Rapid mixing in water treatment is used to disperse coagulant into raw water rapidly, followed by flocculation and filtration process. This process has a strong influence on the overall treatment efficiency. Many researchers and operators have studied a number of rapid mixing parameters, including mixing intensity, time, impeller and mixer shape, coagulant dose, pH, and alkalinity. They are important for the whole operation of coagulation and flocculation process. Among these parameters, coagulant dispersion strongly depends on the properties of hydraulic turbulence and the spatial velocity distribution (i.e., velocity gradients from point to point in a mixer). Also, the properties of hydraulic turbulence and the spatial velocity distribution depend on the pattern of energy dissipation, even at the same amount of mechanical energy being supplied. This is inevitably related with other physical parameters (mixing intensity, impeller and mixer shapes, etc.).
Recognizing the significance of hydraulic turbulence in rapid mix and the need to more closely look at it, this study conducted jar-test, CFD simulation and PIV analysis, using three different shapes of jar. They were a circular jar with squared-baffles, a circular jar with no baffle and a Hudson jar. Also, in order to examine the differences in velocity fields in those jars and evaluate whether G value can represent the differences in velocity fields, the author extended the results of jar-test and PIV analysis to discuss about the adequacy of the G value as having been used for the design and operation of the mixer and coagulation-flocculation until now. PIV data are used to calculate local velocity gradients, which cannot be obtained by any other traditional measurement techniques, for comparison with G values and discussion. Based on the previous studies, the author developed a new way of designing jar-test apparatus and procedure with geometric and dynamic similarities to a full-scale rapid mixer to enhance the capability of jar-test procedure used in many of water treatment plants. To evaluate the new design method, the author conducted a wet test with a real full-scale mixer and two jar-test apparatuses, and mathematically simulated them with CFD software. One of the apparatuses is a Hudson jar with a 2 flat-blade paddle, which is being used in the plant. The other is a jar with a 6 flat blade turbine-type impeller, which is newly developed in geometric similarity to the real full-scale mixer.
From the results of the lab-scale jar-test, it was observed that the performance of rapid mixing in the circular no-baffeld jar was better than the other shapes of jar. Also, the shape of jar is found to be a factor affecting the performance of rapid mixer and ultimately the efficiency of coagulation. The results of CFD simulation and PIV analysis confirmed this by showing that since it forms moderate turbulence throughout a jar and minimizes eddy flow, the circular no-baffled jar produced the most proper mixing conditions among them.
From the comparison between the conventional G value and local velocity gradients, it is evident that there are large difference between them, and G value cannot tell us anything about the factual magnitude and distribution of velocity gradients in a rapid mixer.
The wet tests for determining optimum dosage of coagulant with the two different jar-test apparatuses, the Hudson one in no-similarity and the new one in similarity, illustrated that the difference in distribution of turbulence can yield different optimum dosages. The field tests with the two different dosages confirmed that the Hudson one gave poorer results for field application, by underestimating the optimum dosage, than the new one. As a result, this study concluded that the technique of achieving geometric and dynamic similarities, using Froude and Reynolds numbers, works well to baffled mixers and jars and can be extended to reviewing the adequacy of jar-test apparatuses and procedures currently used for other types of full-scale mixers at water treatment plants. If necessary, designing new ones based on the similarities is also recommended for better performance.