Kinetics of pollutant removal from domestic wastewater in a tropical horizontal subsurface flow constructed wetland system effects of hydraulic loading rate

However, the study shows that domestic wastewater can be treated in horizontal subsurface flow constructed wetland systems to meet even the most stringent Vietnamese standards for discha

Contents lists available atScienceDirect Ecological Engineering

j o u r n a l h o m e p a g e :w w w e l s e v i e r c o m / l o c a t e / e c o l e n g

Kinetics of pollutant removal from domestic wastewater in a tropical horizontal subsurface flow constructed wetland system: Effects of hydraulic loading rate Ngo Thuy Diem Tranga,∗

, Dennis Konnerupb, Hans-Henrik Schierupb, Nguyen Huu Chiema,

Le Anh Tuanc, Hans Brixb

a Department of Environmental Science, College of Environment and Natural Resources, Can Tho University, Campus I, 3/2 Street, Ninh Kieu District, Can Tho City, Vietnam

b Department of Biological Sciences, Aarhus University, Ole Worms Allé 1, DK-8000 Århus C., Denmark

c Department of Environmental and Natural Resources Management, College of Environment and Natural Resources, Can Tho University, Campus I, 3/2 Street, Ninh Kieu District, Can Tho City, Vietnam

a r t i c l e i n f o

Article history:

Received 25 July 2009

Received in revised form 5 November 2009

Accepted 23 November 2009

Keywords:

Ammonium

Background concentration

BOD

COD

First-order kinetics

Nitrogen

Phosphorus

Phragmites vallatoria

a b s t r a c t

The treatment capacity of constructed wetlands is expected to be high in tropical areas because of the warm temperatures and the associated higher rates of microbial activity A pilot scale horizontal subsur-face flow constructed wetland system filled with river sand and planted with Phragmites vallatoria (L.) Veldkamp was set up in the southern part of Vietnam to assess the treatment capacity and the removal rate kinetics under tropical conditions The system received municipal wastewater at four hydraulic loading rates (HLRs) of 31, 62, 104 and 146 mm day−1 Removals of TSS, BOD5and COD were efficient at all HLRs with mean removal rates of 86–95%, 65–83% and 57–84%, respectively Removals of N and P decreased with HLRs and were: NH4-N 0–91%; TN 16–84% and TP 72–99% First-order area-based removal rate constants (k, m year−1) estimated from sampling along the length of the wetland from inlet to outlet at the four HLRs were in the range of 25–95 (BOD5), 22–30 (COD), 31–115 (TSS), 5–24 (TN and TKN) and 41–84 (TP) at background concentrations (C*) of 5, 10, 0, 1.5 and 0 mg L−1, respectively The estimated k-values should not be used for design purposes, as site-specific differences and stochastic variability can

be high However, the study shows that domestic wastewater can be treated in horizontal subsurface flow constructed wetland systems to meet even the most stringent Vietnamese standards for discharge into surface waters

© 2009 Elsevier B.V All rights reserved

1 Introduction

The treatment of domestic wastewater is far from satisfactory

in many subtropical and tropical countries, including Vietnam

As opposed to Europe and North America, where nearly all

wastewater is collected in sewer systems and treated in large

centralized wastewater treatment facilities, in most other parts of

the world, centralized sewer systems and wastewater treatment

plants only exist in larger cities, whereas small and medium-sized

towns generally discharge wastewater untreated into the

environ-ment Hence, for hygienic and environmental reasons, there is an

urgent need to implement appropriate wastewater management

systems in these areas, as for example constructed wetland (CW)

systems, that are low-cost, easy-to-operate, efficient and robust

(Denny, 1997; Haberl, 1999; Kivaisi, 2001) The CWs are currently

∗ Corresponding author Tel.: +84 7103 830 635; fax: +84 7103 730 392.

E-mail address: ntdtrang@ctu.edu.vn (N.T.D Trang).

being studied as a technology that can be used for the treatment

of many types of wastewater in tropical countries, including high-strength wastewater from agricultural activities and municipal wastewater from small communities (Nahlik and Mitsch, 2006; Bojcevska and Tonderski, 2007; Brix et al., 2007a; Konnerup et al., 2009; Kantawanichkul et al., 2009) Treatment of wastewater in

CW systems is based on natural processes, and CWs can tolerate

a high variability in loading rate and wastewater quality Further-more, in addition to treating the wastewater, CW systems planted with ornamental plants can provide a nice-looking environment (Brix et al., 2007a; Zurita et al., 2009)

Horizontal subsurface flow CWs are commonly used for sec-ondary and post-treatment of domestic wastewater in temperate regions, and are efficient in solids and organic matter removal but less efficient in nutrient removal (Masi and Martinuzzi, 2007; El Hamouri et al., 2007; Brix et al., 2007b; Vymazal, 2009) Most performance data available for CW systems are from temperate climates, but the treatment performance is expected to be signifi-cantly higher in tropical areas because of the high temperatures and the associated higher microbial activity The area-specific removal

0925-8574/$ – see front matter © 2009 Elsevier B.V All rights reserved.

capacity of pollutants in CWs is often evaluated using a

first-order reaction kinetics model (Kadlec and Knight, 1996; Kadlec,

2009) Removal rate constants (k) are often estimated based on

inlet–outlet performance data from operational CW systems (e.g

Kadlec, 2009) and may therefore be affected by many factors,

such as wetland configuration, hydraulic loading characteristics,

hydraulic efficiency, and input pollutant concentration Another

approach to estimate removal rate constants that is less affected

by some of these factors is to explore the concentration profiles of

the pollutants from inlet to outlet in the wetland and to use these

profiles to estimate removal rate constants (Kadlec, 2003)

The aim of the present study was to estimate the removal

capacity of horizontal subsurface flow CW systems under

tropi-cal conditions We used an existing experimental CW system at

Can Tho University, Vietnam, that had been loaded with

munici-pal wastewater at a hydraulic loading rate of 31 mm day−1 since

2003 with consistent efficient removals of suspended solids, BOD5,

COD, as well as nitrogen and phosphorus The experimental

sys-tem was equipped with facilities to sample water internal in the

system at different positions along the flow path from inlet to

out-let The system was run at four hydraulic loading rates, and the

first-order area-based removal rate constants (k) were estimated

based on the pollutant concentration profiles within the system

It is to our knowledge the first time that this procedure has been

used in tropical climates to estimate removal kinetics in horizontal

subsurface flow CWs

2 Materials and methods

2.1 Experimental setup

An experimental constructed wetland (CW) system with

hor-izontal subsurface flow (HSF) was built at Can Tho University,

Vietnam (10◦00′N, 105◦46′E), in 2003 (Figs 1 and 2) The CW

received a mixture of domestic sanitary wastewater, grey water

from dormitories and storm water collected in the sewer system

of the campus The CW system contained a 6.4 m3 storage tank

into which the wastewater from the sewer system was pumped,

a 0.96 m3 pretreatment section, a 21.1 m3wetland filter planted

with Phragmites vallatoria (L.) Veldkamp, and finally a 3 m3

efflu-ent section The first part of the pretreatmefflu-ent section was filled

with charcoal placed under a layer of ∅ 40–60 mm gravel for

deodorization and followed by a coconut fiber filter The

wet-land filter (12 m × 1.6 m × 1.1 m; length × width × depth) was filled

with 0.25–0.43 mm (D10–D60) river sand with a saturated hydraulic

conductivity of 9.8 m d−1 Five sampling pipes were placed at the

bottom of the bed at different distances from the inlet for water

sampling (Fig 2) Effluent from the wetland filter passed a

gravel-filled (∅ 10–20 mm) effluent section where the outlet water level

was controlled by a tiltable vertical pipe

Prior to the present experiment, the operation of the system was

stopped for 6 months and the system renovated The sand at the

front 2 m of the wetland filter was washed, and P vallatoria was

subsequently replanted at a density of 25 stems m−2using 60 cm

cut culms from a nearby natural reed bed Then the bed was

sat-urated with wastewater and the plants allowed establishing for a

4-week period prior to continuous wastewater loading

2.2 Wastewater loadings and samplings

From November 2007 to April 2008 four hydraulic loading rates

(HLRs) of 31, 62, 104 and 146 mm day−1were applied to the CW

sys-tem, each for a period of about 40 days starting with the lowest HLR

The loading rates were controlled by filling the inlet storage tank

twice or three times daily with the desired amount of wastewater from the sewer system using a manually controlled pump From the storage tank the wastewater was drained slowly into the system through two valves to achieve even loading over time The outlet level was set just below the bed surface to secure water saturated conditions in the wetland filter

At each HLR, water samples were collected during the last 8 days

of the loading period when performance were assumed to be rep-resentative for the particular HLR Grab samples were taken in the morning every second day at the inlet (WL1), after the pretreat-ment unit (WL2), at the outlet (WL8), and at different intermediate positions in the wetland filter (WL3 to WL7) corresponding to dis-tances from the inlet of 1.9, 3.8, 5.9, 7.9 and 9.9 m using a perforated plastic pipe to withdraw water from the sand bed (Fig 2) All sam-ples were collected and stored at 4◦C in 2 L plastic bottles until analysis in the laboratory

2.3 Water quality analyses The pH, dissolved oxygen (DO), temperature and electric con-ductivity (EC) of the water samples were analyzed on-site using portable pH, DO and EC meters (HI 8424 and HI 9146, Hanna Instru-ments, Romania; Orion 011510, Thermo Electron Corporation, USA) Total suspended solids (TSS) were analyzed by the filtration method using Whatman GF/F filters Ammonium nitrogen (NH4 -N) and orthophosphate (PO4-P) were analyzed in filtered samples using the salicylate method (modified from Lachat, Quikchem No 10-107-06-3-A or B), and standard colorimetric methods (APHA, 1998), respectively Total phosphorus (TP) was analyzed by the ascorbic acid method after acid hydrolysis in an autoclave for

30 min (APHA, 1998) Total Kjeldahl nitrogen (TKN) was analyzed at HLRs of 104 and 146 mm day−1using a Kjeldahl block digestion unit (Kjeldatherm KB 20S, Gerhardt, Germany) and a semi-automatic steam distillation unit (Vapodest 20, Gerhardt, Germany), and total nitrogen (TN) was analyzed at HLRs of 31 and 62 mm day−1

using addition of Devarda’s alloy during distillation For technical reasons, we could not analyze nitrate nitrogen (NO3−), but later analyses of the wastewater in the sewer system and the outlet of the bed indicate that the concentrations were low (<2 mg L−1at

a HLR of 31 mm day−1 and <4 mg L−1 at a HLR of 62 mm day−1) Chemical oxygen demand (COD) was analyzed using a COD reac-tor at 150◦C for 2 h (HACH Instruments, USA) and titration of unreduced K2Cr2O7 with ferrous ammonium sulfate Biochemi-cal oxygen demand (BOD5) was analyzed by the OxyTop method (TS606-G/2-i, Taiwan)

2.4 First-order modeling First-order area-based removal rate constants (k), assuming exponential removal to non-zero background wetland concentra-tions (C*), were estimated for TSS, BOD5, COD, TN (or TKN) and TP Fitted values of k were derived from the following plug-flow k–C* model (Kadlec and Knight, 1996):

lnCo−C∗

Ci−C∗ = −

ky q where k is the first-order area-based removal rate constant (m year−1), Cois the outlet concentration (mg L−1), Ciis the inlet concentration (mg L−1), C* is the irreducible background wetland concentration (mg L−1), q is the hydraulic loading rate (m year−1) and y is the fractional distance from inlet to outlet The equation was rearranged to carry out the fitting procedure:

C =C ∗ +(C −C∗)exp−ky/q

Fig 1 Schematic profile diagram of the pilot scale horizontal subsurface flow constructed wetland.

Fittings were carried out using the non-linear regression

pro-cedure of the software Statgraphics Centurion XV (StatPoint, Inc.,

USA) with Ci and k as unknown parameters Based on the

low-est BOD5 concentrations (4.5 mg L−1) in the data set obtained in

this study, a constant C* of 5 mg L−1 for BOD5 was used in the

model This C* value for BOD5is also suggested byKadlec (2009)

for HSF wetlands with inlet BOD5 concentrations in the range

of 30–100 mg L−1 Typical C* background concentrations of 10, 0,

1.5, and 0 mg L−1for COD, TSS, TN (or TKN) and TP, respectively,

were assumed and applied in the model because the outlet

con-centrations in the present study were far from the expected C*

concentrations (Kadlec and Knight, 1996)

2.5 Statistical analysis

All variables were tested for normal distribution and variance

homogeneity (Levene’s test) and logarithmically transformed if

necessary using Statgraphics Centurion XV (StatPoint, Inc., USA)

Comparisons of inlet and outlet concentrations and removal

effi-ciencies between the four HLRs were performed using one-way analysis of variance (ANOVA) Two-way ANOVA using Type III sum

of squares was used to analyze the effect of HLR and distances (positions along the CW system) on water quality Post hoc com-parisons using Tukey’s Honestly Significant Differences (HSD) were used to identify significant differences between means at the 5% probability level

3 Results 3.1 Inlet and outlet water quality The influent wastewater had concentration levels of nutri-ents corresponding to typical municipal wastewater (Bayley et al., 2003), but was more dilute concerning TSS, BOD and COD Dissolved oxygen concentrations were low in the inlet and all samples taken within the system (<1.6 mg L−1) and slightly higher

in the outlet (1.7–2.9 mg L−1), particularly at the highest HLR of

146 mm day−1, and the water temperature varied between 26◦C at

a HLR of 62 mm day−1and 29◦C at a HLR of 146 mm day−1 (data

Fig 3 Mean (±SD, n = 5) outlet (a) pH, and outlet concentration of (b) TSS, (c) COD and (d) BOD 5 from the constructed wetland at different hydraulic loading rates The horizontal bars above the columns show the mean inlet concentrations.

Fig 4 Mean (±SD, n = 5) outlet concentration of (a) TN or TKN, (b) NH 4 -N, (c) TP and (d) PO 4 -P from the constructed wetland at different hydraulic loading rates The

Table 1 Mean (±1 SD, n = 5) pollutant removal efficiencies (%) at the four hydraulic loading rates of 31, 62, 104 and 146 mm day −1 and results of one-way ANOVA statistic (F-ratio).

31 (mm day −1 ) 62 (mm day −1 ) 104 (mm day −1 ) 146 (mm day −1 )

Different superscript letters within rows indicate significant differences based on a Tukey HSD test (p < 0.05).

Data are for TN at hydraulic loading rates of 31 and 62 mm day −1 and TKN at hydraulic loading rates of 104 and 146 mm day −1

* p < 0.05.

*** p < 0.001.

not shown) The composition of the wastewater varied over time

and the mean inlet concentration differed significantly (p < 0.01)

between the four experimental HLRs, except for TSS and total

nitro-gen (Figs 3 and 4) The pH of the inlet was 7.5–7.8, and decreased

to 6.9–7.2 in the outlet (Fig 3a) Mean inlet concentrations of TSS

were low (33–36 mg L−1) and decreased to <5 mg L−1 in the

out-lets (Fig 3b) Mean inlet concentrations of COD varied between

124 and 169 mg L−1 and decreased to 23–63 mg L−1 in the

efflu-ent (Fig 3c) Mean inlet concefflu-entrations of BOD5varied between 40

and 80 mg L−1and decreased to 16–21 mg L−1in the outlet (Fig 3d)

BOD5mass loading rates were 25, 25, 46 and 72 kg ha−1d−1at the

HLRs of 31, 62, 104 and 146 mm day−1, respectively The

concen-trations of the different forms of N and P in the influent water

also differed significantly between the experimental loading

peri-ods (Fig 4) Corresponding outlet concentrations of total N and

NH4-N increased significantly with HLR, and at the highest HLR of

146 mm day−1the outlet concentrations were just slightly lower,

or at the same level, as in the inlet (Fig 4a and b) Outlet

concen-trations of P were consistently low (<0.2 mg L−1) at the two lowest

HLRs, but the concentrations of total P and PO4-P increased to 3.0

and 3.1 mg L−1, respectively, at the highest HLR of 146 mm day−1

(Fig 4c and d)

3.2 Removal efficiencies

The removal efficiencies of the pollutants calculated as the

percent change in concentrations from inlet to outlet decreased

significantly with HLR (Table 1) Particularly at the highest HLR

of 146 mm day−1, the removal percentages were low because

some overland flow occurred in the planted bed, despite the

fact that the outlet pipe controlling the water level was set low

For COD and BOD5 the removal rates ranged between 57% and

84%, and were lowest for BOD5at 62 mm day−1probably because

inlet BOD5 concentrations were relatively low at this HLR TSS

was removed effectively (>93%) at the three lowest HLRs, and

by 86% at the highest HLR (Table 1) Removal of TN and TKN

also decreased with loading rate to 16% at the highest HLR, and

removal of NH4-N decreased 91% at a HLR of 31 mm day−1 to

no removal at a HLR of 146 mm day−1 Removal efficiencies of

TP and PO4-P were high at all four HLRs but decreased from

>98% at the two lowest HLRs to approximately 75% at the highest

HLR

3.3 First-order reaction rate constants

The concentrations of pollutants generally decreased

exponen-tially to non-zero plateaus along the flow path in the planted

bed, particularly BOD , COD, TSS and TP as shown inFig 5for a

HLR of 62 mm day−1 Nitrogen was removed less effectively and did not approach the background plateau towards the outlet of the system (Fig 5d) We used the plug-flow k–C* model pre-sented by Kadlec and Knight (1996) and constant background concentrations (C*) to fit the exponential decrease in pollutant concentration at each HLR, and the estimated first-order area-based removal rate constants (k) and coefficients of determination (R2) for BOD5, COD, TSS, TN (and TKN) and TP are presented in Table 2 The model fitted well for all the measured parameters except TN at the highest HLR of 146 mm day−1 (Table 2) The estimated k-values of BOD5 were in the range of 25–95 m year−1

and highest at high HLR Removal rate constants for COD were lower, 22–30 m year−1, and for TSS the removal constant was nearly fourfold higher at a HLR of 131 mm day−1compared to a HLR of

31 mm day−1corresponding to the fourfold higher HLR Estimated k-values for TN (or TKN) and TP were variable, in the range of 5–24 and 41–84 m year−1, respectively Rate constants generally increased slightly as HLRs increased for BOD5and COD, but not for

TN and TP, where the estimated k-values were highest at a HLR of

62 mm day−1

Table 2 Estimated (±SE) first-order area-based removal rate constants (k) and coefficients

of determination (R 2 ) for BOD 5 , COD, TSS, TN (or TKN) and TP removal in the con-structed wetlands based on the k–C* model at the four HLRs of 31, 62, 104 and

146 mm day − 1

HLR (mm day −1 ) k (m year −1 ) C* (mg L −1 ) R 2

Constant background concentrations (C*) of 5, 10, 0, 1.5 and 0 mg L −1 for BOD 5 , COD, TSS, TN (or TKN) and TP, respectively, were used in the calculations.

Fig 5 The progression of (a) BOD 5 , (b) COD, (c) TSS, (d) TN and (e) TP concentrations through the constructed wetland at HLR of 62 mm day −1 Fitted curves based on the first-order area-based k–C* model Data points were five measurements in every hydraulic loading rate A typical C* of 5, 10, 0, 1.5 and 0 for BOD 5 , COD, TSS, TN and TP were used to fit the model.

4 Discussion

This study for the first time presents first-order removal rate

constants estimated from the exponential decrease in pollutant

concentrations along the flow path from inlet to exit of a

tropi-cal horizontal subsurface flow CW system and at different HLRs

The influent wastewater was rather dilute concerning TSS, BOD

and COD, which is commonly observed in many tropical areas,

where organic pollutants are degraded relatively fast in sewers and

canals as a consequence of the high temperatures Also, in tropical

areas, where part or whole of the sewer system is open channels,

the composition of the wastewater varies over the year in concert

with the dry and rainy seasons The concentrations of nutrients in

the wastewater corresponded, however, to levels in typical

munic-ipal wastewater We suggest that the results obtained in this study

are representative for primary settled municipal sewage in tropical

areas

The removals of TSS, BOD and COD were generally efficient and

decreased only slightly at the highest HLR where some overland

flow occurred Suspended solids are mainly removed by

physi-cal mechanisms, such as sedimentation and filtration processes, and most solids were removed within the pretreatment unit and the first section of the planted bed BOD and COD removal proba-bly mainly occurred by anaerobic process as oxygen levels were consistently low in the planted bed and as pH dropped during passage through the bed Old and decaying roots from previous studies probably decomposed throughout the bed and may have contributed to the relatively high background levels of BOD and COD recorded in the present study

Nitrogen removal in constructed wetlands occurs through adsorption, assimilation into microbial and plant biomass, ammo-nia volatilization and coupled nitrification–denitrification, and so depends much on the environment inside the system Ammonia volatilization was probably very low, as the pH was in the range of 6.9–7.4 The plant uptake was not quantified in the present study, but as the Phragmites plants were planted only 1 month before starting the study, plant uptake of N might have contributed to the recorded nutrient removals, as have also been documented in other studies (Koottatep and Polprasert, 1997; Greenway and Woolley, 2001; Kyambadde et al., 2005; Konnerup et al., 2009) However,

the main removal process for N was probably denitrification, as

the nitrification process, which is a prerequisite for denitrification,

could occur in the upper layer of the sand bed and at oxic microsites

in the rhizosphere of Phragmites At high HLR the outlet NH4-N

con-centration was at the same level as in the influent, showing that

the capacity of the nitrification–denitrification process had been

exceeded

The major processes responsible for P removal in

horizon-tal subsurface flow constructed wetlands are adsorption to the

medium, precipitation and assimilation into microbial and plant

biomass The initial years after construction of the CW system, the

removal rate of TP was very high, 92–98%, at a hydraulic

load-ing of 31 mm day−1 with a slightly decreasing trend in removal

over time (L.A Tuan, 2005, unpublished) Resting of the system

and the renovation prior to the present experiment probably

reconstituted the P adsorption capacity of the sand medium and

contributed to the very efficient P removal obtained, although

plant uptake of P might also have been important The sand

medium used in the bed was river sand with an unknown

chem-ical composition, but high contents of iron and/or calcium might

be responsible for the efficient P removal Efficient adsorption of

P to the bed medium and associated high P removals have also

been reported byKern and Idler (1999), but removals are likely

to decrease over time due to the saturation of P sorption sites in

the medium (Arias et al., 2001; Arias and Brix, 2005; Vohla et al.,

2007)

The HLR and the consequent detention time of the water are

important for the treatment processes in a CW system At low HLR

the detention time is high, and at high HLRs the water passes rapidly

to the outlet reducing the contact time between the wastewater

and the microorganisms, that are responsible for the degradation

processes The four applied HLRs (31, 62, 104 and 146 mm day−1)

gave nominal detention time of 13.9, 6.9, 4.2 and 3 days,

respec-tively, and thus result in up to fourfold differences in the duration

of contact between pollutants and microorganisms The removal

percentage of nutrients was especially affected by the HLR, with

lower removals at high HLRs, while organic and solids removal

percentages were only slightly affected This is agreement with

results from several other CW systems Tanner et al (1995a,b)

reported that, as the detention time increased from 2 to 7 days

in a subsurface flow gravel-bed system treating dairy

wastewa-ter in New Zealand, the mean removal of all analyzed pollutants

increased Also,Huang et al (2000)reported that NH4-N and TKN

concentrations in a subsurface flow constructed wetland

treat-ing domestic wastewater decreased exponentially with increastreat-ing

detention times, andChung et al (2008)in a CW treating

munic-ipal sewage in Hong Kong found lower removal efficiencies for N

and P at 5 days detention time than at 10 days detention time,

but no difference in COD removal The significantly poorer

perfor-mance at a HLR of 146 mm day−1 in the present study therefore

can partly be explained by low detention time at this loading

rate, but hydraulic overloading leading to short circuiting of flow

between inlet and outlet probably also affected performance

sig-nificantly

First-order area-based removal rate constants (k) and apparent

background concentrations (C*) are commonly used in treatment

wetlands design (Kadlec, 2000) although recently the

tanks-in-series model has been advocated to account for the imperfect flow

patterns that prevail in wetlands (Kadlec, 2009) The estimated

removal rate constants and apparent background concentrations

have been shown to depend strongly on input water quality as well

as hydraulic loading rates (Kadlec, 2000, 2003) Furthermore, the

constituents of lumped parameters such as BOD may be degraded

or removed at different rates within a CW system with

correspond-ing differences in removal rate constants (Kadlec, 2003) In the

present study, we used the internal concentration profiles from inlet to outlet of the CW system as the basis for estimating removal rate constants This procedure counteracts some of the problems listed above and should therefore lead to more reliable estimates

of the rate constants

The k–C* model used fitted the concentrations profiles quite well for all parameters at HLRs up to 104 mm day−1 At the highest HLR of 146 mm day−1the goodness of fit was poorer as indicated

by the lower coefficients of determination (R2) The estimated k-values for BOD5and COD removal differed between HLRs, but not

in a consistent manner, and were in the range of 25–95 m year−1for BOD and 22–30 m year−1for COD, similar to the range reported for many horizontal flow CWs (Kadlec and Knight, 1996; Rousseau et al., 2004; Konnerup et al., 2009; Kadlec, 2009) The estimated rate constants for TSS removal increased nearly fourfold in response to the fourfold increase in HLR, showing that the k–C* model does not adequately describe the TSS removal in the system Most TSS was removed at the front end of the system by filtration and sedi-mentation, and independently on HLR First-order removal models can therefore not be used to describe TSS removal in horizontal subsurface flow CWs

The estimated k-values for TN were in the range of 5–24 m year−1 which is higher than reported byKonnerup et al (2009), who found k-values for TN in the range of 3–6 m year−1

in tropical subsurface flow CWs treating domestic sewage In ver-tical flow CWs treating high-strength synthetic wastewater at HLRs between 20 and 80 mm day−1 in Thailand,Kantawanichkul

et al (2009)found k-values for TKN 30 m year−1 Median rate con-stants for TN in 136 free water surface wetlands are reported

to be 12.6 m year−1 and in 123 horizontal subsurface wetlands 8.4 m year−1 (Kadlec, 2009) However, as total N is a lumped parameter consisting of a mixture of organic-N and different inorganic ions such as NH4-N and NO3-N, removal of total N can in principle not be modeled using a one component k–C* model as in the present study Rather, a sequential conversion model that includes ammonification, nitrification and denitri-fication should be used to better describe removal (Kadlec, 2009)

The k-values for TP were very high, in the range of 41–84 m year−1, which is very much higher than the 2–5 m year−1

found byKonnerup et al (2009)and the 13.5 m year−1reported by Kantawanichkul et al (2009) The median rate constant for TP in

282 free water surface flow wetlands is 10 m year−1, and in 82 hor-izontal subsurface flow wetlands 6 m year−1 (Kadlec, 2009) The main removal process for P at the high loading rates used in the present study is adsorption and/or precipitation to the bed medium (Brix et al., 2001; Del Bubba et al., 2003; Arias and Brix, 2005) The sandy bed medium used in the experimental CW system thus had

a high P sorption capacity compared to the coarser media used in other tropical studies (Konnerup et al., 2009; Kantawanichkul et al., 2009) Further studies are however needed to clarify the longer term performance as the sorption sites are likely to be saturated over time (Brix et al., 2001)

The Vietnamese national standards (TCVN 5945-2005) set by the Ministry of Science, Technology and Environment for effluent discharged into water bodies from which domestic water is taken are 30, 50 and 50 mg L−1for BOD5, COD and TSS, respectively, and

15 mg L−1 for TN, 5 mg L−1 for ammonia-N, and 4 mg L−1for total

P The discharge from the CW systems at the two lowest HLRs of

31 and 62 mm day−1 met these standards, except for

ammonia-N and Tammonia-N which were only slightly higher than the standards at

a HLR of 62 mm day−1 The Vietnamese standards for discharge into water bodies used for transportation, irrigation, fishing, or bathing are less stringent, and actually the effluent from the CW system met these standards at all HLRs Constructed wetland

sys-tems with horizontal subsurface flow thus are a suitable technology

to treat domestic wastewater in Vietnam, and at HLR up to

approx-imately 50 mm day−1(equivalent to 50 L m−2), CW systems of the

type studied here will meet even the strictest Vietnamese effluent

standard

5 Conclusions

The results of this study document that domestic wastewater

can be treated in horizontal subsurface flow constructed wetland

systems to meet even the most stringent Vietnamese standards for

discharge into surface waters Under tropical climatic conditions,

the area-based removal rate constants seem to be significantly

higher than previously reported for CW systems in temperate

areas, and the higher water temperatures are the most plausible

explanation We estimated removal rate constants based on

pol-lutant concentration profiles from inlet to outlet as this procedure

is more robust in relation to variations in hydraulic loading rate

and wastewater composition However, the estimated k-values

should not be used for design purposes, as site-specific

differ-ences and stochastic variability can be high because of factors

such as weather, plant growth, and unpredictable fluctuations

in input water quality and flow The river sand used as growth

substrate in the planted bed had excellent P-removal

proper-ties, but the chemical characteristics of the sand has not yet

been studied and the long-term performance is not known The

present study documents that constructed wetland systems do

perform very well in a tropical setting, and we hope that the

promising results will contribute to stimulate to a broader use

of CW systems as a robust, reliant, cost-effective and efficient

wastewater management systems in Vietnam and other tropical

countries

Acknowledgements

Funding for this project came from the Danish

Interna-tional Development Agency (DANIDA) through the Can Tho

University–University of Aarhus Link in Environmental Science

(CAULES) project

References

APHA, 1998 Standard Methods for the Examination of Water and Wastewater, 20th

ed American Public Health Association, Washington, DC.

Arias, C.A., Brix, H., 2005 Phosphorus removal in constructed wetlands: can suitable

alternative media be identified? Water Sci Technol 51 (9), 267–273.

Arias, C.A., Del Bubba, M., Brix, H., 2001 Phosphorus removal by sands for use as

media in subsurface flow constructed reed beds Water Res 35 (5), 1159–1168.

Bayley, M.L., Davison, L., Headley, T.R., 2003 Nitrogen removal from domestic

efflu-ent using subsurface flow constructed wetlands: influence of depth, hydraulic

residence time and pre-nitrification Water Sci Technol 48 (5), 175–182.

Bojcevska, H., Tonderski, K., 2007 Impact of loads, season, and plant species on the

performance of a tropical constructed wetland polishing effluent from sugar

factory stabilization ponds Ecol Eng 29 (1), 66–76.

Brix, H., Arias, C.H., Del Bubba, M., 2001 Media selection for sustainable

phospho-rus removal in subsurface flow constructed wetlands Water Sci Technol 44

(11–12), 47–54.

Brix, H., Koottatep, T., Laugesen, C.H., 2007a Wastewater treatment in tsunami

affected areas of Thailand by constructed wetlands Water Sci Technol 56 (3),

69–74.

Brix, H., Schierup, H.-H., Arias, C.A., 2007b Twenty years experience with

con-structed wetland systems in Denmark: what did we learn? Water Sci Technol.

56 (3), 63–68.

Chung, A.K.C., Wu, Y., Tam, N.F.Y., Wong, M.H., 2008 Nitrogen and phosphate

mass balance in a sub-surface flow constructed wetland for treating municipal

wastewater Ecol Eng 32 (1), 81–89.

Del Bubba, M., Arias, C.A., Brix, H., 2003 Phosphorus absorption maximum of sands

for use as media in subsurface flow constructed reed beds as measured by the

Langmuir isotherm Water Res 37 (14), 3390–3400.

Denny, P., 1997 Implementation of constructed wetlands in developing countries.

El Hamouri, B., Nazih, J., Lahjouj, J., 2007 Subsurface-horizontal flow constructed wetland for sewage treatment under Moroccan climate conditions Desalination

215 (1–3), 153–158.

Greenway, M., Woolley, A., 2001 Changes in plant biomass and nutrient removal over 3 years in a constructed wetland in Cairns, Australia Water Sci Technol.

44 (11–12), 303–310.

Haberl, R., 1999 Constructed wetlands: a chance to solve wastewater problems in developing countries Water Sci Technol 40 (3), 11–17.

Huang, J., Reneau Jr., R.B., Hagedorn, C., 2000 Nitrogen removal in constructed wet-lands employed to treat domestic wastewater Water Res 34 (9), 2582–2588 Kadlec, R.H., 2000 The inadequacy of first-order treatment wetland models Ecol Eng 15 (1–2), 105–119.

Kadlec, R.H., 2003 Effects of pollutant speciation in treatment wetlands design Ecol Eng 20 (1), 1–16.

Kadlec, R.H., 2009 Comparison of free water and horizontal subsurface treatment wetlands Ecol Eng 35 (2), 159–174.

Kadlec, R.H., Knight, R.L., 1996 Treatment Wetlands Lewis Publishers, CRC Press Inc., Boca Raton, FL.

Kantawanichkul, S., Kladprasert, S., Brix, H., 2009 Treatment of high-strength wastewater in tropical vertical flow constructed wetlands planted with Typha angustifolia and Cyperus involucratus Ecol Eng 35 (2), 238–247.

Kern, J., Idler, C., 1999 Treatment of domestic and agricultural wastewater by reed bed systems Ecol Eng 12 (1–2), 13–25.

Kivaisi, A.K., 2001 The potential for constructed wetlands for wastewater treatment and reuse in developing countries: a review Ecol Eng 16 (4), 545–560 Konnerup, D., Koottatep, T., Brix, H., 2009 Treatment of domestic wastewater in tropical, subsurface flow constructed wetlands planted with Canna and Helico-nia Ecol Eng 35 (2), 248–257.

Koottatep, T., Polprasert, C., 1997 Role of plant uptake on nitrogen removal in con-structed wetlands located in the tropics Water Sci Technol 36 (12), 1–8 Kyambadde, J., Kansiime, F., Dalhammar, G., 2005 Nitrogen and phosphorus removal

in substrate-free pilot constructed wetlands with horizontal surface flow in Uganda Water Air Soil Pollut 165 (1–4), 37–59.

Masi, F., Martinuzzi, N., 2007 Constructed wetlands for the Mediterranean coun-tries: hybrid systems for water reuse and sustainable sanitation Desalination

215 (1–3), 44–55.

Nahlik, A.M., Mitsch, W.J., 2006 Tropical treatment wetlands dominated by free-floating macrophytes for water quality improvement in Costa Rica Ecol Eng 28 (3), 246–257.

Rousseau, D.P.L., Vanrolleghem, P.A., De Pauw, N., 2004 Constructed wetlands in Flanders: a performance analysis Ecol Eng 23 (3), 151–163.

Tanner, C.C., Clayton, J.S., Upsdell, M.P., 1995a Effect of loading rate and planting

on treatment of dairy farm wastewaters in constructed wetlands-I Removal

of oxygen demand, suspended solids and faecal coliforms Water Res 29 (1), 17–26.

Tanner, C.C., Clayton, J.S., Upsdell, M.P., 1995b Effect of loading rate and planting

on treatment of dairy farm wastewaters in constructed wetlands-II Removal of nitrogen and phosphorus Water Res 29 (1), 27–34.

Vohla, C., Alas, R., Nurk, K., Baatz, S., Mander, U., 2007 Dynamics of phosphorus, nitrogen and carbon removal in a horizontal subsurface flow constructed wet-land Sci Total Environ 380 (1–3), 66–74.

Vymazal, J., 2009 The use constructed wetlands with horizontal sub-surface flow for various types of wastewater Ecol Eng 35 (1), 1–17.

Zurita, F., De Anda, J., Belmont, M.A., 2009 Treatment of domestic wastewater and production of commercial flowers in vertical and horizontal subsurface-flow constructed wetlands Ecol Eng 35 (5), 861–869.

Ngo Thuy Diem Trang (born 1976, Vietnam) got her M.Sc.

in 2004 from the University of Putra Malaysia, Malaysia She is currently carrying out her Ph.D at the Depart-ment of EnvironDepart-mental Sciences, College of EnvironDepart-ment and Natural Resources, Can Tho University (Vietnam) and Aarhus University (Denmark), under the supervision of Drs Hans Brix, Hans-Henrik Schierup and Nguyen Huu Chiem The topic of her studies is the use of constructed wetlands in pollution control, particularly in recirculating aquaculture systems.

Dennis Konnerup is carrying out his Ph.D study at Aarhus University, Denmark The topic of his Ph.D work is the ecophysiology of wetland plants and the use of con-structed wetlands for the treatment of wastewater, with particular focus on wetlands in tropical areas.

Hans-Henrik Schierup is an emeritus professor in the

Department of Biological Sciences, Aarhus University

(Denmark) He is specialised in limnology and

ecophys-iology of wetland plants.

Nguyen Huu Chiem is the head of the Department of

Envi-ronmental Sciences, Vice Dean College of Environment

and Natural Resources, Can Tho University, Can Tho City,

Vietnam He is specialised in soil science, social-economic

and wetland ecology and management in the Mekong

Delta.

Le Anh Tuan is a senior lecturer in the Department of Envi-ronment and Natural Resources Management, College of Environment and Natural Resources, Can Tho University, Vietnam He got his Ph.D in Hydrological Environment in Belgium in 2008 He is interested in the design of con-structed wetlands, wetland ecology and climate change.

Hans Brix is a professor in the Department of Biological Sciences, Aarhus University (Denmark) He is specialised

in ecophysiology of wetland plants and use of wetlands in water pollution control.