Preparation of cellulose tri-acetate membranes for high-alcohol beverages via genetic algorithms and high throughput experimentation

Abstract

In fermented beverages, volatile esters are trace compounds with a significant effect on the flavour profile. Some of the most important esters in beer are ethyl acetate (solvent-like aroma) and isoamyl acetate (banana aroma). These esters have a defined threshold which, when exceeded, will be experienced as unpleasant upon tasting. Therefore, the industry needs to selectively remove and fractionate esters by means of adsorption and tune the level of esters in the different process streams.Nanofiltration (NF) and Reverse Osmosis (RO) membranes are known to have a high potential towards removal of low molecular weight components from liquid streams. However, membrane preparation depends on several parameters with complex interactions to obtain the targeted membrane performance. Multi-parameter optimization strategies are thus very useful to reduce material consumption, minimize time and labour during the development of a ‘bottom-up’ preparation method. High throughput (HT) filtration techniques in combination with genetic algorithms (GAs) are known to be very powerful to direct such membrane preparation.This optimization strategy is used to obtain cellulose tri-acetate (CTA) membranes for filtration of high-alcoholic beverages towards ethyl acetate and isoamyl acetate depletion. Before the implementation of the proposed strategy, a successful optimization of the HT operating conditions was setup together with a selection of benchmark membranes by screening the performance of various commercially available nanofiltration (NF) and reverse osmosis (RO) membranes.Next, the non-compositional (solvent impregnation, membrane thickness, solvent evaporation time, annealing time and temperature) and compositional parameters (polymer concentration, solvents and additives selection) used during non-solvent induced phase separation (NIPS) preparation of CTA membranes were determined. Finally, CTA membranes were optimized using GAs.