See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/6037322

A mass balance study on nitrification and deammonification in vertical flow

constructed wetlands treating landfill leachate

Article in Water Science & Technology · August 2007

DOI: 10.2166/wst.2007.503 · Source: PubMed

CITATIONS READS

11 670

2 authors:

Guangzhi Sun David Austin

Chinese Academy of Sciences Jacobs Engineering Group
110 PUBLICATIONS 3,238 CITATIONS 24 PUBLICATIONS 553 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Removal of Pharmaceutical Compounds from Municipal Sewage View project

The Removal of Metals from Mining Wastewater using Subsurface Flow Constructed Wetland View project

All content following this page was uploaded by Guangzhi Sun on 20 November 2014.

The user has requested enhancement of the downloaded file.

A mass balance study on nitrification and

Water Science & Technology Vol 56 No 3 pp 117–123 Q IWA Publishing 2007

deammonification in vertical flow constructed
wetlands treating landfill leachate
G. Sun* and D. Austin**
*Department of Civil Engineering, Monash University, Building 60, VIC 3800, Australia
**North American Wetland Engineering, 4444 Centerville Road, Suite 140, White Bear Lake, MN, 55127,
USA

Abstract A laboratory-scale, mass-balance study was carried out on the transformation of nitrogenous
pollutants in four vertical flow wetland columns. Landfill leachate containing low organic matter, but a high
concentration of ammoniacal-nitrogen, was treated under dissolved oxygen concentrations close to
saturation. Influent total nitrogen (TN) comprised ammoniacal-nitrogen with less than 1% nitrate and nitrite,
negligible organic nitrogen, and very low BOD. Nitrification occurred in three of the four columns. There was
a substantial loss of total nitrogen (52%) in one column, whereas other columns exhibited zero to minor
losses (, 12%). Nitrogen loss under study conditions was unexpected. Two hypotheses are proposed to
account for it: (1) either the loss of TN is attributed to nitrogen transformation into a form (provisionally
termed a-nitrogen) that is undetectable by the analytical methods used; or (2) the loss is caused by
microbial denitrification or deammonification. By elimination and stoichiometric mass balance calculations,
completely autotrophic nitrogen-removal over nitrite (CANON) deammonification is confirmed as responsible
for nitrogen loss in one column. This result reveals that CANON can be native to aerobic engineered
wetland systems treating high ammonia, low organic content wastewater.
Keywords CANON; deammonification; denitrification; engineered wetland; nitrification; reed bed

Introduction
Engineered wetlands have become a popular technical alternative worldwide for
treatment of various wastewaters (Kadlec and Knight, 1996). In the United Kingdom, 840
engineered wetlands have been built since the mid-1980s and recorded in a database
(CWA, 2005), while another 200 may be operating in addition to this record. Despite this
experience, however, nitrogen removal remains a challenge. In Europe, typical percentage
removal of ammoniacal-nitrogen (NH4-N) in long-term, engineered wetland operation is
only in the range of 35– 50% (Verhoeven and Meuleman, 1999).
Classic biological nitrogen removal (BNtR) entails an obligate aerobic, autotrophic
oxidation of ammonia to nitrite and then to nitrate, followed by an obligate anoxic,
heterotrophic reduction of nitrate (Cooper et al., 1996). A classic mass balance on total
nitrogen for a nitrifying reactor will show that TNinfluent ¼ TNeffluent þ D, where D
represents assimilative nitrogen removal or, in soil systems, adsorption of ammonium
(NHþ 4 ), both of which are quantifiable for a given study system.
Recent advances in microbiology have revealed that classic concepts of biological
nitrogen transformations are incomplete. Heterotrophic nitrification and autotrophic
deammonification, and methane oxidation by NO2 and NO3, occur in nature and have
great potential for advancing BNtR reactor theory (Robertson and Keunen, 1990; Hippen
et al., 1997; van Loodstrecht and Jetten, 1998; Raghoebarsing et al., 2006). These novel
mechanisms can explain the apparent failure of nitrogen mass balance studies despite
careful execution of analytical protocols. For example, nitrogen can “disappear” in
doi: 10.2166/wst.2007.503 117
 nitrifying, engineered wetlands (Austin et al., 2003; Sun et al., 2005). Apparent failure,
however, may reveal the presence of novel mechanisms or abiotic processes, such as
NHþ 4 adsorption or heterogeneous struvite (MgNH4PO4z6H2O) precipitation (Wang et al.,
2006). Mass balance analyses force a mechanistic accounting of the fate of all influent
nitrogen. This study uses mass balance analyses to explore the mechanisms of nitrogen
transformations in laboratory-scale, vertical-flow, wetland columns treating a landfill lea-
chate characterised by low biochemical oxygen demand (BOD) and high NH4-N
G. Sun and D. Austin

concentrations.

Methods
Experiments were carried out in four column-scale vertical flow wetlands with identical
dimensions, named columns A, B, C and D. Each was made of a Perspex column of
95 mm in diameter and 900 mm in height and filled with 26.4 ^ 7.2 mm round gravel to
a depth of 150 mm as the bottom layer. The main layer, 650 mm deep, contained
4.4 ^ 1.5 mm river pea gravel where Phragmites australis was planted. The operational
histories of these columns varied. Columns A, B and D were constructed in summer 2002
for studies on the treatment of strong agricultural effluents (Zhao et al., 2004); columns
A and B were operated under high organic loading, while column D was under relatively
lower loading. A substantial, but unquantifiable, residue of biomass and solids had been
deposited in columns A and B prior to the current study in 2005. Column C was built in
summer 2004 and was used for this study only.
Experiments were carried out in two periods. The initial purpose of the experiments
was to investigate the removal rate of ammoniacal-nitrogen from a landfill leachate.
Substantial, unexpected differences were found between the performances of individual
columns during period 1. In response, the study objective changed to investigate routes
of nitrogen transformation in each column. Alterations to the experiment setup were
made at the start of period 2. In period 1, columns A, B and D were arranged as three
reactors in series. In period 2, column C was added, and the columns were rearranged
into two parallel systems in series, columns A þ B and columns C þ D (Figure 1).
Landfill leachate was collected from a municipal waste disposal site in Portadown,
Northern Ireland. In period 1, leachate was pumped into column A and passed through
columns B and C sequentially. In period 2, columns A and C were dosed with the raw
leachate simultaneously; effluents from A and C were pumped to B and D, respectively.
Effluent recirculation was employed individually in each column. The flow rates of feed
and recirculation were carefully controlled by peristaltic pumps set to rotate at required
velocities. The ratio of recirculation to inlet flow rate in each individual column was
fixed at 2.5:1, while the flow rate of raw leachate was controlled at 1.35 L per day during
both experimental periods, giving a net hydraulic loading of 0.19 m3 m22 d21 on each
column. At the start of period 1, a stabilisation time of 14 days was allowed before the
collection of samples began.

118 Figure 1 Schematic diagram of the constructed wetland columns in period 1 (left) and period 2 (right)
 Wastewater samples were collected from the inlet and outlet of each column during
the experiments, six sets in each period. The samples were analysed for total nitrogen
(TN), NH4-N, NO2-N, NO3-N, chemical oxygen demand (COD), PO4-P, suspended solids
(SS), dissolved oxygen (DO) and pH. In period 1, BOD analyses were done on three sets
of samples of raw leachate and column D effluent. BOD analysis was not carried out in
period 2. The values of NH4-N were analysed using a Sension II pH/ISE meter combined
with an ammonia electrode. A Piccolo pH meter was used for the analysis of pH. A

G. Sun and D. Austin

Sension electrode determined DO. A Hach DR2010 spectrophotometer analysed the
remaining parameters per the manufacturer’s methods: TNT (test-in-tube) persulphate
digestion method for TN, ferrous sulphate method for NO2-N, and cadmium reduction
method for NO3-N. It should be noted that the analysis of TN is independent of NH4-N,
NO3-N, and NO2-N analyses. Persulphate digestion oxidises all nitrogen to nitrate, which
is analysed by a colorimetric reagent. After period 2, the roots of Phragmites in each
column were collected, dried at 80 8C for 24 hours, and weighed.

Results and discussion

Overall performance
In both periods, DO in the leachate was close to saturation, as shown in Table 1. The
percentage removals of BOD, PO4-P and SS in different columns varied considerably,
from 0 to 47%, due to different operating history of these columns. Columns A and C had
been heavily used in previous experiments with strong wastewater (Zhao et al., 2004),
whereas the other two columns were not used (C) or only lightly (D) used.
Table 1 clearly indicates that the main pollutant in the leachate is NH4-N. There was
little organic carbon in the leachate as can be seen in the low BOD5 influent concen-
tration. The COD concentrations are attributable to high ammonia concentrations (Hach,
2004) and, possibly, to dissolved recalcitrant carbon. On average, NH4-N was removed
by 36% in period 1 across columns A, B and D that were operating in series (Table 1). In
period 2, NH4-N was removed by 32% across columns A and B and 67% across C and
D. Generally higher average NH4-N removal rates were obtained in period 2 than in
period 1. In columns A and B, there was a general trend of increasing NH4-N removal
rate with time (Figure 2). In contrast, the removal rate decreased steadily with time in
column C. In column D, with the highest treatment efficiency of NH4-N, the removal rate
peaked after five weeks of operation.

Transformation of nitrogen in the wetland columns

Negligible organic nitrogen in the raw leachate is evident from close agreement
(D # 1%) between the average values of persulphate total nitrogen (TN) and the sum of
NH4-N, NO2-N and NO3-N (SN), as shown in Table 2. Concentrations of TN and SN in
the effluents from columns A, B and C are also in close agreement (D # 1.8%).
Therefore, in the absence of organic nitrogen, SN reliably accounts for total nitrogen.
The large difference between persulphate digestion TN and SN in column D effluent
(Table 2) may be attributable to humic substances; a known matrix interference for the
persulphate digestion method (Hach, 2004). Dry weight root mass was greater in column
D (15.1 g) than in columns A (8.2 g), B (7.6 g) or C (7.2 g). Greater humic content in
column D would exert higher persulphate demand, causing a falsely high effluent TN.
Large differences appeared in NH4-N removal rates between columns, despite use of
the same media and operating under the same hydraulic and NH4-N loadings (Figure 2).
Mass balance analysis supports nitrification as the mechanism of NH4-N removal in
columns A and B; rates of nitrification in these two columns are similar (Figure 2). In
column D, high effluent NO3-N concentrations demonstrate that oxidation of ammonia 119
120
G. Sun and D. Austin

Table 1 Mean performance and standard deviation of columns. All values except pH are presented as mg/L. Delta percentage change D% is based on effluent from previous column (EFF) or
influent (INF), whichever is the immediate preceding source. Note that BOD5 values are averages of three sets of data

Parameter Column A Column B Column C Column D

INF EFF D% EFF D% EFF D% EFF D%

Period 1 (n ¼ 6)
NH4-N 490 ^ 15 463 ^ 7 25.5 434 ^ 21 26.2 315 ^ 25 2 27
NO2-N 0.2 ^ .04 7.5 ^ 2.7 þ 3,650 9.7 ^ 2.7 þ 29 8.7 ^ 2.2 2 10
NO3-N 1.6 ^ .06 14 ^ 3.0 þ 775 15 ^ 5.8 þ 7.1 34 ^ 9.1 þ 127
TN 497 ^ 50 492 ^ 45 21.0 467 ^ 29 25.1 438 ^ 46 26.2
BOD5 23 ^ 5 13 ^ 6 2 43
COD 691 ^ 69 618 ^ 92 211 610 ^ 71 21.3 552 ^ 33 29.5
PO4-P 13 ^ 0.9 12 ^ 0.6 27.7 11 ^ 0.6 28.3 11 ^ 1 0.0
SS 22 ^ 2.0 16 ^ 2 227 14 ^ 4 213 9^3 2 36
DO 7.0 ^ 0.4 9.2 ^ 0.4 þ 31 9.7 ^ 0.4 þ 5.4 9.9 ^ 0.3 þ 2.2
pH 8.5 ^ 0.1 8.5 ^ 0.1 0.0 8.6 ^ 0.1 þ 1.2 7.8 ^ 0.3 29.3
Period 2 (n ¼ 6) D% C
NH4-N 483 ^ 22 388 ^ 19 220 328 ^ 28 215 422 ^ 28 213 160 ^ 54 2 62
NO2-N ND 42 ^ 4.5 NA 40 ^ 8 24.8 1.8 ^ 1.3 NA 2.5 ^ 0.8 þ 39
NO3-N 1.4 ^ 0.6 46 ^ 14 þ 3,186 55 ^ 10 þ 20 1.3 ^ 0.6 27.1 43 ^ 14 þ 3,208
TN 485 ^ 27 477 ^ 14 21.6 427 ^ 27 210 430 ^ 50 211 362 ^ 63 2 16
COD 635 ^ 81 571 ^ 47 210 530 ^ 71 27.2 578 ^ 62 29.0 432 ^ 123 2 25
PO4-P 14 ^ 0.9 14 ^ 0.6 0.0 14 ^ 1 0.0 12 ^ 0.5 214 10 ^ 1 2 17
SS 29 ^ 2 22 ^ 4 224 17 ^ 2 223 19 ^ 3 234 10 ^ 3 2 47
DO 6.5 ^ 0.4 9.5 ^ 0.4 þ 46 9.6 ^ 0.3 þ 1.1 9.6 ^ 0.2 þ 48 9.7 ^ 0.1 þ 1.0
pH 8.4 ^ 0.1 7.9 ^ 0.1 26.0 8.1 ^ 0.2 þ 2.5 8.3 ^ 0.2 21.2 7.1 ^ 0.2 2 14
 G. Sun and D. Austin
Figure 2 Variation of ammoniacal-nitrogen removal rates with operation time (the counting of operation time
for each individual column started when the leachate was first pumped into it during this study)

occurred, but differently than in columns A and B. The NO2-N:NO3-N ratio was
0.73, 0.91, and 0.06 in columns A, B and D, respectively. In column C, slight removal of
NH4-N was not accompanied by a rise in NO3-N or NO2-N concentrations, indicating
that nitrifying bacteria were not established. Initially higher, then decreasing, the removal
of NH4-N in column C is consistent with the mechanism of sorption/cation exchange
moving toward saturation of sorption/exchange sites (Connolly et al., 2004).

Nitrogen losses
The loss of total nitrogen (SN) in different columns varied from 1.4 to 52% (Table 3).
Column A exhibited negligible nitrogen loss; a classic nitrifying reactor mass balance
result. In periods 1 and 2, SN losses were 5 and 12% in column B, and 22 and 52% in
column D. The loss in column B may be insignificant, but also could be attributed to
mechanisms at work in column D. Because of adsorption and lack of nitrifying activity,
nitrogen loss in C is not important to this discussion. Column D is clearly different from
the other columns. It has a large missing SN fraction that must be accounted for, which
is provisionally named a-nitrogen because it is uncharacterised.
Alpha-nitrogen may be attributed to (1) experimental error, (2) adsorption/precipita-
tion of soluble nitrogen to solids, (3) off-gassing of unionised ammonia, or (4) biological
conversion of nitrogen to biomass or gas. Experimental error does not account for
a-nitrogen. In columns A, B and C, consistency of methods of total nitrogen analysis
demonstrates precision in measurement. Both methods demonstrated mass recovery of
nitrogen in columns A, B and C, but substantial loss of nitrogen in column D. The same
error repeated by independent means in column D is improbable. Adsorption can be ruled
out by comparison to column C. Struvite (MgNH4PO4·6H2O) formation can be ruled out
because the molar ratio of PO4:NH4 is 1:1, entailing a large loss of ammonium with
complete P removal, which was not observed. The pH in column D was 7.1; thus almost
all ammonia was in ionic form and could not off-gas. Assimilative loss to nitrifying

Table 2 Total nitrogen recovery comparison between persulphate digestion (TN) and SN

Wastewater stream analysed Period 1 Period 2

SN, mg/L Persulphate TN, mg/L D% SN, mg/L Persulphate TN, mg/L D%

Raw leachate 492 497 þ 1.0 484 485 þ 0.2

Effluent from column A 484 492 þ 1.5 477 477 þ 0.0
Effluent from column B 459 467 þ 1.8 422 427 þ 1.2
Effluent from column C – – – 425 430 þ 1.2
Effluent from column D 358 438 þ 22 205 362 þ 76
121
 Table 3 Mean loss of SN in the wetland columns from influent SN entering each column

Waste stream analysed Period 1 Period 2

SN concentration, mg/L Percent loss, % SN concentration, mg/L Percent loss, %

Raw leachate 492 – 484 –

Effluent from column A 484 1.6 477 1.4
Effluent from column B 459 5.2 422 12
G. Sun and D. Austin

Effluent from column C – – 425 12

Effluent from column D 358 22 205 52

bacteria is impossible as it entails yields differing by orders of magnitude in columns D

and A for the same process. By elimination, a-nitrogen is nitrogen gas.
Autotrophic bacteria must be responsible for the conversion of soluble nitrogen to
nitrogen gas due to the lack of biodegradable organic matter in the raw leachate. Comple-
tely autotrophic nitrogen-removal over nitrite (CANON) is consistent with the column D
data. CANON is closely related to Anammox (anaerobic ammonium oxidation) process
in which ammonia is partially oxidised to nitrite, and nitrite together with remaining
ammonia is converted to nitrogen gas by planctomycetes (van Loosdrecht and Jetten,
1998). Unlike Anammox, CANON can occur aerobically in a single-stage reactor
(Third et al., 2001), as under study conditions. Nitrogen mass balance can be calculated
according to Eq (1), by combining the steps of ammonia oxidation to nitrite and Ana-
mmox deammonification (Third et al., 2001). In column D during period 2, mean NH4-N
concentrations at inlet and outlet were 422 mg L21 and 160 mg L21, respectively
(Table 1). Thus 262 mg L21 NH4-N was converted to other forms of nitrogen. Converting
the molar relationships in Eq (1) to mass, 262 mg of NH4-N forms 228 mg N2 and 34 mg
NO3-N per litre of wastewater (compare to 43 mg L21 NO3-N of effluent). If
stoichiometric N2 is added to the actual effluent SN of 205 mg (Table 3), the revised SN
is 433 mg. Compared to the influent concentration of 422 mg L21, the revised influent-
effluent mass balance is within 2.6%. CANON is clearly responsible for the nitrogen loss
in column D.

NHþ 2
4 þ 0:85O2 ! 0:435N2 þ 0:13NO3 þ 1:3H2 O þ 1:4H
þ
ð1Þ

Third et al. (2001) report that the CANON process is subject to competition by nitrite
oxidising bacteria below ammonia loadings of 0.12 kg N m23 d21. Mean study loading
was 0.11 kg N m23 d21, suggesting that the system was operated in a transitional loading
regime. In this regime, subtle differences could explain the dominance of CANON in
column D, nitrification in column A, and slight CANON activity in column B.

Conclusions
Mass balance has demonstrated that the CANON system operates in aerobic subsurface
flow wetlands treating high ammonia wastewater with low organic carbon. CANON is
not an exotic process. Rather, it is native to nitrifying wetlands and undoubtedly will
become a common process in wastewater engineering. The CANON system was not
anticipated at the outset of this study. By accident, the study operated in a transitional
ammonia loading regime, evidently allowing differing competitive outcomes between
N-cycle bacteria consortia in the columns because of subtle differences in initial con-
ditions (operational history). For nitrifying wetlands, an ammonia loading (as N) greater
122 than or equal to 0.11 kg m23 d21 can be considered to stimulate CANON.
References
Austin, D., Lohan, E. and Verson, E. (2003). Nitrification and denitrification in a tidal vertical flow wetland
pilot. In: Proceedings of the Water Environment Federation Technical Conference, Los Angeles, USA.
Connolly, R., Zhao, Y., Sun, G. and Allen, S. (2004). Removal of ammoniacal-nitrogen from an artificial
landfill leachate in downflow reed beds. Process Biochem., 39(12), 1971– 1976.
Cooper, P.F., Job, G.D., Green, M.B. and Shutes, R.B.E. (1996). Reed Beds and Constructed Wetlands for
Wastewater Treatment, WRc Publications, Swindon, UK.

G. Sun and D. Austin

CWA (2005). Constructed Wetlands Database. Version 8.02. Constructed Wetland Association, UK.
Hach Company (2004). Water Analysis Handbook. Hach, Loveland, CO, USA.
Hippen, A., Rosenwinkel, K.H., Baumgarten, G. and Seyfried, C.F. (1997). Aerobic deammonification: a new
experience in the treatment of wastewaters. Wat. Sci. Tech., 35(10), 111 – 120.
Kadlec, R.H. and Knight, R.L. (1996). Treatment Wetlands, CRC Press, Boca Raton, USA.
Raghoebarsing, A.A., Arjan, P., van de Pas-Schoonen, K.T., Smolders, A.J.P., Ettwig, K.F., Rijpstra, W.I.C.,
Schouten, S., Sinninghe Damste, J.S., Op den Camp, H.J.M., Jetten, M.S.M. and Strous, M. (2006). A
microbial consortium couples anaerobic methane oxidation to denitrification. Nature, 440(13), 918 – 921.
Robertson, L.A. and Keunen, J.G. (1990). Combined heterotrophic nitrification and aerobic denitrification in
Thiosphaera pantotropha and other bacteria. Antonie Van Leeuwenhoek, 57(3), 139 – 152.
Sun, G., Zhao, Y. and Allen, S.J. (2005). Enhanced removal of organic matter and ammoniacal-nitrogen in a
column experiment of tidal flow constructed wetland system. J. Biotech., 115(2), 189 – 197.
Third, K.A., Olav Sliekers, A., Kuenen, J.G. and Jetten, M.S.M. (2001). The CANON system (completely
autotrophic nitrogen-removal over nitrite) under ammonium limitation: interaction and competition
between three groups of bacteria. System. Appl. Microbiol., 24(4), 588 –596.
van Loosdrecht, M.C.M. and Jetten, M.S.M. (1998). Microbiological conversions in nitrogen removal. Wat.
Sci. Tech., 38(1), 1 –7.
Verhoeven, J.T.A. and Meuleman, A.F.M. (1999). Wetlands for wastewater treatment: opportunities and
limitations. Ecol. Eng., 12(1/2), 5 – 12.
Wang, J., Burken, J.G. and Zhang, X. (2006). Effect of seeding materials and mixing strength on struvite
precipitation. Wat. Environ. Res., 78(2), 125 – 132.
Zhao, Y.Q., Sun, G. and Allen, S.J. (2004). Anti-sized reed bed system for animal wastewater treatment: a
comparative study. Wat. Res., 38(12), 2907– 2917.

123

View publication stats