Flanders Leads In Water-Quality Measuring

By Koen Triangle, Bart Braem, Greja Brom-Verheijden, and Philippe Michiels

Working with research partners and industry, imec develops and implements sensor, data, and network technology for Internet of Water Flanders (IoW Flanders), a four-year project designed toward fine-grained and real-time monitoring of water quality in Flanders, Belgium.

Internet of Water Flanders is a collaboration between imec, VITO, Flanders Knowledge Center Water (VLAKWA), Flanders Environment Agency (VMM), De Watergroep, and Aquafin, with the support of Flanders Innovation and Entrepreneurship (VLAIO). Over the next four years, the project will install more than two thousand wireless sensors across the whole of Flanders to supplement the monitoring network already in place.

Currently used methods to monitor water quality rely on manual sampling using scoops and on a relatively small number of locations in combination with powerful, yet expensive, multiparameter probes. The IoW Flanders project will take these methods to the next level by further development and large-scale rollout of state-of-the-art sensor, data, and network technology.

Imec sensor (pictured in the beaker) for monitoring various aspects of water quality, with a cable connected to a measuring station for wireless two-way communication with the cloud.

Amongst others, IoW Flanders will use self-learning algorithms and hydrological models to process all relevant data and translate them into information that policymakers can use. The output of these algorithms will soon be used to help support a wide range of water-related decisions.

Koen Triangle, project manager City of Things at imec, says, “The project consists of two main phases. For the first two years, the focus will be on drawing up inventories, research, development, and testing. For the two years after that, we will use the insights gained to implement the system on a large scale across Flanders.”

Sensor Technology for Water-Quality Monitoring

At the moment, IoW Flanders has installed four sensors as part of an initial learning cycle. Greja Brom-Verheijden, senior R&D engineer at imec Netherlands (Holst Centre), points out, “Three sensors are already in place in rivers in the province of West Flanders to measure water salinization. The fourth sensor is located in the village Aartselaar at an Aquafin water treatment plant. We have deliberately placed this sensor in untreated sewage water so that we can test it in an extreme environment. The sensor is positioned just after the coarse dirt filter so that it can avoid any mechanical damage, but before the actual purification filters.”

In order to upscale the process further, it is important to monitor the water quality with a generic sensor, which means it’s independent of the specific conditions of an individual use case. Brom-Verheijden says, “Each sensor module consists of a small printed circuit board (PCB) with a sensor chip. This PCB has a protective epoxy coating in the places where contact with water must be avoided.”

In terms of functionality, each sensor measures the electroconductivity (EC), temperature, and pH. Other measurement parameters yet to be implemented are the oxidation reduction potential (ORP) and the nitrate value. Brom-Verheijden continues, “The ORP may be an important addition for analyzing aspects such as dissolved oxygen concentrations. A low redox potential indicates that a great amount of oxygen is being consumed. Also, we intend to use the ORP values to interpret the data from the pH sensor correctly.”

Imec sensor developed for the IoW Flanders project.

The pH sensor itself is patented imec technology. It operates with a measuring and reference electrode. The reference measurement is taken in a microreservoir on the chip that is connected to the water to be analyzed. To ensure a reliable reading, one must take into account the deviations that, after a certain time, inevitably occur in the reference measurement.

Brom-Verheijden states, “The pH sensor is subject to what is called sensor drift, which is the emergence of diverging readings throughout the lifetime of the sensor. These type of deviations are caused by the dynamics in the composition of the reference fluid. The greater the difference between the measurement fluid and the reference fluid, the faster all kinds of substances will migrate to and from the microreservoir. As a result, more compensation is needed for the altered composition of the reference fluid.”

Deviations may also occur because the electrodes get affected by residues from dirty water. At the same time, and this may sound counterintuitive, the sensors may age even more quickly due to water that is too pure. Brom-Verheijden reveals, “In pure groundwater, for example, the difference in concentration with the reference fluid is greater (more diffusion), which will result in more sensor drift. Also, the metals in the electrodes will dissolve more quickly in purer water, which causes additional degradation.”

Imec compensates for the sensor drift via the data-interpretation software, but only after the raw data have been collected and transmitted. Same as for the interpretation of the EC values. “The EC sensor is based on four electrodes that measure the impedance. By applying a small alternating current to the outer electrodes and calculating the resistance from the resulting voltage on the inner electrodes, you get an indication of the quantity of ions in the water.”

EC measurements are relatively accurate and also fairly robust. Unlike the pH readings, for example, they are less affected by sensor drift or deviations. This means that the raw data, the variations in EC values, tell you that something is happening, but you don’t yet know what.

Contextual information is required for a qualitative analysis or data from nearby sensors. For example, via the synchronous response of adjacent sensors, you can track the evolution of a type of pollution or other event. Also, you can then exclude sensor failure being the cause of deviant readings. Or, the other way round, identify possible faulty sensors and recalibrate them if they are the only ones giving abnormal signals for no apparent reason.

Network & Data Infrastructure for Detailed Water-Quality Monitoring

All of this data communication and interpretation are part of the other areas of expertise that imec is bringing into the IoW Flanders project. The sensors positioned in the water are connected by a cable to a measuring station. This device is located on the surface and houses the necessary infrastructure for wireless two-way communication with the cloud.

Bart Braem, senior business developer at IDLab, an imec research group at the University of Antwerp, informs, “For communication, we use low-power wide area networks (LPWAN) because these are best suited for reliable low-power communication. We can choose between a number of standards to do this, such as LoRa and NarrowBand IoT (NB-IoT). These are part of the standard offering of just about every major telecom provider in Flanders, although they each have their own economic and technical pros and cons. In the end, we have to choose one of them and base our protocols on it.”

To make this choice, imec-IDLab created a network tester. Bart Braem describes, “Our network tester is equipped with six antennas and the necessary radio chips to receive all of the potential communication standards of each provider. We will use the tester to scan the various networks across the whole of Flanders. In parallel, we will conduct communication tests using the sensors. Once completed, we can make an objective choice as to which standard and which provider gives us the best coverage and can assure the most reliable results.”

Network tester the IoW Flanders project will use to scan the spectrum across the whole of Flanders to select the most suitable wireless communication technology and provider.

And there’s more to it than just that. In addition to the standard protocols, imec will program extra algorithms to monitor and upgrade the reliability of the wireless communication to ensure that the required data arrives in the right place.

“Through the wired connection, the sensors deliver raw data to the measuring station. From there, the data travels via the cloud so that it becomes accessible for relevant applications. This may appear straightforward, but in the specific context of IoW Flanders, it actually raises a whole series of challenges,” says Braem. “The sensors and measuring stations are positioned in relatively difficult environments — often in places that have a lot of vegetation, which may obstruct wireless communication. And these locations are not always easy to reach if you need to perform maintenance or repairs. This means that the robustness of the data streams is crucial.”

As default setting, the sensors will measure and transmit data every 15 minutes, a frequency that can be remotely increased or decreased if needed. The aim is also to have both sensor and measuring station run on a single battery charge for six months. This means that the required robustness cannot be achieved by increasing the power. Which is why the IoW Flanders project uses the imec OCTA platform, a proprietary modular development platform to design optimal systems for robust and low-power data transmission from IoT sensors to the cloud.

Philippe Michiels, technical and data architect, City of Things, injects, “Having a suitable and effective design for the data stream is crucial to the whole project. The key is to keep the data stream as compact as possible and yet maintain a full data model in the end.”

To achieve this, a variety of aspects have to be taken into account. For example, how do you accommodate for interruptions caused by sensors that are temporarily out of order or being serviced? And how can you guarantee reliable reference data as a basis for further analyses and calibrations? Michiels continues, “We have learned from VITO that it can make an enormous difference if a sensor changes location in the water, even by half a meter or less. These are aspects that you need to take into account during longer-term monitoring. Or, for example, when a sensor is put back in position after maintenance or repair.”

In addition to robustness and reliability, compatibility is also important. To guarantee interoperability with other systems, IoW Flanders will use widely accepted European standards such as NGSI. This will enable data from the IoW Flanders sensors to be correlated with the existing data from VMM and other parties. It also means that sensors can potentially be plugged into each other’s systems.

All of this requires a thorough analysis of how you actually process the data. Philippe Michiels states, “The multiparameter probes that VMM uses have in-built intelligence for functionalities such as calibration. With imec sensors, calibration is carried out further down the data stream. This and other often subtle differences in data processing need to be catered for by a robust IT design. Ultimately, the IoW Flanders project needs to deliver a data architecture in which every stakeholder is able to collect, calibrate, and process their own data efficiently and securely.”

Other decisions that need to be taken are what you want to store and for how long. Data storage requires substantial resources, and in view of the volume of data generated, a carefully considered retention strategy is a must. And not just steered by technical and business considerations. For example, if the data is used for policy choices, then you need to take account of the legal framework of what types of data may or may not be destroyed.

Challenges Are Not Always Technological

Clearly, the broader context is very important from the outset, and this is manifested in many areas. Brom-Verheijden offers, “For instance, Aquafin asked us to reposition a sensor to a specific location because it was safer for their staff. There was too great a risk in the place where we originally wanted to install the sensor. There, for sensor maintenance, the staff would have had to work in twos and wear protective equipment due to any gas released from the sewers.”

Braem adds, “You also need to take account of the maintenance of the banks where the measuring stations are located. Grass and vegetation are mowed and trimmed regularly, and you do not want your equipment to be damaged.”

Michiels reveals, “Scale is also a factor we shouldn’t lose sight of. If more than two thousand sensors located all over Flanders require maintenance or a new battery every six months, this will mean an average of about twenty sensors to be dealt with daily. That will require a well-organized logistical effort.”

The good news is that the right expertise is on hand. And many of the underlying models and challenges can also use knowledge gained from other City of Things projects. For example, the underlying technical architecture will have a lot of similarities with the one that will be implemented in Mobilidata. Thanks to the knowledge of consortium partners, which will soon be supplemented by private tenders, an infrastructure will be created that at all times provides an accurate insight into the various aspects of the water quality across the whole of Flanders.


Internet of Water Flanders is a collaboration between imec, VITO, Flanders Water Knowledge Centre (VLAKWA), Flanders Environment Agency (VMM), De Watergroep and Aquafin, with the support of the Flanders Innovation and Entrepreneurship Agency (VLAIO). Visit the Internet of Water Flanders website. To learn out more about the monitoring of water quality or about Internet of Water Flanders, read the company’s general blogpost. More details are available about smart environment and other themes, such as how IoT technology can also help flood prediction.

About the Authors

Koen Triangle

Koen Triangle is a project manager at imec City of Things. Through his great interest in technology, the projects he has managed have always had a technological flavor. He has enjoyed pushing boundaries in both an international and national context for the past decade. In a smart street lighting project, he took on the role as a project manager in order to connect the various partners and to support the project in reaching its goals. Before joining imec, Koen worked for Delaware Consulting. This global company focuses on software implementation, helping its clients to convert their strategic goals into applicable software tools. While working at Delaware Consulting, he managed international projects with an internationally based team, with clients based in Israel, America, and Belgium. As a result, he has managed teams in America, China, and Vietnam, converting the goals of our clients into completed projects. It has always been Koen’s role to deliver projects that meet client needs within the scope defined, while taking contextual changes into consideration. To support transparent decision-making in the project, he applies an open, constructive, and inclusive style of communication. This keeps all of the parties informed and enables them to contribute to the project and strategic goals.

Bart Braem


As a senior researcher, Bart Braem contributes to the interplay between research and valorization at imec-IDLab Antwerp. Bart works on research project proposals or bilateral agreements, focused on smart cities, Internet of Things, and Artificial Intelligence at the Edge.




Greja Brom


Greja Brom is currently employed at imec-NL as a senior research and development engineer in the sensor group. Her main task is to lead the sensor development project within the Internet of Water Flanders (IoW Flanders) program. The sensors developed at imec are based on silicon technology, which has been a fundamental part of Greja’s entire career, first as process development engineer (at Royal Philips Electronics) to develop new process technology for the interconnect of CMOS (transistors), followed by designing and processing Micro Electronic Mechanical Systems (MEMS devices) at NXP Semiconductors and, later, by designing, processing, and characterizing sensors at imec-NL.




Philippe Michiels

As technical architect for imec City of Things, Philippe conducts research to establish which technologies are sufficiently reliable and scalable to be able to facilitate smart city data flows and storage. After gaining his master’s degree in computer sciences, Philippe immersed himself in optimization techniques for database systems while studying for his doctorate. After this, he gained numerous years of experience as a software and integration consultant in the private sector, where he was involved in various industrial fields, including chemicals, production, and logistics. He then started his own business and took part in the construction of integration systems and collaboration platforms for industry.