Development of a membrane process for the concentration of fava bean extract for plant-based dairy applications

Introduction. Fava beans are an increasingly popular and viable crop, and an excellent nutritious choice for novel plant-based food products. Fava beans are industrially underexploited as a food ingredient and are today mainly used as animal feed [1]. However, fava beans are an interesting base material for producing plant-based milk. While soy milk dominates the legume-based beverage segment, there is an opportunity to diversify and broaden this market by incorporating fava beans that offer better functionality and nutrition than soy beans [2, 3]. There are very few studies reported in the scientific literature regarding membranes in the plant-based dairy industry, which opens the possibility of a whole new research field. Common processing steps for plant-based beverages (such as high-shear mixing and ultrahigh temperature treatment) can create shear and thermal effects, which can lead to changes in the structure of the proteins, and subsequently affect the stability of the protein emulsions [4]. Ultrafiltration is well establishedfor the cow milk based dairy industries, but also has great potential in producing legume-based products. Thus, ultrafiltration was investigated as a suitable solution for plant-based dairies.Experimental/methodology. A series of different experiments were performed to study the performance of membranes. Initially, a membrane screening of 5 commercial ultrafiltration membranes with low molecular weight cut-offs from Alfa Laval, Microdyn Nadir and Synder were assessed in an Alfa Laval LabStak M20 unit. The membrane screening was performed in concentration mode at a temperature of50°C, a transmembrane pressure of 2 bar, and at a volume flow of 4 L/min. After this, a parameter study and fouling and cleaning assessment of the two most promising membranes was conducted in an Alfa Laval LabStak M10 membrane unit. To identify the optimal operation parameters, at 50°C, the transmembrane pressure was varied between 0.1 and 2.5 bar, the crossflow velocity was changed between 0.1 and 1.0 m/s. Thereafter, the fouling degree and cleaning capacity were determined throughpure water flux measurements, where both alkaline, enzymatic and acid cleaning agents were assessed. Finally, a pilot scale experiment with the most promising membrane candidate was performed in an Alfa Laval M39 Combi unit (3.8” spiral wound), first as parameter study and then as a concentration study. The transmembrane pressure was varied between 0.5-2.5 bar and the flow changed between 0.6-1.2 m3/h. Results and discussion. From the membrane screening, GR90 (Alfa Laval) and UH005 (Microdyn Nadir) were selected as promising candidates. Both achieved good fluxes (14.8 and 12.8 Lm-2h-1, respectively) combined with high protein retentions (79% and 80% total nitrogen retention, respectively). The parameter study showed that a crossflow velocity of 1 m/s and a transmembrane pressure around 1 bar were optimal operational set-points. Although severe fouling occurred on the membrane surfaces, it was possible to recover the pure water flux through a cleaning protocol combining alkaline and enzymatic cleaning agents. When scaling up the process to a 3.8” spiral wound module of the GR90 membrane, a cross flow of 1.2 m3/h and 0.7 bar was applied to concentrate the fava bean base with a volumetric concentration factor of 3.0. The flux varied between 1-6 Lm-2h-1 during the concentration. Overall, this study showed the possibility of ultrafiltration as a suitable unit operation in the future plant-based dairy industries. Acknowledgments.The Green Dairy AB, Sweden, are greatly acknowledged for the collaboration and supply of the fava bean base.References. [1] Crépon, K.; Marget, P.; Peyronnet, C.; Carrouée, B.; Arese, P.; Duc, G. Field Crops Research, 2010, 115 (3), 329-339.[2] Nawaz, M. A.; Singh, T. K.; Stockmann, R.; Jegasothy, H.; Buckow, R. Foods, 2021, 10 (6).[3] Nawaz, M. A.; Tan, M.; Øiseth, S.; Buckow, R. Food Reviews International, 2022, 38 (5), 1064-1102.[4] Qamar, S.; Bhandari, B.; Prakash, S. Food Research International 2019, 116, 1374-1385.