In order to stop global warming, greenhouse gas emissions must be drastically reduced and the removal of already emitted CO₂ from the atmosphere must be pursued in combination with permanent CO₂ storage. In this context, carbon mineralization offers the possibility of producing permanently bound carbonates with calcium and magnesium. One source for the metals are tailings, which are dissolved by acid extraction in form of metal ions. However, the selective extraction of Ca²⁺ and Mg²⁺ ions from the mining waste changes the particle size distribution and the bonds between the particles and thus the filtration properties, depending on the existing material composition.

Another problem is the fluctuating material composition in the feed, which poses a challenge when it comes to the economic efficiency of this process. One possible solution to respond to fluctuating feed conditions is the concept of an autonomous belt filter. However, this requires not only in-situ and online process analysis, but also reduced-order models to describe the filtration properties under fluctuating feed conditions.

In this study, we present a multi-compartment model of the vacuum belt filter. This model is capable of displaying the two zones of cake formation and cake dewatering on a belt filter. This requires continuous measurement technology, such as a camera, which can divide the two zones for calculation purposes. The aim of filtration is to separate as much liquid as possible while specifying a target residual moisture content. To achieve this, the model must predict the filtration time and cake height for a given pressure difference. These two variables can then be used to calculate the required belt speed and feed. Experimental validation demonstrates the validity of our model, and initial variations of the feed suspension are carried out.