Infrastructure AI: How analytics can detect landslips and pipe failures
Artificial intelligence, advanced analytics and digital technologies are reshaping the infrastructure sector.
It’s not simply a matter of automating traditional tasks. Modern digital tools are changing how organisations analyse and understand complex systems. Infrastructure builders use the technologies to better evaluate future options and get the maximum value from existing assets.
That will become increasingly important as New Zealand faces harder choices about which infrastructure projects to fund and how to make limited investment go further.
Eric Peissel, WSP’s global director of transport and infrastructure, says digital technology can help agencies analyse more information and evaluate more options. Yet he warns it should be seen as an essential part of the toolkit, not the answer.
“People often think that digital and AI are some sort of magic wand, but what they really do is help decision making, help optimise things at a scale that previously was a lot more difficult.”
Today, one of the most tangible impacts of these tools is in asset management. They allow infrastructure operators to move from rigid, schedule-based maintenance to a predictive model.
Traditionally, assets like pipes, power lines and roads were maintained on fixed schedules. Advanced analytics and digital sensors can add a more detailed picture of their condition. Agencies can monitor assets in real time, identify early warning signs and predict where failures are most likely to occur.
Peissel says WSP is applying these techniques to detect early signs of landslips, identify potential pipe failures and predict traffic congestion and accident risks. Instead of relying solely on fixed maintenance schedules, agencies can direct scarce maintenance funding towards assets where the need is greatest.
“We use advanced analytics and AI to take a look at a lot of data around real-time information, such as scanning for landslips and for where we’re starting to see potential landslips,” Peissel explains.
“We’re also using the tools with our clients on the water side as well. They have a whole bunch of data saying, ‘Hey, how can we do early detection of where a pipe might fail?’… We’re finding that really being able to deep dive into some of these data sets can actually give you a lot of insights, not only on what happened, but actually start doing some predictive work.”
Peissel says the growing use of advanced analytics and AI does not reduce the need for professional judgment. Digital tools can identify patterns and produce insights from vast amounts of information, but people must still interpret the results and take responsibility for the decision-making.
That is particularly important in engineering, where calculations must be accurate, repeatable and traceable. AI systems can produce misleading results if they are given incomplete or poor-quality information or if the questions put to them are flawed.
“For us, especially in the engineering industry, the human in the loop is such an important piece of our concept because if there isn’t a human validating what’s being done and checking the outcomes, there actually are significant risks.”
Peissel says infrastructure decisions also involve factors that cannot be reduced to data or engineering calculations. Environmental impacts, community views, heritage considerations and the requirements of iwi can all influence where and how a project is built.
“We make the decision of where this needs to go for the soft reasons. Then we figure out the engineering solution to make it work.”
Making these predictive tools work in practice relies on the quality and consistency of the underlying information. Peissel points out that most infrastructure owners aren’t suffering from a lack of data, they are drowning in it. As Peissel puts it, modern infrastructure owners are often “data rich, but information poor.”
Between internet-connected sensors, visual scans and vehicle telemetry, agencies hold vast amounts of data. The challenge is turning data into information that can be acted on, “A lot of making AI and these tools successful is the data analytics and the data quality. It’s not very sexy, but it can make a big difference in terms of your decision making”, Peissel says.
A project WSP undertook with NZ Transport Agency Waka Kotahi (NZTA) illustrates this.
There are around 70 local road-controlling authorities collecting asset management data on their respective networks. Historically, each council gathered, logged and formatted that data differently. The lack of consistency made it nearly impossible for NZTA to get a unified picture of the national road network or evaluate maintenance needs on a like-for-like basis.
WSP worked with the agency to develop a standardised digital process to clean and harmonise that data across all 70 regions.
“Before, it was really each region doing it a little differently. This created a system where they really couldn’t figure out if something being done on the North Island and something being done in Queenstown were comparable”.
“By helping develop a standardised way of looking at and processing the data, they could actually get a full country view of the assets, understand what they were doing, and start having a like-for-like comparison.”
Peissel says getting the foundational level of data quality right requires leadership from the top.
“There needs to be clear regulation, maybe not laws, but clear standards put in place. Governments need to take a certain hand in saying, ‘This is the way we’re going to be collecting data, these are the standards, this is the common data environment I’m looking for.’ Because once the recipe is there, everyone can work through it.”
Without those common standards, agencies risk wasting time and capital trying to reconcile incompatible systems.
“A lot of times people think they’re all using the same recipe, but they’re reading it out of different recipe books. One person’s adding a pinch of salt, another’s putting in two pinches; then we’re trying to figure out why it doesn’t work at the end of the day. Governments need to get into how the sausage is made.”
Peissel warns infrastructure agencies against taking a “salt and pepper approach”, that is trying to spread resources thinly across too many trendy technologies at once.
Instead, he advises concentrating on core asset management foundations and exercising caution with unproven tech.
“In our sector, in the civils world, in the transport world, you want to be on the leading edge, not the bleeding edge. Focus on the good foundational elements that set you up for success. Don’t go chasing the shiny, cool-looking toy, because in the sectors we serve, there is a high opportunity for it not to pan out for a large degree of investment and time that could be much better placed elsewhere.”
WSP is a sponsor of the Herald’s Infrastructure report.