Results and Discussion of Initial Experiments

Since there were problems with the DO probes we used in the initial experimental setup, we switched to measuring the total volume of the bubbles that form inside the filter column. Using MathCad, we used the bubble volume collected to calculate the equivalent DO concentration removed from the water.

The only changes in the setup were to remove the DO probes and instead feed the water leaving the filter column into an inverted graduated cylinder. This cylinder was filled with water and had its mouth submerged in water. As bubbles formed in the filter media, they flowed out into this cylinder and floated to the top, displacing some water and causing the water level inside the cylinder to fall. The air volume was measured every 10 minutes during each run, using the calibrations on the side of the cylinder.

The first experiment run using this bubble collection method used glass beads as the filter media, with a flow rate of 200 ml/min and an unsuspended filter depth of 32 cm. We observed that the performance of the system increased for about 20 minutes, after which the rate of the increase in air volume became relatively constant. This is illustrated in Figure 1, below. After 20 minutes, the line becomes nearly linear. We concluded that the experiment needs run for at least 20 minutes in order for the data to become steady and reliable, and that we would start recording data after at least 20 minutes of runtime.

Next, we experimented with the sand depth. At a flow rate of 200ml/min, we obtained the following results:

Figure 2 shows that a sand depth of 10cm is better than the larger sand depth of 32cm. Logically, this result did not make sense, so we repeated the experiment at a later date. Figures 3 and 4 below illustrate these results, which show that a larger sand depth extracted more DO.

Unable to render embedded object: File (DO removed vs. time, fig 3.png) not found.
Figure 3 DO concentration versus time at varying depths and flow rates.

The DO stripped from the water flowing at 150 mL/min through the 33 cm filter was much greater than at the other flow rates and depths. This was probably because the 150mL/min run was at a greater filter depth, which meant we needed to add glass beads to the column. A lot of air came in with the dry beads, and we most likely did not allow sufficient runtime afterwards in order to allow all the trapped air to escape before we began recording the data. By the time we ran that depth at the lower flow rate, the extra air had left and we were gathering only the air being stripped from the water.

Figure 4, below, shows the same results as Figure 3, omitting the 150 mL/min at 33 cm data.

Unable to render embedded object: File (DO removed vs. time, fig 4.png) not found.
Figure 4 DO concentration versus time at varying depths and flow rates This graph is the same as Figure 3, but the results from the flow rate of 150 ml/min at 33 cm filter depth are excluded.

Figure 4 illustrates the results from this run more clearly than Figure 3, since the extremely high values of the run at 150 mL/min, 33 cm depth are omitted. Clearly, the greater sand depth resulted in the removal of more dissolved oxygen than the lower depth. The flow rate was not varied enough to have much impact on the effectiveness of this method.

We compared the DO removal according to the DO probes and to our calculations using the total collected bubble volume.

Unable to render embedded object: File (DO removed, measured and calculated, 10cm.png) not found.
Figure 5 DO removed, measured by the DO probes as well as by calculating equivalent DO using the collected bubble volume. Sand depth of 10 cm, flow rate of 200 mL/min.

Figure 6 DO concentration removed, measured by the DO probes as well as by calculating equivalent DO using the collected bubble volume. Sand depth of 33 cm, flow rate of 200 mL/min.

Figures 5 and 6 demonstrate the DO concentrations from the DO probe were completely different from the DO concentrations calculated from the air volume. The blue lines represented DO concentrations calculated from the bubble/air volume. The red lines represented the outflow DO concentrations from the DO probe. Figures 6, 7, and 9 showed the DO concentrations calculated from the air volume (blue lines) reached a plateau after 10 minutes. The blue lines in Figures 6, 7, and 9 leveled off like the corresponding DO concentration measured by the DO probe (red lines) in Figures 6, 7, and 9. Figures 6, 7, and 9 have a large distance between the DO concentration from DO probe reading and the DO concentration calculated from the air volume. In Figures 6 and 7, the distance between the red and blue lines was 7 mg/L. For Figure 9, the distance between the red and blue lines was 5 mg/L. This distance was probably due to the initial DO concentration in the water that in not incorporated into the DO concentration calculated from the air volume. Unlike Figures 6, 7, and 9, Figure 8 had the blue line (DO concentration calculated from the air volume) higher than the red line (DO concentration from the DO probe reading). The DO concentration calculated from the air volume looked unreasonable high. Furthermore, the blue line (DO concentration calculated from the air volume) in Figure 8 did not reach a plateau. Instead of reaching a plateau, the blue line in Figure 8 peaked at 10 minutes and decreased rapidly to a lower concentration. This indicated that the high DO concentration was probably caused by the unsettle disturbances in the filter from addition glass beads added to achieve a sand depth of 33 cm.

The data collected in the Table 1 displayed relatively in the same range except for Run 3 which could be partially visualized in Figure 8. The air volume collected is very different causing the calculated DO concentration to be extremely high. We thought that it was due to addition of the sand to the sand depth that cause this difference and error. After the first two runs, the sand depth was increased from 10 cm to 33 cm. During this change, the filter was open to the atmosphere and the additional sand created disturbance into the settled sand. The addition of the sand may have cause the filter to gather more bubble and we did not allow the system to have enough time to settle out before running the experiment.

From the results collected, there were conflicting data on which sand depth was the optimal. At previous runs, the data showed that the sand depth of 10cm was the optimal. However, in the most recent runs, the data indicated that the sand depth of 33 cm out performed the depth of 10cm. This may be due to how the experiments were run. In the previous runs, the larger sand depth was run first and then the shorter sand depth was run after the longer sand depth. There was a period of rest to vacuum the sand out of the column. However, between the experiments, bubbles remained on the top of sand filter from the previous run. These bubbles settlement may have contributed the error in the data collected. In the most current runs, the short depth was run before the longer sand depth. From the data collect, no concrete conclusion could be drawn and more data was needed.

Conclusion

We have not yet drawn any final conclusions. We would be able to analyze the system better when most of the problems of the system are fixed and it was improved. The current results collected were not promising and stable. This was due to many problems, such as the DO probe not reading the concentration correctly.