Published in Soil Sci. Soc. Am. J. 68:1827-1832 (2004).
© 2004 Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
Division S-1—Soil Physics
Surface Clogging in an Intermittent Stratified Sand Filter
M. Rodgers,
J. Mulqueen and
M. G. Healy*
Dep. of Civil Engineering, National Univ. of Ireland, Galway, Ireland
* Corresponding author (markgerardhealy{at}hotmail.com)
 |
ABSTRACT
|
|---|
Accumulation of biomass and deposition of suspended solids at the surface of a sand filter can lead to clogging of the filter media. A laboratory intermittent sand filter column, which included three sand strata, was operated for a period of 806 d before failure occurred through surface clogging. Upon dismantling the column, the cause and effects of the surface clogging were investigated. The main mechanism responsible for sand clogging appeared to be biomass buildup. Maximum loss on ignition of filter media samples was 2.35%, and it occurred in the upper 0.01 m of the sand. There was a reduction in field-saturated hydraulic conductivity in the top 0.01 m of the upper sand stratum from a value of 1.9 x 10–3 ± 1.7 x 10–4 m s–1 (for virgin sand with an effective size, d10, of 0.45 mm) to 3.5 x 10–5 ± 7.5 x 10–6 m s–1. The soil-water characteristic curve, which relates the volumetric water content (
v) to the soil suction, also reflected the changes in the filter media due to clogging. The water-holding capacity greatly increased as biomass accumulated in the filter media. Scanning electron microscopy (SEM) confirmed the existence of a clogging organic layer on the surface of the top sand layer.
Abbreviations: (h), soil-water characteristic curve • v, volumetric water content • Cu, uniformity coefficient, d10, effective size • Kfs, field-saturated hydraulic conductivity • SEM, scanning electron microscopy
|
INTRODUCTION
|
|---|
INTERMITTENT SAND FILTERS have been used successfully to treat high-strength wastewaters in both single-pass and recirculation modes (Gross and Mitchell, 1985; Christopherson et al., 2001; Healy et al., 2004). Biological processes such as organic carbon removal and nitrification occur with the intermittent dosing of wastewater onto the surface of the sand filter. Oxygen transport into the sand filter is achieved mainly by mass transport and diffusion. Although intermittent sand filters appear to offer an economic alternative to land spreading or wetland treatment of dairy wastewater, the capacity of the sand media to treat the wastewater may be exceeded by the hydraulic and chemical loading rates across time, resulting in surface ponding; this reduces the aeration of the sand filter media and may result in failure of the filter. Ponding represents a serious threat to the use of sand filters for the treatment of these wastewaters.
Surface ponding may result from a number of causes. Accumulation of microorganisms and secretions on surfaces as biofilms is believed to be the cause of surface sealing (Siegrist and Boyle, 1987; Vandevivere and Baveye, 1992; Schwager and Boller, 1997; Bouwer et al., 2000). In this process, hydrated extracellular polymers (exopolymers) as well as cells accumulate on the upper layers of the sand media and give rise to a reduction in hydraulic conductivity (Schwager and Boller, 1997). Kropf et al. (1977) found that, although clogging is a surface phenomenon, factors such as the hydraulic and chemical loading rate of the applied wastewater and the filter media characteristics dictate the depth of its accumulation in the filter media and suggested that, once established, the clogging layer develops independent of the filter media characteristics. Siegrist and Boyle (1987) found an accumulation of organic matter in the upper sand layer and suggested that it may have undergone humification and gradually filled the pore space, reducing the hydraulic conductivity. The type of filter media (Jowett and McMaster, 1995) and the deposition of organic and inorganic solids on the surface layer (Daniel and Bouma, 1974; Platzer and Mauch, 1997) have also been considered to cause surface sealing.
Biofilm accumulation has been shown to result in a 95 to 99% reduction in original clean surface hydraulic conductivity (Bouwer et al., 2000). It appears that the rate of infiltration of an intermittent flooding dose of effluent is the best indicator of the development of a clogging mat on an intermittent sand filter; this can be further investigated by the measurement of the field-saturated hydraulic conductivity (Kfs in m s–1), which is calculated by Darcy's law (Craig, 1997):
 | [1] |
where Q is the volume of water flowing per unit time (m3 s–1), A is the cross-sectional area of the specimen (m2), and dH/dZ the hydraulic gradient (m m–1).
Another way of showing the effects of biomass buildup is an analysis of the soil-water characteristic curve [(h)] which is a graph of the v against the pore-water suctions (h) imposed and is dependent on the texture and structure of the media (Fredlund and Rahardjo, 1993). The air entry value is the suction above which air becomes continuous in the specimen; it increases as the particle size of the media decreases (Brooks and Corey, 1966; Geo-Slope International, 2002). As biofilm builds up on each sand grain, the water retention capacity of the media increases. Results obtained by Siegrist (1987) using a filter loaded with domestic septic tank effluent, greywater septic tank effluent, and tapwater demonstrated this effect. Siegrist (1987) attributed the most significant changes in water content near the infiltration surface to the pore size reduction due to biomass buildup and, after 62 mo of operation, the water contents in the upper-4-cm layer for tapwater and domestic septic tank effluent were 26 and 36% by volume, respectively.
The aim of this paper was to investigate the development of a clogging layer in a laboratory intermittent sand filter and to describe its distribution, composition, and effect on field-saturated hydraulic conductivity and water retention capacity. Chemical analysis and SEM were used to characterize the clogging layer, and to assess the influence of biofilm growth on its development.
|
MATERIALS AND METHODS
|
|---|
A laboratory intermittent stratified sand column for treating wastewater was operated for a period of 806 d in single-pass and recirculation modes (Fig. 1)
. The column was 0.9 m deep and 0.3 m in diameter and was designed after Gross (1990), who obtained good contaminant removal efficiency with this design. In the column, a 0.1-m layer of distribution gravel (0.01–0.02 m in size) overlaid a 0.25-m layer of coarse sand media (d10 = 0.45 mm; uniformity coefficient, Cu = 3; particle density = 2.6 g cm–3) and two 0.15-m layers of fine sand (d10 = 0.11 mm; Cu = 1.6; particle density = 2.0 g cm–3) which were separated from each other by 0.075-m layers of pea gravel (0.01–0.02 m in size). The bottom layer of sand was underlain by a 0.1-m layer of pea gravel.
View larger version (26K):
[in this window]
[in a new window]
|
Fig. 1. The recirculating intermittent stratified sand filter used in this study. In single-pass operation, the synthetic effluent was also applied from the feed tank to the sand filter surface but there was no recirculation.
|
|
The sand column was instrumented with an access tube (type ATL1, Delta-T Devices Ltd., Cambridge, UK) to allow the v to be estimated using time domain reflectometry (probe type PR1/6d-02, Delta-T Devices Ltd., Cambridge, UK). Readings were taken in millivolts using a voltmeter (type HH2, Delta-T Devices Ltd., Cambridge, UK) and were converted to units of m3 m–3 using the manufacturer's calibration curve.
Synthetic wastewater (Table 1) with concentrations similar to those measured in dairy parlor wastewaters (Table 2) was made up daily and applied in four doses each day from a feed tank using a peristaltic pump to the sand filter via a spiral distribution manifold at a number of filter hydraulic loading rates and influent concentrations (Table 3). The synthetic effluent was mixed in the feed tank a number of hours before its application to the filter surface, so any adsorption of nutrients onto the bentonite would have taken place before the synthetic effluent was loaded onto the filter. The peristaltic pump was operational for a period of 5 min per dose. At the final occurrence of surface ponding, loading was discontinued and the column was dismantled.
View this table:
[in this window]
[in a new window]
|
Table 3. Water quality parameter concentrations and filter hydraulic, organic, and suspended solids (SS) loading rates applied to the stratified sand filter surface in single-pass (S) and recirculation (R) modes.
|
|
Upon dismantling the column, investigations were made to determine the mechanisms responsible for clogging that was the result of accumulation of substances from the total period of operation. The physical and chemical properties of the sand were characterized for depths at and below the surface. Three undisturbed sand cores, 0.05 m in diameter, were taken at each 0.015- to 0.02-m incremental depth below the surface and used to determine the field-saturated hydraulic conductivity of each layer by the constant-head method (British Standards Institution, 1990c). In this method, a reservoir tank supplied a constant ponded head of water, z, to the open-ended cell containing the sand core of height, l. An overflow weir on the reservoir tank allowed a constant head to be maintained. Taking the base of the open-ended cell as datum, the hydraulic gradient was defined as
 | [2] |
and the constant flow rate, Q, was measured by a graduated cylinder, positioned under the open-ended cell. A virgin sand specimen was used to compare the reduction in field-saturated hydraulic conductivity.
A virgin sample and undisturbed sand cores were also taken at incremental depths from another section of the sand filter column, and the (h) was determined for each depth using the sandbox method (Eijkelkamp Agrisearch Equipment Ltd., The Netherlands). The sandbox method involved the application of incremental water suctions to a saturated sand sample, contained in a stainless steel core, 10–4 m3 in volume and positioned on very fine sand, via an adjustable water table. The core was fully saturated before testing and its weight was recorded at numerous water suctions. The parameters, 
, n, and r of the van Genuchten Equation (van Genuchten, 1980), were calculated from the measured soil-water characteristic curves using the Levenberg-Marquardt minimisation algorithm (Gill et al., 1981), which minimized the sum of the squares of the errors (SSE) of measured and modeled parameters. This was applied directly using SOLVER in MS Excel (Microsoft Corp., Redmond, WA) using initial estimates of the parameters (Wraith and Or, 1998).
The physical mechanism responsible for clogging was investigated in the following manner: On dismantling the column, every 0.01-m layer below the surface was sampled and analyzed for loss on ignition by drying at 105°C, weighing, and then placing in a muffle furnace at 440°C in accordance with the British Standard (British Standards Institution, 1990b). The organic matter burnt off at this temperature, leaving behind a mixture of sand, ash, and bentonite. To disperse the very fine particles following ashing, the filter media was mixed with sodium hexametaphosphate solution at a concentration of 2 g L–1 (British Standards Institution, 1990a) and washed through a 53-µm sieve. This allowed the noncombustible residue, which was composed of bentonite and inert materials dislodged from each layer, to be calculated as a percentage of the total mass of sand grains within that layer. A comparison of loss on ignition gives an indication of biomass distribution within the column.
Scanning electron microscopy was used to view the biofilm buildup on individual sand grains at three locations in the sand layer: at the surface, and at 0.02 and 0.07 m below the surface. The samples were taken using an aluminium stub coated with quick-drying silver paint. When dried, the specimens were gold-coated in an Emscope SC 500 sputter coater (Emscope, Ashford, UK) and were viewed with a scanning electron microscope (Model S-570, Hitachi, Tokyo, Japan) at a 40 x magnification.
Chemical analysis was also performed for every 0.01-m depth increment below the surface. The parameters tested were total-N, total-P, and total-S (after Byrne, 1979), which are indicators of the abundance of organic matter.
|
RESULTS AND DISCUSSION
|
|---|
Surface ponding occurred on three occasions throughout the operation of the column (at filter hydraulic loading rates of 60, 42, and 53.6 L m–2 d–1, Table 3). At the final occurrence of surface ponding, the column was dismantled. The main mechanism responsible for sand clogging appeared to be biomass and secretion buildup in the uppermost part of the coarse sand layer. Loss on ignition was a maximum of 2.35% in the 0- to 0.01-m layer below the coarse sand surface, which was more than double the loss on ignition of the virgin sand (1%), and gradually reduced as it extended into the coarse sand layer (Fig. 2)
. This appeared to confirm the work of Kropf et al. (1977), who suggested that the clogging material penetrates into a coarse media. The gel-like texture of the surface of the uppermost coarse sand layer indicated that a clogging mat formed on the surface, with more permeable sand lying directly underneath. Below the first 0.01 m of coarse sand, the average loss on ignition was 1.3 ± 0.2%, which indicated that the main organic matter buildup occurred in the top 0.01 m of sand and also on the sand surface as a deposit. This general trend was exhibited in the retention of total-N, total-P, and total-S in the upper 0.03 m of sand (Fig. 3)
.
View larger version (21K):
[in this window]
[in a new window]
|
Fig. 2. Loss on ignition and noncombustible material deposition as a percentage of the total dry mass in the upper coarse sand layer of the laboratory stratified sand filter.
|
|
View larger version (16K):
[in this window]
[in a new window]
|
Fig. 3. Deposition of total-N, total-P, and total-S (mg kg–1 of filter media) in the upper coarse sand layer of the laboratory stratified sand filter.
|
|
Measurements of Kfs indicated that clogging occurred in the upper coarse sand filter layer (Fig. 4)
. At 0.015 m below the surface, Kfs was 3.5 x 10–5 ± 7.5 x 10–6 m s–1; this was 2% of the Kfs of the virgin sand (1.9 x 10–3 ± 1.7 x 10–4 m s–1), but was still greater than the filter hydraulic loading rate during the period of clogging (6.3 x 10–7 m s–1). Although the clogging mat occurred at the sand surface, a 0- to 0.015-m deep sample was used to measure the field-saturated hydraulic conductivity in the uppermost coarse sand layer; this depth-averaged Kfs for the surface layer is likely to be much greater than the infiltration capacity at the sand surface in accordance with the expression for a stratified soil (Craig, 1997):
 | [3] |
where K1 and K2 are the field-saturated hydraulic conductivities of the biomas layer, H1, and the underlying sand layer, H2, respectively. Sampling disturbance may also have given rise to some change in the structure of the sand. The reduction in the relative Kfs value in this study was similar to that of Schwager and Boller (1997), who found that, in an intermittently loaded sand filter with a d50 of 0.85 mm and a Cu of 2.1 loaded at 120 L m–2 d–1 with septic tank effluent, the relative hydraulic conductivity reduced to <5% of the Kfs of the virgin sand in the upper 0.04 m of the sand filter.
View larger version (15K):
[in this window]
[in a new window]
|
Fig. 4. Relative hydraulic conductivity variation (Kfs/Kvirgin sand) in the uppermost part of the coarse sand layer with depth after 806 d of operation.
|
|
The reduction in Kfs appeared to extend deep into the upper sand layer (Fig. 4). The Kfs only returned to that of the virgin sand at a depth of 0.165 m below the surface. The reduction in Kfs and the penetration of organic matter into the uppermost coarse sand layer appeared to be correlated to one another; excluding the loss on ignition in the top 0.01 m layer below the coarse sand surface (2.35%), the average loss on ignition was 1.4% in the upper 0.12 m of the coarse sand layer; below this level, the loss on ignition decreased linearly from a value of 1.55% (at 0.13 m below the surface) to 1% at 0.22 m below the surface (Fig. 2). Conversely, the Kfs increased linearly from an average value of 2.8 x 10–4 m s–1 in the upper 0.12 m of the coarse sand layer to the value for virgin sand at 0.18 m below the surface. Loss on ignition was significant for a further 0.04 m into the coarse sand layer, but this appeared to have little effect on the Kfs.
The soil water characteristic curves for several depths in the coarse sand are compared with those for virgin sand in Fig. 5 and the parameters of the van Genuchten equation for each of the sand depths are given in Table 4. The soil-water characteristic curve for the 0- to 0.015-m layer below the coarse sand surface had the highest water content, 50 and 110% greater than that of the virgin sand at 0- and 0.1-m suctions, respectively. This was a reflection of the high loss on ignition of the upper layer of the sand filter. The curves in Fig. 5 exhibit the same characteristics as exhibited by time domain reflectometry measurements taken in the uppermost coarse sand layer of the sand filter and confirm the hydrophilic nature of the biofilm. Using time domain reflectometry, in situ v measured at the top of the upper coarse sand layer of the sand filter used in this study (at a filter hydraulic loading rate of 20 L m–2 d–1) after 189, 240, and 270 d since the start of operation, were 23, 30, and 39% at 5 min after loading.
View larger version (31K):
[in this window]
[in a new window]
|
Fig. 5. The soil-water characteristic curve, (h), for the upper coarse sand layer of the laboratory stratified sand filter.
|
|
Scanning electron microscopy photography showed organic deposits that were in accordance with the indirect quantitative (h), Kfs, and loss on ignition results. Figure 6
illustrates the buildup of clogging material at three different depths in the upper sand layer: at 0.07 m below the surface (Fig. 6B), at 0.02 m below the surface (Fig. 6C), and at the sand surface (Fig. 6D). As a comparison, the virgin sand sample is also illustrated (Fig. 6A). In Fig. 6B and 6C, the sand grains are clearly distinguishable. However, in Fig. 6D (the surface layer), the clogging material makes the sand grains indistinguishable. This confirms that the clogging layer is a surface phenomenon. Although the organic and inert materials were high below the surface of the uppermost coarse sand layer of the filter (Fig. 2), Fig. 6D indicates that the clogging layer developed as a schmutzdecke on the surface.
View larger version (61K):
[in this window]
[in a new window]
|
Fig. 6. Scanning electron microscopy photography on the upper coarse sand layer grains taken at a magnification of 40. (A) virgin sand sample; (B) sample at a depth of 0.07 m below the surface after 806 d of operation; (C) sample at a depth of 0.02 m below the surface after 806 d of operation; (D) surface sample after 806 d of operation.
|
|
The fact that clogging is a surface phenomenon indicates that skimming of the clogging layer to an approximate depth of 0.05 m and its replacement with virgin coarse sand will restore the filter; removal of the top layer to a depth of 0.165 m will completely restore the filter. Where the filter is overlain with distribution gravel in which manifold and lateral loading pipes are embedded, care is required to avoid damage to the pipework. While the gravel can be stockpiled for reuse, the clogged sand layer must be washed or disposed of by, for example, landspreading. To minimize the risk of clogging in a field situation, the primary tank holding the wastewater should be a long rectangular tank to promote sedimentation, and the pumping chamber should be fitted with a screen.
|
CONCLUSIONS
|
|---|
In intermittent sand filters, the main mechanism responsible for sand clogging appears to be biofilm development. The loss on ignition in the upper 0.01 m of the coarse sand layer was 2.35%; this high loss on ignition, the reduction of the field-saturated hydraulic conductivity, the increased water retention, and the nutrient concentration profiles indicate that sand clogging is largely a surface phenomenon. Physical inspection of the filter media and SEM analysis confirmed this. Loss on ignition was significant in the filter media to a depth of 0.22 m below the filter surface. Although the greatest reduction in field-saturated hydraulic conductivity occurred in the uppermost 0.01 m of the coarse sand layer, where a Kfs value of 3.5 x 10–5 ± 7.5 x 10–6 m s–1 was measured, field-saturated hydraulic conductivity only returned to its virgin sand value of 1.9 x 10–3 ± 1.7 x 10–4 m s–1 at a depth of 0.165 m below the filter surface.
|
ACKNOWLEDGMENTS
|
|---|
The authors are grateful to Teagasc for the award of a Walsh fellowship to the third author and for financial support for the work. The authors acknowledge the technical help of N. Donaghue (Department of Medicine, NUI, Galway), S. McCormack (Teagasc, Johnstown Castle), M. Reidy (Teagasc, Moorepark), M. Rathaille, M. O'Brien, and G. Hynes (NUI, Galway).
Received for publication January 7, 2004.
|
REFERENCES
|
|---|
- Bouwer, E.J., H.H.M. Rijnaarts, A.B. Cunningham, and R. Gerlach. 2000. Biofilms in porous media. p. 123–158. In J.D. Bryers (ed.) Biofilms II: Process analysis and applications. Wiley-Liss, New York.
- British Standards Institution. 1990a. Determination of particle size distribution. British standard methods of test for soils for civil engineering purposes. Classification tests. BS 1377:2:1990:9. BSI, London.
- British Standards Institution. 1990b. Determination by mass-loss on ignition. British standard methods of test for soils for civil engineering purposes. Chemical and electro-chemical tests. BS 1377:3:1990:4. BSI, London.
- British Standards Institution. 1990c. Determination of permeability by the constant-head method. British standard methods of test for soils for civil engineering purposes. Compressibility, permeability and durability tests. BS 1377:5:1990:5. BSI, London.
- Brooks, R.H., and A.T. Corey. 1966. Properties of porous media affecting fluid flow. ASCE J. Irrig. Drain. Div. 92:61–88.
- Byrne, E. 1979. Chemical analysis of agricultural materials. An Foras Taluntais, Dublin.
- Christopherson, S.H., J.L. Anderson, and D.M. Gustafson. 2001. Evaluation of recirculating sand filters in Minnesota. p. 207–214. In Karen Mancl (ed.) On-Site Wastewater Treatment: Proc. of the 9th Natl. Symp. on Individual and Small Community Sewage Systems, Fort Worth, TX. 11–14 Mar. 2001. American Society of Agricultural Engineers, St. Joseph, MI.
- Craig, R.F. 1997. Soil mechanics. 6th ed. E&FN Spon, London.
- Daniel, T.C., and J. Bouma. 1974. Column studies of soil clogging in a slowly permeable soil as a function of effluent quality. J. Environ. Qual. 3:321–326.[Abstract/Free Full Text]
- Fredlund, D.G., and H. Rahardjo. 1993. Soil mechanics for unsaturated soils. John Wiley and Sons, New York.
- Geo-Slope International. 2002. Seep/W for finite element seepage analysis. v. 5. User's guide. GSI, Calgary, AB, Canada.
- Gill, P.E., W. Murray, and M.H. Wright. 1981. Practical optimization. Academic Press, London.
- Gross, M. 1990. Virus removal by sand filtration of septic tank effluent. J. Environ. Eng. (Reston, Va) 116:711–720.
- Gross, M., and D. Mitchell. 1985. Biological virus removal from household septic tank effluent. p. 295–304. Proc. of the 4th Natl. Symp. on Individual and Small Community Sewage Systems, New Orleans, LA. 10–11 Dec. 1984. ASAE Publ. 07–85. American Society of Agricultural Engineers, St. Joseph, MI.
- Healy, M.G., M. Rodgers, and J. Mulqueen. 2004. Recirculating sand filters for the treatment of synthetic dairy parlour washings. J. Environ. Qual. 33:713–718.[Abstract/Free Full Text]
- Jowett, E.C., and M.L. McMaster. 1995. On-site wastewater treatment using unsaturated absorbent biofilters. J. Environ. Qual. 24:86–95.
- Kropf, F.W., R. Laak, and K.A. Healey. 1977. Equilibrium operation of subsurface absorption systems. J. Water Pollut. Control Fed. 17:2007–2016.
- Platzer, C., and K. Mauch. 1997. Soil clogging in vertical flow reed beds—Mechanisms, parameters, consequences and ... solutions? Water Sci. Technol. 35:175–181.
- Schwager, A., and M. Boller. 1997. Transport phenomena in intermittent filters. Water Sci. Technol. 35:13–20.
- Siegrist, R.L. 1987. Soil clogging during subsurface wastewater infiltration as affected by effluent composition and loading rate. J. Environ. Qual. 16:181–187.
- Siegrist, R.L., and W.C. Boyle. 1987. Wastewater-induced soil clogging development. ASCE J. Environ. Eng. 113:550–566.
- Vandevivere, P., and P. Baveye. 1992. Effect of bacterial extracellular polymers on the saturated hydraulic conductivity of sand columns. Appl. Environ. Microbiol. 58:1690–1698.[Abstract/Free Full Text]
- van Genuchten, M.Th. 1980. A closed-form equation for predicting the hydraulic conductivity of unsaturated soils. Soil Sci. Soc. Am. J. 44:892–898.[Abstract/Free Full Text]
- Wraith, J.M., and D. Or. 1998. Nonlinear parameter estimation using spreadsheet software. J. Nat. Resour. Life Sci. Educ. 27:13–19.
TOP
ABSTRACT
INTRODUCTION
