for treatment

Intermittent sand filtration is a common and effective method for treating septic tank effluent. However, if the loading rate is too high, clogging and ponding of the sand filter surface layer can occur due to the accumulation of excessive biomass and the deposition of suspended solids. This ponding limits the practicality of sand filtration as it makes it necessary to take the filter out of service for maintenance. The objective of this study was to develop and test, on-site, a new hybrid filter system that would reduce the risk of clogging at an organic loading rate substantially greater than the maximum recommended loading rate for intermittent sand filters. The system comprised a 0.6 m deep horizontal flow biofilm reactor (HFBR) over a 0.85 m deep stratified sand filter. The HFBR consisted of a stack of 20 horizontal corrugated polyvinyl chloride (PVC) sheets, at 32 mm vertical spacings. The sheets were arranged so that the wastewater flowed over and back along alternate sheets down through the stack. The main biofilm growth formed on these sheets. The hybrid filter was loaded with septic tank effluent from an office/garage complex at the rate of 206 L/m 2 •d for a period of 400 days in 2 phases. During the first phase, the effluent volume of 600 L/d was applied in 24 doses/d for 10 minutes per dose, and during the second phase in 6 doses/d for 40 minutes/dose. Biofilms in the HFBR substantially reduced the organic and suspended solids loads that reached the sand filter surface and allowed an average total biochemical demand (BOD T ) loading rate, based on HFBR plan area, of 37 g BOD T /m 2 •d to be applied to the system without clogging. This rate was substantially greater than the maximum recommended loading rate of 24 g BOD T /m 2 •d for intermittent sand filters (US EPA 1980). During both loading phases a BOD T removal of 94% was achieved and nitrification was nearly complete. The average effluent BOD T was 12 ± 4 mg/L during both phases. The hybrid filter system appeared to perform better in terms of suspended solids handling and nitrification during the more frequent dosing phase. The hybrid filtration system offers a more compact alternative to intermittent sand filtration on its own with little risk of clogging.


INTRODUCTION
Intermittent sand filtration has been used to treat septic tank effluent (Furman et al. 1955;Sauer et al. 1976;Piluk et al. 1989). The treatment methodology involves the intermittent dosing of wastewater, following settlement, onto the surface of a sand filter in timed incremental doses. Biological processes such as organic carbon removal and nitrification occur within the sand filter and phosphorus removal occurs through ion exchange and adsorption onto the sand grains.
The US EPA (1980) recommends sand filter depths of 0.6 to 0.9 m and a maximum hydraulic loading rate, based on filter surface area, of 40 L/m 2 • d applied to a covered filter with an effective grain size, d 10 , between 0.35 and 1 mm and a uniformity coefficient, C u , less than 3.5. Anderson et al. (1985) recommend that the maximum organic loading rate should not exceed 24 g BOD T /m 2 •d and a minimum dosing frequency of two doses /d should be applied to single-pass sand filters.
An increase in dosing frequency may have a beneficial effect on the treatment performance of a sand filter (Anderson et al. 1985). In a study of 12 shallow sand filters (0.38 m deep), with an effective grain size (d 10 ) of 0.29 mm and an uniformity coefficient (C u ) of 4.5, Darby et al. (1996) found that increasing the hydraulic load application frequency from 4 to 24 times/d resulted in a slight but statistically significant increase in the removal of turbidity, chemical oxygen demand (COD) and organic-N. In a pilot and field scale experiment on single-layer sand filters (C u =4.4), Boller et al. (1993) showed that small hydraulic flushes in short intervals (12 hydraulic flushes versus 3 hydraulic flushes both totaling 40 L/m 2 • d) enhanced the hydraulic storage capacity, increased the average water retention time by 2 to 3 hours, and prohibited breakthrough of organic matter in the filter.
Clogging of the upper layers of the filter surface and surface ponding may occur over time, due to excessive loading of organic matter and suspended solids. When this occurs, the operation of the sand filter must be temporarily suspended. The occurrence of surface ponding has been attributed to a number of causes, namely, biofilm and secretion build-up (Siegrist and Boyle 1987;Vandevivere and Baveye 1992;Schwager and Boller 1997;Bouwer et al. 2000;Rodgers et al. 2004) and deposition of organic and inorganic solids (Daniel and Bouma 1974;Platzer and Mauch 1997). In a study of a laboratory intermittently loaded stratified sand filter treating synthetic agricultural wastewater, Rodgers et al. (2004) investigated the characteristics of the clogging layer, following the occurrence of surface ponding.
Measurements of field-saturated hydraulic conductivity within the upper layer of the sand filter indicated that the effects of the build-up of organic and inert matter may have extended to a depth of 0.165 m into the layer.
The main aim of this study was to develop and test, on-site, a new hybrid filter system, comprising a stack of 20 horizontal corrugated PVC sheets -spaced 32 mm vertically apart -over a stratified sand filter, for the treatment of septic tank effluent at a loading rate substantially greater than the maximum recommended rate for intermittent sand filters. It was expected that the 32 mm vertical spacing would provide sufficient volume for biofilms to grow without blocking the wastewater flow along the sheet channels. It was also considered that most of the substrates and solids would be removed in the biofilms on the sheets, substantially reducing the loads applied to the sand filter underneath. A secondary aim was to establish if dosing the system every hour would give a better performance than dosing it every 4 hours with the same daily volume of septic tank effluent.

MATERIALS AND METHODS
A horizontal flow biofilm reactor (HFBR) and a stratified sand filter were used to form a 'hybrid filter' and were placed in a pre-cast steel-fibre reinforced concrete tank. The top of the tank had external plan The HFBR module in the hybrid filter had a plan area of 2.91 m 2 and was 0.6 m in depth. The module consisted of 20 corrugated PVC sheets that were levelled and placed over one another using vertical stainless steel threaded bars and nuts at 32 mm intervals along the height of the bars to space the PVC sheets. The 20 sheets that formed the module were enclosed within a metal frame made from galvanized angle iron, which facilitated the removal of the module from the tank in one piece.
Each sheet channel was blocked off at both ends using PVC trimming bolted onto the bottoms of the channels of the sheets; any remaining gaps around the trimming were sealed using silicone. The PVC trimmings used to seal the ends of the sheets were rested upon each other to give support to the individual sheets. There were 32 channels on each sheet, and a 20 mm diameter hole was drilled in each channel at alternate ends of the sequential sheets so that the wastewater flowed along the channels of each sheet from one side to the other and down through the module (Figure 1). The bottom sheet of the module was drilled in a pattern to provide an even distribution of the treated wastewater over a stratified sand filter ( Figure 2).
The diameters of the holes in this bottom sheet were 40 mm. There was an adequate gap at the side of the module parallel to the channels for air entry into the module. A roof was constructed from 10 mm thick plywood and was bolted to the frame that surrounded the module. The roof prevented leaf litter and other organic debris from blocking the top of the module. The top of the roof was covered in impermeable PVC liner to prevent deterioration of the plywood. A length of PVC liner was hung from the front of the roof to allow easy access to the top of the module for inspection of the biofilm growth.
Septic tank effluent from an office/garage complex was applied to this hybrid filter at a rate of 600 L/d.
The daily hydraulic loading rate did not change during the study, but 2 dosing regimes were used -Phase 1 and Phase 2. During Phase 1, which lasted 289 d, the wastewater was dosed 24 times/d for a 10-minute duration and during Phase 2, which lasted 111 d, the wastewater was dosed 6 times/d for a 40 minute duration. The septic tank effluent passed through a baffle filter (Zabel Environmental Technology, Crestwood, Ky., USA) into a feed pump sump. The operation of the feed pump was controlled by an electric timer and float switch that ensured the pump was not activated when the level of wastewater in the sump was very low. A valve was attached to the pump delivery hose so that the flow entering the hybrid filter could be regulated. The flow rate was measured by a flow meter.
The wastewater was delivered to the surface of the HFBR through a 42 mm DYKA pressure distribution pipe. Holes of 2 mm diameter were bored at the top of this distribution pipe at 75 mm intervals. A 250 mm diameter WAVIN pipe was cut in half and placed over the distribution pipe; this insured that the septic tank effluent, on exiting from the distribution pipe, hit the WAVIN pipe, thereby slowing down the speed of the wastewater and spreading it evenly over the top of the HFBR module.
Grab samples of the influent wastewater from the sump (Influent in Tables 1 and 2) were taken by disconnecting the delivery hose from the DYKA pressure pipe and by then turning on the pump to fill a 500 ml container. Samplers, illustrated in Figure 4 and constructed from the stop ends of a 225 mm WAVIN pipe, were added to the ends of one of the channels on the 4 th , 8 th , and 16 th sheet down the HFBR (Tables 1   and 2). Similar samplers also collected water samples in the gravel distribution layer at the top of the sand filter (1 st gravel in Tables 1 and 2), the pea gravel layer between the coarse sand layer and the fine sand layer (2 nd gravel in Tables 1 and 2), and the pea gravel layer between the 2 layers of fine sand (3 rd gravel in Tables 1 and 2). Water samples were also taken at the outlet of the hybrid filter (Effluent in Tables 1 and   2). The locations of these sampling points are illustrated in Figure 3. Weekly grab samples were collected at all these locations. Following treatment by the hybrid filter, the final effluent was discharged to a percolation area. The results presented are for the period when steady state conditions (defined by nearconstant effluent organic and nutrient concentrations) in the filter were achieved.
The water quality parameters measured included BOD (5-day BOD test), chemical oxygen demand (COD) (closed reflux, titrimetric method), ammonium-N (NH 4 + -N) (ammonia-selective electrode method), NO 3 --N (nitrate electrode method), ortho-phosphate P (PO 4 + -P) (ascorbic acid method) and suspended solids (SS) (total suspended solids dried at 103-105 o C). All water quality parameters were tested in accordance with the Standard Methods for the Examination of Water and Wastewater (APHA-AWWA-WEF, 1995). The temperature of the influent and effluent wastewaters was also monitored throughout the study.

RESULTS AND DISCUSSION
The hybrid filter unit comprising the 0.6 m deep HFBR over the 0.85 m deep stratified sand filter was constructed above ground surface and the samplers were located in a south facing position. Both study phases took place under similar climatic conditions.

Carbonaceous oxidation
During Phase 1 of the study (Table 1), when the dosing of the septic tank effluent was hourly, the average HFBR influent total COD (COD T ) was 389±75 mg/L and BOD T was 179±43 mg/L. The corresponding average HFBR organic loading rates were 80 g COD T /m 2 • d and 37 g BOD T /m 2 • d, based on the HFBR plan area. The hydraulic loading rates were 206 L /m 2 • d on the HFBR and 189 L/ m 2 • d on the stratified filter -the difference was due to the HFBR's smaller plan area. In the HFBR, the COD T was reduced by 48 % and the BOD T by 73 % leaving sand-filter loading rates of 38 g COD T / m 2 • d and 9 g BOD T / m 2 • d, which is well below the US EPA (1980)  During Phase 2 (Table 2), when the dosing of the septic tank effluent was every 4 hours, the results for organic carbon removal were similar to those in Phase 1 for both the HFBR and the overall hybrid system e.g., the overall BOD T removal in Phase 1 was 94 % for an influent BOD T of 179±43 mg/L compared to 93 % in Phase 2 for an influent BOD T of 165±30 mg/L.
It seemed that the organic carbon removal capability of the hybrid system for the 2 loading regimes -using the same total daily loading rate -were equally efficient.

Suspended solids removal
During Phase 1 (Table 1) there was a marked continuous reduction in the average SS concentrations down through the HFBR unit from 132±95 mg/L in the influent to 58±20 mg/L in the gravel underlying the HFBR -a decrease of 74 mg/L. This decrease brought about a low solids loading on the sand filter surface reducing the risk of clogging. During Phase 2 (Tables 2) the SS concentrations, while still showing a continuous reduction down through the HFBR, ranged from 89±15 mg/L in the influent to 70±15 mg/L in the gravel underlying the HFBR -a decrease of only 19 mg/L. Biofilm growth was most extensive on the uppermost sheets of the HFBR as a result of the greater organic loading on these sheets. There was also deposition of solids on these sheets from the influent wastewater. The greater the hydraulic dose applied to the HFBR the more likely it was that the biofilm and deposited solids would be suspended in the wastewater flow and transported from sheet to sheet to give a more even distribution of SS on the sheets and higher SS concentrations entering the sand filter, as happened in Phase 2 of the study, when the dose applied was about 3 L /channel -4 times greater than in Phase 1. This increased transport of solids down through the hybrid system could lead to earlier clogging of the sand filter. This suggests that if the SS load on the sand filter is to be kept low, it is better to dose frequently e.g., hourly. This finding is in agreement with studies by Darby et al. (1996) who found that a 99 % SS reduction occurred in sand filters at hourly loadings and a 95 % SS reduction at loadings every 4 hours.
The overall solids removal was excellent in both phases and the final effluent had SS concentrations of 12±4 mg/L in Phase 1 and 13±6 mg/L in Phase 2. On analysing all the SS data down through the hybrid system, it seems likely that the sand filter could clog earlier under the Phase 2 loading than under the Phase 1 loading.

Nitrogen conversion
The concentrations of the influent NH 4 + -N in both phases (Tables 1 and 2) were greater than normal domestic wastewater and were due to cleaning agents used in the office/garage complex. Despite these high concentrations, most of the NH 4 + -N was eliminated in the hybrid filter in both phases.
In Phase 1, the NH 4 + -N concentrations in the HFBR decreased from 120±3 mg/L to 84±26 mg/L -a reduction of 36 mg/L. The corresponding dissolved oxygen (DO) increase in the HFBR was from 1±1 mg/L to 5±1 mg/L ( Table 1). The NH 4 + -N reduction was substantially greater than that of 20 mg/L in Phase 2. The indication of less efficient nitrification in the HFBR in Phase 2 could, as a result of the high volume of wastewater applied in each dose in this phase, be attributed to the redistribution of solids from the uppermost sheets to the lower sheets and consequent adverse conditions for nitrifier growth throughout the HFBR.
For the overall hybrid system in Phase 1, there was an 88 % NH 4 + -N reduction, producing a final effluent of The NH 4 + -N data suggest that the hybrid system is well capable of reducing NH 4 + -N concentrations in normal domestic wastewater to low values and that the hourly loading regime in Phase 1 is more efficient than the Phase 2 regime of loading every 4 hours.

Phosphorus removal
Phosphorus removal was similar in the HFBR and sand filters for both phases (Tables 1 and 2

Bacteria removal
In Phase 1, the coliform content of the influent and effluent wastewater varied greatly. The average total coliform removal in the hybrid filter was 99.7% and the average fecal coliform removal in the hybrid filter was 99.5%.

CONCLUSIONS
The following can be concluded from the on-site testing of a new hybrid filter system comprising a 0.6 m deep horizontal flow biofilm reactor (HFBR) over a 0.85 m deep stratified sand filter: 1. At a hydraulic loading rate of 206 L/m 2 • d dosed hourly for 10 minutes and an average organic carbon loading rate of 37 g BOD T /m 2 • d, the system was capable of producing an effluent with a BOD T of 12 mg/L and SS of 12 mg/l. These loading rates and dosing regime could be used in designs to give a substantially smaller treatment system plan area than an intermittent sand filter on its own.

2.
No clogging was evident in the hybrid system as the main biofilm growth and suspended solids deposition formed in the HFBR where they could be readily accommodated. It appears from the SS data that dosing on an hourly basis would have less risk of clogging than dosing every 4 hours.
3. Nitrification of a high strength NH 4 + -N influent wastewater was 88 % complete in the hybrid system dosed on an hourly basis as against 73 % in the system dosed every 4 hours. This suggests that the more frequent dosing system is better for nitrification.
The system was easy and cheap to construct and run, and would be suitable as a wastewater treatment system for small applications including single houses.      (7) 13 (6) 27 (17)