Natural organic matter is one of the key components to remove from source water during drinking water production. NOM not only negatively influences the colour, taste and smell of the water, it also decreases the effectiveness of common water treatment steps such as (membrane) filtration and UV desinfectation. A big fraction of natural organic matter is negatively charged and because of that it can be removed from water with ion exchange.

Suspended ion exchange is a relatively new type of ion exchange, invented and developed by drinking water company PWN in the Netherlands. The trademark aspect of this process is that ion exchange resin gets dosed directly into the influent in suspension, followed by a filtration and a regeneration step instead of fixed bed reactors. The big advantage of this way of operating is that it can operate raw surface water, it is not prone to clogging and can operate continuously. Therefore it is much easier to implement towards the beginning of the water treatment chain, allowing it to positively impact later treatment steps.

PHREEQC is a computer program developed to do geochemical calculations. It is designed to take into account different chemical processes happening simultaneously in complex ground or water matrices.

This report is about the research done to see if the removal of natural organic matter through SIX can be simulated with a modeling program such as PHREEQC. Therefore a series of jar tests and titrations have been done to determine the relevant chemical properties of four different NOM samples, the influence of pH on ion exchange and the influence of competition between different target anions.

The counter anion exchange equivalence of NOM has shown to be varying from around 1.8 to 8. This can explain why it is observed that the adsorption of natural organic matter is almost unaffected by the presence of competing anions such as sulfate, as this high counter anion exchange equivalence suggests that NOM sits high in the anion selectivity range. Additionally, the adsorption of NOM has been constantly high over a wide pH range. This is remarkable, especially at low pH values where titrations show that the charge of NOM decrease drastically. The release of chloride by the resin does get affected by pH a lot however, although this is minimally related to the changing charge of NOM and rather by the increasing amount of bicarbonate in the system adsorbing to the resin at neutral or high pH ranges.

Using the equilibrium constants calculated with results from the jar tests in PHREEQC give quite accurate results when a system of two anions is simulated. When a system with more than two anions is introduced however, the program values the equilibrium constants too much, resulting in vastly underestimating the adsorption of whichever anion has the lower selectivity in the system.
NOM is currently too complex to accurately represent in a computer model compared to other, more common anions such as chloride, nitrate and sulfate. On top of that PHREEQC needs more parameters beyond exchange equilibria to properly model the exchange reactions in a model.