IOP Conference Series: Earth and Environmental Science

PAPER • OPEN ACCESS

Effect of Sulfate Load on Sulfur Removal in Model Constructed Wetlands

To cite this article: Jiangxin Hou et al 2018 IOP Conf. Ser.: Earth Environ. Sci. 186 012078

View the article online for updates and enhancements.

This content was downloaded from IP address 181.214.44.158 on 12/10/2018 at 13:27

GBEM IOP Publishing
IOP Conf. Series: Earth and Environmental Science 186 (2018)
1234567890 ‘’“” 012078 doi:10.1088/1755-1315/186/3/012078

Effect of Sulfate Load on Sulfur Removal in Model

Constructed Wetlands

Jiangxin Hou, Wenrui Guo, and Yue Wen*

State Key Laboratory of Pollution Control and Resource Reuse, College of
Environmental Science and Engineering, Tongji University, Shanghai 200092, P.R.
China
Corresponding author’s e-mail address: weny@tongji.edu.cn.

Abstract. Four model constructed wetlands (CWs) were designed to investigate the effects of
sulfate load on sulfate and sulfide removal. The results showed that as the sulfate load
increased from 1.42 to 7.01 g S m–3 d–1, the sulfate removal rate increased from 1.42 to 3.16 g
S m–3 d–1, and the sulfide discharge rate increased from 0.08 to 1.46 g S m–3 d–1. The total
sulfur removal rate ranged between 1.29 and 1.74 g S m–3 d–1. The sulfide in the effluent only
accounted for 5.55%–46.9% of the removed sulfate. This indicated that CWs can effectively
reduce sulfide discharge while removing sulfate. The conversion of dissolved sulfide into
deposited sulfur by CW matrix was a main way for sulfide removal. Elemental sulfur, acid
volatile sulfide (AVS), and pyrite-sulfur were the main forms of sulfur deposition in this study.
The accumulations of these three sulfur compounds were 16.6–36.2, 22.3–36.0, and 49.7–63.6
mg S kg–1 gravel, respectively. Sulfur balance analysis showed that 42.9%–71.1% of the
removed sulfate was deposited in the matrix, and only 0.84%–2.34% was absorbed by the plant.

1. Introduction
Domestic sewage and industrial wastewater both contain sulfate. The sulfate concentration in the
municipal sewage is generally 40–200 mg L–1 [1]. In some coastal areas such as Hong Kong, the
sulfate concentration in the municipal sewage is 500mg L–1 due to using seawater to flush the toilet.
Sulfate not only has a corrosive effect on the concrete structure of buildings at high concentrations [2],
but also accelerates the release of phosphorus in riverbeds and aggravates the water eutrophication [3].
In the anoxic condition, the sulfide produced from sulfate reduction would cause black odor, and
damage the health of plants, animals as well as humans [4]. Therefore, the removal of sulfate in
sewage has attracted more and more attention.
Anaerobic biological treatment is widely used for treating sulfate-containing wastewater at present.
The main forms include anaerobic filter, anaerobic sequencing batch sludge process (ASBR), up-flow
anaerobic sludge bed (UASB), and so on. Sulfate reduction is the main sulfate removal pathway in
these technologies. However, sulfide will usually be produced during sulfate reduction, which
accounts for about 70%–93% of the removed sulfate [5, 6]. Therefore, it is necessary to combine with
the biological or chemical oxidation of sulfide to achieve sewage discharge standard [7, 8]. This
greatly increases the construction and operation costs.
Constructed wetlands (CWs) are widely used for treating wastewater due to their low
implementation costs and simple operation [9]. It has been found that the anoxic condition in CWs can
provide favorable conditions for sulfate reduction [10]. Wiessner et al. [11] found that the sulfate
removal efficiency in CWs was up to 90%. Simultaneously, the CWs can effectively control sulfide
Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
GBEM IOP Publishing
IOP Conf. Series: Earth and Environmental Science 186 (2018)
1234567890 ‘’“” 012078 doi:10.1088/1755-1315/186/3/012078

discharge. Wiessner et al. [11] also found that sulfide in the effluent of CWs was 8%–20% of the
removed sulfate. While in the study of Chen et al. [12], the ratio was only 1%–4%. Therefore, using
CWs to treat sulfate-containing wastewater can make up the defects of the traditional anaerobic
biological treatment and save the costs. However, the mechanism of simultaneous removal of sulfate
and sulfide in CWs has not been completely elucidated to date, and further studies are still needed.
Moreover, it has been shown that sulfate load could affect sulfate reduction rate, and therefore affect
the sulfate removal and sulfide discharge [13]. This indicated that the sulfate load is an important
parameter for the performance of CWs in treating sulfate-containing wastewater. Hence, it is
necessary to explore the influence mechanism of sulfate load on sulfate and sulfide removal in CWs.
Therefore, the objective of this study were to quantitatively investigate the mechanism of sulfate
and sulfide removal based on the sulfur analysis, and elucidate the effects of sulfate load on sulfate
and sulfide removal in CWs.

2. Materials and Methods

2.1. Experimental design and operation

Four sequencing batch CWs microcosms (length: 0.3 m, width: 0.3 m, height: 0.5 m) with a pore
volume of 12 L, were set up in this study. Each of them was filled with gravel ( 8‒13 mm, porosity:
0.4, height: 0.4 m), and planted with cattail (Typha latifolia) (density: 22 plants m–2). The wetland
microcosms were located in an air-conditioned greenhouse environment (25 ± 2 ºC).

2.2. Batch experiment

In order to cultivate the plant shoots and microorganisms, the microcosms were operated in batches
over a 30-day pre-incubation period. The batch experiment began after this pre-incubation period and
lasted for 22 batches. Each batch of the pre-incubation and batch experiment lasted for 5 d.
The secondary effluent was collected from a neighboring wastewater treatment plant and its
characteristics are shown in table 1. Concentrated glucose and starch (at a chemical oxygen demand
(COD) ratio of 1:1), potassium nitrate, and sodium sulfate solutions were added to achieve the target
COD, nitrate, and sulfate concentrations for the influent used in the pre-incubation and batch
experiment (table 1).
Table 1. Characteristics of the WWTP secondary effluent and the microcosm influenta.
Concentrationa (mg L‒1)
Parameters COD NO3‒-N NO2‒-N NH4+-N TN SO42‒-S TS
WWTP secondary effluent
19.8 ± 2.2 9.8 ± 1.3 0.9 ± 0.3 1.7 ± 0.3 14.5 ± 2.5 20.5 ± 2.1 21.6 ± 1.5
Microcosm influent for the pre-incubation and batch experiment
W1 421.5 ± 23.6 43.5 ± 2.9 1.2 ± 0.5 1.8 ± 0.3 14.5 ± 3.5 21.3 ± 0.7 21.1 ± 3.5
W2 430.8 ± 32.5 43.6 ± 3.0 0.9 ± 0.3 1.5 ± 0.4 15.0 ± 2.5 41.6 ± 1.5 42.4 ± 5.3
W3 423.2 ± 35.2 43.9 ± 2.6 1.0 ± 0.5 1.8 ± 0.4 14.7 ± 3.5 62.2 ± 1.4 62.9 ± 5.4
W4 419.5 ± 25.5 43.5 ± 2.6 1.0 ± 0.6 1.7 ± 0.4 14.7 ± 3.5 105.2 ± 1.8 107.8 ± 6.5
a
Values are given as mean ± standard deviation (n = 22).

2.3. Analysis of Aqueous Samples

Aqueous samples were taken with a 10 mL syringe at 0, 0.5, 1, 1.5, 2, 4, 8, 12, 22, 28, 36, 46, 52, 60,
70, 76, 84, 96, 100, 108, and 120 h. Sulfide was immediately analysed via the methylene blue method
after the water sample was filtered through 0.45 μm membrane filters. Sulfate, sulfite, thiosulfate, and
nitrate concentrations were determined by a DX ICS–3000 ion chromatography unit (Dionex

2
GBEM IOP Publishing
IOP Conf. Series: Earth and Environmental Science 186 (2018)
1234567890 ‘’“” 012078 doi:10.1088/1755-1315/186/3/012078

Corporation, Sunnyvale, CA, USA). Elemental sulfur in the pore water was determined by extracting
samples with chloroform and analyzing by a HPLC (Agilent 1200, Agilent Technologies, Santa Clara,
CA, USA) [14]. COD was determined by using standard method [15].

2.4. Analysis of Gaseous and Solid Samples

The volatilization rate of hydrogen sulfide was determined by a passive sampler (RAD 170, Supelco,
Bellefonte, PA, USA). Before and after the batch experiment, aboveground and belowground plants
were harvested and separated, followed by washing, drying, milling, and filtering by a mesh filter (200
meshes). Sulfur content in the plant was analyzed with an elemental analyzer (Vario EL III, Elementar,
Hanau, Germany). 100g gravel was evenly weighed from the different depths of each microcosm (10,
20, 30, and 40 cm). After crushing, the gravel powder was used for measuring the form of deposited
sulfur. AVS was extracted by cold diffusion absorption [16]. Acetone washing was used to extract
residual sediment from AVS (5000 rpm centrifuge 5 min). 20 mL acetone was added to the remaining
sediments. Then sealing and shaking the mixture for 24 h [17]. The precipitate of acetone centrifuge
was used to determine pyrite-sulfur. The supernatant was transferred to the 500 mL wide mouth bottle,
sealed, and evaporated to dry for determining the elemental sulfur. Both pyrite-sulfur and elemental
sulfur were determined by Cr-reduction [18].

3. Results and discussion

3.1. Sulfate and sulfide in CWs pore water

As shown in table 2, the sulfate removal rate were1.42, 2.24, 2.53, and 3.16 g S m–3 d–1 in W1–W4,
respectively, which was within the range reported by Moosa et al. [13]. Only W1 achieved complete
sulfate removal, and the sulfate removal efficiency of W2–W4 were 80.9%, 61%, and 45.1%,
respectively. As shown in figure 1(a), the sulfate concentration of W1–W4 was firstly increased from
21.3, 41.6, 62.2, and 105.2 mg S L–1 to 26.9, 47.5, 71.4 and 117.6 mg S L–1, and then decreased to 0,
7.93, 24.3, and 57.8 mg S L–1, respectively. The 0–8 h period in this study was the nitrate-removal
period (data not shown). Previous studies have shown that the nitrate can inhibit the sulfate reduction,
meanwhile, it can also oxidize the reduced deposited sulfur in the gravel to sulfate [12], and this
explained the reason for the increase of sulfate concentration during the initial 8 h. In this study,
nitrate was completely removed at 8 h, which relieved its inhibition on sulfate removal, so the sulfate
concentration decreased during the 8–120 h period.
Table 2. Sulfate removal rate and efficiency in the microcosms
Microcosm Removal ratea(g S m–3 d–1) Efficiency (%)
W1 1.42 100.0
W2 2.24 80.9
W3 2.53 61.0
W4 3.16 45.1
a
The sulfate removal rate was calculated as follows: SO4 removal= (Cin – Cout) Vw V‒1 HRT‒1, where
2‒

SO42‒ removal is sulfate removal rate (g S m–3 d–1); Cin, Cout are the sulfate concentrations in the influent
and effluent (mg S L–1), respectively; VW is the working volume (L); V is the gravel volume (m3);
HRT is the retention time for every batch (d).
As the final product of sulfate reduction, sulfide is also a dominant sulfur compound in the pore
water. As shown in figure 1(b), sulfide could only be detected when sulfate started removal (8 h). The
sulfide concentrations in the W1–W4 firstly reached 3.94, 8.24, 10.5, and 24.7 mg S L–1, at 60, 94, 94,
and 94 h respectively, and then decreased slowly to 1.18, 6.37, 8.36, and 22.3 mg S L–1, respectively.
The decrease in sulfide concentration was due to that its transformation rate outpaced its production
rate. Further analysis showed that the sulfide discharge in each microcosms accounted for 5.53%–49.8%

3
GBEM IOP Publishing
IOP Conf. Series: Earth and Environmental Science 186 (2018)
1234567890 ‘’“” 012078 doi:10.1088/1755-1315/186/3/012078

of the removed sulfate, which meant that the sulfide produced from sulfate reduction could be further
transformed. This indicated that the CW microcosms in this study had a significant sulfide-removal
capacity. The potential reasons included: (1) hydrogen sulfide emission; (2) plant uptake; and (3) the
deposition of sulfur compounds, including calcium sulfate, metal sulfide, and elemental sulfur [11, 19,
20]. In this study, the amount of hydrogen sulfide gas was below the detection limit, which indicated
that hydrogen sulfide emission was negligible. The plant uptake rate for sulfur (0.03–0.04 g S m–3 d–1)
was two orders of magnitude lower than the sulfate removal rate, which meant that plant uptake had a
marginal effect. In addition, the calcium ion content in pore water was too low (< 10 mg L–1) to form a
large amount of calcium sulfate. The gravel of this research contained high iron content (~5.5%),
which could lead to the formation of metal sulfide and elemental sulfur [21]. Therefore, the formation
and deposition of elemental sulfur and metal sulfide might be the main pathway for sulfide removal.
70
W1 (a) W1 (b) W1

Deposited sulfur content (mg S kg-1gravel)

120
W2 W2 25 60 W2
W3 W3 W3
100 W4 W4 50 W4
20
80
SO42－ (mg S L－1)

S2－ (mg S L－1)

40
15
60
30
10
40
20

5
20
10

0 0
0
Elemental sulfur AVS Pyrite-sulfur
0 8 40 80 120 0 8 40 80 120
Time (h) Time (h)

Figure 1. Concentrations of sulfate (a) and sulfide Figure 2. The deposited sulfur contents in the
(b) in the microcosms. Error bars represent ± microcosms. Error bars represent ± standard
standard error of the mean (n = 22). error of the mean (n = 3).

3.2. Deposited sulfur

Previous studies have shown that elemental sulfur, acid volatile sulfide (AVS), and pyrite-sulfur are
the main forms of deposited sulfur in marine and lacustrine sediments [22]. As shown in figure 2, all
the three forms of deposited sulfur accumulated during the batch test. Among them, the accumulation
of elemental sulfur was the lowest, which was 16.6–36.2 mg S kg–1 gravel, and the accumulation of
AVS and pyrite were higher, which were 22.3–36.0 and 49.7–63.6 mg S kg–1 gravel, respectively.
Elemental sulfur is an important product of sulfide oxidation [23]. The presence of elemental sulfur
indicated that the occurrence of sulfide oxidation. Previous study has shown that, in natural sediments,
iron(III) is a ubiquitous electron acceptor for sulfide oxidation [22] (equation (1)):
1
2Fe(OH)3(s) + HS– + 5H+ →8S8(s) + 2Fe2+ + 6H2O (1)
As mentioned above, the wetland matrix contained abundant iron, so it was inferred that the
accumulation of elemental sulfur was the result of sulfide oxidation by iron(III). According to
previous study [24], the rate of the above-mentioned sulfide oxidation was positively correlated with
the sulfide concentration. Therefore, as the sulfide concentration increased (figure 1(b)), the
accumulation of elemental sulfur increased from 16.6 to 36.2 mg S kg–1 gravel.
AVS is produced from the reaction between ferrous ion and sulfide [25]. The ferrous ion can be
produced from the chemical iron(III) reduction by dissolved sulfide (equation (1)). Besides, the
biological reduction of iron(III) by elemental sulfur with the participation of sulfur oxidizing bacteria
(SOB) is also a potential process(equation (2)) [26]. The electron transfer during the latter process is
three times higher than the former, and thus should be the main source of ferrous ion:

4
GBEM IOP Publishing
IOP Conf. Series: Earth and Environmental Science 186 (2018)
1234567890 ‘’“” 012078 doi:10.1088/1755-1315/186/3/012078

1 2– 2+ –
8S8(s) + 6Fe(OH)3(s) → SO4 + 6Fe + 18OH (2)
Obviously, the amounts of dissolved sulfide and elemental sulfur should possess effects on the
production of ferrous ion, and thus affect the accumulation of AVS. As shown in figure 2, the AVS
accumulation in W1 was the lowest (22.3 mg S kg–1 gravel). This was because both the amounts of
sulfide and elemental sulfur in W1 were the lowest (figures 1(b) and 2), and could reduce limited
iron(III) to produce ferrous ion, which limited the accumulation of AVS. However, with the increase
in the amounts of sulfide and elemental sulfur in W2–W4, the accumulation of AVS decreased from
36.0 to 27.1 mg S kg–1 gravel. The cause might be that the gradual increased sulfide concentration
inhibited the activity of SOB [27]. This lowered the iron(III) reduction rate by elemental sulfur, and
negatively affected ferrous ion production, and therefore, the AVS accumulation was decreased.
Under anoxic condition, pyrite is mainly produced by the reaction of AVS and sulfide (equation (3))
[21]:
FeS(s) + H2S→ FeS2(s) + H2 (g) (3)
Burton et al. [21] found that the pyrite formation rate was positively correlated with the content of
AVS and the dissolved sulfide. The accumulation of pyrite-sulfur was the lowest in W1 (49.7 mg S
kg–1 gravel), and that in W2–W4 showed a downtrend from 63.6 to 56.1 mg S kg–1 gravel. It could be
note that the relationship of pyrite accumulation among microcosms was the same as that of AVS
accumulation. This indicated that the AVS content should be the primary control of pyrite
accumulation in this study.
As mentioned above, the sulfide produced from sulfate reduction in the CWs could be further
converted into deposited sulfur and removed from the pore water, through the following pathways in
this study: (1) the oxidation to elemental sulfur by iron(III); (2) reaction with ferrous ions to produce
AVS; (3) reaction with AVS to produce pyrite-sulfur.

3.3. Sulfur balance

As shown in figure 3, 5.55%–46.9%, 1.10%–4.01% and 2.48%–7.27% of the removed sulfate were
discharged from the microcosms in the forms of sulfide, elemental sulfur, and thiosulfate, respectively.
13.0%–15.6%, 9.75%–18.3% and 20.2%–39.9% were converted to elemental sulfur, AVS and pyrite,
and deposited in the gravel. The absorption of sulfate by plants was very limited, only accounting for
0.84%–2.34% of removed sulfate. The sulfur balance showed that the sulfide in the effluent in this
study was less than 50% of the removed sulfate. It showed that the CWs could effectively control the
sulfide discharge while removing sulfate. The gravel of wetland played an important role, which could
convert the sulfide into the deposited sulfur like elemental sulfur, AVS, and pyrite-sulfur through
physical, chemical, and biological processes (e.g. oxidation and precipitation).

5
GBEM IOP Publishing
IOP Conf. Series: Earth and Environmental Science 186 (2018)
1234567890 ‘’“” 012078 doi:10.1088/1755-1315/186/3/012078

100

50

Proportion of S2－ to SO42－ removal (%)

80
Proportion of sulfate removal (%)

40

60
30

40
20
y = -3.938 + 7.116x
20 R2=0.983
10

0
W1 W2 W3 W4 0
Sulfide Suspended elemental sulfur 0 1 2 3 4 5 6 7 8
Thiosulfate-sulfur AVS Sulfate load (g S m－3d-1)
Pyrite-sulfur Deposited elmental sulfur
Plant uptake-S Other-S
Figure 4. The relationship between the proportion
Figure 3. The proportions of various sulfate of discharged sulfide to removed sulfate-sulfur
removal routes. and the sulfate load.

3.4. Engineering Significance

It is an urgent problem that how to efficiently and economically remove sulfate from wastewater and
control sulfide discharge in engineering. The result showed that, as a sewage treatment technology
with good treatment effect and low operating cost, CWs can achieve effective sulfate removal, and
meanwhile reduce the sulfide discharge via the matrix. This makes up for the shortage of traditional
anaerobic biological sulfate wastewater treatment. However, the study also found that as the sulfate
load increased from 1.42 to 7.01 g S m–3 d–1, the proportion of discharged sulfide in the removed
sulfate increased from 5.55% to 46.9%. Figure 4 showed that there was a significant positive
correlation between this proportion and sulfate load (R2 = 0.983). This meant that when the sulfate
load is high, the CWs will not be able to effectively remove sulfide. As mentioned above, the iron in
the wetland matrix played a decisive role in controlling sulfide discharge. A reasonable iron content
could promote the formation of deposited sulfur, and then achieve efficient sulfide removal. In order
to provide theoretical basis and technical support for the treatment of sulfate-containing wastewater
with CWs, it is necessary to further investigate the iron cycle and its interaction with carbon and sulfur
cycles in CWs.

4. Conclusion
As the sulfate load increased from 1.42 to 7.01 g S m–3 d–1, the sulfate removal rate increased from
1.42 to 3.16 g S m–3 d–1, and the sulfide discharge rate increased from 0.08 to 1.46 g S m–3 d–1. The
produced sulfide proportion accounted for 5.55%–46.9% of the removed sulfate, indicating that CWs
can effectively reduce sulfide while removing sulfate.
Wetland matrix can convert sulfide into deposited sulfur, which was a main sink for sulfide
removal. Elemental sulfur, acid volatile sulfide (AVS), and pyrite-sulfur were the main forms of
deposited sulfur in this study, with the accumulation of 16.6–36.2, 22.3–36.0, and 49.7–63.6 mg S kg–1
gravel.
Sulfur balance analysis showed that 42.9%–71.1% of the sulfate removal was deposited in the
gravel, and only 0.84%–2.34% was absorbed by the plant.

Acknowledgments
This work was financially supported by the National Science Foundation of China (51578393).

6
GBEM IOP Publishing
IOP Conf. Series: Earth and Environmental Science 186 (2018)
1234567890 ‘’“” 012078 doi:10.1088/1755-1315/186/3/012078

References
[1] Zhang L, De S P, De G B, De M W, Boon N, and Verstraete W 2008 Water Res. 42 1-12.
[2] Taylor H F W 1966. Chemistry of cements.
[3] Chen M, Li X H, He Y H, Song N, Cai H Y, Wang C, Li Y T, Chu H Y, Krumholz L R, and
Jiang H L 2016 Water Res. 96 94-104.
[4] Simkin S M, Bedford B L, and Weathers K C 2013 Ecosystems 16 1118-29.
[5] Christensen B, Laake M, and Lien T 1996 Water Res. 30 1617-24.
[6] Wang J, Lu H, Chen G H, Lau G N, Tsang W L, and Van Loosdrecht M C M 2009 Water Res. 43
2363-72.
[7] Kobayashi H A, Stenstrom M, and Mah R A 1983 Water Res. 17 579-87.
[8] Wolin M J and Miller T L 1980 Arch. Microbiol. 124 137-42.
[9] Jasper J T, Jones Z L, Sharp J O, and Sedlak D L 2014 Environ. Sci. Technol. 48 5136-44.
[10] Kröpfelová L 2008 Wastewater Treatment in Constructed Wetlands with Horizontal Sub-Surface
Flow: Springer Netherlands. pp 209-27.
[11] Wiessner A, Rahman K Z, Kuschk P, Kastner M, and Jechorek M 2010 Water Res. 44 6175-85.
[12] Chen Y, Wen Y, Zhou Q, Huang J, Vymazal J, and Kuschk P 2016 Water Res. 102 572-81.
[13] Moosa S, Nemati M, and Harrison S T L 2002 Chem. Eng. Sci. 57 2773-80.
[14] Rethmeier J, Rabenstein A, Langer M, and Fischer U 1997 J. Chromatogr. 760 295-302.
[15] APHA 2005 Standard methods for the examination of water and wastewater. Health Laboratory
Science.
[16] Hsieh Y P, Chung S W, Tsau Y J, and Sue C T 2002 Chem. Geol. 182 195-201.
[17] Wieder R K, Lang G E, and Granus V A 1985 Limnology & Oceanography 30 1109-15.
[18] Sullivan L A, Bush R T, and Mcconchie D M 2000 Soil Research 38 729-34.
[19] Baldwin D S and Mitchell A 2012 Water Res. 46 965-74.
[20] Rickard D and Morse J W 2005 Mar. Chem. 97 141-97.
[21] Burton E D, Bush R T, and Sullivan L A 2006 Environ. Sci. Technol. 40 888-93.
[22] Burton E D, Bush R T, Sullivan L A, Hocking R K, Mitchell D R G, Johnston S G, Fitzpatrick R
W, Raven M, Mcclure S, and Jang L Y 2009 Environ. Sci. Technol. 43 3128-34.
[23] Morse J W and Cornwell J C 1987 Mar. Chem. 22 55-69.
[24] Poulton S W 2003 Chem. Geol. 202 79-94.
[25] Chen Y, Wen Y, Zhou J, Tang Z, Li L, Zhou Q, and Vymazal J 2014 Water Res. 59 1-10.
[26] Bo T 2000 Adv. Microb. Ecol. 16 41-84.
[27] Mora M, López L R, Lafuente J, Pérez J, Kleerebezem R, Van Loosdrecht M C, Gamisans X, and
Gabriel D 2015 Water Res. 89 282-92.