Utilities worldwide are under growing pressure to modernise ageing water and wastewater infrastructure while maintaining reliability, affordability and public trust. Digital technologies such as artificial intelligence, machine learning and digital twins are increasingly being positioned as tools to help utilities move from reactive maintenance towards more predictive, data-driven asset management.
In this Q&A, Steve Rupar, a senior water infrastructure specialist at WSP, shares insights into how digital tools are reshaping asset monitoring, predictive maintenance and long-term capital planning across complex water networks. He also discusses the data governance, cybersecurity and organisational challenges utilities must address as they accelerate digital transformation.
Utilities are under growing pressure to modernize ageing infrastructure while maintaining service reliability. How are digital tools like AI, machine learning, and digital twins helping utilities move from reactive to predictive asset management?
Digital tools will be increasingly valuable in helping utilities “see” what is happening underground. Before the digital age, utilities could monitor pressures, flows and water quality only in person at facilities such as pump stations and valve chambers. They could also only assess pipe asset condition when the pipeline was uncovered for other work. By analyzing limited amounts of data and identifying patterns that are not readily observable, AI and machine learning will enable utilities to predict major failures in advance of their occurrence with greater accuracy. Similarly, digital twins are combining the power of GIS, SCADA, AMI and hydraulic models to provide a real-time view of pressure, flow and other features of the underground piping network. These exciting developments will allow utilities to be much more proactive in addressing system issues.
Digital monitoring and predictive maintenance tools are increasingly being deployed across water and wastewater treatment facilities.
Predictive maintenance has become a major focus area for smart utilities. What types of data or monitoring technologies are proving most valuable for detecting early warning signs in critical assets such as pumps and pipelines?
Systems can monitor a wider range of critical assets, enabling more accurate prediction of problems before failure. For pipelines, acoustic networks can be deployed to monitor for leaks, changes in flow direction and even failure of components. Predictive failure analysis tools are constantly improving as well, combining data from new monitoring systems with routine operational data to pinpoint likely failure hot spots in advance. Pumps and motors can now be monitored in real time for conditions such as energy use, temperature, and vibration. This information, when combined with data and machine learning from maintenance and work order systems, can reveal problem areas as they develop and allow for a more proactive response.
Strategic planning is another area being transformed by AI. How can machine learning algorithms support long-term investment decisions, and what challenges remain in adopting these models across large and diverse water networks?
Machine learning is already helping many water systems predict the presence of lead service lines where records don’t exist, based on patterns of occurrence that aren’t readily apparent. In combination with digital twins, AI algorithms can now analyze thousands of potential water or sewer main replacement and/or rehabilitation decisions – such as which mains to replace and where to add new mains to close loops – to resolve system issues like low pressure or inadequate fire flow in a much more rapid manner than the trial and error approach used in traditional hydraulic modeling analysis. This will be increasingly important for large systems, where the optimized mix of potential solutions can be very complex. Additionally, AI can be used to plan treatment plan process selection and layout, at a variety of sites, to give system planners a more effective roadmap for planning. AI models remain a specialty tool, typically requiring outside specialists to operate, but this will be expected to change as models become increasingly sophisticated, and more data is available.
Integrating machine learning with existing systems like SCADA, hydraulic models, and customer billing requires careful data governance. What best practices are you seeing among utilities that are successfully navigating this transition?
As utilities transition to the digital age, they will come under increasing threat of cybersecurity attacks. Establishing protocols, training staff and conducting regular cybersecurity audits and reviews will be critical to protecting systems from bad actors. Data storage and software systems should be duplicated in multiple secure locations. Public utilities should also consider their Freedom of Information Act protocols, to protect truly critical information, while allowing the public to view what they need to monitor utility performance.
Looking ahead, what steps do you believe utilities should take in 2025 and beyond to build the organizational readiness needed to fully leverage digital transformation while maintaining public trust and operational resilience?
Cost savings will not be overnight. Utilities should communicate the upfront costs of deploying AI, machine learning and digital twins to their customers, while emphasizing the long-term payback in terms of reduced operating costs, better customer service, less main failures and other benefits. Internally, utilities need to engage their staff early and assure them that these tools will not reduce jobs but will instead help utilities prepare for increasingly stringent regulatory requirements and rising customer expectations.
