Alum Dosing Theory
Alum, or aluminum hydroxide, is added to raw water in order to coagulate the particles suspended in the water. Before alum is added, the particles have a slight negative surface charge and therefore repel one another. Dosing the water with alum neutralizes the charge on these particles so they are more likely to stick together. The particles are allowed to grow in a flocculator, where the dosed water is thoroughly mixed in order to disperse the alum and promote the flocculation of the particles. These conglomerates are referred to as flocs.
Following the flocculation, the water enters a sedimentation tank, where the flocs begin to settle. This settling creates a "blanket" of flocs, which serves to trap more flocs and provide effective filtration as it grows. The floc blanket is an integral part of the AguaClara technology and enables the system to achieve much lower effluent turbidity.
Underdosing
An underdosed system occurs when the alum dosing is lower relative to the dosing conditions that are considered "ideal." In this situation, the flocs that are formed will be made up of a larger portion of suspended particles, which in our system are clay particles, and a smaller portion of the flocs will be composed of the alum. This results in a smaller floc and, consequently, a more dense floc blanket.
Overdosing
An overdosed system is accomplished by providing an alum dose that is at a higher concentration than what is considered to be the "ideal" dose. In this case, the flocs that are formed will have a higher alum-to-suspended-particle ratio. The flocs are therefore larger and more "fluffy" and the resulting floc blanket is less dense.
Overview of Methods
In these experiments the alum dosage supplied to the flocculation system was varied in order to study how properties of flocs and the floc blanket affect the effluent turbidity produced by the tube settler. The experimental set-up is identical to the one used in Spring 2009, and from our results we hoped to analyze velocity gradient thresholds and possibly investigate how changing influent water chemistry affects the setter's efficiency.
Results and Discussion
Using the Spring 2009 team's process controller methods, we subjected an ideal geometry to non-ideal conditions. Specifically we altered the alum dose to see how different alum doses affected the effluent turbidity. Though the Spring 2009 team had success with a 9.5 mm diameter tube, due to what we think was ineffective air bubble traps in the flocculator, or the addition of a flow accumulator to the method, we experienced failure with this geometry. We achieved an acceptable effluent turbidity (less than 1 NTU) with a 15.1 mm diameter tube that had a length of 30.5 mm. With the good experimental results, we then subjected this tube settler to varying alum dosage to investigate the affect of dosing on tube settler performance. At each alum dosage, the tube settler was tested at a variety of capture velocities and at two different floc blanket levels: the lower level is when the height of the floc blanket falls below the bottom of the tube settler, the high floc blanket level is when the floc blanket height is above that of the bottom of the tube settler.
Experiment 1: Alum Dose = 45 mg/L
Experiment 2: Alum Dose = 35 mg/L
Experiment 3: Alum Dose = 65 mg/L
Experiment 4: Alum Dose = 15 mg/L
Experiment 5: Alum Dose = 105 mg/L
Process Controller Files
Conclusions
Figure 1: Capture Velocity vs. Average Effluent Turbidity shown for each alum dose at low floc blanket level.
Figure 2: Capture Velocity vs. Average Effluent Turbidity shown for each alum dosage at high floc blanket level.
Overall, this system performed well and most of the effluent turbidities were below 1 NTU. The thesis on evaluation on parameters affecting steady-state performance of a floc blanket found that the ideal dosage was 45 mg/L for 100 NTU influent water. The alum doses of 35 mg/L and 65 mg/L performed best. Because the slight underdose and overdose peformed well (average effluent turbidities were under 1 NTU), what was thought to be extreme under and over doses were also tested. We tested 15 mg/L and 105 mg/L to observe how the floc blanket formed under severe non-ideal conditions. The dose of 105 mg/L, despite averaging at less than 1 NTU, resulted in failure since the average effluent turbidity frequently spiked above 1 NTU. The 15 mg/L dose, however, had average effluent turbidity of less than 1 NTU, meaning that cannot be considered a failure.
Originally, an alum dose of 45 mg/L was thought to be an ideal dose, meaning that it was supposed to produce the lowest effluent turbidity. However, this alum dose did not perform as well as was expected. In comparison to the other average effluent turbidities, 45 mg/L should either perform slightly better than an alum dose of 35 mg/L or somewhere between 35 mg/L and 65 mg/L. As shown in the above graphs, 45 mg/L performs worse than all of the other alum doses, including what was supposed to be extreme under and over doses (15 mg/L and 105 mg/L). This should not have happened, possible reasons for these results are air bubbles in the settler tube, other reasons. Experiments at this alum dose should be re-run.
The major cause of failure for an underdose is an incomplete floc blanket as a result of smaller flocs that are formed, but the increased residence time in the flocculator creates larger flocs, which form a floc blanket more quickly and more effectively. Thus, although we expected that the extreme underdose of 15 mg/L would fail, the effluent turbidity fell within the acceptable range. (This paragraph needs to be rewritten. I agree with what you say about the flocculator, but you need to differentiate between underdosing and floc blanket failure. The floc blanket clearly formed and gave performance under 1 NTU, so this was not failure. This was worsened performance compared to a dose of 35 mg/L. Include reasons for why this would be the case.)
In contrast, an alum overdose forms a less dense, more "fluffy" floc blanket, which is not as effective because the flocs are weaker and breaking up in the floc blanket causing spikes in effluent turbidity. The extreme overdose of 105 mg/L shows higher turbidity as a result of these weak flocks in the floc blanket.
Because the effluent turbidity using the alum underdose of 15 mg/L was acceptable once the floc blanket had formed, it seems that the dosage is unimportant once the floc blanket is completely formed, within a certain range of alum doses. It appears that as long as the floc blanket is fully formed, which should occur with a higher dose so that it forms quickly enough, the alum dose can be lowered while still experiencing the same results. However, the alum dose does need to be within a certain range for an effective floc blanket to form.
The water chemistry in our system also contributes greatly to the unexpected results. (were the results unexpected except for 45 mg/L?) The water in the lab is much more alkaline than the water in Honduras. As a result, the pH of the water in Honduras is more sensitive to changes in alum dose. There is an ideal range of pH values where flocculation occurs most effectively, and this range is harder to achieve in Honduras. Thus, the water in the lab allows the system to be more robust and able to achieve acceptable effluent turbidity even with a large range of alum dosages. This observation means that even though the system appears to work successfully regardless of the ranges of alum doses that we tested, the same will most likely not be true in Honduras. A future study could include changing the alkalinity of the water to make the water pH more sensitive to changes in alum dose to confirm the applicability to Honduras.