Filtration is a widely common unit operation in solid-liquid separation and is used in processes across all industries. Filter media play an essential role in the process. Therefore, selecting the right filter media for a specific separation task is an essential step in the design of filtration processes. Due to the large number of filter fabrics available on the market, this is a complex task and is often based on experience. In addition, competing properties such as separation efficiency and flow rate make the selection difficult. For example, the pore size of a filter fabric must be as large as possible for high flow rates, but as fine as necessary for the desired degree of separation. Possible consequences of suboptimal mesh selection are lower filtrate flow rates, clogging of the meshes, and thus increased energy consumption of the filtration process. This, in turn, is directly related to an increase in resource requirements and an increase in particle loading of the filtrate. This can lead to impairment or even failure of following unit operations in the plant network.

Therefore, the aim of this work is to provide a predictive tool to support the selection of the filter mesh. Based on different characterization parameters, a model is developed using machine learning methods. This model allows a computer-aided selection of a suitable filter mesh for a given separation task. The characterization parameters investigated include the separation efficiency in terms of the degree of separation and the flow rate of different filter media. Meshes with different weave types and different geometric pore sizes were considered. In addition, the...