Water-link detects leaks with underground sensors and Machine Learning

Xylos configured the central open cloud platform in which the company’s data are managed. The environment is built according to the DevOps principles, such as scripting the complete environment.  

Using innovation to improve customer service, efficiency and sustainability  
Azure SQL as central database 
PaaS: Platform as a Service configuration: a ready-to-use environment 
Power BI as self-service BI; Xylos trained the users of the system 

To combat water waste, drinking water producer water-link has invested in innovative technology. Efficiency, customer service and sustainability were the keywords for this project. Xylos worked with the company to create a unique Internet of Things environment in which underground sensors carry out measurements in the water pipes.  Additionally, Xylos developed a central cloud platform that collects and analyses data. Unleashing Machine Learning and Power BI on these big data could teach water-link a lot about pressure management in the drinking water pipes.   

Cloud technology  

“When developing an innovation project like this, we first need to determine if the plan is practically viable”, says Martine Similon, the Project Engineer Operations who is responsible for water-link’s innovation policy. “With our test configuration, we wanted to find out if we could use battery-powered sensors to carry out measurements underground, in the water pipes. We also wondered if these data would be relevant and if there was any practical use for them.” 

Nico Kanora, ICT Manager at water-link, adds: “The outcome of innovative projects like this one isn’t always 100% certain. That’s why it’s important to use market standards we can activate quickly, without significant up-front investments. We opted for cloud technology on a pay-as-you-go basis: this helps us provide instant service to our customers.” 

Sensors 

Water-link tasked several companies with the development of battery-powered test devices. “We need several types of sensors,” explains Martine Similon: “sensors that measure the water pressure in the pipes, sensors to build a generic module (temperature, humidity, detection, door opening, volume measurements) and sensors that measure cathodic protection, which is the voltage on the pipes which protects them from corrosion. Cathodic protection is essential to prevent pipe leaks.” 

Bagaar was one of the three parties who developed test devices. It was no easy task, because there was no ready-made solution. Martine Similon explains: “The sensors need a long battery life, they need to be powerful and moisture-resistant and they need to carry out measurements with pinpoint accuracy.” 

Platform 

The sensors collect data, which are sent to a central database through the SIGFOX network. To manage these data properly, water-link asked Xylos to supply the data management system. Nico Kanora, ICT Manager at water-link: “Xylos proposed a PaaS or Platform-as-a-Service approach. We wouldn’t just run the infrastructure through Azure Cloud, like with IaaS, but the platforms as well. This means we’re on the same page, we can support the business quickly and efficiently, we can offer ideas and participate in innovative projects.  Put briefly, we don’t need to waste time maintaining virtual machines with countless modules. Xylos is also a true expert when it comes to Azure and Internet of Things. The company guaranteed that our cloud platform would work with any technology and multiple devices.” 

  • 2,488 km of pipes 
  • 193,029 drinking water customers 
  • 25 sensors in test configuration 

Reports  

Xylos developed the central open cloud platform which will gather and process all data, convert them to the correct format and store them in the central Azure SQL database. After that, water-link users can visualise the data with Microsoft Power BI. “They can generate reports and charts”, says Martine Similon. “Xylos trained the end users, currently about five data analysts, to manage everything efficiently. This is the true meaning of ‘self-service BI’: we don’t have to wait for a supplier or IT manager to help us out whenever we want to create a new report type. This helps us gather information and get even more out of our data.” 

Nico Kanora, ICT Manager at water-link: “This is another example of modern interaction between technology and business. Instead of writing long analyses about what a report should look like, the business can now develop it independently. After all, an image says more than a thousand words. That said, self-service BI doesn’t equal chaos and unlimited access: we’re striving for a bimodal model. In a bimodal model: 

  • The business uses approved and secure data sources. 
  • The business creates personal or public interest reports. 
  • The IT department promotes, fine-tunes and publishes widely used reports on official channels. 

Crucial infrastructure in a new data centre 

Simultaneously with the IoT project, Xylos moved water-link’s internal data centre to a better-equipped floor, where the SAP HANA environment now runs on new HPE infrastructure. The entire database environment is essential to monitor the company’s tens of thousands of customers and countless invoices and meter readings. Xylos developed and installed the redundantly executed data centre with an extremely tight deadline and remains the point of contact for hardware and software support. 

Machine Learning 

As soon as water-link has collected enough data, the company can use Machine Learning to make connections and detect abnormal signals. “With enough relevant data, you can use Machine Learning algorithms to make predictions”, explains Martine Similon. “The average water pressure at certain times of day is a good example. You do have to take several factors into account, such as water being drawn at a measuring location close by, construction works and possible pipe damage.“ 

“Thanks to the Azure technology, we’ve gained more insights into our network and we’re able to make more predictions.” 

Nico Kanora, ICT Manager at water-link