Modeling and process analytics as a basis for future predicting the dynamic behavior of a lab-scale vacuum belt filter

Real-time analysis and prediction of the process performance is a major challenge, as parameters such as e.g. cake porosity cannot be measured directly during filtration and the correlations between parameters are complex. In case of the example cake porosity the empirical correlation depends on particle size distribution, solids volume concentration, particle shape and pressure difference. As a result, it is calculated indirect through measurements of cake height and weight. Unknown changes to parameters in real systems are therefore only detectable in terms of the filtrate volume flow or filtration time. In this context, time-consuming and cost-intensive large-scale studies on the influence of the respective parameters are necessary. One approach is to introduce real-time analysis methods and a model to describe filtration. This paper presents, on the one hand, the implementation of an optical method using a camera to detect the transition between filtration and desaturation. On the other hand, it presents the detailed introduction of a multi-compartment model to describe filtration and desaturation on a lab-scale vacuum belt filter. Experiments have shown that only the bright-field method for optical detection can deliver reliable results as particles within the suspension scatters the light, too. For the filtration model a sufficiently fine spatial resolution of 7.6 mm (50 compartments) enables a realistic representation of the residence times, while too coarse discretization quickly leads to pronounced axial remixing, which does not occur in this form in real belt filters.