A climate-neutral, energy-positive and zero-waste perspective on water
Leakage from drinking water pipes is a global problem – including in Sweden. At VA SYD, non-revenue water accounted for 10% of all water supplied to customers. On a global scale, this figure is quite low, “but when we compare ourselves with our peers, we can still do much better”, says Simon Granath.
Simon Granath is a development engineer at VA SYD, which carried out an innovative project to detect and eliminate water leaks from pipelines using artificial intelligence (AI). The results will be used to help VA SYD achieve its ambitious goals of becoming a climate-neutral, energy-positive water utility with zero unplanned disruptions of service by 2030.
VA SYD is one of Sweden’s largest utilities, supplying more than 546,000 customers with drinking water in the regions around Lund and Malmö in the south of Sweden. In total, VA SYD operates around 5,000 kilometres of pipelines of which 2,000 km are drinking water pipes.
Some of the oldest pipes date from the late 19th century, and with such a vast and heterogeneous infrastructure, pinpointing leaks is a complex task. As Simon Granath explains: “The percentage of non-revenue water varies greatly within our network. Some areas have exceptionally low rates of leaks, while the losses in others are more significant. The issue was that until very recently, we could only detect the largest leaks that led to service disruptions – while it is the small and minor leaks, often from small service pipes, that are responsible for the lion’s share of non-revenue water.” Compared to Sweden’s average of 20% water loss through pipeline leaks – which itself is much lower than the global estimate of 50% non-revenue water – at the beginning of the project VA SYD’s 10% non-revenue water seemed relatively small, but Simon was convinced that VA SYD should be in a position to reduce this number. “Ten per cent non-revenue water, after all, translates into more than five million cubic metres of water per year that is simply lost,” says Simon. Apart from the immediate monetary aspects, non-revenue water also impacts VA SYD’s broader company goals. “This water has been collected and treated, consuming energy and resources in the process. If we want to become climate neutral and energy positive, this kind of waste is simply not acceptable.” Today, the non-revenue water at VA SYD accounts for less than 8%, which is similar to its neighbouring country Denmark.
Finding a smart solution for leakage detection
Some of the oldest pipes date from the 19th century, and with such a vast and heterogeneous infrastructure, pinpointing leaks is complex
Another aspect is improving the service to customers, adds Simon: “We have to maintain a lot of pipelines in a large area. When we have an issue with a pipeline, we need to improve our ability to pinpoint the leak through smart metering zones. And for this purpose, we had been looking at utilizing an AI-based solution because we saw this as a way to make the best use of our installed water meters and improve the hydraulic models that we have for our pipelines.” VA SYD evaluated several systems over the last few years, looking for a solution that was state of the art and made optimum use of the available flowmeters and smart meters in the area that had been chosen for the proof of concept.
Today, the non-revenue water at VA SYD accounts already for less than 8%, which is similar to its neighbouring country Denmark
With SIWA LeakPlus, Siemens offered an excellent solution powered by its cooperation partner BuntPlanet. SIWA LeakPlus ticked all the boxes, says Simon, “but when it came to the implementation concept, we ran into some obstacles. We needed to find a way to use the AI-based leakage detection in an on-premises setup instead of in a cloud environment.” Siemens was able to offer a solution for this requirement as well. “We had some in-depth discussions with Siemens, BuntPlanet, and our IT and OT departments about the best approach to install the SIWA LeakPlus solution in our own data centres. To my knowledge, this is the first time something like this has been done in the Swedish water industry – and we managed to get the job done thanks not least to the support and expertise from Siemens.”
Refining the models and initial results
Nevertheless, the initial setup of the architecture proved to be quite a challenge, says Simon: “You really need to create a suitable concept for the architecture, bring the right people together, and specify exactly what you need and how you want to get there.” After the solution had been installed, the next step was to refine the hydraulic models for the pipelines. “We had to do a dynamic calibration through a series of capacity tests and measure the pipeline pressure at several places in our metering areas. The data were used to create a very fine-meshed model that offers a much higher degree of detail than the hydraulic models we had previously.”
We used the system to simulate different leaks and then evaluated the data. We were able to detect leaks as small as 0.5 litres per second
In parallel to the modeling, the data from the metering systems and pump stations were linked with SIWA LeakPlus using OPC UA. SIWA LeakPlus is one of the Siemens Water (SIWA) applications specifically developed for the water and wastewater industry. This smart solution uses an AI-based algorithm to identify and classify anomalies in the pipe network. To perform reliably, SIWA LeakPlus is first “trained” through historical flow and pressure data so that it learns to separate anomalies from regular operations and assign the correct cause to each anomaly. The application can automatically detect whether an event was caused by some aspect of the regular operation or is due to water theft or a pipeline leak. Alternatively, the operator can also manually assign a cause to an event to refine the algorithm.
With SIWA LeakPlus, Siemens offered an excellent solution powered by its cooperation partner BuntPlanet, ticking all the boxes
After completing the linkage and training, VA SYD was ready to perform the first tests. Simon explains: “We used the system to simulate different leaks and then evaluated the data. We were able to detect leaks as small as 0.5 litres per second – this was quite impressive and a huge improvement over the previous solution, which provided no means of detecting small leaks at all.”
It was especially impressive considering that the available data were still quite limited at that point, adds Simon. “Right now, we are working on Phase 2, which will allow us to improve the leak detection so that we can pinpoint the location of the leak, and we will deploy this solution in the coming weeks. Then we will have the means to really go out there and fix all those small issues efficiently.”
Efficiency that will pay off also on a larger scale
With the proof of concept in its final stages, VA SYD is also working on scaling up the solution to other parts of the network. “The proof of concept was done in a relatively small system with 5,000 consumers per district metered area. This is a general principle here at VA SYD: start small, try things out, and learn from failures so that you can succeed on a larger scale”, says Simon. “I can highly recommend this approach, as it enables you to reduce the risk and refine the solution before you scale up.”
A further area to benefit from SIWA LeakPlus is currently the town of Lund, where VA SYD is installing additional flowmeters to provide more data from the pipeline network. “This is another aspect that we are looking into,” says Simon. “With SIWA LeakPlus, we are able to work with a smarter leakage-detection system that requires less data from the pipelines. The flowmeters on the pipes are quite expensive, and if we can reduce the number of installed meters by having a good dynamic calibration of our hydraulic system, that is something that immediately pays off.”
In parallel to the modelling, the data from the metering systems and pump stations were linked with SIWA LeakPlus using OPC UA
Economic benefits aside, SIWA LeakPlus has already helped VA SYD take the next steps to reach its broader goals, says Simon. “One obvious impact of the new leakage detection solution is that we are not only able to reduce the total amount of leakage, but we can also find and fix a leak before it becomes a burst and affects our customers. That contributes to our goal of zero unplanned interruptions of service. Moreover, the solution enables us to find all those small leaks on small service pipelines that add up to a lot of non-revenue water.”
We are not only able to reduce the total amount of leakage, but we can also find and fix a leak before it becomes a burst
Furthermore, the project at VA SYD has been widely recognized as a best practice in Sweden, leading other municipalities to adopt the same solution for water distribution in their respective areas. One such example is the NSVA water utility in the south-western part of the country. Victor Pellin, along with Simon, played a crucial role as project initiators at VA SYD. Victor later transitioned to NSVA, where he is successfully implementing the SIWA LeakPlus system in their regions. His expertise and valuable experience in leakage detection installations greatly contribute to the success of the project at NSVA.
Treating natural resources with respect is a way of life in Sweden, and water should be no exception to this practice, says Victor: “Even if here in Sweden there is not a water scarcity problem in general, wasting water just feels wrong. Plus, lifting, treating, and pressurizing that water consumes energy, and energy is a major cost factor.”
Consequently, Victor and Simon are aiming even higher than matching the 8% benchmark for non-revenue water from Denmark: “We really want to beat that. There are municipalities that achieve much lower figures, and we want to be able to measure up to the best. And in our opinion, that requires smart leakage detection and AI-based solutions like SIWA LeakPlus.”