- Email: [email protected]
Heterotroph anoxic yield in anoxic aerobic activated sludge systems treating municipal wastewater
Water Research 37 (2003) 2435–2441
Heterotroph anoxic yield in anoxic aerobic activated sludge systems treating municipal wastewater A. Muller, M.C. Wentzel, R.E. Loewenthal, G.A. Ekama* Water Research Group, Department of Civil Engineering, Private Bag, University of Cape Town, Rondebosch, 7701, South Africa Received 19 April 2002; accepted 23 December 2002
Abstract As input to the steady state design and kinetic simulation models for the activated sludge system, the correct value for the heterotroph anoxic yield is essential to provide reliable estimates for the system denitrification potential. This paper examines activated sludge anoxic yield values in the literature, and presents experimental data quantifying the value. In the literature, in terms of the structure of ASM1 and similar models, theoretically it has been shown that the anoxic yield should be reduced to approximately 0.79 the value of the aerobic yield. This theoretical value is validated with data from corresponding aerobic OUR and anoxic nitrate time profiles in a batch fed laboratory scale long sludge age activated sludge system treating municipal wastewater. The value also is in close agreement with values in the literature measured with both artificial substrates and municipal wastewater. Thus, it is concluded that, in ASM1 and similar models, for an aerobic yield of 0.67 mg COD/mg COD, the anoxic yield should be about 0.53 mg COD/mg COD. Including such a lower anoxic yield in ASM1 and similar models will result in a significant increase in denitrification potential, due to increased denitrification with wastewater RBCOD as substrate. In terms of the structure of ASM3, for the proposed substrate storage yields and the aerobic yield of 0.63 mg COD/mg COD, experimental data indicate that the corresponding anoxic yield should be about 0.42 mg COD/mg COD. This is significantly lower than the proposed value of 0.54 mg COD/mg COD, and requires further investigation. r 2003 Elsevier Science Ltd. All rights reserved. Keywords: Activated sludge; Municipal wastewater; Modelling; Heterotrophs; Anoxic yield; Aerobic yield
1. Introduction The considerable advantages of the biologically mediated process of denitrification (e.g. reduced environmental N loads, alkalinity recovery, reduced oxygen demand), have led to its widespread implementation in the single sludge activated sludge system, through the inclusion in the system of anoxic zones/reactors. To aid in the design and optimisation of this system, a number of steady state design (e.g. [1]) and kinetic simulation [2– 8] models have been developed. Critical as input to both sets of models is the value for the ordinary heterotrophic *Corresponding author. Fax: +27-21-689-7471. E-mail addresses: [email protected] (M.C. Wentzel), [email protected] (G.A. Ekama).
organism (OHO) anoxic yield coefficient (YH,NO). In the anoxic utilisation of substrate, this stoichiometric constant determines the proportion of substrate electrons that are passed to the terminal electron acceptor nitrate (reducing it to dinitrogen gas), and the proportion that are used in the synthesis of new cell mass. Hence, the value for the anoxic sludge yield influences both the mass of denitrification possible and the sludge production. However, in activated sludge systems treating municipal wastewaters usually the effect on sludge production is small, since the mass of sludge produced under anoxic conditions is small compared to that produced under aerobic conditions, due to the relatively low influent TKN/COD ratios [6]. The effect on denitrification achievable is more marked. For example, for one unit of COD consumed, with
0043-1354/03/$ - see front matter r 2003 Elsevier Science Ltd. All rights reserved. doi:10.1016/S0043-1354(03)00015-0
2436
A. Muller et al. / Water Research 37 (2003) 2435–2441
Nomenclature
SS,R
ASM
YH,NO
activated sludge model (variously numbered 1, 2, 2d, 3) CFCM continuously fed completely mixed ETC electron transport chain IFFD intermittently fed fill and draw, also called sequencing batch reactor, SBR IWA International Water Association MLE modified Ludzack Ettinger NU nitrate utilised (mg NO3–N/l) OHO ordinary heterotrophic organism OU oxygen utilised (mg O/l) OUR oxygen utilisation rate RBCOD readily biodegradable COD SBCOD slowly biodegradable COD
YH,NO=0.67 mg COD/mg COD then 0.115 mg NO3–N are denitrified and with YH,NO=0.54 then 0.161 mg NO3–N; this represents an increase in denitrification of 40%. Noting that for most anoxic/aerobic sequencing (MLE type) systems the influent readily biodegradable (RB)COD is completely consumed anoxically, this decrease in anoxic yield will result in a substantial increase in denitrification potential with RBCOD as substrate, which will reflect in a similar increase in system denitrification potential. With slowly biodegradable (SB)COD, whether hydrolysed and utilised directly [5], or via RBCOD (ASM1, [3]), or via RBCOD and subsequent substrate storage (ASM3, [8]), the rate limiting process in anoxic SBCOD substrate utilisation usually is calibrated from observed denitrification rates. Hence, for SBCOD if the anoxic yield value is changed, then this has to be compensated for by calibrating the appropriate SBCOD hydrolysis/storage/ utilisation rate to ensure that the denitrification rate remains equal to that observed experimentally. This means that, in calibration of the models, the reduced anoxic yield value must have little influence on the denitrification potential with SBCOD as substrate. For both types of substrates, if anoxic respirometric procedures (i.e. monitoring terminal electron consumption, NO3) are applied for substrate utilisation rate calibration (e.g. [2,9,10]), or wastewater RBCOD characterisation [11,9,12], if the anoxic yield value is in error then the applied measurement similarly will be in error [9]. In ASM1 [3] and similar activated sludge simulation models, the aerobic growth rate on RBCOD and the SBCOD hydrolysis rates are applied under anoxic conditions also, but multiplied by correction factors, Zg for RBCOD and Zh for SBCOD. In ASM3 [8], the aerobic storage of COD and subsequent growth on the stored COD also are applied under anoxic conditions, but both rates are multiplied by the
YH,O2 YSTO,NO YSTO,O2 Zg Zh ZNO
initial RBCOD concentration in the reactor after batch feeding (mg COD/l) heterotroph anoxic yield coefficient (mg COD/ mg COD) heterotroph aerobic yield coefficient (mg COD/mg COD) anoxic stored COD yield coefficient (mg COD/ mg COD) aerobic stored COD yield coefficient (mg COD/mg COD) heterotroph anoxic reduction factor for growth heterotroph anoxic reduction factor for SBCOD hydrolysis heterotroph anoxic reduction factor for storage and growth
correction factor ZNO; the SBCOD hydrolysis rate remains unchanged under anoxic compared to aerobic conditions. Thus, if the anoxic yield is in error, effectively Zg and Zh in ASM1 [9] and ZNO in ASM3 will similarly be in error. Also, if anoxic respirometric procedures are used for wastewater RBCOD characterisation, an error in the anoxic yield value will result in a corresponding error in RBCOD concentration [12]. From the above, the correct value for the anoxic yield is essential in the design and simulation of activated sludge wastewater treatment plants, and in anoxic respirometric wastewater characterisation. This paper examines activated sludge anoxic yield values in the literature and presents experimental data quantifying the value.
2. Value for heterotroph anoxic yield (YH,NO) In the IWA Task Group models ASM1 [3], ASM2 [13] and ASM2d [14] and similar models (e.g. [5,15]) for activated sludge systems, the heterotrophic yield coefficient is assumed to have the same value under anoxic and aerobic conditions. However, when nitrate serves as electron acceptor, ideally (in reality the ATP yields probably are lower, [16]) only 2 mol of ATP are formed per pair of electron moles transferred in the electron transport chain (ETC) to nitrate, compared to the 3 mol when the transfer is to oxygen ([17]; see [18] for a review of heterotrophic respiratory metabolism). This difference reduces the energy captured by the organism in the oxidation of substrate when nitrate serves as electron acceptor compared to oxygen, and hence effectively should reduce the anoxic yield coefficient compared to the aerobic value. Accepting this difference in ATP yield, Orhon et al. [9] theoretically estimated the anoxic to aerobic yield ratio based on bio-energetic principles
A. Muller et al. / Water Research 37 (2003) 2435–2441
set out by McCarty [19–21], and obtained anoxic to aerobic yield ratios of 0.79, 0.80, 0.80 and 0.85 for municipal wastewater, protein, lactate and carbohydrate substrates, respectively. Accepting the heterotroph aerobic yield value of 0.67 mg COD/mg COD conventionally adopted in ASM1 and similar models for activated sludge systems treating municipal type wastewaters, this gives an anoxic yield of 0.53 mg COD/ mg COD. Experimental work supports the theoretical assessment that the anoxic yield is lower than the aerobic yield. McClintock et al. [22] operated aerobic and anoxic (fed nitrate) artificial substrate (Bacto-peptone) fed systems in parallel over a range of sludge ages (1.5 to 15 d). At four of the sludge ages (1.5, 3, 6, 15 d) the anoxic system produced less sludge than the corresponding aerobic system (ratio anoxic to aerobic sludge production 0.61, 0.57, 0.61 and 0.74, respectively), while at the fifth sludge age (15 d) it produced slightly more (ratio 1.04). Using the model of Lawrence and McCarty [23], McClintock et al. derived an anoxic yield value of 0.272 mg VSS/mg COD and an aerobic yield value of 0.503 mg VSS/mg COD, giving an anoxic to aerobic yield ratio of 0.54. They also noted that (in terms of the model of Lawrence and McCarty) the anoxic endogenous decay rate (0.057/d) was significantly less than the aerobic value (0.111/d). The model of Lawrence and McCarty does not take into account endogenous residue generation [24]. Applying the theory of Marais and Ekama [24], which does include endogenous residue generation (as do the IWA Task Group models), to the data of McClintock et al., a heterotroph aerobic yield of 0.55 mg VSS/mg COD and an anoxic yield of 0.35 mg VSS/mg COD can be determined, giving a yield ratio of 0.64 (in both cases an endogenous decay rate of 0.24/d is determined, equal to the ‘‘standard’’ value [1]; different anoxic and aerobic decay rates do not provide good fit to the experimental data, which is contrary to the observations of Siegrist et al. [25]). The yield ratio is significantly lower than the theoretical ratio above, and may be due to the poor COD mass balances of the anoxic systems [6]. Kuba et al. [16] developed two parallel acetic acid fed enhanced cultures of phosphorus accumulating organisms (PAOs), aerobic PAOs in an anaerobic–aerobic sequencing batch reactor (SBR) and denitrifying PAOs in an anaerobic–anoxic SBR. From these studies, the ratio of the anoxic to aerobic sludge production was 0.74, which equals the yield ratio since all conditions in the two systems were identical except for the terminal electron acceptor; this is reasonably close to the theoretical value of Orhon et al. above. . et al. [12] evaluated parallel anoxic and Ubay C , okgor aerobic batch respirometric tests to quantify wastewater RBCOD. With equal heterotroph anoxic and aerobic yield values, they found that the RBCOD concentrations for municipal wastewater derived from the anoxic
2437
respirometric tests were consistently higher than the values derived from the equivalent aerobic tests, by about 1.14 times. They noted that this implied that the heterotroph anoxic yield was lower than the aerobic value. By equating the RBCOD concentrations in the parallel tests, they derived a value for the anoxic yield of 0.37 mg VSS/mg COD (0.53 mg COD/mg COD with . et al. as COD/VSS ratio accepted by Ubay C , okgor 1.42 mg COD/mg VSS) for an aerobic yield value of 0.45 mg VSS/mg COD (0.64 mg COD/mg COD for COD/VSS ratio=1.42 mg COD/mg VSS). This gives an anoxic to aerobic yield ratio of 0.82, which is very close to the theoretical value of Orhon et al. above. Sperandio et al. [26] developed an experimental procedure based on measuring carbon dioxide evolution rates to quantify aerobic and anoxic yields of heterotrophs. With this method, for the artificial substrates acetate, glucose and acetate/starch they obtained a heterotroph aerobic yield of 0.54 mg COD/mg COD for acetate, and 0.66–0.67 for the other two substrates; the latter values are close to the aerobic yield value used as default in most of the activated sludge design and simulation models. For the anoxic yield, values of 0.45 for acetate and (with one exception) 0.54–0.57 for the other substrates were obtained. This gives anoxic to aerobic yield ratios of 0.81–0.85, which spans the range of theoretical values determined by Orhon et al. for artificial substrates. The requirement for a heterotroph anoxic yield which is lower than the aerobic one has been recognised in the development of kinetic simulation models for the activated sludge system. Barker and Dold [6] presented a kinetic model for nutrient removal activated sludge systems in which separate anoxic and aerobic yields were included. Initially the two yields were given the same value of 0.666 mg COD/mg COD, but subsequently in implementation of the model in the BIOWIN computer programme [27] the anoxic yield value was changed first to 0.403 mg COD/mg COD and later to 0.54. This latter value gives an anoxic to aerobic yield ratio of 0.81 which is close to the theoretical value of Orhon et al. above. In ASM3 [8,28], the necessity for a reduced heterotroph anoxic yield was also recognised and included. In this model all COD substrate utilisation proceeds via an intracellular COD storage, with the stored substrate product serving as COD source for heterotroph growth. Both storage and growth processes have yield values, both of which differ between anoxic and aerobic conditions. For the storage process, the cited aerobic yield is 0.85 and the anoxic yield 0.80 mg COD/ mg COD, and for the growth process the respective yields are 0.63 and 0.54 mg COD/mg COD. This gives net aerobic and anoxic yields of 0.54 and 0.43 mg COD/ mg COD, respectively, and a net anoxic to aerobic yield ratio of 0.8. No guidance is given on the source for these values, except that it is accepted that the anoxic to
2438
A. Muller et al. / Water Research 37 (2003) 2435–2441
aerobic energy yield is 0.7. It is stated that these ‘‘values (values for ASM3 constants in general) are provided as examples and are not part of ASM3’’ [8]. The net anoxic to aerobic yield ratio is close to the theoretical value of Orhon et al. above. In summary, theoretically the value for the heterotroph anoxic yield is reduced to approximately 0.8–0.85 of the aerobic yield value. A lower heterotroph anoxic yield has been incorporated in the more recent activated sludge simulation models, which will have an impact on the predictions of system denitrification potentials. Experimentally, with both artificial substrates and municipal wastewaters there is some evidence supporting the lower anoxic yield. However, experimental data quantifying the reduced yield is limited. An experimental investigation to further substantiate and quantify the reduced anoxic yield with municipal wastewater as substrate is reported below.
3. Experimental investigation 3.1. Experimental methods To investigate the effect of feeding conditions on filamentous bulking, Ekama et al. [29] operated two parallel single reactor activated sludge systems, one as an intermittently fed fill and draw (IFFD, also called SBR), the other as continuously fed completely mixed (CFCM) with secondary settling tank and underflow recycle. For both systems, all the system parameters were the same, such as reactor volume (10 l), sludge age (20 d), temperature (20 C) and the mass of COD fed daily (5250 mg COD/d). Also, both systems were started up with the same Mitchell’s Plain wastewater treatment plant (WWTP) activated sludge and fed the same Mitchells Plain settled wastewater. This wastewater was collected in 1 m3 batches from the WWTP and, after maceration, stored in stainless steel tanks in the laboratory cold room at 4 C. The daily wastewater feed volume required was collected from the cold room, warmed to 20 C and split equally between the two systems, so that both systems received the same wastewater daily. (For more details on the systems see Table 1 in [29]). Alternating anoxic aerobic conditions were imposed on the systems, by alternating aeration with nonaeration over a 4 h cycle (3 h aeration on and 1 h aeration off). The feed to the CFCM system was continuous and with the aeration/non-aeration cycle, the sludge was exposed to the influent RBCOD under both anoxic and aerobic conditions. The IFFD system was batch fed once daily and, in order to expose the sludge to influent RBCOD under both aerobic and anoxic conditions, as in the CFCM system, the time of the daily feed was alternated in accordance with the
aeration/non-aeration cycle imposed on the system. On one day, the system was fed at the time it became aerobic and on the next day at the time it became anoxic 1 h earlier. In both the IFFD and CFCM systems, the DO concentration was maintained between 2 and 3 mg O/l during the aerobic periods. The two systems were operated for period of 70 d, during which five different sewage batches were used as influent, and the system performances were monitored daily [29]. Between days 60 and 70, five aerobic and three anoxic in situ profile tests were conducted on the IFFD system for the 2 h period immediately following the wastewater batch feed (for the anoxic profiles, the anoxic period following feeding was increased from 1 to 2 h). For the aerobic profiles, OURs were measured regularly. For the anoxic profiles, grab samples were taken at regular intervals, immediately filtered (0.45 mm) and HgCl2 added to the filtrate. The filtrate was analysed for nitrate and nitrite with the auto analyser industrial methods 33.69W and 35.69W, respectively (Technicon); in all experiments, nitrite concentrations were negligible. Additionally, nitrate concentrations were measured in situ at less than 5 min intervals with an Orion nitrate selective electrode, as mV. The mV-time profiles were calibrated against the nitrate concentrations of the grab samples. 3.2. Results and discussion Typical OUR and nitrate time plots for the aerobic and anoxic profiles are shown in Figs. 1 and 2, respectively. The OUR versus time profiles (Fig. 1) showed an initial rapid approximately constant OUR (80–110 mg O/l h) for the first 25–50 min, followed by a precipitous decrease when the influent RBCOD was completely utilised. Similarly, the anoxic profiles (Fig. 2) showed an initial rapid decrease in nitrate concentration over time for the first 50–80 min until the influent RBCOD was completely utilised, after which a much slower nitrate concentration decrease with time took place. Parallel batch tests on sludge harvested from the CFCM system are not useful for this paper, because the 2 h batch test time period was insufficient to utilise all the influent RBCOD (see [29]). In the aerobic profile results, the area below the OUR–time curve, but above the line extended from the second plateau to the vertical axis, is the oxygen utilised (OU) in the consumption of the influent wastewater RBCOD concentration, see Fig. 1. In the five aerobic batch tests, the OUs were 36.7, 37.3, 38.7, 29.1 and 41.7 giving a mean of 36.6 mg O/l. In the anoxic profiles, the distance on the vertical axis between the lines drawn through the initial rapid and second slower nitrate concentration reduction results is the nitrate utilised (NU) in consumption of the influent RBCOD
A. Muller et al. / Water Research 37 (2003) 2435–2441
2439
In terms of ASM1 (and similar models), for the utilisation of RBCOD:
Fig. 1. Typical OUR versus time profiles following batch feeding under aerobic conditions of IFFD activated sludge system; feed raw municipal wastewater.
OU ¼ SS;R ð1 YH;O2 Þ and
ð1aÞ
NU ¼ SS;R ð1 YH;NO Þ=2:86;
ð1bÞ
where SS,R is the initial RBCOD concentration in the reactor after batch feeding (mg COD/l), OU the oxygen consumed for RBCOD utilisation (mg O/l), NU the nitrate consumed for RBCOD utilisation (mg NO3–N/l), YH,O2 the heterotroph yield under aerobic conditions (mg COD/mg COD), YH,NO the heterotroph yield under anoxic conditions (mg COD/mg COD). Accepting that the initial reactor RBCOD concentration after the batch feed is equal for the anoxic and aerobic profiles, which is reasonable because the same wastewater batch was used in the tests at the same loading rate, and the aerobic and anoxic profile tests were done alternately over the 10 days (see [29]), then equating Eqs. (1a) and (1b) and solving, we get 2:86dNU YH;NO ¼ 1 ð1 YH:O2 Þ: ð2Þ OU Substituting 36.6 mg O/l for OU and 18.3 mg NO3–N/l for NU into Eq. (2) yields YH;NO ¼ 0:43 þ 1:43YH;O2 :
ð3Þ
Eq. (3) is shown plotted in Fig. 3. From Eq. (3), accepting from ASM1 that YH,O2=0.67 mg COD/ mg COD, then YH,NO=0.53 mg COD/mg COD and the anoxic to aerobic yield ratio is 0.79. This anoxic yield value is very close to the range of values measured by Sperandio et al. [26] with artificial substrates, who measured an equivalent aerobic yield of 0.67 mg COD/ mg COD which is equal to the value accepted here. Further, the anoxic to aerobic yield ratio (0.79) is equal to the theoretical value of Orhon et al. [9] for municipal
Fig. 2. Typical nitrate versus time profiles following batch feeding under anoxic conditions of IFFD activated sludge system; feed raw municipal wastewater.
concentration, see Fig. 2. For the three anoxic profiles on the IFFD system, the NUs were 226=16, 5131=22 and 3518=17 mg NO3–N/l, giving a mean of 18.3 mg NO3–N/l.
Fig. 3. Heterotroph anoxic yield versus aerobic yield for the OU and NU values measured in this experimental investigation, in terms of ASM1 [3] and ASM3 [8]. Also shown are the anoxic yield values for the aerobic yield values proposed in the two models.
2440
A. Muller et al. / Water Research 37 (2003) 2435–2441
wastewater (0.79), and reasonably close to the value . et al. [12] also on municipal measured by Ubay C , okgor wastewater (0.82). In terms of ASM3 [8], for the utilisation of RBCOD (assuming complete utilisation of the stored substrate generated from RBCOD uptake): OU ¼ SS;R ð1 YSTO;O2 YH;O2 Þ
and
NU ¼ SS;R ð1 YSTO;NO YH;NO Þ=2:86;
ð4aÞ ð4bÞ
where YSTO,O2 is the stored COD yield under aerobic conditions (mg COD/mg COD), YSTO,NO the stored COD yield under anoxic conditions (mg COD/ mg COD). Similar to the above YH;NO 2:86dNU ð1 YSTO;O2 YH;O2 Þ =YSTO;NO : ð5Þ ¼ 1 OU Accepting the values proposed by the IWA Task Group of YSTO,O2=0.85 and YSTO,NO=0.80 mg COD/ mg COD and the experimentally measured NU and OU, Eq. (5) is shown plotted in Fig. 3 also. From Eq. (5), accepting from ASM3 YH,O2=0.63 mg COD/mg COD, then YH,NO=0.42 mg COD/mg COD, and the anoxic to aerobic yield ratio is 0.67. Both the anoxic yield and yield ratio values are significantly lower than those measured by Sperandio et al. [26] of 0.54–0.57 and 0.81– 0.85, respectively, and those suggested by the Task Group of 0.54 mg COD/mg COD and 0.86, respectively.
experimentally. This can be done by changing the value of the reduction factor for anoxic SBCOD hydrolysis/ utilisation, Zh. The reduction in anoxic yield most likely will have a lesser effect on the net sludge production, since in most conventional N removal activated sludge systems, the mass of sludge produced under anoxic conditions is small compared to that produced under aerobic conditions. Further, a reduced anoxic endogenous respiration rate [25] may compensate for the reduced anoxic yield. In ASM3, accepting the proposed heterotroph aerobic yield of 0.63 mg COD/mg COD, then from the experimental data presented here (which corresponds closely to similar investigations) the anoxic yield should be approximately 0.42 mg COD/mg COD, giving an anoxic to aerobic yield ratio of 0.67. Both values are significantly lower than those suggested by the Task Group, of 0.54 and 0.86, respectively. This requires further investigation.
Acknowledgements This research was supported by the Water Research Commission, the National Research Foundation, THRIP and Water and Sanitation Services South Africa (Pty) Ltd (a subsidiary of Ondeo Services), and is published with their permission.
References 4. Conclusions Both theoretically and experimentally, with both artificial and municipal wastewaters as substrate, there is sufficient evidence that the value for the heterotroph anoxic yield should be reduced compared to the value for the aerobic yield. Theoretically and experimentally, the anoxic to aerobic yield ratio appears to fall into the range 0.78–0.85. In ASM1 (and similar models), accepting the value for the aerobic yield as that conventionally applied, 0.67 mg COD/mg COD, then the anoxic yield value should fall into the range 0.52– 0.57; 0.53 mg COD/mg COD determined here. Changing the anoxic yield value from the current ASM1 aerobic yield value of 0.67 mg COD/mg COD to 0.53 mg COD/ mg COD will result in a significant increase in the denitrification potential of the system, due to increased denitrification with wastewater RBCOD as substrate. If such a change is implemented in ASM1 (and similar models), then it must be remembered to change the SBCOD hydrolysis/utilisation rate in concert, since this rate effectively has been calibrated from the measured denitrification rate, and the predicted denitrification rate with SBCOD must be kept equal to that observed
[1] WRC. Theory, design and operation of nutrient removal activated sludge processes. Water Research Commission, Pretoria, RSA, 1984. [2] Van Haandel AC, Ekama GA, Marais GvR. The activated sludge process 3—single sludge denitrification. Water Res 1981;15(10):1135–52. [3] Henze M, Grady Jr CPL, Gujer W, Marais GvR, Matsuo T. Activated sludge model no. 1. IAWPRC Scientific and Technical Report No. 1, IAWPRC, London, 1987. [4] Dold PL, Ekama GA, Marais GvR. A general model for the activated sludge process. Prog Water Technol 1980;12(Tor):47–77. [5] Dold PL, Wentzel MC, Billing AE, Ekama GA, Marais GvR. Activated sludge simulation programs: Nitrification and nitrification/denitrification systems (Version 1.0). Water Research Commission, Pretoria, South Africa, 1991. TT 52/91, ISBN 0 947447 19 9. [6] Barker PS, Dold PL. General model for biological nutrient removal activated-sludge systems: model presentation. Water Environ Res 1997;69(5):969–84. [7] Murnleitner E, Kuba T, van Loosdrecht MCM, Heinen JJ. An integrated metabolic model for aerobic and denitrifying biological phosphorus removal. Biotechnol Bioeng 1997;54(5):434–50.
A. Muller et al. / Water Research 37 (2003) 2435–2441 [8] Gujer W, Henze M, Mino T, Van Loosdrecht MCM. Activated sludge Model No. 3. Water Sci Technol 1999; 39(1):183–93. . [9] Orhon D, Sozen S, Artan N. The effect of heterotrophic yield on the assessment of the correction factor for anoxic growth. Water Sci Technol 1996;34(5–6):67–74. . . E, Orhon D, Henze M. Respiro[10] Sozen S, Ubay 9Cokgor metric analysis of activated sludge behaviour—II. Heterotrophic growth under aerobic and anoxic conditions. Water Res 1998;32(2):476–88. [11] Ekama GA, Dold PL, Marais GvR. Procedures for determining influent COD fractions and the maximum specific growth rate of heterotrophs in activated sludge systems. Water Sci Technol 1986;18(6):91–114. . E, Sozen . [12] Ubay 9Cokgor S, Orhon D, Henze M. Respirometric analysis of activated sludge behaviour—I Assessment of the readily biodegradable substrate. Water Res 1998;32(2):461–75. [13] Henze M, Gujer W, Mino T, Matsuo T, Wentzel MC, Marais GvR. Activated sludge model No.2. IAWQ Scientific and Technical Report No 3, IAWQ, London, 1995. [14] Henze M, Gujer W, Mino T, Matsuo T, Wentzel MC, Marais GvR, Van Loosdrecht MCM. Activated sludge model no. 2d. Water Sci Technol 1999;39(1): 165–82. [15] Wentzel MC, Ekama GA, Marais GvR. Processes and modelling of nitrification denitrification biological excess phosphorus removal systems. Water Sci Technol 1992; 25(6):59–82. [16] Kuba T, Smolders GJF, van Loosdrecht MCM, Heinen JJ. Biological phosphorus removal from wastewater by anaerobic-anoxic sequencing batch reactor. Water Sci Technol 1993;27(5/6):241–52. [17] Payne WJ. Denitrification. New York: Wiley, 1981. [18] Casey TG, Wentzel MC, Ekama GA. Filamentous organism bulking in nutrient removal activated sludge systems. Paper 10: metabolic behaviour of heterotrophic
[19]
[20]
[21] [22]
[23]
[24]
[25]
[26]
[27] [28]
[29]
2441
facultative aerobic organisms under aerated/unaerated conditions. Water SA 1999;25(4):425–42. McCarty PL. Energetics and bacterial growth. In: Faust SD, Hunter JV, editors. Organic compounds in aquatic environments. New York, NY, USA: Marcell Dekker, 1971. McCarty PL. Energetics of organic matter degradation. In: Mitchell R, editor. Water pollution microbiology. New York, NY, USA: Wiley, 1972. McCarty PL. Stoichiometry of biological reactions. Prog Water Technol 1975;7(1):157–72. McClintock SA, Sherrard JH, Novak JT, Randall CW. Nitrate versus oxygen respiration in the activated sludge process. J WPCF 1988;60(3):342–50. Lawrence AW, McCarty PL. Unified basis for biological treatment design and operation. J Sanit Eng Div ASCE 1970;96(SA3):757–78. Marais GvR, Ekama GA. The activated sludge process Part 1—steady state behaviour. Water SA 1976;2(4): 163–200. Siegrist H, Brunner I, Koch G, Linh Con Phan, Van Chieu LE. Reduction of biomass decay rate under anoxic and anaerobic conditions. Water Sci Technol 1999; 39(1): 129–137. Sperandio M, Urbain V, Audic J-M, Paul E. Use of carbon dioxide evolution rate for denitrifying heterotrophic yield and characterizing denitrifying biomass. Water Sci Technol 1999;39(1):139–46. Envirosim. BIOWIN 32 process simulator. Flamborough, Ontario, Canada, 2001. Henze M, Gujer W, Mino T, Van Loosdrecht MCM. Activated sludge models ASM1, ASM2, ASM2d and ASM3. IWA Scientific and Technical Report No. 9, IWA, London, 2000. Ekama GA, Wentzel MC, Casey TG, Marais GvR. Filamentous organism bulking in nutrient removal activated sludge systems Paper 3—stimulation of the selector effect under anoxic conditions. Water SA 1996;22(2):119–26.
Heterotroph anoxic yield in anoxic aerobic activated sludge systems treating municipal wastewater A. Muller, M.C. Wentzel, R.E. Loewenthal, G.A. Ekama* Water Research Group, Department of Civil Engineering, Private Bag, University of Cape Town, Rondebosch, 7701, South Africa Received 19 April 2002; accepted 23 December 2002
Abstract As input to the steady state design and kinetic simulation models for the activated sludge system, the correct value for the heterotroph anoxic yield is essential to provide reliable estimates for the system denitrification potential. This paper examines activated sludge anoxic yield values in the literature, and presents experimental data quantifying the value. In the literature, in terms of the structure of ASM1 and similar models, theoretically it has been shown that the anoxic yield should be reduced to approximately 0.79 the value of the aerobic yield. This theoretical value is validated with data from corresponding aerobic OUR and anoxic nitrate time profiles in a batch fed laboratory scale long sludge age activated sludge system treating municipal wastewater. The value also is in close agreement with values in the literature measured with both artificial substrates and municipal wastewater. Thus, it is concluded that, in ASM1 and similar models, for an aerobic yield of 0.67 mg COD/mg COD, the anoxic yield should be about 0.53 mg COD/mg COD. Including such a lower anoxic yield in ASM1 and similar models will result in a significant increase in denitrification potential, due to increased denitrification with wastewater RBCOD as substrate. In terms of the structure of ASM3, for the proposed substrate storage yields and the aerobic yield of 0.63 mg COD/mg COD, experimental data indicate that the corresponding anoxic yield should be about 0.42 mg COD/mg COD. This is significantly lower than the proposed value of 0.54 mg COD/mg COD, and requires further investigation. r 2003 Elsevier Science Ltd. All rights reserved. Keywords: Activated sludge; Municipal wastewater; Modelling; Heterotrophs; Anoxic yield; Aerobic yield
1. Introduction The considerable advantages of the biologically mediated process of denitrification (e.g. reduced environmental N loads, alkalinity recovery, reduced oxygen demand), have led to its widespread implementation in the single sludge activated sludge system, through the inclusion in the system of anoxic zones/reactors. To aid in the design and optimisation of this system, a number of steady state design (e.g. [1]) and kinetic simulation [2– 8] models have been developed. Critical as input to both sets of models is the value for the ordinary heterotrophic *Corresponding author. Fax: +27-21-689-7471. E-mail addresses: [email protected] (M.C. Wentzel), [email protected] (G.A. Ekama).
organism (OHO) anoxic yield coefficient (YH,NO). In the anoxic utilisation of substrate, this stoichiometric constant determines the proportion of substrate electrons that are passed to the terminal electron acceptor nitrate (reducing it to dinitrogen gas), and the proportion that are used in the synthesis of new cell mass. Hence, the value for the anoxic sludge yield influences both the mass of denitrification possible and the sludge production. However, in activated sludge systems treating municipal wastewaters usually the effect on sludge production is small, since the mass of sludge produced under anoxic conditions is small compared to that produced under aerobic conditions, due to the relatively low influent TKN/COD ratios [6]. The effect on denitrification achievable is more marked. For example, for one unit of COD consumed, with
0043-1354/03/$ - see front matter r 2003 Elsevier Science Ltd. All rights reserved. doi:10.1016/S0043-1354(03)00015-0
2436
A. Muller et al. / Water Research 37 (2003) 2435–2441
Nomenclature
SS,R
ASM
YH,NO
activated sludge model (variously numbered 1, 2, 2d, 3) CFCM continuously fed completely mixed ETC electron transport chain IFFD intermittently fed fill and draw, also called sequencing batch reactor, SBR IWA International Water Association MLE modified Ludzack Ettinger NU nitrate utilised (mg NO3–N/l) OHO ordinary heterotrophic organism OU oxygen utilised (mg O/l) OUR oxygen utilisation rate RBCOD readily biodegradable COD SBCOD slowly biodegradable COD
YH,NO=0.67 mg COD/mg COD then 0.115 mg NO3–N are denitrified and with YH,NO=0.54 then 0.161 mg NO3–N; this represents an increase in denitrification of 40%. Noting that for most anoxic/aerobic sequencing (MLE type) systems the influent readily biodegradable (RB)COD is completely consumed anoxically, this decrease in anoxic yield will result in a substantial increase in denitrification potential with RBCOD as substrate, which will reflect in a similar increase in system denitrification potential. With slowly biodegradable (SB)COD, whether hydrolysed and utilised directly [5], or via RBCOD (ASM1, [3]), or via RBCOD and subsequent substrate storage (ASM3, [8]), the rate limiting process in anoxic SBCOD substrate utilisation usually is calibrated from observed denitrification rates. Hence, for SBCOD if the anoxic yield value is changed, then this has to be compensated for by calibrating the appropriate SBCOD hydrolysis/storage/ utilisation rate to ensure that the denitrification rate remains equal to that observed experimentally. This means that, in calibration of the models, the reduced anoxic yield value must have little influence on the denitrification potential with SBCOD as substrate. For both types of substrates, if anoxic respirometric procedures (i.e. monitoring terminal electron consumption, NO3) are applied for substrate utilisation rate calibration (e.g. [2,9,10]), or wastewater RBCOD characterisation [11,9,12], if the anoxic yield value is in error then the applied measurement similarly will be in error [9]. In ASM1 [3] and similar activated sludge simulation models, the aerobic growth rate on RBCOD and the SBCOD hydrolysis rates are applied under anoxic conditions also, but multiplied by correction factors, Zg for RBCOD and Zh for SBCOD. In ASM3 [8], the aerobic storage of COD and subsequent growth on the stored COD also are applied under anoxic conditions, but both rates are multiplied by the
YH,O2 YSTO,NO YSTO,O2 Zg Zh ZNO
initial RBCOD concentration in the reactor after batch feeding (mg COD/l) heterotroph anoxic yield coefficient (mg COD/ mg COD) heterotroph aerobic yield coefficient (mg COD/mg COD) anoxic stored COD yield coefficient (mg COD/ mg COD) aerobic stored COD yield coefficient (mg COD/mg COD) heterotroph anoxic reduction factor for growth heterotroph anoxic reduction factor for SBCOD hydrolysis heterotroph anoxic reduction factor for storage and growth
correction factor ZNO; the SBCOD hydrolysis rate remains unchanged under anoxic compared to aerobic conditions. Thus, if the anoxic yield is in error, effectively Zg and Zh in ASM1 [9] and ZNO in ASM3 will similarly be in error. Also, if anoxic respirometric procedures are used for wastewater RBCOD characterisation, an error in the anoxic yield value will result in a corresponding error in RBCOD concentration [12]. From the above, the correct value for the anoxic yield is essential in the design and simulation of activated sludge wastewater treatment plants, and in anoxic respirometric wastewater characterisation. This paper examines activated sludge anoxic yield values in the literature and presents experimental data quantifying the value.
2. Value for heterotroph anoxic yield (YH,NO) In the IWA Task Group models ASM1 [3], ASM2 [13] and ASM2d [14] and similar models (e.g. [5,15]) for activated sludge systems, the heterotrophic yield coefficient is assumed to have the same value under anoxic and aerobic conditions. However, when nitrate serves as electron acceptor, ideally (in reality the ATP yields probably are lower, [16]) only 2 mol of ATP are formed per pair of electron moles transferred in the electron transport chain (ETC) to nitrate, compared to the 3 mol when the transfer is to oxygen ([17]; see [18] for a review of heterotrophic respiratory metabolism). This difference reduces the energy captured by the organism in the oxidation of substrate when nitrate serves as electron acceptor compared to oxygen, and hence effectively should reduce the anoxic yield coefficient compared to the aerobic value. Accepting this difference in ATP yield, Orhon et al. [9] theoretically estimated the anoxic to aerobic yield ratio based on bio-energetic principles
A. Muller et al. / Water Research 37 (2003) 2435–2441
set out by McCarty [19–21], and obtained anoxic to aerobic yield ratios of 0.79, 0.80, 0.80 and 0.85 for municipal wastewater, protein, lactate and carbohydrate substrates, respectively. Accepting the heterotroph aerobic yield value of 0.67 mg COD/mg COD conventionally adopted in ASM1 and similar models for activated sludge systems treating municipal type wastewaters, this gives an anoxic yield of 0.53 mg COD/ mg COD. Experimental work supports the theoretical assessment that the anoxic yield is lower than the aerobic yield. McClintock et al. [22] operated aerobic and anoxic (fed nitrate) artificial substrate (Bacto-peptone) fed systems in parallel over a range of sludge ages (1.5 to 15 d). At four of the sludge ages (1.5, 3, 6, 15 d) the anoxic system produced less sludge than the corresponding aerobic system (ratio anoxic to aerobic sludge production 0.61, 0.57, 0.61 and 0.74, respectively), while at the fifth sludge age (15 d) it produced slightly more (ratio 1.04). Using the model of Lawrence and McCarty [23], McClintock et al. derived an anoxic yield value of 0.272 mg VSS/mg COD and an aerobic yield value of 0.503 mg VSS/mg COD, giving an anoxic to aerobic yield ratio of 0.54. They also noted that (in terms of the model of Lawrence and McCarty) the anoxic endogenous decay rate (0.057/d) was significantly less than the aerobic value (0.111/d). The model of Lawrence and McCarty does not take into account endogenous residue generation [24]. Applying the theory of Marais and Ekama [24], which does include endogenous residue generation (as do the IWA Task Group models), to the data of McClintock et al., a heterotroph aerobic yield of 0.55 mg VSS/mg COD and an anoxic yield of 0.35 mg VSS/mg COD can be determined, giving a yield ratio of 0.64 (in both cases an endogenous decay rate of 0.24/d is determined, equal to the ‘‘standard’’ value [1]; different anoxic and aerobic decay rates do not provide good fit to the experimental data, which is contrary to the observations of Siegrist et al. [25]). The yield ratio is significantly lower than the theoretical ratio above, and may be due to the poor COD mass balances of the anoxic systems [6]. Kuba et al. [16] developed two parallel acetic acid fed enhanced cultures of phosphorus accumulating organisms (PAOs), aerobic PAOs in an anaerobic–aerobic sequencing batch reactor (SBR) and denitrifying PAOs in an anaerobic–anoxic SBR. From these studies, the ratio of the anoxic to aerobic sludge production was 0.74, which equals the yield ratio since all conditions in the two systems were identical except for the terminal electron acceptor; this is reasonably close to the theoretical value of Orhon et al. above. . et al. [12] evaluated parallel anoxic and Ubay C , okgor aerobic batch respirometric tests to quantify wastewater RBCOD. With equal heterotroph anoxic and aerobic yield values, they found that the RBCOD concentrations for municipal wastewater derived from the anoxic
2437
respirometric tests were consistently higher than the values derived from the equivalent aerobic tests, by about 1.14 times. They noted that this implied that the heterotroph anoxic yield was lower than the aerobic value. By equating the RBCOD concentrations in the parallel tests, they derived a value for the anoxic yield of 0.37 mg VSS/mg COD (0.53 mg COD/mg COD with . et al. as COD/VSS ratio accepted by Ubay C , okgor 1.42 mg COD/mg VSS) for an aerobic yield value of 0.45 mg VSS/mg COD (0.64 mg COD/mg COD for COD/VSS ratio=1.42 mg COD/mg VSS). This gives an anoxic to aerobic yield ratio of 0.82, which is very close to the theoretical value of Orhon et al. above. Sperandio et al. [26] developed an experimental procedure based on measuring carbon dioxide evolution rates to quantify aerobic and anoxic yields of heterotrophs. With this method, for the artificial substrates acetate, glucose and acetate/starch they obtained a heterotroph aerobic yield of 0.54 mg COD/mg COD for acetate, and 0.66–0.67 for the other two substrates; the latter values are close to the aerobic yield value used as default in most of the activated sludge design and simulation models. For the anoxic yield, values of 0.45 for acetate and (with one exception) 0.54–0.57 for the other substrates were obtained. This gives anoxic to aerobic yield ratios of 0.81–0.85, which spans the range of theoretical values determined by Orhon et al. for artificial substrates. The requirement for a heterotroph anoxic yield which is lower than the aerobic one has been recognised in the development of kinetic simulation models for the activated sludge system. Barker and Dold [6] presented a kinetic model for nutrient removal activated sludge systems in which separate anoxic and aerobic yields were included. Initially the two yields were given the same value of 0.666 mg COD/mg COD, but subsequently in implementation of the model in the BIOWIN computer programme [27] the anoxic yield value was changed first to 0.403 mg COD/mg COD and later to 0.54. This latter value gives an anoxic to aerobic yield ratio of 0.81 which is close to the theoretical value of Orhon et al. above. In ASM3 [8,28], the necessity for a reduced heterotroph anoxic yield was also recognised and included. In this model all COD substrate utilisation proceeds via an intracellular COD storage, with the stored substrate product serving as COD source for heterotroph growth. Both storage and growth processes have yield values, both of which differ between anoxic and aerobic conditions. For the storage process, the cited aerobic yield is 0.85 and the anoxic yield 0.80 mg COD/ mg COD, and for the growth process the respective yields are 0.63 and 0.54 mg COD/mg COD. This gives net aerobic and anoxic yields of 0.54 and 0.43 mg COD/ mg COD, respectively, and a net anoxic to aerobic yield ratio of 0.8. No guidance is given on the source for these values, except that it is accepted that the anoxic to
2438
A. Muller et al. / Water Research 37 (2003) 2435–2441
aerobic energy yield is 0.7. It is stated that these ‘‘values (values for ASM3 constants in general) are provided as examples and are not part of ASM3’’ [8]. The net anoxic to aerobic yield ratio is close to the theoretical value of Orhon et al. above. In summary, theoretically the value for the heterotroph anoxic yield is reduced to approximately 0.8–0.85 of the aerobic yield value. A lower heterotroph anoxic yield has been incorporated in the more recent activated sludge simulation models, which will have an impact on the predictions of system denitrification potentials. Experimentally, with both artificial substrates and municipal wastewaters there is some evidence supporting the lower anoxic yield. However, experimental data quantifying the reduced yield is limited. An experimental investigation to further substantiate and quantify the reduced anoxic yield with municipal wastewater as substrate is reported below.
3. Experimental investigation 3.1. Experimental methods To investigate the effect of feeding conditions on filamentous bulking, Ekama et al. [29] operated two parallel single reactor activated sludge systems, one as an intermittently fed fill and draw (IFFD, also called SBR), the other as continuously fed completely mixed (CFCM) with secondary settling tank and underflow recycle. For both systems, all the system parameters were the same, such as reactor volume (10 l), sludge age (20 d), temperature (20 C) and the mass of COD fed daily (5250 mg COD/d). Also, both systems were started up with the same Mitchell’s Plain wastewater treatment plant (WWTP) activated sludge and fed the same Mitchells Plain settled wastewater. This wastewater was collected in 1 m3 batches from the WWTP and, after maceration, stored in stainless steel tanks in the laboratory cold room at 4 C. The daily wastewater feed volume required was collected from the cold room, warmed to 20 C and split equally between the two systems, so that both systems received the same wastewater daily. (For more details on the systems see Table 1 in [29]). Alternating anoxic aerobic conditions were imposed on the systems, by alternating aeration with nonaeration over a 4 h cycle (3 h aeration on and 1 h aeration off). The feed to the CFCM system was continuous and with the aeration/non-aeration cycle, the sludge was exposed to the influent RBCOD under both anoxic and aerobic conditions. The IFFD system was batch fed once daily and, in order to expose the sludge to influent RBCOD under both aerobic and anoxic conditions, as in the CFCM system, the time of the daily feed was alternated in accordance with the
aeration/non-aeration cycle imposed on the system. On one day, the system was fed at the time it became aerobic and on the next day at the time it became anoxic 1 h earlier. In both the IFFD and CFCM systems, the DO concentration was maintained between 2 and 3 mg O/l during the aerobic periods. The two systems were operated for period of 70 d, during which five different sewage batches were used as influent, and the system performances were monitored daily [29]. Between days 60 and 70, five aerobic and three anoxic in situ profile tests were conducted on the IFFD system for the 2 h period immediately following the wastewater batch feed (for the anoxic profiles, the anoxic period following feeding was increased from 1 to 2 h). For the aerobic profiles, OURs were measured regularly. For the anoxic profiles, grab samples were taken at regular intervals, immediately filtered (0.45 mm) and HgCl2 added to the filtrate. The filtrate was analysed for nitrate and nitrite with the auto analyser industrial methods 33.69W and 35.69W, respectively (Technicon); in all experiments, nitrite concentrations were negligible. Additionally, nitrate concentrations were measured in situ at less than 5 min intervals with an Orion nitrate selective electrode, as mV. The mV-time profiles were calibrated against the nitrate concentrations of the grab samples. 3.2. Results and discussion Typical OUR and nitrate time plots for the aerobic and anoxic profiles are shown in Figs. 1 and 2, respectively. The OUR versus time profiles (Fig. 1) showed an initial rapid approximately constant OUR (80–110 mg O/l h) for the first 25–50 min, followed by a precipitous decrease when the influent RBCOD was completely utilised. Similarly, the anoxic profiles (Fig. 2) showed an initial rapid decrease in nitrate concentration over time for the first 50–80 min until the influent RBCOD was completely utilised, after which a much slower nitrate concentration decrease with time took place. Parallel batch tests on sludge harvested from the CFCM system are not useful for this paper, because the 2 h batch test time period was insufficient to utilise all the influent RBCOD (see [29]). In the aerobic profile results, the area below the OUR–time curve, but above the line extended from the second plateau to the vertical axis, is the oxygen utilised (OU) in the consumption of the influent wastewater RBCOD concentration, see Fig. 1. In the five aerobic batch tests, the OUs were 36.7, 37.3, 38.7, 29.1 and 41.7 giving a mean of 36.6 mg O/l. In the anoxic profiles, the distance on the vertical axis between the lines drawn through the initial rapid and second slower nitrate concentration reduction results is the nitrate utilised (NU) in consumption of the influent RBCOD
A. Muller et al. / Water Research 37 (2003) 2435–2441
2439
In terms of ASM1 (and similar models), for the utilisation of RBCOD:
Fig. 1. Typical OUR versus time profiles following batch feeding under aerobic conditions of IFFD activated sludge system; feed raw municipal wastewater.
OU ¼ SS;R ð1 YH;O2 Þ and
ð1aÞ
NU ¼ SS;R ð1 YH;NO Þ=2:86;
ð1bÞ
where SS,R is the initial RBCOD concentration in the reactor after batch feeding (mg COD/l), OU the oxygen consumed for RBCOD utilisation (mg O/l), NU the nitrate consumed for RBCOD utilisation (mg NO3–N/l), YH,O2 the heterotroph yield under aerobic conditions (mg COD/mg COD), YH,NO the heterotroph yield under anoxic conditions (mg COD/mg COD). Accepting that the initial reactor RBCOD concentration after the batch feed is equal for the anoxic and aerobic profiles, which is reasonable because the same wastewater batch was used in the tests at the same loading rate, and the aerobic and anoxic profile tests were done alternately over the 10 days (see [29]), then equating Eqs. (1a) and (1b) and solving, we get 2:86dNU YH;NO ¼ 1 ð1 YH:O2 Þ: ð2Þ OU Substituting 36.6 mg O/l for OU and 18.3 mg NO3–N/l for NU into Eq. (2) yields YH;NO ¼ 0:43 þ 1:43YH;O2 :
ð3Þ
Eq. (3) is shown plotted in Fig. 3. From Eq. (3), accepting from ASM1 that YH,O2=0.67 mg COD/ mg COD, then YH,NO=0.53 mg COD/mg COD and the anoxic to aerobic yield ratio is 0.79. This anoxic yield value is very close to the range of values measured by Sperandio et al. [26] with artificial substrates, who measured an equivalent aerobic yield of 0.67 mg COD/ mg COD which is equal to the value accepted here. Further, the anoxic to aerobic yield ratio (0.79) is equal to the theoretical value of Orhon et al. [9] for municipal
Fig. 2. Typical nitrate versus time profiles following batch feeding under anoxic conditions of IFFD activated sludge system; feed raw municipal wastewater.
concentration, see Fig. 2. For the three anoxic profiles on the IFFD system, the NUs were 226=16, 5131=22 and 3518=17 mg NO3–N/l, giving a mean of 18.3 mg NO3–N/l.
Fig. 3. Heterotroph anoxic yield versus aerobic yield for the OU and NU values measured in this experimental investigation, in terms of ASM1 [3] and ASM3 [8]. Also shown are the anoxic yield values for the aerobic yield values proposed in the two models.
2440
A. Muller et al. / Water Research 37 (2003) 2435–2441
wastewater (0.79), and reasonably close to the value . et al. [12] also on municipal measured by Ubay C , okgor wastewater (0.82). In terms of ASM3 [8], for the utilisation of RBCOD (assuming complete utilisation of the stored substrate generated from RBCOD uptake): OU ¼ SS;R ð1 YSTO;O2 YH;O2 Þ
and
NU ¼ SS;R ð1 YSTO;NO YH;NO Þ=2:86;
ð4aÞ ð4bÞ
where YSTO,O2 is the stored COD yield under aerobic conditions (mg COD/mg COD), YSTO,NO the stored COD yield under anoxic conditions (mg COD/ mg COD). Similar to the above YH;NO 2:86dNU ð1 YSTO;O2 YH;O2 Þ =YSTO;NO : ð5Þ ¼ 1 OU Accepting the values proposed by the IWA Task Group of YSTO,O2=0.85 and YSTO,NO=0.80 mg COD/ mg COD and the experimentally measured NU and OU, Eq. (5) is shown plotted in Fig. 3 also. From Eq. (5), accepting from ASM3 YH,O2=0.63 mg COD/mg COD, then YH,NO=0.42 mg COD/mg COD, and the anoxic to aerobic yield ratio is 0.67. Both the anoxic yield and yield ratio values are significantly lower than those measured by Sperandio et al. [26] of 0.54–0.57 and 0.81– 0.85, respectively, and those suggested by the Task Group of 0.54 mg COD/mg COD and 0.86, respectively.
experimentally. This can be done by changing the value of the reduction factor for anoxic SBCOD hydrolysis/ utilisation, Zh. The reduction in anoxic yield most likely will have a lesser effect on the net sludge production, since in most conventional N removal activated sludge systems, the mass of sludge produced under anoxic conditions is small compared to that produced under aerobic conditions. Further, a reduced anoxic endogenous respiration rate [25] may compensate for the reduced anoxic yield. In ASM3, accepting the proposed heterotroph aerobic yield of 0.63 mg COD/mg COD, then from the experimental data presented here (which corresponds closely to similar investigations) the anoxic yield should be approximately 0.42 mg COD/mg COD, giving an anoxic to aerobic yield ratio of 0.67. Both values are significantly lower than those suggested by the Task Group, of 0.54 and 0.86, respectively. This requires further investigation.
Acknowledgements This research was supported by the Water Research Commission, the National Research Foundation, THRIP and Water and Sanitation Services South Africa (Pty) Ltd (a subsidiary of Ondeo Services), and is published with their permission.
References 4. Conclusions Both theoretically and experimentally, with both artificial and municipal wastewaters as substrate, there is sufficient evidence that the value for the heterotroph anoxic yield should be reduced compared to the value for the aerobic yield. Theoretically and experimentally, the anoxic to aerobic yield ratio appears to fall into the range 0.78–0.85. In ASM1 (and similar models), accepting the value for the aerobic yield as that conventionally applied, 0.67 mg COD/mg COD, then the anoxic yield value should fall into the range 0.52– 0.57; 0.53 mg COD/mg COD determined here. Changing the anoxic yield value from the current ASM1 aerobic yield value of 0.67 mg COD/mg COD to 0.53 mg COD/ mg COD will result in a significant increase in the denitrification potential of the system, due to increased denitrification with wastewater RBCOD as substrate. If such a change is implemented in ASM1 (and similar models), then it must be remembered to change the SBCOD hydrolysis/utilisation rate in concert, since this rate effectively has been calibrated from the measured denitrification rate, and the predicted denitrification rate with SBCOD must be kept equal to that observed
[1] WRC. Theory, design and operation of nutrient removal activated sludge processes. Water Research Commission, Pretoria, RSA, 1984. [2] Van Haandel AC, Ekama GA, Marais GvR. The activated sludge process 3—single sludge denitrification. Water Res 1981;15(10):1135–52. [3] Henze M, Grady Jr CPL, Gujer W, Marais GvR, Matsuo T. Activated sludge model no. 1. IAWPRC Scientific and Technical Report No. 1, IAWPRC, London, 1987. [4] Dold PL, Ekama GA, Marais GvR. A general model for the activated sludge process. Prog Water Technol 1980;12(Tor):47–77. [5] Dold PL, Wentzel MC, Billing AE, Ekama GA, Marais GvR. Activated sludge simulation programs: Nitrification and nitrification/denitrification systems (Version 1.0). Water Research Commission, Pretoria, South Africa, 1991. TT 52/91, ISBN 0 947447 19 9. [6] Barker PS, Dold PL. General model for biological nutrient removal activated-sludge systems: model presentation. Water Environ Res 1997;69(5):969–84. [7] Murnleitner E, Kuba T, van Loosdrecht MCM, Heinen JJ. An integrated metabolic model for aerobic and denitrifying biological phosphorus removal. Biotechnol Bioeng 1997;54(5):434–50.
A. Muller et al. / Water Research 37 (2003) 2435–2441 [8] Gujer W, Henze M, Mino T, Van Loosdrecht MCM. Activated sludge Model No. 3. Water Sci Technol 1999; 39(1):183–93. . [9] Orhon D, Sozen S, Artan N. The effect of heterotrophic yield on the assessment of the correction factor for anoxic growth. Water Sci Technol 1996;34(5–6):67–74. . . E, Orhon D, Henze M. Respiro[10] Sozen S, Ubay 9Cokgor metric analysis of activated sludge behaviour—II. Heterotrophic growth under aerobic and anoxic conditions. Water Res 1998;32(2):476–88. [11] Ekama GA, Dold PL, Marais GvR. Procedures for determining influent COD fractions and the maximum specific growth rate of heterotrophs in activated sludge systems. Water Sci Technol 1986;18(6):91–114. . E, Sozen . [12] Ubay 9Cokgor S, Orhon D, Henze M. Respirometric analysis of activated sludge behaviour—I Assessment of the readily biodegradable substrate. Water Res 1998;32(2):461–75. [13] Henze M, Gujer W, Mino T, Matsuo T, Wentzel MC, Marais GvR. Activated sludge model No.2. IAWQ Scientific and Technical Report No 3, IAWQ, London, 1995. [14] Henze M, Gujer W, Mino T, Matsuo T, Wentzel MC, Marais GvR, Van Loosdrecht MCM. Activated sludge model no. 2d. Water Sci Technol 1999;39(1): 165–82. [15] Wentzel MC, Ekama GA, Marais GvR. Processes and modelling of nitrification denitrification biological excess phosphorus removal systems. Water Sci Technol 1992; 25(6):59–82. [16] Kuba T, Smolders GJF, van Loosdrecht MCM, Heinen JJ. Biological phosphorus removal from wastewater by anaerobic-anoxic sequencing batch reactor. Water Sci Technol 1993;27(5/6):241–52. [17] Payne WJ. Denitrification. New York: Wiley, 1981. [18] Casey TG, Wentzel MC, Ekama GA. Filamentous organism bulking in nutrient removal activated sludge systems. Paper 10: metabolic behaviour of heterotrophic
[19]
[20]
[21] [22]
[23]
[24]
[25]
[26]
[27] [28]
[29]
2441
facultative aerobic organisms under aerated/unaerated conditions. Water SA 1999;25(4):425–42. McCarty PL. Energetics and bacterial growth. In: Faust SD, Hunter JV, editors. Organic compounds in aquatic environments. New York, NY, USA: Marcell Dekker, 1971. McCarty PL. Energetics of organic matter degradation. In: Mitchell R, editor. Water pollution microbiology. New York, NY, USA: Wiley, 1972. McCarty PL. Stoichiometry of biological reactions. Prog Water Technol 1975;7(1):157–72. McClintock SA, Sherrard JH, Novak JT, Randall CW. Nitrate versus oxygen respiration in the activated sludge process. J WPCF 1988;60(3):342–50. Lawrence AW, McCarty PL. Unified basis for biological treatment design and operation. J Sanit Eng Div ASCE 1970;96(SA3):757–78. Marais GvR, Ekama GA. The activated sludge process Part 1—steady state behaviour. Water SA 1976;2(4): 163–200. Siegrist H, Brunner I, Koch G, Linh Con Phan, Van Chieu LE. Reduction of biomass decay rate under anoxic and anaerobic conditions. Water Sci Technol 1999; 39(1): 129–137. Sperandio M, Urbain V, Audic J-M, Paul E. Use of carbon dioxide evolution rate for denitrifying heterotrophic yield and characterizing denitrifying biomass. Water Sci Technol 1999;39(1):139–46. Envirosim. BIOWIN 32 process simulator. Flamborough, Ontario, Canada, 2001. Henze M, Gujer W, Mino T, Van Loosdrecht MCM. Activated sludge models ASM1, ASM2, ASM2d and ASM3. IWA Scientific and Technical Report No. 9, IWA, London, 2000. Ekama GA, Wentzel MC, Casey TG, Marais GvR. Filamentous organism bulking in nutrient removal activated sludge systems Paper 3—stimulation of the selector effect under anoxic conditions. Water SA 1996;22(2):119–26.