Analysis And Modeling Of Organic Matter Removal Using ICR Database Of Drinking Water

The newly added organic matter indicator of drinking water standard promulgated in 2006 brought a great challenge to our country’s drinking water treatment process. There is always excess organic matter exsiting in surface water, most of which is natural organic compound, not only causing bad biological stability, but also producing harmful disinfection by-products. Therefore, the study of the removal of organic matter in drinking water has important practical value and theoretical significance for the protection of drinking water quality and safety. In this paper, organic matter removal with different raw water quality under conventional process and advanced process with biological activated carbon(BAC process) were studied through statistical analysis, model building, model analysis, using the ICR database of the United States, with practical suggestions for the operation of water plant process put forward.Through the analysis of organic matter removal of the ICR database: in terms of the median, the TOC removal rate of raw water TOC>2mg/L was 32.9%, TOC removal rate of conventional process of 31.8%, TOC removal rate of BAC process was 39.6%. Under conventional process, raw water with high TOC and low alkalinity tended to have better TOC removal, and hydrophobic organic matter could be selectively remove. The removal of organic matter by Al coagulant was more susceptible to alkalinity, thus Fe coagulant was recommended for raw water with high alkalinity. The TOC removal rate of O3-BAC and general BAC process were 42.6% and 38.8%, respectively, and this gap was more remarkable for high high TOC concentration raw water. The removal of TOC by BAC process was superior to conventional process for raw water of low TOC and high alkalinity. TOC removal rate of three processes: conventional Fe process, conventional Al process, BAC process were predicted with multiple regression analysis model and genetic algorithm optimized BP network model, performance analysis showing the well prediction, which genetic optimized BP model performed better.Based on the established model, first of all, the optimal combination of input parameters to predict the TOC removal rate using BP neural network was obtained: the conventional process was total organic carbon(TOC), alkalinity and coagulant dosage, the BAC process for TOC, alkalinity, temperature and coagulant dosage. Secondly using optimal parameter combination reversely and the USEPA organic matter removal standards, optimal Fe, Al coagulant dosage per TOC of conventional process with raw water under different TOC and alkalinity were calculated. Comparing Fe, Al coagulants, the effects of temperature and alkalinity on Al coagulant were more remarkable than Fe coagulant. Comparing conventional process and BAC process, the effect of temperature on the BAC process is more remarkable, while the alkalinity and UV254 are more notable for the conventional process. Finally, the TOC removal rate of conventional process and BAC process were quantitatively contrasted using the multiple regression model. For raw water with low alkalinity, high TOC or raw water with high SUVA, conventional process was more suitable for the removal of organic matter, while for raw water with low TOC, high alkalinity or raw water with high SUVA, process was more suitable for the removal of organic matter.