- Email: [email protected]
Design configurations affecting flow pattern and solids accumulation in horizontal free water and subsurface flow constructed wetlands
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 4 4 8 e1 4 5 8
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Design configurations affecting flow pattern and solids accumulation in horizontal free water and subsurface flow constructed wetlands A. Pedescoll a,b,*, R. Sidrach-Cardona a,b, J.C. Sa´nchez a,b, J. Carretero c, M. Garfi c, E. Be´cares a,b a
Environmental Institute, University of Leo´n, c/La Serna 56, 24007 Leo´n, Spain Ecology Section, Department of Biodiversity and Environmental Management, University of Leo´n, Campus de Vegazana s/n, 24071 Leo´n, Spain c GEMMA e Group of Environmental Engineering and Microbiology, Department of Hydraulic, Maritime and Environmental Engineering, Universitat Polite`cnica de Catalunya-BarcelonaTech, c/Jordi Girona 1-3, Building D1, E-08034 Barcelona, Spain b
article info
abstract
Article history:
The aim of this study was to evaluate the effect of different horizontal constructed wetland
Received 20 July 2012
(CW) design parameters on solids distribution, loss of hydraulic conductivity over time and
Received in revised form
hydraulic behaviour, in order to assess clogging processes in wetlands. For this purpose, an
5 December 2012
experimental plant with eight CWs was built at mesocosm scale. Each CW presented
Accepted 7 December 2012
a different design characteristic, and the most common CW configurations were all rep-
Available online 19 December 2012
resented: free water surface flow (FWS) with different effluent pipe locations, FWS with floating macrophytes and subsurface flow (SSF), and the presence of plants and specific
Keywords:
species (Typha angustifolia and Phragmites australis) was also considered. The loss of the
Clogging
hydraulic conductivity of gravel was greatly influenced by the presence of plants and
Flow path
organic load (representing a loss of 20% and c.a. 10% in planted wetlands and an over-
Solids distribution
loaded system, respectively). Cattail seems to have a greater effect on the development of
Permeability
clogging since its below-ground biomass weighed twice as much as that of common reed.
Cattail
Hydraulic behaviour was greatly influenced by the presence of a gravel matrix and the
Common reed
outlet pipe position. In strict SSF CW, the water was forced to cross the gravel and tended to flow diagonally from the top inlet to the bottom outlet (where the inlet and outlet pipes were located). However, when FWS was considered, water preferentially flowed above the gravel, thus losing half the effective volume of the system. Only the presence of plants seemed to help the water flow partially within the gravel matrix. ª 2012 Elsevier Ltd. All rights reserved.
1.
Introduction
Constructed wetlands (CWs) are extensive, low energy systems for wastewater treatment which do not require specialised
manpower for their management (Wallace and Knight, 2006). For this reason, and due to the large amount of land required for wastewater treatment, CWs are usually recommended for the sanitation of small communities (Cooper, 2005).
* Corresponding author. Environmental Institute, University of Leo´n, c/La Serna 56, 24007 Leo´n, Spain. Tel.: þ34 987 29 10 00x5227; fax: þ34 987 29 15 63. E-mail addresses: [email protected], [email protected] (A. Pedescoll), [email protected] (R. Sidrach-Cardona), nocardia. [email protected] (J.C. Sa´nchez), [email protected] (J. Carretero), [email protected] (M. Garfi), ebecm@ unileon.es (E. Be´cares). 0043-1354/$ e see front matter ª 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.watres.2012.12.010
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 4 4 8 e1 4 5 8
Despite the advantages that CWs offer, they also present a major operational problem, namely clogging of the gravel media. Clogging is the progressive loss of the initial hydraulic characteristics, mainly porosity and hydraulic conductivity of the granular media. This phenomenon is due to the accumulation of solids from the wastewater, biofilm and plant growth and chemical precipitates (Knowles et al., 2011), and can lead to a decrease in contaminant removal efficiency over time. Although clogging is an unavoidable process (since it is inherent to the treatment), it can be delayed by improving the design (Persson et al., 1999), operation and maintenance (Griffin et al., 2008) of the wetland. In fact, the degree of clogging depends on the accumulation of different nature solids, whereas the distribution of clogging basically depends on the hydraulic behaviour of the system. In order to determine whether intervention to reverse clogging is required, it is essential to assess not only the degree of clogging, but also its spatial distribution throughout the wetland. Since clogging occurs through the accumulation of solids in the interstitial spaces of the gravel, it is obvious that the organic load, and especially the solids loading rate, are of paramount importance in the design. In the early stages after commissioning the plant, clogging occurs mainly at the inlet, spreading progressively towards the outlet. It is at the entrance to the wetland where solids are filtered, thus most of the suspended solids are retained in this zone. Therefore, an overload promotes the increase of sludge in this area, followed by rapid loss of the water infiltration rate and the appearance of ponding (Rousseau et al., 2005). Many authors have shown that the solids loading rate is directly related to sludge accumulation (Tanner et al., 1998; Caselles-Osorio et al., 2007; Pedescoll et al., 2011a). Primary treatments based on physical processes play an important role in decreasing the solids load applied to the system. However, if poorly designed or maintained, they can contribute to an undesirable solids overload at the wetland inlet (Knowles et al., 2010). Many other factors also contribute to undesirable clogging distribution. Water tends to flow along the path of least hydraulic resistance (Pedescoll et al., 2011b), which is influenced by different design parameters and operational aspects, including the length to width ratio (Garcı´a et al., 2005; Jenkins and Greenway, 2005; Persson, 2000) and the inlet and outlet position. Suliman et al. (2006, 2007) observed differences in flow pattern of an experimental flow cell depending on the strategy employed to fill the filter medium in the basin and the inlet-outlet position. The development of below-ground vegetation may have implications for wetland hydraulics (Knowles et al., 2011), because the growth of below-ground plants involves a loss of effective volume, which is occupied instead by the roots and rhizomes. The hydraulic conductivity of the medium in similar systems is greater in unplanted wetlands (Sanford et al., 1995). Edwards (1992) estimated the volume occupied by roots as being 5% of total substrate volume of a subsurface flow wetland (45 cm deep) planted with bulrushes (Scirpus validus), where roots penetrated between 12 and 15 cm below the surface. Moreover, Pedescoll et al. (2011a) observed a drastic decrease in hydraulic conductivity in shallow subsurface flow CWs in which the below-ground biomass penetrated the entire wetland depth (25 cm of water column).
1449
In addition, the growth of roots and rhizomes creates a zone of increased resistance to water flow, promoting the establishment of preferential flow paths. Jenkins and Greenway (2005) found decreased hydraulic efficiencies with increasing vegetation cover and density. The aim of this study was to evaluate the hydraulic performance of eight different configurations of wetlands in terms of clogging development, and to compare design parameters such as organic loading rate, presence or absence of vegetation, plant species and flow type. For this purpose, an experimental plant was built at mesocosm scale and operated for more than 3 years. Each of the 8 wetlands studied incorporated a different design parameter, and pairwise wetland comparisons were conducted to determine their effect. The most commonly used technologies for horizontal flow were evaluated (floating macrophytes, free water surface flow and subsurface flow). Common reed (Phragmites australis) and cattail (Typha angustifolia) were chosen in order to evaluate the effect of plant specie due to their wide use in constructed wetlands. Several clogging indicators were used to characterise the hydraulic behaviour of the wetlands studied, including hydraulic conductivity, solids accumulation, effective volume and hydraulic efficiency. To the best of our knowledge, this is the first time that clogging has been evaluated considering such a wide variety of technologies in CWs under the same climatic and wastewater conditions.
2.
Methods
2.1.
Experimental plant
Eight mesocosm-scale CWs were placed inside the facilities of the Leo´n wastewater treatment plant (WWTP), in the northwest of Spain (Latitude: 42 330 5100 ; Longitude: 5 340 4300 ). Each CW consisted of a fibreglass container (80 cm wide, 130 cm long and 55 cm high). A diagram of the experimental device is shown in Fig. 1. The CWs differed from each other in their design configuration. CW1 and CW5 were constructed as soilless wetlands with floating macrophytes. In these two wetlands, water depth was 30 cm and plant species were supported by garden net cylinders. CW2, CW3 and CW4 were designed as free water surface (FWS) systems, with 25 cm of siliceous gravel (d10 ¼ 4 mm) and 50 cm of water depth. CW2 was a strict FWS with inlet and outlet pipes located at the wetland surface. In contrast, CW3 and CW4 were FWS systems with the outlet placed at the bottom of the container, thus forcing the water to take a partially subsurface pathway. CW6, CW60 and CW7 were designed as subsurface flow (SSF) wetlands, with 50 cm of gravel layer and 45 cm of water depth. CW60 received twice the organic load as CW6. The macrophyte species chosen were Typha angustifolia for CW1, CW2 and CW3, and Phragmites australis for CW5, CW6 and CW60 . Finally, CW4 and CW7 were unplanted systems. Consequently, pairwise comparison of wetlands, differing in only one design parameter, allowed us to evaluate the effect of different design parameters, as shown in Fig. 1. The experimental plant began operating in May 2007 and continued until December 2010. Each wetland was fed with homogenised wastewater from the primary settler at the Leo´n
1450
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 4 4 8 e1 4 5 8
Fig. 1 e Schematic design configuration of the CWs and the effects evaluated through their comparison.
WWTP. The wetlands had a hydraulic loading rate of 50 mmd1 for most of their life span (CW60 received 100 mmd1) with continuous flow rate. This determined an organic loading rate of 3 g BODm2 d1 in summer and 10 g BODm2 d1 in winter and a solids loading rate of 3 g TSSm2 d1 in summer and 15 g TSSm2 d1 in winter (further data on chemical characteristics and removal in Hijosa-Valsero et al., 2010, 2012). However, for the last four months (from September to December 2010) the wetlands were fed with twice the hydraulic loading rate. Note that differences of organic loads between winter and summer were determined by the occurrence of water filtration from irrigation channels for agricultural practises during summer, which diluted the wastewater at the WWTP.
2.2.
2.3.
Solids accumulation
In order to obtain the distribution of accumulated solids, samples of interstitial water were taken in September 2010 from two points at the inlet and two points at the outlet of each wetland and at two different depths, 15 cm and 32 cm. Samples were taken using a syringe attached to a flexible rubber pipe, which was joined to a methacrylate perforated tube with a diameter of 5 mm. 100 mL of interstitial water were carefully sucked out for TSS and VSS analysis according to APHA-AWWA-WPCF (2001). Furthermore, when the CWs were dismantled, the macrophytes were separated from the gravel bed, and aboveand below-ground plant materials were separated, washed, dried and weighed.
Saturated hydraulic conductivity 2.4.
Saturated hydraulic conductivity (K ) over time was measured at four points in each wetland (2 at the inlet and 2 at the outlet) in three sampling campaigns (November 2007, October 2008 and December 2010). The falling head permeameter method described in detail in Pedescoll et al. (2011b) with the modification applied in Pedescoll et al. (2011a) was used. The initial hydraulic conductivity (484 142 md1) considered for the calculation of permeability loss was that given in Pedescoll et al. (2011a), since we used the same gravel in our wetlands as that used in the study cited.
Effective volume
Effective volume was measured in October 2010, coinciding with the end of the experimental plant operational period. For this purpose, the CWs were emptied and the water collected was measured and compared with the theoretical effective volume, which was calculated by means of equation (1): V¼
h ðð2Wh þ Wb ÞLh þ ð2Wb þ Wh ÞLb Þ h 6
(1)
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 4 4 8 e1 4 5 8
where V is the effective volume, in m3; h is the water column height, in m; Wh is the width of the CW container at height h, in m; Wb is the width of the CW container at the base, in m; Lb is the length of the CW container at height h, in m; Lh is the length of the CW container at the base, in m; and h is the porosity of gravel, in a unit basis. Note that, for these systems, a difference of 1 cm in water table involves a difference of 4e10 L depending on the presence or absence of gravel matrix, respectively. This could represent up to 4% of the total volume. Therefore, it can be considered that the effective volume measurement has an error up to 10% and results constitute a rough value in order to observe general trends in effective volume loss.
2.5.
1 s2q ¼ 2D 2D2 1 exp D
ZN ðt sÞ2 EðtÞdt s2q ¼
V Q
(2)
where sn is the nominal HRT, in d; V is the effective volume, in m3 (calculated from equation (1)); and Q is the influent flow rate, in m3 d1. The nominal HRT was compared to the mean actual HRT, which is the average time that wastewater remains in the system, obtained from the hydraulic retention time distribution (RTD) function: ZN s¼
tEðtÞdt
(3)
0
where s is the actual HRT, in hours; t is the time, in hours; and E(t) is the RTD function, in h1. The dispersion number (D) allows determination of how far the hydraulic behaviour of the wetland is from the ideal plug flow. In this study, D was calculated assuming a plug flow model with dispersion in closedclosed boundary conditions using equation (4). According to this model, D values higher than 0.01 are indicative of large deviations from plug flow (Levenspiel, 1999). In CWs, dispersion numbers present a wide range, from 0.07 to 0.35, although these values are within the acceptable mixing range (Kadlec and Wallace, 2009).
0
s
(5)
The squared difference between the theoretical curve and that obtained in the experiment was minimised to estimate the value of the dispersion number by means an iterative procedure using either the Solver or Goal-Seek functions in MS Excel. 4¼
sn ¼
(4)
where D is the dispersion number, dimensionless; and s2q is the normalised variance obtained from the RTD curve as follows:
Hydraulic behaviour determined by tracer test
A tracer test was performed in October 2010 in order to analyse the hydraulic behaviour of the experimental plant. It was conducted for six days (from September 30th at 11:00am to October 5th at 10:00pm) by adding a single-shot injection of 694.4 mg KBrL1 into the inlet tubes for 2 min. Effluent samples were taken from each wetland every 4 h and analysed for bromide concentration by ion chromatography (Dionex ICS-1000, Thermo Fisher Scientific, Sunnyvale, California, USA). At each sampling time, effluent flow rate was measured as well. Thus, evapotranspiration was taken into account for the calculation of actual retention time. According to the plug flow model with dispersion (Garcı´a et al., 2004) and using the tracer response curve, the parameters defining the hydraulic behaviour of the cells were obtained: mean actual hydraulic retention time (HRT), normalised variance, dispersion number, peak tracer concentration time and hydraulic efficiency. Briefly, nominal HRT in horizontal CWs is defined by:
1451
n 2 1X f ðtÞi gðtÞi /0 n i¼1
(6)
where 4 is the error between the model and the experimental data; n is the number of data; f(t)i is the value of the experimental normalised RTD function at time i; and g(t)i is the value of the modelled RTD function at time i. Finally, hydraulic efficiency is a measure that combines the effective volume in the system and the RTD: l¼
tp sn
(7)
where l is hydraulic efficiency, dimensionless; tp is the time taken for the tracer to reach a peak in the experimental RTD function, in hours; and sn is the nominal HRT, in hours. According to Persson et al. (1999), the hydraulic efficiency of a CW can be considered good when l > 0.75, satisfactory when 0.5
3.
Results and discussion
Distribution maps provide a graphic depiction of the water flow through each wetland. In this case, Figs. 2 and 3 present the solids distribution measured at two different depths, 15 cm (Fig. 2) and 32 cm (Fig. 3). Since water tends to flow along the paths of least resistance, regions of preferential flow paths accumulate more solids over time. Accordingly, higher solids accumulation (darker regions in the plot) indicates the preferential water course within the wetland. Other studies have used saturated hydraulic conductivity distribution maps in order to determine flow within the system (Knowles et al., 2010; Pedescoll et al., 2011b). In fact, hydraulic conductivity constitutes an indirect measure of clogging since its relationship with accumulated solids has been proved previously (Pedescoll et al., 2011a). Thus, Fig. 4 shows the saturated hydraulic conductivity and the solids distribution map for two of the 8 wetlands studied. As can be seen, regardless of the values, the pattern was the same. However, as hydraulic conductivity cannot be measured in soilless wetlands (nor on the surface of FWS systems), a solids distribution map was used for the evaluation of flow paths and clogging distribution in the experimental plant. Tracer test results, effective volume and hydraulic conductivity complement the information provided by solids distribution maps.
1452
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 4 4 8 e1 4 5 8
Fig. 2 e Solids distribution map at 15 cm depth for the CWs at the experimental plant. Values are plotted in mgLL1. The arrow indicates the flow direction from inlet to outlet.
The results will be discussed analysing the effect examined through a pairwise comparison of wetlands which only differed in one design parameter, as shown in Fig. 1. The following sections detail which wetlands are compared for each design parameter evaluation.
than hydroponic system (CW5) and FWS combined with SSF (CW4). A detailed discussion of the water quality in the experimental plant is available in Hijosa-Valsero et al. (2010) for the first year of operation, and Hijosa-Valsero et al. (2012) for the whole experimental period.
3.1.
3.2.
Treatment performance of the experimental plant
Contaminant removal efficiencies were in all cases above 75% for BOD5 and 90% for TSS. The contaminant concentrations at the outlet reached values below the limits established by the Council Directive 91/271/EEC (Table 1). However, all CWs presented seasonality in treatment performance. Winter removal efficiencies were lower than summer, partially due to the seasonality of influent concentrations, but also to the environmental conditions (Hijosa-Valsero et al., 2012). Although significantly differences were observed at the beginning of the experimental period, in the long term these differences became less pronounced, which would be partially attributed to clogging processes. With regards to summer results, even if pollutant effluent concentrations kept below discharge limits, removal efficiencies decreased throughout time in all CWs. As far as design configurations are concerned, it was verified that the presence of vegetation improved the treatment, especially for TKN, ammonium and orthophosphate removal (CW3eCW4 and CW6eCW7 comparison). For plant specie performance (CW1eCW5), cattail worked better during the first year. However, at the end of the study, common reed removed nitrogen more efficiently. Regarding the gravel matrix presence and flow type, the strict SSF wetlands (CW6 and CW7) were more efficient in orthophosphate removal
Effect of organic loading rate
The effect of the organic loading rate was determined by comparing CW6 and CW6ʹ. Both wetlands presented strict subsurface flow and were planted with Phragmites australis. CW6 was fed with a continuous flow of 50 mmd1 while CW6ʹ received 100 mmd1 of wastewater. Leo´n wastewater presented a pronounced seasonal variability, in that the water was more diluted in summer than in winter. Accordingly, CW6 received loading rates of 3 g BOD m2 d1 and 3 g TSSm2 d1 in summer and 10 g BOD m2 d1 and 15 g TSSm2 d1 in winter. Thus, CW6ʹ received twice the load. As expected, CW6ʹ accumulated more solids than CW6 (Table 2). On average, solids accumulation in CW6ʹ was quadrupled (0.31 0.27 kg TSSm2 and 0.08 0.02 kg TSSm2 for CW6ʹ and CW6, respectively). Nevertheless, these values are clearly lower than those reported in the literature for treatment systems at full scale (Tanner and Sukias, 1995; Tanner et al., 1998; Caselles-Osorio et al., 2007; Chazarenc et al., 2009). Furthermore, from the amount of solids shown in Fig. 5, the accumulation was more heterogeneous in the overloaded wetland. Whereas accumulated solids in CW6 presented the same range regardless of depth (c.a. 1000e1800 mgL1), in CW6ʹ solids were concentrated in the top inlet region (Figs. 2 and 3). Since water was distributed on the surface, the top inlet zone
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 4 4 8 e1 4 5 8
1453
Fig. 3 e Solids distribution map at 32 cm depth for the CWs at the experimental plant. Values are plotted in mgLL1. The arrow indicates the flow direction from inlet to outlet. Note that CW1 and CW5, in which samples were taken at 15 cm only, are not represented.
constituted the first filter for solids and their accumulation was twice the concentration at the bottom. The same pattern would be expected for CW6, although delayed due to the lower solids concentration in the wastewater. Similarly, saturated hydraulic conductivity was affected by organic and solids loading rate. Although there were no significant differences between these wetlands at the end of the study (hydraulic conductivity at the inlet was 28.3 12.9 md1 and 25.2 9.9 md1 for CW6ʹ and CW6, respectively), the loss of permeability over time was greater for CW6ʹ (Fig. 6). Hydraulic conductivity is a parameter affected by the amount of solids and by their density and packing properties, but also by the amount and size of belowground plant material (Pedescoll et al., 2011a). For these two wetlands, the amount of accumulated roots at the end of the study was similar, but slightly higher for CW6 (Table 2), which might explain the lack of differences in hydraulic conductivity at the end of the study.
3.3.
Effect of vegetation
The effect of presence/absence of plants was evaluated by comparing CW3-CW4 (FWS flow with bottom outlet pipe,
planted and unplanted, respectively) and CW6-CW7 (strict SSF, planted and unplanted, respectively), while the effect of plant species by comparing CW1 and CW5 (FWS flow with floating macrophytes, planted with cattail and common reed, respectively). Many authors have reported the importance of the presence of vegetation in constructed wetlands, affecting both wastewater treatment and the development of clogging (Tanner and Sukias, 1995; Kadlec and Wallace, 2009; Chazarenc et al., 2009). The effect of the growth of roots and rhizomes can clearly be observed in the hydraulic conductivity of CW6 and CW7. At the end of the study, hydraulic conductivity was between 3 and 5 times higher in the unplanted wetland than in the planted one (25.16 9.94 md1 and 124.10 54.38 md1 at the inlet and 93.19 61.89 md1 and 341.48 33.15 md1 at the outlet for CW6 and CW7, respectively), which is in accordance with other studies measuring hydraulic conductivity in planted and unplanted wetlands (Sanford et al., 1995; Pedescoll et al., 2011a). No statistical differences between CW3 and CW4 were found (Fig. 2), probably due to the type of flow in these wetlands. Since these systems combined FWS and SSF, a large part of the solids could have settled on the gravel and decreased the
1454
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 4 4 8 e1 4 5 8
Fig. 4 e Comparison of hydraulic conductivity (plotted in green values in mdL1) and solids accumulation (plotted in brown values in mgLL1) patterns for CW3 and CW7. The arrow indicates the flow direction from inlet to outlet. Note that saturated hydraulic conductivity was measured in the first 15 cm of gravel corresponding to the TSS at 32 cm and 15 cm depth for CW3 and CW7 respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Table 1 e Averaged pollutant concentrations in influent and effluent for the whole study period (Standard deviation in brackets). Design configuration Influent CW1 CW2 CW3 CW4 CW5 CW6 CW6’ CW7
Hydroponic T. angustifolia FWS T. angustifolia FWS-bottom outlet T. angustifolia FWS-bottom outlet unplanted SSF P. australis SSF P. australis SSF P. australis Overloaded SSF unplanted
COD mg O2L1
BOD5 mg O2L1
TSS mgL1
TKN mgL1
NHþ 4 eN mgL1
NO 3 eN mgL1
PO3 4 eP mgL1
177.43 (165.00) 49.92 (31.29)
83.57 (60.60) 17.04 (13.96)
86.57 (99.09) 9.90 (9.76)
20.00 (6.32) 11.71 (7.38)
13.57 (3.21) 10.00 (6.84)
0.75 (0.73) 0.33 (0.55)
1.92 (0.49) 1.65 (1.17)
57.32 (40.11) 39.37 (24.23)
17.30 (13.77) 12.74 (10.89)
12.02 (8.76) 12.04 (8.93)
12.84 (7.66) 11.82 (7.54)
10.58 (7.05) 10.30 (7.13)
0.28 (0.24) 0.15 (0.10)
1.74 (1.07) 1.30 (1.18)
57.67 (35.21)
17.49 (11.79)
14.51 (12.19)
17.07 (7.27)
14.70 (6.69)
0.16 (0.12)
2.33 (1.27)
49.27 (31.91) 39.02 (22.56) 56.64 (32.65)
19.97 (17.23) 16.86 (12.04) 29.28 (20.03)
11.44 (10.17) 29.63 (18.94) 16.15 (11.74)
10.67 (8.65) 8.45 (6.83) 17.73 (6.12)
9.01 (8.60) 7.30 (6.95) 16.24 (5.79)
0.19 (0.10) 0.43 (0.28) 0.25 (0.20)
2.11 (1.30) 0.39 (0.67) 1.98 (1.27)
36.75 (27.39)
14.64 (12.77)
10.36 (8.78)
18.18 (3.75)
17.17 (3.44)
0.23 (0.17)
1.90 (1.25)
16 127.0 150.7
44 37 1 127.0 165.0 120.0 1.30 1.53 1.62
225.9 167.0 191.7
13 291.0 258.5
41 37 181.0 192.0 306.0 306.0 14.4 10.1
9.5 13.4 8.3
5.09 7.78
7.33 8.77 5.11
2.83 1.30
55 102.0 225.9 24.5 9.84
2.48
Theoretical effective volume (L) Below ground Above ground Biomass ratio Below-ground plant dry weight (kgm2) Above-ground plant dry weight (kgm2)
CW7
CW5 CW6 CW6’
CW4
CW2 CW3
a Range of solids concentration.
0.046e0.304
0.024e0.183 0.057e0.109 0.029e0.716
0.003e0.322
0.003e0.108 0.041e1.120
0.030e0.326
Hydroponic T. angustifolia FWS T. angustifolia FWS-bottom outlet T. angustifolia FWS-bottom outlet unplanted SSF P. australis SSF P. australis SSF P. australis Overloaded SSF unplanted CW1
Accumulated interstitial solidsa (kgm2)
1455
hydraulic conductivity measured through a vertical core. In FWS wetlands, TSS removal occurs mainly by settling and flocculation processes in the wetland and also by anaerobic decomposition of the settled organic matter (USEPA, 2000). In accordance with the above, the amount of solids was higher in the unplanted wetland gravel bed (in Fig. 6, the gravel region corresponds to the bottom of CW4), whereas in the region of free water, surface solids were more concentrated in the planted one (CW3). The solids distribution pattern for CW3 and CW4 suggests that flow in the unplanted cell occurred mainly along the surface until the water reached the outlet, where it was forced to infiltrate through the gravel. In contrast, it seems that the presence of plants promoted the partial flow of water through the bottom (note that in the gravel medium, solids concentrated at the inlet in CW3, Fig. 5). In addition to the above, we also found differences between the plant species used in this study. Since belowground plant growth entails a substantial decrease in effective volume and hydraulic conductivity (Edwards, 1992), the species chosen for use in any CW should be carefully considered in terms of the volume which will be occupied by the roots and rhizomes, among other characteristics. The amount of below-ground biomass produced after almost 4 years of operation of the experimental plant is shown in Table 2. In soilless wetlands, T. angustifolia presented more than twice the weight as Phragmites australis. Consequently, there was 10% less effective volume in CW1, planted with cattail. The values for above-ground biomass are in accordance with those reported by Vymazal and Kro¨pfelova´ (2005) for the usual amount of biomass measured in horizontal CWs planted with P. australis (788e5070 g DMm2). On average, the belowground/above-ground ratio was lower in wetlands planted with common reed (2.20 0.80 and 1.48 0.16 for T. angustifolia and P. australis, respectively), with similar values to those reported by Ouellet-Plamondon et al. (2004). Therefore, cattail seems to have a greater effect on the clogging of subsurface flow wetlands. Root density in the overloaded system (CW6ʹ) was lower than in the not overloaded one (CW6) due to much harsher conditions in the first system (lower redox and higher nutrient concentration).
3.4.
Design configuration
Table 2 e Accumulated solids and plant biomass in the CWs and effective volume loss after 44 months of experimental plant operation.
Actual effective volume (L)
Effective volume loss (%)
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 4 4 8 e1 4 5 8
Effect of flow type
The effect of flow type can be evaluated at two levels. Comparison between the CW4 and CW7 tanks enabled us to determine the differences between two unplanted systems, a FWS wetland with the outlet located at the bottom of the cell (CW4) and a strict SSF wetland (CW7). A comparison of the two FWS wetlands with different outlets, a top inlet and top outlet position in tank CW2 vs. FWS with top inlet and bottom outlet in tank CW3 was also conducted. As discussed in the previous section, in CW4 (FWS flow with water leaving at the bottom) the wastewater flowed mainly along the surface, which offers less resistance, and finally infiltrated near the outlet to reach the drainage pipe located at the bottom of the cell instead of using the whole effective volume of the gravel. Consequently, hydraulic efficiency was lower in CW4 than in CW7, in which the water was forced to cross the wetland through the subsurface. Peak tracer concentration time was clearly lower in CW4 (Table 3),
1456
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 4 4 8 e1 4 5 8
Fig. 5 e Amount of solids (TSS) and percentage of volatile solids (VSS/TSS) in the CWs. measured at 15 (surface) and 32 cm depth (bottom).
thus supporting findings for preferential flow on the surface and decreased retention time in the cell. Furthermore, solids accumulation was more homogeneous in CW7 (Figs. 2 and 3) compared to CW4. However, in the unplanted SSF wetland (CW7), the water flowed diagonally in a direct line from the top inlet to the bottom outlet, as has been observed by others (Suliman et al., 2006; Pedescoll et al., 2011b). Since the solids settled on the surface of the gravel in CW4, hydraulic conductivity decreased drastically compared with CW7 (0.33 1.53 md1 and 124.10 54.38 md1 at the inlet and 0.95 0.61 md1 and 341.48 33.15 md1 at the outlet for CW4 and CW7, respectively). Therefore, water only infiltrated the gravel at the outlet in CW4, thus losing half of the effective volume. With regard to the comparison of the FWS tanks, solids accumulation was higher in CW3 than in CW2 (0.056 0.031 Kgm2 and 0.303 0.409 Kgm2 on average for CW2 and CW3, respectively), and specifically at the gravel matrix (Fig. 6). Hence, there is more effective use of the cell volume in CW3, and in fact the water remains longer in this wetland (Table 3), further supporting findings for the effect of the inlet-outlet position on hydraulic behaviour. Suliman et al.
(2007) stated that the best strategy to increase hydraulic efficiency and retention time in constructed wetlands was to employ a bottom-inlet/top-outlet position, in order to force the water flow against gravity.
3.5.
Effect of gravel matrix
The effect of the gravel matrix was evaluated by comparing CW1 (soilless FWS planted with cattail) vs. CW2 (FWS flow planted with cattail) and CW5 (soilless FWS planted with common reed) vs. CW6 (strict SSF planted with common reed). The presence of a gravel matrix should increase the hydraulic efficiency of a system compared to a mixed flow in those wetlands with hydroponic conditions (soilless FWS) (Table 3), especially when strict SSF is considered. However, the higher hydraulic efficiency observed in CW6 seems to be due to the presence of dead zones (note that the peak tracer concentration time was clearly higher than for the rest of the systems studied and the mean HRT was almost 70% higher than the theoretical HRT). Conversely, the wastewater tended to plug flow when solids concentration distribution was considered (Figs. 2 and 3). Solids were accumulated in a decreasing gradient from inlet to
Fig. 6 e Temporal changes in hydraulic conductivity at inlet and outlet zones in the CWs. Note that values are plotted as a percentage of the initial one.
65.8 56.1 83.9 94.8 27 47 35 23 104 104 104 104 4.01 8.63 6.12 4.02
0.10 0.14 0.24 0.35
0.50 1.11 1.55 0.67
84.2 15 7.06 104 0.25
0.24
71.1 50.7 62.8 0.42 0.28 0.28 19 19 19 6.04 104 5.50 104 8.62 104 0.62 1.49 0.23
a Dimensionless.
CW5 CW6 CW6’ CW7
CW4
Conclusions
This study evaluated different design characteristics which mainly affect clogging distribution patterns. Under the conditions tested in the eight mesocosm scale CWs, differences were observed depending on the design parameter tested. Organic loading rate is related to the amount of solids accumulated within the wetland. The overloaded SSF CW presented four times more solids, and the distribution pattern was as would be expected when a wetland operates at recommended loads (w6 g BODm2 d1). The presence of plants entails a substantial decrease in the hydraulic conductivity of the gravel medium, with a loss of permeability which is around 20% higher in planted wetlands than in unplanted ones. The plant species chosen may also have implications for clogging: T. angustifolia seems to have a greater effect since its below-ground biomass weighed twice that of P. australis. Hydraulic behaviour, and therefore water flow paths, was greatly influenced by the presence of a gravel matrix and the outlet pipe position. In strict SSF CWs, the water was forced to cross the gravel and tended to flow diagonally from the top inlet to the bottom outlet (where the inlet and outlet pipes were located). However, when one zone of FWS was considered, water spread preferentially on the surface above the gravel, thus loosing half the effective volume of the system. Only the presence of plants seemed to help the water to flow partially within the gravel matrix.
0.19 0.21 0.39 0.49 37.4 71.1 52.8 41.8 53.6 42.2 22.5 34.6
0.41 54.0 61.5
0.65 0.83 0.39 37.4 29.6 37.8 45.7 69.0 68.0
Hydroponic T. angustifolia FWS T. angustifolia FWS-bottom outlet T. angustifolia FWS-bottom outlet unplanted SSF P. australis SSF P. australis SSF P. australis Overloaded SSF unplanted
1457
outlet in CW6, regardless of depth, whereas in CW5, solids were concentrated mainly on the left side of the cell. In contrast, the presence of the gravel in CW2 reduced the HRT and significantly increased tracer dispersion, thus reducing hydraulic efficiency. This system was a FWS wetland, with the wastewater entering and exiting on the surface. Since water flows along the path of least resistance, the gravel in CW2 merely constituted support for the vegetation because the wastewater mainly flowed through the top 25 cm of the wetland. Solids concentration was in the same range both at the top and bottom of the cell, and was lower than in the soilless wetland (CW1), which suggests that solids settled on the surface of the gravel layer, as discussed in the previous section. Differences between treatments were observed in terms of solids characteristics. As shown in Fig. 5, the percentage of volatile solids (and thus organic matter) was around 60% when FWS was considered, whereas the ratio decreased for solids accumulated in the gravel (that is, in CW6, CW6ʹ, CW7 and the bottom of CW2, CW3 and CW4). This difference in the proportion of volatile solids could be explained by the accumulation of litter from plant decay, together with algal growth ´ lvarez and Be´cares, 2006; Hijosain FWS flow wetlands (A Valsero et al., 2010).
4.
CW1 CW2 CW3
Peak tracer concentration time tp (hours) Error 4 (dima) Dispersion number D (dima) Normalised variance sq (dima) Actual HRT s (hours) Nominal HRT sn (hours) Design configuration
Table 3 e Hydraulic parameters obtained from the tracer experiments carried out after 44 months of experimental plant operation.
Hydraulic efficiency l (dima)
Recovery (%)
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 4 4 8 e1 4 5 8
Acknowledgements This study was funded by the Spanish Ministry of Science through the projects CTM2005-06457-C05-03 and CTM2008-
1458
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 4 4 8 e1 4 5 8
06676-C05-03/TECNO. Anna Pedescoll acknowledges the Juan de la Cierva Programme of the Spanish Ministry of Science and Innovation. Marianna Garfı´ is grateful to the Secretary General of Universities, Ministry of Education (Spain) (Programa Nacional de Movilidad de Recursos Humanos del Plan Nacional de I-Dþi 2008-2011). The authors thank Amando Escudero Marina and Lina Tyroller for his help in the field.
references
´ lvarez, J.A., Be´cares, E., 2006. Seasonal decomposition of Typha A latifolia in a free-water surface constructed wetland. Ecological Engineering 28 (2), 99e105. APHA-AWWA-WPCF, 2001. Standard Methods for the Examination of Water and Wastewater (2001), twentieth ed.. Washington DC, USA. Caselles-Osorio, A., Puigagut, J., Segu´, E., Vaello, N., Grane´s, F., Garcı´a, D., Garcı´a, J., 2007. Solids accumulation in six full-scale subsurface flow constructed wetlands. Water Research 41 (6), 1388e1398. Chazarenc, F., Gagnon, V., Comeau, Y., Brisson, J., 2009. Effect of plant and artificial aeration on solids accumulation and biological activities in constructed wetlands. Ecological Engineering 35, 1005e1010. Cooper, P., 2005. The performance of vertical flow constructed wetland systems with special reference to the significance of oxygen transfer and hydraulic loading rates. Water Science and Technology 51 (9), 81e90. Council Directive 91/271/EEC of 21 May 1991 Concerning urban waste water treatment. Official Journal of the European Communities L 135 of 30 May 1991, pp. 40e52. Edwards, G.S., 1992. Root distribution of soft-stem bulrush (Scirpus validus) in a constructed wetland. Ecological Engineering 1 (3), 239e243. Garcı´a, J., Aguirre, P., Barraga´n, J., Mujeriego, R., Matamoros, V., Bayona, J.M., 2005. Effect of key design parameters on the efficiency of horizontal subsurface flow constructed wetlands: long-term performance pilot study. Ecological Engineering 25 (4), 405e418. ´ lvarez, E., Sierra, J.P., Garcı´a, J., Chiva, J., Aguirre, P., A Mujeriego, R., 2004. Hydraulic behaviour of horizontal subsurface flow constructed wetlands with different aspect ratio and granular medium size. Ecological Engineering 23 (3), 177e187. Griffin, P., Wilson, L., Cooper, D., 2008. Changes in the use, operation and design of sub-surface flow constructed wetlands in a major UK water utility. In: 11th International Conference on Wetland Systems for Water Pollution Control. Indore, India, pp. 419e426. Hijosa-Valsero, M., Sidrach-Cardona, R., Be´cares, E., 2012. Comparison of interannual removal variation of various constructed wetland types. Science of the Total Environment 430, 174e183. Hijosa-Valsero, M., Sidrach-Cardona, R., Martı´n-Villacorta, J., Be´cares, E., 2010. Optimization of performance assessment and design characteristics in constructed wetlands for the removal of organic matter. Chemosphere 81 (5), 651e657. Jenkins, G.A., Greenway, M., 2005. The hydraulic efficiency of fringing versus banded vegetation in constructed wetlands. Ecological Engineering 25, 61e72. Kadlec, R.H., Wallace, S.D., 2009. Treatment Wetlands, second ed. CRC Press, Boca Raton, FL.
Knowles, P., Dotro, G., Nivala, J., Garcı´a, J., 2011. Clogging in subsurface-flow treatment wetlands: ocurrence and contributing factors. Ecological Engineering 37 (2), 99e112. Knowles, P.R., Griffin, P., Davies, P., 2010. Complementary methods to investigate the development of clogging within a horizontal sub-surface flow tertiary treatment wetland. Water Research 44 (1), 320e330. Levenspiel, O., 1999. Chemical Reaction Engineering, third ed. John Wiley and Sons, Inc, Hoboken, USA, p. 668. Ouellet-Plamondon, C.M., Brisson, J., Comeau, Y., 2004. Effect of macrophyte species on subsurface flow wetland performance in cold climate. In: Proceedings of the 2004 Self-sustaining Solutions for Streams, Wetlands and Watersheds Conference. St. Paul, MN; September 12e15, 2004. pp. 8e15. ´ lvarez, E., Garcı´a, J., Puigagut, J., 2011a. Pedescoll, A., Corzo, A., A The effect of primary treatment and flow regime on clogging development in horizontal subsurface flow constructed wetlands: an experimental evaluation. Water Research 45 (12), 3579e3589. Pedescoll, A., Knowles, P.R., Davis, P., Garcı´a, J., Puigagut, J., 2011b. A Comparison of in situ constant and falling head permeameter tests to assess the distribution of clogging within horizontal subsurface flow constructed wetlands. Water, Air and Soil Pollution. http://dx.doi.org/10.1007/s11270011-1021-4. Persson, J., 2000. The hydraulic performance of ponds of various layouts. Urban Water 2 (3), 243e250. Persson, J., Somes, N.L.G., Wong, T.H.F., 1999. Hydraulics efficiency of constructed wetlands and ponds. Water Science and Technology 40 (3), 291e300. Rousseau, D.P.L., Horton, D., Griffin, P., Vanrollenghem, P.A., De Paw, N., 2005. Impact of operational maintenance on the asset life of storm reed beds. Water Science and Technology 51 (9), 243e250. Sanford, W.E., Steenhuis, T.S., Parlange, J.Y., Surface, J.M., Peverly, J.H., 1995. Hydraulic conductivity of gravel sand as substrates in rock-reed filters. Ecological Engineering 4 (4), 321e336. Suliman, F., Futsaether, C., Oxaal, U., 2007. Hydraulic performance of horizontal subsurface flow constructed wetlands for different strategies of filling the filter medium into the filter basin. Ecological Engineering 29 (1), 45e55. Suliman, F., Futsaether, C., Oxaal, U., Haugen, L.E., Jenssen, P., 2006. Effect of the inleteoutlet positions on the hydraulic performance of horizontal subsurface-flow wetlands constructed with heterogeneous porous media. Journal of Contaminant Hydrology 87 (1e2), 22e36. Tanner, C.C., Sukias, J.P.S., Upsdell, M.P., 1998. Organic matter accumulation and maturation of gravel-bed constructed wetlands treating dairy farm wastewaters. Water Research 32 (10), 3046e3054. Tanner, C.C., Sukias, J.P., 1995. Accumulation of organic solids in gravel bed constructed wetlands. Water Science and Technology 32 (3), 229e239. U.S. EPA, 2000. Constructed Wetlands Treatment of Municipal Wastewaters. EPA 625/R-99/010, U.S.. EPA Office of Research and Development, Washington D.C. Vymazal, J., Kro¨pfelova´, L., 2005. Growth of Phragmites australis and Phalaris arundinacea in constructed wetlands for wastewater treatment in the Czech Republic. Ecological Engineering 25 (5), 606e621. Wallace, S.D., Knight, R.L., 2006. Small-scale Constructed Wetland Treatment Systems: Feasibility, Design Criteria and O&M Requirements. Final Report, Project 01-CTS-5. Water Environment Research Foundation (WERF), Alexandria, USA.
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Design configurations affecting flow pattern and solids accumulation in horizontal free water and subsurface flow constructed wetlands A. Pedescoll a,b,*, R. Sidrach-Cardona a,b, J.C. Sa´nchez a,b, J. Carretero c, M. Garfi c, E. Be´cares a,b a
Environmental Institute, University of Leo´n, c/La Serna 56, 24007 Leo´n, Spain Ecology Section, Department of Biodiversity and Environmental Management, University of Leo´n, Campus de Vegazana s/n, 24071 Leo´n, Spain c GEMMA e Group of Environmental Engineering and Microbiology, Department of Hydraulic, Maritime and Environmental Engineering, Universitat Polite`cnica de Catalunya-BarcelonaTech, c/Jordi Girona 1-3, Building D1, E-08034 Barcelona, Spain b
article info
abstract
Article history:
The aim of this study was to evaluate the effect of different horizontal constructed wetland
Received 20 July 2012
(CW) design parameters on solids distribution, loss of hydraulic conductivity over time and
Received in revised form
hydraulic behaviour, in order to assess clogging processes in wetlands. For this purpose, an
5 December 2012
experimental plant with eight CWs was built at mesocosm scale. Each CW presented
Accepted 7 December 2012
a different design characteristic, and the most common CW configurations were all rep-
Available online 19 December 2012
resented: free water surface flow (FWS) with different effluent pipe locations, FWS with floating macrophytes and subsurface flow (SSF), and the presence of plants and specific
Keywords:
species (Typha angustifolia and Phragmites australis) was also considered. The loss of the
Clogging
hydraulic conductivity of gravel was greatly influenced by the presence of plants and
Flow path
organic load (representing a loss of 20% and c.a. 10% in planted wetlands and an over-
Solids distribution
loaded system, respectively). Cattail seems to have a greater effect on the development of
Permeability
clogging since its below-ground biomass weighed twice as much as that of common reed.
Cattail
Hydraulic behaviour was greatly influenced by the presence of a gravel matrix and the
Common reed
outlet pipe position. In strict SSF CW, the water was forced to cross the gravel and tended to flow diagonally from the top inlet to the bottom outlet (where the inlet and outlet pipes were located). However, when FWS was considered, water preferentially flowed above the gravel, thus losing half the effective volume of the system. Only the presence of plants seemed to help the water flow partially within the gravel matrix. ª 2012 Elsevier Ltd. All rights reserved.
1.
Introduction
Constructed wetlands (CWs) are extensive, low energy systems for wastewater treatment which do not require specialised
manpower for their management (Wallace and Knight, 2006). For this reason, and due to the large amount of land required for wastewater treatment, CWs are usually recommended for the sanitation of small communities (Cooper, 2005).
* Corresponding author. Environmental Institute, University of Leo´n, c/La Serna 56, 24007 Leo´n, Spain. Tel.: þ34 987 29 10 00x5227; fax: þ34 987 29 15 63. E-mail addresses: [email protected], [email protected] (A. Pedescoll), [email protected] (R. Sidrach-Cardona), nocardia. [email protected] (J.C. Sa´nchez), [email protected] (J. Carretero), [email protected] (M. Garfi), ebecm@ unileon.es (E. Be´cares). 0043-1354/$ e see front matter ª 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.watres.2012.12.010
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 4 4 8 e1 4 5 8
Despite the advantages that CWs offer, they also present a major operational problem, namely clogging of the gravel media. Clogging is the progressive loss of the initial hydraulic characteristics, mainly porosity and hydraulic conductivity of the granular media. This phenomenon is due to the accumulation of solids from the wastewater, biofilm and plant growth and chemical precipitates (Knowles et al., 2011), and can lead to a decrease in contaminant removal efficiency over time. Although clogging is an unavoidable process (since it is inherent to the treatment), it can be delayed by improving the design (Persson et al., 1999), operation and maintenance (Griffin et al., 2008) of the wetland. In fact, the degree of clogging depends on the accumulation of different nature solids, whereas the distribution of clogging basically depends on the hydraulic behaviour of the system. In order to determine whether intervention to reverse clogging is required, it is essential to assess not only the degree of clogging, but also its spatial distribution throughout the wetland. Since clogging occurs through the accumulation of solids in the interstitial spaces of the gravel, it is obvious that the organic load, and especially the solids loading rate, are of paramount importance in the design. In the early stages after commissioning the plant, clogging occurs mainly at the inlet, spreading progressively towards the outlet. It is at the entrance to the wetland where solids are filtered, thus most of the suspended solids are retained in this zone. Therefore, an overload promotes the increase of sludge in this area, followed by rapid loss of the water infiltration rate and the appearance of ponding (Rousseau et al., 2005). Many authors have shown that the solids loading rate is directly related to sludge accumulation (Tanner et al., 1998; Caselles-Osorio et al., 2007; Pedescoll et al., 2011a). Primary treatments based on physical processes play an important role in decreasing the solids load applied to the system. However, if poorly designed or maintained, they can contribute to an undesirable solids overload at the wetland inlet (Knowles et al., 2010). Many other factors also contribute to undesirable clogging distribution. Water tends to flow along the path of least hydraulic resistance (Pedescoll et al., 2011b), which is influenced by different design parameters and operational aspects, including the length to width ratio (Garcı´a et al., 2005; Jenkins and Greenway, 2005; Persson, 2000) and the inlet and outlet position. Suliman et al. (2006, 2007) observed differences in flow pattern of an experimental flow cell depending on the strategy employed to fill the filter medium in the basin and the inlet-outlet position. The development of below-ground vegetation may have implications for wetland hydraulics (Knowles et al., 2011), because the growth of below-ground plants involves a loss of effective volume, which is occupied instead by the roots and rhizomes. The hydraulic conductivity of the medium in similar systems is greater in unplanted wetlands (Sanford et al., 1995). Edwards (1992) estimated the volume occupied by roots as being 5% of total substrate volume of a subsurface flow wetland (45 cm deep) planted with bulrushes (Scirpus validus), where roots penetrated between 12 and 15 cm below the surface. Moreover, Pedescoll et al. (2011a) observed a drastic decrease in hydraulic conductivity in shallow subsurface flow CWs in which the below-ground biomass penetrated the entire wetland depth (25 cm of water column).
1449
In addition, the growth of roots and rhizomes creates a zone of increased resistance to water flow, promoting the establishment of preferential flow paths. Jenkins and Greenway (2005) found decreased hydraulic efficiencies with increasing vegetation cover and density. The aim of this study was to evaluate the hydraulic performance of eight different configurations of wetlands in terms of clogging development, and to compare design parameters such as organic loading rate, presence or absence of vegetation, plant species and flow type. For this purpose, an experimental plant was built at mesocosm scale and operated for more than 3 years. Each of the 8 wetlands studied incorporated a different design parameter, and pairwise wetland comparisons were conducted to determine their effect. The most commonly used technologies for horizontal flow were evaluated (floating macrophytes, free water surface flow and subsurface flow). Common reed (Phragmites australis) and cattail (Typha angustifolia) were chosen in order to evaluate the effect of plant specie due to their wide use in constructed wetlands. Several clogging indicators were used to characterise the hydraulic behaviour of the wetlands studied, including hydraulic conductivity, solids accumulation, effective volume and hydraulic efficiency. To the best of our knowledge, this is the first time that clogging has been evaluated considering such a wide variety of technologies in CWs under the same climatic and wastewater conditions.
2.
Methods
2.1.
Experimental plant
Eight mesocosm-scale CWs were placed inside the facilities of the Leo´n wastewater treatment plant (WWTP), in the northwest of Spain (Latitude: 42 330 5100 ; Longitude: 5 340 4300 ). Each CW consisted of a fibreglass container (80 cm wide, 130 cm long and 55 cm high). A diagram of the experimental device is shown in Fig. 1. The CWs differed from each other in their design configuration. CW1 and CW5 were constructed as soilless wetlands with floating macrophytes. In these two wetlands, water depth was 30 cm and plant species were supported by garden net cylinders. CW2, CW3 and CW4 were designed as free water surface (FWS) systems, with 25 cm of siliceous gravel (d10 ¼ 4 mm) and 50 cm of water depth. CW2 was a strict FWS with inlet and outlet pipes located at the wetland surface. In contrast, CW3 and CW4 were FWS systems with the outlet placed at the bottom of the container, thus forcing the water to take a partially subsurface pathway. CW6, CW60 and CW7 were designed as subsurface flow (SSF) wetlands, with 50 cm of gravel layer and 45 cm of water depth. CW60 received twice the organic load as CW6. The macrophyte species chosen were Typha angustifolia for CW1, CW2 and CW3, and Phragmites australis for CW5, CW6 and CW60 . Finally, CW4 and CW7 were unplanted systems. Consequently, pairwise comparison of wetlands, differing in only one design parameter, allowed us to evaluate the effect of different design parameters, as shown in Fig. 1. The experimental plant began operating in May 2007 and continued until December 2010. Each wetland was fed with homogenised wastewater from the primary settler at the Leo´n
1450
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 4 4 8 e1 4 5 8
Fig. 1 e Schematic design configuration of the CWs and the effects evaluated through their comparison.
WWTP. The wetlands had a hydraulic loading rate of 50 mmd1 for most of their life span (CW60 received 100 mmd1) with continuous flow rate. This determined an organic loading rate of 3 g BODm2 d1 in summer and 10 g BODm2 d1 in winter and a solids loading rate of 3 g TSSm2 d1 in summer and 15 g TSSm2 d1 in winter (further data on chemical characteristics and removal in Hijosa-Valsero et al., 2010, 2012). However, for the last four months (from September to December 2010) the wetlands were fed with twice the hydraulic loading rate. Note that differences of organic loads between winter and summer were determined by the occurrence of water filtration from irrigation channels for agricultural practises during summer, which diluted the wastewater at the WWTP.
2.2.
2.3.
Solids accumulation
In order to obtain the distribution of accumulated solids, samples of interstitial water were taken in September 2010 from two points at the inlet and two points at the outlet of each wetland and at two different depths, 15 cm and 32 cm. Samples were taken using a syringe attached to a flexible rubber pipe, which was joined to a methacrylate perforated tube with a diameter of 5 mm. 100 mL of interstitial water were carefully sucked out for TSS and VSS analysis according to APHA-AWWA-WPCF (2001). Furthermore, when the CWs were dismantled, the macrophytes were separated from the gravel bed, and aboveand below-ground plant materials were separated, washed, dried and weighed.
Saturated hydraulic conductivity 2.4.
Saturated hydraulic conductivity (K ) over time was measured at four points in each wetland (2 at the inlet and 2 at the outlet) in three sampling campaigns (November 2007, October 2008 and December 2010). The falling head permeameter method described in detail in Pedescoll et al. (2011b) with the modification applied in Pedescoll et al. (2011a) was used. The initial hydraulic conductivity (484 142 md1) considered for the calculation of permeability loss was that given in Pedescoll et al. (2011a), since we used the same gravel in our wetlands as that used in the study cited.
Effective volume
Effective volume was measured in October 2010, coinciding with the end of the experimental plant operational period. For this purpose, the CWs were emptied and the water collected was measured and compared with the theoretical effective volume, which was calculated by means of equation (1): V¼
h ðð2Wh þ Wb ÞLh þ ð2Wb þ Wh ÞLb Þ h 6
(1)
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 4 4 8 e1 4 5 8
where V is the effective volume, in m3; h is the water column height, in m; Wh is the width of the CW container at height h, in m; Wb is the width of the CW container at the base, in m; Lb is the length of the CW container at height h, in m; Lh is the length of the CW container at the base, in m; and h is the porosity of gravel, in a unit basis. Note that, for these systems, a difference of 1 cm in water table involves a difference of 4e10 L depending on the presence or absence of gravel matrix, respectively. This could represent up to 4% of the total volume. Therefore, it can be considered that the effective volume measurement has an error up to 10% and results constitute a rough value in order to observe general trends in effective volume loss.
2.5.
1 s2q ¼ 2D 2D2 1 exp D
ZN ðt sÞ2 EðtÞdt s2q ¼
V Q
(2)
where sn is the nominal HRT, in d; V is the effective volume, in m3 (calculated from equation (1)); and Q is the influent flow rate, in m3 d1. The nominal HRT was compared to the mean actual HRT, which is the average time that wastewater remains in the system, obtained from the hydraulic retention time distribution (RTD) function: ZN s¼
tEðtÞdt
(3)
0
where s is the actual HRT, in hours; t is the time, in hours; and E(t) is the RTD function, in h1. The dispersion number (D) allows determination of how far the hydraulic behaviour of the wetland is from the ideal plug flow. In this study, D was calculated assuming a plug flow model with dispersion in closedclosed boundary conditions using equation (4). According to this model, D values higher than 0.01 are indicative of large deviations from plug flow (Levenspiel, 1999). In CWs, dispersion numbers present a wide range, from 0.07 to 0.35, although these values are within the acceptable mixing range (Kadlec and Wallace, 2009).
0
s
(5)
The squared difference between the theoretical curve and that obtained in the experiment was minimised to estimate the value of the dispersion number by means an iterative procedure using either the Solver or Goal-Seek functions in MS Excel. 4¼
sn ¼
(4)
where D is the dispersion number, dimensionless; and s2q is the normalised variance obtained from the RTD curve as follows:
Hydraulic behaviour determined by tracer test
A tracer test was performed in October 2010 in order to analyse the hydraulic behaviour of the experimental plant. It was conducted for six days (from September 30th at 11:00am to October 5th at 10:00pm) by adding a single-shot injection of 694.4 mg KBrL1 into the inlet tubes for 2 min. Effluent samples were taken from each wetland every 4 h and analysed for bromide concentration by ion chromatography (Dionex ICS-1000, Thermo Fisher Scientific, Sunnyvale, California, USA). At each sampling time, effluent flow rate was measured as well. Thus, evapotranspiration was taken into account for the calculation of actual retention time. According to the plug flow model with dispersion (Garcı´a et al., 2004) and using the tracer response curve, the parameters defining the hydraulic behaviour of the cells were obtained: mean actual hydraulic retention time (HRT), normalised variance, dispersion number, peak tracer concentration time and hydraulic efficiency. Briefly, nominal HRT in horizontal CWs is defined by:
1451
n 2 1X f ðtÞi gðtÞi /0 n i¼1
(6)
where 4 is the error between the model and the experimental data; n is the number of data; f(t)i is the value of the experimental normalised RTD function at time i; and g(t)i is the value of the modelled RTD function at time i. Finally, hydraulic efficiency is a measure that combines the effective volume in the system and the RTD: l¼
tp sn
(7)
where l is hydraulic efficiency, dimensionless; tp is the time taken for the tracer to reach a peak in the experimental RTD function, in hours; and sn is the nominal HRT, in hours. According to Persson et al. (1999), the hydraulic efficiency of a CW can be considered good when l > 0.75, satisfactory when 0.5
3.
Results and discussion
Distribution maps provide a graphic depiction of the water flow through each wetland. In this case, Figs. 2 and 3 present the solids distribution measured at two different depths, 15 cm (Fig. 2) and 32 cm (Fig. 3). Since water tends to flow along the paths of least resistance, regions of preferential flow paths accumulate more solids over time. Accordingly, higher solids accumulation (darker regions in the plot) indicates the preferential water course within the wetland. Other studies have used saturated hydraulic conductivity distribution maps in order to determine flow within the system (Knowles et al., 2010; Pedescoll et al., 2011b). In fact, hydraulic conductivity constitutes an indirect measure of clogging since its relationship with accumulated solids has been proved previously (Pedescoll et al., 2011a). Thus, Fig. 4 shows the saturated hydraulic conductivity and the solids distribution map for two of the 8 wetlands studied. As can be seen, regardless of the values, the pattern was the same. However, as hydraulic conductivity cannot be measured in soilless wetlands (nor on the surface of FWS systems), a solids distribution map was used for the evaluation of flow paths and clogging distribution in the experimental plant. Tracer test results, effective volume and hydraulic conductivity complement the information provided by solids distribution maps.
1452
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 4 4 8 e1 4 5 8
Fig. 2 e Solids distribution map at 15 cm depth for the CWs at the experimental plant. Values are plotted in mgLL1. The arrow indicates the flow direction from inlet to outlet.
The results will be discussed analysing the effect examined through a pairwise comparison of wetlands which only differed in one design parameter, as shown in Fig. 1. The following sections detail which wetlands are compared for each design parameter evaluation.
than hydroponic system (CW5) and FWS combined with SSF (CW4). A detailed discussion of the water quality in the experimental plant is available in Hijosa-Valsero et al. (2010) for the first year of operation, and Hijosa-Valsero et al. (2012) for the whole experimental period.
3.1.
3.2.
Treatment performance of the experimental plant
Contaminant removal efficiencies were in all cases above 75% for BOD5 and 90% for TSS. The contaminant concentrations at the outlet reached values below the limits established by the Council Directive 91/271/EEC (Table 1). However, all CWs presented seasonality in treatment performance. Winter removal efficiencies were lower than summer, partially due to the seasonality of influent concentrations, but also to the environmental conditions (Hijosa-Valsero et al., 2012). Although significantly differences were observed at the beginning of the experimental period, in the long term these differences became less pronounced, which would be partially attributed to clogging processes. With regards to summer results, even if pollutant effluent concentrations kept below discharge limits, removal efficiencies decreased throughout time in all CWs. As far as design configurations are concerned, it was verified that the presence of vegetation improved the treatment, especially for TKN, ammonium and orthophosphate removal (CW3eCW4 and CW6eCW7 comparison). For plant specie performance (CW1eCW5), cattail worked better during the first year. However, at the end of the study, common reed removed nitrogen more efficiently. Regarding the gravel matrix presence and flow type, the strict SSF wetlands (CW6 and CW7) were more efficient in orthophosphate removal
Effect of organic loading rate
The effect of the organic loading rate was determined by comparing CW6 and CW6ʹ. Both wetlands presented strict subsurface flow and were planted with Phragmites australis. CW6 was fed with a continuous flow of 50 mmd1 while CW6ʹ received 100 mmd1 of wastewater. Leo´n wastewater presented a pronounced seasonal variability, in that the water was more diluted in summer than in winter. Accordingly, CW6 received loading rates of 3 g BOD m2 d1 and 3 g TSSm2 d1 in summer and 10 g BOD m2 d1 and 15 g TSSm2 d1 in winter. Thus, CW6ʹ received twice the load. As expected, CW6ʹ accumulated more solids than CW6 (Table 2). On average, solids accumulation in CW6ʹ was quadrupled (0.31 0.27 kg TSSm2 and 0.08 0.02 kg TSSm2 for CW6ʹ and CW6, respectively). Nevertheless, these values are clearly lower than those reported in the literature for treatment systems at full scale (Tanner and Sukias, 1995; Tanner et al., 1998; Caselles-Osorio et al., 2007; Chazarenc et al., 2009). Furthermore, from the amount of solids shown in Fig. 5, the accumulation was more heterogeneous in the overloaded wetland. Whereas accumulated solids in CW6 presented the same range regardless of depth (c.a. 1000e1800 mgL1), in CW6ʹ solids were concentrated in the top inlet region (Figs. 2 and 3). Since water was distributed on the surface, the top inlet zone
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 4 4 8 e1 4 5 8
1453
Fig. 3 e Solids distribution map at 32 cm depth for the CWs at the experimental plant. Values are plotted in mgLL1. The arrow indicates the flow direction from inlet to outlet. Note that CW1 and CW5, in which samples were taken at 15 cm only, are not represented.
constituted the first filter for solids and their accumulation was twice the concentration at the bottom. The same pattern would be expected for CW6, although delayed due to the lower solids concentration in the wastewater. Similarly, saturated hydraulic conductivity was affected by organic and solids loading rate. Although there were no significant differences between these wetlands at the end of the study (hydraulic conductivity at the inlet was 28.3 12.9 md1 and 25.2 9.9 md1 for CW6ʹ and CW6, respectively), the loss of permeability over time was greater for CW6ʹ (Fig. 6). Hydraulic conductivity is a parameter affected by the amount of solids and by their density and packing properties, but also by the amount and size of belowground plant material (Pedescoll et al., 2011a). For these two wetlands, the amount of accumulated roots at the end of the study was similar, but slightly higher for CW6 (Table 2), which might explain the lack of differences in hydraulic conductivity at the end of the study.
3.3.
Effect of vegetation
The effect of presence/absence of plants was evaluated by comparing CW3-CW4 (FWS flow with bottom outlet pipe,
planted and unplanted, respectively) and CW6-CW7 (strict SSF, planted and unplanted, respectively), while the effect of plant species by comparing CW1 and CW5 (FWS flow with floating macrophytes, planted with cattail and common reed, respectively). Many authors have reported the importance of the presence of vegetation in constructed wetlands, affecting both wastewater treatment and the development of clogging (Tanner and Sukias, 1995; Kadlec and Wallace, 2009; Chazarenc et al., 2009). The effect of the growth of roots and rhizomes can clearly be observed in the hydraulic conductivity of CW6 and CW7. At the end of the study, hydraulic conductivity was between 3 and 5 times higher in the unplanted wetland than in the planted one (25.16 9.94 md1 and 124.10 54.38 md1 at the inlet and 93.19 61.89 md1 and 341.48 33.15 md1 at the outlet for CW6 and CW7, respectively), which is in accordance with other studies measuring hydraulic conductivity in planted and unplanted wetlands (Sanford et al., 1995; Pedescoll et al., 2011a). No statistical differences between CW3 and CW4 were found (Fig. 2), probably due to the type of flow in these wetlands. Since these systems combined FWS and SSF, a large part of the solids could have settled on the gravel and decreased the
1454
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 4 4 8 e1 4 5 8
Fig. 4 e Comparison of hydraulic conductivity (plotted in green values in mdL1) and solids accumulation (plotted in brown values in mgLL1) patterns for CW3 and CW7. The arrow indicates the flow direction from inlet to outlet. Note that saturated hydraulic conductivity was measured in the first 15 cm of gravel corresponding to the TSS at 32 cm and 15 cm depth for CW3 and CW7 respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Table 1 e Averaged pollutant concentrations in influent and effluent for the whole study period (Standard deviation in brackets). Design configuration Influent CW1 CW2 CW3 CW4 CW5 CW6 CW6’ CW7
Hydroponic T. angustifolia FWS T. angustifolia FWS-bottom outlet T. angustifolia FWS-bottom outlet unplanted SSF P. australis SSF P. australis SSF P. australis Overloaded SSF unplanted
COD mg O2L1
BOD5 mg O2L1
TSS mgL1
TKN mgL1
NHþ 4 eN mgL1
NO 3 eN mgL1
PO3 4 eP mgL1
177.43 (165.00) 49.92 (31.29)
83.57 (60.60) 17.04 (13.96)
86.57 (99.09) 9.90 (9.76)
20.00 (6.32) 11.71 (7.38)
13.57 (3.21) 10.00 (6.84)
0.75 (0.73) 0.33 (0.55)
1.92 (0.49) 1.65 (1.17)
57.32 (40.11) 39.37 (24.23)
17.30 (13.77) 12.74 (10.89)
12.02 (8.76) 12.04 (8.93)
12.84 (7.66) 11.82 (7.54)
10.58 (7.05) 10.30 (7.13)
0.28 (0.24) 0.15 (0.10)
1.74 (1.07) 1.30 (1.18)
57.67 (35.21)
17.49 (11.79)
14.51 (12.19)
17.07 (7.27)
14.70 (6.69)
0.16 (0.12)
2.33 (1.27)
49.27 (31.91) 39.02 (22.56) 56.64 (32.65)
19.97 (17.23) 16.86 (12.04) 29.28 (20.03)
11.44 (10.17) 29.63 (18.94) 16.15 (11.74)
10.67 (8.65) 8.45 (6.83) 17.73 (6.12)
9.01 (8.60) 7.30 (6.95) 16.24 (5.79)
0.19 (0.10) 0.43 (0.28) 0.25 (0.20)
2.11 (1.30) 0.39 (0.67) 1.98 (1.27)
36.75 (27.39)
14.64 (12.77)
10.36 (8.78)
18.18 (3.75)
17.17 (3.44)
0.23 (0.17)
1.90 (1.25)
16 127.0 150.7
44 37 1 127.0 165.0 120.0 1.30 1.53 1.62
225.9 167.0 191.7
13 291.0 258.5
41 37 181.0 192.0 306.0 306.0 14.4 10.1
9.5 13.4 8.3
5.09 7.78
7.33 8.77 5.11
2.83 1.30
55 102.0 225.9 24.5 9.84
2.48
Theoretical effective volume (L) Below ground Above ground Biomass ratio Below-ground plant dry weight (kgm2) Above-ground plant dry weight (kgm2)
CW7
CW5 CW6 CW6’
CW4
CW2 CW3
a Range of solids concentration.
0.046e0.304
0.024e0.183 0.057e0.109 0.029e0.716
0.003e0.322
0.003e0.108 0.041e1.120
0.030e0.326
Hydroponic T. angustifolia FWS T. angustifolia FWS-bottom outlet T. angustifolia FWS-bottom outlet unplanted SSF P. australis SSF P. australis SSF P. australis Overloaded SSF unplanted CW1
Accumulated interstitial solidsa (kgm2)
1455
hydraulic conductivity measured through a vertical core. In FWS wetlands, TSS removal occurs mainly by settling and flocculation processes in the wetland and also by anaerobic decomposition of the settled organic matter (USEPA, 2000). In accordance with the above, the amount of solids was higher in the unplanted wetland gravel bed (in Fig. 6, the gravel region corresponds to the bottom of CW4), whereas in the region of free water, surface solids were more concentrated in the planted one (CW3). The solids distribution pattern for CW3 and CW4 suggests that flow in the unplanted cell occurred mainly along the surface until the water reached the outlet, where it was forced to infiltrate through the gravel. In contrast, it seems that the presence of plants promoted the partial flow of water through the bottom (note that in the gravel medium, solids concentrated at the inlet in CW3, Fig. 5). In addition to the above, we also found differences between the plant species used in this study. Since belowground plant growth entails a substantial decrease in effective volume and hydraulic conductivity (Edwards, 1992), the species chosen for use in any CW should be carefully considered in terms of the volume which will be occupied by the roots and rhizomes, among other characteristics. The amount of below-ground biomass produced after almost 4 years of operation of the experimental plant is shown in Table 2. In soilless wetlands, T. angustifolia presented more than twice the weight as Phragmites australis. Consequently, there was 10% less effective volume in CW1, planted with cattail. The values for above-ground biomass are in accordance with those reported by Vymazal and Kro¨pfelova´ (2005) for the usual amount of biomass measured in horizontal CWs planted with P. australis (788e5070 g DMm2). On average, the belowground/above-ground ratio was lower in wetlands planted with common reed (2.20 0.80 and 1.48 0.16 for T. angustifolia and P. australis, respectively), with similar values to those reported by Ouellet-Plamondon et al. (2004). Therefore, cattail seems to have a greater effect on the clogging of subsurface flow wetlands. Root density in the overloaded system (CW6ʹ) was lower than in the not overloaded one (CW6) due to much harsher conditions in the first system (lower redox and higher nutrient concentration).
3.4.
Design configuration
Table 2 e Accumulated solids and plant biomass in the CWs and effective volume loss after 44 months of experimental plant operation.
Actual effective volume (L)
Effective volume loss (%)
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 4 4 8 e1 4 5 8
Effect of flow type
The effect of flow type can be evaluated at two levels. Comparison between the CW4 and CW7 tanks enabled us to determine the differences between two unplanted systems, a FWS wetland with the outlet located at the bottom of the cell (CW4) and a strict SSF wetland (CW7). A comparison of the two FWS wetlands with different outlets, a top inlet and top outlet position in tank CW2 vs. FWS with top inlet and bottom outlet in tank CW3 was also conducted. As discussed in the previous section, in CW4 (FWS flow with water leaving at the bottom) the wastewater flowed mainly along the surface, which offers less resistance, and finally infiltrated near the outlet to reach the drainage pipe located at the bottom of the cell instead of using the whole effective volume of the gravel. Consequently, hydraulic efficiency was lower in CW4 than in CW7, in which the water was forced to cross the wetland through the subsurface. Peak tracer concentration time was clearly lower in CW4 (Table 3),
1456
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 4 4 8 e1 4 5 8
Fig. 5 e Amount of solids (TSS) and percentage of volatile solids (VSS/TSS) in the CWs. measured at 15 (surface) and 32 cm depth (bottom).
thus supporting findings for preferential flow on the surface and decreased retention time in the cell. Furthermore, solids accumulation was more homogeneous in CW7 (Figs. 2 and 3) compared to CW4. However, in the unplanted SSF wetland (CW7), the water flowed diagonally in a direct line from the top inlet to the bottom outlet, as has been observed by others (Suliman et al., 2006; Pedescoll et al., 2011b). Since the solids settled on the surface of the gravel in CW4, hydraulic conductivity decreased drastically compared with CW7 (0.33 1.53 md1 and 124.10 54.38 md1 at the inlet and 0.95 0.61 md1 and 341.48 33.15 md1 at the outlet for CW4 and CW7, respectively). Therefore, water only infiltrated the gravel at the outlet in CW4, thus losing half of the effective volume. With regard to the comparison of the FWS tanks, solids accumulation was higher in CW3 than in CW2 (0.056 0.031 Kgm2 and 0.303 0.409 Kgm2 on average for CW2 and CW3, respectively), and specifically at the gravel matrix (Fig. 6). Hence, there is more effective use of the cell volume in CW3, and in fact the water remains longer in this wetland (Table 3), further supporting findings for the effect of the inlet-outlet position on hydraulic behaviour. Suliman et al.
(2007) stated that the best strategy to increase hydraulic efficiency and retention time in constructed wetlands was to employ a bottom-inlet/top-outlet position, in order to force the water flow against gravity.
3.5.
Effect of gravel matrix
The effect of the gravel matrix was evaluated by comparing CW1 (soilless FWS planted with cattail) vs. CW2 (FWS flow planted with cattail) and CW5 (soilless FWS planted with common reed) vs. CW6 (strict SSF planted with common reed). The presence of a gravel matrix should increase the hydraulic efficiency of a system compared to a mixed flow in those wetlands with hydroponic conditions (soilless FWS) (Table 3), especially when strict SSF is considered. However, the higher hydraulic efficiency observed in CW6 seems to be due to the presence of dead zones (note that the peak tracer concentration time was clearly higher than for the rest of the systems studied and the mean HRT was almost 70% higher than the theoretical HRT). Conversely, the wastewater tended to plug flow when solids concentration distribution was considered (Figs. 2 and 3). Solids were accumulated in a decreasing gradient from inlet to
Fig. 6 e Temporal changes in hydraulic conductivity at inlet and outlet zones in the CWs. Note that values are plotted as a percentage of the initial one.
65.8 56.1 83.9 94.8 27 47 35 23 104 104 104 104 4.01 8.63 6.12 4.02
0.10 0.14 0.24 0.35
0.50 1.11 1.55 0.67
84.2 15 7.06 104 0.25
0.24
71.1 50.7 62.8 0.42 0.28 0.28 19 19 19 6.04 104 5.50 104 8.62 104 0.62 1.49 0.23
a Dimensionless.
CW5 CW6 CW6’ CW7
CW4
Conclusions
This study evaluated different design characteristics which mainly affect clogging distribution patterns. Under the conditions tested in the eight mesocosm scale CWs, differences were observed depending on the design parameter tested. Organic loading rate is related to the amount of solids accumulated within the wetland. The overloaded SSF CW presented four times more solids, and the distribution pattern was as would be expected when a wetland operates at recommended loads (w6 g BODm2 d1). The presence of plants entails a substantial decrease in the hydraulic conductivity of the gravel medium, with a loss of permeability which is around 20% higher in planted wetlands than in unplanted ones. The plant species chosen may also have implications for clogging: T. angustifolia seems to have a greater effect since its below-ground biomass weighed twice that of P. australis. Hydraulic behaviour, and therefore water flow paths, was greatly influenced by the presence of a gravel matrix and the outlet pipe position. In strict SSF CWs, the water was forced to cross the gravel and tended to flow diagonally from the top inlet to the bottom outlet (where the inlet and outlet pipes were located). However, when one zone of FWS was considered, water spread preferentially on the surface above the gravel, thus loosing half the effective volume of the system. Only the presence of plants seemed to help the water to flow partially within the gravel matrix.
0.19 0.21 0.39 0.49 37.4 71.1 52.8 41.8 53.6 42.2 22.5 34.6
0.41 54.0 61.5
0.65 0.83 0.39 37.4 29.6 37.8 45.7 69.0 68.0
Hydroponic T. angustifolia FWS T. angustifolia FWS-bottom outlet T. angustifolia FWS-bottom outlet unplanted SSF P. australis SSF P. australis SSF P. australis Overloaded SSF unplanted
1457
outlet in CW6, regardless of depth, whereas in CW5, solids were concentrated mainly on the left side of the cell. In contrast, the presence of the gravel in CW2 reduced the HRT and significantly increased tracer dispersion, thus reducing hydraulic efficiency. This system was a FWS wetland, with the wastewater entering and exiting on the surface. Since water flows along the path of least resistance, the gravel in CW2 merely constituted support for the vegetation because the wastewater mainly flowed through the top 25 cm of the wetland. Solids concentration was in the same range both at the top and bottom of the cell, and was lower than in the soilless wetland (CW1), which suggests that solids settled on the surface of the gravel layer, as discussed in the previous section. Differences between treatments were observed in terms of solids characteristics. As shown in Fig. 5, the percentage of volatile solids (and thus organic matter) was around 60% when FWS was considered, whereas the ratio decreased for solids accumulated in the gravel (that is, in CW6, CW6ʹ, CW7 and the bottom of CW2, CW3 and CW4). This difference in the proportion of volatile solids could be explained by the accumulation of litter from plant decay, together with algal growth ´ lvarez and Be´cares, 2006; Hijosain FWS flow wetlands (A Valsero et al., 2010).
4.
CW1 CW2 CW3
Peak tracer concentration time tp (hours) Error 4 (dima) Dispersion number D (dima) Normalised variance sq (dima) Actual HRT s (hours) Nominal HRT sn (hours) Design configuration
Table 3 e Hydraulic parameters obtained from the tracer experiments carried out after 44 months of experimental plant operation.
Hydraulic efficiency l (dima)
Recovery (%)
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 4 4 8 e1 4 5 8
Acknowledgements This study was funded by the Spanish Ministry of Science through the projects CTM2005-06457-C05-03 and CTM2008-
1458
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 4 4 8 e1 4 5 8
06676-C05-03/TECNO. Anna Pedescoll acknowledges the Juan de la Cierva Programme of the Spanish Ministry of Science and Innovation. Marianna Garfı´ is grateful to the Secretary General of Universities, Ministry of Education (Spain) (Programa Nacional de Movilidad de Recursos Humanos del Plan Nacional de I-Dþi 2008-2011). The authors thank Amando Escudero Marina and Lina Tyroller for his help in the field.
references
´ lvarez, J.A., Be´cares, E., 2006. Seasonal decomposition of Typha A latifolia in a free-water surface constructed wetland. Ecological Engineering 28 (2), 99e105. APHA-AWWA-WPCF, 2001. Standard Methods for the Examination of Water and Wastewater (2001), twentieth ed.. Washington DC, USA. Caselles-Osorio, A., Puigagut, J., Segu´, E., Vaello, N., Grane´s, F., Garcı´a, D., Garcı´a, J., 2007. Solids accumulation in six full-scale subsurface flow constructed wetlands. Water Research 41 (6), 1388e1398. Chazarenc, F., Gagnon, V., Comeau, Y., Brisson, J., 2009. Effect of plant and artificial aeration on solids accumulation and biological activities in constructed wetlands. Ecological Engineering 35, 1005e1010. Cooper, P., 2005. The performance of vertical flow constructed wetland systems with special reference to the significance of oxygen transfer and hydraulic loading rates. Water Science and Technology 51 (9), 81e90. Council Directive 91/271/EEC of 21 May 1991 Concerning urban waste water treatment. Official Journal of the European Communities L 135 of 30 May 1991, pp. 40e52. Edwards, G.S., 1992. Root distribution of soft-stem bulrush (Scirpus validus) in a constructed wetland. Ecological Engineering 1 (3), 239e243. Garcı´a, J., Aguirre, P., Barraga´n, J., Mujeriego, R., Matamoros, V., Bayona, J.M., 2005. Effect of key design parameters on the efficiency of horizontal subsurface flow constructed wetlands: long-term performance pilot study. Ecological Engineering 25 (4), 405e418. ´ lvarez, E., Sierra, J.P., Garcı´a, J., Chiva, J., Aguirre, P., A Mujeriego, R., 2004. Hydraulic behaviour of horizontal subsurface flow constructed wetlands with different aspect ratio and granular medium size. Ecological Engineering 23 (3), 177e187. Griffin, P., Wilson, L., Cooper, D., 2008. Changes in the use, operation and design of sub-surface flow constructed wetlands in a major UK water utility. In: 11th International Conference on Wetland Systems for Water Pollution Control. Indore, India, pp. 419e426. Hijosa-Valsero, M., Sidrach-Cardona, R., Be´cares, E., 2012. Comparison of interannual removal variation of various constructed wetland types. Science of the Total Environment 430, 174e183. Hijosa-Valsero, M., Sidrach-Cardona, R., Martı´n-Villacorta, J., Be´cares, E., 2010. Optimization of performance assessment and design characteristics in constructed wetlands for the removal of organic matter. Chemosphere 81 (5), 651e657. Jenkins, G.A., Greenway, M., 2005. The hydraulic efficiency of fringing versus banded vegetation in constructed wetlands. Ecological Engineering 25, 61e72. Kadlec, R.H., Wallace, S.D., 2009. Treatment Wetlands, second ed. CRC Press, Boca Raton, FL.
Knowles, P., Dotro, G., Nivala, J., Garcı´a, J., 2011. Clogging in subsurface-flow treatment wetlands: ocurrence and contributing factors. Ecological Engineering 37 (2), 99e112. Knowles, P.R., Griffin, P., Davies, P., 2010. Complementary methods to investigate the development of clogging within a horizontal sub-surface flow tertiary treatment wetland. Water Research 44 (1), 320e330. Levenspiel, O., 1999. Chemical Reaction Engineering, third ed. John Wiley and Sons, Inc, Hoboken, USA, p. 668. Ouellet-Plamondon, C.M., Brisson, J., Comeau, Y., 2004. Effect of macrophyte species on subsurface flow wetland performance in cold climate. In: Proceedings of the 2004 Self-sustaining Solutions for Streams, Wetlands and Watersheds Conference. St. Paul, MN; September 12e15, 2004. pp. 8e15. ´ lvarez, E., Garcı´a, J., Puigagut, J., 2011a. Pedescoll, A., Corzo, A., A The effect of primary treatment and flow regime on clogging development in horizontal subsurface flow constructed wetlands: an experimental evaluation. Water Research 45 (12), 3579e3589. Pedescoll, A., Knowles, P.R., Davis, P., Garcı´a, J., Puigagut, J., 2011b. A Comparison of in situ constant and falling head permeameter tests to assess the distribution of clogging within horizontal subsurface flow constructed wetlands. Water, Air and Soil Pollution. http://dx.doi.org/10.1007/s11270011-1021-4. Persson, J., 2000. The hydraulic performance of ponds of various layouts. Urban Water 2 (3), 243e250. Persson, J., Somes, N.L.G., Wong, T.H.F., 1999. Hydraulics efficiency of constructed wetlands and ponds. Water Science and Technology 40 (3), 291e300. Rousseau, D.P.L., Horton, D., Griffin, P., Vanrollenghem, P.A., De Paw, N., 2005. Impact of operational maintenance on the asset life of storm reed beds. Water Science and Technology 51 (9), 243e250. Sanford, W.E., Steenhuis, T.S., Parlange, J.Y., Surface, J.M., Peverly, J.H., 1995. Hydraulic conductivity of gravel sand as substrates in rock-reed filters. Ecological Engineering 4 (4), 321e336. Suliman, F., Futsaether, C., Oxaal, U., 2007. Hydraulic performance of horizontal subsurface flow constructed wetlands for different strategies of filling the filter medium into the filter basin. Ecological Engineering 29 (1), 45e55. Suliman, F., Futsaether, C., Oxaal, U., Haugen, L.E., Jenssen, P., 2006. Effect of the inleteoutlet positions on the hydraulic performance of horizontal subsurface-flow wetlands constructed with heterogeneous porous media. Journal of Contaminant Hydrology 87 (1e2), 22e36. Tanner, C.C., Sukias, J.P.S., Upsdell, M.P., 1998. Organic matter accumulation and maturation of gravel-bed constructed wetlands treating dairy farm wastewaters. Water Research 32 (10), 3046e3054. Tanner, C.C., Sukias, J.P., 1995. Accumulation of organic solids in gravel bed constructed wetlands. Water Science and Technology 32 (3), 229e239. U.S. EPA, 2000. Constructed Wetlands Treatment of Municipal Wastewaters. EPA 625/R-99/010, U.S.. EPA Office of Research and Development, Washington D.C. Vymazal, J., Kro¨pfelova´, L., 2005. Growth of Phragmites australis and Phalaris arundinacea in constructed wetlands for wastewater treatment in the Czech Republic. Ecological Engineering 25 (5), 606e621. Wallace, S.D., Knight, R.L., 2006. Small-scale Constructed Wetland Treatment Systems: Feasibility, Design Criteria and O&M Requirements. Final Report, Project 01-CTS-5. Water Environment Research Foundation (WERF), Alexandria, USA.