This study aimed at determining the treatment performances of a full-scale vertical flow constructed wetlands designed to treat wastewater from a food-processing industry (cookie factory), and to study the influence of the organic loading rate. The full-scale treatment plant was designed with a first vertical stage of 630 m², a second vertical stage of 473 m² equipped with a recirculation system and followed by a final horizontal stage of 440 m². The plant was commissioned in 2011, and was operated at different loading rates during 16 months for the purpose of this study. Treatment performances were determined by 24 hour composite samples. The mean concentration of the raw effluent was 8,548 mg.L−1 chemical oxygen demand (COD), 4,334 mg.L−1 biochemical oxygen demand (BOD5), and 2,069 mg.L−1 suspended solids (SS). Despite low nutrients content with a BOD5/N/P ratio of 100/1.8/0.5, lower than optimum for biological degradation (known as 100/5/1), mean removal performances were very high with 98% for COD, 99% for BOD5 and SS for the two vertical stages. The increasing of the organic load from 50 g.m−2.d−1 COD to 237 g.m−2.d−1 COD (on the first stage) did not affect removal performances. The mean quality of effluent reached French standards (COD < 125 mg.L−1, BOD5 < 25 mg.L−1, SS < 35 mg.L−1).

INTRODUCTION

In agro-food industries, the consumption of large amounts of water and the production of highly concentrated effluent in organic biodegradable matter represent an important issue. Effluents from the agro-food industry are generally deficient in nutrients, which can be problematic for treatment.

Vertical flow constructed wetlands (VFCWs) are now one of the most popular technologies in France for the treatment of raw domestic wastewater for small communities (Paing et al. 2015), but there are still very few applications for industrial wastewater. Furthermore, several studies have shown that VFCW could be efficient for high organic content wastewater. In agricultural areas, some examples can be cited for the treatment of swine effluent (Kato et al. 2013) or milking parlor washing effluents (Liénard et al. 2003). For the agro-food industry, VFCWs were already applied for small slaughterhouse effluent (Weedon 2004; Soroko 2007), dairy and cheese factory (Comino et al. 2011; Sharma et al. 2011; Kato et al. 2013) and winery effluent (Masi et al. 2007; Serrano et al. 2011; Masi et al. 2015). The highest organic content (mean chemical oxygen demand (COD) of 14 120 mg.L−1) was reported for olive mill wastewater by Herouvim et al. (2011). To our knowledge, no references were found concerning the use of constructed wetlands to treat effluent from a cookie factory.

All of these studies presented VFCWs with one or several stages, most of the time in association with horizontal flow constructed wetlands. In contrast to the French system, which treats raw wastewater directly (Molle et al. 2005; Paing et al. 2015), most of the cited studies showed CWs combined with pretreatment including septic or storage tanks (Liénard et al. 2003; Masi et al. 2007; Soroko 2007; Kato et al. 2013), trickling filter (Herouvim et al. 2011) and anaerobic digester (Serrano et al. 2011).

In that particular context, the aim of our work was to evaluate the treatment performance of the French vertical flow system for agro-food industry effluent and to define the optimal organic loading rate (OLR). Our study was performed in a full-scale system consisting of two vertical stages followed by a horizontal one, and fed with effluent from a cookie factory at different OLRs during 2 years. Based on our results, and on the assessment of sludge accumulation and energy consumption, the economical and environmental value of this process has been evaluated.

MATERIAL AND METHODS

The CWs system at Contres

The cookie factory of Saint Michel at Contres (47°43′N; 1 °44′E) is testing new recipes of industrial pastries. The effluents for this factory are mainly made of leftovers from the production line (including butter, sugar, flour, eggs, chocolate, almond powder, flavors, and non-hydrogenated vegetable oils), alkaline detergent and acid products for disinfection.

All the effluents from the cookie factory are treated in a CW system built in 2011 by the Jean Voisin company. The CW system is made of three stages, a first stage of vertical flow accounting for 630 m² and divided into four beds, a second stage of vertical flow with a total surface of 473 m², divided into three beds, and a horizontal flow with a total surface of 440 m² divided into two beds. All beds in the three stages are fed in alternance. The vertical CWs of the first stage are filled with Mayennite®, an expanded schist (see Prigent et al. 2013a). The system is equipped with recirculation, enabling it to pump back effluent from the second stage to the first stage fixed at 100% during the study. The depth of both stages was 75 cm for the first stage, which was mainly filled with Mayennite® and for the second stage, which was mainly filled with sand (Table 1). Regarding the pipe system for the feeding: the initial system was as for the first stage, eight feeding points inox 76 mm/bed; and for the second stage, 20 diffusion points in a PVC pipe 50 mm/bed. After bed surface reduction and wall implementation: first stage, four feeding points inox 76 mm/bed; and second stage, 10 diffusion points in a PVC pipe 50 mm/bed.

Table 1

Description of the layers of the VF beds

First stage 
15 cm Mayennite® 2/4 mm – D10 = 2 mm, UC = 1.8 
25 cm gravel 2/6 mm – D10 = 3 mm, UC = 1.5 
15 cm gravel 10/20 mm – D10 = 12 mm, UC = 1.3 
20 cm gravel 20/50 mm – D10 = 24 mm, UC = 1.5 
Second stage 
40 cm sand 0/4 mm – D10 = 0.3, UC = 4 
15 cm gravel 4/10 mm – D10 = 5, UC = 2 
20 cm gravel 10/20 mm – D10 = 12 mm, UC = 1.3 
First stage 
15 cm Mayennite® 2/4 mm – D10 = 2 mm, UC = 1.8 
25 cm gravel 2/6 mm – D10 = 3 mm, UC = 1.5 
15 cm gravel 10/20 mm – D10 = 12 mm, UC = 1.3 
20 cm gravel 20/50 mm – D10 = 24 mm, UC = 1.5 
Second stage 
40 cm sand 0/4 mm – D10 = 0.3, UC = 4 
15 cm gravel 4/10 mm – D10 = 5, UC = 2 
20 cm gravel 10/20 mm – D10 = 12 mm, UC = 1.3 

In September 2013, all the filters were equipped with walls to reduce the surface fed with effluent to test the effect of an increased organic loading on treatment performances. The surface of the first stage was reduced to 157 m² divided into two beds; the second stage was reduced to 157 m² divided into two beds, and the last horizontal stage was reduced to 220 m² in one bed.

Effluent sampling and analysis

A 24 hour composite sample was first taken in November 2012 (18 months after commissioning), then nine 24 hour composite samples were taken between April and June 2013. After the surface reduction, 10 composite 24 hour samples were taken from December 2013 up to July 2014.

Samples were taken proportionally to the flow at four points: inlet, outlet of the first stage, outlet of the second stage, and outlet of the CWs system. Parameters including pH, COD, biochemical oxygen demand (BOD5), and suspended solids (SS), were analyzed on all samplings according to micro-methods (Merck kit for COD and OxiTop flask using nitrification inhibitors, Bioblock Scientific, Rungis, France). Samples were regularly sent to an independent certified laboratory to control the results with an error always smaller than 15% on the results. Total nitrogen (TN), total Kjeldahl nitrogen (TKN) and total phosphorus (TP) were analyzed once every 3 months by an independent certified laboratory according to AFNOR (2005). The hexane extractable substances (HES) have been analyzed in one study according to AFNOR (2005).

Core sampling and analysis

Sampling

Cores were sampled (July 2014) within the first stage to estimate the amount of accumulated matter on top of the filter, and the effect of sludge accumulation in relation to the OLR. Cores were sampled at 1 m distance from the inlet point in the fed bed, and in the partitioned bed (not fed since the surface reduction in September 2013) and at 0.5, 1 and 2 m distances in the bed in the others. Cores were sampled using a PVC pipe (ø 7.5 cm) that was pushed inside the filtration matrix at up to 40 cm depth. In the core, three layers were sampled: the surface layer corresponding to the accumulated sludge, the first layer of Mayennite® (Ø2–4 mm) down to 15 cm, and finally a layer of gravel (Ø2–6 mm) down to 25 cm. Cores were put in plastic bags and stored at 4°C before analysis.

Analyses

Core samples were first washed using a buffer solution (pH = 7.2; K2HPO4 9.3 g/L; KH2PO4 1.8 g/L as in Prigent et al. (2013a)) and then screened on a 0.5 mm filter to remove media and pieces of rhizome. The solution was then dried at 105 °C and finally, volatile matter was measured by loss after incineration at 550 °C, after removing the mass of P-salt employed in the buffer solution.

RESULTS AND DISCUSSION

Treatment performances

Average concentrations, together with treatment performances (based on mass balance), are presented in Table 2.

Table 2

Average concentrations and treatment performances observed at the CWs system of Contres between November 2012 and July 2014

    Inlet VF1 VF2 HF Treatment performances Treatment performances 
Parameters Units mg.L−1    VF1 + VF2 % CWs syst. % 
COD Mean ± SD 8,548 ± 2,524 1,186 ± 937 107 ± 46 41 ± 21 98.4 ± 1.2% 99.2 ± 0.8% 
20 20 20 18 20 18 
BOD5 Mean ± SD 4,334 ± 1,388 634 ± 566 17 ± 18 5 ± 5 99.2 ± 1.3% 99.8 ± 0.4% 
20 20 19 17 19 17 
SS Mean ± SD 2,069 ± 1,463 149 ± 130 13 ± 13 12 ± 11 99.2 ± 0.9% 99.2 ± 1.3% 
20 19 19 18 19 18 
TKN Mean ± SD 75 ± 17 6.6–12.9 2.2–2.5 2 ± 1.2 96.5% 97.6 ± 2.6% 
TN Mean ± SD 78 ± 21 7.2–13.5 5.1–13.6 2.2 ± 0.1 70–97.1% 97.5 ± 2.6% 
TP Mean ± SD 21 ± 5 2.8–4.1 2.9–3.1 1.5 ± 0.5 80.0% 92.5 ± 7.2% 
    Inlet VF1 VF2 HF Treatment performances Treatment performances 
Parameters Units mg.L−1    VF1 + VF2 % CWs syst. % 
COD Mean ± SD 8,548 ± 2,524 1,186 ± 937 107 ± 46 41 ± 21 98.4 ± 1.2% 99.2 ± 0.8% 
20 20 20 18 20 18 
BOD5 Mean ± SD 4,334 ± 1,388 634 ± 566 17 ± 18 5 ± 5 99.2 ± 1.3% 99.8 ± 0.4% 
20 20 19 17 19 17 
SS Mean ± SD 2,069 ± 1,463 149 ± 130 13 ± 13 12 ± 11 99.2 ± 0.9% 99.2 ± 1.3% 
20 19 19 18 19 18 
TKN Mean ± SD 75 ± 17 6.6–12.9 2.2–2.5 2 ± 1.2 96.5% 97.6 ± 2.6% 
TN Mean ± SD 78 ± 21 7.2–13.5 5.1–13.6 2.2 ± 0.1 70–97.1% 97.5 ± 2.6% 
TP Mean ± SD 21 ± 5 2.8–4.1 2.9–3.1 1.5 ± 0.5 80.0% 92.5 ± 7.2% 

As expected for industrial effluent, large fluctuations of concentrations were observed in COD (8,548 ± 2,524 mg.L−1), BOD5 (4,334 ± 1,388 mg.L−1) and SS (2,069 ± 1,463 mg.L−1), and this was linked to variations in production.

The efficiency was very good after the two vertical flow stage with 98.3%, 99.2% and 99.2% for COD, BOD5 and SS, respectively, enabling achievement of the threshold imposed for the cookie factory of Contres (COD < 300 mg.L−1, BOD5 < 100 mg.L−1, SS < 100 mg.L−1) and an even more stringent discard limit generally imposed for larger industrial sites (COD < 125 mg.L−1, BOD5 < 25 mg.L−1, SS < 35 mg.L−1). Beyond having a polishing effect, we hypothesized that the third horizontal stage also enabled the remove of residual COD, which is generally slowly biodegradable, due to its large residence time. The overall performance of the CWs system enabled very good efficiency of 99.2%, 99.8% and 99.2% for COD, BOD5 and SS, respectively.

The ratio of BOD5/N/P was estimated on average at 100/1.8/0.5 respectively, in the raw effluent, which is quite low in nutrients compared to the ratio of 100/5/1 generally recommended for intensive biological processes. Despite the lack of nutrients, good removal efficiencies have been observed, which may be due to the effect of larger or more complex ecosystems found in CWs, with increased interactions between plans, microbes and microorganisms.

The performances observed for the first two VFs, greater than 98% for COD, BOD5 and SS, were superior to those reported in many other studies on industrial effluents (Soroko 2007; Sharma et al. 2011; Kato et al. 2013). These authors measured, respectively, COD removals of 87, 81 and 97%, on the first two vertical stages, for a respective OLR of 119, 123 and 48 g COD.m−2.d−1 in the first stage. The observed performances are also better than those reported by Paing et al. (2015) based on 169 full-scale French constructed wetlands (mean COD removal efficiencies of 93% for an OLR ranging from 5 to 100 g COD.m−2.d−1).

These very good performances can be explained by several factors, including the use of expanded schist as filling media in the first stage (Prigent et al. 2013a) favoring water retention and microorganism development, the influence of effluent recirculation (Prigent et al. 2013b), enabling a partial dilution of the effluent, and finally, the concentration of inlet effluent, as it has been observed that the performance was proportional to the inlet concentration in other studies (Paing et al. 2006). Nutrients are also well treated, as presented in Table 2, with mean removal efficiencies for the CW systems of 97.6%, 97.5% and 92.5%, respectively, on TKN, TN and TP. Those performances are also higher than those reported for domestic effluent (Paing et al. 2015), which can most likely be explained by the lack of these elements leading to an overconsumption by plants and microorganisms. It is also worth mentioning that the third stage enables improved TN and TP removal efficiencies.

Finally, the removal efficiencies observed on fat matter were also very good, and reached up to 91.7% removal, measured in March 2014, even with an influent with high concentration (808 mg.L−1 fat matter); it is worth mentioning that fat matter accounts for about one-third of the total COD at Contres.

Influence of organic loading rate

The following mean loading rates for COD, BOD5 and SS were applied in the first stage during the first monitoring period: respectively, 50, 25 and 13 g.m−2.d−1 and then 237, 130 and 40 during the second monitoring period after the wall was implemented, which corresponds to a reduction of four times the surface in this stage.

The change of loading rate did not significantly impact on the removal efficiency, as presented in Figure 1 for COD and SS. A similar trend was observed for BOD5 (results not shown). An increase in the removal rate has even been observed in the first stage for, respectively, COD, BOD5 and SS, going from 73%, 72% and 86% under the low-OLR period up to 84%, 85% and 90% during the high-OLR period (Figure 1). This increase in performance is most likely due to the accumulation of organic matter on top of the filter, which increases the filtration, and increases the contact time between effluent and depolluting organisms (Chazarenc & Merlin 2005; Molle 2014). The slight decrease of performance observed in stage two is a consequence of the better efficiency of the first stage, which reduced the pollution load to be treated.

Figure 1

Comparison of removal performance of the first and second vertical stage (VF1 and VF2) during both monitoring periods: low organic loading rate and high organic loading rate.

Figure 1

Comparison of removal performance of the first and second vertical stage (VF1 and VF2) during both monitoring periods: low organic loading rate and high organic loading rate.

The OLRs applied in the two stages VF1 + VF2 versus the amount treated are presented in Figure 2. It can be seen that the maximum loading rates can be much larger, up to five times that generally recommended for domestic wastewater (100 g COD.m−2.d−1 equiv. to 1.2 m²/PE first stage; Molle et al. 2005 state that 60 g COD.m−2.d−1 is the overall loading rate in two stages with a surface of 2 m²/PE). In our study, the overall removal efficiency remained very high whatever the OLR applied. The average organic load applied during the high period on the first stage (237 gCOD.m−2.d−1) is larger than most of the values found in the literature for agro-food effluent treatment: 56 gCOD.m−2.d−1 for winery effluent (Masi et al. 2007), 48 and 87 gCOD.m−2.d−1 for slaughterhouse effluent (Weedon 2004; Soroko 2007), 120 and 136 g COD.m−2.d−1 for a dairy and cheese factory (Sharma et al. 2011; Kato et al. 2013). Nevertheless, some authors also reported a higher organic load: between 43 and 466 g COD.m−2.d−1 for winery effluent by Serrano et al. (2011), between 88 and 6,589 g COD.m−2.d−1 for olive mill effluent by Herouvim et al. (2011). Conversely, the French system was shown to treat some very high organic loading during a limited period of time, for example in a French campsite with a high seasonal tourist population (up to 200 g COD.m−2.d−1 reported by Boutin et al. (2010)).

Figure 2

Relationship between applied and treated organic loading rates on the two vertical stages (VF1 + VF2).

Figure 2

Relationship between applied and treated organic loading rates on the two vertical stages (VF1 + VF2).

Accumulated solids

Although accumulation matter is directly connected to clogging processes, and the deposit layer on the surface is known to be a key factor in the performance of vertical systems (Molle 2014), few studies have investigated this phenomena. In our study, the mean solids (dry matter (DM)) accumulation rate was found to be 4.2 kg DM.m−2 after 3 years (28 months at low OLR and 8 months at high OLR). The mean sludge height was measured at 1.2 cm with a content of DM of 350 kg/m3. This accumulation is lower than those observed in the literature. Chazarenc & Merlin (2005) measured an accumulation matter ranging from 20 kg DM.m−2 (3 and 4 years operating) to 80 kg DM.m−2 (8 years) on vertical French filters for domestic wastewater. Prigent et al. (2013a) studied a compact vertical flow constructed wetland and estimated that accumulation matter was 5 ± 2 kg DM.m−2 after 12 months of operation. The low accumulation rate measured in Contres could be explained by the low OLR applied during the first 2 years of operation. Indeed, there was no deposit on the surface at the end of this period, as proved by the core on the ‘partitioned bed’, not fed since September 2013.

Figure 3 showed that the accumulation of DM was more important near the sprinkler, up to a distance of 1 m. It appeared that accumulation matter on the surface decreased with the distance of the sprinkler while, within the substratum, the distribution of DM is most heterogeneous. Most of the DM (57%) accumulated on the surface and in the first 15 cm of filtering material (Mayennite® Ø 2–4 mm), as shown by Prigent et al. (2013a) and Chazarenc & Merlin (2005).

Figure 3

Solids accumulation in the top layers of the first stage vertical flow system at different distances from the feeding point.

Figure 3

Solids accumulation in the top layers of the first stage vertical flow system at different distances from the feeding point.

In our study, the ratio of sludge accumulated on the surface represents only 0.05 kg DM/kg BOD5 treated. This result is considerably lower than those obtained with an intensive system with, generally, a production of sludge of 0.8 kgDM/kg BOD5 treated. This difference could be explained by the mineralization process, which take place on the surface of constructed wetlands, but also by the accumulated DM within the wetland. This mineralization process is clearly visible with the comparison between the ratio of organic matter measured in the influent (93% DM) and the one measured in the sludge sampling on the surface (31% DM). Moreover, we can mention the absence of overproduction of microorganisms, unlike in the conventional activated sludge process, which are favored by the resting periods operated in vertical filters.

Table 3

Estimation of the mass of DM accumulated in the first stage VF1

  Distance of the sprinkler
 
 
 Units <0.5 m 0.5 m–1.0 m 1.0 m–2.0 m >2.0 m Total 
Surface m² 5.7 18.6 72.7 61.0 158 
Surface rate 3.6% 11.8% 46.0% 38.6% 100% 
Mass of DM per m² kg DM.m² 4.2 4.6 4.2 3.4 – 
Mass of DM of sludge kg DM 23.7 85.8 305.7 205.3 620.5 
  Distance of the sprinkler
 
 
 Units <0.5 m 0.5 m–1.0 m 1.0 m–2.0 m >2.0 m Total 
Surface m² 5.7 18.6 72.7 61.0 158 
Surface rate 3.6% 11.8% 46.0% 38.6% 100% 
Mass of DM per m² kg DM.m² 4.2 4.6 4.2 3.4 – 
Mass of DM of sludge kg DM 23.7 85.8 305.7 205.3 620.5 
Table 4

Estimation of the mass of TSS eliminated from the influent by the first stage VF1

Period Duration d Mean flow m3.d−1 Mean TSS concentration mg.L−1 TSS treatment performance TSS eliminated by VF1 tons 
Low OLR 850 0.8 2.596 86.3% 1.52 
High OLR 260 4.1 1.717 89.7% 1.64 
Total     3.16 
Period Duration d Mean flow m3.d−1 Mean TSS concentration mg.L−1 TSS treatment performance TSS eliminated by VF1 tons 
Low OLR 850 0.8 2.596 86.3% 1.52 
High OLR 260 4.1 1.717 89.7% 1.64 
Total     3.16 

The total of total suspended solids (TSS) eliminated from the influent by the first stage VF1 was estimated at about 3.16 tons after 37 months operation. The comparison with the total DM accumulated shows that the mineralization was very important (Table 3 and 4). Indeed, only 0.62 ton of dry accumulated matter was measured within the filter layer and on the surface, which represents only 20% of the TSS eliminated. Finally, considering the short operation time of less than 1 year with the high organic load, more results are needed to estimate the potential clogging, and to define accurate design parameters.

Energy consumption

The electrical energy input was estimated by considering the consumption of the different devices of the wetland plant (principally the pumps used for feeding the two vertical stages). Energy requirements were estimated to 0.14 kWh.m−3. This consumption is lower than the energy required for an activated sludge process, which is about 1 kWh.m−3. Moreover, the specific energy consumption was estimated to be 0.04 kWh/kg BOD5 treated. By comparison, electrical usage for a conventional intensive treatment plant can reach 2 kWh/kg BOD5 treated.

CONCLUSIONS

The monitoring of the full-scale treatment plant of the Contres cookie factory showed very good treatment efficiencies with, respectively, 99.2%, 99.8% and 99.2% for COD, BOD5 and SS.

Despite high concentrations of the industrial influent in mg.L−1 (8,548 COD, 4,334 BOD5, and 2,069 SS), the mean quality of effluent reached French standards in mg.L−1 (COD < 125, BOD5 < 25, SS < 35) at the outlet of the second vertical stage. The increasing of organic load from 50 g.m−2.d−1 COD to 237 g.m−2.d−1 COD (in the first stage) did not affect removal performance. The efficiencies of biological treatment were not affected by the low nutrients content with a BOD5/N/P ratio of 100/1.8/0.5, lower than the known optimum for biological degradation. The accumulation of DM on the surface and in the two first layers of filtering material presented a positive effect on the performance removal.

The sludge accumulation on the surface of the first stage and the energy consumption were much lower compared to activated sludge processes (0.05 versus 0.8 kg DM/kg BOD5 treated, and 0.04 versus 2 kWh/kg BOD5 treated), so that the operational cost and environmental impact of such CW systems are advantageous.

ACKNOWLEDGEMENTS

This study was partially supported financially by the French ministry of Industry (PHYTORIA project). We thank the factory of St Michel at Contres who helped us with the monitoring of the treatment plant and accepted the experiments.

REFERENCES

REFERENCES
AFNOR
2005
Receuil Normes et Réglementation Environnement. Qualité de l'eau (Standard Collection and Regulatory Environment. Water Quality), Vols 1 and 2
.
Boutin
C.
Prost-Boucle
M.
2010
Etude des filtres plantés de roseaux dimensionnés pour des campings (Study of vertical flow constructed wetlands designed for campsites). Rapport final ONEMA/CEMAGREF. 69 pp
.
Chazarenc
F.
Merlin
G.
2005
Influence of surface layer on hydrology and biology of gravel bed vertical flow constructed wetlands
.
Water Science and Technology
51
(
9
),
91
97
.
Herouvim
E.
Akratos
C. S.
Tekerlekopoulou
A.
Vayenas
D. V.
2011
Treatment of olive mill wastewater in pilot-scale vertical flow constructed wetlands
.
Ecological Engineering
37
(
6
),
931
939
.
Kato
K.
Inoue
T.
Ietsugu
H.
Koba
T.
Sasaki
H.
Miyaji
N.
Kitagawa
K.
Sharma
P. K.
Nagasawa
T.
2013
Performance of six multi-stage hybrid wetland systems for treating high-content wastewater in the cold climate of Hokkaido, Japan
.
Ecological Engineering
51
,
256
263
.
Liénard
A.
Esser
D.
Houdoy
D.
Sabalçagaray
P.
2003
Conception et performances des filtres plantés de roseaux pour le traitement des eaux de lavage de salles de traite
(Design and performances of vertical flow constructed wetlands for the treatment of washing milking parlour effluents).
Ingénieries-EAT
34
.
Masi
F.
Martinuzzi
N.
Bresciani
R.
Giovannelli
L.
Conte
G.
2007
Tolerance to hydraulic and organic load fluctuations in constructed wetlands
.
Water Science and Technology
56
(
3
),
39
48
.
Masi
F.
Rochereau
J.
Troesch
S.
Ruiz
I.
Soto
M.
2015
Wineries wastewater treatment by constructed wetlands: a review
.
Water Science and Technology
71
(
8
),
1113
1127
.
Molle
P.
Lienard
A.
Boutin
C.
Merlin
G.
Iwema
A.
2005
How to treat raw sewage with constructed wetlands: an overview of the French systems
.
Water Science and Technology
51
(
9
),
11
21
.
Paing
J.
Dugue
L.
Gonzalez
H.
Hendou
M.
2006
Removal performances of vertical flow reed beds for new design guidelines. Proceedings of the tenth International Conference on Wetland Systems for Water Pollution Control, 23–29 September 2006
,
Ministerio de Ambiente, do Ordenamento do Territori e do Desenvolvimento Regional (MAOTDR) and IWA, Lisbon, Portugal
, pp.
651
660
.
Serrano
L.
De la Varga
D.
Ruiz
I.
Soto
M.
2011
Winery wastewater treatment in a hybrid constructed wetland
.
Ecological Engineering
37
(
5
),
744
753
.
Sharma
P. K.
Inoue
T.
Kato
K.
Ietsugu
H.
Tomita
K.
Nagasawa
T.
2011
Potential of hybrid constructed wetland system in treating milking parlor wastewater under cold climatic conditions in northern Hokkaido, Japan
.
Water Practice & Technology
6
(
3
),
doi:10.2166/wpt.2011.052
.
Weedon
C. M.
2004
Treatment of abattoir effluent using a compact vertical flow reed bed system. 9th International Conference on Wetland Systems for Water Pollution Control, 29–30 September, Avignon, France
.