(Soft) sensor for coagulation step in treatment

The increasing volume of collected data, developments in data science techniques and growing computing power present great opportunities to improve water treatment processes. By setting up ‘digital twins’, a virtual treatment can run in tandem with the physical treatment. This offers numerous benefits: treatment processes can be better monitored, the consequences of changes in water quality become apparent, and operations can be controlled directly on the basis of process parameters. This results in a better (or more stable) drinking water quality. The use of digital twins can also reduce the use of chemicals and make treatment scenario studies possible. In the present project we are investigating the possibility of using (soft) sensors to regulate the coagulation step in the treatment.

Model-based approach

To make a digital twin of a complete treatment, each treatment process needs to be incorporated into a model or (soft) sensor. By combining different signals from physical sensors with each other, and possibly supplementing them with models, a (soft) sensor is capable of determining parameters that are not directly measurable. Several initiatives in this area already exist at the water utilities. For example, Vitens applies a digital twin for rapid filtration, and models are also being developed for various treatment processes within the Joint Research Programme of KWR and the water utilities. The (soft) sensors/models make the picture of water quality more complete, allowing for the creation of better measurement programmes. This boost in water-quality knowledge places water utilities in the position to set precise treatment objectives in order to meet the water-quality requirements.

Coagulation process

The coagulation process is the first element to be tackled within DPWE. In this process, the addition of iron chloride (flocculant) and sodium hydroxide (pH correction) leads to the removal of turbidity and a portion of the DOC.

We begin with an inventory of the key treatment objectives for the coagulation process, which associated water-quality parameters are measured, which variations occur, and how the dosage of chemicals is determined (which treatment objective the water utilities have in mind). We will also identify the sensors and models available for coagulation at drinking water utilities, as well as the necessary parameters (input parameters, control parameters).

This will be followed by experimental or model-based research. The experimental research will look for connections between water-quality sensors (e.g., zeta potential, biomass) and the functioning of the coagulation. The results will provide a basis for the control of the coagulation process. In the model-based research a phreeqC model will be set up for coagulation. This describes the chemistry of the coagulation process. Based on the influent water quality, the floc formation is modelled using ‘population balances’ and calculated with effluent water quality. The removal of DOC (fractions) is also added to this (based on kinetics and adsorption isotherms) as well as possible complexation with Fe (III).

Required water-quality parameters and sensors

Coagulation involves the use of large amounts of chemicals every year. Fixed settings are often used in treatment practice. By monitoring and controlling the process, the water quality can be improved and/or the use of chemicals reduced.