(534c) Modeling Vibratory Nanofiltration System for Coffee Extract Preconcentration
AIChE Annual Meeting
Tuesday, November 9, 2021 - 3:30pm to 5:00pm
Vibratory membrane processes alleviate the issues on membrane fouling that is advantageous when integrated into food and beverage processes. Its application as a coffee extract preconcentration alternative further opens opportunity for sustainable soluble coffee production that may be adapted in other food and beverage industries. However, the unique dynamic nature of the vibratory membrane system challenges conventional approaches for understanding and predicting the mechanisms of the process. This process gap limits the overall transferability of the technology to broader industry sectors. This study demonstrates two alternative methods to model the vibratory membrane performance based on important operating parameters like feed concentration (Co), transmembrane pressure (TMP), and module vibratory frequency (F) and displacement (d). Membrane screening and parametric studies were conducted to establish the membrane type and operating conditions of the vibratory membrane process. In one approach, a semi-empirical resistance-in-series model was investigated to correlate flux enhancement and rejection efficiencies based on theoretical membrane surface concentrations and fouling resistances. Vibrations thinned the boundary layer that increased the mass transfer coefficient by a factor of 3.5; and reduced concentration polarization effects by 60% compared with crossflow (CF) operation. These reductions enhanced permeate flux by about 2 to 3 times that of CF operation, with low flux decline. Osmotic pressure resistances were found dominant under low feed concentrations and TMP. However, concentration polarization resistances tend to exceed osmotic pressure resistances under high feed concentration and TMP operation. Real rejections relative to membranes surface chemical oxygen demand were used to predict theoretical permeate conditions that were in reasonable agreement with experimental data. In the second approach, statistical modeling was employed using response surface methodology in conjunction with a Box-Behnken experimental design to determine the effects and interactions of the operating parameters on the performance of the vibratory NF operation. Mathematical models were determined from multivariate regression analysis on permeate flux, permeate quality, and rejection efficiencies. These correlations were used to determine optimum conditions and experimental verification in preconcentrating coffee extracts showed reasonable agreement in terms of vibratory membrane performance. Overall, the models developed in this study are not only useful in managing membrane fouling in vibratory systems, but also in optimizing and developing alternative approaches on similar lines and their scale-up to promote other industrial applications.