【作者】 陈琳；

【导师】 田禹；

【作者基本信息】 哈尔滨工业大学 ， 环境工程， 2012， 博士

【摘要】 膜生物反应器（MBR）利用膜分离装置代替传统的二沉池，实现了固液的有效分离，被视为最具发展潜力和应用前景的污水处理新工艺之一。然而膜污染是限制MBR进一步商业化应用的瓶颈问题，如何减缓膜污染问题是目前膜技术研究的热点。本文借助数学模拟手段，解析了关键膜污染物溶解性微生物产物（SMP）在MBR中的生成代谢途径及其组分的膜污染潜能，实现了MBR系统生物处理过程及泥饼层对污染物截留行为的模拟，探明了不同形态活性污泥的胞外聚合物（EPS）-膜体系XDLVO的相互作用能，研发了污泥形态学参数图像分析技术，构建了膜孔污染与膜面泥饼层污染的分析模型，弥补了MBR在数值模拟方面研究的不足。本研究结合序批实验及数学模拟手段分析SMP的生成代谢途径，发现微生物利用底物产生的溶解性产物（UAP）和微生物死亡产生的溶解性产物（BAP）均具有生物可降解性，但其数均分子量（M_n）均高于20kDa而无法被微生物直接利用，在此基础上提出了基于生成-水解-利用途径的SMP生成代谢模型。针对SMP污染问题，利用膜过滤实验分析了BAP和UAP的过滤性能，BAP和UAP的MFI-UF指数分别为4.37×10³和1.63×10⁴s/L²；结合BAP和UAP的过滤模式分别符合泥饼层过滤型（R²=0.988）和完全堵塞型（R²=0.997），表明微生物利用底物产生的UAP更容易引起膜污染。因此，控制MBR适宜的操作条件，降低UAP的生成量，可有效减缓由SMP引起的膜污染问题。在传统ASM3模型基础上引入异养菌同时贮存利用有机底物机制和SMP生成代谢机制，构建了适用于MBR的扩展ASM3-SMP模型，其对MLSS，COD，SMP和S_NO均具有较好的预测能力，相关系数R²值分别为0.83，0.51，0.58和0.74，标准偏差RMSE值分别为134.1mg/L，4.82mg/L，2.23mg/L和1.72mg/L。在此基础上，首次利用扩展傅立叶幅值敏感性测试（EFAST）算法对扩展ASM3-SMP模型进行敏感性和不确定性分析，得出随着SRT的延长输出值受参数主敏感度的影响逐渐增强且模型趋于线性化的结论。此外，依据质量守恒、受力平衡及一阶传递原理，构建了泥饼层形成及污染物在泥饼层-膜过滤过程中的截留模型，避免了采用固定截留系数值带来的模拟偏差，更为真实地模拟了MBR的膜出水水质。在分析正常活性污泥（NS）和丝状菌膨胀污泥（BS）的膜污染特性基础上，研发了污泥絮体形态图像分析系统，构建了形态学参数与污泥指数（DSVI）的自回归历遍模型（ARX），其中AR（或R），EFLI/FAI和FF的组合是最佳输入参数，以[EFLI/FAI,FF, R]为输入参数时，ARX的预测值与实验值间的R_adj²达到0.57-0.86，且延迟时间nk偏离零值。为分析NS与BS膜污染特性的差异，本论文进一步探讨了不同形态活性污泥的EPS-重构膜体系间的XDLVO作用能，EPS-膜系统总界面能U XDLVO123为负值，PVDF膜与BS-EPS的U₁₂₃^XDLVO具有相对较大的负值；此外膜表面的粗糙度极大地改变了EPS-膜体系间的相互作用能，相比于光滑膜面，NS-EPS和BS-EPS的膜临界通量分别降低了27.4%和58.2%，结果表明EPS-膜系统之间会发生自吸附行为，其中BS-EPS更容易引起膜污染。由于膜污染划分为膜孔污染及膜面泥饼层污染，本文分别基于膜面泥饼层对污染物的截留构建膜孔污染模型以及基于质量守恒、受力平衡、塌缩效应构建膜面泥层污染模型。TMP变化的模拟结果与试验结果达到了良好的拟合度，证实了模型对膜污染过程预测的实用性。结合扩展ASM3-SMP模型及膜污染模型，考察系统运行参数对膜污染的影响并进行优化，结果表明通量是影响膜污染速率的最重要因素，水力停留时间的较优控制范围为7-9h，污泥停留时间的较优控制范围为20-40d，曝气量的较优控制范围为1.2-1.8m³/h。更多还原

【Abstract】 Submerged membrane bioreactor （MBR） is considered to be a promising biologicaltreatment technology that uses membrane to replace the conventional sedimentationbasin for the solid-liquid separation of mixed liquor. However, membrane fouling is stilla major problem that hinders MBR’s more widespread application, thus the mitigation ofmembrane fouling has become a challenge of membrane technology. In this work, theproduction-degradation approach of SMP and their fouling potentials were elucidated.Thebilogical process in the MBR and the organic interception by cake layer were wellsimulated. The XDLVO interaciton energy between extracellular polymeric substance（EPS） and membran were investigated. The morphological parameter of sludge withimage analysis was developed, and a fouling model including pore blocking and cakelayer formation was proposed. These compensated the deficiency of MBR research suchas the numerical simulation, enabled in-depth understanding the characteristics ofmembrane fouling, provided valuable information for MBR operation, and promoted thesustainable development of water quality and environment.Characterization and modeling of the SMP were conducted towards a betterunderstanding of the production and degradation of SMP in a MBR. Based on thechemical analysis, both utilization associated products （UAP） and biomass associatedproducts （BAP） exhibited the biodegradability and had a large fraction of molecularweight greater than20kDa, thus the approach of production–hydrolysis-utilization wereproposed for the formation and metabolism of SMP. With respect to the membranefouling, a series of stirred dead-end filtration tests were conducted to investigate thefouling potential of BAP and UAP, showing that the respective mean MFI-UF valueswere estimated to be4.37×103and1.63×104s/L2. Modeling work well indicated that themain fouling mechanisms for BAP and UAP filtration were cake filtration and completeblockage. These results showed that the UAP produced in the cell proliferation phaseexhibited the stronger fouling propensity than BAP. With respect to the biological process in MBR, extended ASM3-SMP model wasproposed based on the platform of Activated Sludge Model3（ASM3） with two mainimprovements:（i） adopting the concept of simultaneous growth and storage of organicsubstrates;（ii） introducing the formation and degradation of soluble microbial products.the model showed an acceptable agreement with MLSS, COD, SMP and S_NO, and thecorrelation coefficients between measured and simulated data were0.83,0.51,0.58and0.74, respectively. A complete global sensitivity and uncertainty analysis of the fullmodel is performed using the variance-based extended FAST method, which showed thatthe model discussed herein generally appeared to be much additive as SRT extended andoutputs were predominantly dependent on first-order effects. Additionally, a mechanisticmodel was developed to demonstrate the formation of dynamic cake layer by sludgelower than critical diameter, and the modeling results showed that the external cake layerof rejected sludge particles played an important role in COD removal before the physicalmembrane filtration. Therefore, the removal of organic substance in the MBR systemwas due to in-series phenomena: biological degradation with biomass, interception in thecake layer and filtration of primary membrane.Comparison of sludge morphology was made to elucidate the different filtrationcharacteristics, and image analysis was successfully developed in monitoring sludgestatus and gaining morphological parameters. The combinations of EFLI/FAI, FFand the floc elongation related parameter （AR or R） were the preferred performinginput candidates for ARX model. The performance of the ARX （[EFLI/FAI, FF, AR]）showed theR_adj²were0.81and0.86（respective for MBR I and MBR II） inidentification;0.65and0.57（respective for MBR II and MBR I） in validation. Mostoften, the inputs gave rise to a delay value, differing from zero, thus improving the intime prediction of the sudden rise in DSVI values.The interaction energy, represented by XDLVO potential, was calculated, allowingexploring the interaction energy profiles for EPS-reconstructed membrane surface. Thefree energy of BS-EPS was lower than that of NS-EPS with respect to PVDF in an aquatic environment, suggesting BS-EPS has higher fouling tendency in terms ofadsorption onto the membrane surface. Additionally, the XDLVO interaction energybetween the EPS and a smooth membrane surface was significantly altered by membranesurface topology （roughness）. The accurate assessment of membrane-EPS interactionsallowed identification of critical flux for effective membrane process operation, resultingin more than0.27and0.58decrement of critical flux for NS-EPS and BS-EPS with theincorporation of roughness effect.A mathematial model was developed to model membrane fouling in the MBR. Themodel for pore fouling was based on the organic interception by cake layer, and the massbalance, force balance and collapse effect was considered in the model of cake layerformation. The results of the simulation compared fairly well with the experimentalresults that were obtained with lab-scaled MBRs, which validated the applicability of thefouling model. The effects of operational variables on the membrane fouling wereconducted with the combination of extended ASM3-SMP and membrane fouling model.The simulation results demonstrated that filtration flux was the most significant factorcausing fouling problems. Based on the simulation results for this study, the hydraulicretention time had an optimum range of7-9h, solid retention time of20-40d andaeration rate of1.2-1.8m³/h from the fouling control and economic point of view.更多还原