The role of inorganic carbon limitation in biological nitrogen removal of extremely ammonia concentrated wastewater

Water Research 37 (2003) 1100–1110

The role of inorganic carbon limitation in biological nitrogen removal of extremely ammonia concentrated wastewater Bernhard Wett*, Wolfgang Rauch Department of Environmental Engineering, University of Innsbruck, Technikerstr.13, A-6020 Innsbruck, Austria Received 26 September 2001; received in revised form 10 June 2002; accepted 26 August 2002

Abstract It is clear from the fundamental biochemical processes that nitrification of extremely concentrated ammonia loads requires—among others—(1) sufficient alkalinity to buffer acidification and (2) bicarbonate as the substrate for the autotrophic biomass. However, at low pH values the aeration process causes CO2 stripping and consequently a decrease of the available inorganic carbon. In order to analyse such complex interactions, we suggest in this paper an enhanced version of the widely acknowledged IWA (formerly IAWQ) activated sludge models. These model enlargements comprise an ion-balance for the calculation of the pH value and of dissociation species, a balance of inorganic carbon and a more detailed description of the relevant N-elimination processes and their inhibitions. The model was successfully employed to optimise a treatment strategy for rejection-water and landfill leachate (500–2000 mg ammoniaN l
1, COD/N ratio of 0.25–4). Detailed data from two full-scale rejection-water treatment plants were used for systems identification, model calibration and validation. The results suggest that inhibition and limitation by nitrous acid (HNO2) and unionised ammonia (NH3) have often been overestimated. In this investigation the bicarbonate concentration proved to be crucial for the process. The optimisation of the bicarbonate concentration in the reactor
1
1 could improve the nitrozation rate up to 100 mg NH+ h . 4 -N l r 2002 Elsevier Science Ltd. All rights reserved. Keywords: Ammonia; Modelling; Kinetics; Rejection-water; Bicarbonate limitation; Inhibition; Carbonate system; Nitrification; Gas stripping

1. Introduction It is common knowledge that nitrification is a twostep process where ammonia is first oxidised by Nitrosomonas species to nitrite (NO
2 )—the so-called nitrozation process—and subsequently the nitrite to nitrate (NO
3 ) mainly by Nitrobacter species—a process denoted nitration. Denitrification is then the reduction
of NO
3 to NO2 and further on to N2 by the catabolism of heterotrophic bacteria. In municipal wastewater treatment, however, NO
2 as an intermediate of nitrogen removal processes occurs hardly. The reason is that—at usual wastewater temperatures of 10–151C and in the *Corresponding author. Tel.: +43-512-507-6926; fax: +43512-507-2911. E-mail address: [email protected] (B. Wett).

absence of inhibition and limitations—Nitrobacter are growing almost twice as fast as Nitrosomonas. Consequently nitrozation is commonly the rate-limiting nitrification process. Similarly, in denitrification the reduction of NO
2 is about 1.8 times faster than nitrate reduction [1]. As the overall nitrite consumption is almost twice as high as the production, NO
2 cannot accumulate in common municipal wastewater treatment. Hence, it is common practice in activated sludge modelling to describe nitrification and denitrification not in this complexity but to reduce the processes to one comprehensive step each [2]. However, for the description of nitrogen removal in case of extremely ammonia loaded wastewater—e.g. from rejection-water of digested sludge, landfill leachate or piggery wastewater—this simplified assumption is no longer applicable. As typically various inhibitions and/

0043-1354/03/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 0 4 3 - 1 3 5 4 ( 0 2 ) 0 0 4 4 0 - 2

B. Wett, W. Rauch / Water Research 37 (2003) 1100–1110

or limitations occur it is no longer obvious which of the steps described above determines the entire process. Consequently, more sophisticated model assumptions are required for an adequate description of the system performance. Former research on that issue concentrated predominately on inhibition effects due to high concentrations of NH3 and HNO2 (e.g. [3,20]) and on limitation effects due to a lack of NH3 and HNO2 as the actual substrates for nitrification [4,5]. In our paper, we attempt to analyse systematically the complex interactions that dominate acidification and limit inorganic carbon as a substrate of the autotrophic biomass. We start with a thorough description of the systems identification process which is based on the data derived from monitoring a full-scale rejection-water treatment process. In a second step the resulting calibrated model is applied for process optimisation. The investigation clearly identified bicarbonate limitation as the relevant rate limiting effect.

2. Model development In a nutshell, system identification is the process to suggest a set of models based on prior knowledge of the system and to select then the most adequate of these models by favouring both simplicity of the model structures and a good fit with the measured data. Subsequently, the parameter values of the model must be estimated (calibration) before being able to apply the model for forecasting. Reichert and Omlin [6] elaborate further the problem to identify the ‘true’ model structure. They argue that there are typically several different model structures that describe the system in an adequate way and cannot be discriminated. In order to reduce the uncertainty related to model structure selection, it is useful practice to break a complex process into several steps that can be individually identified, e.g. by means of experiments. However, as outlined earlier, the process of nitrogen elimination is characterised by several interlinked inhibition and limitation effects. Consequently an individual analysis of the sub-processes is neither purposeful nor practicable and, hence, model structure identification had to rely on the observations of the overall process under varying, dynamic conditions. The overall process could be monitored as it actually occurred as a result of strategies proposed by the authors generating a huge amount of data for a proper system identification. 2.1. Model structure The structure of the presented numerical model is determined by the time scale of the involved processes:

*

*

1101

Biochemical processes (reaction) and physical processes (transport) proceed at a certain rate and are calculated by mass balance equations. The resulting set of differential equations needs to be integrated over time in order to calculate the gradual change of concentrations. Transport denotes here liquid flow, gas transfer (aeration and gas stripping) and sedimentation. In contrast to these time dependent processes, dissociations are assumed to happen instantly and are calculated by equilibrium chemistry-based algorithms. Dissociation depends on the current pH value, temperature and on ionic strength and is rather a state than a process. In each time step the ionic charges of the electrolytes in the solution are balanced with no effect on the total concentration of an electrolyte.

The total concentration of the electrolyte (e.g. total inorganic carbon) and the ionic charge (i.e. alkalinity) are calculated according to the principle of mass conservation. Then the dissociation equilibria are balanced and the concentration of each ion is determined. But a direct conclusion from alkalinity to the pH value cannot be drawn, because the power of the current buffer systems has to be taken into account. The dissociation equilibria [3,7] are calculated for two conditions—under the estimated pH value and at neutrality. Then all ion differences are summed up according to their valence which leads to the new alkalinity (dissociation products of water, carbonic acid, nitric acid, nitrous acid and ammonia are considered). The iteration loop is repeated with a varied pH value until the exit-criterion is met. In the next time step, the biokinetic reaction term is calculated under the inhibitory and limiting influence of the ion concentrations according to the current pH value (Fig. 1). The model, which has been applied on the simulation examples below, distinguishes 14 compounds, 8 ionssub-compounds and 10 biochemical processes. In the following section only those model components are outlined, which describe potential rate limiting phenomena—that are inhibition kinetics of nitrogen elimination processes and bicarbonate limitation including inorganic carbon balance and outgassing. 2.2. Four-step nitrogen elimination and its inhibitions A detailed description of the metabolism pathway of nitrogen elimination enlarges the common model assumptions of Henze et al. [2] significantly: Beside the four oxidation and reduction processes separate decay processes for Nitrosomonas and Nitrobacter have to be implemented (see e.g. [8]). Additional compounds are to be inserted in order to balance Nitrosomonas, Nitro-

B. Wett, W. Rauch / Water Research 37 (2003) 1100–1110

1102

Fig. 1. Scheme of the model structure.

bacter, nitrite and nitrate. Inhibitions can be caused by substrates and products of the nitrogen elimination processes—by NH3 and HNO2. Both substances need not be individually included in the compound-process matrix but can be calculated at any instant as subcompounds according to their dissociation. Anthonisen et al. [3] investigated the range of inhibiting concentrations of nitrification. In their work three inhibition zones are defined, first the inhibition of Nitrobacter by unionised ammonia (NH3>0.1– 1.0 mg l
1), second the inhibition of Nitrosomonas by unionised ammonia (NH3>10–150 mg l
1) and third the inhibition of Nitrobacter by nitrous acid (HNO2>0.2– 2.8 mg l
1). Anthonisen points out that acclimation, number of nitrifiers and temperature may affect the inhibitory concentrations. Much of the same reason Abeling [1] defines the inhibition parameters of nitrous acid as a function of the operation level, i.e. the HNO2 concentration to which the sludge is exposed over a period of time. The same author also gives a comprehensive survey on the wide range of values found in the literature for inhibition parameters and boundary concentrations. Compared to this parameter uncertainty the exact shape of the inhibition function is of minor influence on the accuracy of the description of the inhibitory influence. Thus for the NH3 and HNO2 inhibition the simplest form of inhibition kinetic was chosen: rinhibited ¼ r

ki ki þ I

ð1Þ

where i stands as suffix for both substances. The inhibited process rate rinhibited is calculated from the uninhibited process rate r and the concentration I of the inhibitory substance and the inhibition constant ki (process rates are given in Table 1). Model calibration is based on data from full-scale experiments at the WWTP Strass [9]. With respect to the relations described above calibration revealed no inhibition due to nitrous acid. This is not surprising, as first the sludge was adapted to relative high nitrite concentrations (average nitrite concentration of 100 mg
1 NO
2 -N l ) and second the concentration of nitrous
1 acid (100 mg NO
at pH 7.3 results into 0.07 mg 2 -N l
1 HNO2-N l ) was below the inhibition threshold given by Anthonisen (0.2–2.8 mg HNO2-N l
1). In contrast to HNO2, unionised ammonia had a strong influence on the nitration process. The NH3 concentration in the pilot plant reactor was at 1 mg
1 NH3-N l
1 in average (equivalent to 100 mg NH+ 4 -N l
1 at a pH value of 7.3) and maximum 3 mg NH3-N l . An initial inhibition constant ki of 1.6 mg NH3-N l
1 achieved the best fit during calibration. However, when the process was operated under stable conditions the inhibition constant increased to 20 due to sludge adaptation. For nitrozation on the other hand no inhibition could be detected. These observations confirm that the sensitive ammonia concentration zone for nitrozation is at least two orders of magnitude higher than for nitration. Consequently, nitration inhibition by unionized ammonia remains the only relevant inhibitory influence of the investigated nitrogen compounds. This

Stripping CO2 stripping

NO3 reduction

Anoxic growth NO2 reduction

Decay NB

Nitrobacter Nitration

Decay NS

Nitrosomonas Nitrozation

Process

SNO3 SNO3 =SNO2 kO2 Ss SNH kSh þ Ss kNHh þ SNH kNOh þ SNO3 kO2 h þ SO2 ðSNO3 =SNO2 Þ þ 1

SNO2 kO2 Ss SNH kSh þ Ss kNHh þ SNH kNOh þ SNO2 kO2 h þ SO2



  HO2 Vfluid Qair ¼ Rð1
xÞ = O2air
O2fluid kstrip O2 rO2   HCO2 dCO2fluid Qair ¼ CO2air
CO2fluid kstripCO2 RxO2 dt rCO2 Vfluid

mh Xh Z

1:77mh Xh Z

SO 2 kO2 b þ SO2

expððSHCO3
kHCO3 Þ=a kNH3 nb SNO2 SO2 kHNO2 kNOnb þ SNO2 kO2 nb þ SO2 expððSHCO3
kHCO3 Þ=a þ 1 kNH3 nb þ SNH3 kHNO2 þ SHNO2

 bnb Xnb 0:5 1 þ

mnb Xnb

SO2 0:1 þ SO2



expððSHCO3
kHCO3 Þ=aÞ kNH3 ns SO2 kHNO2 SNH kNHns þ SNH kO2 ns þ SO2 expððSHCO3
kHCO3 Þ=aÞ þ 1 kNH3 ns þ SNH3 kHNO2 þ SHNO2

 bns Xns 0:5 1 þ

mns Xns

Process rate

Table 1 Kinetic parameter set used in given simulation examples

kstripO2 ¼ 1:15; kstripCO2 ¼ 0:042

x ¼ 0:2

Z ¼ 0:8; ks ¼ 20 kNHh ¼ 0:5; kNOh ¼ 0:5; kO2 h ¼ 0:2

mh ¼ 6e0:069ðT
20Þ

bnb ¼ 0:20e0:098ðT
20Þ

kNOnb ¼ 0:3; kO2 nb ¼ 1:0; kNH3 nb ¼ 20

a ¼ 10; kHCO3 ¼ 50 [mg C/l], kHNO2 ¼ 2:8 mnb ¼ 1:06T
15

kO2 ns ¼ 0:4 [mgO2/l], kNH3 ns ¼ 3000 bns ¼ 0:23e0:098ðT
20Þ

mns ¼ 0:6  1:1T
15 ½1=d kNHns ¼ e0:117ðT
20Þ ½mgN=l

Parameters

B. Wett, W. Rauch / Water Research 37 (2003) 1100–1110 1103

B. Wett, W. Rauch / Water Research 37 (2003) 1100–1110

1104

inhibition can shorten the metabolism route by two steps since nitrite is reduced directly and not via nitrate. From the economical point of view nitration inhibition is beneficial, since 25% of the stoichiometric oxygen demand for nitrification and 40% of the required organic carbon for denitrification can be saved. In the literature controversial opinions are found concerning the inhibition of the denitrification process. Nowak et al. [10] report that anoxic growth of heterotrophic biomass on NO
2 is favoured over that on NO
3 , whereas Abeling’s [1] experiments indicate the opposite. Moreover, she found that in case of a decrease
1 of the nitrate concentration to about 10 mg NO
3 -N l , the nitrate reduction stagnates and the nitrite reduction starts. Our measurements at the pilot plant in Strass, where the sludge was adapted to concentrations as stated above, confirm the reported preference of nitrite. The ratio of 1.8 of the process rates as mentioned above gave a good fit to the model performance. Nevertheless, a decrease of the nitrate reduction rate was measured as
soon as NO
3 concentration came close to 10 mg NO3
1 N l . Consequently, a dampening term with the competitive ratio of nitrite and nitrate concentrations was implemented in the kinetic function of the nitrate reduction rate r: rcompetitive ¼ r


NO
3 =NO2 :

ðNO3 =NO2 Þ þ 1

ð2Þ

2.3. Balance of inorganic carbon and its limiting influence Usually, the aerobic degradation of organic matter by the heterotrophic biomass produces CO2 in abundance. At a COD/NH+ 4 -N ratio of about 20 (which is common in municipal wastewater), it is not to be expected that the carbon supply for autotrophic growth could become the limiting factor. Nevertheless, the ammonia-nitrogen concentration in rejection-water from anaerobic sludge treatment can be four times higher than the COD concentration. In this case respiration in an activated

sludge system is dominated by nitrifying organisms and a limitation of bicarbonate—the actual substrate—can occur. Inorganic carbon limitation was primarily investigated in natural water systems. Model studies have pointed out the importance and the relationship of pH on the inorganic carbon limitation of algal biofilm growth [11,12]. Grguric et al. [13] have balanced outgassing of CO2 due to acidity from nitrification at the New Jersey State Aquarium. Concerning wastewater treatment, inorganic carbon limitation has been observed in nitrifying biofilms [14] and a stimulation of nitrifier growth has been achieved by bicarbonate addition [15]. The range of growth limiting bicarbonate concentrations is much higher than the ones of ammonia and oxygen. According to our measurements and simulations following function describes the HCO
3 influence on nitrifier’s growth (Fig. 2): vlimited ¼ v

eððHCO3





kÞ=aÞ

ððHCO3

kÞ=aÞ

e

: þ1

ð3Þ

The concentration of inorganic carbon results from a balance of assimilation, respiration and CO2 stripping due to aeration. Under aerobic conditions autotrophic organisms depend on the inorganic carbon which has been oxidised by the heterotrophic biomass. CO2 stripping displaces significantly the C/N ratio in the reactor. The exchange between the liquid and gas phases occurs at the water surface and at the boundary of the air bubbles in case of pressurised aeration. The low partial pressure of CO2 in the aerated liquid (atmospheric CO2 concentration 0.035%) limits the CO2 concentration of the solution according to Henry’s law. Thus the rising air bubbles strip more inorganic carbon than they can transfer into the water. Fig. 3 shows schematically the influences on inorganic carbon production and consumption. Additionally, the equilibrium between carbon dioxide and bicarbonate depends on the alkalinity balance. This dissociation equilibrium

Fig. 2. Types of kinetics applied in the model assuming the same inhibition/saturation constant k:

B. Wett, W. Rauch / Water Research 37 (2003) 1100–1110

1105

Fig. 3. Flow-scheme of inorganic carbon during wastewater treatment.

is the most important buffer mechanism within the relevant pH range between 6.5 and 8. For the description of the CO2 stripping due to aeration a semi-empirical approach with a lumped stripping parameter is inevitable. In another field of application Bisset et al. [16], respectively Wanninkhof [17], calculated the gas transfer velocity of CO2 at the air–seawater interface as a function of wind speeds. Musvoto et al. [18] considered gas exchange in their 3phase mixed weak acid/base kinetic model. Stripping rates for CO2 have been observed and calculated two orders of magnitude higher than that for NH3. The flux of CO2 across the open boundary of the water–air interface is described by FluxCO2 ¼ k LCO2 ðpCO2w
pCO2a Þ;

ð4Þ

where k is the gas transfer velocity, LCO2 is the solubility of CO2 expressed in units of concentration/pressure, and pCO2 is the partial pressure of CO2 in air and water [17]. The gas flux is defined as a mass flow per unit of time and unit of interfacial area. Therefore, the gas concentration CO2w is calculated by dCO2w FluxCO2 a
RCO2 ¼ Vw dt ðpCO2w
pCO2a ÞLCO2 k a
RCO2 ; ¼ Vw

ð5Þ

where a is the water–gas interface area, Vw the water volume of the reactor and RCO2 the biokinetic reaction rate of CO2. The interfacial area a obviously depends on the air-flow Qair ; the size of the air bubbles, the aerator type and the water depth. These complex influences and the gas transfer velocity k have been lumped to one parameter—the stripping coefficient kstrip : Calibration revealed that gas stripping is not linear to the air-flow which led to the implementation of a power law based on the air-flow or on the oxygen respiration rate RO2 : The product of partial pressure pCO2 and solubility LCO2 results in the actual gas concentration in the

water CO2w : dCO2w ðCO2w
pCO2a LCO2 Þ ðkstrip CO2 RxO2 ÞQair ¼ dt Vw
RCO2 ; dO2w ðO2w
pO2a LO2 Þ ðkstrip O2 RxO2 Þ Qair ¼
RO2 ; dt Vw Qair ¼

Rð1
xÞ Vw O2 : kstrip O2 ðO2w
pO2a LO2 Þ

ð6Þ ð7Þ

ð8Þ

Since the gas flux is calculated from the difference between the partial pressure in the atmosphere and in the water, the Eqs. (4)–(6) address gas entry and exit at the interface and are also applicable for the oxygen transfer due to aeration (Eq. (7)). In order to couple the aeration to the actual CO2 stripping, the air-flow Qair is calculated from (7) by (8) and can be inserted in Eq. (6). This suggested description of the relevant gas–liquid interactions requires the calibration of three parameters—the stripping coefficients of O2 and CO2 and the exponent x for the influence of respiration on the aeration efficiency.

3. Results and discussion 3.1. Parameter calibration and process optimisation The presented model was developed in order to investigate and optimise the separate biological SBRtreatment of rejection-water at the WWTP Strass (200.000 PE). The rejection-water from dewatering digested sludge contains ammonia concentrations from 1400 to 2000 mg N l
1. The sludge is conditioned by lime before dewatering which results in a pH of 12.7 at an alkalinity of about 130 mmol l
1. The treatment process is characterised by a strong variation of alkalinity which can only be handled by a reliable pH control [9]. Some

1106

B. Wett, W. Rauch / Water Research 37 (2003) 1100–1110

of the experimental data from the full-scale SBR pilot plant is analysed and simulated in following operation examples. The process strategy performs three SBR operation cycles per day. In each cycle the aeration phase is scheduled within a fixed time frame of 320 min; however, actual aeration is subject to a simple on-off controller based on the pH signal. During the whole aeration phase rejection-water is fed to the reactor at a constant flow rate. The pH value increases towards the upper pH setpoint due to the alkaline rejection-water and denitrification, then the aerator starts. The pH value in the reactor decreases from acidification during nitrification until the lower setpoint of the pH control is reached and the aerator stops (Fig. 4). This control mechanism proceeds until the storage tank is empty or the aeration phase ends. During the short mixing phase (30 min)

primary sludge is pumped into the reactor to enhance denitrification, which continues during the sedimentation (100 min) and the drawing off phase (30 min). Initially the pH-operation level was set between 7.0 and 7.3. This control strategy obtained nitrification rates far below the expectations and the influence of a limitation and/or inhibition was obvious. The process performance could not be improved until it was recognised, that at this pH level too much CO2 is stripped by intensive aeration. The following increase of the pH-operation level from 7.0opHo7.3 to 7.3opHo7.6 almost tripled the nitrification rate. The reason being that the pH increase from 7.0 to 7.3 reduced the portion of CO2—in fact the only strippable dissociation species of carbonic acid—from 19.0% to 10.5% of the total inorganic carbon. Decreased CO2 stripping caused an accumulation of inorganic carbon

Fig. 4. Simulation example with pH-, nitrogen- and oxygen profiles: Intermittent aeration between pH-setpoints 7.3opHo7.6 (Vmax ¼ 640 m3, NH3 influent=1450 mg N l
1, SS=16 g l
1, SRT=50 d, SVI=40 ml g
1, T¼ 231C).

B. Wett, W. Rauch / Water Research 37 (2003) 1100–1110

during the subsequent operation cycles. Hence the bicarbonate was no longer within the rate-limiting concentration range. The concentrations of nitrate and nitrite at the presented day of the simulation example (Fig. 4) start from zero because of the preceding weekend break. Due to insufficient load equalisation all nitrogen compounds exhibit strong variations over the period. Due to the limited aerator capacity and the high respiration rate, the oxygen setpoint of 2 mgO2 l
1 could not be reached (Fig. 5). The pH value in the reactor is mainly determined by the balance of the added alkaline rejection-water and the hydronium ion production during nitrification. Hence, nitrification (aeration respectively) is interrupted by the pH control as soon as the available alkalinity is consumed. No further oxidation of ammonia is possible until alkalinity is recovered by the rejection-water flow and denitrification. In other words, the maximum ammonia-elimination rate is determined by the initial alkalinity of the rejection-water unless an external OH
source is added. At the end of the SBR the discharge of treated wastewater with high inorganic carbon concentrations at a high pH value results in a significant loss of alkalinity to the system and should be avoided. A decrease of the pH-operation level at the end of the aeration phase causes CO2 stripping and reduces the total inorganic carbon concentration. The inherent alkalinity of the carbonic buffer system is utilised for nitrification before the liquid is discharged. Model prediction (Fig. 6) revealed that these effects cause an increase of the maximum ammonia-elimination rate from 75% to 89%. When the pH-decrease (6.5opHo6.8) at the end of the cycle was implemented in the pH-control, measurements indicated an elimination rate of 86% which confirmed the theory. To sum up, since the pH control maintains the bicarbonate

1107

concentration just above the limiting range until the end of the aeration phase, almost all of the available bicarbonate is used to buffer nitrification.

3.2. Model validation The presented model was developed and calibrated with long-term data sets from various control strategies run at the pilot plant in Strass. Validation was achieved by applying the model to another full-scale system at the WWTP Salzburg. The treated wastewater is there composed from two components—rejection-water from sludge digestion and leachate from a landfill. Since the digested sludge is conditioned by polymers instead of lime, the pH value of the wastewater is near 8.0 and the dissolved inorganic carbon concentration DIC is 800 mg C l
1 (compare Strass: pH=12.7, DIC=50 mg l
1). Because of the relatively low NH+ 4 -N/COD ratio (640/ 2500 during the measurement period) no external carbon source or primary sludge was required. The treatment strategy and process control implemented at WWTP Salzburg used a different pHoperation level as described before. Due to the high DIC concentration the pH control of the aeration system has been set at 6.8opHo7.1, i.e. a pH range which would cause excessive CO2 stripping and consequently a lack of inorganic carbon in Strass. But here the high DIC flux during wastewater addition equalises the CO2 stripping. In order to investigate the range of bicarbonate limitation of nitrifiers, the pH control was switched off. Without pH control the aeration interval lasted from 80 min after the beginning till the end of the aeration phase. Fig. 7 presents the pH- and nitrogen profiles and gives a close-up of the second operation cycle together with oxygen and inorganic carbon profiles. It should be noted that simulations have been

Fig. 5. On-line measurement of pH value, air supply and oxygen concentration (compare Fig. 4) (oxygen sensor was interfered by high NOx-level during the anoxic settling period of the 3rd operation cycle).

1108

B. Wett, W. Rauch / Water Research 37 (2003) 1100–1110

Fig. 6. Simulation example after the implementation of the pH decrease at the end of the aeration phase.

run with the same set of parameters as in Strass (also the highly sensitive stripping parameters). After the aeration was switched on, the pH value drops quickly to about 7.2 where the bicarbonate buffer starts to take effect. Both the calculated and the measured inorganic carbon concentration in the reactor are close to 200 mg C l
1 and the nitrification rate is at its maximum (100 mg N l
1 h
1). When the bicarbonate concentration is decreased by stripping below 100 mg C l
1 the nitrification rate starts to decline and reduced respiration affects the controlled air-flow. Constant ammonia concentration at this state of the operation cycle indicates a balance between nitrification and ammonia input and between CO2 stripping and bicarbonate input from the wastewater flow. After 270 min the wastewater influent flow is stopped and pH value and ammonia concentration decrease to the next level of balance. Here at a bicarbonate concentration of 25 mg C l
1 nitrification is slowed down close to zero and the air-flow is dropped to a quarter of the maximum. The overall behaviour of the process confirms the assumptions of the model. 3.3. Discussing inorganic carbon limitation against NH3 limitation In literature an alternative approach is suggested in order to interpret low nitrification rates at relatively low pH values. Wiesmann [5] and Hellinga et al. [4] presented detailed investigations of the kinetics of the nitrogen elimination from concentrated ammonia flows. They argue that biomass can only transport the uncharged NH3 over its membrane and therefore NH3 and not NH+ serves as the actual electron donor 4 (substrate). Consequently, both authors formulate the substrate limitation of the ammonium oxidation by a

Monod term considering the actual NH3 concentration which is calculated from the pH-dependent ammonia dissociation. Hellinga et al. [4] quantify the NH3 half saturation constant (Monod constant) to 0.46 mg NH3N l
1 at a temperature of 351C. At pH 7 this is
1 equivalent to 40 mg NH+ 4 -N l , which is a very high value in comparison to the typical Monod constant of
1 1.0 mg NH+ at 101C and 201C as stated by Henze 4 -N l et al. [2]. Wiesman [5] reports from nitrification tests in a stirred tank reactor fed by a synthetic wastewater with
1 ammonia concentrations of 1000 mg NH+ and 4 -N l no relevant organic compounds. The measured NH+ 4 -N concentration in the reactor reached 750 mg N l
1 at a controlled pH value of 7.2. After the pH value had been increased to 7.8 an almost total ammonium removal was observed which was explained by higher NH3 concentrations at a higher pH. It is remarkable that here analogous boundary conditions are described as observed at WWTP Strass (pH-operation level between 7.0 and 7.3 and a lack of carbon). Similar to the experiments from Wiesman also in Strass an increase of the pH led to a significant increase of the nitrification rate. However, the results presented in this paper suggest that this effect is not caused by the increased NH3 concentration but instead due to decreased CO2 stripping and the dominant role of the inorganic carbon limitation.

4. Conclusion It is common knowledge that ‘‘growth and activity of nitrifying bacteria decrease dramatically below neutrality’’ [19]. However, the pH value is a parameter indicating a potential limitation but is not necessarily a limiting factor itself. Still it is difficult to distinguish the

B. Wett, W. Rauch / Water Research 37 (2003) 1100–1110

1109

Fig. 7. Rejection-water treatment at the WWTP Salzburg (600.000 PE) in a carbon limiting range when the pH control was switched
1
1
1 off: Vmax ¼ 1275 m3, NHþ 4in ¼ 642 mg N l , SS=13 g l , T¼ 261C; SVI=50 ml g , SRT=10 d.

direct impact of a low pH value from bicarbonate limitation caused by a low pH. Two model-based case studies of rejection-water treatment in Strass and Salzburg with completely different inorganic carbon concentrations and pH-operation levels could clearly identify bicarbonate limitation as a very relevant ratelimiting influence on nitrification. Suggested model approaches for bicarbonate limitation- and CO2 stripping kinetics proved reliable mathematical descriptions and extended the range of validity of a nitrogen

elimination model to extremely high ammonia concentrations at varying pH values.

References [1] Abeling U. Nitrogen elimination from industrial wastewater—denitrification via nitrite. Publ. of Inst. f. Water Management and Waste Techn. Univ. Hannover, No. 86, 1994 (in German).

1110

B. Wett, W. Rauch / Water Research 37 (2003) 1100–1110

[2] Henze M, Grady Jr, CPL, Gujer W, Marais GvR, Matsuo T. Activated sludge model No.1. IAWQ Scientific and Technical Reports 1, 1987. [3] Anthonisen AC, Loehr RC, Prakasam TBS, Srinath EG. Inhibition of nitrification by ammonia and nitrous acid. J WPCF 1976;48/5:835–52. [4] Hellinga C, Van Loosdrecht MCM, Heijnen JJ. Model based design of a novel process for nitrogen removal from concentrated flows. Math Comput Modellling Dyn Syst 1999;5/4:351–71. [5] Wiesmann U. Biological removal from wastewater. Fiechter A, editor. Advances in biochemical engineering, vol. 51, Springer, Berlin, 1994, p. 113–54. [6] Reichert P, Omlin M. On the usefulness of overparameterized ecological models. Ecol Modelling 1997;95(2-3):289–99. [7] Stumm W, Morgan JJ. Aquatic chemistry: chemical equilibria and rates in natural waters, 3rd ed.. New York: Wiley, 1996. [8] Koch G, Egli K, Van der Meer JR, Siegrist H. Mathematical modelling of autotrophic denitrification in a nitrifying biofilm of a rotating biological contactor. Water Sci Technol, 2000;41/4-5:191–8. [9] Wett B, Rostek R, Rauch W, Ingerle K. PH-controlled reject-water-treatment. Water Sci Technol 1998;37/12:165–72. [10] Nowak O, Svardal K, Schweighofer P. The dynamic behaviour of nitrifying activated sludge systems influenced by inhibiting wastewater compounds. Water Sci Technol 1995;31/2:115–24. [11] James DH, Lee GF. Model of inorganic carbon limitation in natural waters. Water Air Soil Pollut 1974;3/3:315–20.

[12] Liehr SK, Eheart JW, Suidan MT. A modelling study of the effect of pH on carbon limited algal biofilms. Water Res 1988;22/8:1033–41. [13] Grguric G, Sondey CJ, DuVall BM. Carbon and nitrogen fluxes in a closed seawater facility. Sci Total Environ 2000;247/1:57–69. [14] Flora EMCV, Suidan MT, Flora JRV, Kim BJ. Speciation and chemical interactions in nitrifying biofilm. II: sensitivity analysis. J Environ Eng 1999;125/9:878–84. [15] Byong-Hee J, Yasunori T, Hajime U. Stimulating accumulation of nitrifying bacteria in porous carrier by addition of inorganic carbon in a continuous flow fluidized bed wastewater treatment reactor. J Biosci Bioeng 2000;89/ 4:334–9. [16] Bisset WP, Walsh JJ, Dieterle DA, Carder KL. Carbon cycling in the upper waters of the Sargasso Sea: I. Numerical simulation of differential carbon and nitrogen fluxes. Deep-Sea Res I 1999;46:205–69. [17] Wanninkhof R. Relationship between wind speed and gas exchange over the ocean. J Geophys Res 1992;97/C5: 7373–82. [18] Musvoto EV, Wentzel MC, Ekama GA. Integrated chemical–physical processes modelling—II. Simulating aeration treatment of anorganic digester supernatants. Water Res 2000;34/6:1868–80. [19] Hagopian DS, Riley JG. A closer look at the bacteriology of nitrification. Aquac Eng 1998;18:223–44. [20] Winter R, Krauth K. Nitrification and denitrification of the split sewage flow from anaerobic sludge treatment. Korrespon Abwass 1996;6:1032–39.