Study On Soy Protein Wastewater Treatment With Ultrafiltration Process

The rapid development of soy protein production in China in the last two decades has produced a great deal of wastewater which contains highly concentrated organic matters and its typical BOD and COD values are in the range of 5000~8000mg l-1 and 18000~20000mg l-1 respectively, of which the main organic components are protein and oligosaccharides. Up to now, the approach carried out has been focused on a biological process, through which the protein and oligosaccharides are removed in the form of pollutants directly. As protein and oligosaccharides are known to be the attractive food additives with much functionality, if they are reclaimed efficiently, the goals of reclamation of resources and purification of wastewater can be obtained, and the environmental and economical benefits can be achieved as well. In this paper, a pilot scale in situ experiment was conducted to study the application of UF membranes in the treatement of soy protein wastewater and the reclamation of soy protein and oligosaccharides from the wastewater tapped from a soybean protein manufacturing process. The performance and operating conditions of cross-flow UF using modified PS membrane was investigated for treating soy protein wastewater. To begin with, pretreatment strategies, including temperature pasteurization at 95℃, adjustment of pH to 4.5, precipitation for 90min and fine filtration with pore diameter of 0.5 μm was employed for treating the raw wastewater, and after that, the concentration of protein was reduced to 0.26% compared to 0.54% in the raw wastewater, and COD reduces from 17749.0mg/L to 11000mg/L, whereas, the remove ratios of oligossacharides and salinity were less than l0%. Secondly, plasma polymerization technology was adopted in modifying PS membrane, by which a hydrophilic group as -COO-Na+ was introduced to the membrane surface. As the modified PS membrane was used to treat soy protein wastewater, its characteristics of fouling resistance were apparently superior to that of PS membrane. Moreover, the flux of the modified PS membrane was higher than the domestic PP membranes in filtrating soy protein wastewater. In comparison with the membranes of PTG10, which was purchased from Millipore Co., the modified PS membranes still showed a little shortage. Thirdly, three modified PS membranes with different MWCOs were used in the experiment according to the molecular weight distribution of solutes in the wastewater. The membrane with MWCO of 10kDa was selected as the optimal membrane for further investigations because it had a protein retaining ratio of more than 90%, an oligosaccharides retaining ratio of less than 10% and relatively higher permeate flux. Moreover, COD of the permeate was similar to that of the membrane with MWCO of 6kDa as 7000mg/L. Finally, orthogonal experimental design was conducted to determine the optimal operating parameters. It showed that a relatively high permeate flux and long operation duration could be obtained under the condition pH 4.5, temperature 42℃, transmembrane pressure 0.2MPa and feeding flux of 1.0L/s. During the process of ultrafiltration, the variation of volume concentration ratio (VCR) was found to be relative with the operating time linearly. The mechanisms of membrane permeate flux decline, such as concentration polarization and gel layer, osmosis pressure function, pore blocking and cake layer, et al. were discussed based on experimental phenomena respectively in the process of soy protein wastewater ultrafiltration operated in a cross flow mode. And the effect of adsorption fouling of solutes on membranes was investigated. A semi-empirical model of permeate flux of UF membrane as a function of operation time (t) or/and volume concentration factor (VCR) was carried out based on the discipline of permeate flux variation in the process of soy protein wastewater UF in which the permeate flux curve was divided in two domains, and the latter one was divided in two periods as well. The three sections corresponded to the initial flux decline, slowly decline period and relative steady period respectively. A program was devised using the software MATLAB for simulating the variation of permeate flux vs. VCR. The results revealed that the simulation of the three periods regarding the permeate flux curve tally with the experimental data closely. From the analysis of simulation results, the reason caused the decline of permeate flux in the domain 1 was not only involved with adsorption fouling, but also concentration polarization and pore blocking; and the ones involved in the afterward periods could be affected by membrane characteristics, membrane nominal molecular weight cut off, and operating conditions of the system.The decline of UF permeate flux in the process could be slowed down by pure water dynamic flushing. The dynamic flushing required to be operated with surface flushing and back flushing simultaneously under the condition of 50℃of temperature, 0.3MPa of pressure, 5 minutes of duration and 15minutes of interval. While the permeate flux was less than the standards as 6L/m2h, chemical cleaning was adopted. It seemed that 0.5% NaOH was a good selection for cleaning UF membrane fouled by soy protein wastewater as that its permeate flux recovery rate could reach 90% and more. Higher concentration of cleaning agents might change the characteristics of membrane. SEM analysis revealed that long-term chemical cleaning could alter the structure as well as the filterability of membrane. However, provided chemical agents were selected properly, severe alteration of membrane characteristics could be avoided. A dual-step ultrafiltration system that combined air blast dynamic ultrafiltration (MWCO, 50kDa) with crossflow ultrafiltration (modified PS membrane with 10kDa of MWCO) was introduced for treating soy protein wastewater. The optimal operating conditions of the air blast vibratory ultrafiltration were determined as 0.1 MPa of air blasting density and 0.05MPa of transmembrane pressure, and the continue filtration duration of the dual-step system could be extended to 22 hours and more, which was almost two times of that of the system including microfiltration and crossflow ultrafiltration. Moreover, the module of air blast ultrafiltration was convenient for cleaning in that the water-permeate-recovery (WFR) of dynamic flushing was about 70% and the efficiency of chemical cleaning was better than that of the module of cross flow UF. Economic analysis indicated that, the investment of the dual-step UF system was about 70% less than that of imported systems, while enormous profits could be obtained by using the system compared with treating soy protein wastewater using biological processes. The variation of permeate flux (J) of the air blast dynamic UF vs. time (t) was simulated using the semi-empirical model, and the results showed that dividing the flux curve in three periods was true of the experimental results. The factors which induced the mechanisms of the decline of the permeate flux in the process of soy protein wastewater filtration with air blast dynamic UF were considered to be adsorption fouling, pore blocking,and cake layer. A dynamic model of permeate flux decline was build in the light of the mechanisms of pore blocking and cake layer. Once the time scale for pore blocking was considered much smaller than that of cake growth, the model was identical with the dual-step equation of the semi-empirical model; otherwise, the model obtained via Taylor approximation was similar with the tri-step equation of the semi-empirical model. The aspects caused the permeate flux decline of the air blast dynamic membrane were analyzed according to the two models, and pore blocking and cake layer were thought to be of controlling ones which could be affected mainly by air blast density and transmembrane pressure.