underground pipelines.”

Another focus of the research group Bohorquez is part of is to develop the technology on a commercial scale. So far, they have developed a platform that will work on most pipes. The goal is to create a solution that allows for the relevant data to be plugged in so that companies know where the leaks are in their pipes.

“We can create a model that can be customised for each system. It allows us to train the data for their systems to help their algorithm learn quickly and adjust to the intricacies of their network,” she said.

Her future

Bohorquez has continued her postdoctoral research part-time at the University of Adelaide. She is also working at Inside Infrastructure, a consultancy and advisory firm that is part of Ricardo PLC. Her work in consultancy has allowed her to see what else is out there in the water industry.

“I really want to get to the point where my communications skills are at the forefront of what I do,” she said. “I want to improve the industry and help solve problems in a way that provides many opportunities for me to progress.”

Related Articles:

Send this to a friend