Influence Of Nanofiltration Membrane Fouling Behaviour In Public Water Supply Mechanism Study

Nanofiltration can effectively remove pollutants from water bodies and retain some beneficial elements for human body,and is increasingly practiced in public water supply.However,nanofiltration membrane in the process of drinking water treatment,due to membrane contamination brought about by the membrane flux decline,operating energy consumption increase and membrane life becomes shorter and other problems,limiting the development of nanofiltration process.In order to alleviate the membrane contamination rate,do a good job in membrane contamination prevention and control,and guarantee the long-term stability and high flux operation of nanofiltration process,it is of practical significance to study the influence mechanism of nanofiltration membrane contamination behavior.In this study,a one-year pilot study of nanofiltration membranes in public water supply was carried out using East Taihu Lake water as the raw water.The paper firstly investigated the removal effect of two types of nanofiltration pretreatments,conventional and conventional-ozone/biological activated carbon（O₃/BAC）process,on pollutants in raw water;compared and analyzed the effect of two combined processes,conventional+nanofiltration and conventional+O₃/BAC+nanofiltration,on the removal of pollutants in raw water;then focused on the analysis of the growth law of nanofiltration membrane transmembrane pressure difference（TMP）on the basis of typical nanofiltration membrane contamination model.Finally,a neural network-based TMP prediction model for nanofiltration was established based on experimental water quality data and nanofiltration membrane operating conditions data to further explain the influence mechanism of nanofiltration membrane pollution behavior.From the above studies,the following main conclusions were drawn:（1）During the test period,the raw water quality of Taihu Lake was stable throughout the year,and the turbidity,p H,conductivity and algal cell concentration showed a small seasonal variation pattern.Turbidity remained at a low level throughout the year,with an average turbidity of 2.2 NTU.The concentration of organic matter in raw water also showed some regular changes with the seasons,with a small change in COD_Mnbetween 2.28mg/L and3.94mg/L throughout the year;DOC between 2.19mg/L and 6.80mg/L,with a small increase in DOC in spring;UV₂₅₄between 0.025cm^-1and 0.079cm^-1,with low biological activity of aquatic plants and microorganisms in winter and UV₂₅₄in winter.activity was low,and UV₂₅₄was relatively low in winter.（2）The O₃/BAC process has better effluent quality than the conventional process and can guarantee relatively stable nanofiltration influent water quality.The turbidity of nanofiltration influent can be controlled at a low level,with the turbidity of effluent ranging from0.089-0.155NTU,with an average turbidity of 0.12NTU,while the turbidity of effluent from conventional process ranges from 0.082-0.365,with an average turbidity of 0.173NTU.The conventional-O₃/BAC process can remove dissolved organic pollutants in water better,with the average removal rates of CO The average removal rates of COD_Mn,DOC and UV₂₅₄were75.98%,95.03%and 99.25%,respectively,while the conventional process was not effective in removing organic pollutants in water,with the average removal rates of COD_Mn,DOC and UV₂₅₄organic indexes being 20.65%,12.97%and 27.36%,respectively.（3）Both nanofiltration combined processes can effectively improve water quality,and conventional+O₃/BAC+nanofiltration has better effect on organic removal.The turbidity can be controlled within 0.1NTU by nanofiltration,and the average turbidity of conventional+nanofiltration is 0.050NTU,and the average turbidity of conventional+O₃/BAC+nanofiltration is 0.054NTU.The conventional+O₃/BAC process can enhance the effect of nanofiltration on water quality,and the average removal rate of COD_Mn,DOC and UV₂₅₄organics in raw water by conventional+nanofiltration is 76.03%,83.94%and 86.94%,respectively.The average removal rate of COD_Mn,DOC and UV₂₅₄organic indexes by conventional+nanofiltration was 76.03%,83.94%and 86.44%,respectively,and the average removal rate of COD_Mn,DOC and UV₂₅₄organic indexes by conventional+O₃/BAC+nanofiltration was 86.66%,96.61%and 99.60%,respectively.（4）Analysis of the mechanism of the influence of each index on the nanofiltration TMP growth revealed that the TMP growth of NF90 and NF270 showed a significant negative correlation with temperature,a significant positive correlation with conductivity and TDS,and no significant correlation with the remaining influent water quality indexes.The difference is that the contamination rate of NF90 is faster than that of NF270,and the TMP of NF90 shows explosive growth,while the TMP of NF270 shows a stepwise growth.NF90 has strong salt interception ability,and inorganic ions accumulation will form inorganic scale on the membrane surface,and will also complex with organic matter to form a dense pollution layer,which will lead to the explosive growth of its TMP;while NF270 has a general effect on Ca²⁺/Mg²⁺retention,which can effectively slow down the formation rate of inorganic-organic composite pollution layer;The NF270 membrane is in mesh format,which can change the hydraulic conditions of the nanofiltration membrane module and slow down the microbial contamination rate,making the contamination rate of NF270 much lower than that of NF90 under low temperature in winter.（5）The mathematical model can better fit the trend of NF90 nanofiltration membrane TMP in the rapid growth phase,while the fitting effect for NF270 TMP is insufficient,mainly because the NF270 TMP growth is the result of multiple mechanisms,and the mathematical model cannot analyze the mechanism of its TMP rise from the whole process.Further analysis of the contamination behavior of NF270 by pollutant factors showed that the intermittent decrease of temperature was consistent with the stepwise growth of TMP,and the linear fit r value was-0.8197;the correlation between total hardness and organic matter index and TMP was weak,but the correlation between conductivity and TDS was strong,which was due to the retention effect of NF270 on divalent Ca²⁺/Mg²⁺.In general,and the conventional+O₃/BAC pretreatment process can remove some hydrophobic organics,which slows down the rate of forming a composite contamination layer of organics and Ca²⁺/Mg²⁺on the membrane surface to a greater extent.Therefore,the NF270 has a slower contamination rate and a stepwise TMP growth.（6）In order to effectively solve the problem that mathematical models cannot fit nonlinear parameters and are prone to overfitting,this paper establishes a membrane pollution prediction model based on long and short-term memory neural network（LSTM）based on previous water quality and operating conditions data,and the results show that the prediction model with 1 h as the time interval can accurately predict the TMP growth trend in the next 30hours with root mean square error（RMSE）and correlation coefficient（R²）of 0.0175 and0.9861,respectively.When temperature and conductivity were used as input variables for the LSTM and GRU networks,the prediction model could accurately predict the trend of nanofiltration TMP in the next time period based on the measured water temperature or conductivity data in a certain time period（3 days）with an error RMSE of 0.0046 and 0.0092,respectively;and the prediction model could also predict the trend of nanofiltration TMP in long-term operation TMP data set,also demonstrated good prediction performance.In addition,the neural network model can accurately make sudden water quality warning in time water quality data past changes in the law,the prediction of the peak appears less deviation,the deviation of turbidity is±1.6 NTU,the deviation of water temperature is±0.3℃.