【作者】 鞠兴华；

【导师】 彭党聪；

【作者基本信息】 西安建筑科技大学 ， 环境工程， 2008， 博士

【摘要】 利用数学模型和数值计算对城市污水处理厂生物处理系统进行系统分析、模拟、设计、控制和诊断是目前水污染控制领域研究的热点。本文以国际水协会（IWA）1号模型（ASM1）为对象,首先对其系统构成进行解析,然后进行组分和参数敏感度分析以及模型多参数估计,在此基础上,探讨模型组分和参数的测定方法,最后采用ASM1对四级分段进水A/O生物脱氮系统进行了模拟研究。建立了敏感度方程,采用COST模拟基准,对ASM1模型参数和组分进行了敏感度分析,获得了参数不确定性对系统整体性能影响的定量分析结果:5个化学计量参数中,Y_H对模拟结果的影响最大,敏感度最高;14个动力学参数中,敏感度较大的参数依次为μ_A、b_H和k_h;模型组分中对出水COD、SS、氨氮和总氮影响最大的分别为S_I、X_s、S_NH和X_s。敏感度分析结果为ASMl应用中参数的选择、取值提供了可行的依据。针对ASM1中参数多,且与出水水质关系的复杂性,采用遗传算法对其进行最优估计。确定了遗传算法的目标函数,设计了遗传策略。在COST模拟基准条件下,进行了三参数和四参数估计算例的验证和分析,参数估计平均误差小于5%。与人工估计方法相比,采用遗传算法估计ASM1模型参数,具有较高的可靠性和搜索效率。对ASM1组分及参数的测定方法进行了全面探讨,在此基础上,对西安市北石桥污水处理厂水质特性进行了分析和活性污泥系统ASM1模型参数的测定。结果表明该厂进水溶解态惰性有机物S_I、快速可生物降解有机物S_s、慢速可生物降解有机物X_s和颗粒态惰性有机物X_I分别占污水总COD的比例平均为4.10%、17.60%、51.60%和13.28%;氨氮、溶解态有机氮和颗粒态有机氮占总氮的比例平均为64.94%、7.35%和27.71%。DE氧化沟和四级分段进水A/O生物脱氮工艺的污泥产率系数Y_H别为0.72和0.697;活性污泥混合液COD与MLSS单位转换系数分别为0.808gCOD/gMLSS和0.777gCOD/gMLSS。DE氧化沟异养菌衰减系数b_H为0.612d^-1（25℃）。在上述研究结果的基础上,对四级分段进水A/O生物脱氮工艺试验系统进行了模拟,模拟出水COD与实测结果的平均误差为9.3%,说明本研究提出的组分和参数的测定方法、敏感参数的选择原则、参数的估计方法是合理的。最后采用了遗传算法对四级分段进水A/O生物脱氮工艺进水流量进行了优化分配。以模拟出水水质与实际出水水质的误差最小为目标函数,得到各级进水流量分配比例为37.2%、27.4%、23.2%和12.2%。将此结果应用于实际污水处理试验中,可以得到较高的污染物去除效率。通过对活性污泥系统模型及模拟的理论分析和应用研究,为深入理解活性污泥工艺的动态行为和规律,优化污水处理系统设计和运行方案,提高经济社会效益,提供了理论基础和技术手段。更多还原

【Abstract】 Mathematical modeling and simulation are valuable tools in various aspects of the wastewater treatment practice,such as optimization of the plant operation,planning and design of new facilities,or development of new treatment processes.Based on the activated sludge model No.1（ASM1）developed by International Water Association,the experimental methods for characterizing wastewater composition and parameters of ASM1 were firstly determined.Then the sensitivity analysis of components and parameters of ASM1 were discussed and parameters estimation of ASM1 by the genetic algorithm has been extensively investigated.Finally,the activated sludge process model was introduced to simulate carbon and nitrogen removal in a 4-stages step-feed biological nutrient removal pilot plant.In order to determine the key parameters and components of ASM1,sensitivity analysis was adopted based on COST wastewater treatment plant benchmark.The results showed that the most sensitive one is heterotrophic yield Y_H among 5 stoichiometric parameters.At the same time three other parameters,b_H,μ_A and k_h have high sensitivity in 14 kinetic parameters. The soluble inert organic matter S_I is sensitive to effluent COD,slowly biodegradable substrate X_S is sensitive to suspended solids in effluent,ammonia S_NHand Slowly biodegradable substrate X_S are sensitive to ammonia and total nitrogen in effluent.Genetic algorithm adopts the least absolute difference between experimental results as the fitness function.To confirm the applicability of the method,parameters were estimated using the proposed methods under the COST simulation benchmark.Simultaneous estimation of three or four parameters is made and error is less than 5%.The genetic algorithm can accomplish the complicated parameter estimation of the activated Sludge model perfectly and effectively in contrast manual parameter estimation.The characteristics of municipal wastewater in xi’an Beishiqiao sewage treatment plant were analyzed with proposed methods.Total COD consists of 4.1%of S_I,17.6%of S_S,51.6% of X_S,and 13.28%of X_I,the total nitrogen consists of 64.94%of S_NH,7.35%of S_ND,and 27.71%of X_ND.The average heterotrophic yield coefficient Y_H is 0.72 in xi’an Beishiqiao sewage treatment plant,and is 0.697 in a four stages pilot plant of step-feed biological nutrient removal by batch OUR respirometric experiment.The relationship between particulate components of the ASM1 and SS in the following clarifier should be transformed by a transformation coefficient f_COD-SS,which is 0.808gCOD/gMLSS for xi’an Beishiqiao sewage treatment plant,and 0.777gCOD/gMLSS for the 4-stage step-feed biological nutrient removal pilot plant.The value of the decay rate of the heterotrophic biomass b_H at 25℃is 0.612d^-1in xi’an Beishiqiao sewage treatment plant.In the 4-stage step-feed biological nutrient removal pilot plant,model calibration with estimated parameters and experimental data showed a good agreement within 9.3%error range.Using the genetic algorithm,it is possible to estimate active sludge model parameters efficiently.Application of the genetic algorithms to optimize influent flow distribution in a 4-stage pilot plant of step-feed biological nutrient removal system was discussed.The objective function of optimization was designed to minimize the difference between estimated and required effluent concentrations.The optimized parameter for influent distribution ratios are 37.2%,27.4%,23.2%and 12.2%respectively.Good performance was achieved with the optimized influent distribution ratios.In conclusion,activated sludge process modeling and simulation offer an important theoretical fundamental and a convenient approach to get more detailed understandings to the dynamic behaviors and characteristics of activated sludge system.It is an effective tool to research on the more reasonable and optimal process control strategies.It may hopefully promote the design and management of wastewater treatment system up to a new scientific and technological level in order to reach more strict environmental objectives more economically.更多还原