Research helps to reduce the impact of Substances of Very High Concern on drinking water supplies

Models and data serve as a basis for predictions of treatment efficiency with the Aquapriori tool

As part of the collective research for the water utilities (BTO), research is being conducted into Substances of Very High Concern (ZZS), which may pose a threat to drinking water quality. During a webinar on 16 March, interim results were presented on modelling the toxicity of ZZS and their behaviour in soil and treatment, and the automatic searching of publications for relevant data about ZZS. These models and data will be included in the Aquapriori web tool, which helps water utilities to predict the effects of treatment processes on ZZS.

Substances of Very High Concern (ZZS) are substances that are hazardous for people and the environment. The water sector is increasingly running up against ZZS – examples being pesticides, industrial substances, narcotics waste or trace medication – and new emission pathways. Sound knowledge and information are required about the presence and behaviour of substances, their emission pathways, their impact on sources and whether they can be removed during the production of drinking water.

Cross-thematic project

For all these reasons, the cross-thematic BTO project Impact of Substances of Very High Concern in the Environment is working on areas that include sampling campaigns, prediction methods for the behaviour of ZZS in the subsurface and treatment, and the impact of these substances on human and environmental health. This project will continue until late 2023. During a webinar on 16 March, sixty BTO participants listened to an overview of the current status of this project.


Renske Hoondert (KWR) gave a presentation of the work on modelling ZZS toxicity. Machine learning has been used to investigate the relationship between the molecular properties and structures of ZZS on the one hand and toxicity on the other (as a response value to a range of bioassays). This was presented in conjunction with the clustering that was done earlier in the project, when the substances were ranked by chemical structure. A representative substance was selected from each cluster and used in the monitoring campaign. The focus was on two toxicity predictors: a traditional indicator, the octanol-water partition coefficient, and the chemical structure (structural alerts). Chemical structure proves to be a good predictor of toxicity, but which of these alerts is the strongest predictor of toxicity depends to a major extent on the type of bioassay used to examine toxicity. Relationships between substance properties/structures and toxicity will be further explored and explained mechanistically in future research in order to make it possible to prioritise substances on the basis of expected toxicity.

Behaviour in treatment

Bas Wols (KWR) gave a presentation on the behaviour of ZZS in treatment. The results of a sampling campaign at the full-scale treatment plants showed that most measured ZZS were removed during treatment but that the concentrations in the raw water were often too low to arrive at a proper estimate of treatment efficiency. That purification efficiency is needed for model development. A monitoring campaign has therefore now been established in which ZZS are continuously fed to activated carbon filters on a pilot scale. These measurements will be used to improve the models further. The models will be made available through the AquaPriori web tool, which can be used to estimate the removal of any organic micropollutant.

Behaviour in the subsurface

Bas van der Grift (KWR) gave a presentation on the behaviour of substances in the subsurface. This information can be used to estimate the transport, breakdown and/or adsorption of a ZZS in different types of extraction method. Measurement data for a large number of substances from actual extraction methods and from the sampling campaign with ZZS are used here to develop the subsurface models further. These models are also added to the AquaPriori web tool.

Text mining

Xin Tian (KWR) gave a presentation on the use of natural language processing (a form of artificial intelligence to parse human language with a computer) to automatically extract a large amount of data from the scientific literature (text mining). This information can be used in the treatment and subsurface models to predict the behaviour of ZZS.

ZZS pilot setup