gradients as the water passes down through the flocculator reduces floc-
shearing and yields lower settled water turbidities.
8. The performance of buoyant coarse media floccuiation is also dependent upon the depth of the bed. Longer media beds result in higher head losses,
and somewhat improved turbidity removal. Again, bed depth must be within the
limits for head loss accumulation.
9. As was expected, temperature had a significant impact on flocculator performance, and any design or pilot-plant study must take this into account. An 8°C decrease in influent water temperature resulted in a 20% decrease in turbidity removal from the pilot plant influent at the Williams Plant.
10. Air-scouring, even with vary low air flow rates, will effectively reduce
head loss and clean the beds of media. It should be possible to operate a buoyant coarse media flocculator for at least 24 hours between air-scourings.
11. Sludge removal capability may be required after both the flocculator and
the overflow column as significant deposition was observed following passage
through each bed.
Recommendations
1. Additional bench and pilot-scale testing are highly recommended before
constructing any full-scale buoyant coarse media flocculators.
2. Further studies on a full-scale buoyant coarse media flocculator are
needed to address the issue of possible scale effects on performance. The
changes in the degree of wall friction between pilot-scale and full-scale could have an effect on head loss buildup and media bed separations.
3. Further research into optimizing coagulation chemical dosages
specifically for this type of treatment could result in improved performance and possible cost-savings.
4. Future tests should also examine the effect of varying the angle at which
the media bed is tapered to determine an optimal angle. These tests should
incorporate tapering of the upflow clarification bed as well, to determine an optimal angle of tapering for the overall treatment scheme. The possible
benefits of tapered upflow clarification following tapered BCM floccuiation have
5. Long term operation of the buoyant coarse media flocculator on either the pilot-scale or demonstration-scale is needed to determine the lifespan of the media bed. The possibility of media fouling due to sludge deposition or metals precipitation is a concern that should be investigated.
6. An iterative combined process simulation or pilot study would be required to make definitive statements regarding the optimal treatment for a
given water.6 The absence of data regarding filter performance following
buoyant coarse media flocculation makes it difficult to make a meaningful comparison of BCM flocculator performance with other types of treatment. It is not easy to effectively compare single processes within the process train of a treatment plant. Thus, there is a certain degree of uncertainty in the evaluation of buoyant coarse media flocculation as a superior alternative to traditional
References
1) Hudson, H. E. Water Clarification Processes: Practical Design and
Evaluation. Van Nostrand Reinhold Company, New York, 1981.
2) Moffett, J. W. The Chemistrv of Hioh-Rate Water Treatment. Jour. AWWA,
60:1255 (Nov. 1968).
3) Hudson, H. E. Jr. & Wolfner, J. P. Design of Mixing and Flocculation Basins. Jour. AWWA, 59:1257 (Oct. 1967).
4) Vrale, L & Jorden, R. M. Rapid Mixing in Water Treatment. Jour. AWWA,
63:52, (Jan. 1971).
5) TeKippe, R. J. & Ham, R. K. Velocity Gradient Paths in Coagulation. Jour.
AWWA, 63:439 (July 1971).
6) Wiesner, M. R. & Mazounie, P. Raw Water Characteristics and the Selection
of Treatment Configurations for Particle Removal. Jour. AWWA, pp.80-89, (May
1989).
7) Sanks, R.L Water Treatment Plant Design.. Ann Arbor Science, Ann Arbor,
1979.
8) O'Melia, C. R., "Coagulation and Flocculation," in W. J. Weber(ed.)
Phvsiochemical Processes for Water Quality Control. Wilev-lnterscience. New
York (1972).
9) Camp, T. R. Flocculation and Flocculation Basins. ASCE Transactions, 120:1
(1955).
10) Okun, D. & Schuiz, C. Surface Water Treatment for Communities in Developing Countries. Wilev. New York (1984).
11) Andreu-Villegas, R. & Letterman, R. D. Optimizing Fiocculator Power Input. Journal of Environmental Engineering Division ASCE, 102:251 (April 1976).
12) Kerri, K.D. & Beard, J.D. Jar Tests Fine-Tune Your Plant's Performance. Opflow, Vol.13, No. 4 (April 1987).
13) Richter, C. A. & Balkowiski, C. S. Water Treatment Plant for Small Communities. Sanepar. Brazil, 1983.
14) Schroeder, E. D. Water and Wastewater Treatment. McGraw-Hill. New
York, (1977).
15) Ramaley, B.L Lawler, D.F. Wright, W.C. & O'Melia, C.R. Integral Analysis
of Water Plant Performance. Journal of Environmental Engineering Division ASCE, 107:547 (June 1981).
16) James M. Montgomery Consulting Engineers Inc., Water Treatment
Principles & Design. Wiley-lnterscience, New York (1985).
17) Cleasby, J. L Is Velocity Gradient a Valid Turbulent Flocculation
Parameter? Journal of Environmental Engineering Division ASCE, 110:857
(Oct. 1984).
18) Han, M. & Lawler, D.F. The (Relative) Insignificance of G in Flocculation.
Appendices
The complete results from each of the flocculation runs performed during the field testing are presented in this section of the report. To help the reader interpret the raw data presented in this section, the complete results from a representative test (the sixth run at the OWASA Water Treatment Plant) are explained in detail on the following pages.. Every run is identified by a number and by the date on which it was started. The sixth OWASA run was started on
July 18, 1991, and lasted for approximately four days.
The top section of the table includes all of the properties and chemical additions to the feed water for the pilot unit. The turbidity given is the average instantaneous feed-water turbidity sampled from the piping between the pump and the flow meters. The coagulant line lists the type and dose of coagulant used by the plant during that run as recorded by plant operators. Polymer refers to the amount of cationic polymer that is added during the rapid mix. PAC is the amount of powdered activated carbon added to the raw water by the plant operators to improve treatment and reduce taste and odor problems. KMNO4 is the amount of potassium permanganate added to the water. The temperature given is the average temperature of the water entering the pilot unit during that run. Temperature typically varied by no more than one or two degrees Celsius over the course of any given run. Therefore, the average temperature is a very accurate representation for any given run. The flow rate listed in this section refers to the flow rate through the downflow flocculator at the start of the run. Any changes to that flow rate, as well as the flow rate through the overflow column, are listed in the remarks section at the end of the table.
The media section of the table gives the size distribution, type, and bed depth of each media used in a run. The direction of flow is also given, with downflow referring to media used in the flocculation unit, and upflow referring to media used in the overflow column. For example, in this run the first line
identifies the media used as being 3M ceramic media of 1/2-inch to 3/4-inch in diameter. The media formed a 3-foot deep bed, and like all of the runs
performed at the OWASA plant, utilized a tapered media bed in the flocculator. This was followed by a bed of 1-inch diameter hollow polyethylene NORPAK media, which was 4.1 feet deep, and was located in the overflow column as
Indicated by the upflow designation. The overflow column used in all runs was a straight column which was 6 inches by 12 inches in cross section.
The next section of the table contains all of the results that were
measured while sampling during the run. The first column is the sample number, in this run 17 samples numbered 0 through 16 were taken. The second column gives the cumulative run time in minutes until that set of samples was taken. This run totalled 5830 minutes(slightly more than four days) in duration.
The third through fifth columns give the head loss measured in
millimeters at three different depths within the media bed. The column labelled TOP head loss measures the head loss across approximately the top six inches
of ceramic media in the flocculator. This value started at near zero, and
increased to a maximum of 100 mm, before eventually stabilizing at
approximately 60 mm two days into the run. The column labelled MID PT lists the head loss across approximately the top 18 inches of bed. This value was also too small to measure accurately when the bed was clean, but increased to a maximum of 185 mm, before stabilizing at approximately 118 mm two days into the run. The column labelled O'ALL gives the head loss across the entire 3 foot bed of ceramic media in the flocculator. This head loss ranged from 5 mm at the start of the run to a maximum of 255 mm (about 10 inches) before
stabilizing at approximately 190 mm (7.5 inches) two days into the run.
The sixth and seventh columns give the estimated G value in the top and middle sections of the downflow media bed. The root mean-square velocity gradient values could not be measured directly, therefore these values were computed using the head losses measured across the media bed, and the equation for velocity gradients in media beds as given earlier. The average G value in the top section of the flocculator increased steadily to a maximum of
220 sec""! and finally stabilized at approximately 175 sec'l. In the middle
section of the bed, the average G value reached a maximum of 84 sec'' and
stabilized at approximately 70 sec''. From these two columns, we can clearly
see that the velocity gradient is tapered from high to low as the water passes
through the flocculator.
The six remaining columns contain the turbidity measurements taken during the run. The column labeled "effluent" (under the "flocculator" heading) represents instantaneous turbidity samples taken from below the media bed in
variation in the magnitude of this sample was due to floe accumulation at the sampling point in the bottom of the flocculator. The "settled" flocculator sample in the next column is the sample in the previous column after 20 minutes of settling. This value ranged from 0.8 to 2.8 NTU. There was less variability in the settled sample because the floes that accumulate at the sample port and cause the large changes in the instantaneous effluent turbidity are typically also
large enough to settle within 20 minutes.
The columns under "overflow turbidity" give the instantaneous and settled water turbidity at the top of the overflow column. This is the water
leaving the pilot plant after going through both treatment units. The value of the instantaneous turbidity ranged from 0.5 NTU to 2.5 NTU, while the settled water turbidity never exceeded 1.0 NTU. It should be noted that in most instances, if the instantaneous turbidity was below 1.0 NTU, the sample would not be
settled.
The next to last column, labeled "feed settled", is the turbidity of the pilot- plant influent after being flocculated using the standard jar test and then being settled for 20 minutes. The settled feed water samples were typically in the range of 0.4 NTU to 0.7 NTU, with the exception of one measurement which was 1.4 NTU. The purpose of this column is to insure that the influent to the pilot plant is properly coagulated, and is capable of being readily flocculated. The final column of data is simply a measurement of the instantaneous turbidity of the influent to the pilot plant. This value ranged from 6.5 NTU to 12.5 NTU. Some of that variability could be due to scouring of turbidity from the walls of the piping feeding the flocculator as the flow through the pipe was slightly higher
while this sample was being taken.
The final section of the data sheets is the remarks section. This section is
where any changes in operating conditions, or interesting observations made during the run, are listed. It will also typically contain the turbidities of the full scale flocculation trains at the treatment plant as well. In addition to changes in plant hydraulics, the remarks also indicate when the media beds were cleaned
May 15 to June 30, 1991
This report covers one and one-half months of pilot plant work in both Baity Lab and at OWASA's Jones Ferry Road Water Treatment Plant. Six floccuiation runs were made with the tapered bed filled with ceramic media or ceramic media over Norpak.. For five of these runs, the overflow column was filled with Norpak media. Ferric chloride was used as the coagulant for the two runs conducted in Baity Lab. The four runs conducted at OWASA used water from the rapid mix basin to which alum, polymer, PAC, and potassium permanganate had been added. Water temperatures ranged from 23.5 to 27.0OC The results
are presented below as bullets.
Of the two runs conducted at Baity Lab, the first run was with two layers of ceramic media over a layer of 1" Norpak, all in the downflow configuration. Flow rate was 6 GPM. FeCIa concentration was 14 mg/l. Effluent turbidities averaged 5.0 NTU, settling to 3.5 NTU after the bed had ripened. The run was terminated after the Norpak media separated from the rest of the bed after 92.5 hours of run time. Before the Norpak separated, settieable floe was exiting the
bed. The second run was conducted with three ceramic layers in the
flocculator, followed by Norpak in an upflow mode in the overflow column. Flow rate to the flocculator was 6 GPM for most of the run. Twelve mg/l of FeCIs was used. There were not sufficient data points in this run to draw any significant conclusions. The run was made to demonstrate flocculator operation to CDM and EIMCO during the technical review meeting. Runs 20-22 were conducted
in this configuration and are documented in the April report.
The next four runs were made at the OWASA Treatment Plant using water from the rapid mix basin to which OWASA had already added chemicals. Chemical dosages were 35 mg/l alum, 0.1 mg/l polymer, 4.5 mg/l PAC, and 0.5
mg/I KMNO4. All runs were made with ceramic media in the flocculator and Norpak in the overflow column. During the first run three layers of ceramic were in the flocculator, flow rate was 6 GPM, and the loading rate to the overflow