by Eliza Booth, Journalist, Infrastructure magazine
As technology continues to develop faster than ever before, and with the Coronavirus pandemic highlighting the need for improved digital solutions, machine learning and artificial intelligence are set to play an even bigger role in the future of asset management. At the recent Asset Management for Critical Infrastructure Virtual Conference, part of the 2020 Critical Infrastructure Summit, infrastructure and utility leaders gathered for a panel about the future of machine learning, artificial intelligence and digital twins in asset management. Here are the key points from the discussion.
Speakers on the panel included Nicola Belcher, Director of Rail Assets, Projects and Compliance, Victorian Department of Transport; Russell Riding, Automation Team Leader, Melbourne Water; and Bruce Thompson, Executive Director Spatial Services, NSW Department of Customer Service.
How the pandemic impacted asset management
Since COVID-19 hit, the utility and infrastructure sectors have needed to change some aspects of the way they work very quickly, including keeping critical assets running while navigating constant changing restrictions and lockdown measures.
During the panel, the three speakers discussed the ways their work has changed this year. Ms Belcher said that the rail sector provides a critical service for the public and that while Victoria was in Stage 4 lockdown following a second COVID-19 wave, the rail industry needed to keep operating to get frontline and essential workers to their workplaces.
Ms Belcher also explained that while the majority of office staff were able to transition to remote working, projects like the Level Crossing Removal project and the Melbourne Metro Tunnel project are considered critical, with work continuing despite the lockdown restrictions.
“The biggest change to maintenance and renewal activities has been the work practices out on-site. So in Victoria we need COVID-Safe plans for people to be able to continue to work on-site and that has required high-risk areas to consider which people they have on-site, and how long they spend on-site,” she said.
Ms Belcher said the Victorian Department of Transport recognised a big opportunity for the uptake of technology on sites, including utilising drones for further workplace monitoring – something her department is looking further into for the future.
Similarly, Mr Riding said that due to the critical nature of water supply, Melbourne Water’s field staff have been required to continue working through lockdown, all work was carried out following up-to-date state health advice including specific hygiene and social distancing protocols.
“We’ve had to rely on the use of newer technologies such as video conferencing, Skype and Microsoft teams to be able to provide that support to the asset management team,” he said.
This is working well. Mr Riding said apps, which were already in use at Melbourne Water, had become more critical to the team during this time and allowed the utility to utilise, and become more familiar with, technology they already had in place.
Mr Thompson said that the NSW Department of Customer Service was fairly well prepared for the changing work environment, having trialled remote working before the pandemic hit, so processes were already in place to assist with the transition.
“We ran up about 70 additional virtual machines in our cloud-based server centre and we’re getting productivity gain out of that because large image processing jobs, and other things that we do in that knowledge space, are now taking less time,” Mr Thompson said.
Machine learning for critical assets
Machine learning has become a major tool for infrastructure and utility companies in recent years with the need for autonomous technology to help monitor and manage critical assets.
During the panel, Mr Riding discussed one of Melbourne Water’s first machine learning projects, which focused on pump selection.
“We have a treatment plant, Winneke Treatment Plant, up in the north east of Melbourne and there’s a major pump station there that supplies water to that treatment plant – there’s actually six different pumps there,” Mr Riding said.
“We implemented a machine learning solution that basically looks at historical data and determines the most efficient combination of pumps and speeds for any particular flow rate, and it then directs the control system on-site to select those pumps in that manner.”
Mr Riding said that since the implementation of the machine learning technology approximately two years ago, Melbourne Water has seen a 20 per cent (or around $200,000) saving in the energy cost for running the pump station at the Winneke Treatment Plant.
Artificial intelligence and image recognition
During a follow up question from one of the delegates, Mr Riding also explored the role that apps and artificial intelligence can play in efficient water operations.
“Because [Melbourne Water] is field based, it’s easier for us to collect data via an app and have it in real time. We’ve already invested in certain apps to capture calibration data and similar sorts of asset information so that we can understand if the asset is performing correctly and do some reporting to understand where we have issues and address them correctly,” Mr Riding said.
“Overall, having those apps allows us to provide a higher level of service, whether it’s water supply, sewage treatment, or even drainage.”
Mr Riding also talked about the innovative IoT and camera technology Melbourne Water has implemented to manage blocked drains in the utility’s network. Images are taken of stormwater grates that regularly get blocked with rubbish and debris.
These images are then uploaded to a cloud platform where artificial intelligence and image recognition technology is used to alert the asset management team of blockages in the network, who then mobilise teams to clean the blocked grates on an as-needed basis.
This reduces the resources needed to manually check for blockages after storms as the technology will identify specific assets that need attention, thereby improving overall efficiency.
When it comes to other technologies having significant impacts on asset management, Mr Thompson talked about the benefits of digital twins in the sector.
“The spatial digital twin is a way of bringing together people’s data in an accurate and reliable way so that we can position private sector infrastructure, whether it’s a building or an asset, against government infrastructure and assets,” Mr Thompson said.
“The other benefit is the temporal nature; being able to look at the world at the point in time a transport model was generated, maybe five years ago, and understand what the road network was like then rather than what it’s like now.
Then being able to move forward in time and put down multiple proposals or competing proposals – a road corridor against a train line, or a road corridor against doing nothing – and compare what’s happening.
“It’s really about bringing that together on this digital workbench.”
You can watch the full panel discussion, Asset Management for Critical Infrastructure Virtual Conference, or any of the Critical Infrastructure Summit events for free on-demand. Visit www.critical-infrastructure.com.au.