Developments in areas such as digital twins, instrumentation and artificial intelligence all connect to open up digital opportunities for utilities. The Source hears from contributors on these themes to the IWA digital white paper series.
Digitalisation in the water industry has progressed apace since the advent of digital data collection, to the extent that generally only a small fraction of the increasing quantities of the data gathered is put to use. However, advances in sensors and artificial intelligence, and the move to create ‘digital twins’ of systems, means that this data is now beginning to prove its value.
The series of white papers emerging from the International Water Association’s Digital Water Programme underlines the considerable potential (see box for details). Professor Zoran Kapelan, of Delft University of Technology in the Netherlands, is one of the authors of the recent paper on artificial intelligence (AI) based solutions for the water sector. He notes that work is being undertaken in many different areas, including early warning systems for the detection of bursts/leaks in drinking water systems, but also addressing many other issues in these, wastewater and other systems.
Professor Kapelan himself has undertaken much work on a range of AI applications in the water sector, with some of the solutions ultimately being used by water companies. Part of his work includes development of new methods for real-time modelling antiestrogeni and management of water systems, pushing in the direction of digital twins – virtual digital replicas of a system.
A new project about to start at TU Delft, with research institute KWR and utility Waternet, will use AI and machine learning to improve different aspects of real-time operational management of wastewater treatment works. Another project, with start-up company Noria and infrastructure agency Rijkswaterstaat, will look into AI-based monitoring of plastics in urban waterways; the Netherlands has many open channels, which Kapelan says the government is keen to monitor and remove plastics from. “AI has a lot of applications – it can be used to address a wide range challenges in the water sector, as shown in the recent IWA AI white paper,” he observes.
Many start-up companies are appearing in the area of AI, he adds, headed by people graduating into, or already present in, the water sector, but also others from a predominantly IT background. “People are sensing the opportunities – the market is developing fast. If someone decides to start up a company, it is not surprising – the water infrastructure in the Western world is old and deteriorating, and will need major investment in the coming years, of around a trillion dollars worldwide,” he says. “The big numbers become interesting for start-up companies to get in there and address these challenges.”
Most are applying different machine learning methods to extract useful information from various data sources, says Kapelan, using this information for improved operational management and long-term planning of water systems. “There are new sensors for water quality, new computer vision-based methods to assess the condition of sewers, and drone-based technologies to perform effective inspections and other tasks; companies are trying to develop and integrate a range of these new technologies with existing ones. Integrated solutions help water companies to do business better. There is a lot going on – it is a very exciting time for AI in the water industry.”
Kapelan sees a need for a greater focus on actuators in future – the devices that make it possible to implement control of the systems. “If you want to automate a system, sensors are only part of the story. If you want to control a system well, you need actuators, in addition to smart algorithms. There is lot more that can be done there,” he says, adding: “Digital twins, I believe, are definitely a way to improve things in the future.”
Automated or semi-automated systems will still very much rely on past experience and knowledge, Professor Kapelan says: “We will still see human operators in control of water and wastewater systems, but assisted with digital water technologies encapsulated via advanced software and hardware helping them to make better decisions.”
One of future challenges includes ensuring wider acceptance of AI in the water sector, which is recognised as conservative and risk averse. “One way to overcome these challenges is to further educate people about digital water and related opportunities, methods and solutions, providing an environment where new technology becomes more acceptable,” says Professor Kapelan.
The instrumentation connection
Achieving this increasingly digital future and evolving current systems and assets into smart water systems and digital twins will require close integration with the many, increasingly ubiquitous and inexpensive, sensors. Oliver Grievson, Technical Lead at Z-Tech Control Systems (and author of another white paper in the IWA digital series), says that the challenge is, to a large extent, about how the industry collaborates. Cross-industry relationships and feedback help to develop technologies that are available at an appropriate price point.
For the operator, electronic verification that instrumentation is working correctly is crucial, Grievson notes. An operator can be based hundreds of miles from some sites, and a large company can have upwards of 20,000 pieces of instrumentation for its wastewater systems, any of which can fail on any given day. The control system needs to be able to diagnose the fault and provide sufficient information for the operator to send out the correct responder. Grievson notes that key to successful use of instrumentation is how to make proactive and reactive maintenance as simple as possible. “Alarms go off all the time; there has got to be priority,” he adds.
This is where the digital transformation comes in. “Instrumentation is the eyes and ears of the system – if it is not working, you can’t see or hear what’s going on,” says Grievson. It fits within the various layers in a smart treatment system – the physical assets, the instrumentation and control, communications, digitalisation, and artificial intelligence, he explains.
The system needs to include an element such as SCADA to simplify the data avalanche, he notes. “If you’re producing thousands to tens of thousands of pieces of data on a large treatment works, what does the operator need to understand; what do they need to do in the network? What is the data to information ratio?”
Clarity on this use feeds back to the operator’s attitude to the instrumentation. “Once an operator is using the data, they know it comes from instrumentation and they will value it; they will look after it better,” says Grievson. It impacts the initial choice of instrumentation for a capital scheme and its subsequent maintenance. “When instrumentation is installed badly, you have a vicious downward spiral – if the data is poor, the operator will not look after the instrument, and it will gradually decline,” he adds.
Such issues are particularly critical in places such as the UK, where operational manning levels are relatively low compared to some areas of the world. Grievson notes also that, in some countries, the cost of a sensor and its maintenance can outweigh the cost of personnel.
“Fundamental for me in digital transformation is using data and acting on it. It is a huge resource,” says Grievson. This does not have to be a complex process, he notes. “You need to ask what you want to find out. For me, the first part of digital transformation is stakeholder engagement.”
Each stakeholder – from the CEO, to the operator, to the treatment manager on the ground – has different information needs. Instrumentation is at the heart of this “Situational awareness is one of the most powerful things instrumentation on the ground can provide. It can tell the operating company what’s going on,” Grievson says.
True situational awareness can minimise the consequences of an issue and identify priorities, he observes. “If an instrument does not have a duty, don’t install it – it will do more harm than good. If you don’t install the right thing in the right way in the right place, what is the point? If there is no value to the data an instrument produces, what is the point?”
The future is moving towards real-time control, such as running a wastewater treatment works from a live model and using sewers intelligently to minimise polluting discharges. “At the very heart of it is instrumentation. It’s already being done in Copenhagen. To me, it will go a step further; a smart wastewater network in dry weather will use the inlet control system to balance flows if it has the capacity.” This requires knowledge of the weather, both current rain gauge data and historical data, to decide what to pass forward to smooth out diurnal peaks, Grievson explains.
While this is the future, he notes that “the technologies are probably already ready to use or just need adaptation”. The industry and its supply chain will have to collaborate, he adds, with suppliers providing the R&D if they know what is required. “The supply chain is brilliant at that, but the industry has to say ‘this is what we want’ to realise digital transformation. Again, it comes back to identifying informational needs and stakeholder engagement – it is all interlinked.”
Digital twin ambitions
The concept of digital twins is perhaps the ultimate ambition of digitalisation. Professor Ingmar Nopens, of Ghent University’s Faculty of Bioscience Engineering, and co-author of an IWA white paper on digital twins, says that “a digital twin is a virtual representation of a process or system – you can best compare it to a flight simulator, where pilots are trained to use a real system in a virtual environment”. Extending the analogy, the plane’s autopilot is a further example of the concept in which an autonomous system takes control of a real-life activity.
“The water sector could regard different applications of digital twins – as a tool to train operators or technicians to make decisions, or they could use it in real time, put the digital twin into a SCADA and enable an automatic command and control system,” Nopens says.
In his view, the key area of debate is over accuracy, and whether the digital twin can represent a real system. “If it’s a lousy digital twin, it will not have a lot of predictive power; it will come with a lot of uncertainty, and will hamper decision-making,” he says. The first examples of digital twins are “on the verge of implementation”, he adds.
In a process such as a biological wastewater treatment works, where living organisms are involved, there is greater complexity than in a purely physical system, such as a pipe network, says Nopens. He envisages digital twins being trialled in simple systems, with utilities learning from the challenges encountered there, leading to real-time applications in treatment works in the future.
“Digital twins set mathematical equations that allow the use of predictive power to model, to see what is going to happen, to change the parameters of the system, and do something about it,” he notes. With the surge in development and deployment of sensors over the past decade, there is a vast amount of data available, much of which is not currently used. “An operator looks at one screen, one graph; the brain is not prepared to look at 50 graphs. You need a computer model algorithm to help – a digital twin can help you come up with the smartest decision.”
Nopens adds that “a lot of pieces of the puzzle are lying around; it’s a case of getting them together and putting them into practice”. “I believe there will be a stepwise approach, but, in the end, I’m not sure that the industry will be completely autonomous. At the moment, a lot of work is done manually. Digital twins will complement this in the end. The operator will take the final decision – she can use the digital twin and its processes over a span of time and build up experience with it, and gain faith and trust.”
Another application not on the radar for many people is the idea of virtual pilots, using computer models to design systems. A pilot – which is, in essence, a miniaturised version of a treatment system – could be replaced by a virtual pilot, which could run many more scenarios, the results of which could be incorporated into the final built process. “To go to virtual design of a system, if you want to make the leap to a circular economy – a plant that recycles materials – I see this coming,” says Nopens. “I now think this is also part of digitalisation, and it should come into the water sector – it is undervalued. A lot of time is wasted on lengthy procedures. I think this is really important.”
A smarter future
These elements – AI, instrumentation and digital twins – combine to help craft raw data into usable insights and integration that will improve its current value significantly. The future is set to be increasingly smart, with these technologies supporting decisions across a widening span of complexity.
Widening awareness of water’s digital opportunities
The white paper series of the IWA Digital Water Programme explores some of the key themes of the digital world, helping to raise awareness of the current and future opportunities for the water sector. These focus areas include ‘digital twins’, instrumentation, and artificial intelligence.
The recent paper ‘Digital Water: Artificial Intelligence – Solutions for the Water Sector’ was authored by Zoran Kapelan, of Delft University of Technology, Emma Weisbord, of Royal Haskoning DHV, and Vladan Babovic, of the National University Singapore.
The paper aims to introduce readers to tangible solutions (rather than technologies) that were developed to address specific challenges in real-life water systems. The authors hope that, by using applied examples, the topic of AI and its applications in the water industry will be made clearer to a wider group of interested readers. The examples included were selected to demonstrate that AI-based solutions can address real challenges and provide tangible benefits to the water sector.
For more details of IWAís Digital Water Programme, on-demand webinars, and the latest on the white paper series, see: iwa-network.org/programs/digital-water
