Single-Stage Deammonification Process Performance - MBBR Versus IFAS Configurations

Hong Zhao1, Romain Lemaire2, Magnus Christensson3, Glenn Thesing1, Frederic Veuillet4, Juan
Ochoa4, Daniel Lamarre5 and Alain Gadbois5
1
Kruger Inc., 4001 Weston Parkway, Cary, NC 27513
2
Veolia Water Technical Department, 1 rue Giovanni Battista Pirelli, 94417 St-Maurice, France
3
AnoxKaldnes AB, Klosterängsvägen 11A, 226 47 Lund, Sweden
4
Veolia Environment Research and Innovation, Chemin de la Digue – BP 76, 78603 Maison
Laffitte Cedex, France
5
John Meunier Inc. 4105 Sartelon, St.-Laurent, Quebec H4S 2B3

ABSTRACT

ANITA™Mox is a Veolia process using moving bed biofilm reactor (MBBR) technology tested
and validated in full-scale plants for energy- and cost-effective autotrophic nitrogen removal
from Sidestream effluent using anammox bacteria. In order to increase the ANITA™Mox
process performances under different operating conditions, substrate transport and accessibility
inside the biofilm must be enhanced. Numerous studies are currently investigating hybrid
systems using combined suspended cultures and biofilm called Integrated Fixed-Film Activated
Sludge (IFAS) for wastewater treatment. In this work, (i) two lab-scale biofilm ANITA™Mox
reactors were operated using the two configurations (IFAS and MBBR), (ii) the distribution of
the Anammox (AnAOB) and Ammonia Oxidiser Bacteria (AOB) in the suspended sludge and
the biofilm was characterized using molecular tools (qPCR), and (iii) a full scale demonstration
with the IFAS configuration was conducted in Sjölunda WWTP at Malmo, Sweden.

The lab study and the full-scale demonstration showed that in IFAS configuration,
ANITA™Mox process achieved very high N-removal rate (up to 8 gN/m².d), which was 2-4
times higher than that achieved in the pure MBBR mode. The high concentration of suspended
solids (MLSS) in the bulk obtained within the IFAS mode induces a very efficient bacterial
distribution between AOB and AnAOB population. AnAOB activity mainly occurs in the
biofilm (96% of total AnAOB in the reactor), whereas Nitritation by AOB mostly takes place in
the suspended phase (93% of total AOB). This spatial distribution observed in the IFAS reactor
results from a natural selection due to more easily substrate accessibility for AOB in the bulk
(NH4, O2) creating higher nitrite concentration in the bulk liquid compare to pure MBBR mode.
The efficient control of MLSS level in the IFAS reactor is a key parameter to enhance the nitrite
production by AOB and increase the substrate availability in the AnAOB-enriched biofilm
leading to higher N-removal rate. These promising results obtained using IFAS configuration
open new roads for the widespread application of very compact and robust ANITATMMox
process for Sidestream but also Mainstream cost-effective N-removal.

KEYWORDS
Anammox, ANITA Mox, IFAS, MBBR, Nitrogen removal, Sidestream treatment
INTRODUCTION

ANITA™Mox is a cost- and energy-effective 1-stage deammonification MBBR process

developed for autotrophic nitrogen removal (i.e. anammox) and recently implemented at full-
scale in four plants in Europe (Sjölunda, Växjö - Sweden; Holbæk, Grindsted - Denmark) with
two more under-construction in the US (James River, Virginia and City of Durham, North
Carolina) for sidestream treatment of N-rich effluent (Christensson et al., in Press). The high
NH4-removal rate achieved (1.2 kgN/m3.d) is explained by synergetic interaction in biofilm-
carrier between Anammox bacteria (internal biofilm layers) and Ammonia Oxidising Bacteria
(AOB) (external layer). This microbial synergy is strongly influenced by transport limitations
and substrate availability inside the biofilm which are dependent on different environmental
factors such as morphology, density and thickness of the biofilm, temperature, substrate
concentration (NO2-, NH4+, O2) and shear stress. Therefore, in order to improve the
ANITA™Mox process performances under different operating conditions (e.g. Mainstream and
Sidestream application), substrate transport must be enhanced.

Several studies investigated the effect of combining suspended cultures and fixed biomass into
one system called Integrated Fixed-Film Activated Sludge (IFAS) for municipal (Al-Sharekh and
Hamoda, 2001; Ødegaard et al., 2000) and industrial (Wessman et al., 2004) wastewater
treatment. More specifically, Paul et al., (2006) have reported that a clear spatial distribution of
microbial population between floccular biomass (heterotrophs) and fixed biomass (nitrifiers)
leads to higher COD and N-removal performances.

The objectives of this paper are, (i) to demonstrate the advantages of IFAS configuration for the
ANITA™Mox process using comparative lab-scale MBBR and IFAS reactors and full scale
plant, (ii) to study the impact of critical operating parameters on the process performance, and
(iii) to determine the distribution of the Anammox and AOB bacteria between suspended sludge
and biofilm-carrier using molecular tools.

MATERIAL AND METHODS

Lab-Scale Experiments
Experimental Setup. Two 7 liter lab-scale reactors were operated in parallel to compare
ANITA™Mox process performances under pure MBBR (R1) and IFAS (R2) configuration with
a settling tank of 1.9 liter for sludge recycling (Figure 1A). Each reactor was equipped with
sensors for measuring: Temperature, pH with InPro 3250 (Mettler Toledo), Dissolved oxygen
(DO) with InPro 6050 (Mettler Toledo), NH4-N and NO3-N with Varion+® 700 IQ (WTW). The
temperature was controlled at 30°C ± 0.5 via jackets of the double-walled reactors, the pH
maintained between 7 to 8 by acid and base dosing, NH4-N in the treated effluent controlled
between 10-150 mgN/L and DO kept in the range of 0.1 to 1 mgO2/L. In both reactors aeration
was supplied by diffusion through small holes in a plastic tube at an air flow rate of 0-5 L/min.
Mixing of liquid and carriers in the two reactors was ensured by air flow and also additional
nitrogen gas flow (2-4 L/min). This hydrodynamic condition was sufficient to have a good
mixing of the carriers in both reactors. The suspended sludge SRT in R2 (IFAS) was maintained
around 5 days ± 2 through manual wasting of excess sludge.
Figure 1 - (A) IFAS and MBBR lab-scale reactors. (B) Anammox seeded K5 AnoxKaldnes
media used as inoculums.

In order to control the ammonia concentration in the outlet of the reactors and ensure the same
effluent quality in both reactors (and therefore the same substrate diffusion limitation in the
biofilm), a simple regulation based on the on-line NH4-N sensor was used. If the NH4-N level in
the reactor was higher than the desired setpoint, the feeding pump was switched off until the
NH4-N level decreased below this setpoint.

Carrier media. K5 plastic carriers from AnoxKaldnes were used for biofilm growth in both
reactors (Figure 1B). K5 carriers have a high protected surface area corresponding to 800 m2/m3.
The carrier filling degree was 43% in both reactors. This filling degree was chosen to guarantee a
good mixing in the relatively narrow lab-reactors but higher filling degree (up to 55%) can be
used in full-scale ANITA™Mox with K5 carriers. As a result of this filling degree the total
protected surface area in each reactor was 2.41 m2.

Influent. Influent used in this study was reject water from a mesophilic anaerobic sludge digester
collected in a WWTP near Paris. The reject water was sampled once a week and stored for up to
72 hours at 4°C. The mean reject water characteristics are shown in Table 1.

Table 1 - Reject water characteristics used in the lab-scale study.

NH4-N COD BOD TSS Alkalinity
pH
mg/L mg/L mg/L mg/L (mg/L CaCO3)
Concentration 907 ± 200 414 ± 150 29 ± 5 400 ± 200 4200 ± 1000 8 ± 0.5

This sidestream effluent is very high in ammonia and low in COD with a ratio COD/N=0.45.
However, the refractory soluble COD represents in average 70% of the soluble COD meaning
that the soluble biodegradable COD (sbCOD) to N ratio was around 0.25-0.3. PO4 was very low
in the reject water (< 2 mgP/L) and KH2PO4 was added in the feed to be non-limiting with a
minimum of 2.0 mgP-PO4/L in the treated effluent.
Analytical methods. NH4-N, NO2-N, NO3-N, PO4-P, COD and sCOD were measured using a
colorimetric micro method with kits Dr Lange (HACH). Samples were filtered at 0.45µm for
soluble compounds analysis. Alkalinity was determined by titration with a standardised sulphuric
acid solution (0.02N). Mixed Liquor Suspended Solids (MLSS) and the Mixed Liquor Volatile
Suspended Solids (VSS) were measured according to Standard Methods.

Biofilm characterisation. In order to monitor the biofilm development, two K5 carriers were
taken out each week from IFAS and MBBR reactors and placed in the water to take photographs
using stereo microscope.

Sample collection and DNA extraction. Carriers samples were collected and immediately frozen
in liquid nitrogen and stored at -80°C before DNA extraction. The genomic DNA was extracted
from carrier biofilm using Fast DNA spin kit for Soil (MP Biomedicals, IllKirch, France)
following the manufacturer’s instructions.

Quantitative PCR. Quantification of AnAOB, AOB and total bacteria was performed with
primers described in Table 2 using Sso-Fast™ EVAGreen® (Bio-Rad, USA) containing
nonspecific fluorophore EVAgreen. PCR was performed in 10 µl reaction mixtures with primers
at a final concentration of 0.4 µM each. Between 5 and 20 ng of sample DNA were added to the
mixture. Quantitative PCR were run with the following protocol: 3min at 95°C to activate DNA
polymerase followed by 40 cycles of denaturation at 95°C for 10s and annealing/extension at
57°C for 30s with fluorescence capture at the end of this step. Melt curve analysis was performed
after the final cycle of extension by increasing temperature from 70 to 95°C in 0.25°C
increments every 5s. Fluorescence was measured 5s after each temperature increment.
EVAGreen qPCR was performed in triplicate on each serial dilution of plasmid solution and in
duplicate on each DNA sample.

Table 2 - PCR Primers used for qPCR quantification.

Primer & Target Tm
Sequence (5’→3’)a Reference
probe name organism (°C)
V3F Total CCTACGGGAGGCAGCAG 60 Muyzer (1993)
Bacteria
V3R (16S) ATTACCGCGGCTGCTGG 57.6
HzoF AnAOB CATGGTCAATTGAAAGRCCACC 59.3 Park (2009)
HzoR (hzo) GCCATCGACATACCCATACTS 59.8
amoA-1F AOB GGGGTTTCTACTGGTGGT 56 Rottehauwe
amoA-2R (amoA) CCCCTCKGSAAAGCCTTCTTC 62.8 (1997)

F1 nxrA NOB CAGACCGACGTGTGCGAAAG Poly (2008)

(nxrA)
R2 nxrA TCCACAAGGAACGGAAGGTC
a
M = A or C; R = A or G; K = G or T; S = G or C

Standard curve were generated by using dilution series of plasmid solutions ranging from
approximately 1x108 to 1x102 copies per reaction. A plasmid containing fragment of 16S RNA
gene from Escherichia coli was used for quantification of total bacteria. A plasmid containing
partial sequence of hzo gene from Brocadia anammoxidans was used for quantification of
anaerobic ammonia oxidising bacteria (AnAOB). A plasmid containing partial amoA gene from
Nitrosomonas europea was used for ammonia oxidising bacteria (AOB) quantification. A
plasmid containing partial nxrA gene from Nitrobacter was used for nitrite oxidising bacteria
(NOB) quantification. All plasmids were synthesized by Eurofins MWG Operon (Ebersberg,
Germany). PCR reactions were run on a CFX 96 thermocycler (Bio-Rad, USA)

Operational conditions. IFAS and MBBR were started with pre-colonized K5 carriers coming
from a full-scale ANITA™Mox plant in Sjölunda, Sweden. Colonized media were collected and
stored in effluent at 4°C before introduction into the reactors. To prevent unwanted biofilm
formation on the reactor wall surfaces, walls were manually cleaned two times a week. A
summary of the experiments is provided in Table 3.

Table 3 - Operational conditions in lab-scale study.

Stage DO Temperature N-NH4out
Reactor pH
(d) (mg/L) °C (mgN/L)
I R1 : MBBR
0.8 – 1 30 7.5±3 150±50
0 – 47 R2 : MBBR
II R1 : MBBR 0.4 – 0.8
30 7.5±3 150±50
48 – 177 R2 : IFAS 0.1 – 0.2
III R1 :MBBR 0.8
30 7.5±3 70±10
178 – 212 R2 : IFAS 0.1 – 0.2
IV R1 : MBBR 0.6
30 7.5±3 30±5
213 – 330 R2 : IFAS 0.1 – 0.2
V R1 : MBBR 0.6
30 7.5±3 10±2
331 – 388 R2 : IFAS 0.1 – 0.25

For each operational period, nitrogen and COD mass balances were determined from water
samples collected daily. The following stoichiometric equation for AnAOB activity was used:

NH4+ + 1.32NO2- + 0.066HCO3- + 0.13H+

 1.02N2 + 0.26NO3- + 0.066CH2O0.5N0.15 + 2.03H2O (1)

Full-Scale IFAS Demonstration

The full-scale demonstration of IFAS configuration was conducted in the full scale
ANITA™Mox plant at Sjölunda WWTP (550,000 PE), Malmo, Sweden. The full scale plant
consists of four separate 50 m3 insulated and covered fibre glass cylindrical tanks, which are 6
meters high with a water depth of 5.5 meters and a diameter of 3 meters. All four tanks were
operated in MBBR configuration. The main design and operating parameters and the reject water
composition were detailed by Christensson et al. (2011).

One of the four MBBR tanks was converted to an IFAS configuration by installing a conical, in-
tank, clarifier for sludge retention as shown in Figure 2 between Jan-Feb of 2013. The clarifier
had to be inside of the existing MBBR due to space constraint on site. The clarifier has a surface
area of 5.0 m2 and a total volume of 7.0 m3. K5 carriers (also used in the lab-scale study) with
carrier filling degree of about 50% were used in the demonstration. Compared to typical
operation for MBBR ANITATMMOX, the IFAS rector was operated at lower DO (0.2 to 0.6 mg-
O2/L versus 1.0 to 1.3 mg-O2/L) and at higher MLSS level (2,000 to 4,000 mg/L versus 20 to
400 mg/L).

Figure 2 - Retrofit of Existing MBBR ANITATMMox to IFAS Configuration.

RESULTS AND DISCUSSION

The results for comparison of the MBBR and IFAS reactors (R1 and R2) at lab-scale are
presented with emphasis on (i) ammonia, nitrite and TSS influence on N-removal efficiencies,
(ii) biofilm morphology and (iii) the AnAOB and AOB population distribution between
suspended and fixed biomass.

Reactors (R1 and R2) were operated for 388 days using real anaerobic sludge digester centrate
during which four different NH4 concentrations in the treated water were tested. As shown in
Table 3, the experiments were divided into five stages. During Stage I (Day 0-47) both reactors
were run as pure MBBR and colonised K5 media was introduced in the reactors at Day 0. In this
stage, NH4 concentration in the outlet was controlled at 150±50 mgNH4-N/L while N-removal
performances and biofilm development on carriers stabilised. During Stage II (Day 48-177), R2
operation was shifted to IFAS configuration after the installation of a settler unit to recycle the
suspended sludge. In Stage II, NH4 outlet concentration was maintained at 150±50 mgNH4-N/L
but the applied N-load on each reactor was different and automatically adjusted based on N-
removal performances. During the following Stages, the NH4 outlet concentration was decreased
in both reactors to 70±10 mgNH4-N/L in Stage III, 30±5 mgNH4-N/L in Stage IV and finally
10±2 mgNH4-N/L in Stage V.

Nitrogen removal performances

Figure 3 shows the evolution of the N-removal rate in the MBBR (R1) and IFAS (R2) reactors
and the MLSS in the IFAS during the entire experiments period. During Stage I, N-removal in
both reactors increased from 1.2±0.1 gN/m2.d to 2.0±0.2 gN/m2.d due to some loss of activity
during the carrier transportation and the initial reseeding procedure. Steady-state performances in
both reactors operated as pure MBBR ANITA™Mox were obtained after 50 days using a DO
around 0.8 mgO2/L. On Day 48, R2 was switched to IFAS mode while R1 remained in pure
MBBR mode with no sludge recycling. In Stage II, N-removal rate for R1 remained around
1.9±0.3 gN/m².d (Figure 3) while for R2 (IFAS), the positive effect of the IFAS configuration is
very clear with N-removal rate increasing from 2 to 6-8 gN/m².d between Day 56 and 84 with
MLSS level reaching up to 3 g/L with the use the clarifier to recycle part of the sludge. These
very good performances remained constant for R2 during the stabilized period between Day 86
to 122.

Figure 3 - Comparison of N-removal rate in MBBR (R1) and IFAS (R2) reactors and MLSS in
IFAS (R2). Points A, B, C, D, E, F are identified as perturbations periods in IFAS mode.
In order to evaluate the effect of MLSS on the N-removal rate in IFAS configuration, trials under
lower MLSS level were performed (points A, B, C in Figure 3). The decrease of the suspended
sludge concentration immediately impacted the N-removal rates in IFAS but quickly recovered
to reach back 6-8 gN/m2.d when MLSS was increased back to its previous level (2-3 g/L).
From Day 178 (Stage III) the NH4 outlet concentration was controlled at 70 mgNH4-N/L in both
reactors. The N-removal rates in IFAS stayed constant during Stage III at 8 gN/m2.d with MLSS
increasing slightly to 2-3 g/L. The robustness of the process was confirmed once again when
MLSS was suddenly decreased from 3 to 1.5 g/L before reaching 3 g/L again few days after
(point D on Figure 3) with N-removal quickly reaching back 8gN/m2.d. N-removal for R1
(MBBR) was rather constant at 2gN/m2.d during Stage III.

During Stage IV, NH4 outlet concentration was further reduced to 30 mgNH4-N/L to verify if the
IFAS reactor could maintain a high N-removal rate under lower NH4 concentration in the reactor
bulk liquid. N-removal initially decreased to 4 gN/m2.d with MLSS slightly below 3 g/L but
when higher MLSS level was maintained in R2 between Day 245 and 260, N-removal of 7
gN/m2.d could still be achieved. After Day 260, an unexpected change in the reject water
composition was observed (sudden increase of salinity and possible toxicity likely due to on-
going upgrade work on the WWTP digesters) which perturbed the operation of both reactors for
50 days. The suspended sludge in R2 (mostly constituted of AOB, see later) started to
deflocculate and to be washed out of the clarifier as shown by the clear drop of MLSS in R2
down to 0.8 g/L resulting in a much lower N-removal of 1.3 gN/m2.d (point F in Figure 3). For
R1 (MBBR), N-removal also decreased from 2 gN/m2.d before this pollution event to 0.7
gN/m2.d. The main composition of the collected reject water (i.e. NH4, COD, TSS) did not
change during this period but the overall salinity did. The composition of reject water collected
on site went back to normal after Day 314 (point F in Figure 3) and MLSS in R2 quickly
increased back to 2.6 g/L at the end of Stage IV due to better floc formation and improved sludge
separation in the clarifier. As a result, N-removal rate quickly increased to 5 gN/m2.d in 12 days,
and reached back 7 gN/m2.d after 30 days when MLSS reached 4-5 gMLSS/L. It further
demonstrates that IFAS process is very robust and resilient even after this unexpected event
which lasted 50 days.

In Stage V, NH4 outlet concentration was finally decreased to 10 mgNH4-N/L. For R1, the N-
removal dropped to 1 gN/m2.d likely due to substrate (i.e. NH4) diffusion limitation in the
biofilm which impacted the AnAOB activity. N-removal rate in R2 was proportional to MLSS
concentration in the reactor. The low NH4 concentration in the bulk liquid resulted in slightly
lower N-removal rate (i.e. 4-6 gN/m2.d) but the effect of NH4 diffusion limitation into the
biofilm was less visible than for pure MBBR operation. The NH4 bulk liquid concentration under
which N-removal rate is impacted due to substrate diffusion limitation in the biofilm is related to
the boundary layer thickness between the liquid and biofilm phases which is dependent of the
reactor hydrodynamics condition.

Effect of nitrite concentration in the bulk liquid

NO2 is a substrate of Anammox bacteria (Eq.1) but NO2 accumulation can also inhibit their
activity (Fernandez et al. 2012; Lotti et al. 2012). Therefore, NO2 level in the bulk liquid should
be high enough to be able to fully diffuse into the deep layer of the biofilm where the AnAOB
are mostly located but not too high to avoid inhibition of AnAOB. Figure 4 shows the relation
between NO2 level measured in the bulk liquid and the N-removal rate observed in each reactor
during the entire experiment. N-removal in R1 was higher when NO2 level was above 2 mgN-
NO2/L but NO2 level rarely reached more than 4 mgN-NO2/L in the bulk liquid (Figure 4a). This
relation between NO2 level and N-removal rate is also observed for R2 with a very clear
correlation between the two parameters up to 25 mgN-NO2/L (Figure 4b). The more MLSS in R2
(IFAS), the more AOB nitritation activity was obtained leading to higher NO2 level in the bulk
liquid and therefore better diffusion into the deep layer of the biofilm to reach the AnAOB.
It clearly demonstrates that 1-stage deammonification biofilm processes are often limited by the
AOB activity and the mass transfer limitation of substrate (i.e. NO2) towards the AnAOB.
Operating the ANITA™Mox in IFAS mode can therefore improve the process performance by
enhancing the AOB activity and reducing the mass transfer limitation of NO2 into the biofilm.

Figure 4 - Effect of Nitrite concentration on N-removal rate under (A) MBBR and (B) IFAS
modes.

Biofilm Morphology
Biofilm structure was macroscopically observed at different time of the study (Figure 5: Biofilm
morphology – MBBR/IFAS). Initially, seeded carriers coming from Sjölunda ANITA™Mox
full-scale plant were used to start-up both reactors. The biofilm was characterized by an irregular
and thin surface (around 300µm thickness) with the Anammox specific red color. After 147 days
of operation (i.e. 100 days after having switched R2 to IFAS mode), R2 biofilm showed a
double-layer, the inner-one with a red color (400-600 µm) and the outer-one with a brown/grey
color and heterogeneous aspect (200-300 µm thickness). For R1 (MBBR), the same double-layer
morphology was observed but the difference between both layers was less clear. At Day 261,
both biofilm had colonized the entire voids of the plastic carrier, however, the double-layer
structure was still visible in the MBBR with an inner red biofilm of 400-500 µm thickness
(mostly composed of AnAOB). This double-layer structure was not observed in the IFAS biofilm
with AnAOB dominating the overall biofilm surface as characterized by its red color. The initial
observation at Day 0 showed that the biofilm on the seeded carriers coming from the full-scale
plant was thinner than that obtain after a long operation in the lab-scale reactors. The N-removal
rate achieved in both full-scale and lab-scale (R1) pure MBBR mode were very similar so these
morphological differences are likely due to different hydrodynamic conditions in full-scale and
lab-scale systems. Indeed, shear-stress from aeration is more important at full-scale than in the
lab reactors and lead to (i) higher removal of the outer-layer of the biofilm by detachment and
(ii) to a densification of the biofilm basal layer.

Figure 5 - Biofilm morphology: MBBR / IFAS.

Quantification of AnAOB, AOB, NOB and Total Bacteria in the MBBR and IFAS reactors
Bacterial populations found in the biofilm were similar for both reactors (Figure 6). The
abundance of total bacteria as well as microorganisms involved in N-removal stayed very stable
during the entire study. Even the transition towards the IFAS mode in reactor R2 did not induce
major changes in the composition of the bacterial population in the biofilm (Figure 6B). AnAOB
are the most abundant microorganisms composing both biofilm. Considering the average copies
number of 16S rRNA gene per genome (i.e. 4.2 copies), the proportion of AnAOB reached
between 50 and 75% of the total bacteria for both IFAS and MBBR modes. In contrast, NOB
which are well known competitors of AnAOB for their NO2 substrate represented less than 0.1%
of total bacteria in both biofilm.
Figure 6 - Quantification of AnAOB, AOB, NOB and Total Bacteria in the biofilm of the
carriers from (A) MBBR R1 and (B) IFAS R2. Dotted line indicates the switch to IFAS mode in
R2.

After R2 was switched to IFAS mode on Day 47, total bacteria in the liquid phase increased 10
fold between Day 42 and Day 147 (Figure 7B), whereas it stayed relatively stable in the control
pure MBBR reactor R1 (Figure 7A).
Figure 7 - Quantification of AnAOB, AOB, NOB and Total Bacteria in the liquid phase from
(A) MBBR and (B) IFAS reactors. Dotted line represents the switch to IFAS mode in R2.

Concerning the bacteria species involved in N-removal, AOB increased continuously in the
liquid phase between Day 49 and 147, from 4.5E+7 to 1.5E+10 copies of amoA gene per ml of
sample to finally represents more than 10% of the total bacterial population in the liquid phase
against about 0.1% before transition towards IFAS mode (Figure 7B). It is also important to note
that introduction of the IFAS mode did not increase the NOB population in the liquid phase as it
was the case for the AOB (Figure 7B). Furthermore, their proportion in the total biomass
decrease in Reactor R2, meaning that NOB can still be efficiently repressed in IFAS mode due to
the very low DO applied (<0.2 mgO2/L, Table 3) and the control of the suspended sludge SRT in
the reactor.

Regarding AnAOB, their abundance in the liquid phase increased significantly after transition to
IFAS mode due to detachment phenomena but they still remained mainly located in the biofilm.
Indeed, taking into account the liquid volume of the reactor, the protected surface of the carrier
and the filling ratio it was possible to calculate the repartition of the different bacterial
population. Table 4 shows the calculated repartition of AnAOB and AOB in the biofilm and
liquid phase for each reactor on Day 147. The IFAS configuration induces a spatial separation of
AOB and AnAOB in the reactor: AnAOB are located in the biofilm where they are protected
from the low DO and from being washed-out of the system due to longer biofilm SRT whereas
AOB are located in liquid phase in close contact with DO and NH4 with shorter SRT that is
controlled via the WAS stream from the clarifier. This spatial repartition of the metabolic
function, partial nitritation and Anammox reaction, leads to better N-removal performances than
in pure MBBR due to higher nitritation activity in the liquid phase.

NOB were always detected in both R1 and R2 but their activity was not impacting the overall
deammonification process as indicated by the measured ratio of N-NO3 produced : N-NH4
removed that was always lower that 11% (i.e. stoichiometric ratio for deammonification if no
NOB and denitrification activity present). This ratio was close to 6% in R2 (IFAS) reactor during
the test period meaning that some heterotrophic denitrification took place using the small amount
of COD available in the reject water together with hydrolysed particulate organics. The
combination of very low DO in the bulk liquid and shorter SRT for the suspended sludge was
very successful in washing-out NOB from the system while maintaining high AOB activity.

Table 4 - Repartition of AnAOB and AOB (% of total bacteria) in biofilm and liquid phase for
each reactor on Day 147.
AnAOB AOB
Reactor
Liquid Biofilm Liquid Biofilm
R1 (MBBR) 1% 99% 1% 99%
R2 (IFAS) 4% 96% 93% 7%

Dynamic model of spatial bacterial distribution in the MBBR and IFAS mode
Bacterial population distribution and specific role for both pure MBBR and IFAS
deammonification configuration are schematically represented in Figure 8. In the MBBR, the
high proportion of AnAOB and AOB in the biofilm on the carrier and lower microbial growth in
the bulk is the result of two mechanisms: Firstly, the low SRT of the suspended biomass
(SRT=HRT=20-24h) limits the autotrophic biomass growth in the suspended phase due to
continuous wash-out. This condition promotes the growth of autotrohic bacteria (AnAOB and
AOB) in the biofilm. Secondly, the differences of maximum specific growth rates between
AnAOB (0.04-0.08d-1) and AOB (0.8-1.0d-1) can also explain the spatial distribution in the
biofilm. Typically in the ANITA™Mox pure MBBR mode, the AOB are located in the outer
layers of the biofilm to access oxygen, while AnAOB are located in the inner anoxic layer of the
biofilm. This way the AnAOB are protected from oxygen, which is consumed in the external
layers by AOB. Excess NH4 in the bulk liquid and NO2 produced by AOB in the biofilm are
transported to deeper layers of the biofilm by diffusion and by convection in the biofilm voids
and micro-channels.

Figure 8 - Dynamic model of spatial distribution of bacterial populations involved in N-removal

and biofilm structure in pure MBBR versus IFAS ANITA™Mox configurations.
In IFAS configuration, higher SRT for suspended biomass (controlled around 5 days) enhances
the AOB growth as flocculated biomass. Substrate diffusion limitation in flocs is less apparent
than in biofilm which are thicker and denser. This leads to better substrate accessibility for AOB
(oxygen, NH4) in the suspended solids meaning that AOB that pre-existed on the outer-layer of
the biofilm gradually disappear from the biofilm due to a lack of oxygen (i.e. DO in bulk liquid
is lower in IFAS than MBBR) that is now mostly consumed by AOB in the liquid phase. The
biofilm is therefore almost exclusively composed of AnAOB with a very fine top layer of
oxygen scavengers like AOB or normal heterotrophs. The larger AOB population in IFAS
configuration improves the overall flux of NO2 produced for AnAOB but also the residual
concentration in the bulk improving the diffusion of the NO2 through the basal layer of the
biofilm where AnAOB are located and therefore increasing their active fraction in the biofilm. In
IFAS and MBBR configuration, the global amount of AnAOB was similar, but the improvement
of nitrites flux and production in IFAS mode lead to N-removal rate near the maximum values
obtained during fully anoxic AnAOB activity batch test with non-limiting substrate levels (i.e.
NH4 and NO2).

Full-scale demonstration
Figure 9 presents the performance results obtained in the IFAS ANITA Mox full-scale reactor at
Sjölunda WWTP before and after the installation of the clarifier. The clarifier was installed on
Day 960 and a stable performance was achieved after Day 985. During the stable period, average
ammonia removal efficiency was about 95% and average TN removal efficiency was about 85%.
The percentage of nitrate production out of ammonia removed was about 8.5%.
Figure 9 - Full scale performance before and after switching to IFAS configuration: A) nitrogen
and temperature profiles; B) ammonia and nitrogen removal efficiencies and % of nitrate
production. Line indicates the day of switching.

Figure 10 presents the comparison on applied nitrogen surface loading and surface removal rates
before and after switching to the IFAS configuration. As shown, the surface loading and removal
rates were doubled due to the switch in process configuration (for example 5-6 g-N/m2/d in IFAS
compared to 2-3 g-N/m2/d in the MBBR). However, the loading and removal rates were not
stable and were related to the MLSS level in the IFAS reactor.

Due to intensive aeration in the reactor tank below the clarifier and constant foaming problem,
the TSS spikes were observed in the clarifier effluent. As results of the TSS spikes, the MLSS in
the IFAS reactor was built up slowly with a variation between 1,000 to 4,000 mg/L. From Day
1010 (June 2013), the TSS spikes have been minimized, which leads to an increase in MLSS and
an increase in nitrogen surface removal rates.

Despite of the MLSS variation, nitritation was enhanced and taken place even the bulk DO
concentration was between 0.2 to 0.6 mg/L. Compared to the MBBR operation (1 to 3 mg-N/L),
a higher nitrite build-up between 4 to 8 mg-N/L was maintained in the IFAS configuration. The
higher nitrite buildup would lead to an improvement on the anammox process because of less
substrate limiting condition. Due to the improvements on both nitritation and anammox, the
nitrogen surface removal rate in the IFAS configuration has doubled that in the MBBR
configuration. As indicated by the ratio of nitrate production out of ammonia removal, the low
DO condition in the IFAS configuration was able to repress the NOB growth in the suspended
growth under a higher nitrite level.
Figure 10 - Comparison on applied surface loading and surface removal rates before and after
switching to IFAS configuration. Line indicates the day of switching.

CONCLUSIONS

 In IFAS configuration, ANITA™Mox process reached very high N-removal rate of up to 8

gN/m².d. These performances are 3-4 times higher than those achieved in pure MBBR design
with similar high TN and NH4 removal efficiency (90% and 95% respectively).
 IFAS mode induces a spatial differentiation of the bacterial N-removal activity: Anammox
activity mainly occurs in the biofilm (96% of total AnAOB), whereas nitritation mostly takes
place in the suspended phase (93% of total AOB).
 The efficient control of MLSS in the IFAS reactor is a key parameter to enhance the nitrite
production by AOB and increase the substrate availability in the AnAOB-enriched biofilm
leading to higher N-removal rate.
 N-removal performance quickly recovered after perturbation periods confirming the
robustness of the IFAS configuration
 Using the IFAS ANITA™Mox configuration at full-scale seems very promising for
sidestream treatment with N-removal rate up to 3 kgN/m3react.d expected. The easy physical
separation between AOB-rich suspended sludge and AnAOB-rich biofilm carriers through
the use of non-clogging sieve is a clear advantage to secure the retention of AnAOB in the
system.
 These promising results obtained with the IFAS ANITA™Mox process for sidestream
treatment open new attractive solutions for Mainstream N-removal application due to the
easy retrofit of IFAS process into activated sludge systems with existing clarifiers.
 The full-scale demonstration at Sjölunda WWTP has proved that the IFAS configuration has
doubled the nitrogen removal rate compared to the MBBR configuration.
 As observed in the lab-scale study, the full scale study also indicates that the nitrogen
removal rate was related to the MLSS level in the IFAS reactor. The full-scale demonstration
is still ongoing with the modified clarifier to improve biomass retention and nitrogen removal
performance.
REFERENCES

Al-Sharekh, H.A., Hamoda M.F. 2001 Removal of organics from WW using a novel biological
hybrid system. Water Sci. Tech., 43, 1, 321-326.
Christensson M., Ekström S., Andersson Chan A., Le Vaillant E., Lemaire R. in press
Experience from start-ups of the first ANITA Mox plants. Water Sci. Tech.
Christensson M., Ekström S., Lemaire R., Le Vaillant E., Bundgaard E., Chauzy J., Stalhandske
L., Zhao H., Ekenberg M. 2011 ANITATM MOX – A BioFarm solution for fast start-up of
deammonifying MBBRs. WEFTEC 2011, Los Angeles, CA, USA.
Fernandez I., Dosta J., Fajardo C., Campos J.L., Mosquera-Corral A., Mendez R. 2012 Short-
and long-term effects of ammonium and nitrite on the Anammox process. J. of Env. Manag., 95,
170-174.
Lotti T., van der Star W.R., Kleerebezem R., Lubello C., van Loosdrecht M.C. 2012 The effect
of nitrite inhibition on the anammox process. Water Res., 46 (8), 2559-2569.
Muyzer, G., Teske, A., Wirsen, C.O., Jannasch, H.W. 1993 Phylogenetic relationships of
Thiomicrospira species and their identification in deep-sea hydrothermal vent samples by
denaturing gradient gel electrophoresis of 16S rDNA fragments. Appl. and Env. Microb., 59 (3),
695-700.
degaard H., Gisvold B., Strickland J. 2000 The influence of carrier size and shape in the
moving bed biofilm process. Water Sci. Tech., 41, 4-5, 383-391.
Park H., Rosenthal A., Ramalingam K., Fillos J., Chandran K. 2010 Linking community profiles,
gene expression and n-removal in anammox bioreactors treating municipal anaerobic digestion
reject water. Env. Sci. and Tech., 44 (16), 6110-6116.
Paul E., Wolff D.B., Ochoa J., Ribeiro da Costa R.H. 2006 Recycled and virgin plastic carriers in
hybrid reactors for wastewater treatment. Water Environ Res., 79(7), 765–774.
Rotthauwe J.H., Witzel K.P., Liesack W. 1997 The ammonia monooxygenase structural gene
amoA as a functional marker: Molecular fine-scale analysis of natural ammonia-oxidizing
populations. Appl. and Env. Microb., 63 (12), 4704-4712.
Wessman F., Tuegen E.Y., Zheng Q., He G., Welander T., Rusten B. 2004 Increasing the
capacity for treatment of chemical plant wastewater by replacing existing suspended carrier
media with Kaldnes Moving BedTM media at a plant in Singapore. Water Sci. Tech., 49, 11-12,
199-205.