Integration And Optimization Of Drinking Water Treatment Processes

Along with more and more serious contamination of raw water and the advance of drinking water quality standards, the demand of treatment processes is becoming more and more stringent. And so, with the constraints of water quality, the integration and optimization of drinking water treatment processes is an important problem in water-supply field.It is necessary to select the optimal treatment process because of the change of raw water quality. Tianjin raw water could represent raw water characters of North-China. In this article, some studies have been finished based on Tianjin raw water, and the main results and conclusions are as follows.Clustering Analysis is introduced to divide Tianjin raw water into different periods for the first time. The results show that Luanhe raw water was divided into 3 periods(including low-temperature raw water, natural raw water, high algae raw water), and Huanghe raw water into 2 periods(including mildly contaminated raw water and badly contaminated raw water). At the same time, characters of each period (level) were analyzed and explained, which provide the theory basis of the optimization of water treatment processes.Based on laboratory-scale jar test, the optimal running parameters of coagulation-sediment units were obtained. The results include: rotation velocity of coagulation n=200rpm, or Gt=11280～12000; rotation velocity of the first and second flocculation units n₁=80rpm and n₂=40rpm; coagulant dose m=8mg/L, settling time t=20min. For economic consideration, the priority-order is m>G₂>G>t>G₁. From a pilot trial of filter bed grading, it is shown that the anthracite-sand filter has slower increase of head loss(3.5cm/h), and it is the optimal filter bed grading of Tianjin raw water.Non-uniformity coefficient (k) of water distribution of Inclined-tube Settling Tank has been put forward and the influences of configuration parameters on k have been discussed as well. The results show that: k will increase sharply when L/B>4 and the height of water distribution area (h₁) should not be lower than 1.3m. The influences of k on sedimentation efficiency (Critical Settling Velocity, CSV) have also been studied. The results show that: CSV increases with L/B; under the conditions of q=10 and L/B≥6, the total force on the sliding flocs(F_total) is nearly equal to zero; the feasible height of water distribution area (h₁) is in range of 1.2～1.6m; the mean CSV increases with inclined tube diameter (d), which need the enhancement of coagulation effect. For the down-sliding flocs, different surface load(q) will need different minimum diameters(d). That is, q=15m/h, min(d)=18mm; q=30m/h, min(d)=65mm. Meanwhile, certain diameter will determine the corresponding maximal loads (q). For example, when J=35mm, the maximal surface load should be 27m/h.According to the four periods of raw water quality(without Luanhe low temperature section during the experiment), Artificial Neural Network (ANN) models of water treatment processes in each period have been set up to predict the effects and water quality of the conventional systems and advanced systems for process evaluation. For ANN models of conventional system, the correlation coefficient of turbidity is bigger than 0.85 and the coefficient of CODMn is bigger than 0.89. As Huanghe raw water is badly contaminated, advanced treatment is needed to improve the treated water quality, and the correlation coefficient of CODMn is bigger than 0.80. The simulated coefficients of ANN are much bigger than the critical coeffient (R_0.1), which indicates the simulation accuracy and prediction stability.Based on the changing raw water, Genetic Algorithms (GA) has been first combined with the ANN models to select the optimal running parameters of water treatment processes. The results show that, with high algae-laden raw water of Luanhe river, CODMn of treated water with optimal running parameters are not bigger than 3.0mg/L; comparing with non-optimal parameters, turbidity is reduced by 0.10～0.16NTU and cost of water product is reduced by 0.017～0.049Yuan/m³; HPAC (High-efficiency Poly-Aluminium Chloride) is the optimal coagulant and PPC, KMnO₄ or O₃ is the optimal pre-oxidant. With badly contaminated raw water of Huanghe river, the optimal process is "Coagulation-DAF-Filter-O₃-BAC"; FeCl₃ is better than HPAC and PPC is the optimal pre-oxidant.Based on the Analytical Hierarchy Process (AHP), the optimal treatment processes are selected from the aspects of economy, management and technology. For Luanhe raw water and lightly contaminated Huanghe raw water, the order of optimal conventional treatment processes is: "Coagulation (HPAC) + Flocculation + DAF + Filter"> "Pre-oxidation (PPC) + Coagulation (HPAC) + Flocculation + DAF + Filter"> "Pre-adsorption (PAC) + Coagulation (FeCl₃) + Flocculation + DAF + Filter". With the badly contaminated Huanghe raw water or worse raw water quality, the order of optimal advanced treatment processes is: "Coagulation Flocculation + DAF + Filter + GAC"> "Pre-oxidation (PPC) + Coagulation (FeCl₃) + Flocculation + DAF + Filter + GAC"> "Coagulation (FeCl₃) + Flocculation + DAF + Filter + O3-BAC". For the enhancement of safe drinking water, it is necessary to select advanced treatment, and the optimal process is "Coagulation + Flocculation+ DAF + Filter + O₃-BAC".At present, the selected optimal treatment process, "Pre-oxidation +Coagulation +Flocculation +DAF +Filter + Disinfect", has been applied to a water plant in Tianjin. The running parameters and some configuration parameters are discussed in this article.