Spring 2009 Turbidity Profiles

Methods

• Hot-swapping Baffle Configurations

Two adjustable baffle configurations were built this semester. Each consists of two 6' rows of baffles, one row for each of the first two channels of the flocculator. While one configuration (2 rows) is in the flocculator and being tested, the other configuration (which is not in use) can have its baffle spacing altered. Because that is possible, instead of needing to wait between experiments while baffle spacing is readjusted, the new baffle spacing can be ready on the second baffle configuration and the two configurations can be simply swapped. Then, while the next experiment is running, the first baffle configuration can have its spacing altered to be ready for the subsequent experiment. Since each configuration also has different length spacers (and new spacer of other lengths can be readily fabricated), flocculator arrangements where one channel has a certain baffle spacing while the other channel has a different baffle spacing, echoing the current plant design, can be tested as well. In previous tests, many data samples needed to be collected in one sitting to avoid having large fluctuations in turbidity, which required experimenting and determining new alum dosages, and often resulted in data that could not be compared between days of testing. The new adjustable baffle system will do much to ameliorate this problem by allowing the pilot plant team to perform experiments faster, and thus perform more experiments in a day.

• PAC dosing

Polyaluminum chloride (PAC) is the coagulant of choice of the Cornell Water Treatment Plant as well as many other plants across the United States. According to plant workers PAC is much more forgiving than alum in terms of dosing and forms flocs better in colder water, which is important when testing in Ithaca. Dosing with PAC was set up at the Pilot Plant. Our goal was to collect turbidity profiles of the tank based upon energy dissipation (the spacing of the baffles). To facilitate this goal and remove a large source of potential error, we eliminated alum dosing in favor of dosing the same proportion of PAC that the treatment plant is dosing (the plant doses PAC in ppm of the total flow). This should make the profiles more comparable and the effects of baffle spacing more clear, since temperature and non-optimal dosing effects have been greatly reduced by adopting PAC.

• Running configurations

With each different baffle configuration, spacers were placed on the PVC pipes to adjust distance between baffles. While one experiment was running in the floccuator, a spare baffle system was being adjusted to avoid wasting time. Two different flow rates, 50L/min and 100L/min, were used with each baffle configuration.

Tube settlers were placed at three different, relatively evenly spaced locations throughout the flocculator, while another turbidimeter measured influent turbidity. Tube settlers had to be filled with water before being connected to the pump to pulling air through the pump into the turbidimeters instead of water from the location where the settler was located. The turbidimeters were cleaned with each change in baffle configuration with distilled water and wipes. The configuration was then allowed to run for several residence times so flow could develop fully and the PAC dosage would be distributed to the system. After this turbidity would be recorded by Process Controller into an Excel spreadsheet file from all of the turbidimeters every 5 seconds. This data, after removing outliers, was used to plot turbidity profiles of the mean turbidity for each settler location in the flocculator as well as the statistical standard deviations above and below the mean.

Calculations

{float:left}
h6. Equation for energy dissipation:
!energydissform.JPG!
{float}

{float:left}
h6. Equation for flocculator residence time:
!thetafloc.JPG!
{float}

Length of flocculator channel = 1.8288 m
K _baffle = 4
Π _cell = 4
w = .305 m
h = .764 m

Turbidity Profiles

Samples of the Excel files, containing turbidity data for each of the turbidimeters used to create the turbidity profiles can be downloaded via the links for Configurations 4, 6, and 8.

Results

...

Wiki Markup
{float:right|border=2px solid black|width=600px} [!Turbidity Trends as Function of Energy Diss.png|width=600px!|^Turbidity Trends as Function of Energy Diss.png] h5. Turbidity Trends as Function of Energy Dissipation. {float}

We were able to eliminate very small baffle spacings (and thus very high energy dissipation rates) from the range of efficacy for turbidity removal. As can be seen from the graph, more research needs to be done within the lower range of dissipation rates as there was great variance of the data. Some of the best configurations were Configuration 6 at 50 L/min, which had .051 m spacing in both channels for an energy dissipation rate of 1.507 mW/kg. This configuration provided a 61% reduction in turbidity to a turbidity of about 0.7 NTU. Configuration 8 at 50 L/min had .051 m spacing in first channel (ε = 1.507 mW/kg) and a .102 m spacing in second channel (ε = .094 mW/kg). Effluent turbidity was reading at under 0.7 NTU, an overall turbidity reduction of 59%. The other exceptional configuration was Configuration 4 at 50 L/min, which had .127 m spacing in both channels for an energy dissipation rate of .039 mW/kg. Turbidity was reduced 65% to an effluent turbidity of about 0.8 NTU.

Areas of Future Study

Challenges and Areas of Research for Fall 2009

Conclusions

Dosing with PAC and the new baffle system made it much easier to test a wide array of configurations in a shorter period of time than was possible before. Through the testing of these configurations we were able to eliminate very high energy dissipation rates as being effective at reducing turbidity, but also saw a large range of rates (between 0.03 mW/kg to 1.5 mW/kg) that worked well. It is hard to determine if these rates work well for all influent turbidity however. The problem with these tests was that influent turbidity was often times coming in at a turbidity level well under 5 NTU, which would likely not even be dosed in Honduras. With turbidity this low it is also hard to gauge how effective the configuration would be at higher turbidities. We recommend finding a way to work with higher influent turbidity in future tests, even if that means using a stock tank of water with clay mixed in to keep turbidity levels higher instead of water coming from Fall Creek. Using water other than Fall Creek water will necessitate monitoring of PAC dosage, since the same dosage that the treatment plant is using could not be used if the pilot plant were running off of a different water source. It was evident however that using PAC caused a reduction in turbidity for almost all configurations and turbidity was able to be lowered at the pilot plant to levels that had previously not been reached when using alum. For the first time, effluent turbidity in the pilot plant was seen below 1 NTU, even reaching at low as 0.6-0.7 NTU.