Fuzzy Optimum Selection Of Drinking Water Treatment Process

Along with the more and more serious contamination of the sourcewater quality and the advance of the drinking water quality standard, the demand to the treatment process is more stringent. So optimization of the treatment process which can meet the drinking water quality standard is the important problem in water-supply field.The change of the source water quality requires reasonable selection of the treatment process. As the northern city, T city's surface water could represent the northern sourcewater partially. Based on the sourcewater quality and the pilot's running dada, how to select the optimal process is our main point. The change of source water is analysised; and then based on the 6 main components, the study divides the source water into 9 levels by using Kmeans-Clustering, which provides the base for optimal selection of treatment process. The results show that the Kmeans-Clustering is fit for multitudinous swatch's clustering analysis, and the division is impersonality, credibility, which is the important reference to control the running process.To select the optimal treatment process is to select the most reasonable one that could meet the drinking water quality standards. However, the different water quality items have the different weights, which means that the problem is fuzzy. So the fuzzy idea is used to select the different optimal process and different optimal running parameters under different water quality levels, which is based on the trial data. The results show that the sourcewater quality of No.5 and No.6 level are the disadvantage ones, and the complicated treatment process is "Coagulation - DAF - Filtration - Ozonation - BAC". Some treatment unites could be surpassed under better sourcewater quality.The treatment units' running controls are optimized. Above all, the ANN model of Coagulant-DAF System models its reducing efficiency of turbidity and UV254 ideally, with correlation (R1) values ranging fora 0.84 to 0.78. When modeling the Conventional Water Treatment System, ANN model's forecasting precision is ideal compared with the Partial Least Square Regression (PLS) model. All the models are prepared for the next optimal control and selecting the optimum running parameters.Secondly, aiming at the non-uniformity of water distribution when changing horizontal flow tanks into inclined pipe sedimentation tanks, we analysis the problem theoretically on the basis of hydraulic concept. The results show that there are several factors (the ratio {LIB) of tank's length (Z) to width (B), the height of the water distribution area, diameter of the inclined pipes) affecting the non-uniformity (k value); k value is proportional to {LIB) 3, and the non-uniformity becomes severe when LIB is exceeding 4. In the end, some possible suggestion is put forward.