Smart Water Utility – how smart do you want to be?

However, there are major differences: the time scale at which the production and distribution of water have to be coordinated and the role of clients in that respect. During the meeting – which took place remotely because of the corona crisis – the topics discussed included the smart use of sensors, inspiration from the energy sector and data science.

Smart use of sensors

It is expected that a better understanding of the system, both the medium and the infrastructure, will allow water companies to operate more efficiently, flexibly and sustainably, and enhance operational reliability at the same time. These improvements can be achieved by establishing close connections between individual system components in a cyber-physical environment that has also been dubbed ‘Water 4.0’. Sensors are at the heart of a system of this kind because the numbers tell the story.

During the knowledge exchange meeting, Peter van Thienen (KWR) showed how sensors can indeed provide the basis required for decision-making about interventions but that they are not a silver bullet. To know what is going on in the entire system, you need to measure everything and that is not (yet) feasible in physical or cost terms. As a result, it is important to use the available sensors intelligently. A good problem definition can serve as a basis for designing an optimal positioning strategy using models. Any blind spots can be filled in with models (soft sensors, digital twins), with the available measurements serving as a basis for predicting the status of sections where there are no sensors. Van Thienen emphasised the importance of being aware of the limitations and uncertainties that these solutions entail.

Inspiration from the energy sector

Electricity grid operators will be facing a huge challenge in the years to come. The energy transition will place increasing strain on their grids: some heating in homes has to be supplied using electricity because more sustainable energy production will be organised locally and it will be necessary to meet the energy requirements for electric transport. Traditional energy grids cannot yet cope with these challenges.

Walter Hulshorst (Smart Energy NL) talked about ways grid operators are addressing these challenges and showed how smart grids will be an essential part of the grid structure of the future. This means, in particular, that more and more sensors are being installed in the grids to monitor the load and to locate malfunctions – or prevent them altogether – more quickly. Smart meters, initially at the level of distribution stations, are essential here. In addition, the Smart Cable Guard is used. It can detect minuscule malfunctions in cables and repair them before a power failure can occur, for example when they are damaged during digging work. These applications are currently at the stage of large-scale demonstrations in preparation for large-scale roll-out.

The Vitens excavation damage app will be used to resolve low-tech problems with high-tech machine learning

Applying data science

A third contribution during the knowledge exchange meeting shed light on an important building block for the Smart Utility: data science. Data science is a field that derives insights from structured and unstructured data. It is already being applied in various ways at water companies.

Mario Maessen, who works at Vitens, explained how all kinds of new initiatives around data science are being developed here in cross-departmental ways. He gave as an example the design of an app that automatically estimates the risk of excavation damage to water pipes associated with KLIC reports (mandatory reports for mechanical excavation work).

Vitens is also working on the analysis of customer contacts by using text analysis for chat traffic. And lead pipes are being identified using image recognition and Optimal Character Recognition for over two million drawings from the archives, an activity that would be far too time-consuming to do with manual scanning.

Labelling deer in real time with a deep learning model

In the session on data science, Alex van der Helm, who works at Waternet, talked about the Waternet Datalab. The goal of the Datalab team is to develop, implement and manage data science applications in the operational processes, and to inspire and advise employees in their work on Artificial Intelligence (AI) solutions.

As a practical application, Van der Helm described the counting of fallow deer using the automatic analysis of thermal images collected with a drone. That has led to the mapping out of an area that was not easily accessible. For Waternet, as the manager of the water supply dunes, this is an important additional tool for counting numbers of fallow deer more accurately. A second application is the use of sensors and AI to optimise aeration in the Amsterdam West wastewater treatment plant and therefore to cut nitrous oxide emissions and energy consumption. This is part of the H2020 Fiware4Water project.

Hydroinformatics Platform continues despite coronavirus

Obviously, coronavirus was no obstacle to the meeting of the Hydroinformatics Platform, which was organised digitally with a GoToMeeting. A survey was sent out to the participants in advance using Google Forms to generate input. And the usual network lunch before the meeting took place on Zoom so that experiences could also be exchanged informally. Working from home in times of coronavirus provided enough material for interaction, both on- and off-topic.