The role of Cayuga Lake Seston Extract in SSF
By
Monroe L. Weber-Shirk*
School of Civil and Environmental Engineering
Cornell University
Ithaca, New York 14853-3501
Phone: (607) 255-8445
Fax: (607) 255-9004
E-mail:
Draft Tuesday, August 29, 2000
to Water Research
The role of Cayuga Lake Seston Extract in SSF
Abstract
Cayuga Lake Seston Extract (CLSE) was obtained from Cayuga Lake (Ithaca, NY) and applied to glass bead filter beds at different application rates. Biological activity in the filters was inhibited with 3 mM sodium azide. The filters were challenged with a synthetic raw water containing E. coli. The CLSE fed filters removed up to 99.9999% of the influent coliforms.
Introduction
Ever since 1885 when Percy F. Frankland discovered that London’s slow sand filters removed bacteria[1] the mechanisms responsible have been a mystery. In previous work[2] we showed that bacteria removal could be achieved by a Chrysophyte that efficiently preys on bacteria in suspension. Chrysophyte predation does not appear to be amenable to engineering manipulation since enhanced populations of the Chrysophyte could only be supported by increasing the influent bacteria population. In addition, the 4-mm-diameter Chrysophyte is smaller in size than several of the pathogens that are of significant concern.
We also showed that particle removal improves even when filters were continuously poisoned with sodium azide[3]. This suggested that at least one other particle removal mechanisms is operative in slow sand filters. The hypothesis that bacteria are removed by attachment to sticky biofilms was suggested by Huisman and Wood,[4]Bellamy et al.,[5] and Hirschi and Sims[6], however, our previous research [7] indicated that biofilms were not significant in slow sand filters fed Cayuga Lake water.
We knew from previous research that Cayuga Lake seston applied to slow sand filters could improve particle capture. It was possible that the physical chemical ripening mechanisms was simply due to a reduction in pore size as particles accumulated in the voids. However, head loss calculations indicated that even a thin layer of clogged voids would create excessive head loss if the voids approached 1 mm in diameter. Thus it seemed likely that the seston increased the stickiness of the media surfaces.
Additional unpublished findings included the observation of a polymer tightly attached to slow sand filtration media even after only a few hours of operation and the observation that Cayuga Lake water would rapidly clog a glass fiber filter, but that the head loss through the filter could be reduced by acid washing the filter. This suggested that an acid-dissolvable polymer was present in lake water and so we set out to extract the polymer to test its ability to modify filter media surfaces.
Recent discovery of transparent exopolymer particles (TEP), in oceans [8] and lakes [9] and their role in enhancing particles stickiness [10], suggested that TEP might be responsible for physical-chemical particle removal in slow sand filters.
Materials and Methods
Cayuga Lake Seston Extract source
Approximately 140 L of Cayuga Lake water were obtained from Bolton Point Water Treatment Plant (Ithaca, NY). The raw water was obtained from a line that is used to continuously sample the raw water quality. The continuous low flow in the line allows sedimentation and the initial flushing of the line contained a large quantity of particles. The line was flushed on two separate occasions and the samples were used as CLSE sources. The CLSE was extracted from the lake water by first allowing the particles to settle, discarding the supernantant, adding an equal volume of 1 N HCl per volume of the particle suspension, centrifuging the acidified suspension, and finally collecting the supernatant. The supernatant was yellow and would immediately flocculate when neutralized with NaOH. The CLSE was further concentrated by neutralizing the acidified CLSE with NaOH, centrifuging the suspension to form a CLSE gel and then dissolving the CLSE in 0.5 N HCl. The extracted CLSE was dissolved in a total of 1.5 L. A 100 mL sample of this stock was neutralized with base and centrifuged to form a polymer gel. The CLSE gel had a wet mass after centrifugation at 3300 g for 5 minutes of approximately 55 g/L of the stock and a dry mass (dried at 105°C) of 2.58 g/L. Thus the centrifuged gel had approximately 20 g of water per g of CLSE. The dried CLSE consisted of approximately 56% volatile solids based on combustion at 550°C. The nonvolatile residual is not soluble in distilled water. It is slowly soluble in 5 N HCl and when dissolved has a yellow color. From ICP analysis it is primarily aluminum, sodium, and iron. As the pH is raised above 7 it appears to become clear. At neutral pH it forms flocs and has a slightly pinkish color.
CLSE analysis
The CLSE was analyzed for metals using the inductively coupled plasma technique. The primary element was aluminum. The sodium was undoubtedly from the sodium hydroxide used to neutralize the acid extract. The total mass identified was 797 mg/L. The total inorganic dry weight was 1.14 g/L and the total organic dry weight was 1.44 g/L.
Figure 1. ICP analytical results of CLSE by mass.
Filtration apparatus
The filtration apparatus has already been described[3]. The filters are 10 cm in diameter, 18 cm deep, and are filled with glass beads. The glass beads passed a 0.42 mm sieve and were retained on a 0.30 mm sieve. The filter approach velocity was 10 cm/hr. All filters were fed from a common feed that was made by combining distilled water with 500x concentrates of E. coli, ions, and sodium azide.
E. coli source
Filter performance was evaluated based on Escherichia coli removal. E. coli were added to all filters at an influent density of approximately 2/mL. E. coli were grown on Nutrient broth, protected against cell damage by adding 15% glycerin and then frozen in liquid nitrogen in 1 mL cryovials. Each cryovial contained approximately 2x109 E. coli as measured using membrane filtration. One cryovial of E. coli was diluted to 2 L using buffered dilution water. The 2 L stock of E. coli was metered into the distilled water feed at a dilution of 1:500.
Major Ions
The synthetic feed was made using distilled water amended with 250 mM CaCl2, 100 mM NaNO3, 150 mM MgSO4, and 150 mM NaHCO3. These 4 salts were added according to the concentrations used in Fraquil, a synthetic water designed to mimic the trace element composition of fresh waters (Morel, et al., 1975). Two 500x concentrated stock solutions were created to eliminate precipitation. These stocks were blended continuously with the distilled water feed. The influent pH was between 6 and 7.
Biological Inhibition with Azide
All filters were initially poisoned with azide at 3 mM (0.02%) to ensure that E. coli removal was only due to abiotic processes. As previously documented [7] E. coli enumeration was not affected by exposure of the sample to sodium azide. Azide was washed away in the membrane filtration procedure and thus the reversibility of the azide inhibition[11] made it possible to count E. coli in samples containing azide. An azide 500x stock solution was used to meter in azide to the distilled water feed shared by all of the filters. The azide feed was discontinued after 3.6 d to verify that filter performance was unaffected by the azide.
Application of CLSE
CLSE was added to the filters using two techniques. Two filters received initial doses of CLSE at 22 and 108 g/m2. We initially pumped the neutralized CLSE suspension into the filters in downflow mode. However, the CLSE clogged the filters as evidenced by excessive head loss. To circumvent this problem we inverted and rolled the filters to mix the CLSE with the filter media.
Three filters received continuous feeds of acidified CLSE and base at different CLSE concentrations. The normality of the base was adjusted to equal that of the acidified CLSE and all the feeds were metered by a single multi-channel peristaltic pump. The concentrations of the CLSE feeds and resulting application rates are shown in Table 1. The medium and high application rates had to be terminated early because the maximum head loss of 1 m was exceeded.
Table 1. CLSE application rates (by dry weight) and total application given actual duration of the CLSE feed.
CLSE concentrationmg/L / CLSE flux
g/(m2/day) / total CLSE applied
mg / total CLSE applied
g/m2
low / 258 / 0.62 / 48.4 / 6.2
medium / 1290 / 3.10 / 137 / 17.5
high / 6450 / 15.5 / 148 / 18.8
The CLSE application rates were chosen based in part on earlier experiments that indicated that a slow sand filter fed Cayuga Lake water accumulated organic carbon in the filter at a rate of approximately 0.2 mg/m2/day.
Filter performance evaluation
Filter performance was evaluated based on Escherichia coli and particle removal. E. coli were chosen as test particles because they do not multiply significantly under the low nutrient conditions of slow sand filters and because their removal is one of the objectives of water treatment. The E. coli were enumerated in the filter influent and effluent using the membrane filtration technique as described in Standard Methods for the Examination of Water and Wastewater[12]. Since all the filters shared a common feed the influent was measured with a single sample from each filter influent (6 samples). Filter effluents were measured with 2 or more membrane filters. An attempt was made to place slightly less than 80 E. coli on each membrane filter by adjusting the sample volume from approximately 50 mL for influent samples up to 2 L for some effluent samples. Die off did not contribute significantly to E. coli inactivation or removal as demonstrated by observing that approximately 50% of the influent E. coli were present in the control effluent.
Results and Discussion
The two filters that received an initial application of CLSE had somewhat improved removal of E. coli. The filter that received 22 g/m2 had an average E. coli removal of 90%. The filter that received 108 g/m2 removed up to 99% of influent E. coli, however, bacteria growth within the filter interfered with the membrane filtration test and made E. coli enumeration difficult. Although the initial application of CLSE improved filter performance it was not as effective as the continuous addition of the acidified CLSE.
Filter performance is illustrated in Figure 2. The time zero measurement was taken prior to adding any CLSE to the filters. The filters receiving 3.1 and 15 had no detected E. coli in 0.8 mL on day 1 and thus their actual performance was not quantified. Subsequent sample volumes were increased to reduce the detection limit. The filter receiving 0.62had no effluent E. coli in a 2 L sample on days 8 and 9 even though the influent E. coli concentration was approximately 0.8/mL and thus the removal likely exceeded 6 log. The break in CLSE feed to the filter receiving 0.62 on the 3rd day was the result of a leak in the CLSE line. This corresponds to a decrease in E. coli removal efficiency that may have been caused by The E. coli removal in the 3 filters that received acid-dissolved CLSE exceeded 6 log (99.9999%) removal (). These remarkable results were achieved after an application of only 4 g of CLSE per square meter of filter area.
From the filter that received 15 it is apparent that filter performance continues to be excellent for several days even after the CLSE feed was discontinued. Filter performance decreased steadily from approximately 6-log removal when the CLSE feed was ended to 2-log removal 7 days later. This suggests that CLSE improves filter performance by changing the surface properties of the filter media and thus enhancing the E. coli attachment efficiency. This is also consistent with the gradual improvement in filter performance for the filter receiving 0.62.
The CLSE is apparently fairly resistant to biodegradation
Head loss increased with addition of CLSE as indicated in Figure 2.
The removal of E. coli was abiotic as indicated by the lack of any change after the cessation of the azide feed.
Figure 2. E. coli removal as a function of time and CLSE application rate.
Figure 3. Head loss as a function of time.
Conclusion
Small mass of CLSE enhances filter removal.
Future research required determining amount of CLSE that maintains a desired level of pathogen removal.
Future research required characterizing CLSE and identifying the active components.
Acknowledgments
I thank Andrew and JC for conducting the trial run of this research and isolating the first CLSE. I also thank Seokoon and Natalie for preparing the stock of E. coli.
References
1. Baker, M.N., The Quest for Pure Water. Vol. 1. 1981, Denver: American Water Works Association.
2. Weber-Shirk, M.L. and R.I. Dick, Bacterivory by a Chrysophyte in Slow Sand Filters. Water Research, 1999. 33(3): p. 631-638.
3. Weber-Shirk, M. and R.I. Dick, Physical-Chemical Mechanisms in Slow Sand Filters. Journal American Water Works Association, 1997. 89(1): p. 87-100.
4. Huisman, L. and W.E. Wood, Slow Sand Filtration. 1974, Geneva: World Health Organization.
5. Bellamy, W.D., D.W. Hendricks, and G.S. Logsdon, Slow Sand Filtration: Influences of Selected Process Variables. Journal American Water Works Association, 1985. 77(12): p. 62.
6. Hirschi, S.D. and R.C. Sims. Particles and Microorganisms in Slow Rate Sand Filtration. in AWWA/UNH Slow Sand Filtration Workshop. 1991. Durham, N.H. Oct. 27-30.
7. Weber-Shirk, M. and R.I. Dick, Biological Mechanisms in Slow Sand Filters. Journal American Water Works Association, 1997. 89(2): p. 72-83.