CAFÉ computer aided food processes (KB-15-002-004)

Beer microfiltration is a complex process, mainly due to the fact that the unfiltered beer contains many type of particles. This results in multiple, interacting fouling modes. Insight in the complexity of the process has been achieved by the two main building blocks of this project. First, the development of a knowledge representation (ontology) and data management system resulting in clear terminology and in unambiguous formulation of the relation between many different aspects of the beer filtration process. Also, the data management system makes it possible to couple data to this knowledge and to manage experimental data in a transparent and safe way. Secondly, the realisation of a model-based optimization strategy showed how the process can be improved. The model is a state-of-the-art physical model. Use of such a model results in more insight in the fouling processes and provides means to actually link different raw beer characteristics to parameter values. The resulting control strategy can be applied on the entire filtration sequence: it provides the continuous process parameter settings and the scheduling of the cleaning events (mix-integer problem). Until know, beer microfiltration operates with continuous settings for the cross flow and permeate flux. The project has shown that on optimal policy with varying cross flow and permeate flux results in a cost reduction of 10-20%.

Given the similarity of beer with biotechnological broths and micro-algae suspensions, which all can be viewed as a mixture of cells, macromolecules and colloids, the developed control strategy can be transferred to these other applications. We are currently seeking research opportunities for that.