Prediction models for the removal of organic micropollutants

Removing organic micropollutants from the water

The removal of OMPs from the water remains an important focus of attention. Both old and new substances regularly occur in both surface water and groundwater. This raises the important question as to whether the current treatment is capable of removing these substances and, if not, which treatment steps are needed to achieve this for a specific substance. In the present project the focus is on modelling the most commonly used barriers against organic micropollutants, namely: activated carbon filtration, membrane filtration and oxidation.

In a previous Joint Research Programme project, existing models for activated carbon filtration and membrane filtration were further developed, and oxidation (both based on UV/H₂0₂ and ozone or ozone/H₂0₂) was added. A substance list was compiled which covers the range of properties of micropollutants in water. These were tested for activated carbon filtration and a number of different membranes. The models were in this way validated. The role of natural organic matter (NOM) in both the adsorption of OMPs during activated carbon filtration (ACF) and the degradation of OMPs in oxidation processes remains an important research question, as does the role of substance load in the membrane models. In addition, the oxidation models need to be validated.

Improving models for OMP removal

In this project we address the following research questions:

1. The presence of NOM plays an important role in adsorption on activated carbon. It appears that different NOM fractions have different influences on when a substance degrades. Additional experiments need to be carried out with water with different NOM fractions. The isotherms of different NOM fractions (and of possible different water matrices) can be used for the NOM and OMP isotherms in the ACF model. The addition of NOM fractions improves the applicability of the models for various water matrices (each with its own NOM composition).
2. Validation experiments are needed for the oxidation modelling, especially for ozone processes. Besides the transformation of OMPs, it is also important to include competition with NOM (reaction of ozone with NOM and/or NOM fractions) in the experiments, so that the model is also applicable to other water matrices. Furthermore, the oxidation processes based on UV require better predictions of the photolysis constants for new substances.
3. In the membrane models, differences in the removal of OMPs between traditional NF/RO membranes are scaled on the basis of salt retention. To improve the prediction for other membranes, this is deepened further by taking load differences into account (of the OMPs, but also for instance between NaCl or MgSO₄). This load interaction is certainly important for the separation in the case of (capillary) NF membranes. To better incorporate the load interaction into the model, we aim in the follow-up to evaluate other types of capNF membranes as well. We are also working on a translation of OMP retention data of capNF membranes on a lab scale to full-scale application, using larger recoveries (with a feed&bleed system). This will be validated through experiments.
4. A picture is required in the case of all models of the uncertainties in the parameters and the predictions.

Assessing treatment impact

When a new substance is encountered, a rapid assessment can be made with the models as to whether it is removed by the current treatment and/or whether urgent measures are required. A prediction can also be made about what type of additional process would possibly be the most effective. In the design of new treatment processes or renovations, it is important to know which treatment processes and conditions are best for the removal of a specific OMP.