The Fluoride Auto subteam seeks to develop a sustainable, inexpensive fluoride removal system for implementation in upcoming AguaClara plants in India. Using the apparatus developed in previous semesters, the team continued running experiments testing how various concentrations of PACl affect fluoride removal. The team then analyzed the variability of the Langmuir isotherm generated by the summer 2018 Fluoride team. This analysis factored into the overarching goal of developing a model to predict an optimal coagulant dosage given both influent and target effluent fluoride concentrations.
With 85.0% of its drinking water sourced from groundwater, India is the largest user of groundwater in the world. In stark contrast to tap water sources in the United States that are supplemented with fluoride, India’s groundwater sources often display excess levels of fluoride due to rocks in aquifers that leak large amounts of fluoride into the water. Consequently, villagers who rely on these well sources for water are at a high risk of overexposure to fluoride. The prevalence of dental fluorosis, an indicator of excessive fluoride concentrations, differs across India, but has been shown to range from 13-91 percent depending on the age group in question and the water source supplying the state or municipality (Arlappa et al., 2013).
In accordance with AguaClara's mission to create affordable, reliable, and sustainable water treatment solutions, the goal of the subteam is to treat groundwater that contains excessive fluoride concentrations. Teams from previous semesters analyzed the efficiency of fluoride removal by passing a coagulant, polyaluminum chloride (PACl), and a solution of fluoride through a sand filter. However, the sand filter was an inefficient method of removal because of the buildup of headloss from particles deposited in the sand filter that eventually prevented water from flowing through (Dao et el., 2015). Thus, instead of using a traditional sand filter, the team researched a similar relationship between PACl and fluoride via a floc blanket reactor. Flocs are aggregates of particles that are created through collision and precipitate out into the water. In the floc blanket reactor, flocs of PACl and clay adsorbed the fluoride from the influent water. The flocs then overflowed into a floc weir as the floc blanket grew and the purified water flowed out the top of the reactor. This reactor was modeled after the floc blanket, floc weir, and plate settlers setup in the sedimentation tank of a typical AguaClara water treatment plant. The team expected that the floc blanket reactor would be able to remove fluoride with a significantly higher efficiency than the sand filter from previous semesters and could run for extended periods of time due to the absence of headloss buildup in the reactor. The flocs in the floc blanket could exit the floc weir while the fluoride in the sand filter would build up and saturate the sand in a short amount of time (Longo et al., 2016) (Cheng et al., 2016).
The main goal of Fluoride Auto subteam thus far has been to analyze fluoride removal by researching and constructing Langmuir isotherms. Langmuir isotherms are used to explain adsorption mechanisms. They are based upon the assumptions that there are a fixed number of adsorption sites and each site can hold one adsorbed molecule. The team has fit a Langmuir isotherm to previous data obtained in Fall 2017 and Summer 2018 by Fluoride subteams. After discovering issues with various data points, the Langmuir isotherm plots were constructed with the removal of some data points. Once the Langmuir isotherm is verified, the information can be applied to other dissolved species removal teams in AguaClara.
Over-consumption of fluoride can lead to arthritis, dental fluorosis, crippling fluorosis, bone deformation and ligament calcification (Roholm, 1937). Fluoride can cause irritation through inhalation, digestion, and touch, and can cause damage to both eyes and exposed skin (NJ Department of Health, 2010). Though there isn’t an established average level of fluoride in India, literature suggests that fluoride levels are seldom above 5
According to the National Research Council (NRC), the maximum contaminant level (MCL) of fluoride in drinking water is 4
One common type of water treatment consists of a series of coagulation, flocculation, and clarification. During coagulation, raw water is mixed with a positively charged coagulant (typically an aluminum salt or iron salt), altering or destabilizing any negatively charged particles or dissolved and colloidal contaminants (EPA, 2016). Depending on the dose of coagulant, there are two methods of particle destabilization. The first, charge neutralization, occurs with a lower coagulant dose and happens as the negative colloids are attracted to the positively charged coagulant particles. The second method, sweep flocculation, requires a high coagulant dose and transpires when the contaminants are caught by precipitates as they settle in the suspension (EPA, 2016). The destabilized particles then proceed through flocculation, where additional mixing increases the rate of particle collision, forming larger precipitates. Following the formation of flocs, clarification removes the agglomerated particles through sedimentation or other removal processes (EPA, 2016).
In recent years, polymerized forms of aluminum salts have been used increasingly to replace standard aluminum salt coagulants (Ingallinella and Pacini, 2001). Polyaluminum chloride, a partially hydrolyzed aluminum salt, is one of the most widely used, as it delivers results similar to aluminum sulfate (alum) coupled with a polyelectrolyte (Ingallinella and Pacini, 2001). The main advantages of using polyaluminum chloride instead of alum include a reduction in sulfates added to treated water, lower sludge production, reduced odor problems, and higher overall removal efficiency (Gebbie, 2001). In the Daylesford Water Filtration Plant, a dose of 45
Many sources have shown that PACl seems to be the optimal coagulant for the purposes of removing fluoride from influent water. (Zonoozi et al, 2009). PACl has been and continues to be used as the most common form of dye removal at the industrial scale. Despite its cost, a main reason as to why PACl is so popular as opposed to other coagulants is that PACl does not cause any leeching of hazardous chemicals into the water from dye decomposition (Golob et al, 2005). A drawback of PACl is that once its coagulation capabilities are saturated, the dye coagulation drops drastically while other coagulants are able to reach a maximum capacity and plateau. The team has been using PACl as the coagulant since the start of the project. Therefore the reasons listed above the team will continue to use PACl as its choice of coagulant for experiments. A future experiment may be to optimize choice of coagulant in order to maximize fluoride removal, but other more important parameters must be optimized before approaching this issue.
Several techniques currently use PACl to reduce high fluoride levels. The Nalgonda technique is a popular fluoride removal method that involves a combination of rapid mixing, flocculation, sedimentation, filtration and disinfection. However, fluoride removal is usually done through co-precipitation (Bailey and Fawell, 2004), which has traditionally been done using aluminum sulfate, but more recent experiments have proven that PACl can be an effective substitute (Kumbhar and Salkar, 2014). The Nalgonda technique typically utilizes a "batch filtration" method, where large quantities of water are treated in buckets. This technique does not utilize continuous flow, and requires a series of treatments to obtain decontaminated water for extended periods of time. For this reason, the Nalgonda technique has been largely introduced as a household treatment method, and has been introduced to various Indian villages, including those in Nalgonda and in the state of Telangana. It is also currently being studied at the pilot scale in Kenya, Senegal and Tanzania (Dahi et al., 1996). In addition to the restrictions implied by batch treatment, the Nalgonda method requires a high dosage of aluminum sulfate to aggregate with fluoride and precipitate. A study conducted by (Dahi et al., 1996) suggests that 13
In regards to other filtration methods, a study by Inganiella achieved 33.3% removal of fluoride using a combination of a gravel pre-filter and a sand rapid filter to capture granules of fluoride, PACl, NaClO and
Floc blankets develop when vertical flow sedimentation tanks form a fluidized bed of particles. An example is shown in the figure below. A floc blanket then facilitates particle removal by “increasing particle-particle interactions that lead to flocculation and filtration occurring in the floc blanket” (Hurst, 2010).
Figure 1: Floc blanket forming in the reactor
The process of forming flocs requires both the precipitation of aluminum hydroxide from the coagulant and the contact with raw water colloidal particles (Hurst, 2010). Once the combination of precipitation and mixing forms small particles, these new flocs collide to form larger, more porous flocs that can then be used for clarification (Hurst, 2010).
This floc blanket clarification is considered hindered settling, which is a form of sedimentation (Gregory et al., 1996). Sedimentation processes are characterized by the removal of suspended particles, e.g. flocs, sand, and clay, from water. Removal is possible due to the differences in density between water and the suspended particles, but is also dependent on the size of the suspended particles, water temperature, turbulence, stability of flow, bottom scour, and flocculation (Sun, 2004). Floc blanket clarification, however, is primarily driven by upflow velocity of the water and by floc concentration, while the other parameters show a weaker correlation with the creation of the floc blanket. (Gregory et al., 1996). The relationship between upflow velocity, concentration and water quality can be combined into the mass rate of settling, which is equal to the product of upflow velocity and concentration. Within a range of mass fluxes, a distinct interface is established between clear water and the floc blanket and thus one can deduce appropriate values of velocity and floc concentration. At concentrations above that ideal mass flux range, the aggregation of flocs becomes thick enough that compression settling occurs, which refers to when over a quarter of the volume of the coagulant will enter the floc blanket within an hour. At concentrations below that appropriate range, flocs are not inhibited by other particles and a suspension with different settling velocities is formed (Gregory et al., 1996).
Examples of floc blanket clarification are shown throughout the world as further proof of its potential in water purification. In Taiwan, a process of pre-sedimentation, floc blanket clarification, and sand filtration is used to reduce 100 NTU water down to potable levels (Lin et al., 2004). Floc blankets have also been used extensively to purify water of algae, protozoa and specific virus strains (LeChavallier and Au, 2004). Therefore, it is believed that the adaptability of this method in conjunction with the use of PACl would allow for effective fluoride removal.
In the Spring of 2016, the team determined that the sand filter system was inefficient and decided to move towards the idea of a single floc blanket reactor (Longo, 2016). Although the sand filter provided cheaper, more adequate removal of fluoride per milligram of PACl used, a key issue that arose with the sand filter was the system run time. The sand filter became saturated with PACl and fluoride too quickly and the head loss built up rapidly. In a matter of a couple of hours, the floc blanket was completely saturated to the point where it was no longer efficient or providing adequate removal of fluoride. Consequently, the system had to be backwashed too frequently to be an effective process (about every couple hours). On a much larger scale, such as a full size plant, the level of maintenance required to backwash each sand filter would be infeasible. In order to address this, the team fabricated a new reactor as seen in Figure 2 mirroring that of the floc blanket and plate settlers in the current AguaClara plants. The team set up a new apparatus fit with stock tanks, a reactor, a turbidimeter, a flocculator, and stock and waste pumps, referencing research previously conducted on the relationship between the amount of coagulant added and head loss accumulation (Dao, 2015).
Figure 2: Pump Powered Setup
The Spring 2016 team then developed a MathCad file to calculate flow rates of pumps from a given set of parameters including upflow velocity, tubing diameters, and reactor concentrations. The team also created a ProCoDA method file to turn the flow rates into RPMs for the pumps so that the process of changing PACl and fluoride concentrations within the reactor was more user-friendly. Transitioning the MathCad file completely to Python is still underway, but might not be necessary if our gravity powered setup is successful.
In the Fall of 2016, the team fabricated a new bottom insert to prevent the accumulation of flocs that clogged the bottom of the reactor. The newly fabricated geometry with a smooth sloped bottom as shown in Figure 3 allowed for gradual flow expansion and the recirculation of the flocs that would have settled to the bottom of the reactor with the old bottom geometry. The formation of gels during experiments in Fall 2018 has many potential causes, one of which was thought to be the bottom geometry. Further research into that is currently underway.
Figure 3: Bottom insert geometry that prevented clogging of the reactor
The team also determined the minimum length of the reactor needed to save resources and space. A shorter reactor, with a height of 5 cm below the weir, was tested with concentrations of 25
Over the summer of 2017, tests performed during the previous semester were repeated in order to finally determine whether the second reactor in series made enough significant improvements to be implemented on a larger scale. The fluoride team in the summer of 2017 stopped adding clay to the system and sought to determine the optimal PACl concentration necessary for high fluoride removal rates (Longo, 2017). Initial tests showed that the reactors would fail within 10 hours due to sludge buildup regardless of PACl concentration, so the team switched to a new reactor designed by the summer 2017 High Rate Sedimentation team. The new reactor increased the time to failure and allowed for higher upflow velocities. After running various tests, the team determined an upflow velocity of 1.5
Figure 4: Langmuir Model Comparison to Experimental Data
After creating the adsorption model, the team observed a significant difference in the fluoride probe readings for tap water versus the readings in deionized water. The findings suggested that the amount of fluoride ions in tap water is actually higher than the readings from the fluoride probe state. Therefore, the team decided the adsorption model was incorrect and should be remade taking this discrepancy into consideration.
Experiments were conducted with a new fluoride probe in the Spring of 2018, which measured the effluent concentrations of fluoride after treatment with different concentrations of PACl. (When a new probe was ordered, the difference in voltage readings between tap and DI water were negligible; therefore, it was concluded that the discrepancy last semester may have been caused by a probe issue) The results are summarized in the table below.
Table 1: A summary of the experimental results using the fluoride probe . Initial fluoride concentrations, as well as PACl concentrations were varied to pinpoint an optimal coagulant concentration for fluoride removal.
The goal of both the Fall 2017 and Summer 2018 team was to measure different effluent concentrations of fluoride with the probe for different concentrations of PACl to construct an adsorption model. This model would be used to predict the PACl dosage needed for a desired effluent fluoride concentration and would also give an adsorption density value for the mass of F- adsorbed, or W. Since the desired fluoride level set by the WHO is 1.5
At a certain maximum PACl dosage, the Summer 2018 team found that the reactor would clog due to insufficient shear to break down the PACl flocs, thereby causing the bed to fluidize. If the adsorption model calculates an excessive PACl dosage, a second reactor would be necessary to decrease the max concentration of PACl necessary to run the system. This will be shown again in experiments run in the Fall of 2018, where gel systems were formed at high PACl concentrations.
However, the Summer 2018 team found that results were inconclusive. For example, influent concentrations of 10
The Summer 2018 team began experiments with the new probe delivered in May. Coagulant dosing was changed based on the isotherm developed during the spring of 2018, which showed that coagulant dosing at 6.25 mg/L will not create a floc blanket. The coagulant dosing ranged from 10 mg/L to 50 mg/L with the influent fluoride concentration ranging from 3 mg/L to 20 mg/L. The team also identified that the influent fluoride concentration did not match the ideal concentration set by the flow rate and stock tank concentration. Therefore, the team began to take a manual reading of the influent fluoride before it reached the flocculator to compare to the effluent fluoride reading. These manual readings were used by the Fall 2018 in the Langmuir isotherm derivation.
The construction of this apparatus was completed last fall and the fabrication of the reactor was completed over the summer. For information on how to run an experiment, refer to the manual at the end of this report.
- Fluoride is pumped from the stock tank and mixed with tap water.
- PACl was pumped from the PACl stock tank and mixed with the diluted fluoride stream
- Mixture was sent to the flocculator to make flocs.
- The PACl-fluoride mixture flowed into the reactor.
- Flocs settled out through the floc weir.
- Fluoride probe was immersed in effluent reactor stream to measure the effluent fluoride concentration.
- The effluent of the tube settler runs through the turbidimeter and then to the sink.
- The floc weir effluent line goes is pumped to the waste line by a waste pump
- Two 600 RPM pumps and one 100 RPM pump
- Transparent 2.54 cm (1") PVC piping
- Flexible and hard 0.635 cm (1/4") tubing and Microbore tubing
- Turbidimeter
- 1,000 ppm Polyaluminum Chloride (PACl) stock solution
- 1,000 ppm Fluoride stock Solution
- Connectors and buckets for stock tanks
- Two stir plates with stir bars for the stock solutions
The initial challenge for the Fluoride Auto Fall 2018 subteam was to analyze the variability of the Summer 2018 and Fall 2017 Langmuir Isotherm. A Langmuir isotherm was developed from the Summer 2018 and Fall 2017 data. Generally, adsorption occurs when gas or liquid particles, in this case, fluoride, adhere to binding sites of an adsorbent, in this case, PACl. This forms a thin film on the surface of the adsorbent as the weak chemical reaction between Fluoride and PACl takes place, as follows:
A Langmuir Isotherm models operates under two assumptions:
- Each binding site on the PACl molecule is chemically equivalent
- Each Fluoride molecule can only adhere to one binding site
The uptake is defined as the amount fluoride adsorbed over the mass of the adsorbent, PACl. This describes the adsorption density, and is defined as follows:
As the concentration of effluent fluoride increases, W decreases because the adsorbent reaches its capacity,
W was calculated for the experimental data from the Summer 2018 and Fall 2017 teams. A nonlinear regression method was used to compare a theoretical Langmuir Isotherm model to the experimental data. Values for
Figure 5. The experimental data, (dots), plotted against the Langmuir isotherm (solid).
One method for reducing variability was by introducing a new bottom geometry into the reactor system. The Summer 2018 team observed that flocs aggregated at the bottom of the reactor. A different geometric piece was added to the reactor entrance to force flocs to flow parallel to each other. Significant numbers of flocs would not accumulate at and flow upwards within the reactor without a singular point for flocs to collide. A CAD file of this design is shown below.
Figure 6. AutoCAD drawing of the Bottom Geometry piece based on fabrication principles proposed by Dr. Weber-Shirk
This design is based on the principle that any constriction of a pipe diameter must be done with a linear flow profile. The two different contractions of the reactor radius were linear profiles at 60 degrees. The geometry inlet converges to a square because of the linear wedges. The inlet tube is a circle, which prevents a perfect fit from the inlet tube to the bottom of the geometry. Water flowed effectively through the geometry. Experimentation with this new bottom geometry resulted in gel formation and thus, experimental failure.
Figure 6. Bottom view of AutoCAD drawing of Bottom Geometry
Gel formation occurs because of the formation of a super saturated solution. Gel formed throughout the entire reactor for the first experiment, which was run with 40 mg/L PACl and 10 mg/L fluoride. There were two hypotheses as to the root cause of the gel formation. These theories were that the concentration of PACl was too high and that the ratio of fluoride to PACl was too high. Previous experiments were run with up to 50 mg/L PACl concentration without gel formation. An aluminum fluoride complex could also form and repel other aluminum particles within the solution. While past teams have conducted experiments at higher fluoride concentrations, few experiments were conducted at concentrations of 10 mg/L or higher. The second theory is consistent with this semester's experiments in that the inclusion of the bottom geometry increases gel formation. Below are two images of gel formation. Dye was injected to visualize the gel location during the second experiment because the gel was not present throughout the entire reactor for this experiment.
Figure 7 and 8. Images of dye injected within the reactor system for gel visualization
The team concluded that gel first formed in the floc hopper and accumulated until it reached the top of the reactor because of the flow of dye throughout the reactor. The full spread of gel is shown in Figure 7 while Figure 8 shows the gel after the reactor was flushed with a high flow rate of water. Gel breaks up within the reactor when the upflow velocity is 8 mm/s. The gel within the floc weir is harder to break because it is controlled by a single speed pump. The team must further understand what parameters reduce the accumulation of PACl and flocs and promote formation of gel to prevent further gel formation in experiments.
Further experimentation was conducted by replacing the fluoride stock with red dye to produce more visible flocs. The first experiment with the red dye showed that flocs formed but also aggregated into a gel at the entrance of the floc weir. This experiment was run with 20 mg/L PACl. The team hypothesized that the floc weir protruded into the sedimentation tube and created a surface for flocs to accumulate and form gel. A new sedimentation tube was fabricated to test this hypothesis. The red dye experiment was replicated to test whether gel would form in this new sedimentation tube with the same experimental conditions. Gel formed in this experiment, however, flocs entered the floc weir without aggregating at the entrance as they had in the previous reactor. The team hypothesized that gel formed when the flocs that were above the floc weir descended down the sedimentation tube and collided with other flocs, which created a super saturated solution necessary for a gel.
The team continued to run tests at low concentration of PACl, or a maximum of 20 mg/L, and high concentrations of red dye given the dilute stock solution of red dye. However, each experiment gelled, and thus, produced no useful data for further validating a Langmuir adsorption model.
The Fall 2018 Fluoride Auto subteam has concluded thus far that the Langmuir isotherm model fits previous experimental data with low variability. The team has also concluded that the current reactor setup leads to uniform gel formation. The new bottom geometry and reactor reduced gel formation; however, gel still formed with the new reactor. Further experimentation and research must be conducted to understand the flocculation and gel formation processes occurring within the team's setup.
The next steps include determining the root cause of gel formation and exploring whether floc blanket formation is necessary for fluoride removal. The new bottom geometry and reactor did not reduce gel formation. The team will also explore re-introducing clay into the system, which was phased out of testing in the Fall 2017 semester. This is because gel formation might be impossible to prevent without clay. Reintroducing clay could also increase floc blanket formation, however, it is still unknown whether a floc blanket is necessary for the system. The Fluoride Auto team could also be remerged with Fluoride gravity given that the principal task for this semester's team, validating the Langmuir Isotherm model, was achieved.
Arlappa, N., Aatif Qureshi, I., and Srinivas, R. (2013). Fluorosis in India: an overview. Int J Res Dev Health, 1(2).
Bailey, K. and Fawell, J. (2004). Fluoride in Drinking-water.
Cheng, M., Longo, A., and Vidal, B. (2016). Fluoride, Fall 2016.
Dahi, E., Mtalo, F., Njau, B., and Bregnhj, H. (1996). Defluoridation using the Nalgonda Technique in Tanzania. In Reaching the Unreached: Challenges for the 21st Century, New Delhi, India.
Dao, K., Desai, P., and Longo, A. (2015). Fluoride, Fall 2015.
Dokko, J. and Espada Fraile, J. (2016). Countercurrent Stacked Floc Blanket Reactor, Fall 2016.
EPA (2016). Water Treatability Database.
Gebbie, P. (2001). Using Polyaluminum Coagulants in Water Treatment.
Golob, V., Vinder, A., and Simoniˇc, M. (2005). Efficiency of the coagulation/flocculation method for the treatment of dyebath effluents. Dyes and pigments, 67(2):93–97.
Gregory, R., Head, R. J. M., and Graham, N. J. D. (1996). The Relevance of Blanket Solids Concentration in Understanding the Performance of Floc Blanket Clarifiers in Water Treatment. Chemical Water and Wastewater Treament, IV.
Hurst, M. W. (2010). Evaluation of Parameters Affecting Steady-State Floc Blanket Performance. Degree of Master of Science, Cornell University, Ithaca, New York.
Ingallinella, A. M. and Pacini, V. A. (2001). Simultaneous removal of arsenic and fluoride from groundwater by coagulation-adsorption with polyaluminum chloride. Journal of Environmental Science and Health, Part A.
Kokko, J. P. and Rector, F. C. (1972). Countercurrent multiplication system without active transport in inner medulla. Kidney international, 2(4):214–223.
Kumbhar, V. S. and Salkar, V. D. (2014). Use of PAC as a Substitute for Alum in Nalgonda Technique. International Journal of Emerging Technology and Advanced Engineering, 4(10).
LeChevallier, M. W. and Au, K.-K. (2004). Water Treatment and Pathogen Control.
Lin, W., Chen, L. C., Chung, H. Y., Wang, C. C., Wu, R. M., Lee, D. J., Huang, C., Juang, R. S., Peng, X. F., and Chang, H.-L. (2004). Treating High Turbidity Water Using Full-Scale Floc Blanket Clarifiers. Journal of Environmental Engineering, 130(12):1481–1487.
Longo, A., Desai, P., and Dao, K. (2016). Fluoride, Spring 2016.
Longo, A., Zhang, V., and Cheng, M. (2017). Fluoride, Fall 2017
NJ Department of Health (2010). Sodium Fluoride.
Roholm, K. (1937). Fluorine Intoxication: A Clinical-Hygienic Study. Copenhagen, Denmark.
Sun, S. F. (2004). Sedimentation. In Physical Chemistry of Macromolecules: Basic Principles and Issues, Second Edition, pages 243–266. John Wiley & Sons, Inc.
Yang, Z. L., Gao, B. Y., Yue, Q. Y., and Wang, Y. (2010). Effect of pH on the coagulation performance of Al-based coagulants and residual aluminum speciation during the treatment of humic acid–kaolin synthetic water. Journal of Hazardous Materials, 178(1):596–603.
Zonoozi, M. H., Moghaddam, M. A., and Arami, M. (2009). Coagulation/flocculation of dye-containing solutions using polyaluminium chloride and alum. water science and technology, 59(7):1343–1351.
The reactor in the system as shown in Figure 8 was developed by the summer 2018 Fluoride Gravity subteam. Details on how to construct it may be found here.
Figure 6: The sedimentation tube used for removal of the fluoride-containing flocs created during flocculation.
The apparatus the team is currently running experiments with is uses a basic flocculator-settling tube design. Coagulant, and water are pumped into the flocculator, and the pumps are controlled by ProCoDA. The Fluoride pump is not controlled by ProCoDA so we manually set the flow rate before each experiment. The pumps all pump through the flocculator and the outlet of the flocculator enters the bottom of the sedimentation tube. Ideally, flocs in the sedimentation tube will recirculate as fluid flows up the tube, and if the floc blanket gets large enough, flocs will fall down the floc weir. The outlet of the sedimentation tank goes into the in-line fluoride concentration bottle, and then flows into the turbidimeter and then into the sink. ProCoDA measures data at constant time intervals for voltage, and turbidimeters.
- Open tap water valve and close the fluoride valve as to not pump fluoride into the system during backwash
- Close the waste line
- Turn on Just Water process in ProCoDA and fill system completely with water.
- Make sure the sample bottle with the fluoride probe does not overflow from the high flow rate of the water
- Continue to run water until turbidimeter reads less than 0.5 NTU and fluoride concentration is less than 0.5
$\mathrm{\frac{mg}{L}}$ . - Record the initial voltage reading to make sure the initial concentration of fluoride in the sample bottle is about 0
$\mathrm{\frac{mg}{L}}$
- Fill stock tanks with appropriate concentration of PACl, and fluoride hours
- Calculate the flow rates of the PACl, water and fluoride pumps from the MathCAD file and run the pumps at the appropriate RPM using ProCoDA
- Run the waste pump at the appropriate RPM as calculated from ProCoDA.
- Calibrate the fluoride probes and record the initial concentration of fluoride in the sample bottle
- Place the fluoride probe into the in-line concentration measurement bottle
- Waste line is open (System will explode if this is not open)
- No leaks anywhere in system
- Pumps are all turned on and running at the correct RPM (Check ProCoDA)
- Pumps are all pumping water in the correct direction (in the direction of the flocculator and reactor)
- Put ProCoDA into the "Run Experiment" mode #####During test
- Recheck everything periodically to ensure it is running how it should be and that there are no water leaks. Make sure liquid in the continuous fluoride measurement bottle doesn't overflow. #####End of test
- Change the process to the OFF state.
- Clean the reactor using the "Cleaning Procedure" after the experiment is completed.
- Put a piece of sponge in the tube between the flocculator and PACl insert.
- Run a high velocity jet through the tube to purge the flocculator of excess clay buildup.
- Drain the reactor through the valves at the bottom.
- Flush water through the system.
- If there is not a noticeable amount of floc blanket buildup, (a) and (b) can be skipped.
- Make the stock calibration concentrations of .1
$\mathrm{\frac{mg}{L}}$ , 1$\mathrm{\frac{mg}{L}}$ , 10$\mathrm{\frac{mg}{L}}$ , and 20$\mathrm{\frac{mg}{L}}$ in small bottles. Individually pipette fluoride stocks into all four bottles, do not use serial dilutions. - Rinse the fluoride probe with DI water and carefully dab the end of the probe on a Kimwipe. If any sediments from prior experiments remain, rub off with polishing
- Insert the probe into one of the calibration solutions.
- Swirl the probe around, then let it settle. Record the voltage once it reaches a steady state
- Make sure to record the voltage at the minimum voltage (the voltage will spike first and eventually reach a steady state voltage before increasing again).
- Repeat with the other fluoride concentrations and record the values in Google Docs (labeled "Fluoride Calibration").
- The R squared value, slope, and y-intercept will be updated as the voltages are updated (make sure the R squared value is at least .99 to ensure accurate fluoride calibrations).
- If R squared value is not 0.99 or higher, rinse the probe and let it settle in 5
$\mathrm{\frac{mg}{L}}$ solution for 5 minutes, then rinse thoroughly with DI water. Repeat procedure.
If the fluoride probe requires replacement, use this link to purchase a new one: https://www.coleparmer.com/i/cole-parmer-ion-selective-electrode-ise-fluoride-double-junction-bnc/2750219?PubID=UX&persist=true&ip=no&gclid=Cj0KCQjw-o_bBRCOARIsAM5NbIPaSgDmo2SsaOPk9Nmz8tdYYPRbkZ7sxqVKuY_9hdBmA0lDkKri_JcaAkr3EALw_wcB
ProCoDA is a process control system that was developed by Monroe Weber-Shirk in order to set process parameters through a computerized system. It can be adjusted to different system states that control the system pumps depending on what flow rates are desired. Additionally, ProCoDA collects the data from probes, allowing for compilation of dye concentration data.
To begin the ProCoDA method file, four states were made: ON and OFF, Just Water and Run Experiment. In the OFF state, all the valves were closed and no pumps were on. In the ON state, all the pumps were ON and all valves were opened, in the Just Water state, only the water valve was open. ProCoDA turned this pump on and off via a normal valve control, so long as the pump was already set to a proper flow rate. In the Run Experiment state, the pump flow rates are updated to proper values and shut off after 4 hours of testing.
The method file was set to control the revolutions per minute (RPM) of the PACl/dye pump and the tap water pumps. This was done using the peristaltic pump ProCoDA file available in the AguaClara server as well as inputs for desired flow rate and tubing size. For the PACl and fluoride pump heads, inputs of
import math as m
import numpy as np
from aide_design.play import*
import aide_design.floc_model as fm
Q=.76*(u.milliliter)/(u.second)
Q.to(u.m*u.m*u.m/u.s)
D=(1/8)*u.inch
D.to(u.m)
A=np.pi*(D**2)/4
print(A)
#Area given our diameter of tubing
v=(Q/A).to(u.m/u.s)
print(v)
#velocity through our tubing given volumetric flow and area
L=9.43*u.m
T = 298 * u.degK
vis = pc.viscosity_kinematic(T)
print(vis)
#kinematic viscosity given the temperature
Gcoil=206.907*(1/u.s)
g=9.81*(u.meter)/(u.second)/(u.second)
hf=((Gcoil**2)*vis*L/v/g).to(u.m)
print(hf)
R_c = 0.05*u.m
shearG = fm.g_coil(Q,D,R_c,T)
print(shearG)
#Shear gradient in flocculator given curvature and volumetric flow
hfModel=((shearG**2)*vis*L/v/g).to(u.m)
print(hfModel)
#headloss due to the flow through the flocculator
deltah2=((v**2)/2/g+hf).to(u.m)
print(deltah2)
#height difference necessary for the velocity the team wants
Qpacl=.00475*(u.milliliter)/(u.second)
Qpacl.to(u.m*u.m*u.m/u.s)
Dpacl=.0005588*(u.m)
#Use microbore inner diameter of 1/50 inches
Apacl=np.pi*Dpacl*Dpacl/4
print(Apacl)
vpacl=Qpacl/Apacl
print(vpacl)
headP=.05*(u.meter)
Lpacl=headP*g*Dpacl*Dpacl/32/vis/vpacl
Lpacl.to(u.m)
print(Lpacl)
#length of microbore tubing necessary for the given headloss