You are on page 1of 14

Available online at www.sciencedirect.

com

ScienceDirect
Journal of Hydro-environment Research 12 (2016) 91104
www.elsevier.com/locate/jher

Research papers

Hydraulic performance of a modified constructed wetland system through


a CFD-based approach
Emily Elizabeth Rengers 1,*, Jhonatan Barbosa da Silva, Paula Loureiro Paulo,
Johannes Grson Janzen
Universidade Federal de Mato Grosso do Sul, Centro de Cincias Exatas e Tecnologia (CCET), Programa de Ps-Graduao em Tecnologias Ambientais
(PGTA), Campus Universitrio, 79070-900 Campo Grande, MS, Brazil
Received 23 April 2015; revised 6 April 2016; accepted 14 April 2016
Available online 19 April 2016

Abstract
Low-cost household technologies, as horizontal subsurface flow constructed wetlands, are important to address water and sanitation needs in
the Asia-Pacific region in a more integrated and sustainable manner, and a better understanding of these technologies would benefit their
engineering design. Computational Fluid Dynamics (CFD) simulations of a modified constructed wetland system (EvaTAC) were undertaken to
determine empirical effects of geometric and flow parameters on the hydraulic performance and the effluent pollutant fraction. The CFD model
was validated by comparing the computed residence time distribution (RTD) with experimental results. RTD functions were then used to quantify
hydraulic indexes: short-circuiting, mixing, and moment. The EvaTAC is composed of an evapotranspiration and treatment chamber (CEvaT) and
a horizontal subsurface flow constructed wetland (HSSF-CW). For the CEvaT, length and the interaction between length and flow rate were the
most important factors for the hydraulic efficiency. For the effluent pollutant fraction, the most important factor was flow rate. For the HSSF-CW,
the strongest influence on the hydraulic efficiency was the length. Baffles and the interaction between length and baffles also had significant
statistical influence on the hydraulic efficiency. Furthermore, the results showed that flow rate, length, and the interaction between flow rate and
length influenced the effluent pollutant fraction significantly. Finally, a poor correlation between hydraulic indexes and effluent pollutant fraction
was obtained, indicating that the hydraulic indexes are not good predictors of the effluent pollutant fraction.
2016 International Association for Hydro-environment Engineering and Research, Asia Pacific Division. Published by Elsevier B.V. All rights
reserved.
Keywords: Residence time distribution (RTD); Hydraulic efficiency; Constructed wetland; Tracer test; Evapotranspiration and treatment chamber (CEvaT);
Flow rate

1. Introduction
Improving global access to clean drinking water and safe
sanitation is one of the least expensive and most effective
means to improve public health and save lives. The developed
countries, where water and sanitation services are nearly universal, significantly reduced water-, sanitation-, and hygienerelated diseases by the start of the 20th century by installing
water and sanitation systems. However, in developing regions,

* Corresponding author. Universidade Federal de Mato Grosso do Sul, Centro


de Cincias Exatas e Tecnologia (CCET), Programa de Ps-Graduao em
Tecnologias Ambientais (PGTA), Campus Universitrio, 79070-900 Campo
Grande, MS, Brazil. Tel.: +55 6799263120.
E-mail address: emilyrengers1@gmail.com (E.E. Rengers).
1
Permanent address: Rua Cel Zzimo, n. 222, Bloco G Ap. 14, Bairro Cel
Antonino, Campo Grande, MS 79010-340, Brazil.

water and sanitation services are still severely lacking. In the


Asia-Pacific region, for example, the percentage of persons
without access to safe drinking water and basic sanitation was
estimated to be, respectively, 8% and 40% in 2012 (United
Nations Publication, 2014). Consequently, nearly half of all
people have infections or diseases associated with inadequate
water supply and sanitation (Bartram et al., 2005). Unsafe
water, inadequate sanitation, and insufficient hygiene account
for an estimated 20% of deaths of people under age 14
(Prss-stn et al., 2008). The vast majority of people still
affected by poor access to improved drinking water and sanitation are people living in rural areas and rapidly growing,
unplanned, periurban areas, without centralized water and sanitation systems (Choffnes and Mack, 2009).
To improve these statistics researchers are developing lowcost household technologies in order to address water and sanitation needs in a more integrated and sustainable manner.

http://dx.doi.org/10.1016/j.jher.2016.04.002
1570-6443/ 2016 International Association for Hydro-environment Engineering and Research, Asia Pacific Division. Published by Elsevier B.V. All rights
reserved.

92

E.E. Rengers et al. / Journal of Hydro-environment Research 12 (2016) 91104

Nomenclature
C
C2
C()
D
d
E()
k
M
M0
M1
M2
MDI
MI
p
Q
RTD
Si
t
t10
t90
V
X
x

10

tracer concentration (mg L1)


inertial resistance coefficient (m1)
dimensionless RTD function
diffusion coefficient (m2 s1)
mean particle diameter (mm)
normalized residence time distribution (RTD)
decay rate constant
tracer mass at the inlet of the system (mg)
zeroth moment of the dimensionless RTD
function about the origin
first moment of the dimensionless RTD function
about the origin
second moment of the dimensionless RTD
function about the origin
Morril dispersion index (t90/t10)
moment index
pressure (Pa)
system volumetric flow rate (L min1)
residence time distribution
momentum source term (N m3)
dimensional time (min)
dimensional time necessary for 10% of the
tracer to leave the system
dimensional time necessary for 90% of the
tracer to leave the system
volume of fluid in the system (L)
fraction of pollutant remaining over time for
first-order reactions
fluid direction
permeability (m2)
void fraction (dimensionless)
dimensionless time
dimensionless time necessary for 10% of the
tracer to leave the system
magnitude of the fluid velocity through porous
media (m s1)
fluid density (mg m3)
theoretical residence time (min)

Among the various developments, the EvaTAC (a horizontal


subsurface flow constructed wetland) represents an innovative
and emerging solution for household treatment and reuse of
gray water (Silva et al., 2014). It removes pollutants through
time dependent processes, including sedimentation, sorption,
plant uptake, and chemical or biological reductions (Haberl
et al., 2003; Kadlec and Wallace, 2009; Konnerup et al., 2009;
Paulo et al., 2009; Vymazal and Krpfelov, 2009). The more
time pollutants remain in the wetland, the greater amount of
pollutants are removed. Hence, it is important to design household wetlands such that the flow regime in the wetland is a plug
flow, because this leads to maximum residence time for all
parcels of water entering the wetland and, consequently,

maximum pollutant reduction. Under plug flow, every parcel of


water entering the inlet reaches the outlet at some nominal time,
, described as:

V
Q

(1)

where V represents the volume of fluid in the system and Q is


the system volumetric flow rate. The nominal time is the time
required for a complete volume exchange within a wetland
(Kadlec and Knight, 1996; Persson et al., 1999). In practice,
plug flow, i.e. a single nominal residence time, is impossible to
achieve. Each parcel of water may have a unique residence time
affected by physical properties (e.g. water density), flow type
(e.g. flow rate), geometric parameters (e.g. wetland length) and
physiological processes (e.g. plant transpiration) of a wetland.
This is attributed to complex hydrodynamics in wetlands,
mainly short-circuiting, mixing, and stagnant or recirculation
zones. Therefore, estimation of residence time distribution
(RTD) for each water parcel is important, because it is used to
calculate expected hydraulic and water quality efficiencies for
specific designs. Although laboratory, field, and computational
studies on the RTDs of constructed wetlands have been
conducted to investigate the influence of design factors
(Alcocer et al., 2012; Fan et al., 2008; Garca et al., 2004), they
are inconclusive on at least two issues.
Firstly, considering experimental design, many of the studies
used the one-factor-at-a-time strategy. In this strategy, design
factors such as inlet configuration, distribution and type of
vegetation and evapotranspiration, type of porous medium,
length to width ratio, and flow rates are analyzed by changing
one factor at a time while holding the rest constant. However, a
thorough investigation of the effects of the factors, along with
their mutual interaction, is desirable for a better understanding
of the subject. For example, Garca et al. (2004) found that the
main effect of aspect ratio (length to width) is not an important
factor in controlling the wetland efficiency, in comparison with
flow rate, granular medium size and water depth. Therefore, we
would be tempted to conclude that there is no effect of aspect
ratio upon wetland efficiency. However, the effect of aspect
ratio may depend on the levels of another factor such as flow
rate. If this is the case, then knowledge of the interaction
between aspect ratio and flow rate is more useful than knowledge of the main effect of aspect ratio. A significant interaction
can mask the significance of main effects (Berthouex and
Brown, 2002). Consequently, Garca et al. (2004) could have
observed that the main effect of aspect ratio may not have much
meaning if interaction is present. A factorial design strategy is
the only way to discover interactions between factors, allowing
the researcher to iteratively and quickly move toward more
efficient wetlands if they exist (Berthouex and Brown, 2002). In
addition, the relative importance of all the factors can be evaluated simultaneously with a fewer number of experiments.
Secondly, in regard to modeling, there appears to be three
schools of thought, grouped according to the models used to
describe the wetland hydraulics. The black box model simplifies reality to an ideal wetland with ideal hydraulic behavior
(Kadlec and Wallace, 2009). A plug flow wetland is one

E.E. Rengers et al. / Journal of Hydro-environment Research 12 (2016) 91104

example of such a model. The opaque box model group splits


a wetland into zones such as mixing, plug flow or shortcircuiting, that resemble reality better that represents prominent
features of the wetland. The glass box model group models
the mass and momentum conservation equations and not on a
priori assumptions of flow patterns. To our knowledge, only
computer-modeling studies using one- and two-dimensional
approaches have been conducted to investigate the velocity
patterns and RTDs inside subsurface wetlands (Fan et al., 2008;
Persson et al., 1999; Rizzo et al., 2014). These models are not
able to capture the complex three-dimensional flow patterns
that may be the single most significant cause for impairment of
a wetlands hydraulic efficiency, such that three-dimensional
modeling tools are required to allow detailed assessments of
the actual flow characteristics inside wetlands. Computational
Fluid Dynamics (CFD) is a tool that is able to provide this new
insight into the detailed flow field in wetlands and the RTDs. It
is worth mentioning that by simply evaluating an RTD, one can
identify phenomena such as short-circuiting, but the RTD
cannot guide the designer toward improving the hydraulic efficiency; only by evaluating both the velocity and tracer concentration distribution, can new insight be gained into the behavior
of wetland hydraulics and possible remedial action be taken.
Furthermore, CFD uses a generalization of the NavierStokes
equations and of Darcys law, while existing codes to simulate
flow in constructed wetlands commonly rely on Darcys law via
resolution of either the Richards equation (e.g. HYDRUS software as in Rizzo et al., 2014) or in the time-deformable HF
saturated domain (e.g. BIO_PORE software as in Samso et al.,
2013). Under appropriate assumptions (e.g. inertial effects can
be neglected) the NavierStokes equations reduce to the
Darcian equation on the macroscopic level. Hence, CFD furnishes equations that are more general. Another important point
to mention is that CFD allows simulating both fluid and porous
domains. This is important, since the EvaTAC contains both
fluid and porous domains. Finally, CFD is a powerful approach
because it is low cost, can be scaled up, and can be used under
many different design constraints (Fan et al., 2008). Hence, it is
evident that CFD modeling can enhance our understanding of
the hydrodynamic behavior of a constructed wetland.
Therefore, in order to benefit the engineering design of subsurface flow constructed wetlands developed for household
treatment and reuse of gray water and the further use of CFD to
simulate the transport processes that occurs inside constructed
wetlands, the objectives of this work were: (a) to validate a CFD
model of a modified constructed wetland system and use it to
calculate the flow field and the RTD of the system; and (b) to
investigate the effect of two geometric parameters (length and
baffles) and a flow parameter (flow rate) on hydraulic indexes
(short-circuit, mixing, and moment index) and an effluent pollutant fraction using a full factorial experiment.
2. Materials and methods
2.1. RTD and hydraulic efficiency
The analysis of the RTD function originating from tracer
studies is a major tool for the analysis of wetlands (Werner and

93

Kadlec, 2000). RTD can be obtained by an instantaneous injection of a known quantity of tracer mass, M, at the inlet section
of the wetland and the subsequent measuring of the tracer
concentration, C, with time, t, at its outlet section. In order to
allow direct comparison of measured RTDs having dissimilar
conditions (e.g. different volumes and flow rates, mass tracer),
they are usually normalized (Wahl et al., 2010). The dimensionless RTD function, C(), and the dimensionless time, , can be
defined, respectively, as:

C ( ) =

C ( ) Q
= E ( )
M

(2)

(3)

where t is the dimensional time and E is the dimensional RTD.


The zeroth (M0), first (M1), and second (M2) moments of the
dimensionless RTD function about the origin are respectively:

M0 =

M1 =

M2 =

C ( ) d

(4)

C ( ) d

(5)

( M1 ) C ( ) d
2

(6)

The zeroth moment provides the fraction of tracer mass


recovered. If the inlet tracer mass is substituted by the outlet
tracer mass, M0 is always unity. When M0 = 1, the first moment,
M1, is the centroid of the RTD. The second moment, M2, is the
variance of the RTD, which accounts for the spread of the tracer
over time.
In order to characterize the performance of a wetland, it
is useful to reduce the RTD to a single number to represent
hydraulic efficiency. Hydraulic efficiency describes hydraulic
performance in terms of departure from ideal plug flow
(Thackston et al., 1987). A metric of hydraulic efficiency may
quantify the amount of short-circuiting, mixing, or a
combination of both (Persson et al., 1999). In the present work,
four characteristics were used, which are grouped into three
broad categories of indicators: (a) short-circuiting, (b) mixing,
and (c) moment index. As a short-circuit indicator (category a),
we used 10 = t10/, which is the nondimensional time necessary
for first 10% of the tracer to leave the wetland. In an ideal
plug flow, this ratio is 1. As mixing indicators (category b),
the dispersion index, M2, and the Morril dispersion index,
MDI = t90/t10 (where t90 is the time necessary for 90% of the
tracer to leave the wetland), were used while for ideal plug
flow, M2 = 0 and MDI = 1 were utilized (Wahl et al., 2010). The
indexes chosen for categories (a) and (b) were suggested by
Teixeira and Siqueira (2008) after evaluating different shortcircuit and mixing indexes on three criteria: the correlation of
the index to the physical phenomenon it is said to represent;
the capability of the index to detect variation; and statistical

94

E.E. Rengers et al. / Journal of Hydro-environment Research 12 (2016) 91104

Fig. 1. Geometry configurations of the bench scale models used for validation: (a) Evapotranspiration and Treatment Chamber (CEvaT) and (b) Horizontal
Subsurface Flow Constructed Wetland (HSSF-CW).

variability of the index. The moment index (category c) is


(Wahl et al., 2010):

MI = 1

(1 )C ( ) d

(7)

The moment index assumes that residence times of a completely efficient wetland will meet or exceed the nominal residence time. The portion of tracer exiting the wetland prior to the
nominal residence time adversely impacts hydraulic efficiency
being considered inefficient with more weight assigned to the
more severely premature residence times. If the bulk of tracer
exiting has a close proximity to the nominal residence time,
then hydraulic efficiency is high. As more tracer exits earlier,
hydraulic efficiency is low.

the system. Water samples were collected at the outlet of each


system, starting at the time of injection. Sodium chloride concentration was measured using a conductivity probe with intervals increasing during the crest of the tracer curve and
decreasing as it descended (Headley and Kadlec, 2007) and two
RTD plots were derived. The experiment was considered complete after 5 and 3.5 nominal RTDs for the CEvaT and the
HSSF-CW, respectively. The tracer recovery was above 90% for
both systems. The experiment in the CEvaT had a duration of
311 minutes and an average temperature of 27C. The experiment in the HSSF-CW had a test duration of 193 minutes and
an average temperature of 24C. Discharge was measured by
direct volumetric measurements at the outlet. The average value
of the flow rate was Q = 1.25 L/min.

2.2. Experimental setup

2.3. CFD setup

Fig. 1 is a sketch of the bench scale modified household


constructed wetland system, which was used for the validation
of the numerical model. The EvaTAC was composed of an
evapotranspiration and treatment chamber (CEvaT) and a
horizontal subsurface flow constructed wetland (HSSF-CW)
in series. The dimensions of both systems were 0.98 m
long 0.28 m wide 0.60 m high with a bottom slope of 0.6
degrees. The average water depth was approximately 0.55 m.
The CEvaT contained a 150 mm tube throughout its bottom,
denoted as the Anaerobic Chamber (AnC) for its capacity to
digest organic matter and retain coarse material. The gray water
raised and passed through the holes of the AnC, went through a
layer of coarse gravel (D60 = 14 mm) and exited through a
32 mm manifold pipe. The CW-HSSF consisted of a fine gravel
bed (D60 = 6 mm) and had a 32 mm manifold pipe for both the
inlet and outlet. The void fractions () of the coarse and fine
gravel, defined as the volume of the empty spaces divided by
the volume of the packed bed region, were 0.43 and 0.46
respectively.
A pulse tracer study was experimentally conducted in each
unit separately following the protocol from AWWA Research
Foundation (1996). A mass of sodium chloride (5850 mg for the
CEvaT; 5460 mg for the HSSF-CW) was injected for approximately 1% of the nominal residence time into the inlet pipe of

2.3.1. Numerical methods


Initially the flow fields were determined through 3D steady
state simulations by solving the discretized mass conservation
and momentum conservation equations using the commercial
CFD (Computational Fluid Dynamics) code ANSYS CFX. The
conservation equations are given by:

ui
=0
xi

(8)

u j ui
1 p
2 ui
=
+
+ Si
x j
xi
x j x j

(9)

where i or j = 1, 2, or 3, x1, x2, and x3, denote the cross-stream


(x), spanwise (y), and streamwise directions (z), respectively, u1,
u2, and u3, are the corresponding velocity components, p is the
pressure, is the fluid density, and Si is the source term.
In CFD simulation, the porous media model does not resolve
the flow field at every point at the microscopic scale. Instead, an
equivalent spatially-averaged flow field is determined. The
porous media effects on the macroscopic flow (such as pressure
drop) is incorporated using an extra momentum loss term, Si, in
the NavierStokes equations (Eq. 9) (ANSYS Inc, 2014):

E.E. Rengers et al. / Journal of Hydro-environment Research 12 (2016) 91104

95

Fig. 2. Computational grid for the Evapotranspiration and Treatment Chamber (CEvaT): (a) main computational domain; (b) close-up view of the Anaerobic
Chamber (AnC) near the inlet.

1
Si = ui + C2 u ui

(10)

where u is the magnitude of the fluid velocity through porous


media, is the permeability (m2), and C2 is the inertial
resistance coefficient (m1). The source term, Si, is composed of
two parts, a viscous loss term (Darcys law), and an inertia lose
term, contributing to the pressure gradient in the porous domain
that is proportional to the fluid velocity through the porous
domain. The permeability coefficient of resistance and inertia
can be defined as:

d602
3
150 (1 )2

(11)

3.5 1
d60 3

(12)

C2 =

where d is the mean particle diameter and is the void fraction.


Boundary conditions were defined at the borders of the computational domain. A uniform flow was imposed at the inlet. At
the outlet, an average static reference pressure of 0 Pa was
specified. A no-slip boundary condition was applied at the walls
and a free-slip boundary condition was applied at the water
surface.
Once the steady-state flow field was obtained, the transport
of a scalar quantity C was simulated by solving the threedimensional advectiondiffusion equation:

C
C C C

+U j
=
D
t
x j xi x j

(13)

where D is the diffusion coefficient. Equation (13) is based on


the values of Uj, which were obtained from the converged
steady-state hydrodynamics simulation. The RTD was
calculated from the solution of Equation (13). The injection
of the tracer was represented by a square step input, having
duration equal to approximately 1% of the nominal residence
time. The solute transport simulations were carried out using a
time step of 15 s. The tracer transport simulations were carried
out for a period that ranged between 2.5 and 6.5. Recovery
rates of 95% and greater were achieved for the CEvaT and the
HSSF-CW. A root mean square (RMS) residual of 106 was
used in the simulation to get a high level of convergence of the

simulated solution. Additionally, mass conservation in the fluid


domain was met for a relative error (comparison between
volume flow rate at the inlet and outlet sections) of less than
1%.
The numerical model was validated using the experimental
measurements presented in Section 2.2. The main domain of the
CEvaT and a zoomed-in view of the AnCs numerical grid are
shown in Fig. 2. The grid for both the CEvaT and the HSSF-CW
had a finer spacing in regions of larger gradients (near the wall
and the holes) and coarser spacing in the regions of low velocity
gradients. The grid was unstructured with about 340,000 elements and used a minimum element size of 0.003 m. The
number of elements was defined after an evaluation of the
influence of the number of elements on the simulation results.
The process included the procedures presented by Celik et al.
(2008). Details of the grid independence study are not repeated
here for brevity; more details can be found in Rengers (2014).
The discretization error in average was around 2%. Similar
grids were used for all other simulations.
2.3.2. Factorial experimental design
After the numerical model was validated with the experimental results presented in Section 2.2, a two-level factorial
experimental design was established for the CEvaT and the
HSSF-CW for household treatment and reuse of gray water.
The CEvaT computational experiments were based on a 22
factorial experimental design (see Table 1). The independent
variables (factors) chosen at two levels were the wetland
length (L, 1 m and 2 m) (Fig. 3) and flow rate (Q, 7 L/min and
30 L/min). The AnC configuration had a triangle format with a
base of 0.5 m and a height of 0.87 m (Fig. 3). A triangular AnC
was used in this section in effort to imitate in-situ configurations and for brevity it was the only configuration used. The
HSSF-CW simulations were based on a 23 factorial experimental design (see Table 2). The independent variables (factors)
chosen at three levels were the wetland length (L, 1 m and 2 m),
flow rate (Q, 7 L/min and 30 L/min), and baffles (absent and
present) (Fig. 4). Internal baffles, 10 mm thick, divide the tank
into three compartments for L = 1 m and five compartments for
L = 2 m, which are 0.32 m and 0.39 m wide respectively.
The variables chosen to be varied in the experimental design
have been shown to distinctly influence hydraulics in ponds and
wetlands (Alcocer et al., 2012; Garca et al., 2004; Persson

96

E.E. Rengers et al. / Journal of Hydro-environment Research 12 (2016) 91104

Table 1
Design matrix of the 22 factorial experimental design, levels of independent variables (L and Q) and observed responses (10, M2, Mo, MI, and X) for the
evapotranspiration and treatment chamber (CEvaT).
Experiment

Length
L (m)

Flow rate
Q (L/min)

Short-circuit
indicator
10

Variance
index
M2

Morril
index
MDI

Moment
index
MI

Effluent pollutant
fraction
X*

1
2
3
4

1
1
2
2

7
30
7
30

0.25
0.25
0.23
0.24

0.56
0.62
0.65
0.58

7.91
8.02
9.04
8.28

0.72
0.71
0.70
0.71

0.91
0.98
0.83
0.96

* Assuming k = 1.8 day1 for calculating X using Eq. (15).

et al., 1999). The ranges of data tested were chosen based on


characteristics found in typical Brazilian homes; they were not
intended to be representative of the entire range of field conditions, but rather intended to determine if the wetland length,
flow rate, and the presence of baffles had an effect on the
responses. The response variables (dependent variables) were
10, M2, MDI, MI, and X. X is the fraction of pollutant remaining
over time. For many individual wetland treatment processes,

X can be calculated by means of a first-order rate function


(Kadlec and Knight, 1996), as follows:

X = ekt

(14)

where k is the reaction rate coefficient (in units of time1),


expressing the actual fraction of pollutant that is converted per
unit time. Values of k vary widely in the literature (0.30 < k < 6.11,
k in units of day1) (Stein et al., 2006). The decay rate constant

Fig. 3. Geometries of the Evapotranspiration and Treatment Chamber (CEvaT) 22 factorial experimental design with the triangular Anaerobic Chamber (AnC).
(a) L = 1 m and (b) L = 2 m.

Table 2
Design matrix of the 23 factorial experimental design, levels of independent variables (L, Q, and baffles) and observed responses (10, M2, Mo, MI, and X) for the
horizontal subsurface flow constructed wetland (HSSF-CW).
Experiment

Length
L (m)

Flow rate
Q (L/min)

Baffles

Short-circuit
indicator
10

Variance
index
M2

Morril
index
MDI

Moment
index
MI

Effluent pollutant
fraction
X*

1
2
3
4
5
6
7
8

1
1
1
1
2
2
2
2

7
30
7
30
7
30
7
30

No
No
Yes
Yes
No
No
Yes
Yes

0.54
0.59
0.60
0.62
0.66
0.74
0.70
0.68

0.21
0.21
0.11
0.12
0.09
0.07
0.06
0.07

2.95
2.72
2.37
2.35
2.13
1.86
1.87
1.97

0.82
0.82
0.86
0.85
0.88
0.89
0.89
0.88

0.94
0.99
0.94
0.99
0.89
0.97
0.89
0.97

* Assuming k = 1.8 day1 for calculating X using Eq. (15).

E.E. Rengers et al. / Journal of Hydro-environment Research 12 (2016) 91104

97

Fig. 4. Geometries of the Horizontal Subsurface Flow Constructed Wetland (HSSF-CW) 23 factorial experimental design. (a) L = 2 m, without baffles; (b) L = 2 m,
with baffles.

k was considered equal to 1.8 day1 throughout the text. For


first-order reactions, X can be determined directly from the
RTD (Wahl et al., 2010):

X=

E (t ) ekt dt

(15)

where E(t) is the RTD. All other simulation parameters were the
same as those in Section 2.2.
3. Results and discussion
3.1. Validation of the computational model
For the HSSF-CW, there is a good agreement between
experimental and computational RTDs (Fig. 5). The data show
that the maximum difference between numerical and experimental results was in general within experimental uncertainty.
For the CEvaT, there is a satisfactory agreement between the
two curves (Fig. 5). At the first part of the curve (ascending
part), the peak of the CFD simulation is higher and lags behind
the experimental results. At the second part of the curve
(descending part), the two curves almost coincide and have an
exponential shape.

3.2. CEvaT
Fig. 6 presents the dimensionless velocity magnitude and
streamlines for the different CEvaT configurations. The velocities are normalized with inlet velocities. It can be seen that the
inlet jet travels through the bottom of the triangle with high
velocity. The areas of recirculation, which occurs once the jet
hits the back of the AnC, dissipates the energy of the jet. Once
in the porous medium, lower velocities are found, and the fluid
moves up and toward the outlet. Though the order of magnitude
is greater for the system with peak flow, the behavior of the
velocity remains consistent. Note that the flow structure may
not be fully captured if a 2D model is considered.
In order to enhance our understanding of the CEvaT hydrodynamics discussed above, hydraulic indexes are calculated
to benchmark the CEvaTs across a spectrum of flow rates and
lengths (see Table 1). Table 3 shows the main effects of Q
and L and their interactions on 10, M2, MDI, and MI. It can be
noted that 10 remains almost unchanged to changes in Q and
decreases slightly (from 0.25 to 0.23) as L increases from 1 to
2 m. The interaction Q and L on 10 is low and equal to 0.007.
The sequence of the main and interaction effects with respect
to 10 is found to be L > Q > LQ. The values of 10 are low

Fig. 5. Comparison of the simulated (solid line) nondimensional concentration C() along nondimensional time () with the experimental data (solid line). (a)
Evapotranspiration and Treatment Chamber (CEvaT) and (b) Horizontal Subsurface Flow Constructed Wetland (HSSF-CW). Horizontal and vertical bars represent
the experimental uncertainty.

98

E.E. Rengers et al. / Journal of Hydro-environment Research 12 (2016) 91104

Fig. 6. Isocontours and streamlines of dimensionless velocities, V/Vinlet, for the Evapotranspiration and Treatment Chamber (CEvaT) (see Table 1): (a) Experiment
1; (b) Experiment 2; (c) Experiment 3; (d) Experiment 4.

(10 0.25), indicating the presence of intense short-circuiting.


This is consistent with the results presented in Fig. 6, where the
inlet jet went toward the opposite wall with high velocity, and
part of the flow was diverted toward the system outlet, so that
significant volumes of water exit the system in a much lower
time than the theoretical residence time, . When the jet
impinged on the opposite wall, the flow was re-directed from a
horizontal into a vertical direction inducing counter-clockwise
circulations that consist mainly of vertical two-dimensional
circulation cells in the back half of the tank (opposite the inlet
wall).

In general, short-circuiting is never desirable in a treatment


wetland because treatment area is effectively lost (Thackston
et al., 1987) and, consequently, wetland performance is
reduced. The Pareto chart (Fig. 7) gives the relative importance
of the main and interaction effects. The vertical line in the chart
indicates the minimum statistically significant effect magnitude
for a 95% confidence level. None of the main effects of Q and
L and their interactions on 10 are statistically significant.
In the case of the mixing indexes, the average M2 and MDI is,
respectively, 0.6 and 8.31. M2 increases from 0.56 at low flow
and to 0.62 at high flow, while MDI changed from 7.91 at low

Table 3
Average and main effects of Q and L and their higher order interactions of the 22 factorial design on the 10, M2, MDI, and Moment index for the evapotranspiration
and treatment chamber (CEvaT).
Effect

Average effect
Main effects
Flow rate Q
Length L
Two-factor interactions
QL

Short-circuit
indicator
10

Variance
index
M2

Morril
index
MDI

Moment
index
MI

Effluent pollutant
fraction
X*

0.240

0.600

8.310

0.710

0.920

+0.008
0.018

0.003
+0.027

0.327
+0.700

+0.005
0.007

+0.098
0.051

+0.007

0.062

0.433

+0.008

+0.029

* Assuming k = 1.8 day1 for calculating X using Eq. (15).

E.E. Rengers et al. / Journal of Hydro-environment Research 12 (2016) 91104

99

Fig. 7. Pareto chart of the effects of the 22 factorial design on the (a) 10, (b) M2, (c) MDI, (d) MI and (e) X for the Evapotranspiration and Treatment Chambers
(CEvaTs). The vertical line in the chart indicates the minimum statistically significant effect magnitude for a 95% confidence level.

flow to 8.02 at high flow. Increase in L from 1 to 2 m increases


M2 and MDI by 0.09 and 1.13, respectively. The interaction Q
and L on the mixing indexes is greater than the main effects, i.e.
knowledge of the interaction between Q and L is more useful
than knowledge of the main effect. In other words, the effect of
Q depends on the levels of L. Considering the calculated mixing
indexes (M2 > 0, MDI >> 1), the hydraulic efficiency is compromising (Van der Walt and Haarhoff, 2000). Changes in
mixing scale do not directly translate into changes in treatment
efficiency (Holland et al., 2004). Nevertheless, for reaction
rates of first-order kinetics or greater, the wetland treatment
efficiency is maximized with plug flow, i.e. for M2 = 0 and
MDI = 1 (Holland et al., 2004). None of the main effects of Q
and L and their interactions on mixing indexes M2 and MDI are
significant (Fig. 7). The sequence of the main and interaction
effects with respect to 2 and MDI are found to be QL > L > Q
and L > QL > Q, respectively.
The average MI is 0.71. MI decreases from 0.72 at low flow
to 0.71 at high flow and decreases to 0.70 as L increases from 1
to 2 m. The interaction flow rate and length on MI is low and
equal to 0.008. Since for ideal plug flow conditions MI = 1 and
a value of MI approaching zero indicates a high degree of
mixing or short-circuiting (Wahl et al., 2010), our CEvaTs have
an average to high performance. None of the main effects of Q
and L and their interactions on MI are significant (Fig. 7).
Nevertheless, the sequence of the main and interaction effects
with respect to MI are found to be QL > L > Q.
Considering the decay rate constant k = 1.8 day1, effluent
pollutant fractions X were computed using Eq. (15) and presented in Table 1. It can be observed that X increased from
0.91 to 0.98 as Q changes from 7 L/min to 30 L/min, and X
decreased from 0.91, at L = 1 m, to 0.83 at L = 2 m. Hence, an
increase in L has a positive effect on wetland treatment efficiency, while an increase in Q has a negative effect on wetland

treatment efficiency. The interaction of flow rate and length on


X is lower than the main effects and equal to 0.029. None of the
main effects of Q and L and their interactions on X are significant (Fig. 7). The sequence of the main and interaction effects
with respect to MI are found to be Q > L > QL.
To enhance water treatment by modifying the hydraulic
regime of a constructed wetland to be effective, a strong correlation between hydraulic efficiency and pollutant reduction
must exist. Therefore, a hydraulic index with a strong correlation to pollutant reduction would be able to give a good prediction of water treatment (Wahl et al., 2010). None of the indexes
demonstrate good correlation to treatment (Fig. 8), which
differs from the findings of Wahl et al. (2010). This is in agreement with our results that show that the length and the interaction between the length and the flow rate are the most important
parameters for hydraulic behavior while flow rate is the most

Fig. 8. Hydraulic indices versus effluent pollutant fraction, X, for the Evapotranspiration and Treatment Chambers (CEvaTs). () MI; () M2; () 10.

100

E.E. Rengers et al. / Journal of Hydro-environment Research 12 (2016) 91104

important for wetland treatment efficiency. Since the parameters that influence the hydraulic behavior vary from those that
influence the fraction of pollutant X, we infer that the hydraulic
indexes do not demonstrate good correlation to X. This result
may be true only for our design rather than a generic feature of
a horizontal subsurface flow constructed wetlands.
3.3. HSSF-CW
Fig. 9 presents isocontours of the dimensionless velocity
magnitude and streamlines for the various configurations of
the HSSF-CW. Again, the velocities are normalized with inlet

velocities. For HSSF-CWs without baffles, the flow structure is


in general similar. Part of the flow goes directly toward the
outlet via a short-circuiting route, causing volumes of water to
exit the system in a much lower time than . High flow velocities are observed in the inlet and outlet regions. Lower velocities are found in the corners, noting that these low velocity areas
are smaller in the systems with a peak velocity. The flow structure may be considered as mainly 2D. As the length of the
HSSF-CW is increased, the velocity of the water becomes
nearly constant and parallel to the bottom across any crosssection. In other words, higher lengths increased the ideal plug
flow behavior of the system and the short-circuiting route. For

Fig. 9. Isocontours and streamlines of dimensionless velocities, V/Vinlet, for the Horizontal Subsurface Flow Constructed Wetland (HSSF-CW) (see Table 2): (a)
Experiment 1; (b) Experiment 2; (c) Experiment 3; (d) Experiment 4; (e) Experiment 5; (f) Experiment 6; (g) Experiment 7; (h) Experiment 8.

E.E. Rengers et al. / Journal of Hydro-environment Research 12 (2016) 91104

101

Table 4
Average and main effects of Q, L, and baffles, and their higher order interactions of the 23 factorial design on the 10, M2, MDI, and Moment index for the horizontal
subsurface flow constructed wetland (HSSF-CW).
Effect

Average effect
Main effects
Flow rate Q
Length L
Baffles
Two-factor interactions
QL
Q Baffles
L Baffles
Three-factor interactions
Q L Baffles

Short-circuit
indicator
10

Variance
index
M2

Morril
index
MDI

Moment
index
MI

Effluent pollutant
fraction
X*

0.64

0.12

2.27

0.86

0.95

+0.033
+0.108
+0.018

+0.000
0.086
0.059

0.105
0.640
0.275

+0.001
+0.048
+0.018

+0.062
0.033
+0.001

0.003
0.033
0.028

0.004
+0.009
+0.039

+0.020
+0.145
+0.200

+0.004
0.009
0.015

+0.019
0.000
+0.001

0.018

+0.006

+0.040

0.007

0.000

* Assuming k = 1.8 day1 for calculating X using Eq. (15).

HSSF-CWs with baffles, after entering into the system, the flow
follows a plug flow path parallel to the baffles and finally
exits via the outlet pipe. In spite of the higher velocities, the
short-circuiting time is higher (in comparison to the system
without baffles) because of the longer flow path. In the corners,
velocity is low. As the HSSF-CW without baffles, the shortcircuiting time in the HSSF-CW with baffles increases with the
increase of the length because the flow path is longer. Again, the
flow structure may be considered as mainly 2D.
In the same way as the CEvaT, we analyzed the hydrodynamic performance of the HSSF-CW via the study of hydraulic
indexes discussed previously (see Table 2). Table 4 shows the
main effects of Q and L and their interactions on 10, M2, MDI,
and MI for the HSSF-CW.

In all the HSSF-CW configurations, 10 is between 0.54 and


0.74, denoting compromising hydraulic efficiency (Van der Walt
and Haarhoff, 2000). The sequence of the most important effects
with respect to 10 is found to be L > Q Q Baffles. The
importance of the length to the short-circuiting of the HSSF-CW
is consistent with the velocity field described previously. Our
results are also found to be in good agreement with recent results
obtained by Alcocer et al. (2012), who observed that higher
aspect ratios (length to width) and higher flow rate improved the
hydraulic behavior of the system by reducing the effect of
short-circuiting. It is worth mentioning that Alcocer et al.
(2012), differently from our study, increased the aspect ratio by
varying length and width simultaneously. All interactions and
the three main effects are statistically insignificant (Fig. 10).

Fig. 10. Pareto chart of the 23 factorial design on the (a) 10, (b) M2, (c) MDI, (d) MI and (e) X for the Horizontal Subsurface Flow Constructed Wetlands
(HSSF-CWs). The vertical line in the chart indicates the minimum statistically significant effect magnitude for a 95% confidence level.

102

E.E. Rengers et al. / Journal of Hydro-environment Research 12 (2016) 91104

In the case of the mixing indexes, the average M2 and MDI


over the eight simulations are, respectively, 0.12 and 2.27. M2
varied between 0.06 and 0.21 (Table 2), agreeing well with the
range of 0.070.35 observed by Garca et al. (2004), King
et al. (1997), Seeger et al. (2012), and Simi and Mitchell
(1999). Considering the calculated mixing indexes (M2 0.21,
MDI < 3), the hydraulic efficiency is excellent and near ideal
plug flow (Van der Walt and Haarhoff, 2000). Two main effects
are significant for M2: length and baffles (Fig. 10). As the length
of the HSSF-CW is increased, the mixing decreases. This agrees
with the results of Garca et al. (2004), who observed a decrease
in M2 in an increase of the aspect ratio (length to width). Using
baffles gives a lower mixing index than without baffles. One
interaction is statistically significant for M2: length and baffles.
In other words, the effect of L depends on the presence or
non-presence of baffles. The effect of baffles upon mixing
indexes is much stronger for a short system (L = 1 m) than for a
long system (L = 2 m). None of the other effects are statistically
significant. The sequence of the most important effects with
respect to mixing is found to be L > Baffles > L Baffles.
Regarding the Moment Index, its average value is 0.86. The
main effect of length is significant and none of the interactions are
significant (Fig. 10). Nevertheless, the interaction between the
length and baffles and the presence or not of baffles are the two
most important effects after length. Since for ideal plug flow
conditions MI = 1 and a value of MI approaching zero indicates a
high degree of mixing or short-circuiting (Wahl et al., 2010), our
HSSF-CWs have an average to high performance.
For the fraction of pollutant X (assuming k = 1.8 day1 for
calculating X using Eq. 15), flow, length, and their interactions
are significant. X increases as Q rises, while X decreases as L
rises. Hence, an increase in L has a positive effect on wetland
treatment efficiency, while an increase in Q has a negative effect
on wetland treatment efficiency. The interaction of the flow rate
and length on X (0.019) is lower than the main effects. However,
it is worth mentioning that the effect of flow rate on the fraction
of pollutant X is much stronger for a long system (L = 2 m) than
for a short system (L = 1 m).
In summary, our results indicate that the strongest influence on
the hydraulic efficiency was the length.This is similar to the results
of Alcocer et al. (2012), who observed that the construction of a
wetland with a higher aspect ratio is the most significant parameter
upon the hydraulic efficiency of the system. The baffles and the
interaction between the length and baffles significantly affected
the hydraulic efficiency. The effect of baffles upon hydraulic
efficiency is much stronger for a short system (L = 1 m) than for a
long system (L = 2 m). Finally, the influence of the flow rate on
hydraulic efficiency is statistically insignificant. Our results are
consistent with the observations of Fan et al. (2008) and Alcocer
et al. (2012), who found that the effect of flow rate on the hydraulic
efficiency is slight, although there is some increase in the
hydraulic efficiency with its increase.
Regarding the effluent pollutant fraction, the order of significance with respect to influence on X was found to be flow
rate, length and interaction flow rate and length. Considering
the importance of flow rate, our result is similar to Garca et al.
(2004), who observed that flow rate is a very important factor in

Fig. 11. Hydraulic indices versus effluent pollutant fraction, X, for the Horizontal Subsurface Flow Constructed Wetlands (HSSF-CWs). () MI; () M2;
() 10.

controlling the treatment efficiency of the wetland. Considering


the length, our result is in reasonable agreement with Garca
et al. (2004), who found that the aspect ratio (length to width)
is not important in comparison with flow rate, granular medium
size and water depth. Although the length is not as important as
the flow rate, our data show that it is still statistically significant.
Finally, the interaction between flow rate and length is an
important factor that should be taken into account for horizontal subsurface flow constructed wetlands design and for predictive performance models. Although not statistically significant,
our results indicate a positive association between the presence
of baffles and X. This is consistent with Tee et al. (2012), who
observed that the baffled wetland achieved better performance
(X) versus the planted unbaffled wetland. The better performance of the baffled wetland was explained by the longer
pathway due to the up-flow and down-flow conditions sequentially thus allowing more contact of the wastewater with the
rhizomes and micro-aerobic zones.
Since the order of importance of the factors that influence
the hydraulic efficiency vary from those that influence the water
quality efficiency, we expect that the hydraulic indexes do not
demonstrate good correlation to effluent disinfectant fraction X.
This is confirmed by the low correlation between the hydraulic
indexes and X (see Fig. 11). Hence, although the hydraulic
efficiency affects the water quality of a wetland, none of the
hydraulics indexes demonstrated good correlation to X.
4. Conclusions
A CFD model was employed to calculate the flow field and
the RTD for a CEvaT and a HSSF-CW in order to evaluate
hydraulic performance. Prior to its application, the CFD model
was validated by comparison of the computed RTD with experimental RTDs observed in a bench scale system of each. A
satisfactory agreement was observed between the calculated
and the experimental curve for the CEvaT, and a good agreement was noted between the computational and the experimental data for the HSSF-CW. Hence, CFD can be used to model
horizontal subsurface flow constructed wetlands. 2D CFD

E.E. Rengers et al. / Journal of Hydro-environment Research 12 (2016) 91104

simulations seem to be sufficient to model the HSSF-CW;


however, a 2D model cannot identify the complex flow structure
that occurred in the CEvaT due to the presence of the Anaerobic
Chamber (AnC). It is worth mentioning that the results presented in this study are an important first step of any CFD study
of constructed wetlands. Future studies will add additional
physics (or chemistry) to build-up to the complex model. The
presence of vegetation, for example, seems to improve flow and
removal of pollutants (Chazarenc et al., 2003; Mesquita et al.,
2013); furthermore, evapotranspiration seems to be beneficial
and to improve all performances (Chazarenc et al., 2003).
Hence, future studies are needed to explore the effects of these
and other factors upon hydraulic and treatment efficiencies.
For the CEvaT, the most important factor for hydraulic performance was the length and the interaction between length and
flow rate, while the less important factor was the flow rate. For
the effluent pollutant fraction, the most important factor was the
flow rate, while the less important factor was the interaction
between length and flow rate.
For the HSSF-CW, the most significant effect for the hydraulic
indexes was length followed by baffles and the interaction between
baffles and length. Concerning the interaction between baffles and
length, the effect of baffles upon hydraulic performance is much
stronger for a short system than for a long system. With respect to
the effluent pollutant fraction, the order of statistically significant
effects was found to be flow rate > length > interaction flow rate
and length. Concerning the interaction between flow rate and
length, the effect of flow rate on the fraction of pollutant X is much
stronger for a long system than for a short system.
Although the hydraulic efficiency affects the water quality of
a wetland, none of the hydraulics indexes demonstrated good
correlation to the fraction of pollutant (X) for both the CEvaT
and the HSSF-CW. It is worth mentioning that we only considered a first-order reaction for the reduction reaction.
Our results will benefit the engineering design of subsurface
flow constructed wetlands developed for household treatment
and reuse of gray water in the Asia-Pacific region and the
further use of CFD to simulate the transport processes that
occurs inside this kind of wetlands.
Acknowledgements
This study was supported by a grant from the Brazilian
National Council for the Improvement of Higher Education
(CAPES) under grant 1091480, and a Brazilian Funding
Authority for Studies and Projects Public Announcement:
Water and Energy in Social Housing grant 01.10.0507.00
(FINEP-07/2009).
References
Alcocer, D.J.R., Vallejos, G.G., Champagne, P., 2012. Assessment of the plug
flow and dead volume ratios in a sub-surface horizontal-flow packed-bed
reactor as a representative model of a sub-surface horizontal constructed
wetland. Ecol. Eng. 40, 1826.
ANSYS Inc., 2014. CFX-Pre Users Guide Release 14.5.
AWWA Research Foundation, 1996. Tracer Studies in Water Treatment
Facilities: A Protocol and Case Studies. AWWA American Water Works
Association Research Foundation, USA.

103

Bartram, J., Lewis, K., Lenton, R., Wright, A., 2005. Focusing on improved
water and sanitation for health. Lancet 365 (9461), 810812.
Berthouex, P.M., Brown, L.C., 2002. Statistics for Environmental Engineers,
second ed. Lewis Publishers/CRC Press, Boca Raton, FL, p. 489.
Celik, I., Ghia, U., Roache, P.J., Freitas, C.J., Coleman, H., Raad, P.E., 2008.
Procedure for estimation and reporting of uncertainty due to discretization
in CFD applications. J. Fluids Eng. 130 (7), 14.
Chazarenc, F., Merlin, G., Gonthier, Y., 2003. Hydrodynamics of horizontal
subsurface flow constructed wetlands. Ecol. Eng. 21, 165173.
Choffnes, E.R., Mack, A., 2009. Global Issues in Water, Sanitation, and Health:
Workshop Summary. The National Academies Press. 328 pp.
Fan, L., Hai, R., Wang, W., Lu, Z., Yang, Z., 2008. Application of computational
fluid dynamic to model the hydraulic performance of subsurface flow
wetlands. J. Environ. Sci. 20, 14151422.
Garca, J., Chiva, J., Aguirre, P., lvarez, E., 2004. Hydraulic behaviour of
horizontal subsurface flow constructed wetlands with different aspect ratio
and granular medium size. Ecol. Eng. 23, 177187.
Haberl, R., Grego, S., Langergraber, G., Kadlec, R.H., Cicalini, A.R., Dias,
S.M., et al., 2003. Constructed wetlands for the treatment of organic
pollutants. J. Soils Sediments 3 (2), 109124.
Headley, T.R., Kadlec, R.H., 2007. Conducting hydraulic tracer studies of
constructed wetlands: a practical guide. Ecohydrol. Hydrobiol. 7 (34),
269282.
Holland, J.F., Martin, J.F., Granata, T., Bouchard, V., Quigley, M., Brown, L.,
2004. Effects of wetland depth and flow rate on residence time distribution
characteristics. Ecol. Eng. 23 (3), 189203.
Kadlec, R.H., Knight, R.L., 1996. Treatment Wetlands. CRC Press, Boca
Raton, FL. 893 pp.
Kadlec, R.H., Wallace, S.D., 2009. Treatment Wetlands, second ed. New York,
New York.
King, A.C., Mitchell, C.A., Howes, T., 1997. Hydraulic tracer studies in a pilot
scale subsurface flow constructed wetland. Water Sci. Technol. 35, 189196.
Konnerup, D., Koottatep, T., Brix, H., 2009. Treatment of domestic wastewater
in tropical, subsurface flow constructed wetlands planted with Canna and
Heliconia. Ecol. Eng. 35, 248257.
Mesquita, M.C., Albuquerque, A., Amaral, L., Nogueira, R., 2013. Effect of
vegetation on the performance of horizontal subsurface flow constructed
wetlands with lightweight expanded clay aggregates. Int. J. Environ. Sci.
Technol. (Tehran) 10, 433442.
Paulo, P.L., Begosso, L., Pansonato, N., Shrestha, R.R., Boncz, M.A., 2009.
Design and configuration criteria for wetland systems treating grey water.
Water Sci. Technol. 60 (8), 20012007.
Persson, J., Somes, N., Wong, T., 1999. Hydraulics efficiency of constructed
wetlands and ponds. Water Sci. Technol. 40 (3), 291300.
Prss-stn, A., Bos, R., Gore, F., Bartram, J., 2008. Safer Water, Better
Health: Costs, Benefits and Sustainability of Interventions to Protect and
Promote Health. World Health Organization, Geneva.
Rengers, E.E., 2014. Otimizao da eficincia hidrulica de um sistema tipo
wetland construdo usando cfd. Dissertation (Unpublished masters
dissertation). Universidade Federal de Mato Grosso do Sul, Campo Grande,
Brazil (in Portuguese).
Rizzo, A., Langergraber, G., Galvo, A., Boano, F., Revelli, R., Ridolfi, L.,
2014. Modelling the response of laboratory horizontal flow constructed
wetlands to unsteady organic loads with HYDRUS-CWM1. Ecol. Eng. 68,
209213.
Samso, R., Garcia, J., 2013. BIO PORE, a mathematical model to simulate
biofilm growth and water quality improvement in porous media: application
and calibration for constructed wetlands. Ecol. Eng. 54, 116127.
Seeger, E.M., Maier, U., Grathwohl, P., Kuschk, P., Kaestner, M., 2012.
Performance evaluation of different horizontal subsurface flow wetland
types by characterization of flow behavior, mass removal and
depth-dependent contaminant load. Water Res. 47 (2), 769780.
Silva, J.B., Magalhes Filho, F.J.C., Menezes, C.S., Paulo, P.L., 2014.
Hidrodinmica no Desenvolvimento de Ecotecnologia para o Tratamento
de guas Cinza. In: Org: Silva, G.F., Leite, N.A., 4 workshop rede de
pesquisa: Uso racional de gua e eficincia energtica em habitaes de
interesse social. Editora da Universidade Federal de Sergipe cap. 9, pp.
203220 (in Portuguese).

104

E.E. Rengers et al. / Journal of Hydro-environment Research 12 (2016) 91104

Simi, A.L., Mitchell, C.A., 1999. Design and hydraulic performance of a


constructed wetland treating oil refinery wastewater. Water Sci. Technol. 40
(3), 301307.
Stein, O.R., Biederman, J.A., Hook, P.B., Allen, W.C., 2006. Plant species and
temperature effects on the KC* first-order model for COD removal in
batch-loaded SSF wetlands. Ecol. Eng. 26, 100112.
Tee, H.C., Lim, P.E., Seng, C.E., Nawi, M., 2012. Newly developed baffled
subsurface- flow constructed wetland for the enhancement of nitrogen
removal. Bioresour. Technol. 104, 235242.
Teixeira, E.C., Siqueira, R.N., 2008. Performance assessment of hydraulic
efficiency indexes. J. Environ. Eng. (New York) 134, 851859.
Thackston, E.L., Shields, F.D., Schoroeder, P.R., 1987. Residence time distributions of shallow basins. J. Environ. Eng. (New York) 113 (2), 219223.

United Nations Publication, 2014. Statistical Yearbook for Asia and the Pacific.
United Nations.
Van der Walt, J.J., Haarhoff, J., 2000. Is a reservoir really that simple? A CFD
investigation into the internal hydraulics of reservoirs. In: Proceedings
WISA 2000 Conference. Sun City, South Africa.
Vymazal, J., Krpfelov, L., 2009. Removal of organics in constructed wetlands
with horizontal sub-surface flow: a review of the field experience. Sci. Total
Environ. 407, 39113922.
Wahl, M.D., Brown, L.C., Soboyejo, A.O., Martin, J., Dong, B., 2010.
Quantifying the hydraulic performance of treatment wetlands using the
moment index. Ecol. Eng. 36, 16911699.
Werner, T.M., Kadlec, R.H., 2000. Wetland residence time distribution
modeling. Ecol. Eng. 15 (12), 7790.

You might also like