Study On Membrane Flux Prediction And Optimization Of Operating Conditions In Dead-End Microfiltration

In this paper, the influence of different operating conditions on the yeast suspension membrane flux was studied by using homemade experimental equipment for stirring dead-end microfiltration, such as agitation velocity, temperature, operating pressure and suspension concentration. In addition, a linear multi-regression model was used to further study the degree of the influence of these operating conditions on the flux. The results showed that: (1) Agitation velocity(ω), temperature(T) and operating pressure(P) had remarkable influence on the flux when operating pressure was in the range of 0.04MPa~0.10MPa, and the percentages of influencing degree of operating condition were 64.9%, 14.4%, 12.9% and 7.8% respectively; (2) The relationship between permeation flux and different operating conditions was quantificationally expressed by the regression equation, which wasFurthermore, a statistical method (multivariate linear regression model) was used to quantificationally analyze the effects of operating conditions on the efficiency of domestic wastewater treatment and the performance of membrane filtration in a membrane bioreactor. The results showed that: (1) Transmembrane pressure (TMP), activated sludge concentration, dissolved oxygen (DO) concentration, hydraulic retention time (HRT), temperature and pH were all influencing factors of accumulated volume of membrane permeate, while sludge retention time (SRT) had no effect. The mathematic expression obtained by the multivariate linear regression model between accumulated volume of membrane permeate and operating conditions was V = 249 .64TMP-0.056MLSS+31.69DO+55.06pH-5.25T-8.82HRT+16.24; (2) TMP had no effect on the quality of effluent. In terms of chemical oxygen demand (COD), the influencing degrees of activated sludge concentration, DO, temperature, sludge retention time (SRT) and HRT were 5.8%, 2.0%, 63.9%, 13.6% and 14.7%, respectively. The mathematic expression between COD and operating conditions was COD = 0. 0073MLSS-5.64DO+2.86T+0.099SRT+2.48HRT-88.82; (3) For ammonia nitrogen (NH3-N), the influencing degrees of activated sludge concentration, DO, pH, temperature and SRT were 40.5%, 9.6%, 26.7%, 7.5% and 15.7%, respectively. The mathematic expression between NH3-N and operating conditions was NH 3 - N=0.017MLSS-12.99DO-28.86pH+0.90T+0.13SRT+201.30. Moreover, the resistance-in-series model was applied to investigate the fouling characteristics of polyacrylonitrile (PAN) flat-sheet microfiltration membrane in membrane bioreactors. Because of various foulants in membrane bioreactors had different effects on membrane filtration resistances, the contributions of various foulants to different types of filtration resistances were quantitatively analyzed by using a multivariate linear regression method. The results showed that: (1) Foulants mostly existed in the sludge cake layer and gel layer, so the sludge cake resistance and gel layer resistance were main filtration resistances, in which the sludge cake resistance played a predominant role; (2) The gel layer resistance Rg only depended on SMP concentration, and their relationship was Rg = 1 .4695×1011SMP; (3) The contributions of MLSS, EPS and colloid to the pore blocking resistance Rp were 21.7%, 17.6% and 60.7%, respectively, and to the sludge cake resistance Rs were 22.1%, 44.4% and 33.5%, respectively. The quantificational expressions of them were Rp = -1. 4878×107 MLSS+8.7254×108EPS+6.1774×109ColloidandRs = 1 .1309×109 MLSS+8.4635×108EPS+1.2142×1011Colloid, however, SMP had no effects on the pore blocking resistance and the sludge cake resistance.Finally, a cake model that predicts permeation flux in dead-end microfiltration under stirred condition is proposed, and it is established based on a particle mass balance on the membrane surface in terms of Darcy's law. In this model, denudation coefficient K is introduced to quantitatively describe the back-transport of particles from the membrane surface by agitation. Denudation coefficient K has a direct proportion with agitation velocity and suspension concentration, while it has an inverse proportion with transmembrane pressure. The model is validated by experimental data, and the results show good agreement between model prediction and experimental observation, it indicates that agitation can decrease cake layer thickness and consequently improve permeation flux, and the cake model can be used to predict permeation flux during dead-end microfiltraton.