AquaConnect researchers on a tour of infiltration ponds in the Epe water extraction area. Vitens infiltrates up to 6 million cubic meters of clean water annually from the Klaarbeek and Grift streams, reclaiming it after passing through the ground.
AquaConnect researchers on a tour of infiltration ponds in the Epe water extraction area. Vitens infiltrates up to 6 million cubic meters of clean water annually from the Klaarbeek and Grift streams, reclaiming it after passing through the ground.
In the very dry years 2018, 2019 and 2022, there were shortages in all components of the water system in the high sandy soils of the southern, central and eastern Netherlands – in agricultural soils, stream discharges and groundwater. This was because the water system has an intensive drainage system to quickly drain surplus water [1]. In 2021, the AquaConnect programme started research into key technologies that contribute to the transition to a different, circular water system: retaining water longer and promoting reuse to better withstand prolonged droughts. AquaConnect is funded by the Dutch Research Council (NWO) and 36 end users from the water sector (see www.aquaconnect.nu), including drinking water company Vitens and the regional water authority ‘Vallei en Veluwe’.
Operational problems arise during extreme drought and with low groundwater levels. Vitens supplies drinking water to consumers and businesses in the eastern and central Netherlands. More than 95% of this drinking water is produced from groundwater. The Vallei en Veluwe water authority is responsible for water management and wastewater treatment on the sandy soils of the Veluwe and the Gelderse Vallei. In the past, there was enough clean groundwater to meet demand, but now both organisations face new challenges:
Within AquaConnect, replenishment of groundwater resources with wastewater treatment plant effluent has been identified as a possible solution to make the water system on high sandy soils more resilient to drought. This effluent is produced continuously and in many places is now discharged into major watercourses, even though this water could be useful to replenish groundwater reserves. The Veluwe, for example, potentially lends itself to the active infiltration and storage of large amounts of freshwater. In AquaConnect, important new scientific insights related to active infiltration of sewage treatment plant effluent (including finding suitable infiltration sites), micropollutants and consumer perception were obtained.
Suitable infiltration sites
Valdrich Fernandes (WUR) trained three machine learning models (U-net, Attention U-net and encoder-decoder) to quickly and accurately predict the effect of active infiltration on groundwater levels [2]. He did this by comparing these models with the AMIGO groundwater model of the Baakse Beek catchment. AMIGO, or Actueel Model Instrument Gelderland Oost (v3.1), is in use by the regional water authority ‘Rijn en Ijssel’, Vitens and the Province of Gelderland. The machine learning models were trained based on six different input parameters: groundwater depth (m), infiltration volume (mm/day), depth below the riverbed (m), drainage resistance, aquitard (poorly permeable layer) resistance and aquifer permeability. Ultimately, the models provide three key insights regarding infiltration: groundwater level response (m) (Figure 1), spatial response (km2) and volumetric response (m3). The spatial response refers to the area for which a higher groundwater level is predicted, and the volumetric response refers to the total volume of infiltrated water.
Figure 1. The modelled groundwater level response around Doetinchem by the AMIGO model and the machine learning model U-net for infiltration areas of 1 km2 with an infiltration rate of 15 mm/day. Adapted from Fernandes et al. (2025) [2].
Of the three machine-learning models, U-net was found to be the best at approximating AMIGO's prediction, but 3,000 times faster than AMIGO (0.24 s versus 1,290 s) [2]. With U-net, hydrologists can therefore evaluate infiltration scenarios much faster and identify suitable locations for active infiltration.
Micropollutants
With active infiltration, water quality is also important. Jan Specker (UvA) conducted a literature review on the risks of micropollutants with different types of reuse of wastewater treatment plant effluent (including active infiltration) [3]. He found 48 publications from around the world on field-scale reuse of wastewater treatment plant effluent in which the presence and removal of micropollutants was analysed. In these 48 studies, 1,051 different micropollutants were measured. The top 10 most frequently encountered micropollutants and transformation products are shown in Table 1.
Tabel 1. On the left, the top 10 most detected micropollutants, and on the right, the top 10 most detected transformation products in wastewater treatment plant effluent. For each micropollutant, the number of studies in which it was found/measured (total = 48) is given [3].
Using the observed concentrations of these substances and the 'predicted no effect concentrations' (PNECs) from the NORMAN database, it was possible to calculate a risk quotient for ecological and human risks when reusing sewage treatment plant effluent. The result was a list of micropollutants with ecological (PFOS, chlorpyrifos, triclocarban, ethinylestradiol) and human (PFOS, PFOA) risks [3].
Jan's study shows the importance of monitoring and/or removing certain micropollutants before reuse. In the 48 studies analysed, different techniques were applied, ranging from natural post-treatment (soil passage) to a double reverse osmosis barrier. The latter has the highest removal efficiency [3]. Within AquaConnect, technological and natural removal methods have also been investigated – nanofiltration and soil passage, respectively.
Removal through soil passage.
Alessia Ore and Jill Soedarso (WUR) have improved our understanding of the sorption and biological transformation of micropollutants during soil passage. In a wet (2017) and a dry (2019) year, they used non-target screening to monitor effluent from the Haaksbergen wastewater treatment plant before and after subsurface irrigation for 89 micropollutants and associated transformation products. Unlike target screening, non-target screening measures known and hitherto unknown substances [4]. This screening showed that a large proportion of the substances detected were transformation products (2017: 43 out of 56; 2019: 72 out of 90) [5].
Additional research has been conducted on the removal and transformation of micropollutants during soil passage of surface water (for drinking water production, Figure 2a) [6]. Almost two-thirds of the micropollutants in the infiltration water were largely (>90%) removed. However, highly persistent substances such as artificial sweeteners reached groundwater abstraction points. Here, non-target screening also revealed new substances (Figure 2c) that had been left out of the picture with target screening. These included transformation products formed in the aquifer, demonstrating that the observed removal was partly due to biological transformation [6].
Figure 2. Detected number of micropollutants and transformation products (TP) over a soil passage route (A) using target screening and non-target screening. B = target screening; C = non-target screening. I = infiltration pipe; M = monitoring pipe; A = withdrawal pipe; DW = drinking water. PFAS was only measured using target screening. Adapted from Ore et al. (2025b) [5].
Jill studied the sorption of 54 micropollutants to 5 sandy soils; these were representative of natural purification systems in the Netherlands. The aim was to identify substance properties that influence the sorption of micropollutants to sandy soils, which would then allow better prediction of their leaching to groundwater. Recent lab experiments show that 32 of the 54 micropollutants tested did not or hardly adsorb to sandy soils. For the 22 micropollutants that did adsorb well, it was examined whether there was a relationship with properties of those substances and characteristics of the soil material such as (i) organic matter content and (ii) capacity to bind positively charged ions. That turned out not to be the case. Jill's further research focuses on finding the properties of soil and micropollutants that do explain sorption in order to model leaching risks.
Alessia and Jill's research shows that biodegradation and sorption do not remove all micropollutants. Combining soil passage with membrane filtration (such as nanofiltration) can ensure that micropollutants do not contaminate groundwater sources.
Figure 3. Schematic representation of an 'inside-out' and 'outside-in' nanofiltration membrane
Nanofiltration
Tjerk Watt (UTwente) researched new configurations for coated hollow fibre nanofiltration membranes. With these nanofiltration membranes, it is possible to retain micropollutants that are not removed during soil passage as a result of sorption and biodegradation. In addition, these membranes block divalent ions but allow monovalent ions to pass through, producing a less saline concentrate.
The application of several layers of charged coatings to a support membrane is crucial to this. Normally, such a coating is applied to the inside of the support membrane ('inside-out') (Figure 3). Tjerk investigated whether this coating could also be applied to the outside ('outside-in'). This results in an increased active surface area, allowing the production of narrower and stronger fibres that fit into a smaller module. By reversing the porous structure of the support membrane in this way, a hollow fibre nanofiltration membrane with similar retention to the same membrane with the coating on the inside could successfully be developed [7]. However, the active surface of the membrane with the coating on the outside is 3.8 times larger, so a smaller module is needed to treat a given amount of water.
The role of consumers
When reuse of wastewater treatment plant effluent is discussed, consumer perception is often cited as a major barrier to a transition to a circular water system. A recent H2O article showed that Vitens employees are generally positive about water reuse [8], but does this also apply to consumers? Research by Noelle Lasseur (WUR) among 2,000 respondents showed that most people are not against the reuse of treated wastewater in principle, but they are rarely actively involved in developments. This lack of engagement leads to wrong assumptions about public perception. In her research, Noelle also interviewed water professionals about the challenges in transitioning to a circular water economy with decentralised water treatment and local water reuse. Two main challenges emerged:
Noelle concludes that better economic assessment of the positive side effects of a circular water system, such as an improved local water regime, could accelerate the transition process to a circular water system.
Conclusions
The AquaConnect research programme provided important new insights into active infiltration of wastewater treatment plant effluent into the groundwater system:
Active infiltration of sewage treatment plant effluent after removal of micropollutants through natural as well as technological treatment, combined with innovative monitoring, provides a useful application for the effluent. It can contribute to water security on the high sandy soils in the Netherlands and beyond.
In the recent dry summers, groundwater levels on the high sandy soils dropped to such an extent that drinking water production, nature and agriculture were compromised. At drinking water company Vitens and the Vallei en Veluwe water authority, the question arose whether infiltration of wastewater treatment plant effluent is an option. Within the AquaConnect programme, this has been explored from different fields of expertise. In this article, we list new insights into finding suitable infiltration sites, the presence and removal of micropollutants and consumer perception.