【作者】 范吉；

【导师】 吕树光； Peter A.Vanrolleghem；

【作者基本信息】 华东理工大学 ， 环境工程， 2011， 博士

【摘要】 活性污泥法是目前污水生物处理最主要的方法,其数学模型的建立及工艺优化过程中如何把各种反应过程及机理正确地表达在数学模型中是关键。目前人们对溶解性微生物产物(SMP)的研究越来越重视。SMP产生于污水处理系统中与微生物直接相关的细胞新陈代谢过程中,在许多方面影响着污水处理工艺,如它是构成出水中化学需氧量(COD)的重要组成部分,限定了污水的最低处理极限,影响出水的排放指标等。同时,研究者发现微生物在利用有机质进行生长的过程中并不是如国际水质协会(IWA)提出的活性污泥数学模型3号(ASM3)所描述的先把进水有机物贮存、再利用胞内贮存物进行生长的过程,而是存在着利用基质同时贮存与生长(Simultaneous substrate storage and growth, SSSG)的现象。因此,如何把上述两种机理正确地反映在活性污泥数学模型中,并验证新建模型的有效性是本课题的研究重点。本课题以国际水质协会IWA提出的活性污泥数学模型3号(ASM3)为平台,首先把SMP的概念融入到ASM3模型中,建立了SMP-ASM3模型。其中SMP形成与降解机理分为两部分：一是基质利用相关型产物(Substrate Utilization Associated Products, UAPs),它与基质消耗及微生物生长有关,其产生速率与基质利用率成正比；另一个是生物量相关型产物(Biomass Associated Products, BAPs),它与微生物的内源代谢有关,其产生的速率与微生物的浓度成正比。在该模型中SMP降解过程的动力学是以Monod方程的形式表达的。其次,将微生物利用有机物进行同时贮存与生长的机理融入到SMP-ASM3模型中,建立了SSSG-SMP-ASM3模型。在此新建立的模型中,微生物利用有机物进行同时贮存与生长,即微生物在基质充足时贮存一部分有机物为胞内贮存物,同时,直接利用额外的进水中易生物降解有机物进行生长；在基质不足时,微生物利用胞内贮存物进行生长。在此基础上,对SSSG-SMP-ASM3模型建立的机理进行了探讨,特别是“同时贮存与生长机理”,即认为微生物在基质充足期代谢有机物的过程有三这途径：直接消耗一部分进水有机物进行生长、贮存额外的进水有机物和消耗一部分胞内贮存物进行生长,这三个过程是同时进行的;另外,本研究也对"SMP的降解机理”进行了简化,认为SMP的分子量很大,不能直接透过细胞膜进入微生物内部,因而假设SMP先被胞外酶水解成小分子状的有机物,之后再被微生物代谢。据此,建立了可实用化的ASMP模型。对模型参数进行评估,识别新建模型中灵敏度高的参数,从而为模型的校准提供依据。本研究采用全局灵敏度分析法的一种,即区域灵敏度分析(RSA),来评估参数的随机变化对出水水质浓度的影响幅度。结果表明：对溶解性COD (SCOD)、氨氮(SNH)和硝态氮(SNO)影响最大的参数,即影响系统中SCOD代谢的参数主要有23个,分别是基质贮存比例系数(fSTO)、微生物利用基质直接进行生长的好氧产率系数(Y1,O)、颗粒性缓慢水降解有机物的水解速率常数(kH)、贮存物的好氧产率系数(YSTO,O)、好氧贮存速率系数(kSTO,O)、微生物基于基质生长过程中UAP的缺氧产率(kUAP,NO)、自养菌硝态氮饱和常数(KA,NO)、控制XSTO代谢的常数(K1)、UAP的水解速率常数(kH,UAP)、基于胞内贮存物的好氧最大比生长速率(μH2,O)、微生物利用胞内贮存物生长过程中UAP的缺氧产率(kUSTO,NO)、异养菌缺氧产率系数(Y1,No)、微生物利用胞内贮存物生长过程中UAP的好氧产率(kUSTO,O)、自养菌氧饱和常数(KA,O)、微生物好氧内源代谢过程中BAP的产率(kBAP,O)、基于基质的好氧生长速率(μH,O)、基于基质的缺氧最大比生长速率(μH,NO)、自养菌碱度饱和常数(KA,ALK)、异养菌的缺氧内源呼吸速率(bH,NO)、微生物直接利用进水基质进行生长过程中UAP的好氧产率(kUAP,O)、基于胞内贮存物的缺氧生长速率(μH2,NO)、饱和常数(K2)和饱和常数(KS)；影响系统中SNH的参数主要有21个,分别是fSTO、自养菌最大比生长速率(μA)、YSTO,O、微生物利用胞内贮存物进行生长的好氧产率系数(Y2,O)、Y1,O、kUAP,NO、kUSTO,NO、kH、μH2,O、μH,NO、氧饱和常数(KO)、Y1,NO、硝态氮饱和常数(KNO)、缺氧贮存速率系数(kSTO,NO)、异养菌的好氧内源呼吸速率(bH,o)、自养菌产率(YA)、KA,ALK、微生物缺氧内源代谢过程中BAP的产率(kBAP,NO)、自养菌氨氮饱和常数(KA,NH)、kSTO,O和饱和常数(KNH)；影响出水SNO变化的参数主要有16个,分别是fSTO、kH、μA、KA,ALK、KO、kH,UAP、μH2,O、kSTO,NO、Y1,O、Y1,NO、KA,NH、K1、μH2,NO、自养菌硝态氮饱和常数(KA,NO)、K2和饱和常数(KNo)。模型机理评估及模型参数校准是检验新建模型的有效性和准确性的重要部分。本研究提出了一种新的适用于评估新建模型机理和参数的校准方法,主要由以下几个步骤组成：a、在文献值的基础上对参数进行灵敏度分析,找出显著影响模型预测结果的参数；b、对进水组分进行划分；c、用耗氧呼吸速率(OUR)法来评估模型的机理；d、用实际工艺的数据进一步检验和校准模型；e、用该工艺的另一种不同运行条件下的实验数据验证模型校准的结果,如改变污泥停留时间(SRT)、曝气方式、进水流量等;f、若第e步不成功,则用第e步的工艺校准模型,并用第d步的工艺运行方式验证校准的结果；g、用另外一种活性污泥工艺进一步验证模型校准的结果；h、若验证不成功,重复d、e、f步骤,直到验证成功。根据新提出的校准方案,本研究用从OUR实验中获得的数据对模型进行评估。模拟结果指出ASMP模型比SSSG-SMP-ASM3模型更准确地模拟活性污泥系统中OUR的动态变化(相关系数R分别为0.893和0.848,误差平方和SSE分别为0.013和0.023);ASMP模型比ASM3模型更准确地模拟SCOD的动态变化(R分别为0.981和0.977,SSE分别为79.2和354.9),说明：a、在基质充足时微生物消耗基质的过程有三种途径,即直接利用一部分进水基质进行生长、额外的基质被贮存、微生物利用胞内贮存物进行生长,这三个过程是同时发生的；b、有必要将SMP的形成与降解机理结合到ASM3模型中。同时,用从小试规模的序批式间歇反应器(SBR)中获取的一周期内水质(SCOD、SNH和SNO)的动态变化数据和稳态出水数据(SCOD、UAP、BAP、SNH、SNO、TN和混合液悬浮固体浓度MLSS)校准及模拟三种模型(ASMP、SMP-ASM3和ASM3),结果表明ASMP模型比SMP-ASM3和ASM3模型更准确地模拟SCOD(R分别为0.939、0.876和0.929,SSE分别为737.1、1757.7和8370.3)、SN(R分别为0.992、0.991和0.979,SSE分别为1.72、12.85和12.12)和SNO(R分别为0.992、0.972和0.952,SSE分别为1.19、10.8和11.53)的动态变化,和稳态出水的SCOD(SSE分别为0.25、4和571.2)、UAP(SSE分别为0.01、0.49和—)、BAP(SSE分别为0.16、1.69和—)、SNH(SSE分别为0.01、0.09和0.09)、SNO(SSE分别为0.09、4.41和4)、TN(SSE分别为1.69、8.41和7.84)和MLSS(SSE分别为7395、67081和123000)。另外,用相同进水水质的连续流完全混合式反应器(CSTR)的出水数据检验SBR的校准结果,用ASMP模型对稳态出水SCOD、UAP、BAP、SNH和SNO的模拟结果表明模型获得了良好的校准,其SSE分别为0.04、0.25、0.09、0.01和1.69。在此基础上,用ASMP模型对SBR工艺进行了优化及不同运行条件下的出水水质进行了预测。采用ASMP和ASM3模型分别对上海市某污水处理厂3—6月份的动态出水水质进行了模拟研究,结果表明ASMP模型可以很好地模拟该污水处理厂的出水水质,其实测出水COD、SNH和SNo动态变化值与模拟值的平均偏差分别为0.8(-16.1—12.5)、0.3(-3.6—7.6)和-5.4(-21.7—1.7),优于实测值与ASM3模型模拟值的平均偏差(分别为8.6(-6.9—19.8)、2.0(-4.3—8.5)和-7.0(-28—1.7))；同时,模拟结果也表明出水中SMP占有相当部分比例(约25%),是构成出水中COD的重要组成部分。更多还原

【Abstract】 Activated sludge process is currently one of the main biological methods for wastewater treatment. It is a key point to correctly reflect various reaction processes and mechanisms in modeling establishment and process optimization in simulation procedure. So far, more and more researches are focused on the soluble microbial products (SMPs). SMP which is generated in wastewater treatment process is relative to biomass metabolism. It influences wastewater treatment process in various aspects, for instance, consisting of important part in effluent COD, affecting effluent discharge standard, limiting the lowest treatment threshold, etc. On the other hand, unlike the mechanism of initial storage then growth for substrate described in activated sludge model no.3 (ASM3) proposed by IWA, many researchers found that simultaneous substrate storage and growth happened. Hence, the aim of this project is to propose a new activated sludge model with the above two mechanisms involved and further to calibrate its availability.In this project, firstly, a new proposed SMP-ASM3 model was established with the combination of SMP concept into ASM3. In which SMP was categorized into two parts:one was substrate utilization associated products (UAP) which is related to both substrate utility and biomass growth, and its producing rate is in proportion to substrate utilization rate. The other is biomass associated products (BAP) which is related to biomass respiration, and its producing rate is in proportion to biomass concentration. And the expression of SMP degradation is based on Monod model. Secondly, SMP-ASM3 model was further modified into SSSG-SMP-ASM3 model by combination of the simultaneous substrate storage and growth mechanism into SMP-ASM3. In which storage and growth occur simultaneously in the feast phase, and only after the depletion of the primary substrate the microorganisms would utilize the stored polymers as a carbon and energy source in the famine phase. Then, based on SSSG-SMP-ASM3 model, a further ASMP model was focused on the evaluation of the simultaneous storage and growth process, which indicated that the consumption of substrate by biomass should contain three processes which occurred simultaneously, i.e. substrate storage, biomass ditrect growth on substrate and biomass growth on storage products. In addition, from the practical application point of view, the mechanism of SMP degradation was simplified, i.e. SMP is first hydrolyzed into small molecular weight organic matters before being utilized by biomass. The reason for the assumption was due to the big molecular weight of SMP that can not get through the cell membrane.It is necessary to assess and indentify the high sensitive model parameters in order to provide guidances for model calibration. In this study, regional sensitivity analysis (RSA) was used to evaluate the effects of parameter random disturbances on effluent quality. It was found that the parameters fSTO,Y1,O,kH,YSTO,O,kSTO,O,kUAP,NO,KA,NO,K1,kHUAP,μH2,O,kUSTO,NO, Y1,NO,kSTO,NO,KA,O,kBAP,O,μH,O,μH,NO,KA,ALK,bH,NO,kUAP,O,μH2,NO and K2 affected evidently on effluent COD; the parametersμA,μH1,μH2,fSTO, kSTO,kSTOU and kSTOB contributed more to oxygen profile;the parameters fSTO,μA, YSTO,O, Y2,O, Y1,O,kUAP,NO,kUSTO,NO, kH,μH2,O,μH,NO,KO, Y1,NO,KNO,kSTO,NO,bH,O,YA, KA,ALK,kBAP,NO,KA,NH and kSTO,O affected evidently on effluent SNH and the parameters fSTO, kH,μA,KA,ALK, KO,kHUAP,μH2,O,kSTO,NO, Y1,O, Y1,NO,KA,NH, K1,μH2,NO,KA,NO and K2 took more responsibility for SNO concentration.It is an important part in model evaluation and calibration procedure to assure the model’s validation and accuracy in its application. In this study, a very applicable method was proposed for the new proposed model calibration which mainly consists of the following eight steps:a. identify the sensitive parameters; b. classify the influent components using experiments according to the model requirements; c. evaluate the model mechanism with OUR tests; d. calibrate and assess the proposed model with activated sludge processes; e. change the operation parameters and compare the experimental data with the simulated results by the calibrated model; f, if there is a obvious deviation, use the data in step e to recalibrate the model, and use the data in step d to check the model simulation; g. validate the calibrated model with another kind of activated sludge process with the same influent characteristic; h. if there is a obvious deviation in step g, repeat steps d-f, till the validation is successful.According to the new proposed protocol, the evaluations of models with data from OUR tests have been carried out. The results indicated that ASMP model was more accurate than SSSG-SMP-ASM3 in simulating OUR dynamic variation (the correlation coefficient R were 0.893 and 0.848, and the sum of squared errors (SSE) were 0.013 and 0.023) respectively; ASMP model was more accurate than ASM3 in predicting SCOD dynamic variation (R were 0.981 and 0.977, and SSE were 79.2 and 354.9 respectively), which means:a) there are three processes occurred simultaneously for biomass growth, i.e. substrate storage, growth direct on substrate and growth on storage products; b) it is necessary to add SMP mechanism into ASM3. Meanwhile, models (ASMP, SMP-ASM3 and ASM3) have also been calibrated and evaluated with data from a lab-scale sequential batch reactor (SBR). The simulations of SCOD, SNH and SNO dynamic variations during one cycle of SBR and effluent SCOD, UAP, BAP, SNH, SNO, TN and mixed liquid suspended solid (MLSS) were focused on. The results indicated that ASMP was more accurate than SMP-ASM3 and ASM3 in predicting the dynamic variations of SCOD (R were 0.939,0.876 and 0.929; SSE were 737.1,1757.7 and 8370.3 respectively), SNH (R were 0.992,0.991 and 0.979; SSE were 1.72,12.85 and 12.12 respectively) and SNO (R were 0.992,0.972 and 0.952; SSE were 1.19,10.8 and 11.53 respectively), and in predicting of effluent SCOD (SSE were 0.25,4.0 and 571.2), UAP (SSE were 0.01,0.49 and—), BAP (SSE were 0.16,1.69 and—), SNH (SSE were 0.01,0.09 and 0.09), SNO (SSE were 0.09,4.41 and 4), TN (SSE were 1.69,8.41 and 7.84) and MLSS (SSE were 7395,67081 and 123000). In addition, the effluent data (SCOD, UAP, BAP, SNH and SNO) from a lab-scale completely stirred tank reactor (CSTR) with same influent characteristic were used to validate the calibrated ASMP model by SBR data, and the comparative results between measured and simulated values demonstrated the well calibration of the model as the SSE were 0.04,0.25,0.09,0.01 and 1.69 respectively. Based upon above results, the optimization of SBR process was conducted with ASMP model and the predicted effluent qualities under various operational conditions were investigated.Eventually, ASMP and ASM3 models were applied to simulate the dynamic variations of effluent from March to June of one Wastewater Treatment Plant (WWTP) in Shanghai. The simulated results suggested that ASMP was more accurate than ASM3 in predicting effluent COD,SNH and SNO, as the average deviations between measured and simulated values by ASMP were 0.8 (-16.1-12.5),0.3 (-3.6-7.6) and-5.4 (-21.7-1.7) less than that by ASM3 which were 8.6 (-6.9-19.8),2.0 (-4.3-8.5) and-7.0 (-28-1.7), respectively. It also indicated that SMP consisted to part of effluent COD (about 25%) and be an important component in effluent COD.更多还原