Mathematical Model for Assessing Wort Filtration Performance Based on Granularity Analysis



Jufei Xu and Jiang Kang (1), College of Life Science and Technology, Faculty of Food Science, Xinjiang University, Urumqi, 830046, China; Deliang Wang (1), China National Research Institute of Food and Fermentation Industries, Beijing, 100015, China; Qian Qin and Guohua Liu, Yanjing (Guilin Liquan) Brewery Co., Ltd., Guilin, 541002, China; Zhiping Lin, Beijing Yanjing Brewery Co., Ltd. Beijing, 10091, China; Martin Pavlovic, Slovenian Institute of Hop Research and Brewing, Zalskega tabora 2, SI– 3310 Zalec, Slovenia; and Pavel Dostalek, Department of Biotechnology, Institute of Chemical Technology Prague, Praha 616628, Czech Republic. J. Am. Soc. Brew. Chem. 74(3):191-199, 2016.


(1) Corresponding authors. E-mails: 1468501684@qq.com (J. Kang); wdlpost@163.com (D. Wang)

A laboratory filtration system equipped with a 0.80 µm nitrocellulose membrane for wort filtration and a 0.45 µm mixed cellulose ester mi­croporous membrane for beer filtration was used for this study. The criti­cal granularity distributions of wort and beer before and after filtration were determined by the Beckman Coulter Multisizer 3 counter and parti­cle size analyzer. Using Matlab 7.0 software, a mathematical model of the wort filtration performance and granularity relationship was established, which is applicable for predicting and improving the wort filtration per­formance. The granularity distribution and percentages of particle num­bers and volumes of the wort before and after filtration had great impact on the wort filtration performance. The percentages of particle numbers and particle volumes were used to characterize the wort filtration perfor­mance based on a mathematical model. The correlation between real and predicted wort filtration performance values was very close. R2, the rela­tive error of real value and prediction value on wort filtration perfor­mance, was defined as (predicted value – real value)/real value. When we used the particle number percentage, a relative variation R2 was between 1.67 and 3.01%. When we used the particle volume percentage, the rela­tive variation R2 was between 2.89 and 3.87%. Keywords: Wort filtration performance, Granularity distributions, Matlab 7.0, Mathematical model, Wort filtration, Particle number percent­age