A comparison of the treatment performance of freely-draining subsoils
and horizontal subsurface-flow wetland systems receiving on-site
wastewater effluent
N.D. O’Luanaigh, L.W. Gill, B.D.R. Misstear and P.J. Johnston
Department of Civil, Structural and Environmental Engineering, University of Dublin, Trinity College,
Dublin 2, Ireland
(Email: noluanai@tcd.ie; gilll@tcd.ie; bmisster@tcd.ie; pjohnston@tcd.ie)
Abstract
On-site research was carried out in Ireland to assess and compare the treatment capabilities of freely
draining subsoils receiving both septic tank and secondary effluents in tandem with a treatment
assessment of a horizontal subsurface-flow reed bed. This is due to recent concern being expressed as
to whether a sufficient level of treatment is being achieved in the subsoil’s vadose zone coupled with
doubts being raised as to the treatment capabilities of horizontal flow reed beds as stand-alone
secondary treatment systems. Results over an 18-month period have shown the reed bed to remove
only 52% of the organic load but achieve 2-3 log removal in total coliforms and E. coli. The majority
of nitrification is seen to occur in the first 0.3m of subsoil for all sets of percolation trenches and,
coupled with poor denitrification throughout the subsoil, has compromised groundwater quality in the
form of nitrate infiltration. The research also shows that in general the septic tank effluent has
received a comparable quality to the secondary effluent in terms of E. coli by the time the point of
discharge to groundwater is reached in the subsoil.
Keywords
On-site wastewater treatment; STE, septic tank effluent; SE, secondary effluent; percolation area;
freely draining; horizontal subsurface flow (HF) reed bed
INTRODUCTION
Domestic wastewater from over one third of the population in Ireland, or 400,000 dwellings, is
treated by on-site systems (Department of the Environment et al., 2004). Given that over 25% of all
water supplies is currently provided by groundwater, there is a great need to protect this resource
from the risks posed by contamination arising from both diffuse (generally agricultural) and point
(manure and silage storage, on-site wastewater treatment systems) sources. Traditionally the
dominant management approach for on-site systems has been to settle wastewater in a septic tank
prior to subsurface distribution via a percolation area (soil treatment unit). Whilst the prevailing
anaerobic conditions in a septic tank provide limited treatment and breakdown of organics and
nutrients, it is the remaining biological constituents of the wastewater, namely pathogenic bacteria,
infectious viruses and protozoa that are, as Daly (2003) states, the single greatest threat to
groundwater in Ireland. As such, it is role of the percolation area and its inherent subsoil
characteristics to provide a sufficient contaminant attenuation zone and ensure long-term protection
from groundwater pollution.
A freely draining soil is classified as having a “T-value” of between 1 and 5, obtained from an onsite percolation test based on the standard falling head percolation test (Mulqueen and Rodgers,
2001). At present, one of the growing concerns amongst water authorities, practitioners and other
bodies in Ireland surrounds the discharge of on-site effluent into these highly permeable soils,
where although permitted under the current guidelines, there exists an intuitive feeling that the
effluent is percolating at an excessive rate. This ultimately creates the risk of an insufficient level of
nutrient and pathogen removal and an increased potential for groundwater contamination. Many
studies have highlighted the importance in the formation and development of an organic biomat
layer at the soil-gravel interface along the base and wetted sides of the percolation trenches and
deemed it to be a crucial factor in providing extensive treatment of the constituents in the applied
effluent (Bitton and Gerba, 1984; Siegrist, 1987; Gill et al., 2005). The level of its development is
thought to be dependent on a combination of both the subsoil’s grain size and the organic strength
of the applied effluent being treated.
In cases where septic tank effluent (STE) is deemed unsuitable for immediate discharge to a
percolation area (e.g. high water table, shallow bedrock), secondary effluent (SE) can be achieved
by an alternative system such as a horizontal subsurface-flow (HF) reed bed prior to further tertiary
treatment in the subsoil. Despite this, there is reluctance in Ireland amongst a number of water
authorities to permit the use of wetlands as stand-alone secondary treatment systems given the
scarcity of reliable long-term performance data relevant to Ireland and its climate coupled with their
apparent poor winter treatment performance (Healy and Cawley, 2002). HF reed beds, with gravel
as its chosen media, are regarded as particularly good in the removal of BOD5 and SS. However, the
majority of research to date has been carried out using domestic wastewaters with low (50 -150 mg
L-1) BOD5 inflow concentrations (Vymazal, 2001), probably due to the use of combined (surface,
rain and wastewater) sewer systems (Griggs and Grant, 2001). Undiluted STE, in comparison, has a
BOD5 of between 150 and 400 mg L-1. Little nitrification is also achieved across the bed, due to its
poor oxygenation capacity, which may have a detrimental impact on groundwater quality if the
receiving subsoil fails to provide sufficient treatment.
A three-year research project, funded by the Irish EPA, was undertaken to examine and compare the
performance of freely-draining subsoils, with a T-value of between 1 and 5, in treating both septic
tank effluent (STE) and secondary effluent (SE). A gravel-based horizontal subsurface-flow reed
beds was also assessed in providing secondary treatment to the fed effluent. This paper details the
findings from the initial 18-month period of sampling (April 2006 – September 2007).
MATERIALS AND METHODS
Site Selection, Design and Construction Layout
Two study sites were chosen subject to compliance with Groundwater Protection Schemes
(Department of the Environment et al., 2004) and EPA guidelines (EPA, 2000) for acceptability.
The first site, Site A, serving 6 no. residents and posting a T-value of 4, received STE which was
evenly split two-ways via a distribution device to a set of percolation trenches and an on-site
constructed HF reed bed. The secondary effluent from the reed bed was thereafter discharged to a
set of percolation trenches (Figure 1). Site B, serving 4 no. residents and yielding a T-value of 5,
received SE from a rotating biological contactor (RBC) with half the flow split to percolation
trenches and the other half to a smaller HF reed bed for tertiary treatment (Figure 2). Soil moisture
analysis was carried out via suction lysimeters (Soilmoisture Equipment Corporation) which were
installed in the percolation areas (as per Figures 1 and 2) at nominal depths of 0.3, 0.6 and 1.0 m
below the invert of the percolation trenches. Groundwater monitoring boreholes immediately
upstream (U/S) and downstream (D/S) of the percolation areas were also drilled to assess the
pollutant effect of the system on the underlying aquifer. To obtain a profile of soil moisture tension,
tensiometers (Soil Measurement Systems) were installed to the same three depths as the lysimeters.
In order to assess the meteorological effects (evapotranspiration and precipitation levels) on both
the reed bed and percolation area, a weather station (Campbell Scientific) and rain gauge (Casella)
were mounted on site to collect the appropriate data.
Design of the HF reed bed was based on WRc European Guidelines (Cooper et al., 1996), a
document which outlines the design features, methodology, removal mechanisms and operational
requirements of a number of different constructed wetland configurations. Combining the theory of
a plug-flow BOD5 removal model (first-order kinetics) and Darcy’s law yields an areal bed design
figure of 5 m2 per person for secondary treatment. The dimensions for the reed bed on Site A were
set at 6 x 2.5 x 0.6 m (l x b x h) accordingly. An impervious butyl rubber liner was used to seal the
reed bed before it was filled with 5 – 15 mm washed pea gravel and planted with Phragmites
Australis (common reed) in blocks of 4 per m2.
U/S Groundwater BH
Lysimeter
Tensiometer
Distribution
Box 3
HF Reed Bed
Septic Tank
16m
16m
Percolation Trenches Receiving STE
Distribution
Box 1
Weather Station
Percolation Trenches Receiving SE
Rotating
Biological
Contactor
Rain Gauge
Distribution
Box 2
Percolation Trenches Receiving SE
HF Reed
Bed
Weather
Station
Rain
Gauge
Distribution
Box
20m
D/S Groundwater BH
Figure 1. Schematic diagram of Study Site A and
instrumentation layout (6 PE)
Figure 2. Schematic diagram of Study Site B
and instrumentation layout (4 PE)
ANALYSIS AND RESULTS
Method of Soil Moisture Sample Analysis
As chloride (Cl) does not play a significant role in any geochemical reactions, its indicative use in
the research was as a tracer to ascertain the extent of the wastewater effluent spread across the
percolation area. The results of the laboratory analysis for Cl at the various sample positions along
all sets of monitored trenches on both sites were averaged at the same depth plane at which they
were recorded. A clear indication could thus be found of any potential differences in loading rates
between sampling distances on the same depth plane aswell as highlighting any variances between
the depth planes.
Analysis of the Cl data across the percolation trenches receiving STE on Site A revealed consistent
effluent loading at the 0, 5 and 10 m sample positions along all three depth planes. This was in
contrast to the 15 and 20 m sample positions where steadily low Cl levels suggested the effluent had
not yet reached these points (Figure 3). Further scrutiny of the soil moisture tension values from the
tensiometers installed on the percolation trenches at the front (2.5 m), middle (7.5 m) and end (17.5
m) showed that the effect of dilution by recharge events on contaminant concentrations on the first
half of the percolation area (0 – 10 m) was minimal. As such, physical, chemical and biological
processes were the predominant mechanisms in the attenuation of effluent at the 0, 5 and 10 m
sample positions. This was in contrast to the 15 and 20 m sample positions where it was suspected
little effluent had been recorded. At these positions in the percolation area, the tensiometers reacted
to the variations in effective rainfall over the sampling period (Figure 4).
350
140
0m (0.3m)
Soil Moisture Tension (mBar)
-1
Cl Loading (g d
70
5m (0.3m)
10m (0.3m)
250
15m (0.3m)
20m (0.3m)
200
150
100
50
8.00
0
16.00
-70
-140
24.00
-210
32.00
-280
0.3m
0.6m
1.0m
05/10/2007
06/08/2007
07/06/2007
08/04/2007
07/02/2007
09/12/2006
10/10/2006
11/08/2006
12/06/2006
04/12/2007
05/10/2007
06/08/2007
07/06/2007
08/04/2007
07/02/2007
09/12/2006
10/10/2006
11/08/2006
12/06/2006
13/04/2006
12/02/2006
Figure 3. Average Cl loading rates at 0.3m
depth plane across all sample positions
on percolation trenches receiving STE (Site A)
40.00
13/04/2006
-350
0
Effective Rainfall (mm)
)
300
0.00
eff. rain
Figure 4. Soil moisture tension plotted
against effective rainfall for the 12.5 m
tensiometer position (Site A)
Similar Cl analyses was also carried out on the percolation trenches of Sites A and B receiving
secondary effluent which showed that the majority of the effluent had been recorded between 0 5m and 0 - 2.5m respectively. This trend of increased concentrations at the front end of these
trenches, one suspects, is due to the reduced organic loading as a result of the additional secondary
treatment provided by the reed bed (Site A) and RBC (Site B). This has inhibited the formation of a
biomat preventing distribution of the effluent along the base of the trenches. A recent CCTV
investigation of the pipework has endorsed these findings where a mass of organic material was
only visible along the base of the pipework on those areas predicted to be receiving effluent.
While the attenuation of the percolating effluent is mostly due to physical, chemical and biological
processes within the subsoil, the contribution of dilution by recharge must also be considered. The
effective rainfall reaching the depths of the trenches was calculated from which more rigorous
loading rates of the different parameters in the subsoil could be generated. The reduction in Cl
concentration between the STE (or SE) and the different depth planes was examined as another
method of quantifying this dilution effect. This reduction in Cl concentration with depth, however,
is not entirely due to dilution as it also results from the removal of colloidal matter in the
percolating effluent by the combined straining action of, and adsorption by, the subsoil and
associated biomat. The mean reductions in Cl concentration between the STE (or SE) and the 0.3 m,
0.6 m and 1.0 m depth planes were therefore determined over a period when the contribution of
effective rainfall was zero. Accounting for this loss in Cl concentration the additional reduction in
concentration could then be attributed to dilution by effective rainfall. The average dilution of the
effluent at the different depths on those days that received effective rainfall can be seen in Table 1.
Further mass balance calculations were carried out to determine the zone of contribution of
effective rainfall at each depth plane and thus to estimate the dispersion of the effluent plume below
each percolation trench (Table 2). As expected, the zone of contribution increases with depth given
the natural dispersion of the effluent plume.
Table 1. The average reduction in Cl
effluent concentration due to dilution
on those days receiving effective
rainfall
Depth
Plane
Site A
(STE)
Site A
(SE)
Site B
(SE)
0.3 m
0.6 m
19%
30%
32%
35%
33%
34%
1.0 m
37%
42%
34%
Table 2. Calculated length and dispersion of effluent
plume at both sites
Width of plume (m) from edge
of trench at depth plane
Length
(m)
0.3 m
0.6 m
1.0 m
Site A (STE)
11
0.2
0.5
0.75
Site A (SE)
7
0.5
0.75
0.9
Site B (SE)
2
0.35
0.4
0.4
Chemical Analysis
Tables 3 to 6 list the average loading rates of each chemical parameter across the reed bed and the
designated depth planes through the percolation areas on each site. In the case of the subsoil, the
loads at each depth for all parameters were calculated as before taking into account the effect of
dilution by effective rainfall over the zone of contribution.
Table 3. Average loading rates of chemical parameters across the percolation trenches receiving STE (Site A)
Parameter
Loading Rate (g d-1)
STE
0.3 m plane
0.6 m plane
1.0 m plane
COD
200
39.3
33
29.8
NH4-N
28.7
5
2
0.6
NO2-N
0.1
0.1
0.1
0.1
NO3-N
0.4
9.3
10
11
Ortho-P
5.6
0.3
0.2
0.2
pH
7.5
7.9
8.0
8.0
Table 4. Average loading rates of chemical parameters across the horizontal flow reed bed and subsequent
percolation trenches (Site A)
Parameter
Loading Rate (g d-1)
Reed Bedin Reed Bedout
0.3 m plane
0.6 m plane
1.0 m plane
COD
173
83
23.6
22.9
18.5
NH4-N
23.9
22.3
0.6
0.5
0.3
NO2-N
0.1
0.02
0.1
0.1
0.03
NO3-N
0.3
0.1
13.9
11.4
8.6
Ortho-P
4.4
2.6
0.3
0.3
0.2
pH
7.5
7.4
8.3
8.1
8.1
Table 5. Average loading rates of chemical parameters across the percolation trenches receiving secondary
(RBC) effluent (Site B)
Parameter
Loading Rate (g d-1)
SE
0.3 m plane
0.6 m plane
1.0 m plane
COD
26.1
9.1
5.9
6.4
NH4-N
2.3
0.1
0.1
0.1
NO2-N
0.1
0.1
0.4
0.2
NO3-N
4.6
10
7.6
7.9
Ortho-P
4
0.4
0.1
0.2
pH
7.3
7.0
7.1
7.8
On Site A, the concentrations were averaged over the 0, 5 and 10 m sample positions for the
percolation trenches receiving STE and over the 0, 2.5 and 5 m sample positions for the percolation
trenches receiving the reed bed discharge. The 0 m sample position concentrations alone were
averaged for the percolation trenches receiving SE on Site B. Loading rates at the inlet and outlet of
the reed bed were calculated according to a mass balance study to incorporate the effects of
precipitation and evapotranspiration.
The installation of the constructed reed bed as a secondary treatment system on Site A has reduced
the COD loading on the subsequent percolation area by 52%. When compared with comparable
stand alone secondary treatment systems on the Irish market (e.g. peat filter) (Gill et al. 2005), the
reed bed appears to under-perform considerably with regard to organic removal. An average COD
effluent concentration of 230 mg L-1 suggests it to be on the high side of a secondary treated
effluent. A subsequent dye tracer test using rhodamine WT has shown there to be significant
channelling within the reed bed which has compromised its full treatment capabilities. On all
percolation trenches of sites A and B, an organic load reduction of at least 65% was seen to have
occurred in the first 0.3 m of the subsoil. In the case of the percolation trenches receiving STE this
breakdown is presumably due to the attenuation and filtering capacity of the extensively formed
biomat layer, while it is assumed predominant aerobic heterotrophic biodegradation and
physiochemical processes are responsible for the rapid load reduction in the trenches treating SE.
TKN measurements at the septic tank outlet on Site A clearly showed that the majority of the
nitrogen (80%) in the raw effluent underwent ammonification in converting to its inorganic NH4+
form. Within the first 0.3m of subsoil, extensive nitrification also occurred on all percolation areas
while the majority (80% and 59%) of the overall Total-N load reduction in the subsoil of Site A
occurred at this same depth in the trenches receiving STE and SE respectively. Significantly, the
research has highlighted the poor levels of denitrification that occur with depth in these freelydraining subsoils; most likely due to the effluent’s short retention time in anoxic zones of the soil
owing to its excessive percolation rate. Little nitrification occurred, as expected, in the HF reed bed
with minimal reduction in the Total-N load to the subsequent percolation trenches. As a result and
of worry, was the effect the whole treatment system of Site A had on nitrate (NO3-) levels in the
groundwater immediately downstream. Almost an eightfold increase (5.3 mg l-1 to 39.7 mg l-1 NO3-)
in average concentrations was measured between U/S and D/S groundwater monitoring points
during the sampling period.
The attenuation of ortho-P within the subsoil treatment system is a function of the soil’s texture and
mineralogy while its removal from the percolating effluent is controlled by soil adsorption and
mineral precipitation. The ability of a soil to fix phosphorous is highly dependent on its clay
content. It was rather surprising, therefore, to find low concentrations of ortho-P in the top layer of
the subsoil and throughout given its freely draining nature on both sites. A reduction in load of 95%
was measured between the STE and the 0.3 m plane (Site A) coupled with a 90% (Site A) and 97%
(Site B) drop between the SE and 0.3 m plane. Arias et al. (2001) has suggested that the P-removal
efficiency with respect to sands is often high initially but then decreases after some time after the Psorption capacity of the sand is used up. The more likely reason in the initial P-reduction on Site A,
however, can be found in the on-site trial hole where, on closer inspection, it was noted that an
overlying layer of sandy gravely SILT/CLAY had slightly protruded to below the invert of the
percolation trenches and as such the clay content of the subsoil initially receiving effluent was
increased locally. A combination of sorption onto the media and biomass growth is said to enhance
P-removal in the early stages of a wetland which may be suggestive of the 40% decrease in load
across the reed bed to date. The nutrient uptake, mobilisation and storage by the macrophyte roots
and rhizomes is thought to be negligible given their limited growth and expansion in the bed
substrate.
Bacteriological Analysis
Table 7 presents the E. coli concentrations measured at the septic tank and reed bed outlet in
tandem with measurements on all sets of percolation trenches on both sites.
Table 7. Concentrations of and reductions in E. coli across the reed bed and depth planes (Site A and B)
No. of Samples with E.coli concentration (MPN 100mL -1 )
E.coli count
Average
(MPN 100mL-1)
SITE A
STE
Reed Bed (SE)
5.62 x 106
6.92 x 104
No. of
Samples
<10
10-100
100-1000
>1000
0.3m plane
0.6m plane
1.0m plane
47
37
41
42
32
39
4
2
2
1
3
0
0
0
0
SE
0.3m plane
0.6m plane
1.0m plane
42
57
47
40
55
43
2
2
1
0
0
1
0
0
2
SE
0.3m plane
0.6m plane
1.0m plane
24
19
21
20
17
19
3
2
1
0
0
1
1
0
0
log10 Reduction
STE
~2
SITE B
The installation of the HF reed bed on Site A has greatly reduced the bacterial load on the
subsequent percolation trenches with an average 2-3 log removal (99.5%) in total coliforms and 2
log removal (99%) in E. coli reported across the bed. Despite this impressive performance,
relatively high concentrations of pathogens still remain in the SE which would deem it unsuitable
for discharge to groundwater prior to further (tertiary) treatment in the subsoil.
The bacteriological results from the percolation trenches underline the clear ability of the highly
permeable subsoil and its associated biomat to remove enteric bacteria from the percolating
effluent. There were, however, isolated incidences of E. coli contamination with depth in both sites.
Two of the lysimeters on the 0.6 m depth plane of the percolation area receiving STE on Site A
experienced high E. coli counts (>100 MPN 100 mL-1) on an early sampling date. However, a
reduction in enteric bacteria concentration with time was evident at this position suggesting the
initial occurrence may have been due to the presence of a preferential flowpath which initiated the
free movement of the bacteria to the lysimeter’s porous cup but which became blocked by biofilm
over time thereafter. Negligible E. coli readings were also recorded with depth along the trenches
receiving SE on Site A apart from isolated incidences of bacteriological contamination on the 1.0 m
depth plane. BS5930 (British Standards Institute, 1981) analysis of soil samples at this depth
showed the lysimeter cup to be on the boundary of a sandy gravelly SILT and sandy GRAVEL.
With the subsoil having a high sand content, it’s possible that with an associated large percentage of
pore spaces, bacteria migration through the local subsoil may have occurred. Coupled with this is
the likelihood that the assumed reduced biomat formation under these trenches had the effect of
increasing the hydraulic conductivity and load per unit area. The occurrence of such an incident had
a resulting negative impact on the groundwater quality at this time with a high of 360 MPN 100mL1
recorded at the downstream groundwater monitoring point.
CONCLUSIONS
Analysis has shown the biomat development under the percolation trenches of an on-site
wastewater treatment system to differ with applied effluent strength. SE, with its reduced organic
content, reached only the front (0 – 5 m at most) of the percolation trenches in contrast to the
percolation trenches receiving highly concentrated STE where the flow had spread to a distance of
at least 10m. Close examination of the reduction and removal of both chemical constituents and
microbial indicator organisms from separately applied STE and SE at various depths across the
freely-draining subsoil show there to be similar performance despite contrasting organic and
bacterial influent loadings. The evidence from this research indicates that the majority of organic,
ortho-P and Total-N load removal occurs in the distribution gravel around the percolation pipe and
across the first 0.3m in the subsoil, while denitrification is minimal with depth in such soil types.
The performance of the HF reed bed to date as a stand-alone secondary treatment system has been
shown to be very limiting in an Irish context with regard to organic removal when the STE is highly
concentrated (max. COD concentration in the STE measured at 940 mg l-1). Given the associated
lack of nitrification and nominal Total-N load removal across the bed, the only true benefit to date
in the installation of a reed bed prior to a receiving polishing filter is in the removal of the STE’s
indicator bacteria.
ACKNOWLEDGEMENTS
The authors wish to acknowledge the Irish Environmental Protection Agency for funding this
research study under the Environmental Research Technological Development and Innovation
(ERTDI) Programme as part of the National Development Plan 2000-2006.
REFERENCES
Arias, C.A., Del Bubba, M. and Brix, H. (2001). Phosphorous Removal by Sands for Use as Media in Subsurface Flow
Constructed Reed Beds. Wat. Resources, 35(5), 1159-1168
Bitton, C. and Gerba, C.P. (1984). Groundwater Pollution Microbiology, John Wiley & Sons Inc.
British Standards Institute. (1981). BS 5930-1981 Code of Practice for Site Investigations. British Standards Institute,
London.
Cooper, P.F., Job, G.D., Green, M.B. and Shutes, R.B.E. (1996). Reed Beds and Constructed Wetlands for Wastewater
Treatment. Water Research Centre (WRc), Swindon, U.K.
Daly, D. (2003). Editorial. p.1. In Geological Survey Ireland Groundwater Newsletter. No. 42. GSI Dublin, Ireland
Department of the Environment and Local Government (DoELG), Environmental Protection Agency (EPA) and
Geological Survey of Ireland (GSI). (2004). Groundwater Protection Schemes.
EPA. (2000). Wastewater Treatment Manuals: Treatment Systems for Single Houses. Environmental Protection
Agency, Wexford, Ireland
Gill, L., Súilleabháin, C. Ó., Johnston, P. and Misstear, B. (2005). An Investigation into the Performance of Subsoils
and Stratified Sand Filters for the Treatment of Wastewater from On-Site Systems. Research Project 2000-MS-15M1 Final Report. Environmental Protection Agency, Wexford, Ireland
Griggs, J. and Grant, N. (2001). Reed Beds for the Treatment of Domestic Wastewater. CRC, London.
Healy, M. and Cawley, A.M. (2002). Wetlands and Aquatic Processes – Nutrient Processing Capacity of a Constructed
Wetland in Western Ireland. Journal of Environmental Quality 31, 1739-1747
Mulqueen J., and Rodgers, M. (2001). Percolation Testing for Hydraulic Conductivity of Soils for Percolation Areas.
Water Research, 35(16), 3909-3915.
Siegrist, R.L. (1987). Soil Clogging During Subsurface Wastewater Infiltration as Affected by Effluent Composition
and Loading Rate. Journal of Environmental Quality 16:181-187
Vymazal, J. (2001). Removal of BOD5 in Constructed Wetlands with Horizontal Subsurface Flow : Czech Experience.
Water Science and Technology, 40(3), 133-138