From lab to sensor: learning about the possibilities together

The TKI project ‘A fast picture of microbiological water quality’ was completed recently. Practical experience was acquired with four sensors, each of which analyses microbiological water quality in a different way: the number of bacterial cells (BactoSense, bNovate), the active biomass (BugCount Guardian, LuminUltra), the enzyme activity of bacteria (BACTcontrol, microLAN) or the presence of bacterial nutrients that determines the rate of biofilm formation (CBM, Milispec BV).

Finger on the pulse

“As a drinking water utility, you want to know whether these sensors are advanced enough to use in your operations,” says Marang. “That allows you keep your finger on the pulse better than the usual analyses in the lab during the renewal or adaptation of processes.” KWR researcher Nikki van Bel adds: “Some drinking water utilities in other countries are already using some of the sensors we have tested. But chlorine is often added to the drinking water there. Chlorine kills off the microbiology: you’re talking about very different processes. That’s why it’s a good thing to take a closer look at sensors in the Dutch situation – without chlorine – as well as to explore what these technologies can do for our drinking water sector.”

Practical obstacles

Evides was the first pilot location in the project. With the exception of the BugCount Guardian (the sensor that measures active biomass), the sensors were tested at a drinking water production location. Marang: “We were already working on our own research into the effects of an additional treatment step during drinking water production. The question was what additional information the sensors could provide.” Before the sensors could do their job, there were practical obstacles to overcome, Marang says. “In the area of ICT, for example. The main challenge we came across was that, as a drinking water utility, we are very secure. But obviously, you want to link the sensors to the Internet so you can look at the results remotely. The proper arrangements were put in place in the end but this is definitely something you need to keep in mind.” The operators’ handling of the sensors also required training, says Van Bel. “Things like changing glass tubes or knowing how to control and monitor different sensors. In the light of experience, modifications were made to the sensors where required and they were then tested at the same pilot location or the next one. Lessons from practice like this are also very useful for suppliers.”

Signal value

Pilot projects at a range of drinking water utilities have shown that the added value of the sensors depends on the location. “At Evides, for example, the results from the BactoSense were a good match with the results from the lab measurements,” says Van Bel. “The results with BACTcontrol were more problematic. In the final pilot project, the sensors were tested at a renovated Oasen production location, where we managed to establish a good link between the results from both sensors and operational changes in the treatment.” The dependency on the location makes it difficult to compare locations straightforwardly and to come up with a universal signal value, for example, explains the researcher. “A normal value for one type of drinking water may be very high for another. We were often able to explain why a sensor detects a rise or fall in the number of bacteria but, in this project, this wasn’t an early warning system you could use in practice. In addition, water at drinking water production locations meets the guidelines naturally. Where the measurements were made, the statutory parameters were not exceeded during the project. So that makes it very difficult to find a signal value.”

Follow-up project

Given the results, Evides does not yet believe there is a permanent application for the sensors at production locations, says Marang. “But we are still looking at how we can use the sensors. In the follow-up TKI project “Peak-controlled source detection for Meuse water quality”, which started last year, we want to test the BACTcontrol sensor with surface water taken in as untreated water to produce drinking water. If it performs well, it could provide us with an option to use microbiological parameters in order to control the intake of water from the Meuse.”

Industrial water

Even for industrial water, sensor performance appears to be closely related to the type of water source and how the water system is structured. Nienke Koeman, a KWR researcher, supervised this part of the project. “At Dow in Terneuzen and at BASF in Antwerp, we facilitated the implementation of the pilot projects and worked up the data. At Dow, the BactoSense turned out to be a source of useful information in the system that was tested. At BASF, the BACTcontrol was more suitable. You really do need tailor-made solutions.”

More frequent and direct monitoring

At Dow, all four sensors were tested on cooling water, says Niek van Belzen, who works at this manufacturer of products such as plastics and chemicals. “We use the cooling water in our cooling towers. It consists of pretreated surface water and treated effluent from our biological water treatment plants. There is a risk of biological fouling in the heat exchangers and we tackle that using chemicals. Given our sustainability goals, we want to reduce the use of those chemicals. We are also aiming to reuse water and more. So we are looking for ways to monitor the microbiological water quality of cooling water more frequently and directly. Sensors could be a good solution.”

Positive results

Van Belzen explains that Dow already uses a lot of sensors but that microbiological sensors are new. “Of course, the conditions are different for industrial water and drinking water. Our cooling water contains more microorganisms, for example. I was pleasantly surprised by the outcomes. We have established a clear picture of the data generated by the various sensors. And what may constitute valuable information for us. For example, the BactoSense provided the most immediate results. That allowed us to see clearly the impact of changes in our process. As soon as we added another source to the cooling water, we saw an immediate change in the measurement data. That’s a lot more sensitive than the lab measurements, which we do once a week. In addition, this sensor also did well in terms of robustness and reliability. We could run it for a long time without checking it.”

A lot of knowledge produced

After the project, Dow kept the BactoSense in use for a time to measure other water flows, such as the water in storage tanks. “We wanted to know about the bacteriological water quality there as well,” says Van Belzen. “The whole process has produced a lot of knowledge for us. We were primarily interested in the possibilities for the medium and long terms. Before we purchase a sensor of this kind, we need a picture of the costs and benefits. And we have to make an assessment of how urgent it is. A TKI project like this makes it possible to generate a lot of data for a relatively low investment in time and money. With a research institute like KWR, you can squeeze out a lot more information than if you work on your own. And so this type of research is very suitable as a way of seeing how we can improve our processes. Not only in Terneuzen but also worldwide. We are continuing to monitor developments in microbiological sensors closely.”