How predictive maintenance can protect critical infrastructure during extreme weather

Lightning storm over Melbourne CBD (powercor)
Image: Shutterstock

By George Dragatsis, Chief Technology Officer Australia and New Zealand, Hitachi Vantara

As wild weather continues to play havoc with ageing infrastructure assets, inspection and maintenance strategies and approaches are once again a focus for discussion.

The recent collapse of six transmission towers in Victoria, causing power outages for half a million people, has once again brought into focus the impact that climatic conditions can have on critical infrastructure, and the crucial role of maintenance crews in being able to detect at-risk infrastructure before something catastrophic can occur.

Related article: Victoria’s power outage could have been far worse. Can we harden the grid against extreme weather?

In the latest incident, transmission towers were knocked over or bent by a wild thunderstorm with wind gusts of up to 122 km/h. It’s not unprecedented. In January 2020, six transmission towers near Cressy in Victoria similarly collapsed in a thunderstorm.

An investigation report into the 2020 occurrence stated that “severe convective downbursts near the electricity transmission infrastructure caused the collapse of the towers. The wind speed at the time of the incident was determined, based on the tower damage and damage to trees and other structures in the vicinity, to be in excess of 125km/h … beyond the design specification of the towers”.

Maintenance inspections—cyclic and ad hoc—had been performed to standard.

But the question today is whether existing standards and approaches could benefit from being updated to make monitoring of infrastructure degradation more continuous, and maintenance more predictive.

The uneven impacts of localised climate conditions

Traditionally, maintenance inspections of critical infrastructure are often manual, requiring field crews to either climb infrastructure or to conduct visual inspections via helicopter or, increasingly, using drones. This is not only costly but time-consuming due to the vast geographic reach of infrastructure in Australia, and the sheer number of towers to inspect.

But the age of the infrastructure, combined with the increased frequency of environmental conditions that test dated design limitations, is driving a need to inspect more often in order to detect degradation of the steel or components that put the infrastructure at risk of failure.

There are hundreds of thousands of kilometres of transmission lines around Australia. Within that network, assets are at various stages of degradation.

Climate change and other environmental conditions are causing them to degrade at different times and at different locations. Different degrees of exposure to the elements— wind, moisture and temperature—produces different rates of degradation and different maintenance requirements. At scale, and using predominantly manual techniques, this can be a hard challenge to solve.

Aged infrastructure is not only at-risk of collapse due to extreme weather and fires.

Operators are also acutely aware that a failure of their infrastructure due to age or other causes can cause fires as well. As the industry notes, “Electricity can start bushfires when infrastructure is damaged or foreign objects contact powerlines. While the number of bushfires ignited by electricity is very low, once started they have the potential to burn large areas. Mitigating the vulnerability of networks to damage and faults is essential.”

The potential of AI-powered predictive maintenance systems

Forward-thinking operators in this space are increasingly turning to AI-powered predictive maintenance systems to improve the resiliency of their infrastructure.

This trend has, in part, been brought about by the increased frequency of extreme weather events, such as devastating storms, floods, heatwaves and droughts. “Once-in-a-century” occurrences now happen more frequently, and as the climate crisis worsens, these events will only multiply—as will disruption to critical national infrastructure.

In many instances, these are faults and weaknesses that might have been spotted or fixed during the next round of scheduled maintenance. When extreme weather strikes, however, it is too late to fix things, and the costs—both in terms of money and human impact—can escalate rapidly.

Consideration of weak points that could potentially fail during extreme weather has to be one of the business drivers for a predictive maintenance program.

Predictive maintenance systems use AI to crunch data from sensors atop drones, satellites and cameras to make calculated bets on when, where and how infrastructure might fail under the stress of extreme weather.

Using long-established techniques like computer vision, image analytics, deep learning and basic pattern recognition, these applications can alert maintenance teams to anomalies or problems that humans wouldn’t be able to spot without assistance.

Related article: Report says renewables require less maintenance

Many utilities already capture images of their infrastructure as part of the inspection process, but could benefit from the ability to analyse tens of thousands of images in seconds. The AI in this instance can narrow down the vast amount of image data into a shortlist of infrastructure with potential defects, flagging them for verification by experienced maintenance personnel. The model gets more accurate over time, enabling an ever-increasing cadence of higher quality defect detection.

Adaptation and mitigation have become two of the key objectives as our world readies for the impact of climate change. The growing shift towards technology-driven solutions is crucial for ensuring safety and reducing economic losses in an era of increasingly unpredictable climate patterns. As such, predictive maintenance will play a vital role in supporting our objectives as we work towards making our world more resilient.

Previous articleActewAGL opens 2024 Community Grants program
Next articleNew streetlight EV charger being trialled in regional NSW