Vol.

7
No 3-4, 191-200
2007

Wastewater treatment
Constructed wetlands for grey water treatment
in wetlands:
Theoretical and
practical aspects

Lisa M. Avery1,4, Ronnie A.D. Frazer-Williams1, Gideon Winward1,

Chris Shirley-Smith2, Shuming Liu3, Fayyaz A. Memon3, Bruce Jefferson1
1School of Water Sciences, Cranfield University, Cranfield, Bedfordshire SG17 5RJ.
e-mail: [email protected]
2Water Works UK, Water Works UK Ltd, 21 Cornwall Avenue, London, N22 7DA.
e-mail: [email protected]
3School of Engineering, Computing and Mathematics, Harrison Building, University of
Exeter, North Park Road, Exeter, EX4 4QF.
e-mail: [email protected]
4Current address: Macaulay Insitute, Craigiebuckler, Aberdeen, AB15 8QH.
e-mail: [email protected]

Abstract
A vertical flow and horizontal sub-surface flow wetland were compared with a novel
system (the GROW green roof water recycling system) for treatment of low-organic
strength domestic grey water for reuse. The vertical and horizontal flow wetlands
were planted with Phragmites australis in a sand:soil:compost medium. The GROW
system was planted with a variety of marginal plants selected for treatment and aes-
thetic characteristics. BOD removal did not differ significantly between the three
systems which all consistently met USEPA standards for grey water for reuse.
However, the GROW system was most effective at removal of suspended solids and
turbidity (mean removal 91.2% and 98.2% respectively) and both GROW and VF
were more effective at removing pathogens (4.2 and 4.8 log reduction of total coli-
forms respectively) than the HF (2.7 log reduction).
Key words: E. coli, green roof water recycling system, indicator microorganisms,
recycling, reuse, water quality

1. Introduction the exception of toilet waste (black water),

although kitchen waste is often excluded due to
Increasing pressure on water resources its high oil and food waste content. This can be
world-wide has raised interest in recycling and more difficult to degrade and can have a higher
reuse of wastewater as a demand management pathogen content. Grey water tends to be more
strategy. One option for implementing the reclai- lightly polluted than black water, particularly in
mation of wastewater is through reuse of domes- terms of suspended solids and pathogens
tic grey water for non-potable applications such (Jefferson et al. 2000) and it therefore should
as toilet flushing within small new developments require less extensive treatment prior to reuse.
or multi-occupancy buildings. Approximately one-third of domestic water con-
Grey water can include all domestic waste sumption is used for bathing, showering and
water (washing, laundry and kitchen water) with hand-washing; thus treatment and a further third
192 L. M. Avery et al.

tends to be used for toilet flushing (Dixon et al. 2. Materials and methods
1999). Consequently, reuse of grey water for toilet
flushing provides considerable savings in potable Constructed wetlands
water.
While water for non-potable applications Three small-scale constructed wetlands sys-
such as toilet flushing or garden irrigation need tems were established in June 2004 at Cranfield
not meet the stringent water quality requirements University. One horizontal sub-surface flow (HF)
of potable water, untreated grey water can still reed bed and one vertical flow (VF) reed bed
carry risks and some degree of treatment is neces- (‘RIBS’ – Oceans-ESU, Bradford, UK) were
sary (Yates, Gerba 1998). Water quality guidelines planted with Phragmites australis in a
for reuse of domestic grey water vary worldwide. sand:soil:compost medium (ratio 65:25:10) with
The USEPA standards are regarded as stringent, coarse (20 mm) gravel around the inlet zone for
and even stricter targets are imposed in some US the HF bed and around the collection/outlet zones
states, for example California. Legislation varies for both beds (Fig. 1 a and b). Both beds were
throughout Europe, with some countries including otherwise identical with a surface area of 6m2 and
the UK providing no water quality targets and lit- media to a depth of 0.7 m.
tle guidance regarding the level of treatment A third and novel constructed wetland sys-
required for reuse. Where standards exist, they tem, the ‘GROW’ Green Roof Water Recycling
focus primarily upon microbial indicator organ- System (WWUK, London, UK) was established
isms (total/faecal coliforms; E. coli), organic con- concurrently. The GROW system is designed to
tent (BOD), turbidity/suspended solids and pH. sit on a pitched roof suitable for use in urban new
Treatment of domestic grey water for reuse developments where ground space is limited.
should be sustainable and affordable and the low However, with the use of a wooden frame, GROW
capital and maintenance requirements could make may be used on a flat roof or at ground level. The
constructed wetlands a suitable small scale treat- test rig comprised five rows of two troughs con-
ment option for domestic greywater. nected in series and placed onto a tiered wooden
This study formed part of a larger project in framework to represent a sloping roof. The first
which a range of leading-contender treatment trough was approximately 1m above the ground
technologies were being assessed for treatment of and the lowest was 0.7m above ground (Fig. 2). A
grey water. The current paper reports on the per- baffle and a weir within each trough are intended
formance of three differently configured con- to force the flow through the whole of the media,
structed wetland systems treating domestic grey reducing the potential for short-circuiting. The
water over the first nine months of operation. troughs were aerated for one hour each day via a

To inspection caps
a.
110 mm slotted pipe
Gravel (20-40mm)
Outlet

Pipe location

To inspection caps

b. Fig. 1. System schematic outline:

a. Horizontal flow collection pipes. Pipes
are buried in a 20–40 mm φ washed gravel
mound, followed by sand/soil/compost
Outlet mix (ratios – 65/25/10) to 100 mm from
top (adapted from literature supplied by
Oceans-ESU Ltd.),
b. Vertical flow collection pipes. Pipes are
110 mm slotted buried in 150mm (from base) 20–40 mm φ
To overflow cowl pipe washed gravel, followed by around 700
mm sand/soil/compost mix (ratios –
65/25/10). (adapted from literature sup-
plied by Oceans-ESU Ltd.).
 Constructed wetlands for grey water treatment 193

tem was increased to 480 dm3 d–1

applied on the same basis. Nominal
hydraulic retention times are given
in Table I.

Grey water source

Influent grey water was collect-
ed from eighteen specially plumbed
student flats on the Cranfield
University campus. Water from
baths, showers and bathroom basins
drained to a communal sump from
which it was pumped underground to
two inter-connected holding tanks
using a submersible pump with a
level control. Both the sump and the
holding tanks had overflow connec-
Fig. 2. Computer-aided representation of ‘GROW’ Green roof water tions to sewer. The holding tanks
recycling system (by kind permission of Frank Parry – Proteus Graphic were mixed using a submersible
Design). The system comprises 5 tiered rows each comprising two pump and a recirculation system. The
troughs. Troughs contain ‘Optiroc’ expanded clay medium to a depth of system provided a real grey water
aprox. 10 cm topped with 10–20 mm diameter washed gravel. source on which any technology
could be tested.
porous hose. Troughs were filled to approximately
10 cm depth with Optiroc (light-weight expanded Sampling and analysis
clay) and topped with gravel chippings (10–20mm
diameter) to 16 cm. Troughs were planted with the Influent and effluent streams were sampled
following aquatic species; trough 2 with 6 plants twice-weekly between 09:00–10:00 hours. Water
of Iris pseudocorus; trough 3 with 8 plants of quality parameters monitored were BOD5, COD,
Veronica beccabunga; trough 4 with 6 plants of SS, turbidity, water temperature, pH, dissolved
Glyceria variegates; trough 5 with 6 plants of oxygen and indicator organisms (total coliforms,
Juncus effusus; trough 6 with 6 plants of Iris ver- E.coli and faecal coliforms). Additional parameters
sicolor; trough 7 with 7 plants of Caltha palustris; measured forntnightly included NH4-N, NO3–N
trough 8 with 8 plants of Lobelia cardinalis; and PO4–P. Temperature was determined in-situ
troughs 9 and 10 each with 7 plants of Mentha whilst pH and dissolved oxygen of water samples
aquatica. The outflow pipe carrying the treated were analysed using the Hanna H1 8424 micro
GROW effluent is at the end of trough 10 which computer pH meter and Jenway 9071 portable dis-
also contains a media restrainer mesh. The entire solved oxygen meter immediately after sample col-
system was covered with a reinforced membrane lection. All other analyses were performed within
to prevent rainwater entering the system. The one day of sample collection. COD, NH4–N,
influent water enters the system via the inlet well NO3–N and PO4–P were analysed using the Merck
into the first trough. Upon reaching the end of cell vial spectroquant method. BOD5 was deter-
each row, it flows from a well, down a weir to the mined by the 5 day incubation method (Environment
subsequent row. At the inlet of the first trough is a Agency, 1988). Indicator organisms were initially
removable filter which removes materials such as enumerated using the Colilert method but this
hair and other debris that might cause clogging. method was replaced in November 2004 by the
Following a start-up period during which all membrane filtration method (Environment Agency,
three systems received 160 dm3 d–1 grey water 2002). A comparison of both methods yielded simi-
applied on a continuous flow basis for HF and lar results. Rainfall was determined using an on-site
GROW and supplied as ten batches over 24 hours rain gauge and weather data was collected daily for
for the VF system; the hydraulic load on each sys- the site at Cranfield Airfield.
Table I. Nominal hydraulic retetion times for each CW. Statistical analyses
Constructed Period 1 HRT Period 2 HRT
wetland (d) (d) Treatment performance during monitoring
HFRB 10.5d 3.5d period two was evaluated using analysis of vari-
VFRB 1.05d * 8.4h * ance to compare influent greywater with effluents
GROW 2.33d 18.6h from each constructed wetland. Statistical analy-
* per batch ses were performed using Genstat 7 (Statistics
194 L. M. Avery et al.

Department, Rothamsted Research, Harpenden, grey water is dosed with concentrated synthetic
UK) Microbial indicator data, turbidity data and grey water must be undertaken to test the systems
suspended solids data were log10 transformed to on higher organic strength greywater. Importantly,
met the assumptions of ANOVA. Fisher’s LSD the collection system is designed as would be
test was used to identify significant differences. expected for a fully functional system therefore the
influent is likely to reflect grey water generated in
similarly plumbed multi-occupancy buildings.
3. Results and discussion
Influent water quality Monitoring periods
Grey water is usually similar in organic The first monitoring period covered the start-
strength to domestic wastewater but has lower up of the reed beds at a low hydraulic loading rate
suspended solids and turbidity. It is also deficient (160 dm3d–1) during which vegetation and micro-
in macronutrients (N and P). These factors can organisms were allowed to acclimitise. During
limit the effectiveness of biological treatment period two, the hydraulic load was increased to
(Jefferson et al. 2000). 480 dm3 d–1. Aproximately 77 dm3 of grey water
The influent grey water during this study was is produced per person per day in a residence
weak in terms of organic strength with a maximum occupied by 2 people (Dixon et al. 1999) thus this
BOD of 19.0±0.9 mg dm-3 measured during the provided each constructed wetland with grey
monitoring period. Mean influent water quality water equivalent to that produced by approximate-
parameters for the experimental period are shown ly six people. Differences between monitoring
in Table II, which also provides a comparison with periods one and two were not tested statistically
separate grey water samples obtained from the due to the shorter and less intensive sampling dur-
shower of one individual collected at two different ing period one (August-September 2004).
locations. This highlights the variability of grey
water. Literature values for the organic strength Effluent water quality
(BOD) of grey water range from 33–300 mg dm-3
(Jefferson et al. 2004). The COD:BOD ratio of Organics
influent grey water to the constructed wetlands There was a significant difference in BOD
tended to be higher than that of fresh individual between the influent grey water and treated effluent
shower water, suggesting that some degradation from all three wetlands systems during monitoring
was occurring in pipework prior to reaching the period 2 (Table III). There was no significant differ-
CWs. Grey water has been shown to follow first- ence in effluent BOD concentrations between treat-
order decay kinetics equating to a 50% reduction ment wetlands and effluent BOD was consistently
BOD5 over a 4-h period (Jefferson et al. 2000) below 10 mg dm-3 (Fig. 3), which is the USEPA
period. Dixon et al. (2000) showed that storage of standard for water for in-building re-use, on all but
grey water can improve water quality through set- one sampling date during the first monitoring period.
tling of suspended material, and aerobic microbial Effluent BOD was similar during both sampling
growth and anaerobic release of soluble COD. The periods. Mean BOD removal rates were 83.1, 88.2
COD strength of grey water collected from the stu- and 84.0 (period one) and 87.8, 93.1 and 91.5 (period
dent flats in this study was toward the lower range two) for HF, VF and GROW respectively. Influent
of values reported in the literature. Mean COD val- BOD load was generally higher during the first four
ues reported for different sites range from 40 to 371 months and appeared to drop in December remain-
mg dm-3 with similar vari-
ation recorded within indi- Table II. Grey water characteristics for Cranfield student flats and individual shower
waters
vidual sites (Jefferson et
al.2000; Jefferson et al. Water quality parameter Student flats Individual shower waters
2004). The low organic Mean ± SEM
strength may be due in part BOD5 (mg dm-3) 19.0±0.9 100 186
to powerful showers in the COD (mg dm-3) 81.9±4.1 292 498
student accommodation Turbidity (NTU) 19.1±2.0 21 44
leading to high dilution of Suspended Solids (mg dm-3) 25.1±3.8 – –
the grey water. The impli- Total coliforms [log10 (y+1)] 5.5±0.1 – –
cations for this study are E. coli [log10 (y+1)] 3.0±0.1 – –
that in order to ascertain Faecal Enterococci (log10) 2.9±0.2 – –
the true potential of con- pH 7.3±0.0 7.2 7.5
struced wetlands for grey Nitrate (mg dm-3) 1.6±0.3 7.5 6.3
water treatment, further Ammonia (mg dm-3) 1.2±0.3 0.4 1.0
work in which influent Phosphate (mg dm-3) 0.7±0.1 0.3 1.3
 Constructed wetlands for grey water treatment 195

ing lower for the following five months (Fig. 3; R2 during period two. Effluent from GROW showed
0.15; P<0.01). There is no clear explanation for this little change in turbidity, exhibiting superior per-
trend, but it may be attributable to changes in occu- formance compared with the reed beds and meet-
pancy of the flats and/or seasonal changes in water ing the USEPA standard of 2 NTU consistently
or product use behaviour. There was no evidence (Fig. 4). The VF reed bed removed on average
that this trend was reflected in effluent BOD content. 41.5 and 31.5% of the turbidity during monitoring
Effluent BOD was, however, notably greater during periods one and two respectively. The HF reed
August 04 than during any other month (August 04 bed performed poorly during the first monitoring
means ± SEM; 3.2±1.0, 2.3±1.8, and 2.4±1.8 for HF, period, actually increasing the effluent turbidity
VF and GROW respectively). compared with the influent grey water (Table III).
COD measurements during the second moni- A qualitative observation was made that precipi-
toring period reflected a similar trend to BOD tates of iron oxide appeared in the effluent, par-
whereby influent COD was significantly greater ticularly during the start-up period. This was
than effluent for all three constructed wetlands, attributed to operational difficulties with the HF
however GROW appeared to be slightly more bed including a leak which led to periods during
effective at removing COD than the HF (Table III; which the bed was not saturated. It seems likely
P>0.001). Although there was a trend towards that this allowed the development of biofilms of
more effective removal of COD by VF than HF, iron-oxidising bacteria at the interface between
there was no significant difference overall. aerobic and anaerobic zones (Weiss et al.1995).
Interestingly, effluent COD was noticeably greater Once rectified, effluent turbidity improved (period
during monitoring period one, and the change in 1 mean turbidity 42.2±9.1; period two mean tur-
removal rates during the second period was more bidity 10.3±1.0). Neither the VF or HF reed beds
pronounced for all three wetlands (Table III) than reliably met the USEPA standard (Fig. 4). It seems
for BOD (percentage removal for HF, VF and likely that this was a reflection of the media uti-
GROW were 35.5, 45.5 and 58.6 (period one) and lized in the reed beds. Finer particulates present in
69.9, 80.9 and 80 (period two) respectively. This the sand:soil:compost mix of the reed beds are
indicates that either the change in hydraulic load, likely to be passed into the effluent. The larger
development of vegetation and microbial commu- diameter media of GROW were less likely to con-
nities or changes in environmental factors led to tribute to effluent turbidity. Furthermore, the inlet
more efficient removal of some of the more filter on the GROW system may reduce turbidity
recalcitrant organics. Influent COD tended to on entry to the rig, although accumulation of fine
decrease over time, but the trend was not signifi- floc-like material similar in appearance to that
cant (R2 =0.03; P>0.05). found in the influent grey water was evident par-
ticularly within the first (unplanted) trough of
Turbidity and suspended solids GROW. The incorporation of baffles and wells
Differences in both influent and effluent tur- with overflow weirs into the GROW design may
bidity were apparent between the two monitoring provide a physical barrier to finer particulates
periods. Influent turbidity decreased from a mean being carried through the bed. For example, at the
of 27.9±9.2 NTU in period one to 17.4±1.6 NTU low flow rates employed during the study period it
14 100
90
12
80
10
BOD (mg dm-3)

70
Turbidity (NTU)

8 60
50
6
40

4 30
20
2
10

0 0
04
4

5
04
4
04
4

05

05

5
5

04
04
4

05

05

5
04

-0
-0

r-0
-0

-0
-0

r-0
-0

-0

v-

c-

b-
v-

n-
p-
b-
c-

n-
p-

ct
g

ar
ct
g

ar

ay

Ap
No
Ap

De
Au

Fe
No

De

Ja
Se
Au

Fe

O
Ja
Se

M
M

1-

1-
1-

1-

1-

1-

1-

1-
1-

1-

1-
1-

1-

1-

1-

1-
1-

1-

1-

Date Date

Fig. 3. Effluent BOD following treatment of grey water Fig. 4. Effluent turbidity following treatment of grey water
through HF reed bed (filled circles), VF reed bed (open through HF reed bed (filled circles), VF reed bed (open
circles) and GROW (grey squares) on each sampling occa- circles) and GROW (grey squares) on each sampling occa-
sion. Solid line represents USEPA standard for reuse. sion. Solid line represents USEPA standard for reuse.
196 L. M. Avery et al.

Table III. Mean influent and effluent characteristics.

Period 1 Period 2
Parameter Sample Removal ANOVA
Mean ± SEM mean Mean ± Removal
efficien- F P
(n) SEM (n) efficiency
cy* value
Influent 18.1±3.7 (7) n/a 19.1±1.0 (47) a n/a 311.6 <0.001
HF effluent 3.1±1.6 (7) 83.1 2.3±0.2(47) b 87.8
BOD
(mg dm-3) VF effluent 2.1±1.6 (7) 88.2 1.3±0.1(47) b 93.1
GROW 2.9±1.6 (7) 84.0 1.6±0.2(47) b 91.5
effluent
Influent 85.0±13.4 (10) n/a 80.9±3.6 (32) a n/a 104.8 <0.001
HF effluent 54.7±8.6 (10) 35.5 24.3±3.1(32) b 69.9
COD VF effluent 46.3±11.3 45.5 15.5±2.6 (32) b 80.9
(mg dm-3) (10)
GROW 35.2±8.7 (10) 58.6 16.2±2.9 (23) b 80.0
effluent
Influent 27.9±9.2 (10) n/a 17.4±1.6 (50) a n/a 208.9¶ <0.001
b
HF effluent 42.2±9.1 (10) –51.3 10.3±1.0 (50) 40.6
Turbidity
(NTU) VF effluent 16.3±1.4 (10) 41.5 11.9±1.4 (50) b 31.5
GROW 1.9±0.4 (10) 93.3 0.3±0.0 (50) a 98.2
effluent
Influent 36.6±16.6 (9) n/a 23.1±3.5 (50) a n/a 78.8¶ <0.001
c
Suspended HF effluent 16.2±4.1 (9) 55.8 4.7±0.6 (50) 79.7
solids VF effluent 3.8±1.0 (9) 89.7 2.0±0.3 (50) b c 91.5
(mg dm-3) GROW 4.1±1.4 (9) 88.9 2.0±0.5 (50) b 91.2
effluent
Influent 7.1±0.1 (10) n/a 7.3±0.0 (50) a n/a 43.2 <0.001
HF effluent 7.0±0.1 (10) n/a 7.1±0.0 (50) b n/a
pH VF effluent 6.8±0.0 (10) n/a 7.0±0.0 (50) b n/a
GROW 7.5±0.1 (10) n/a 7.3±0.0 (50) a n/a
effluent
Influent 5.0±0.3 (9) n/a 5.7±0.1 (35) a n/a 261.8 <0.001
Total coliforms HF effluent 2.3±0.2 (9) 2.7 3.0±0.1 (34) c 2.7
(log10 VF effluent 1.1±0.1 (9) 4.0 0.8±0.1 (34) b 4.8
CFU/100cm3) GROW 2.9±0.2 (9) 2.1 1.5±0.1 (34) b 4.2
effluent
Influent 3.1±0.2 (9) n/a 2.9±0.1 (36) a n/a 181.9 <0.001
E. coli
HF effluent 1.6±0.3 (9) 1.4 0.7±0.1 (36) b 2.2
(log10
(γ+1)CFU/100 VF effluent 0.3±0.0 (9) 2.8 0.1±0.0 (36) c 2.8
3
cm ) GROW 0.4±0.1 (9) 2.6 0.2±0.1 (36) b c 2.6
effluent
Influent 2.0±0.2 (8) n/a 3.1±0.2 (36) a n/a 70.7 <0.001
Faecal
HF effluent 2.2±0.1 (8) –0.2 1.4±0.1 (36) c 1.6
enterococci
(log10CFU/100 VF effluent 2.0±0.1 (8) 0.0 0.6±0.2 (36) b c 2.4
cm3) GROW 1.1±0.2 (8) 0.9 0.5±0.1 (36) b 2.6
effluent
Influent 2.3 (2) † n/a 1.5±0.2 (14) a b 5.6 0.002
HF effluent 2.6 (2) † †
1.0±0.1 (14) a 29.3
Nitrate
(mg dm-1) VF effluent 4.2 (2) † †
2.7±0.5 (14) b –81.1
† †
GROW 2.3 (2) 1.9±0.2 (14) a b –25.2
effluent
Influent 0.9 (2) † n/a 1.3±0.4 (14) a n/a 7.13 <0.001
HF effluent 0.3 (2) † †
1.0 ±0.2 (14) a b 20.6
Ammonia
(mg dm-3) VF effluent 0.2 (2) † †
0.1±0.1 (14) b 89.3
GROW 0.1 (2) † †
0.7±0.3 (14) a b 48.8
effluent
Influent 0.6 (2) † n/a 0.7±0.1 (14) a n/a 22.7 <0.001
HF effluent 0.2 (2) † †
0.3±0.1 (14) b 62.6
Phosphate
(mg dm-3) VF effluent 0.1 (2) † †
0.1±0.1 (14) b 84.4
GROW 0.2 (2) † †
0.8±0.1 (14) a –15.1
effluent
*calculated as % removal except for microbial indicators which were calculated as log reduction. †SEM and removal effi-
ciency not calculated as too few replicate samples for period 1. Similar superscript letters denote no significant differ-
ence based on Fisher’s LSD test. N.B. ANOVA was performed on period two data only. ¶Log-transformed data used for
ANOVA but means represent untransformed data.
 Constructed wetlands for grey water treatment 197

35 4.5

4
Suspended solids (mg dm-3)

30
3.5

Log10 total coliforms

25
3
20 2.5

15 2

1.5
10
1
5
0.5

0 0
-D 4
-N 4
4

8- - 05
4

-M 5
10 -0 5
-A 4

05

-D 4
04

21 t -04
4
5

8- - 05
4

-M 5
10 -0 5
-A 4

13 - 05
24 -04

5
-0
-0

-0
0

-0

-0

-0
-0

-0
0

-0

-0

-0
g-

g-
ov
ct
ug

pr
ec

ar
eb
ep

ay
an

ov
ug

pr
ec

ar
eb
ep

ay
an
c
Au

Au
-O

-A

-O

-A
M
-F
-S

-J

-N

M
-F
-S

-J
24
1-

1-
21

19

16

13

13
29

26

19

16

13
29

26
Date Date
Fig. 5. Effluent suspended solids following treatment of Fig. 6. Total coliforms (log10)in effluent following
grey water through HF reed bed (filled circles), VF reed treatment of grey water through HF reed bed (filled
bed (open circles) and GROW (grey squares) on each circles), VF reed bed (open circles) and GROW (grey
sampling occasion. squares) on each sampling occasion.

is likely that the wells at the end of each trough were frequently similar to or slightly greater than
provided the opportunity for sedimentation to influent concentrations although the HF bed tend-
occur. There was visual evidence of accumulation ed to remove more nitrate than the VF and GROW.
of some particulate material within these wells. GROW appeared to introduce a small amount of
During monitoring period two, there was no phosphate into the effluent, but again both influ-
statistically significant difference between the ent and effluent concentrations were very low and
influent grey water and effluent from HF reed bed unlikely to pose a problem for reuse applications.
(influent mean 17.4±1.6; HF mean 10.3±1.0.).
GROW and VF reed bed both reduced turbidity Indicator microorganisms
significantly (Means: VF effluent 11.9±1.4; Indicator microorganisms feature strongly in
GROW effluent 0.3±0.0; P>0.001) with GROW reuse standards worldwide. It is accepted that to
exhibiting the best overall performance. provide reassurance over health risk associated
All three CWs reduced the influent suspend- with grey water reuse, there is a necessity to fol-
ed solids content significantly but to differing low treatment with a further disinfection step.
degrees; GROW being most effective and HF least Thus while residual indicator microorganisms
effective (Fig. 5) present in treated grey water effluent are likely to
be eliminated prior to reuse, concurrent reduction
pH, nitrate, ammonia and phosphate in their numbers and organics removal during
Both reed beds facilitated a significant, albe- treatment through CWs reduces the subsequent
it small decrease in pH (period two means 7.3±0.0, disinfection demand.
7.1±0.0 and 7.0±0.0 for influent, HF and VF Total coliforms (Fig. 6) were reduced sig-
respectively; P>0.001) with the decrease most nificantly (P<0.001) after passage through each of
pronounced in the VF bed. GROW effluent pH the three rigs, however effluent concentrations
did not differ significantly from that of the influ- were also significantly different from each other,
ent (GROW period two mean 7.3±0.0). The ranking VF>GROW>HF from most efficient to
USEPA define a pH range of 6–9 as acceptable for least (period two mean log reductions of 2.7, 4.8,
treated water for re-use. Both the influent and and 4.2 respectively). E. coli (Fig. 7) was also
effluents in this study fell consistently within this reduced significantly by treatment through all
range. three rigs (P<0.001) and again VF and GROW
Nitrate, ammonia and phosphate were pres- were more effective that HF, although there was
ent only in low concentrations in influent grey no significant difference between VF and GROW
water. The consequently high C:N ratio can limit (mean log reductions of 2.2, 2.8, and 2.6 for HF,
the treatment efficiency of biological systems VF and GROW respectively). Reductions of fae-
(Jefferson et al. 2000). All three CWs demonstrat- cal coliforms in CWs usually exceed 99% (2-log
ed some level of ammonia removal, although units) (Kadlec, Knight 1995) and this was the case
there was frequently overlap between influent and for all three CWs, with the exception of the HF
effluent concentrations (Table III). Similarly, reed bed which only achieved a mean reduction of
effluent concentrations of nitrate and phosphate 97% during period one.
 198 L. M. Avery et al.

3.5 4

3 3.5

Log10 faecal enterococci

2.5
Log10 (y +1) E. coli

2.5
2
2
1.5
1.5

1
1

0.5 0.5

0 0

4
4
4

5
5
4

5
4

5
4
-D 4
-N 4
4

8- - 05
4

-M 5
10 -0 5
-A 4

-F 5

-0
04

-0
5

-0
-0
-0

-0

-0

-0

-0
-0
-0
-0

-0
0

-0

-0

-0

ov
ct
g

ug

pr
ec

ar
eb
ep

an
g-

-
ov
ct
ug

pr
ec

ar
eb
ep

ay

Au
an

a
-O

-M

-A
-N

-D
Au

M
-A

-F
-S

-J
-O

-A

M
-S

-J

24
1-

10

8-
21

19

16

13

13
29

26
24
1-

21

19

16

13

13
29

26

Date
Fig. 7. E. coli (log10) in effluent following treatment of Fig. 8. Faecal enterococci (log10) in effluent following
grey water through HF reed bed (filled circles), VF reed treatment of grey water through HF reed bed (filled
bed (open circles) and GROW (grey squares) on each circles), VF reed bed (open circles) and GROW (grey
sampling occasion. squares) on each sampling occasion.

All three technologies also significantly of Enterococci in the reed beds in the present
reduced numbers of Faecal Enterococci (Fig. 8) in study. The larger diameter media in the GROW
influent grey water (P<0.001). Mean effluent con- system provide less surface area and larger pore
centrations ranked GROW (0.5±0.1 log10 CFU spaces thus are bacterial shape is perhaps less
100 cm-3) <VF (0.6±0.2 log10 CFU 100 cm-3) likely to be as influential in determining removal.
<HF (1.4±0.1 log10 CFU 100 cm-3) although only It is possible that actual die-off times for the dif-
GROW and HF were significantly different from ferent bacteria would become more influential in
each other. Mean log reductions were 2.6, 2.4 and this system. Furthermore, bacteria can be consid-
1.6 for GROW, VF and HF respectively. ered as particulates, indeed, particlulate content
Enterococci appeared to be removed less effi- has been proposed as a surrogate for the microor-
ciently than E. coli during passage through both of ganism content of wastewater (Chavrez et al.
the reed beds, but this was not so for the GROW 2004). Since microorganisms can be considered to
system (Table III). The shape and surface charac- behave as particulates, it is perhaps not surprising
teristics of the two different bacterial genera are that GROW is more efficient in their removal than
likely to evoke different interactions with CW the reed beds, since the same trend was evident for
media. Adhesion to solid surfaces exerts a major turbidity and suspended solids. It is possible that
influence on the transport of microbial cells again, the presence of barriers and wells within
through porous media. Although it is not clear the GROW system enhances removal of microor-
why attachment would lead to preferential reten- ganisms.
tion of cells of certain shapes and sizes, some Planting regime may also be an important
reports suggest that cell attachment to solid sur- factor in the physical and biological removal of
faces may indeed be greater for elongated cells contaminants from grey water. GROW differs
than for spherical cells. Fontes et al. (1991) dem- from the reed beds in that it possesses a variety of
onstrated that small coccoid cells had a much different plants which will not only have diverse
higher recovery rate in soil column effluents than root architechture but are also likely to support a
larger, rod-shaped cells. Conversely, Camper et al. diverse microbial populations. It is well docu-
(1993) found no correlation between cell size and mented that individual plant species support dif-
recovery in soil column experiments. Other cell ferent rhizosphere microbial communities (e.g.
surface characteristics such as electrostatic charge Bergsla-Vlami et al. 2005). Such diversity may
and hydrophobicity are likely to influence adher- facilitate more efficient degradation of organic
ence to media and thus retention in the bed, how- material along with increased competition, pro-
ever Gannon et al. (1991) found cell size to be duction of antimicrobial compounds and grazing
important and observed no relationship between pressure against pathogens in influent grey water.
retention in soil columns and these characteristics. Furthermore, the final two troughs of GROW are
It has been shown that the length to width ratio of planted with Mentha aquatica, the essential oils of
cells is also important in transport of bacteria which are reported to have antimicrobial activity
through porous media. Near-spherical shape cells (Mimica-Dukic et al. 2003). Diffusion of aromatic
were retained least within sand columns (Weiss et compounds into the aqueous phase may reduce
al. 1995). This is comparable to the lesser removal indicator organisms within the effluent. In contrast
 Constructed wetlands for grey water treatment 199

the reed beds support a single macrophyte species VF at reducing suspended solids and turbidity.
and a potentially lower microbial diversity. The GROW system and the VF reed bed were
Hydraulic retention time and path length of more effective than the HF reed bed in terms of
media through which the influent passes are major indicator micro-organism removal. Overall, the
parameters affecting effluent quality. For example, GROW system showed the greatest potential for
Vymazal (2005) demonstrated that enteric microbe meeting more stringent USEPA standards for in-
removal efficiency in CWs with emergent macro- building re-use. However, these results should be
phytes is primarily influenced by hydraulic load- interpreted with caution because the influent grey
ing rate (HLR) and the resultant hydraulic resi- water represented a low-organic strength source
dence time (HRT) and the presence of vegetation. and further work is required to determine treat-
In this study, the nominal HRT was least (per ment capability of each CW under higher organic
batch) in the VF reed bed and greatest in the loadings and under lower hydraulic retention
HFRB. This did not reflect the most effective pol- times.
lutant removal, however it is well established that
in practice short-circuiting and preferential flow
lead to variable retention times and poorer perfor- 5. References
mance. Further work will determine the actual
HRT of the CWs. Bergsma-Vlami, M., Prins, M.E., Raaijmakers, J.M. 2005.
Influence of plant species on population dynamics,
genotypic diversity and antibiotic production in the
Selection of media rhizosphere by indigenous Pseudomonas spp. FEMS
Expanded clay (Optiroc) media was selected Microbiology Ecology 52, 59–69.
for the GROW system sue to its light weight. Camper, A.K., Hayes, J.T., Sturman, P.J., Jones,
Weight is an important design condsideration for W.L., Cunningham, A.B. 1993. Effects of motility
this system as it is intended for situation on pitched and adsorption rate coefficient on transport of bacte-
roofs. Gravel chippings were used to top the ria through saturated porous media. Applied and
Optiroc to prevent it from floating and blocking Environmental Microbiology 59, 3455–3462
weirs within the system. Chavrez, A., Jiminez, B., Maya, C. 2004. Particle size dis-
A sand/soil/compost mix was supplied for tribution as a useful tool for microbial detection.
Water Science and Technology 2, 179-186.
the reed beds. The original design for European
wetlands utilized soil-based systems which are Dixon, A., Butler, D., Fewkes, A. 1999. Water Saving
potential of domestic water reuse systems using grey
still used in Europe (Kadlec, Knight 1995). Soil water and rain water in combination. Water Science
reed beds tended to be superceded by gravel beds and Technology 39, 25–32.
principally because they tended to clog easily with Dixon A., Butler D., Fewkes A., Robinson M. 2000.
high organic and particulate loads (Reed, Brown Measurement and modelling of quality changes in
1992). However, grey water composition differs stored untreated grey water. Urban Water 1, 293–
from that of tertiary sewage containing less sus- 306.
pended material therefore finer media such as Environment Agency 1988. 5 Day Biochemical Oxygen
sand/soil based beds are likely to be less suscep- Demand (BOD5) (with dissolved oxygen in waters).
tible to clogging with this type of wastewater. 2nd Ed. Methods for the Examination of Waters and
associated Materials. Standing Committee of
Manios et al. (2002) reported that channel forma- Analysts.
tion and short-circuiting is more prevalent in soil
Environment Agency 2002. The Microbiology of
based reed beds compared with gravel beds and Drinking Water 2002. Parts 4 and 5– Methods for the
further work is required to determine whether this isolation and enumeration of coliforms, Escherichia
is the case for sand/soil/compost mixture reed coli (including E.coli 0157:H7) Methods for the
beds treating grey water. Furthermore, slow sand Examination of Waters and associated Materials.
filters, which may be considered as comparable to Standing Committee of Analysts.
the vertical flow reed bed, are successfully used to Farooq, S., Al-Yousef, A.K., Al-Layla, R.I., Ishaq, A.M.
treat effluents of comparable or higher organic 1994. Tertiary treatment of sewage effluent via pilot
load and turbidity (e.g. Farooq et al. 1994). scale slow sand filtration. Environmental Technology
15, 15-28. 1994.
Fontes, D.E., Mills, A.L., Hornberger, G.M., Sherman J.S.
1991. Physical and chemical factors influencing trans-
4. Conclusions port of microorganisms in porous media. Applied and
Environmental Microbiology 59, 2473–2481.
Constructed wetlands could potentially pro- Gannon, J.T., Manilal, V.B., Alexander, M. 1991.
vide a useful low-tech option for treatment of Relationship between cell surface properties and
domestic grey water for reuse. All three wetlands transport of bacteria through soil. Applied and
removed between 83 and 92% of the BOD from Environmental Microbiology 57, 190–193
influent grey water and consistently met the strin- Jefferson, B., Laine, A., Parsons, S., Stephenson, T., Judd,
gent USEPA standard of 10 mg dm-3 BOD. The S. 2000. Technologies for domestic wastewater recy-
GROW system was more effective than HF and cling. Urban Water 1, 285–292.
200 L. M. Avery et al.

Jefferson, B., Palmer, A., Jeffrey, P., Stuetz, R., Judd, S. Reed, S.C., Brown, D.S. 1992. Constructed wetland
2004. Grey water characterisation and its impact on design – the first generation. Water and Environment
the selectilon and operation of technologies for Research 64, 776–781.
urban reuse. Water Science and Technology 50, 157- Vymazal, J. 2005. Removal of enteric bacteria in con-
164. structed wetlands with emergent macrophytes: A
Kadlec, R.H., Knight, R.L. 1995. Treatment Wetlands. review. Journal of Environmental Science and Health
Lewis Publishers, Florida. Part A-Toxic/hazardous substances & environmental
Manios, T., Stentiford, E.I., Millner, P.A. 2002. The engineering 40, 1355–1367.
removal of indicator microorganisms from primary Weiss, T.H., Mills, A.L., Hornberger, G.M., Herman, J.
treated wastewater in subsurface reed beds using 1995. Effect of bacterial cell shape on transport of bac-
different substrates. Environmental Technology 23, teria in porous media. Environmental Science and
767–774. Technology 29, 1737–1740.
Mimica-Dukic, N., Bozin, B., Sokovic, M., Mihajlovic, Yates, M.V., Gerba, C.P. 1998. Microbial considerations
B., Matavulj, M. 2003. Antimicrobial and antioxidant in wastewater reclaimation and reuse. In: Asano, T.
activities of three Mentha species essential oils. Planta [Ed.] Wastewater Reclamation and reuse. Technomic
Medica 69, 413–419. Publishing Company Inc, Pennsylvania, USA. (Vol
10) pp. 437–520.