Modelling Greenhouse Gas Emissions of a Hybrid Fixed-film Anammox Process Treating Sludge Dewatering Centrate in Wastewater Treatment

The aim of this study is to estimate and optimize greenhouse gas (GHG) emissions of a process in wastewater treatment, which utilizes anaerobic ammonium oxidation (Anammox). The single-stage nitritation-Anammox process applies fixed biofilm carriers and treats the centrate of sludge dewatering. GPS-X biokinetic modelling tool was used for quantifying the specific nitrous oxide, carbon dioxide and methane emissions at various operational conditions. In general, the amount of biology related GHG production was estimated to be higher than that of indirect emissions, by three orders of magnitude. Of direct emissions, nitrous oxide gas production should be taken into account primarily. Based on the simulations, feasible options of minimising N2O emissions include applying an operational temperature of 30-35°C, and increasing airflow to reduce the effect of oxygen limitation. To release less N2O, the process should also preferably be operated as an IFAS application with a low concentration of suspended solids (1.5-2 g/L), or even without sludge recycle.


INTRODUCTION
Anaerobic ammonium oxidation (Anammox) is a commonly used process in wastewater treatment, where ammonium nitrogen is converted into gaseous dinitrogen under anoxic conditions with nitrite as the electron acceptor [1]. Prior to the Anammox reaction, ideally a Original Research Article mixture of nitrite and ammonia is prepared in a so-called Sharon process, where part of the influent ammonia-N is oxidized to nitrite-N, by ammonia oxidizing bacteria [2]. To avoid the conversion of nitrite-N into nitrate-N by nitrite oxidizers, the Sharon process operates at a high temperature (>30°C) and pH of 7-8, generally without sludge retention [3]. Anammox microorganisms are inhibited reversibly by the presence of oxygen. Research studies have shown that aerobic nitrifiers do not play an important role in the main process itself [4].
Anammox is applied as a side-stream technology to reduce the load of reject water from sludge management to the bioreactors treating municipal wastewater, since 15-20% of the inlet nitrogen load is recycled with the return liquors from sludge dewatering [5]. There are examples of Anammox solutions from the whole spectrum of the leading wastewater treatment technologies; e.g. activated sludge in sequence batch reactors [6], rotating biological contactors [7], moving bed biofilm reactors [8], membrane bioreactors [9] and technologies applying upflow anaerobic sludge blanket [10]. The integrated fixed-film activated sludge (IFAS) construction can also be applied for Anammox treatment, with the use of clarifiers for sludge recycle; allowing suspended biomass to be retained in the bioreactors, apart from the biofilm bound to the carrier media. The slightly aerated suspended phase is ideal for implementing nitritation as part of the Sharon process, while Anammox microorganisms generally reside in the inner, anaerobic biofilm layers [11]. Although many studies describe the design and operation of such a system, during in-depth literature review no specific papers were found investigating GHG emissions of Anammox applications in terms of operational parameters. In this paper the authors' contribution to Anammox technology is to optimize GHG emissions originating from this process.
The commonly discussed greenhouse gas emissions in wastewater treatment are those of carbon dioxide, methane and nitrous oxide. Biology related carbon dioxide emissions derive from the degradation of organic matter and the aerobic respiration of biomass [12]. During anaerobic processes, methane is generally produced concurrently with carbon dioxide. The quantity of CH 4 depends on the amount of organic matter in wastewater; as well as the temperature and the type of treatment system applied [13]. Nitrification involves autotrophic (ammonia-oxidizing) bacteria, which convert the NH 4 + ions into the intermediate compound of NH 2 OH, followed by NO 2 -ions. Because of the latter step, NO and N 2 O are released as byproducts. Nitrite oxidizers transform NO 2 ions to NO 3 -. The build-up of NO 2 ions can lead to an increased production of N 2 O gas. Furthermore, due to low dissolved oxygen conditions, ammonia oxidizers can also consume NO 2 as a source of oxygen, which is then reduced into NO, then N 2 O, as a result of the process known as autotrophic denitrification [14]. However, nitrifiers also consume CO 2 as an inorganic carbon source [15]. Denitrification is a heterotrophic process involving four metabolic stages, during which NO 3 ions are formed into NO 2 ions, NO, N 2 O, and then N 2 gas. Lower C/N ratios of wastewater can cause higher emissions of nitrous oxide [14].
Operation of the Sharon-Anammox process also emit N 2 O, especially due to nitrite accumulation, which results in higher N 2 O concentrations in the off-gas [16]. The emissions can be reduced by applying a combined single-stage technology for the nitritation and Anammox processes (such as IFAS), partly preventing the accumulation [17]. Efficient control of aeration is required in such systems, since oxygen limitation is generally assumed to cause increased N 2 O emissions [16]. It is proposed that N 2 O emissions from singlestage nitritation-anammox reactors can also be minimised by operating the technology under conditions where anaerobic activity exceeds aerobic activity [18].

Model Used for Estimating GHG Emissions -Mantis 3
Mantis 3, a biokinetic model developed for GPS-X by Hydromantis, was used for quantifying GHG productions of the wastewater treatment plantfocusing on the hybrid Anammox reactor -, in CO 2 equivalents. The Petersen matrix and mathematical scheme of Mantis 3 is built based on the development of ASM2d. It covers the biological, physical and chemical processes experienced in wastewater engineering, such as hydrolysis, as well as metabolisms involving heterotrophs, autotrophs and phosphorus accumulating. It interprets nitrification and denitrification as two-step processes. It also incorporates denitrification by autotrophic bacteria, utilizing NO 2 as the electron acceptor instead of O 2 . A major function in terms of GHG emissions is the simulation of gas-liquid transfer processes: apart from the exchange of oxygen between the gas and liquid phase, it also takes into account the absorption and desorption of CO 2 , N 2 , CH 4 , H 2 and N 2 O, based on K L a volumetric mass transfer coefficients and their related saturation concentrations [19].
In Mantis 3, the most important feature for estimating GHG emissions is the integrated Carbon Footprint (CF) module. It classifies gas productions into three types. The sources of these emissions are detailed as follows.
Direct emissions related to biology: -CO 2 discharges from anaerobic, anoxic, and aerobic biological processes; -N 2 O production of nitrification and denitrification -CH 4 emitted by anaerobic processes.
Indirect emissions related to energy consumption: -Emissions attributed to pumping energy requirement; -Emissions caused by the energy demand of aeration; -Miscellaneous, energy use emissions.

Material emissions:
-Emissions caused by usage of chemicals; -Emissions brought about by use of materials, such as membranes or media; -Emissions related to transportation of materials.
The CF module also includes offsets of treatment plants, that help reduce net emissions.
The following offsets can be applied for sequestering direct emissions: -Biogenic capture of CO 2 (by nitrification, for example); -Flaring of CH 4 ; -Applying CH 4 for heating and energy production.
Practically, there is no way of offsetting the two other types of emissions mentioned.
Eq. (1) is used as a method of estimating direct emissions, at a given time of t: Where, concentration of dissolved gas i in a reactor (g/m 3 ); K L a i (t): volumetric global warming potential of gas i (-); 25 in case for CH 4 , and 298 for N 2 O [20].
The amount of indirect emissions is deduced by using the following formula, Eq. (3): Where, E scope2, i (t): indirect emissions of gas i in CO 2 equivalent (g/d), f elec, i : gas i emission factor for electricity generation (-).
The applied emission factor is a region specific value found in databases. This presumes that the three examined gases are produced in a set ratio, based on the generally known energy production processes of a given region. Certain regions can be chosen in GPS-X, of which the US national was selected for our simulations, to generalize GHG calculations [19]. Mantis 3 can also be used for estimating emissions by use of materials. In this study however; effects of chemical dosing, or replacement of carriers over time were not included.

Mathematical Model Setup of the Treatment Plant and Anammox Process
The wastewater plant -utilizing the Anammox process for centrate treatment -uses a fixed bed biofilm reactor cascade for biological wastewater treatment. The process layout was built in GPS-X (Fig. 1). The design capacity of the plant is 76 000 m 3 /d, the average influent and effluent water quality parameters are summarized in Table 1. Disc filter units with a pore size of 30 µm are used for phase separation after the biological reactors. They provide solids removal of 97%, and require counter-flow backwashing to remove the residual filter cake. The sludge removed by backwash from the filter units -with a quantity of 2280 m 3 /d -is forwarded to a gravity thickener, from which the supernatant is recycled to the bioreactors. Approximately 240 m 3 /d thickened sludge is generated, and further treated by an anaerobic digester applying a hydraulic residence time of 30 days. The digested sludge is dewatered by a centrifuge, from which the sludge cake is temporarily contained, then transported away.
Prior to being recycled to the reactor cascade, the centrate from dewatering is treated by a fixed-film Anammox system, to lower its high nitrogen content due to anaerobic digestion. This process is based on an integrated fixed-film activated sludge (IFAS) Anammox reactor, with a volume of 150 m 3 , and filling ratio of 0.071 m 3 /m 3 ; utilizing the same fixed bed biofilm carrier as the sewage treatment cascade. A dissolved oxygen concentration of 0.4 mg/L, and a temperature of 35°C is applied for operation of this single-stage process. A clarifier is used for recycling the suspended biomass into the Anammox reactor. The effluent is forwarded to the bioreactors, and so is the wasted sludge, containing active nitrifying biomass that can be used for continuous inoculation of the cascade. The IFAS Anammox reactor receives -in average -a 210 m 3 /d hydraulic inflow of centrate. The main influent and clarifier effluent characteristics of the centrate are revised in Table 2.

Modell Calibration and Validation
Based on IWA Good Modelling Practice Guidelines [21] model calibration is a step-wise procedure whereby different aspects of the plant model (in this case: sludge production, nitrification, denitrification, oxygen transfer) are calibrated in sequence. The procedure consists of characterization and fractionation of the influent wastewater, the specification of operational variables (e.g. flow rates), and the adjustment of key model parameters (e.g. target biofilm thickness) in order to minimize the error between measured and calculated data. The changes to the influent fractions were made based on the influent data and effluent soluble COD. The particulate COD/volatile suspended solids ratio in the model was adjusted match the influent VSS. The main calibration parameters of fixed film processes were as follows: anoxic growth reduction factor for ordinary heterotrophic organisms, biofilm mass per surface area ratio, rate of diffusion of pollutants into the biofilm. As the result of the calibration the standardized residuals for the variables were examined and the model results were within two standard deviations of the measured values. The whole procedure and a more detailed evaluation approach is published in the literature [21].

Evaluating Emissions of the Treatment Plant and the Anammox Process
Direct and indirect GHG productions were modelled, interpreting the standard operational conditions of the wastewater treatment plant, to compare the emissions from the Anammox treatment to the other biological treatment units. The summarized results are stated in Table 3, as specific emissions relative to the influent wastewater flow, in CO 2 equivalents.
The majority of GHG emissions originates from the sewage treatment cascade, the Anammox reactor's direct emissions are approximately 10 percent of the cascade's biology related GHG production. Indirect emissions from the cascade are more substantial, only lower by two orders of magnitude than its direct emissions; compared to the Anammox process where the quantity of indirect emissions is lower by three orders of magnitude than the biological emissions. This is mostly due to the fact that the nitritation-Anammox reactor operates at a reasonably lower DO concentration, requiring a lower specific airflow. The anaerobic digester is the unit with the lowest amount of direct emissions, comprising mostly of CO 2 , due to the combustion of most of the CH 4 gas. The simulation software implies that in case of the digester, there is no need to account for indirect emissions, as the energy requirement of heating can be covered by burning the CH 4 content of the biogas. Indirect emission from Anammox process (g CO 2 /m 3 water) 0.6 The N 2 O production of the IFAS Anammox process is much more significant compared to CO 2 and CH 4 emissions, than in the case of the bioreactor cascade, since reactors applying the nitritation and Anammox processes mainly involve nitrogen removal; and are not purposely intended for removal of organic carbon, like sewage treatment operations are.

Effects of the IFAS Anammox Operational Parameters on GHG Production
Five process variables of the nitritation-Anammox system were analysed regarding CO 2 , CH 4 and N 2 O production, by running steady-state simulations. This study focuses on biology related emissions since the indirect emissions are negligible in comparison, as mentioned previously.
The specific gas productions -relative to the volumetric flow of centrate -were quantified in CO 2 equivalents and they should be minimized besides meeting the target effluent quality. The characteristics of the centrate given in Table 2 were specified as a baseline for the simulations, varying one IFAS operational parameter at a time. Input parameters related to other units of the wastewater treatment plant have not been modified during the analyses.The ranges of operational parameters were selected based experiences, the hydraulic residence time is between 10.5 and 22.5 hours, the temperature is between 30 and 50°C, the DO setpoints are between 0.2 and 1.0 mg/l, the filling ratio is between 0.04 and 0.08 mg/l and the mixed liquor suspended solids concentration is between 1400 and 4000 mg/l. The effect of the parameters on GHG emission were analyzed individually, while one was being varied the other remained constant. Complex optimization algorithms with varying all of the parameters simultaneously have not been carried out, since the operators of wastewater treatment plants have limited capability of controlling all of these parameters at the same time. Emission of N 2 O was plotted on a secondary axis, because it was, in all cases, higher than CO 2 and CH 4 emissions by at least one order of magnitude.

Effect of hydraulic residence time
Applying a higher reactor volume proved advantageous for increasing the amount of Anammox biomass, lowering the effluent NH and NO 2 -N concentrations. Results of the GHG emission estimationsresidence times -are illustrated on Fig. 2. N production was shown to rise as a result of more intensive Anammox activity, however, the emission appears to stop increasing above a HRT of 17 hours. According to the simulations, altering the residence time apparently has no effect on CH 4 and CO 2 emissions. Thus selecting a higher Anammox reactor volume is rather to be considered economically than in the terms of GHG emissions.

Effect of reactor temperature
Raising the temperature of the water phase was also shown to be beneficial for Anammox bacteria, but disadvantageous for nitrifying biomass. Based on the modelling, the unit needs to be heated to at least 30°C to provide operational conditions for Anammox metabolism.
The effects of centrate temperature on GHG productions are summarized by Fig. 3.
Though higher temperatures provide better nitrogen removal, a gradually higher amount of N 2 O is released. CO 2 and CH increase, too, on a smaller scale, due to 255 time. Input parameters related to other units of the wastewater treatment plant have not been modified during the analyses.The ranges of operational parameters were selected based on experiences, the hydraulic residence time is between 10.5 and 22.5 hours, the temperature is between 30 and 50°C, the DO setpoints are between 0.2 and 1.0 mg/l, the filling ratio is between 0.04 and 0.08 mg/l and the mixed liquor tration is between 1400 and 4000 mg/l. The effect of the parameters on GHG emission were analyzed individually, while one was being varied the other remained constant. Complex optimization algorithms with varying all of the parameters simultaneously ot been carried out, since the operators of wastewater treatment plants have limited capability of controlling all of these parameters at O was plotted on a secondary axis, because it was, in all cases, emissions by at least

Effect of hydraulic residence time
Applying a higher reactor volume proved advantageous for increasing the amount of Anammox biomass, lowering the effluent NH 4 -N N concentrations. Results of the direct at different are illustrated on Fig. 2. N 2 O production was shown to rise as a result of more intensive Anammox activity, however, the emission appears to stop increasing above a g to the simulations, altering the residence time apparently has no emissions. Thus selecting a higher Anammox reactor volume is rather to be considered economically than in the terms of ture Raising the temperature of the water phase was also shown to be beneficial for Anammox bacteria, but disadvantageous for nitrifying biomass. Based on the modelling, the unit needs to be heated to at least 30°C to provide ammox metabolism. The effects of centrate temperature on GHG productions are summarized by Fig. 3.
Though higher temperatures provide better nitrogen removal, a gradually higher amount of and CH 4 emissions scale, due to anaerobic degradation processes promoted by higher temperatures. Considering GHG production, and from a financial point of view, it is safer to operate Anammox reactors at lower temperatures.

Effect of dissolved oxygen
Raising the dissolved oxygen level oxygen-limited zones in the reactor, and helps avoid the autotrophic denitrification process by nitrifiers, that produces N 2 O as they consume NO 2 -N. As shown by Fig. 4, a large amount of N 2 O emission can be spared by increasing DO levels, though this does not sensibly affect nitrogen removal.
Moreover, DO levels above this range can influence Anammox microbes by diffusion into the biofilm, risking the ammonia oxidizers outcompeting them, the accumulation of NO and worse treatment efficiency. Based on t model, a DO level of at least 0.2 mg/L is needed for suitable nitritation.
; Article no. BJECC.2016.024 anaerobic degradation processes promoted by higher temperatures. Considering GHG production, and from a financial point of view, it is safer to operate Anammox reactors at lower

Fig. 3. Direct GHG emissions of the IFAS Anammox process at altered temperatures
Raising the dissolved oxygen level minimises limited zones in the reactor, and helps avoid the autotrophic denitrification process by O as they consume N. As shown by Fig. 4, a

Effect of filling ratio
Larger media surfaces intensify biofilm activity, promoting nitrogen removal andother operational variables -moderately raising N 2 O production, as seen on Fig. 5. Increasing the filling grade is beneficial for Anammox organisms, that generally reside in the biofilm; thus selecting a high enough surface area is essential for stable operation of the single process, to remove both nitrite-N and ammonia N sufficiently. Applying a higher filing ratio virtually does not influence the removal of organic carbon, having no apparent effect on the other two examined GHG emissions.  Increasing the RAS flow mainly promotes nitrification in the suspended phase. The process can be operated quasi as a pure biofilm reactor Larger media surfaces intensify biofilm activity, compared to moderately raising O production, as seen on Fig. 5. Increasing the filling grade is beneficial for Anammox organisms, that generally reside in the biofilm; thus selecting a high enough surface area is essential for stable operation of the single-stage N and ammonia-N sufficiently. Applying a higher filing ratio virtually does not influence the removal of organic carbon, having no apparent effect on the other two examined GHG emissions.  Increasing the RAS flow mainly promotes nitrification in the suspended phase. The process can be operated quasi as a pure biofilm reactor without sludge retention, with an MLSS concentration of approximately 1400 mg/L. The introduction of recycled sludge provides a steep increase in NH 4 -N removal, but also sharply increases N 2 O production -as it can be seen on Fig. 6. As higher MLSS concentrations lower the oxygen transfer efficiency, more oxygen limited zones are present, also contributing to the increase in nitrous oxide emission.

CONCLUSION
Regarding the operation of an integrated fixed film Anammox process, modelled emissions are negligible compared to biology related emissions. Emissions of N most considerable, and suggested to be minimized by applying a sensibly low MLSS, and high enough DO concentration to evade oxygen limitation and autotrophic denitrification. Reactor temperature within 30-35°C also keeps the production of N 2 O at a lower level. without sludge retention, with an MLSS concentration of approximately 1400 mg/L. The rovides a steep N removal, but also sharply as it can be seen on Fig. 6. As higher MLSS concentrations lower the oxygen transfer efficiency, more oxygen limited zones are present, also contributing to the in nitrous oxide emission.