WO2017207011A1

WO2017207011A1 - Control of n2o-emissions by aeration

Info

Publication number: WO2017207011A1
Application number: PCT/DK2017/050186
Authority: WO
Inventors: Riccardo BOIOCCHI; Gürkan SIN; Krist V. GERNAEY
Original assignee: Danmarks Tekniske Universitet
Current assignee: Danmarks Tekniske Universitet
Priority date: 2016-06-03
Filing date: 2017-06-06
Publication date: 2017-12-07
Anticipated expiration: 2018-12-03

Abstract

Disclosed herein is a method and a system for controlling N₂O emission from a water-based solution containing nitrifying organisms, wherein an input variable for a controller is the ratio between NO₃ ^- produced and NH₄ ⁺ oxidized as determined by measuring NO₃ ^- and NH₄ ⁺ concentration both at the inlet and at the outlet of the aerated section containing the water-based solution with nitrifying organisms and a scaled output variable is obtained from the controller as a function of said ratio by a non-linear algorithm, so that the oxygen input and/or the aeration time is varied based on an integration of the deviations of said output variable.

Description

Control of N₂0-emissions by aeration

The invention relates to a method for controlling the emission of N₂0 during biological removal of NH₄ ⁺ from a water-based solution such as waste water, drinking water or another aqueous solution having too high a content of NH₄ ⁺. Biological removal means that microorganisms decompose unwanted substances. The invention also relates to a system comprising elements allowing reduced emission of N₂0 during treatment of the water-based solution. In particular the invention relates to a control strategy providing control of the microbial turnover of NH₄ ⁺ and resulting in a reduced emission of N₂0. Background

Wastewater treatment plants are used for cleaning wastewater and in that process (among others) remove ammonium from the wastewater. Currently, the main focus when cleaning wastewater is to reduce - or preferably remove - the amount of potential harmful substances in the wastewater. Little focus has been on the gasses being released during such a cleaning process.

Nitrous oxide (N₂0) is well-known as an ozone-depleting substance and a harmful greenhouse gas with a global warming potential 300 times higher than carbon dioxide. Measurement campaigns at wastewater treatment plants (WWTPs) have revealed that a considerable amount of N₂0 is normally emitted. Developing control strategies aiming at reducing the emissions of the greenhouse gasses, in particular N₂0, is therefore of interest.

During biological wastewater treatment processes, N₂0 is found to be produced by two different microbial groups: ammonia-oxidizing bacteria (AOB) and heterotrophic bacteria (HB). In particular, AOB have shown their capability in using the produced N0₂ ^" instead of 0₂ as electron acceptor for the oxidation of NH₄ ⁺. The reduction of N0₂ ^" carries N₂0 as end product. Another possible pathway is the incomplete hydroxylamine oxidation by AOB.

Intermediate compounds accumulated during this process can lead to N₂0. With regard to the role of HB, N₂0 is produced as intermediate compound during the reduction of nitrogen oxides like N0₃ ^" and/or N0₂ ^" into N₂. If the reduction rate of N₂0 into N₂ is slower than the reduction rate of N oxides into N₂0, an accumulation of N₂0 can occur. The minimization of N₂0 emissions can therefore be achieved by slowing down both the AOB-mediated N₂0-production processes and the net HB-mediated N₂0

production process.

There is thus a need for a method for slowing down both the AOB-mediated N₂0- production processes and the net HB-mediated N₂0 production process.

Description of the invention

Disclosed herein is a method for controlling N₂0 emission from a water-based solution containing nitrifying organisms. The term water-based solution containing nitrifying organisms is meant to include wastewater treatment plants that employs a mixed culture of microorganism in particular Ammonium Oxidizing Bacteria (AOB) and Nitrite Oxidising Bacteria (NOB) in various reactor configurations such as granular or biofilm or suspended mixed culture systems or a hybrid system based on any combination of configuration employing nitrifying organisms. Systems that employ mixed culture of microorganisms that performs nitrification (AOB and NOB) - also referred to as microbial conversion of NH4-N - include as examples drinking water, process water, industrial wastewater treatment, soil treatment, are also included in the term water- based solution containing nitrifying organisms. Thus, examples of water-based solution containing nitrifying organisms include a wastewater solution with nitrifying organisms, a drinking water solution with nitrifying organisms, an industrial process water, an industrial wastewater solutions with nitrifying organisms, and similar.

The method comprises the step of providing an aerated section containing the water- based solution with nitrifying organisms. The aerated section may comprise one or more units in form of closed or open tanks or reactors.

By aerated is meant an "oxygen containing fluid", which is provided by various aeration supply systems such as air, pure oxygen, surface aerators, bubble aeration and similar. The aerated section is having an inlet for an oxygen containing fluid providing oxygen input to the aerated section for an aeration time, an inlet for leading the water-based solution into the aerated section, and an outlet for leading the water-based solution out of the aerated section.

The water-based solution with nitrifying organisms may flow through the aerated section at an aerated section flow rate. The method further comprises the steps of:

• measuring the concentration of NH₄ ⁺ ([NH₄ ⁺]|_N) and N0₃ ^" ([N0₃ ^"]|_N) at the inlet of the aerated section or upstream of the aerated section, and

• measuring the concentration of NH₄ ⁺ ([ΝΗ₄ ⁺]₀υτ) and N0₃ ^" ([Ν0₃ ^"]ουτ) at the outlet of the aerated section or downstream of the aerated section;

The method also comprises the step of providing a process or input variable R_NatAmm for a controller, the process or input variable corresponding to the ratio between N0₃ ^" produced and NH₄ ⁺ oxidized inside the aerated section.

In one or more embodiments, R_NatAmm is calculated by the equation:

;^a¾£A5¾Sii H rfS .

The method further comprises the steps of:

• obtaining a scaled output variable (A_sk_La, k_La, AUDO_SP, AF,T_a) from the

controller as a function of RNatAmm by a non-linear algorithm, the value of the scaled output variable being between -1 and +1 ;

• obtaining a scaling factor (SF_AskLa, SF_KLA, SFAUDOSP, SF_AF, SF_TA) determined by the chosen output variable

• multiplying the scaling factor with the scaled output variable in order to

determine a deviation of the output variable, and

• providing a signal to an actuator varying the oxygen input to the aerated section based on an integration of the deviations of the output variable.

Alternatively, the method further comprises the steps of:

controller as a function of R_NatAmm by a non-linear algorithm, the value of the scaled output variable being between -1 and +1 ;

• multiplying the scaling factor with the scaled output variable in order to

determine a deviation of the output variable, and

• providing a signal to an actuator varying the aeration time. By varying the aeration time, i.e. the duration of the intermittent aeration, the time for the aerated periods and the time for unaerated periods in the aerated section is controlled. In intermittent aeration, there is a duration when the controller is active, i.e. switched on, and there is a duration when the controller is inactive, i.e. switched off. Therefore, these systems involve a cyclic operation where aerobic and anoxic phases alternate in each cycle by means of intermittent aeration. The total cycle time, T_c, thus includes an aerated time, T_A, where nitrification is allowed to proceed with NH₄ ⁺ consumption and conversion to nitrite and/or nitrate; and the following anoxic phase, T_D, denitrification is allowed to proceed with consumption of N0₃ ^" and build-up of NH₄ ⁺ . The aerated fraction, AF, defines the fraction of time within a cycle in which the reactor is aerated, AF= T_A/T_C..

Intermittent aeration, periodically aeration and aeration time is meant to refer to the same parameter.

In one or more embodiments, the oxygen input is kept constant and the aeration time is varied. In one or more embodiments, the oxygen input is kept constant and the aeration time parameters such as T_c and or AF is varied.

In one or more embodiments, the aeration time is fixed and the oxygen input is varied. Yet alternatively, the method further comprises the steps of:

• obtaining a scaled output variable (A_sk_La, k_La, AUDO_SP AF,T_a) from the

• obtaining a scaling factor (SF_AskLa, SF_KLA, SF_AUDOSP_, SF_TAI SF_af) determined by the chosen output variable

• multiplying the scaling factor with the scaled output variable in order to

determine a deviation of the output variable, and

• providing a signal to an actuator varying the oxygen input to the aerated

section, and/or the aeration time based on an integration of the deviations of the output variable. Disclosed herein is also a system for treatment of a water-based solution containing nitrifying organisms.

The system comprises an aerated section having:

• an inlet for leading the water-based solution into the aerated section;

• an outlet for leading the water-based solution out of the aerated section,

• an inlet for an oxygen containing fluid providing oxygen input to the aerated section, and

an outlet for gas emission;

The water-based solution with nitrifying organisms may flow through the aerated section at an aerated section flow rate.

The oxygen containing fluid may be provided by different aeration supply systems such as air, pure oxygen, surface aerators, bubble aeration, membrane aeration systems and similar.

The system also comprises:

• a first unit measuring an inlet value for N0₃ ^" concentration ([N0₃ ^"]|_N);

· a second unit measuring an outlet value for N0₃ ^" concentration ([N0₃ ^"]₀UT);

• a third unit measuring an inlet value for NH₄ ⁺ concentration ([NH₄ ⁺]|_N); and

• a fourth unit measuring an outlet value for NH₄ ⁺ concentration ([ΝΗ₄ ⁺]₀υτ)-

The system further comprises a non-linear controller unit adapted for:

- receiving the measured values for [Ν0₃ ^"]ι_Ν, [Ν0₃ ^"]ουτ, [NH₄ ⁺]|_N and [ΝΗ₄ ⁺]₀υτ; calculating a process or output variable RNatAmm, i.e. the ratio between N0₃ ^" produced and NH₄ ⁺ NatAmm is e.g. calculated by using

calculating an error variable E_RNatAmm, and

providing an output variable for an actuator determining the supply of the oxygen containing fluid to the aerated section.

Alternatively, the system further comprises a non-linear controller unit adapted for:

- receiving the measured values for [N0₃ ^"]|_N, [Ν0₃ ^"]₀υτ, [NH₄ ⁺]|_N and [ΝΗ₄ ⁺]₀υτ; - calculating a process or output variable RNatAmm, i.e. the ratio between N0₃ ^" produced and NH₄ ⁺ oxidized inside the aerated section, wherein RNatAmm is e.g.

calculated by using the

- calculating an error variable E_RNatAmm, and

- providing an output variable for an actuator determining the aeration time.

Yet alternatively, the system further comprises a non-linear controller unit adapted for:

- receiving the measured values for [Ν0₃ ^"]ι_Ν, [Ν0₃ ^"]ουτ, [NH₄ ⁺]|_N and [ΝΗ₄ ⁺]₀υτ;

- calculating a process or output variable R_NatAmm, i.e. the ratio between N0₃ ^" produced and NH₄ ⁺ oxidized inside the aerated section, wherein R_NatAmm is e.g. calculated by using the equation:.... **^".'^ I. ¾fcS¾^')iirai.;

- calculating an error variable E_RNatAmm, and

- providing an output variable for an actuator determining 1) the supply of the oxygen containing fluid to the aerated section and 2) the aeration time.

By the method and system according to the above, it is possible to control the emission of N₂0 during treatment of a water-based solution where the primary purpose is to remove NH₄ ⁺ from the solution. This in particular useful in wastewater treatment plants where a reduction of the N₂0 production can be reduced with at least 70-80 % without increasing the costs very much.

Brief description of the drawings

Figure 1 shows a layout of a Benchmark Simulation Model N°2 for Nitrous oxide. Figure 2 is a block diagram for the implementation of a controller in the mainstream activated sludge unit.

Figures 3a-b show an embodiment of membership functions for RNatAmm (figure 3a) and ASkLa (figure 3b).

Figure 4 shows the dynamic and steady-state of influent TKN load simulated using the BSM2N model shown in figure 2. Figures 5a-j show the dynamic results for both open-loop and closed-loop

configurations of the N₂0 emission (figure 5a, 5b, 5c), RNatAmm (figure 5d, 5e, 5f), the oxygen mass transfer coefficient of the first aerobic tank (figure 5g), and the oxygen-to- nitrogen loading ratio (figure 5h, 5i, 5j).

Figure 6 shows a layout of another Benchmark Simulation Model N°2 for Nitrous oxide.

Figure 7a-b fuzzy-logic model used for R_NatAmm_,sp used in the layout of figure 6. Figure 8a-c fuzzy-logic model used for set point for the oxygen control used in the layout of figure 6.

Figures 9a-c show the N₂0 emission using a regular open-loop configuration and the fuzzy-logic controller of figure 6.

Figure 9d-f show the set point of dissolved oxygen and the actual dissolved oxygen.

Figure 10 is a table showing the N₂0 emission both for TKN_in and TKN_rem, the TN and TKN rate, and the aeration cost.

Figure 1 1 shows the NH₄effluent concentration membership functions.

Figures 12a-b show 3D plots with the AUDO_SP on the z axis and NH₄ ⁺eff and E_RnatAmm in the other two axes.

Description of preferred embodiments

In one or more embodiments, the system comprises a unit measuring dissolved oxygen concentration in the aerated section or at the outlet from the aerated section. Since the conversion rate of organic nitrogen into nitrate is increased by an increased operating temperature, the optimal value for R_NatAmm is expected to increase in function of operating temperature. To take this into account, temperature adaptation of the set points for RNatAmm may be included using a separate fuzzy-logic module, using the measured influent temperature as single input variable. Thus, in one or more embodiments, the system comprises a temperature transmitter positioned at the inlet to the aerated section or upstream of this inlet, and a second non-linear controller unit configured to receive a signal from the temperature transmitter.

In one or more embodiments, the system comprises an oxygen-transmitter registering the level of oxygen in the flow of water-based solution inside or at the outlet of the aerated section and a linear controller unit configured to receive a signal from the oxygen-transmitter.

For the development of a control strategy applied to biological wastewater processes, a fuzzy-logic approach may be the most suitable to be adopted. As a matter of fact, given their interactive nature and their high non-linearity, biological wastewater treatments can be more suitably controlled by non-linear controllers like fuzzy-logic controllers than by linear controllers like the Proportional Integrative Derivative (PID) ones. Fuzzy-logic controllers (FLCs) present the additional possibility of incorporating expert knowledge about the processes to be controlled (R. Boiocchi et al., J. Process Control. 30 (2015) 22-33). FLCs are also not affected by the capability of a mathematical model in describing realistically the processes to be controlled, since their design is independent from the model used. Furthermore, adopting a fuzzy-logic approach easily allows including the fuzziness associated with the control objectives, typically related to the wastewater treatment processes.

For these reasons, a FLC aiming at minimizing the N₂0 emissions by balancing the activity of the different microbial groups is investigated further here.

In one or more embodiments, the non-linear algorithm is a fuzzy-logic model having membership functions for the process or input variables and the output variables, wherein the input variable R_NatAmm is considered to be "GOOD" when it is around 1 , i.e. higher than 0.95 and preferably higher than 0.99 and lower than 1.3 and preferably lower than 1.2.

In one or more embodiments, the non-linear algorithm is a fuzzy-logic model having membership functions for the process or input variables and the output variables, wherein the input variable R_NatAmm is considered to be "LOW" when it is below 0.95, or below 0.99. In one or more embodiments, the non-linear algorithm is a fuzzy-logic model having membership functions for the process or input variables and the output variables, wherein the input variable RNatAmm is considered to be "HIGH" when it is higher than 1.3 or preferably 1.4.

If the input variable RNatAmm is around 1 , the process is not accumulating N0₂ and it should not be necessary to regulate the oxygen input to the aerated section. Even if the input variable R_NatAmm is slightly above 1 , i.e. between 1.0 and lit should not be necessary to regulate. However, if R_NatAmm is below 0.99 it is considered necessary to increase the oxygen input to the aerated section and if R_NatAmm is above 1.2 it is necessary to reduce the oxygen input to the aerated section.

As an alternative to increasing or reducing the oxygen input, the duration of the aeration during intermittent (or periodically) aerated sections may be decreased or increased, respectively. By prolonging or shortening the time when the aeration is switched on in the aerated section, an effect similar to the effect obtained by controlling the oxygen input continuously can be observed.

By varying the aeration time, i.e. the duration of the intermittent aeration, the time for the aerated periods and the time for unaerated periods in the aerated section is controlled. In intermittent aeration, there is a duration when the controller is active, i.e. switched on, and there is a duration when the controller is inactive, i.e. switched off. Therefore, these systems involve a cyclic operation where aerobic and anoxic phases alternate in each cycle by means of intermittent aeration. The total cycle time, T_c, thus includes an aerated time, T_A, where nitrification is allowed to proceed with NH₄ ⁺ consumption and conversion to nitrite and/or nitrate; and the following anoxic phase, T_D, denitrification is allowed to proceed with consumption of N0₃ ^" and build-up of NH₄ ⁺ . The aerated fraction, AF, defines the fraction of time within a cycle in which the reactor is aerated, AF= T_A/T_C. Thus, a variation of the duration of the aeration (T_A orA_F ) may be combined with the variation of the oxygen input rate (DeltaDOsp, DeltaKIa or kla).

MATLAB/Simulink may be used as computer environment for the development of the fuzzy-logic control strategy. The model employed for testing its performance is the Benchmark Simulation Model N°2 for Nitrous oxide (BSM2N), developed by Boiocchi et al. (Extending the benchmark simulation model no2 with processes for nitrous oxide production and side-stream nitrogen removal, in: 12th Int. Symp. Process Syst. Eng. 25th Eur. Symp. Comput. Aided Process Eng., Elsevier, 2015: pp. 2477-2482). On this platform, dynamic simulations may be performed in order to evaluate the capability of the controller in reducing the N₂0 emissions and its contextual effect on effluent quality and operational costs.

The BSM2N describes the physical and biochemical mainstream and side-stream processes occurring in a typical pre-denitrification WWTP. An overview of the configuration of the BSM2N is shown in Figure 1. As can be seen in the figure, the biological mainstream unit consists of two well-mixed anoxic tanks 101 , 102 followed by three aerated tanks 103, 104 105. The biological processes occurring in the reactors 101 , 102, 103, 104, 105 are described by an upgraded version of the Activated Sludge Model Ν (ASM1) by Henze et al. (Water Res. 21 (1987) 505-515.), namely the Activated Sludge Model for Greenhouse gases N°1 (ASMG1) by Guo et al. (Bioprocess Biosyst. Eng. 37 (2014) 151-163).

Wastewater enters the tank system at an inlet position 106 and exits the tank system 101-105 at an outlet position 107. Sludge for disposal is led out at separate position 108. The state variables of the ASMG1 are:

• oxygen (S₀₂);

• readily-biodegradable and slowly-biodegradable carbon (S_s and X_S,

respectively);

• ammonium, nitrite, nitrate, dinitrogen, nitric and nitrous oxide nitrogen (S_NHx,

SNO2. SNO3. SN2, SNO and SN2O. respectively);

• soluble and particulate inerts (S| and X_H respectively);

• soluble and particulate organic nitrogen (S_ND and X_ND, respectively);

• alkalinity (S_A_K), and

• heterotrophic, ammonia-oxidizing and nitrite-oxidizing bacteria (X_HB, XAOB and XNOB, respectively).

The processes included in the ASMG1 are:

• aerobic growth of X_HB;

• hydrolysis of X_S, hydrolysis of X_ND;

· ammonification of S_ND;

• aerobic growth of X_AOB; • aerobic growth of X_NOB;

• anoxic growth of X_HB on S_No3, on S_No2, on S_No and on S_N2o;

• anoxic growth of X_AOB on S_No2 and on S_No, and

• decay of X_HB, XAOB and X_NOB-

The production of N₂0 is modelled to be produced according to two pathways: 1) during HB denitrification and 2) during nitrite reduction (via nitric oxide) by AOB. The production of N₂0 as a consequence of incomplete hydroxylamine (NH₂OH) oxidation by AOB is not included in the model, as a recently-developed model including the incomplete NH₂OH reveals that the amount of N₂0 possibly produced during this process seems to be rather low compared to the amount of N₂0 produced during AOB and HB denitrification (M. Pocquet et al., Water Res. 88 (2015) 948-959.).

A fuzzy-logic inference system suitable for controlling the system of figure 1 consists of the following three main sequential operations:

Fuzzification, where numerical (crisp) inputs are converted into linguistic (fuzzy) inputs, according to predefined membership functions (MFs),

Fuzzy inference, where fuzzy outputs are deduced on the basis of the fuzzy inputs deduced in point 1), according to specified linguistic rules,

Defuzzification, where fuzzy outputs are converted into numerical (crisp) outputs according to predefined MFs and a chosen defuzzification method.

As can be deduced, the main parameters to be decided during the design of a fuzzy- logic controller are the MFs of input and output variables and the linguistic rules linking input to output variables.

The controller according to this invention is designed in such a way that it minimizes the production of N₂0 by both AOB and HB. Heterotrophic denitrification can only occur under low oxygen conditions. The process can therefore occur in the anoxic zone in tanks 101 and 102 and at very poor oxygenation regimes in the aerobic zone in tanks 103, 104 and 105 (see figures 1 and 2).

The contribution on total N₂0 by HB however is stronger in the aerobic zone 103, 104, 105 due to the higher mass transfer capability. In the anoxic zone 101 , 102, the HB- produced N₂0 stays in the liquid phase and thus is more likely to be reduced into N₂ rather than to strip.

With regard to the production of N₂0 during AOB denitrification, it is well-established that low oxygen concentrations and/or high nitrite (N0₂ ^") availability promote the use of N0₂ ^" itself by AOB as electron acceptor for the oxidation of ammonium (NH₄ ⁺). Hence, for its minimization, in the aerobic zone, the oxygen has to be in a sufficiently high concentration to enhance the consumption of the AOB-produced N0₂ ^" by the coexisting microbial group, namely NOB. Enhancing the NOB activity by means of oxygen supply increase would contextually minimize the amount of N₂0 produced by HB. Thus, the control of NOB activity can be beneficial for the reduction of the N₂0 produced by both AOB and by HB.

For these reasons, monitoring the NOB activity over the one of AOB has been identified as a potential strategy for the minimization of the production of N₂0. The variable identified to be controlled at this purpose is the ratio between the nitrate produced by NOB and the ammonium consumed by AOB in the aerobic zone, expressed as RNatAmm in Eqn. (1). s™ ^" SH^SSH ⁽¹⁾

A value of R_NatAmm around 1 theoretically indicates that all the nitrite produced by AOB is subsequently consumed by NOB. If RNatAmm is lower than 1 , not all the AOB- produced N0₂ ^" is consumed by NOB. Thus N0₂ ^" starts accumulating in the system, which would in turn enhance AOB denitrification. This is a typical situation resulting from low oxygen availability.

On the other side, values of the RNatAmm parameter significantly higher than 1 represent oxygen inhibition of heterotrophic denitrification, where an accumulation of HB- produced N0₂ ^" occurs. A too high oxygen supply would in part directly inhibit the HB- produced N0₂ ^" reduction and consume a larger amount of organic carbon needed by HB. In this scenario, the oxygen supply needs to be turned down to avoid that the accumulated N0₂ ^" to be reduced into N₂0. In virtue of these considerations, the control strategy use measurements from the influent and effluent of the aerobic zone of NH₄ ⁺ and N0₃ ^" for the calculation of RNatAmm, namely the controlled variable. The oxygen mass transfer coefficient (k_La) is used as a manipulated variable since changes in the oxygen availability forms the only available actuator that is able to induce the due shift on NOB activity. The fuzzy-logic controller described above is implemented in the mainstream aerobic zone as depicted in Figure 2. As can be seen, the concentration of NH₄ ⁺ and N0₃ ^" before the wastewater enters the aerobic tanks 103, 104, 105 are measured at 1 10 and the concentration of NH₄ ⁺ and N0₃ ^" after the wastewater exits the aerobic tanks 103, 104, 105 are measured at 1 11 and the values used for calculation R_NatAmm a a measuring unit 112.

The controller 113 will use the value of R_NatAmm as direct input variable and will deduce a scaled deviation of the oxygen mass transfer coefficient (A_sk_La) as output variable. The latter will have a value between -1 and +1. The oxygen mass transfer coefficient (A_sk_La) is then multiplied by a scaling factor (SF_kLa) at the unit 1 14 in order to obtain its physical dimension. The deviations of k_La are integrated in time and then added up to their respective nominal values.

In one or more embodiments, the membership functions for the input and output variables are defined as represented in Figures 3a-b. Table 1 shows the linguistic rules to enable deducing on the basis of the values of RNatAmm, the variation of the k_La to be actuated on the three aerobic tanks.

Table 1 : linguistic rules for a membership function

IF THEN

RNatAmm is considered to be "GOOD" when it is comprised within 0.99 and 1.2. The reason for allowing a value of R_NatAmm slightly higher than 1 is due to the fact that a fraction of the incoming organic nitrogen, after being hydrolysed and ammonified, is usually oxidized into N0₂ ^" and then N0₃ ^". An amount of N0₃ ^" higher than the NH₄ ⁺ (i RNatAmm^>1) consumed would therefore result. When RNatAmm is in this range, no changes of k_La are designed to be actuated.

On the contrary, the maximal positive change of k_La (i.e. A_sk_La= 1) is decided to occur when RNatAmm is equal or below 0.95, a scenario which would indicate that NOB need more oxygen for the conversion of N0₂ ^" into N0₃ ^". When RNatAmm is equal or higher than 1.4, inhibition of heterotrophic denitrification is considered to occur. In this case, the maximal negative change of k_La (i.e. A_sk_La=-1) is defined to be inferred by the control system.

A value for the k_La_NoM equal to 120 d^"1 is assigned for the first two aerobic reactors 103, 104 and a value of 60 d^"1 was used for last tank 105. A 240 d^"1 value corresponding to the difference between the saturation limit of k_La (i.e. 360 d^"1) and the nominal value in the first two tanks (120 d^"1) is used as value for the scaling factor.

Simulations of a BSM2N open-loop and closed-loop configuration are performed with the aim of addressing the changes due to the implementation of the control strategy disclosed herein. In particular, the BSM2N is simulated with a 609-day long dynamic influent. The dynamic and the steady-state of the influent Total Kjeldahl Nitrogen (TKN) load for the last month are depicted in Figure 4.

Corresponding to the same period of time, Figures 5a-j show the dynamic results for both open-loop and closed-loop configurations of the N₂0 emission factor in figure 5a calculated as percentage of N₂0 emitted per unit of TKN in the influent, RNatAmm in figure 5d, 5e, 5f, the oxygen mass transfer coefficient of the first aerobic tank 103 in figure 5g, and the oxygen-to-nitrogen loading ratio (RO) in figure 5h, 5i, 5j, calculated according to Eqn. (2) as the ratio between the oxygen supplied into the three aerobic tanks 103, 104, 105 and the TKN in the influent of the first anoxic tank 101.

The present quantity indicates the typical aeration regime of the plant.

Following the BSM2 protocol for the benchmarking of control strategies from both the open-loop and closed-loop simulation results of the last 52 weeks, which allows a comparison more unbiased with regard to plant initial conditions and nominal value of k_La, the following average values are found:

N₂0 emission factors (N₂O_ef),

- total nitrogen removal efficiency (HTN),

RNatAmm,

N0₂ ^" and N0₃ ^" effluent loads,

Effluent limit violations in percentages of operating time for ammonium and total nitrogen (V_NH4 and V_TN, respectively),

- Effluent Quality Index (EQI), calculated by taking into account the amount of the different pollutants in the effluent. The higher EQI is, the worst the effluent quality is,

- the average aeration energy (AAE), proportional to the oxygen mass transfer coefficients.

An overall evaluation of the impact of the controller implementation on the plant performance can thus be achieved and is summarized in Table 2.

Table 2: Open-loop and closed-loop plant performance evaluation.

As can be noted from table 2, the controller is able to reduce the average N₂0 emitted by 35% by keeping the controlled variable, namely R_NatAmm, a a higher value. Thus NOB activity is enhanced and, consequently, a higher consumption of N0₂ ^" results, leading to lower N0₂ ^" load in the effluent. The results of table 2 show that due to the higher NOB activity, the load of N0₃ ^" in the effluent is increased, which explains in turn the reduced TN removal efficiency (from 70.8% to 60.7%). The effluent quality index is therefore increased accordingly. Also the percentage of operating time in which TN violations (V_TN) are recorded is higher for the closed-loop configuration, although the percentage of operating time in which NH₄ ⁺ violations (V_NH) occur is reduced due to the higher AOB resulting from the higher aeration. Since the controller has enhanced the NOB activity by increasing the amount of oxygen supplied, the average aeration costs have increased by approximately 30% compared to the open-loop case.

The results presented in figures 1-5 and in the table 1 and 2 show the capability of the controller in reducing significantly the average amount of N₂0 emitted by speeding up the NOB activity, which prevents the AOB-produced N0₂ ^" from being reduced to N₂0. The results also suggest that reduction in the total N₂0 emitted, which can be considered satisfactory, can be improved by further enhancing NOB activity.

Figure 6 shows an updated version of the model shown in figure 1 , where the aerobic zone/section 1 15 may contain one or more open of closed aerobic tanks, e.g. the three aerobic tanks 103, 104, 105 as shown in figure 1.

The control system used in figure 6 adapts the aeration regime 115 in order to minimize the difference between a set-point of RNatAmm (RNatAmm.sp) and the actual RNatAmm- The control concept is implemented using a feedback fuzzy logic control framework 123 similar in origin to the one used and described above for figures 1-5. However, compared to the previous implementation, the membership functions were finely tuned and significantly improved.

The set-point R_NatAmm_,sp changes automatically with the temperature of the influent wastewater at 106.

In one or more embodiments of the invention, the method comprises the step of measuring a temperature T of the water-based solution at one or more of the following locations:

· upstream of the inlet of the aerated section

• at the inlet of the aerated section • inside the aerated section

• at the outlet of the aerated section

• downstream of the outlet from the aerated section.

The value is set to vary from 1.1 in cold seasons to 1.3 in warm seasons for a wastewater treatment plant in Europe/Scandinavia. This is included in a separate fuzzy-logic module 124. The control system manipulates the oxygen set point to be tracked by a Proportional Integrative (PI), which in turn manipulates the air supply (k_La).

In one or more embodiments of the invention, the method comprises the step of using the temperature T of the water-based solution for obtaining a set point (RNatAmm.sp) for the variable RNatAmm, wherein RNatAmm.sp is a function of T given by:

NatAmm.SP

In order to optimally regulate the air supply, measurements of the following are needed:

- [N0₃ ^"] and [NH₄ ⁺] 110 in to the aerobic zone 115;

- [N0₃ ^"] and [NH₄ ⁺] 11 1 out of the aerobic zone 115;

- [DO] 125 of the aerobic zone 1 15;

Temperature of the influent water 106 or the water in aerobic zone 1 15.

The fuzzy-module 124 used with the system in figure 6 is used to update the set point for RNatAmm.sp as a function of the temperature is shown in figures 7a-b, where figure 7a shows the input variable being the temperature and the output variable obtained therefrom in figure 7b.

The linguistic rules implemented in order to enable the module inferring the RNatAmm.sp on the basis of a measured temperature are shown in table 3.

Table 3. Linguistic rules implemented in order to enable the module inferring the

RNatAmm.SP

IF THEN

"Temperature is LOW" "RNatAmm.SP IS LOW" "Temperature is MEDIUM" "RNatAmm.SP IS MEDIUM"

"Temperature is HIGH" "RNatAmm.SP IS HIGH"

The fuzzy-logic module 123 used to infer the set point for the DO concentration is designed as shown in figure 8a, and the rules applying thereto in figure 8b. An alternative fuzzy-logic module 123 used to infer the set point for the DO concentration is designed as shown in figure 8c.

The input variable used in the model 123 is given as:

ERNatAmm ^—

(3)

Thus, in one or more embodiments of the invention, the method comprises the step of calculating an error function defined by E_RNatAmm = RNatAmm.sp - RNatAmm.

In one or more embodiments, E_RNatAmm is considered to be:

· "GOOD" when it is around 0, i.e. higher than -0.1 and lower than 0.1 ,

• "LOW" when it is below -0.2 and maybe up to -1 , and

• "HIGH" when it is higher than 0.1 and up to 1.4.

Alternatively, in one or more embodiments, E_RNatAmm is considered to be:

· "GOOD" when it is around 0, i.e. higher than -0.05 and lower than 0.05,

• "LOW" when it is below -0.4 and maybe up to -1 , and

• "HIGH" when it is higher than 0.2 and up to 1.4.

Alternatively, in one or more embodiments, E_RNatAmm is considered to be:

· "GOOD" when it is around 0, i.e. higher than -0.1 and lower than 0.1 ,

• "LOW" when it is below -0.1 and maybe up to -1 , and

• "HIGH" when it is higher than 0.1 and up to 1.4.

Alternatively, in one or more embodiments, E_RNatAmm is considered to be:

· "GOOD" when it is around 0, i.e. higher than -0.05 and lower than 0.05,

• "LOW" when it is below -0.2 and maybe up to -1 , and

• "HIGH" when it is higher than 0.2 and up to 1.4. In one or more embodiments, NH₄ ⁺eff is considered to be:

• "GOOD" when it is lower than 1.5, and

• "HIGH when it is higher than 2. This is illustrated in figure 1 1. NH₄ ⁺eff is normally obtained using the sensor 126 as shown in figure 6.

The output value is denoted AuDO_Sp. The linguistic rules implemented in order to enable the control module 123 inferring the unitary variation of the oxygen set point on the basis of the difference between R_NatAmm_,sp and measured R_NatAmm are given in table 4.

In one or more embodiments, the linear controller receives a signal from the second non-linear controller (AuDO_Sp), wherein the signal provides a set point for the measured level of oxygen (DOsp) in the flow of water-based solution leaving the aerated section.

Table 4. Linguistic rules used in the module 123.

In one or more embodiments of the invention, the method comprises the step of providing an output variable (AuDO_Sp), wherein AuDO_Sp is scaled and provided as a set point to a linear feedback controller such as a P, PI or PID controller, wherein the linear feedback controller receives a process or input variable corresponding to a dissolved oxygen-content of the treated water-based solution and then provides a signal to an actuator determining an extent of aeration input to the aerated section.

For both the RNatAmm.sp fuzzy-logic module 124 and the RNatAmm fuzzy-logic control module 123, the implication method chosen is the correlation-minimum, the

aggregation method is the disjunction and the defuzzification method is the Center-of- Area. The scaling factor is a scalar number which regulates only the speed of the controller. The scaling factor has to be of the same order of magnitude of the nominal value. A first attempt value for SF_DOsp is 1. The system is tested on the BSM2 simulation model comparing control strategies for domestic wastewater treatment plants. The simulation results on the BSM2 using different mathematical models for N₂0 production show a 70-80% N₂0 emission reduction compared to the corresponding open-loop scenario. This is valid both in terms of N₂0 emission per unit of influent Total Kjeldahl Nitrogen (TKN) and in terms of N₂0 emission per unit of TKN removed. In figures 9a-c, the N₂0 emission when using a regular open-loop configuration (OL) is compared to the closed-loop fuzzy-logic controller (CL). Figures 9d-f show the set point of dissolved oxygen and the actual dissolved oxygen. The increment in aeration energy costs is minimal (2-6%). The control system is robust against sensor and actuator noises.

This is a large improvement of the results shown in figures 1-5 and proves the effect of the added robustness of the system achieved by: (I), finely tuning the membership functions for the input variable, and (II) including the temperature element, and (III) manipulating the DO_Sp rather than the k_La in a cascade configuration which is in turn used by linear controller to calculate back the k_La .

Figure 10 is a table showing performance of the fuzzy logic controller with respect several performance metris: the percentage of N₂0 emission relative to TKN_in and TKN_rem the percentage of TN removed and percentage of TKN removed, and the aeration cost (AEC) for the following four controller configurations:

(1) using the traditional well-known open-loop model of BSM2,

(2) the regulatory model, where the fuzzy logic controller acts directly on the k|a (figures 1-5),

(3) the cascade model, where the fuzzy logic controller acts on DOsp, which is then sent as input to the linear controller which acts on then k|a, and

(4) the sens and act model, which uses the cascade control configuration together with sensor and actuator models - adding sensor/actuator dynamics and noise as defined by BSM2. The results are shown for three different mathematical models used for describing the biological processes occurring in a WWTP. Model A corresponds to a ASMG1 model, model B corresponds to a ASMG1 model with a first set of AOB production pathways, and model C corresponds to the ASMG1 model with another set of AOB production pathways.

As can be seen in the table in figure 10, the N₂0 emission is greatly reduced with the method and system of the invention compared to the open-loop model. At the same time cost are comparable for the cascade model and the open-loop model.

The mathematics behind the fuzzy-logic modules used in this application can be described as follows:

The steps in the fuzzy logic model approach may follow the following scheme: Step i) Calculation of R_NatAmm given by: R_NatAnn(t =

Step ii) Calculation of the set point for RNatAmm as a function of the temperature T

Step iii) Calculation of fuzzy-logic controller input given by:

Step iv) Calculation of the unitary variation of dissolved oxygen set point given by:

^_1 if ERNatAmm(t)≤—OA

2.61^■ E_RNatAmm(t) + 0.079 if -0.4 < E_RNatAmm(t) < -0.05

AuDO_SP(t) 0 if -0.05≤ E_RNatAmm(t)≤ 0.05

6.36^■ E_RNatAmm(t) - 0.272 if 0.05 < E_RNatAmm(t) < 0.2

1 if E_RNatAmm(t)≥ 0.2

Step v) Calculation of actual variation of the dissolved oxygen set point given by:

ADO_SP(t) = SF_D0SP ^■ AuDO_SP(t)

Step vi) Calculation of the dissolved oxygen set point given by:

DO_SP(t = DO_SPinom + J ADO_SP(t)^■ dt Step vii) Calculation of the error between DOsp and measured oxygen concentration given by: e(t) = DOsp(t) - DO(t) Step viii) Calculation of k_La given by: k_La(t) = K_p ^■ e(t) + K,^■ / e(t)^■ dt

Alternatively the calculation of the unitary variation of dissolved oxygen set-point in step iv) is given by figures 12a-b, which are a 3D plot with the AUDO_SP on the z axis and NH₄ ⁺eff and E_RnatAmm in the other two axes. It describes graphically the control law according to which the deltaUDOsp is derived on the basis of measured NH₄ ⁺eff and ERnatAmm- This is an alternative way of the mathematical expression of the control law, needed to be graphic because the inputs are two and not on as the case before where only E_RNatAmm is used. As shown in figure 6, the fuzzy-logic control module 123 may also receive input from the sensor 126 regarding the NH₄ ⁺, effluent concentration at the outlet of the aerated section or downstream of the aerated section. While keeping R_NatAmm a its optimal value and thus reducing N₂0 production, it is important to ensure a sustained ammonium conversion with the aim of respecting ammonium effluent limits. For this reason, the effluent ammonium concentration (NH₄ ⁺, effluent) may be used as additional input to the control system to support taking the decision on the proper control action.

By this approach, one is able to directly take into account effluent ammonium concentration in the effluent. This adds on to the approaches focusing mainly on having a good ratio of RNatAmm- In this approach, the flexibility is allowed to consider controlling effluent NH₄ ⁺ quality in addition to having a good ratio of RNatAmm-

The NH₄ ⁺, effluent concentration may be integrated in an updated model, providing the linguistic rules used in the module 123 as shown in table 5.

Table 5. Linguistic rules used in the module 123

ERNatAmm

LOW GOOD HIGH

(NH₄ ⁺)eff GOOD NEGATIVE ZERO POSITIVE

With regard to the manipulated variable used to achieve the control objective, either the oxygen supply, here represented by the oxygen mass transfer coefficient {k_La), or the oxygen concentration set point (DO_Sp) of a Proportional Integral (PI) controller, may be adopted to regulate the oxygen input in the aerobic zone. The control structure in terms of disturbances (d), states (x), manipulated (u) and controlled variables around an aerated zone may be summarized as: d = \N H^_in, S_{s in}, T_in, X_{s in}, Xp in_> N 0₂ i_n, N 0_{3 in}, Q_in, NO_in, N₂ O_in, N₂ in, Xi in, S! _in] (4) x = [NH+, S_s, T, X_s, Xp, N0₂ , NO3 , NO, N₂0, N₂, X„ S,] (5) = [RNatAmm>

(6) u = [k_La] or [DO_SP] (7) Based on the two manipulated variables shown in Eqn. (7), two generic control structures may be implemented in an aerobic section for the minimization of N₂0 emissions.

The physical constraints on the controlled variables RNatAmm and [NH₄ ⁺]_eff as shown in the figures may be expressed in Eqn. (8). As can be seen, both of them have a lower limit of zero. The upper limit for R_NatAmm is given by the sum of the unit, which is achieved when all the influent ammonium is converted into nitrate, and the ratio between the total amount of influent organic nitrogen (N_org,inAER) and the amount of influent ammonium consumed. As a matter of fact, as previously described, the organic nitrogen influent to the aerobic zone can be, under high oxygenation regimes, quickly hydrolysed and ammonified, oxidized into nitrite and finally converted into nitrate by NOB. Thus R_NatAmm can have values higher than unity. TKN_{in AER} is the Total Kjeldahl Nitrogen in the influent to the aerobic zone which is used to represent the maximal amount of effluent ammonium that could potentially be released. As a matter of fact, at rather poor oxygen levels, the influent ammonium is not consumed and, in addition, the influent organic nitrogen is hydrolysed and ammonified.

All disturbances and state variables other than the oxygen concentrations in the aerobic zone can have theoretical values comprised between 0 and <+∞. With regards to the effluent ammonium concentration ([NH₄ ⁺], eff), although a typical legal limit for effluent ammonium concentrations is 4 mg N. L^"1 , in order to stay on the safe side, in an embodiment it may be decided that values higher than 2 mg N. L^"1 are considered the worst system operation requiring maximal control action. Optimal values for [NH₄ ⁺-N]_eff may be chosen to be equal or lower than 1 .5 mg N. L^"1.

Eqns. (9-12) express the critical points for the vector of controlled variables identified for each system operation, optimal and worst, for different scenarios.

UB for OPTIMAL SYSTEM OPERATION => y_{UB 0PT} = [1.2 0] (9)

LB for OPTIMAL SYSTEM OPERATION (10)

YUB, WORST— [1-2 2]

VLB, OPT = [1-4 0]

UB for WORST SYSTEM OPERATION VIB.OPT = [i o]

(1 1) VLB.OPT = [1-4 2]

νΐΒ,ορτ— [1 2] LB for WORST SYSTEM OPERATION

VUB, WORST — [l-2 TKN_{IN AER}]

org, in AER

VuB.WORST \ 1.5

ΔΝΗ4 J

(12)

J B, WORST ⁼ [0 1-5]

V AER

^TKNin'

yUB.WORST — [O TKN_{in AER}\

With regards to the shape of the membership functions for input and output variables, triangular and trapezoidal shapes may be chosen for the sake of simplicity.

The last step consists in the assignment of the degree of membership to fuzzy sets previously identified for each numerical value of RNatAmm and [NH₄ ⁺-N_eff]. A degree of membership to the identified fuzzy sets equal to 1 may be assigned to the critical points defined in subsection 2.2 as expressed in Eqns. (13-16). A degree of membership equal to 0 to the other fuzzy sets may be assigned to the same critical points.

UB for OPTIMAL SYSTEM OPERATION (13)

(fiNatAmm)GOOD — 1-15 and 1.25

LB for OPTIMAL SYSTEM OPERATION (14)

= 1.5

GOOD

(^NatAmm)wW

UB for WORST SYSTEM OPERATION (.^NatAmm) HIGH (15)

HIGH

NatAmm LOW = 0

N, org, in AER (16)

LB for WORST SYSTEM OPERATION NatAmm)HIGH = 1 + -

ANH

(NH+_eff) = TKN_inAER Abbreviation list

WWTP Wastewater treatment plants

AOB Ammonia-oxidizing bacteria

NOB Nitrite-oxidizing bacteria

HB Heterotrophic bacteria

FLC Fuzzy logic controller

BSM2N Benchmark Simulation Model no 2 for Nitrous oxide

ANOX Anoxic tank

AER Aerobic tank

ASMG1 Activated Sludge Model for Greenhouse gases no1

RNatAmm Ratio between N0₃ ^" produced and NH₄ ⁺ oxidized

RNatAmm.SP Set point of R_NatAmm

MF Membership functions

k_La Oxygen mass transfer coefficient

A_sk_La Unitary deviation of k_La

DOSP Set point of dissolved oxygen

ASDOSP Unitary deviation of DO_Sp

SFDOSP Scaling factor for DO_Sp

k_LaNOM Nominal value for k_La

SFk_La Scaling factor for k_La

TKN Total Kjeldahl Nitrogen

TN Total Nitrogen

ΠΤΝ TN removal efficiency

EQI Effluent Quality Index

Claims

Method for controlling N₂0 emission from a water-based solution containing nitrifying organisms, the method comprising the steps of:

• providing an aerated section containing the water-based solution with nitrifying organisms, the aerated section having:

- an inlet for an oxygen containing fluid providing oxygen input to the aerated section for an aeration time;

- an inlet for leading the water-based solution into the aerated section, and

- an outlet for leading the water-based solution out of the aerated section;

• measuring the concentration of NH₄ ⁺ ([NH₄ ⁺]|_N) and N0₃ ^" ([N0₃ ^"]|_N) at the inlet of the aerated section or upstream of the aerated section;

• providing a process or input variable RNatAmm for a controller, the process or input variable corresponding to the ratio between N0₃ ^" produced and NH₄ ⁺ oxidized inside the aerated section, wherein R_NatAmm is e.g. calculated by the equation: ...^^ . I ¾}.ii4 : ¾ . .raL,

• obtaining a scaling factor (SF_AskLa, SF_KLA, SF_AUDOSP, SF_AF, SF_ta) determined by the chosen output variable;

• multiplying the scaling factor with the scaled output variable in order to

determine a deviation of the output variable, and

• providing a signal to an actuator varying the oxygen input to the aerated

section, and/or the aeration time based on an integration of the deviations of the output variable.

2. Method according to claim 1 , wherein RNatAmm is calculated by the equation:

3. Method according to any preceding claim, wherein the oxygen input is kept

constant and the aeration time varied. Method according to the claims 1-2, wherein the aeration time is fixed and the oxygen input is varied.

5. Method according to any preceding claim, wherein the non-linear algorithm is a fuzzy-logic model having memberships functions for the process or input variables and the output variables, wherein the input variable RNatAmm is considered to be "GOOD" when it is around 1 , i.e. higher than 0.95 and preferably higher than 0.99 and lower than 1.3 and preferably lower than 1.2.

6. Method according to any preceding claim, wherein the non-linear algorithm is a fuzzy-logic model having memberships functions for the process or input variables and the output variables, wherein the input variable R_NatAmm is considered to be "LOW" when it is below 0.95, or below 0.99.

7. Method according to any preceding claim, wherein the non-linear algorithm is a fuzzy-logic model having memberships functions for the process or input variables and the output variables, wherein the input variable RNatAmm is considered to be "HIGH" when it is higher than 1.3 or preferably 1.4.

8. Method according to any preceding claim further comprising the step of measuring a temperature T of the water-based solution at one or more of the following locations:

• upstream of the inlet of the aerated section

• at the inlet of the aerated section

• inside the aerated section

• at the outlet of the aerated section

• downstream of the outlet from the aerated section.

9. Method according to claim 8 comprising the step of using the temperature T of the water-based solution for obtaining a set point (RNatAmm.sp) for the variable RNatAmm, wherein RNatAmm.sp is a functio of T given by

R NatAmm.SP

10. Method according to claim 9 comprising the step of calculating an error function defined by ERNatAmm ⁼ RNatAmm.SP - RNatAmm.

1 1 . Method according to claim 10, wherein E_RNatAmm is considered to be:

· "GOOD" when it is around 0, i.e. higher than -0.1 and lower than 0.1 ,

• "LOW" when it is below -0.2 and maybe up to -1 , and

• "H IGH" when it is higher than 0.1 and up to 1 .4.

12. Method according to any of the claims 1 1 further comprising the step of providing an output variable (AUDO_SP), wherein AUDO_SP is scaled and provided as a set point to a linear feedback controller such as a P, PI or PI D controller, wherein the linear feedback controller receives a process or input variable corresponding to a dissolved oxygen-content of the treated water-based solution and then provides a signal to an actuator determining an extent of aeration input to the aerated section.

13. System for treatment of a water-based solution containing nitrifying organisms, the system comprising:

• an aerated section having:

- an inlet for leading the water-based solution into the aerated section; - an outlet for leading the water-based solution out of the aerated

section,

- an inlet for an oxygen containing fluid providing oxygen input to the aerated section for an aeration time, and

- an outlet for gas emission;

· a first unit measuring an inlet value for N0₃ ^" concentration ([N0₃ ^"]|_N);

• a second unit measuring an outlet value for N0₃ ^" concentration ([N0₃ ^"]₀UT);

• a third unit measuring an inlet value for N H₄ ⁺ concentration ([N H₄ ⁺]|_N); and

• a fourth unit measuring an outlet value for N H₄ ⁺ concentration ([ΝΗ₄ ⁺]₀υτ),

• a non-linear controller unit adapted for:

o receiving the measured values for [N0₃ ^"]|_N, [Ν0₃ ^"]₀υτ, [N H₄ ⁺]|_N and

o calculating a process or input variable R_NatAmm, i.e. the ratio between N0₃ ^" produced and N H₄ ⁺ oxidized inside the aerated section, wherein RNatAmm is e.g. calculated by using the equation: jS¾^HS¾ .; o calculating an error variable E_RNatAmm, and

o providing an output variable for an actuator determining:

a) the supply of the oxygen containing fluid to the aerated section, and/or

b) the aerating time.

14. A system according to claim 13, wherein RNatAmm is calculated by the equation: *^tA!i!" ¾ rf¾k ^" 15. A system according to claim 13 or 14 further comprising a unit measuring

dissolved oxygen concentration in the aerated section or at the outlet from the aerated section.

16. A system according to any of the claims 13-15, wherein the system further

comprises a temperature transmitter positioned at the inlet to the aerated section or upstream of this inlet, and a second non-linear controller unit configured to receive a signal from the temperature transmitter.

17. A system according to claim 16, wherein the temperature is used to calculate a setpoint variable R_NatAmm_,sp according to claim 9.

18. A system according to any of the previous claims 13-17, wherein E_RNatAmm

⁼ R NatAmm ,SP" R NatAmm■ 19. A system according to any of the previous claims 13-18, wherein the system

further comprises an oxygen-transmitter registering the level of oxygen in the flow of water-based solution inside or at the outlet of the aerated section and a linear controller unit configured to receive a signal from the oxygen-transmitter. 20. A system according to claim 19, wherein the linear controller receives a signal from the second non-linear controller (AuDO_Sp), wherein the signal provides a set point for the measured level of oxygen (DOsp) in the flow of water-based solution leaving the aerated section.