Water Air Soil Pollut (2012) 223:22632275

DOI 10.1007/s11270-011-1021-4

A Comparison of In Situ Constant and Falling Head

Permeameter Tests to Assess the Distribution of Clogging
Within Horizontal Subsurface Flow Constructed Wetlands
A. Pedescoll & P. R. Knowles & P. Davies &
J. Garca & J. Puigagut

Received: 29 June 2011 / Accepted: 9 November 2011 / Published online: 9 December 2011
# Springer Science+Business Media B.V. 2011

Abstract Clogging is the main operational problem assessment of clogged areas originated from prefer-
associated with horizontal subsurface flow constructed ential flow paths within full-scale HSSF CWs.
wetlands (HSSF CWs). The measurement of saturated Discrepancy between methods (either in magnitude
hydraulic conductivity has proven to be a suitable and pattern) aroused from the vertical hydraulic
technique to assess clogging within HSSF CWs. The conductivity profiles under highly clogged conditions.
vertical and horizontal distribution of hydraulic conduc- It is believed this can be attributed to procedural
tivity was assessed in two full-scale HSSF CWs by differences between the methods, such as the method
using two different in situ permeameter methods (falling of permeameter insertion (twisting versus hammer-
head (FH) and constant head (CH) methods). Horizontal ing). Results from both methods suggest that clogging
hydraulic conductivity profiles showed that both develops along the shortest distance between water
methods are correlated by a power function (FH= input and output. Results also evidence that the design
CH0.7821, r2 =0.76) within the recorded range of and maintenance of inlet distributors and outlet
hydraulic conductivities (070 m/day). However, the collectors appear to have a great influence on the
FH method provided lower values of hydraulic pattern of clogging, and hence the asset lifetime of
conductivity than the CH method (one to three times HSSF CWs.
lower). Despite discrepancies between the magnitudes
of reported readings, the relative distribution of Keywords Reed beds . Waterflows . Clogging .
clogging obtained via both methods was similar. Design . Saturated hydraulic conductivity
Therefore, both methods are useful when exploring
the general distribution of clogging and, specially, the
1 Introduction
A. Pedescoll : J. Garca : J. Puigagut (*)
GEMMA, Department of Hydraulic, Clogging is considered to be the main operational
Maritime and Environmental Engineering, problem associated with subsurface flow constructed
Universitat Politcnica de Catalunya,
BarcelonaTech, c/ Jordi Girona 1-3, Building D1,
wetlands (SSF CWs) for wastewater treatment.
08034 Barcelona, Spain Clogging occurs because treatment results in the
e-mail: [email protected] accumulation of solids, biofilm, plant matter and
chemical precipitates which reduce the porosity of
P. R. Knowles : P. Davies
Sustainable Environment Research Group, School of
the gravel filter media over time (Brix 1997; Tanner et
Engineering and Applied Science, Aston University, al. 1998; Nguyen 2000; Wallace and Knight 2006;
Birmingham B4 7ET, UK Garca et al. 2007). In advance stages of SSF CW
2264 Water Air Soil Pollut (2012) 223:22632275

clogging, symptoms may include sludge accumula- conductivity for assessing the clogging within hori-
tion (usually near the inlet), overland flow and zontal subsurface flow constructed wetlands. Two
decreased treatment efficiency (Platzer and Mauch constructed wetlands in relatively contrasting stages
1997; Rousseau et al. 2005; Caselles-Osorio et al. of clogging (advanced clogging and unclogged) were
2007; Kadlec and Wallace 2009). evaluated with each method to highlight whether the
Solutions to reverse clogging include gravel extrac- obtained results are influenced by which method is
tion and cleaning, in situ application of chemical employed. Furthermore, hydraulic conductivity pattern
oxidants or filter medium replacement (Cooper et al. along studied wetlands is also discussed according to
2005; Murphy et al. 2009; Nivala and Rosseau 2009; design factors.
Pedescoll et al. 2009). In all cases, the financial
investment involved can represent a substantial fraction
of the cost of a new system (Cooper 2009), and for this 2 Methods
reason, it is essential to assess the degree of clogging in
SSF CWs to accurately determine whether intervention To assess saturated hydraulic conductivity two differ-
is required. Measures of clogging include the analysis of ent tests were used (falling head and constant head
accumulated solids in filter media (Caselles-Osorio et al. tests) which have been specifically developed to
2007; Pedescoll et al. 2009), hydrodynamic visual- measure the potentially high hydraulic conductivities
isations by means of tracer tests (Bowmer 1987; associated with wetland gravels. Moreover, the
Knowles et al. 2010) and determination of the repeatability and accuracy of the methods used have
hydraulic gradient between points in the filter media been previously described in Knowles and Davies
(from which the media hydraulic conductivity can be (2009) and Pedescoll et al. (2011). Both tests measure
estimated) (Sandford et al. 1995; Rodgers and the hydraulic conductivity through vertical cores of
Mulqueen 2006; Suliman et al. 2006). gravel, although by interpolating between sample
Ultimately, it is desirable to directly measure the points, it would be possible to predict a horizontal
hydraulic conductivity of the media, which none of conductivity pattern based on the assumption that
the aforementioned methods can achieve. However, flow would behave identically in both planes. It is
the gravel media in constructed wetlands is non- necessary to point out that if only the permeability of
cohesive, so it is difficult to extract unaltered samples the wetted gravel is of interest, the unsaturated layer
of material to perform standardised and controlled should be removed from the permeameter cell before
laboratory tests. For this reason, in situ procedures starting the experiment and the subsequent analysis
have been developed to directly measure the hydrau- modified to reflect the new test conditions.
lic conductivity of constructed wetlands (Reynolds et
al. 2000). The most commonly applied methods 2.1 Falling Head Permeameter
include the Guelph permeameter (Mastrorilli et al.
2001; Langergraber et al. 2003; Ranieri 2003), a The falling head permeameter is based on Lefrancs
constant head (CH) method (Knowles and Davies test with falling head (NAVFAC 1986) and has been
2009) and a falling head (FH) method (Caselles- used to measure the saturated hydraulic conductivity
Osorio et al. 2007; Pedescoll et al. 2009). Each of constructed wetlands (Pedescoll et al. 2009). A
method disturbs the medium to a certain degree; scheme of the experimental set-up is given in Fig. 1.
although this disturbance is small compared with that Hereinafter, the labels in square brackets refer to the
caused during the extraction, storing and transporta- components illustrated in this figure. The test consists
tion of a gravel core to be tested in lab facilities. of timing a falling column of water between points
Ideally, these methods are able to measure a wide along the height of a permeameter cell (100 cm long
range of media hydraulic conductivities, as governed and 10.5 cm diameter steel tube (Fig. 1 [a]) with a
by media properties and the stage of clogging (i.e. 0 perforated wall over the length which is immersed
500 m/day might be typical of the transition between into the gravel). The cell is hammered into granular
completely clogged and clean media). medium until the desired wetted depth within the bed
This work aims to compare two different methods is reached. A pressure probe (Fig. 1 [b]) (TNS119,
for the in situ measurement of saturated hydraulic Desin Intruments S.A., Barcelona) connected to a
Water Air Soil Pollut (2012) 223:22632275 2265

iterative procedure that can be done easily using either

the Solver or Goal-Seek functions in MS Excel.
X
tn
h 2  f h2 2 ! 0 2
t0

where,
h2 is the height of the water table level inside
the tube at time t, in metres
f(h2) is the modelled data, in metres
t is the time, in seconds
In order to obtain vertical profiles of hydraulic
Fig. 1 Experimental set-up for in situ determination of the
hydraulic conductivity with falling head permeameter
conductivity (at different depths), measurements using
the falling head permeameter can be achieved via two
laptop (Fig. 1 [c]) by means a data logger (Fig. 1 [d]) techniques: (1) measuring the hydraulic conductivity at a
(Datataker DT50, Thermo Fisher Scientific Australia first depth (K1), extracting the gravel above this depth
Pty Ltd, Melbourne), is placed inside the pipe on the and repeating the test at a second depth by further
surface of the bed and used to measure the water level submersing the tube (K2); or (2) inserting the tube at
inside the tube. The pressure value at atmospheric two different depths (K1 and KT) without extracting
pressure must be checked before the test starts. A the gravel, and deducing K2 by using Eq. 3:
minimum number of 100 measurements are recommen- 1 1 1
ded for accurate calculation of hydraulic conductivity; 3
KT K1 K2
therefore, it is recommended to take values every second
or more depending on the hydraulic conductivity Where:
expected. A pulse of water added to the tube will yield KT is the total hydraulic conductivity (obtained with
a negative exponential curve between water level and the second test), in metres per second
time, the slope of which is related to the hydraulic K1 is the hydraulic conductivity at the first depth
conductivity according to Lefrancs formula: (obtained with the first test), in metres per
d 2 ln2L=d h1 second
K ln 1
8Lt h2 and
Where: K2 is the hydraulic conductivity at the second
depth, in metres per second
K is the hydraulic conductivity of the studied
material, in metres per second
d is the diameter of the tube, in metres
2.2 Constant Head Permeameter
L is the submerged length of the tube, in metres
h1 is the height of the water table level inside the
A constant head permeameter has also been used to
tube at time zero, in metres
assess the saturated hydraulic conductivity in constructed
h2 is the height of the water table level inside the
wetland filter media (Knowles et al. 2010). A permea-
tube at time t, in metres
meter cell made of a non-perforated PVC tube (50 cm
and length) is inserted into the gravel by twisting until a
40-cm core is encapsulated. A water reservoir (with a
t is time, in seconds
graduated measuring tube) discharges water into the
permeameter cell via the Mariotte Siphon technique
The squared difference between the theoretical curve (Fig. 2), thus enabling a constant head to be maintained.
and that obtained in the field is minimised to estimate The hydraulic conductivity of the encapsulated gravel
the value of the hydraulic conductivity by means an core can be measured via Darcys Law, by timing the
2266 Water Air Soil Pollut (2012) 223:22632275
Fig. 2 Experimental set-up a. Air Inlet Pipe
a
for in situ determination of d b. Filling port
hydraulic conductivity b c. End cap
f c d. Air Port
with the constant head
permeameter method (not to e e. Fluid Level in reservoir
scale: reservoir stands f. Graduated measuring tube
u g l 1 Bar g. Reservoir
approx. 1 m off the ground). c h. 50cm long Permeameter Cell
Note: this figure has been i
j i. Drain Port
adapted from Knowles and t m j. Down pipe
Davies (2009) k k. Tripod
n l. Graduated Depth Gauge
o
m. Digital Differential Manometer
p n. Levels in manometer takeoff tubes
q o. Gravel Surface
p. Unsaturated gravel
s q. Free water surface
r r. Saturated gravel
h s. Manometer takeoff tubes
t. Mariotte Siphon Level
u. Stop cap

discharge from the reservoir required to maintain the n is the number of sections or core divisions (in
constant head. this case 2)

qL
hn is the vertical head loss across each depth, in
CELL
KT metres
ACELL hT
hT is the water level into the permeameter cell at
Where: constant head, in metres
KT is the hydraulic conductivity of the gravel
and
core, in metres per second
q is the flow in the reservoir, in cubic metres KT is the hydraulic conductivity of the gravel core,
per second in metres per second
LCELL is the submerged length of the permeameter
cell, in metres Four measurements per depth are required to
ACELL is the cross-sectional area of the permea- deduce the vertical hydraulic conductivity profile
meter cell, in square metres along the gravel core, as depicted by Fig. 3. Before
the experiment begins, record the digital manometer
and readings after they have stabilised (static water level)
hT is the total head loss across the permeameter cell and the distance the probes have been inserted into
at constant head, in metres the take-off tube. Any readings obtained should be
roughly equal. Any disparity between the readings
Inserting two piezometer take-off tubes into the will be caused by minor differences between the
permeameter cell at different depths (in this study, 20 vertical alignments of the top of the take-off tubes,
and 40 cm), within which are inserted two digital and therefore, recording the different static water level
differential manometer depth probes, allows the readings will allow these discrepancies to be
vertical hydraulic conductivity profiles to be calculat- accounted for. Once the experiment has started, record
ed from the following formula: the water level inside the permeameter cell so that the
1 total applied head across the permeameter cell can be
Kn   calculated. Finally, record the digital manometer
n hthKn T readings after they have re-stabilised (dynamic water
level). This stage must be completed before the
Where:
reservoir empties. Further details about the exact
Kn is the hydraulic conductivity at each depth, in experimental procedure and apparatus depicted in
metres per second Fig. 2 can be found in Knowles and Davies (2009).
Water Air Soil Pollut (2012) 223:22632275 2267

A Before applying constant head B After applying constant head

1 Bar 1 Bar

Sampling Depth
1
2 3

Fig. 3 Measurements that are taken during the experiment, has been omitted from the graphic. a Before applying constant
depicted for one take-off tube. Corresponding readings will head; b After applying constant head. Note: this figure has been
need to be taken in each individual take-off tube. N.B. For adapted from Knowles and Davies (2009)
clarity, the reservoir device which maintains the constant head

2.3 Wetlands Description and Measurements gravel was extracted and replaced with clean media
(notably, the replacement media was of a larger
The two tests were used to profile the hydraulic specification than the original media). All the wet-
conductivity of two different SSF CWs operated by lands are planted with common reeds (Phragmites
Severn Trent Water (a UK water utility provider) australis). One CW was chosen from each site and
in Warwickshire, UK, which are used for the each hydraulic conductivity test applied across a 44
tertiary treatment of municipal wastewater in matrix of sampling points (Fig. 4), measuring at two
decentralised wastewater treatment plants (Table 1). different depths (20 and 40 cm). Since both methods
The treatment plants at Fenny Compton (FC) and have been proven to be repeatable (Knowles and
Moreton Morrell (MM) include parallel networks of Davies 2009; Pedescoll et al. 2011), only one
tertiary SSF CWs which provide downstream support measurement per point was conducted. However,
for secondary treatment Rotating Biological Contac- because the repeatability of the falling head method
tors, although MM serves a population roughly was proven only under laboratory conditions (Pedes-
double that of FC (1,500 vs. 800 PE). The four coll et al. 2010), three measurements were carried out
CWs at MM and two CWs at FC have been in with the falling head permeameter at the inlet and
operation since 19931994, although those at FC outlet of the Fenny Compton facility. (at 20 and
underwent a partial refurbishment in February 2007 40 cm depth) in order to determine its repeatability in
due to clogging issues, whereby the top 20 cm of the field.

Table 1 Characteristics of the

wetlands studied Fenny Compton Moreton Morrell

Wastewater treatment Tertiary Tertiary

Population (PE) 800 1,500

Set in operation 1993 (replacement of the top 1994
20 cm of gravel in 2008)
Flow (m3/day) 120 134
Surface area (m2) 450 225
Length to width ratio 2.9 1.0
Depth (m) 0.6 0.6
Gravel media size (mm) 612 upper 0.2 m 39
36 lower 0.4 m
2268 Water Air Soil Pollut (2012) 223:22632275

Fig. 4 Plan view of the two studied wetlands and the matrix of sampling points. Distances are shown in metres

3 Results FC (Fig. 5), a region of preferential clogging (probably

originated from a preferential flow path) extends from
3.1 Horizontal Profile Maps where flow enters the system at the far right of the inlet
distributor (circa point A1), diagonally through the
Horizontal profile maps depicted in Figs. 5 and 6 media, towards where the water exits the system at the
represent the nominal hydraulic conductivity of the far left of the effluent collection pipe (circa point D4).
entire 40 cm gravel cores measured using each This clogged path corresponds to values of hydraulic
method, and do not delineate variations in vertical conductivity ranging from 0 to 12.5 m/day and 0.071.2
hydraulic conductivity. Horizontal hydraulic conduc- for CH and FH methods, respectively. In the case of
tivities range between 070 and 035 m/day for FC MM (Fig. 6), a low conductivity path exists along
and MM, respectively. However, the average core transect A (from point A1 to point A4), with both
conductivity of MM was three times lower than those methods returning hydraulic conductivity values
of FC (5.229.26 and 18.6521.21 m/day measured below 1 m/day. It is assumed that before the start up
with CH method for MM and FC, respectively). of the systems, the whole filter media had the same
At the studied sites both methods describe the hydraulic conductivity, since is a property related to
same hydraulic conductivity pattern. Specifically at both gravel size and shape.

Fig. 5 The horizontal 1

hydraulic conductivity A
profile (metres per day) of 2
Length

Fenny Compton obtained

with a falling head
3
permeameter and b constant
head permeameter 0 - 10 m/d
4
10 - 20 m/d
D C B A
Width 20 - 30 m/d
30 - 40 m/d
1 40 - 50 m/d
50 - 60 m/d
B
2
Length

4
D C B A
Width
Water Air Soil Pollut (2012) 223:22632275 2269
Fig. 6 The horizontal 1 1
hydraulic conductivity
A B
profile (metres per day) of 0 - 1 m/d 0 - 5 m/d
Moreton Morrell obtained 2 1 - 2 m/d 2 5 - 10 m/d

Length

Length
with a falling head 2 - 3 m/d 10 - 15 m/d
permeameter and b constant 3 - 4 m/d 15 - 20 m/d
4 - 5 m/d 20 - 25 m/d
head permeameter 3 3
25 - 30 m/d

4 4
A B C D A B C D
Width Width

3.2 Vertical Profile Maps transect, conductivities lower than 1 m/day were
registered across the top 20 cm, whereas conductivity
The obtained vertical hydraulic conductivity profiles ranged between 0 and 5 m/day in the bottom 20 cm of
for FC differ depending on the method employed, gravel. The highest conductivities were recorded
both in terms of magnitude and distribution of along transect D, with values ranging from 0.05
clogging (Fig. 7). The FH method describes that the to17.4 m/day in the top 20 cm, and from 0.15 to
hydraulic conductivity of the top 20 cm is ca. 10 100 m/day in the bottom 20 cm. According to the FH
times higher than the bottom 20 cm of gravel (65450 method, the lowest conductivities were also registered
and 0.0595 m/day, respectively). However, these along transect A (with values from 0.45 to 6.53 m/day
differences were less pronounced according to the CH in the top 20 cm and 0.020.09 m/day in the bottom
method, (090 and 0190 m/day for the top 20 cm 20 cm). The highest conductivities were recorded
and bottom 20 cm, respectively). along transect D; ranging from 0.97 to 11.75 m/day in
At MM, the vertical hydraulic conductivity profiles the top 20 cm and between 0.01 and 6.07 m/day in the
(Fig. 8) also differed depending on the method bottom 20 cm. Generally, the highest conductivities
applied. According to the CH method, a preferential were recorded at the outlet areas of both studied
flow path was registered along transect A. Along this wetlands, irrespective of the applied method.

A B
0 0
Depth

20 20 1
40 40

0 0
Depth

20 20 2
40 40

0 0
Depth

20 20 3
40 40

0 0
Depth

20 20 4
40 40
D C B A D C B A
Width Width

0 - 50 m/d 0 - 20 m/d
50 - 100 m/d 20 - 40 m/d
100 - 150 m/d 40 - 60 m/d
150 - 200 m/d 60 - 80 m/d
80 - 100 m/d

Fig. 7 The vertical hydraulic conductivity profiles (metres per day) obtained at Fenny Compton with a falling head permeameter and
b constant head permeameter
2270 Water Air Soil Pollut (2012) 223:22632275

A B
0 0
Depth

Depth
20 20 1

40 40
0 0
Depth

Depth
20 20 2

40 40
0 0
Depth

Depth
20 20 3

40 40
0 0
Depth

Depth
20 20 4

40 40
A B C D A B C D
Width Width
0 - 4 m/d 0 - 20 m/d
4 - 8 m/d 20-40 m/d
8 - 12 m/d 40-60 m/d
12 - 16 m/d 60-80 m/d
16 - 20 m/d 80 - 100 m/d

Fig. 8 The vertical hydraulic conductivity profiles (metres per day) obtained at Moreton Morrell with a falling head permeameter and
b constant head permeameter

4 Discussion both methods are useful when trying to identify

general patterns of clogging and preferential flow
4.1 Comparison of Methods paths in the primary (longitudinal) flow direction.
However, as emphasised in Fig. 9, the FH method
Profile maps were obtained by interpolating from generally produces lower hydraulic conductivity
discrete points in which hydraulic conductivity was measurements than the CH method, especially for
measured once. The repeatability and accuracy of the more clogged CW at MM. This phenomenon may
constant head permeameter was demonstrated in a be attributable to the differing manner by which the
previous study carried out at the same systems permeameter cell is immersed into gravel during each
(Knowles and Davies 2009). For the falling head method (Carter and Ball 1993). In the CH method, a
permeameter, a good repeatability was found at twisting action is used which may act to shear clog
laboratory scale with two different types of filter formations and hence artificially increase hydraulic
media (Pedescoll et al. 2011). Even so, the hydraulic conductivity of clogged media. Conversely, the FH
conductivity was checked in triplicate at two points
(at the inlet of the bedB1and at the outletD4) Table 2 Hydraulic conductivity values (in metres per day) for
in FC facility and at 20 and 40 cm depth. Results repeatability tests carried out at Fenny Compton
showed that this method is repeatable since standard
deviation is always below the order of magnitude of Sample point Depth (cm) Max Min Average SD
the averaged hydraulic conductivity and never exceed B1 20 184 153 174 17
the 30% of the average value (Table 2). 40 27 14 22 7
For each system studied, the horizontal hydraulic
D4 20 215 131 176 42
conductivity distributions obtained via each method
40 23 21 22 1
are similar (Figs. 5 and 6). This result suggests that
Water Air Soil Pollut (2012) 223:22632275 2271

1A 1B
80 35
Hydraulic conductivity (m/d) 30
60 constant head
25
falling head
20
40
15

10
20
5

0 0
0 20 40 60 80 100 0 20 40 60 80 100

Cumulative relative Cumulative relative

frequency (%) frequency (%)

2A 2B
300 400
Hydraulic conductivity (m/d)

Falling Head
250 300
Constant Head
200 200

100
150
20
100
10
50

0 0
Total depth 20 cm 40 cm Total depth 20 cm 40 cm

Fig. 9 Cumulative relative frequency distribution of the hydraulic conductivity values for each method at different
hydraulic conductivities obtained for both methods (n=16 per depths (total, 20 and 40 cm depth) in 2A Fenny Compton and
plot) in 1A Fenny Compton and 1B) Moreton Morrell. Mean 2B) Moreton Morrell

method uses a hammer action which may compact gravel strata (especially because the replacement
clog formations, thus artificially decreasing the gravel was of a larger size than the original gravel).
hydraulic conductivity of clogged media. This would However, these differences were only detected with
also explain why the disparity between the methods the FH method (with hydraulic conductivity values of
was not so apparent in the relatively unclogged media 65450 and 0.0595 m/day for the top and the
of FC. Unfortunately, no experiments were conducted bottom, respectively) (Fig. 7). Contrastingly, MM
to test this hypothesis, and therefore, further research demonstrated advanced symptoms of clogging in the
should be carried out to determine the extent of the upper gravel media, including surface sludge forma-
effect of permeameter cell insertion on hydraulic tion and overland flow, and it would therefore be
conductivity discrepancies between both methods. logical to assume that the upper gravel strata would be
Discrepancies between the methods are more less conductive than the lower gravel strata (according
clearly emphasised by comparison of the vertical to the theory regarding horizontal SSF CW clogging
hydraulic conductivity profiles, which demonstrated mechanisms described in Kadlec and Wallace (2009)).
variability in both distribution and magnitude (Figs. 7 However, this trend was only identified with the CH
and 8). At FC, the top 20 cm of gravel was replaced method, with the FH method generally suggesting
1 year before the experiments were carried out, and that values in the upper strata were more conductive.
therefore, it would be logical to assume that the upper Again, these observations may be artefact of
gravel strata would be more conductive than the lower procedural differences between the methods. The FH
2272 Water Air Soil Pollut (2012) 223:22632275

method repeats tests at different depths to obtain the between both methods since it may vary according
vertical conductivity profile. The multiple insertions to the infiltration properties of the media tested and
and potential for solids washout in the upper layers the clogging status of the wetland analysed.
during the first test may explain why the results of the Figure 10 compares the results of the horizontal
FH method suggested that the upper media at MM hydraulic conductivity profile obtained by each method
was more conductive than the lower media. In the CH to elucidate whether the discrepancy between methods
method, the values at different depths are calculated can be quantified. As evident from Fig. 10, across the
from the difference between the dynamic water levels measured range of hydraulic conductivities (from 0 to
inside each piezometer take-off tube. Thus, in order to 70 m/day), the results of the FH and CH methods are
obtain a reliable value the level inside the take-off related via a power function (FH=CH0.7821), with a
tubes must be stable; however, at very high media regression of r2 =0.76. From this correlation, it is
conductivities, it is possible that the recharge reservoir possible to calculate that differences in hydraulic
empties before the manometer completely stabilises. conductivity measurements between the two methods
To avoid this inconvenience, the reservoir should be generally differ by a factor of 13; which is an
enlarged. acceptable discrepancy (Elrick and Reynolds 1992;
Interestingly, Pedescoll et al. (2011) reported a Reynolds and Zebchuk 1996; Elrick et al. 2002).
disparity between laboratory versions of the FH and Moreover, it is worth noticing that exact values of
CH permeameter test which may support the above hydraulic conductivity are difficult to be obtained
findings. The laboratory FH test provided lower since it depends on many factors such as the
conductivity values than the CH method for low technique used or the physical characteristics of the
conductive material (a D50 of 0.9 mm sand with a media (Reynolds et al. 2000; Wilson et al. 2000).
hydraulic conductivity of 35 m/day); however, the Overall, the power function elucidates that the discrep-
opposite trend was observed in highly conductive ancy between FH and CH results increases as the
materials (a D50 of 7.1 mm gravel with a hydraulic hydraulic conductivity value increases. Furthermore,
conductivity of 265 m/day), with the CH method this function means that in clogged systems with lower
returning lower hydraulic conductivity values than the hydraulic conductivities, the CH method will be able to
FH method. Therefore, the observed effects, i.e. the describe small differences in hydraulic conductivity
CH method detecting low conductivities in the upper with a greater resolution than the FH method. Therefore,
strata of FC and the FH method detecting high the FH method is a simple application tool, useful for
conductivities in the upper strata of MM, may be roughly horizontal profiles of clogging in a wetland,
explainable due to a similar effect. Evidently, care while the CH method can further refine the measure-
must be taken while extrapolating the relation ments and obtain vertical profiles with a single

Fig. 10 Regression curve 1000.000

and 95% confidence
intervals of the comparison
between experimental 100.000 FH = CH0.7821
values of hydraulic r2= 0.7611
Falling Head (m/d)

conductivity obtained via 10.000

the falling head and
constant head methods
(n=28). Note that axes are 1.000
represented in logarithmic
scale
0.100
CH vs FH
Regression
0.010
95% confidence intervals

0.001
0.001 0.010 0.100 1.000 10.000 100.000 1000.000

Constant Head (m/d)

Water Air Soil Pollut (2012) 223:22632275 2273

measurement. Thus, the choice of each method depends all of the risers with the exception of that near point
on the type of desired information. A1 were completely clogged.
The horizontal flow patterns observed in the
present work for the surveyed systems (Figs. 5 and
4.2 Clogging Distribution 6) are in accordance with previous studies in two
different HSSF CWs for secondary treatment in Spain
The horizontal hydraulic conductivity profile for FC (Sams et al. 2009). These authors attributed the
indicate a range of values between 0 and 70 m/day, observed preferential flow paths to uneven influent
whereas the range of values at MM was 035 m/day; distribution at the inlet and to the aspect ratio (W/L)
aptly reflecting that MM was in a more advanced of the beds. They concluded that aspect ratios less
stage of clogging than FC (Figs. 5 and 6). These than one can compensate for uneven influent distri-
values are in accordance with previous studies bution by damping the influence of transverse short-
performed on the same wetlands (Knowles and circuiting. Fenny Compton and Moreton Morrell have
Davies 2009; Knowles et al. 2010) and with studies aspect ratios of 3 and 1, respectively (Table 1), and
performed at wetlands with similar characteristics in resultantly, the observed clogging profiles suggest
terms of treatment type, dimensions and substrate size that transverse short-circuiting is prevalent in these
(Caselles-Osorio et al. 2007). systems. Consequently, aspect ratio and flow distri-
Results indicated that clogging paths had devel- bution systems will greatly influence the clogging
oped in both wetlands, which generally corresponded pattern within a wetland.
to the shortest distance between where flow enters Regarding the vertical distribution of clogging in
and exits the system and thus supports the idea that each system at FC, the hydraulic conductivity at the
flow finds the path of least resistance (diagonally surface of the wetland (top 20 cm) was 10 times
from point A1 to point D4 at FC, and along transect A higher than at the bottom (due to a complete
at MM; Figs. 5 and 6). replacement of the surface gravel 1 year before the
It is thought that these flow paths are exacerbated present study) (Fig. 7a). However, at MM the relation
by inlet distributor and effluent collector design. At was opposite, with lower hydraulic conductivity at the
FC flow is distributed by a pipe with seven points and surface of the system (Fig. 7b). This can be ascribed
water flows within the pipe from transects A to D to the surface-based influent distributors that feed
(Fig. 4). Since the pipe is flatly emplaced on the these systems and result in the formation of a clog
gravel surface, there is no compensation for internal matter layer on the surface of the gravel that thickens
pipe losses and the majority of the water tends to flow with time. Indeed, during the field campaign, sludge
onto the bed near point A1. Thus, at point A1 the accumulation and overland flow were visually
registered hydraulic conductivity was the lowest for detected at the inlet area of MM. Subsurface-based
the entire inlet area, irrespective of the method influent distributors, such as infiltration chambers,
employed (Figs. 5 and 6) (0.1 and 0.07 m/day for have been recommended by some authors to prevent
constant head and falling method, respectively). this occurrence (Wallace and Knight 2006).
Water entering the effluent collection pipe would Sludge layer formation can be exacerbated by
flow along the pipe in the direction from transects A uneven influent distribution. This was evident at
to D in order to exit the system. Again, no provision MM where the inlet area in the vicinity of the only
is made for internal pipe losses and the water tends to functional riser pipe (point A1) corresponded to the
exit at point D4 where hydraulic conductivity was the thickest sludge layer accumulation. For this reason,
lowest for the entire outlet area, irrespective of the Severn Trent Water (the water utility provider) have
method used (12.57 and 0.99 m/day for constant and identified that maintenance of influent distributors is
falling head method, respectively). At MM water is paramount, and newly designed systems have incor-
distributed by a pipe with four vertical distribution porated trough style influent distributors which
risers and the influent flows along the pipe from improve distribution uniformity and are easier to
transects A to D. Larger solids tend not to travel up maintain than vertical risers (Griffin et al. 2008).
the risers and instead are conveyed towards the end of Other factors that may affect the accumulation of
the pipe, with the result that during the field campaign sludge at the inlet of HSSF CWs include deposition of
2274 Water Air Soil Pollut (2012) 223:22632275

reeds (which are not harvested) and the solids loading head method produced more logical results at Moreton
rate (Cooper et al. 2005). Note that these wetlands are Morrell (hydraulic conductivity is lowest in the upper
systems for tertiary treatment which should receive gravel layers were surface layer accumulation occurs). It
influent with relatively low BOD and TSS concen- is believed this is inherent in the difference between
trations (Cooper 2009). However, typical TSS loads constant head and falling head methods, although more
to HSSF CWs for tertiary treatment in the UK range work is required to ascertain the exact cause of this
from 2 to 30 g TSS/m2day (Knowles et al. 2011), variability.
which is similar to that described for secondary Both methods provide results which support that
treatment units (between 2.2 and 10 g TSS/m2day) the clogging pattern in these systems is highly
(Tanner et al. 1998; Caselles-Osorio et al. 2007). The influenced by the design and maintenance of the
atypically high BOD and solids loads to these tertiary influent distribution and effluent collection systems.
systems are due to the regular presence of biological Generally speaking, flow follows the shortest path
flocs (especially during wet weather) that are (that of least resistance) between entering and exiting
sloughed from secondary treatment units upstream the system, which is due to the large aspect ratios (W/L)
of the wetland. Therefore, possible solutions to and uneven influent distribution at FC and MM results
clogging include for example: (1) prevent recalcitrant in acute transverse short-circuiting. Therefore, this study
solids from entering the wetlands via filtration and emphasises the need for improved HSSF CW designs in
screening, and (2) design the wetland to ameliorate order to ameliorate the hydraulic behaviour and longev-
the hydraulic behaviour in order to control where ity of these systems.
sludge is deposited.
Acknowledgements This research was supported by the Span-
ish Ministry of Science and Innovation; projects NEWWET
(CTM2005-06457) and NEWWET2008 (CTM2008-06676).
5 Conclusions Anna Pedescoll kindly acknowledges the Spanish Ministry
of Science for her scholarships. Paul Knowles would like to
Preferential flow paths and corresponding clogging acknowledge the guidance of Dr. Paul Griffin from Severn
Trent Water and joint funding from Severn Trent Water Plc.
profiles in CWs can be assessed by the in situ
(UK) and a CASE studentship granted by the ESPRC UK
application of either constant head or falling head (ref. CASE/CNA/06/28).
permeameter methods. However, discrepancies between
the apparatus and procedures involved in each method
can affect the range of obtained results. Accordingly, for References
the horizontal hydraulic conductivity profiles (range of
values 070 m/day), the falling head method provides Bowmer, K. H. (1987). Nutrient removal from effluents by
values one to five times lower than those obtained by the artificial wetland: influence of rhizosphere aeration and
constant head method. This suggests that the CH preferential flow studied using bromide and dye tracers.
Water Research, 21(5), 591599.
method offers finer resolution than the FH method
Brix, H. (1997). Do macrophytes play a role in constructed
when studying small variations of hydraulic conductiv- treatment wetlands? Water Science and Technology, 35(5),
ity in clogged media. Despite this, the datasets obtained 1117.
via the two methods can be well-correlated by a power Carter, M. R., & Ball, B. C. (1993). Soil porosity. In M. R.
Carter (Ed.), Soil sampling and methods of analysis (pp.
function (FH=CH0.7821, r2 =0.76). It is believed that
581588). Boca Raton: CRC.
the differing method of permeameter insertion (twist- Caselles-Osorio, A., Puigagut, J., Seg, E., Vaello, N., Grans,
ing versus hammering) is largely accountable for the F., Garca, D., et al. (2007). Solids accumulation in six
discrepancy. full-scale subsurface flow constructed wetlands. Water
Research, 41(6), 13881398.
Vertical profiles of hydraulic conductivity mea- Cooper, P., & What can we learn from old wetlands? (2009).
sured with the falling head and constant head methods Lessons that have been learned and some that may have
differ both in terms of magnitude and relative been forgotten over the past 20 years. Desalination, 246,
distribution. It appears that the falling head method 1126.
Cooper, D. J., Griffin, P., & Cooper, P. F. (2005). Factors
produced more logical results at Fenny Compton affecting the longevity of subsurface horizontal flow
(hydraulic conductivity was highest in the recently systems operating as tertiary treatment for sewage effluent.
replaced upper gravel layers), whereas the constant Water Science and Technology, 51(9), 127135.
Water Air Soil Pollut (2012) 223:22632275 2275

Elrick, D. E., & Reynolds, W. D. (1992). Methods of analyzing hydraulic conductivity used to asses clogging in subsur-
constant-head well permeameter data. Soil Science Society face flow constructed wetlands. Ecological Engineering,
of American Journal, 56, 320323. 35, 12161224.
Elrick, D. E., Angulo-Jaramillo, R., Fallow, D. J., Reynolds, W. Pedescoll, A., Sams, R., Romero, E., Puigagut, J., & Garca, J.
D., & Parkin, G. W. (2002). Analysis of infiltration under (2011). Reliability, accuracy and repeatability of the
constant head and falling head conditions. In P. A. C. Raats et falling head method for hydraulic conductivity measure-
al. (Eds.), Environmental mechanics: water, mass and ments under laboratory conditions. Ecological Engineering,
energy transfer in the biosphere. Geophysical Monograph; 37(5), 754757.
Series, vol. 129 (pp. 4753). Washington, DC: AGU. Platzer, C., & Mauch, K. (1997). Soil clogging in vertical flow
Garca, J., Caselles-Osorio, A., Story, A., De Pauw, N., & reed bedsmechanisms, parameters, consequences and
Vanrolleghem, P. A. (2007). Impact of prior physico- solutions? Water Science and Technology, 35(5), 175181.
chemical treatment on the clogging process of subsurface- Ranieri, E. (2003). Hydraulics of sub-superficial flow con-
flow constructed wetlands: model-based evaluation. Water, structed wetlands in semi arid climate conditions. Water
Air, and Soil Pollution, 185, 101109. Science and Technology, 47(9), 4955.
Griffin, P., Wilson, L., & Cooper, D. (2008). Changes in the Reynolds, W. D., & Zebchuk, W. D. (1996). Hydraulic
use, operation and design of sub-srface flow constructed conductivity in a clay soil: two measurement techniques
wetlands in a major UK water utility. 11th International and spatial characterization. Soil Science Society of
Conference on Wetland Systems for Water Pollution America Journal, 60, 16791685.
Control. Indore, India, 419426. Reynolds, W. D., Bowman, B. T., Brunke, R. R., Drury, C. F.,
Kadlec, R. H., & Wallace, S. D. (2009). Treatment Wetlands & Tan, C. S. (2000). Comparison of tension infiltrometer,
(2nd ed.). Boca Raton: CRC Press. pressure infiltrometer, and soil core estimates of saturated
Knowles, P. R., & Davies, P. A. (2009). A method for the in- hydraulic conductivity. Soil Science Society of America
situ determination of the hydraulic conductivity of gravels Journal, 64, 478484.
as used in constructed wetlands for wastewater treatment. Rodgers, M., & Mulqueen, J. (2006). Field-saturated hydraulic
Desalination and Water Treatment, 1(5), 257266. conductivity of unsaturated soils from falling-head well
Knowles, P. R., Griffin, P., & Davies, P. A. (2010). Comple- tests. Agricultural Water Management, 79, 160176.
mentary methods to investigate the development of Rousseau, D. P. L., Horton, D., Vanrolleghem, P. A., & De
clogging within a horizontal sub-surface flow tertiary Pauw, N. (2005). Impact of operational maintenance on
treatment wetland. Water Research, 44, 320330. the asset life of storm reed beds. Water Science and
Knowles, P., Dotro, G., Nivala, J., & Garca, J. (2011). Clogging in Technology, 51(9), 243250.
subsurface-flow treatment wetlands: occurrence and contrib- Sams, R., Pedescoll, A., & Garca, J. (2009). Clogging
uting factors. Ecological Engineering, 37(2), 99112. distribution in two horizontal subsurface flow constructed
Langergraber, G., Haberl, R., Laber, J., & Pressl, A. (2003). wetlands through measurements of hydraulic conductivity.
Evaluation of substrate clogging processes in vertical flow Proceedings of the 3rd Wetland Pollutant Dynamics and
constructed wetlands. Water Science and Technology, 48 Control, WETPOL. Barcelona, Spain, 2024 September
(5), 2534. 2009.
Mastrorilli, M., Ranieri, E., & Simeone, V. (2001). Evaluation Sandford, W. E., Steenhuis, T. S., Parlange, J. Y., Surface, J.
of hydraulic conductivity in a Phragmites wastewater M., & Peverly, J. H. (1995). Hydraulic conductivity of
treatment plant. Water Pollution VI Modelling, Measur- gravel sand as substrates in rock-reed filters. Ecological
ing and Prediction (pp. 105111). Boston: WIT. Engineering, 4, 321336.
Murphy, C., Cooper, D., & Williams, E. (2009). Reed bed Suliman, F., French, H. K., Haugen, L. E., & Svik, A. K.
refurbishment: a sustainable approach. Proceedings of the (2006). Change in flow and transport patterns in horizontal
3rd International Symposium on Wetland Pollutant Dy- subsurface flow constructed wetlands as a result of
namics and Control, WETPOL, 2024 September 2009. biological growth. Ecological Engineering, 27(2), 124
Barcelona, Spain. 133.
NAVFAC (1986). Soil Mechanics. Design Manual 7.01. Naval Tanner, C. C., Sukias, J. P. S., & Upsdell, M. P. (1998). Organic
Facilities Engineering Command, Alexandria, Virginia, matter accumulation and maturation of gravel bed con-
USA, 389 pp. structed wetlands treating dairy farm wastewaters. Water
Nguyen, L. M. (2000). Organic matter composition, microbial Research, 32(10), 30463054.
biomass and microbial activity in gravel-bed constructed Wallace, S. D., & Knight, R. L. (2006). Small-scale constructed
wetlands treating farm dairy wastewaters. Ecological wetland treatment systems: feasibility, design criteria and
Engineering, 16(2), 199221. O&M requirements. Final Report, Project 01-CTS-5, Water
Nivala, J., & Rosseau, D. P. L. (2009). Reversing clogging in Environment Research Foundation (WERF). Alexandria,
subsurface flow constructed wetlands by hydrogen perox- USA.
ide treatment: two case studies. Water Science and Wilson, M. A., Hoff, W. D., Brown, R. J. E., & Carter, M. A.
Technology, 59(10), 20372046. (2000). A falling head permeameter for the measurement
Pedescoll, A., Uggetti, E., Llorens, E., Grans, F., Garcia, D., & of the hydraulic conductivity of granular solids. The
Garca, J. (2009). Practical method based on saturated Review of Scientific Instruments, 71(10), 39423946.