- Email: [email protected]
Nitrous oxide emissions from the oxidation tank of a pilot activated sludge plant
w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 3 5 6 3 e3 5 7 3
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Nitrous oxide emissions from the oxidation tank of a pilot activated sludge plant Adriana Maria Lotito a,b,*, Pascal Wunderlin a, Adriano Joss a, Marco Kipf a, Hansruedi Siegrist a a b
Eawag, Swiss Federal Institute of Aquatic Science and Technology, U¨berlandstrasse 133, P.O. Box 611, 8600 Du¨bendorf, Switzerland Department of Water Engineering and Chemistry, Politecnico di Bari, via Orabona 4, 70125 Bari, Italy
article info
abstract
Article history:
This study discusses the results of the continuous monitoring of nitrous oxide emissions
Received 3 February 2012
from the oxidation tank of a pilot conventional wastewater treatment plant. Nitrous oxide
Received in revised form
emissions from biological processes for nitrogen removal in wastewater treatment plants
27 March 2012
have drawn great attention over the last years, due to the high greenhouse effect. However,
Accepted 31 March 2012
even if several studies have been carried out to quantify nitrous oxide emission rates from
Available online 9 April 2012
different types of treatment, quite wide ranges have been reported. Only grab samples or continuous measurements over limited periods were considered in previous studies, which
Keywords:
can account for the wide variability of the obtained results. Through continuous moni-
Nitrous oxide emissions
toring over several months, our work tries to fill this gap of knowledge and get a deeper
Activated sludge
insight into nitrous oxide daily and weekly emission dynamics. Moreover, the influence of
Sludge age
some operating conditions (sludge age, dissolved oxygen concentration in the oxidation
Dissolved oxygen
tank, nitrogen load) was studied to determine good practices for wastewater treatment
Nitrogen load
plant operation aiming at the reduction of nitrous oxide emissions. The dissolved oxygen
Stress conditions
set-point is shown to play a major role in nitrous oxide emissions. Low sludge ages and high nitrogen loads are responsible for higher emissions as well. An interesting pattern has been observed, with quite negligible emissions during most of the day and a peak with a bell-like shape in the morning in the hours of maximum nitrogen load in the plant, correlated to the ammonia and nitrite peaks in the tank. ª 2012 Elsevier Ltd. All rights reserved.
1.
Introduction
The enforcement of more restrictive regulations concerning nitrogen discharge from wastewater treatment plants has led to the implementation of processes for nitrogen removal, using both innovative treatment technologies (e.g., moving bed, biofilters, supernatant treatment with nitritation/ anammox) and sensor control for optimal operation of conventional activated sludge plants in order to achieve
maximum nitrification capacity and biological nitrogen removal (BNR). Nitrous oxide (N2O) is one of the intermediates produced in both nitrification and denitrification processes and it can be released into the atmosphere, especially when incomplete nitrification or denitrification take place. As its greenhouse effect estimated over a time horizon of 100 years is about 300 times the one of CO2 and it is really persistent (lifetime of 114 years) (IPCC, 2001), the analysis of its emission from
* Corresponding author. Department of Water Engineering and Chemistry, Politecnico di Bari, via Orabona 4, 70125 Bari, Italy. Tel.: þ39 340 3005056. E-mail address: [email protected] (A.M. Lotito). 0043-1354/$ e see front matter ª 2012 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2012.03.067
3564
w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 3 5 6 3 e3 5 7 3
wastewater treatment plants is of paramount importance. Moreover, N2O is currently the single most important ozonedepleting gas and is expected to remain the largest throughout the 21st century (Ravishankara et al., 2009). Several attempts to quantify nitrous oxide emissions from wastewater treatment plants have been carried out over the past years: different treatment schemes have been scrutinized, ranging from conventional activated sludge plants to sequencing batch reactors to modified bioreactors for enhanced nitrogen removal (extended aeration, modified Ludzack Ettinger, four stage Bardenpho, etc.) (Czepiel et al., 1995; Su¨mer et al., 1995; Benckiser et al., 1996; Kimochi et al., 1998; Foley et al., 2010; Hu et al., 2010). However, wide emission ranges have been reported so far, also due to the different methods used for N2O capture and measurement and to the adopted normalization (emissions with reference to influent nitrogen load, population equivalent, influent flow rate, etc.), which makes results not easily comparable. Furthermore, different results have been obtained in lab-scale analyses or full-scale measurements in wastewater treatment plants. On the grounds of the results of Czepiel et al. (1995), the Intergovernmental Panel on Climate Change proposed in its Guidelines of 2006 to use an annual default emission factor of 0.0032 kgN2O per population equivalent (IPCC, 2006), which corresponds to about 0.35& of influent nitrogen when assuming a wastewater nitrogen load of 16 g per population equivalent per day (Foley et al., 2010). Nevertheless, in a review of numerous studies, a higher median value of 10& of influent nitrogen, within a range of values from 0.3& to 30& (10e90th percentiles), has been computed (Foley and Lant, 2008). Such huge variability is also confirmed in another review by Kampschreur et al. (2009), both in lab-scale and full-scale studies. None of previous studies refers to continuous N2O monitoring over a long time lapse. Grab samples or continuous online measurements on limited periods were always used for full-scale analyses, while longer on-line measurements were adopted only for batch tests or continuous processes with synthetic wastewaters (Kampschreur et al., 2009). For example, Czepiel et al. (1995) measured nitrous oxide fluxes at roughly weekly intervals at approximately the same time of the day for a period of about 4 months. Foley et al. (2010) performed four more intensive sampling rounds of 2e4 h in the morning and in the afternoon of two consecutive days for each of the plants considered, taking grab samples. Ahn et al. (2010) examined the diurnal variability in N2O emissions using continuous measurements over a 24-h period, but over just one day. On the contrary, this paper proposes the results of the continuous monitoring of N2O emissions from the oxidation tank of a conventional pilot wastewater treatment plant over several months in order to obtain a more extensive and complete data-set about N2O emission in terms of both total emission during a day and temporal variability. Two plant configurations have been considered (Fig. 1): first, a two-stage biological treatment (pre-denitrification and nitrification) and, second, a three-stage process (pre-denitrification, first oxidation, second oxidation). As both process design and operating conditions are likely to influence the magnitude and variability of N2O generation (Foley et al., 2010), some operational parameters (sludge age, dissolved oxygen, nitrogen load), supposed to be fundamental in N2O production according to
Fig. 1 e Pilot plant sketch. Wastewater was directly pumped from the sewer system and reached the biological compartment after preliminary treatments and primary settling. Two plant configurations were tested: a) two-tank configuration (activated sludge system with predenitrification); b) three-tank configuration (predenitrification followed by two oxidation tanks).
previous studies, have been varied to assess their influence. In fact, sub-optimal growth conditions favor N2O production in the nitrification stage: more specifically, both low concentrations of dissolved oxygen and short sludge retention time were found to promote nitrous oxide build-up by nitrifying activate sludge; furthermore, N2O production proved to be proportional to both nitrogen and organic loads in fully aerated activated sludge plants (Colliver and Stephenson, 2000). In addition to monitoring normal plant operation, specific tests have been performed to study the behavior of the system under stressed conditions (ammonia overloading, lowering of dissolved oxygen), in order to understand which factors play a major role in nitrous oxide emissions.
2.
Material and methods
2.1.
Pilot plant
The pilot plant under examination perfectly reproduces a fullscale plant (Fig. 1). Municipal wastewater is directly pumped
3565
w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 3 5 6 3 e3 5 7 3
from the combined sewer system of the city of Du¨bendorf (Switzerland) and undergoes pre-treatments for the removal of grit and sand before going into the primary settler. The biological compartment is made up of two or three tanks (depending on the plant configuration adopted), each with a volume of 7.5 m3. After the biological compartment, a secondary settler removes solids before discharge. A sludge recycle from the secondary settler to the denitrification tank, set to be proportional to the influent flow through a sludge recycle factor, maintains sludge concentration in the tanks and brings back oxidized nitrogen in the denitrification tank, from where excess sludge is removed. When a two-tank biological compartment configuration was adopted, the first tank was used for pre-denitrification, while the second one (oxidation tank) was aerated in order to oxidize both organic carbon and ammonia (Fig. 1a). Aeration was controlled to keep a constant concentration of dissolved oxygen (DO) in the reactor. In the three-tank configuration (Fig. 1b), the first tank was still used for pre-denitrification, while both the others were aerated. A different DO set-point was fixed for the two tanks, with a higher value in the first oxidation tank and a lower in the second one. An internal recycle of 1.5 m3/h was added from the second oxidation tank into the pre-denitrification one. The influent flow rate into the biological compartment was variable, with changes every 2 h to simulate the daily flow fluctuations typical in sewer systems. Two different daily average values of about 1.0 and 1.2 m3/h were adopted (in particular, the maximum flow rate was set to be 1.3 or 1.6 m3/h in the period 8:00e10:00 a.m., while the minimum one was 0.6 or 0.7 m3/h in the period 4:00e6:00 a.m.). These influent flow rates corresponded to about 80 or 100 population equivalents (PEs) (120 gCOD PE1 d1 or 0.3 m3 PE1 d1).
2.2.
Analytical methods
The plant was equipped with various analytical instruments, including sensors for ammonia in the primary settler (Ion Selective Electrode Endress&Hauser) and DO in the biological tanks (Optical LDO, Hach Lange; Optical LDO, Endress&Hauser). All equipment was connected to a central programmable logic controller (PLC) linked to a supervisory control and data acquisition (SCADA) system for direct access and data storage, with data taken every 2 min, except for DO and air flux (data taken every 10 s). Laboratory analyses were carried out three times per week on 24-h flow proportional composite samples of influent and effluent and on grab samples from the tanks to determine total nitrogen TN (commercial photochemical test kits, Hach Lange), ammonia (Foss FIAstar flow injection 5000 analyzer), nitrites and nitrates (ion chromatography; 761 compact IC, Metrohm). Total and soluble chemical oxygen demand (COD) were measured using commercial photochemical test kits (Hach Lange), while total suspended solids (TSS) in the influent and in the effluent and mixed liquor suspended solids (MLSS) were measured according to standard methods (APHA, 1998). Off-gas concentrations in terms of O2 and CO2 (Rosemount Analytical Binos 100 2M Dual-Channel Gas Analyzer) and
N2O (Rosemount Analytical X-Stream X2) were continuously measured from the oxidation tank, which was perfectly sealed and regularly checked to avoid any loss of gas (see Supplementary data for accuracy of N2O measurements, Figures S1eS3). Data taken every 20 s were logged for further elaborations. Gas analyzers were calibrated three times per week with a two point calibration, using N2 for zero point calibration, pure gas with known concentration for N2O (200 ppm) and CO2 (2%), and ambient air for O2. Off-gas N2O measurements were processed to compute the mass of nitrogen emitted as N2O by multiplying them by the air flow rate (see Supplementary data, Figure S4). At standard conditions, 1 ppm N2O corresponds to 1:249 mgN2 O N=m3air . Sludge age was computed dividing the total amount of solids present in the biological tanks by the amount of solids discharged every day through effluent and excess sludge removal; hence, it was referred to the biological compartment, excluding the sludge contained in the secondary settler.
2.3.
Operational schedule
In order to establish steady-state equilibrium, the plant was operated for several weeks at constant sludge age before the beginning of off-gas measurements. During the first part of the monitoring campaign (two-tank configuration; period A), covering about two months, high sludge ages were tested (higher than 15 days). Different operational conditions were tested, varying some parameters as reported in Table 1 (excess sludge removal, sludge age, DO set-point). Therefore, three phases can be distinguished in period A. In the second part of the monitoring campaign (period B, three-tank configuration) sludge ages were reduced. Before starting experiments, the plant was operated for more than two months to be sure to have steady-state conditions in the new configuration. Nevertheless, N2O emissions were continuously measured even during the transitional period.
Table 1 e Operational conditions applied in period A (two-tank configuration) and B (three-tank configuration). Parameters
Period A A1
A2
Period B A3
B1
B2
Duration [d] 14 8 28 14 28 1.0 1.0 1.0 1.0 1.2 Average influent flow rate [m3/h] Return sludge 2 2 3 2.5 2.5 flow ratio MLSS [g/l] 4.1 4.1 4.3 2.3 1.6 0.6 0.6 0.75 1.15 1.80 Excess sludge [m3/d] Sludge retention 22 20 16 15 10e12 time (SRT) [d] DO set-point in 1.5 2.0 3.0 2.0 2.0 oxidation 1 [mg/l] DO set-point in e e e 0.5 0.5 oxidation 2 [mg/l] Stress tests Ammonia addition Tested Tested Tested Tested Tested DO ¼ 0.5 mg/l Tested e e e e DO ¼ 1.0 mg/l Tested e Tested Tested Tested
3566
w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 3 5 6 3 e3 5 7 3
Moreover, some stress tests were run as indicated in Table 1, consisting in the instantaneous addition of ammonia (in form of ammonium bicarbonate) or in the reduction of the DO set-point for 2 h in the oxidation tank (in period B, such tests were performed on the first oxidation tank). Unfortunately, due to bulking problems in the secondary settler, it was necessary to stop phase A2 before carrying out the other planned stress tests. The measurements were stopped for about one week in order to change the operational conditions and allow the plant to recover between phases A2 and A3. The oxidation tank was spiked with ammonia in every phase in order to observe the behavior of the system overloaded with ammonia: 480 g of (NH4)HCO3 were added instantaneously to increase the ammonia concentration by about 11.5 mgN/l. The salt was added at 10:10 a.m., in the period of maximum nitrogen influent load. Grab samples were taken from the oxidation tank to monitor nitrogen conversion. Dissolved oxygen reduction tests consisted in lowering the DO set-point from the normal value to 1.0 mg/l in the period 9:30e11:30 a.m., which corresponded to the phase of maximum nitrogen load to the tank. This value was chosen as it has been indicated in previous studies as the value that maximizes N2O emission (Ce´bron et al., 2005; Tallec et al., 2006). Only during phase A1 a second test with a DO setpoint of 0.5 mg/l was performed. Like in ammonia addition tests, grab samples were taken to follow nitrogen patterns.
3.
Results and discussion
3.1.
Nitrification performance of the plant
Influent TN values in the range between 12.9 and 43.3 mgN/l were measured during the whole monitoring campaign (mean value of 26.5 5.7 mgN/l), with average ammonia around 17.9 4.7 mgN/l and nitrites and nitrates lower than 1.0 mgN/ l. Average influent temperature was 14.9 1.4 C in period A and 19.3 0.8 C in period B, with pH being 8.0 0.4 and 7.4 0.5 for the two periods, respectively. During period A, the plant exhibited complete nitrification, with both ammonia and nitrite concentrations in the effluent below 0.5 mgN/l, and progressive reduction of nitrates from 5 mgN/l to less than 2 mgN/l. In period B, complete nitrification was achieved as well. However, ammonia was not completely removed in the first oxidation tank, especially during the morning peak in influent concentration, when some ammonia was removed in the second oxidation tank.
3.2.
Nitrous oxide emissions during normal operation
Nitrous oxide emissions from the oxidation tank during normal operation in period A (two-tank configuration) were always rather low (Fig. 2). In fact, the amount of nitrogen emitted per day from the oxidation tank as N2O (related to the total nitrogen load entering the biological compartments) was in most of the cases lower than 0.40&. Examining the trends for every week, N2O emissions generally increased during the week and reached the highest value on Sundays, with a peak around 1.15& during phase A1.
Fig. 2 e Daily fraction of the total influent nitrogen emitted as N2O during the three phases of period A. Filled markers refer to normal operation of the plant, while hollow markers refer to the results obtained during stress tests (ammonia addition or DO reduction).
Differences were observed among the three phases in which period A was divided, due to the different set-points of DO in the oxidation tank. As can be seen in Fig. 2, higher fractions of nitrogen were emitted as N2O with lower DO values (i.e., in phase A1). Discarding the results obtained with stress tests, average values of 0.41 0.32& during phase A1, 0.13 0.11& during phase A2 and 0.12 0.06& during phase A3 were measured. However, the small variation between phases A2 and A3 is due to the absence in phase A2 of data from week-end days or holidays (during which higher loads were normally observed) due to the previously mentioned bulking problems: average values calculated without such days amount to 0.29 0.16& for phase A1 and 0.04 0.02& for phase A3, respectively. The low recorded emissions can be explained as the outcome of the operation of the plant at constant DO. In the study by Ahn et al. (2010) about different plant configurations, it was observed that, in general, wider ranges of operating DO concentrations in the aerobic zones correspond to higher fractions of nitrogen emitted as nitrous oxide. Moreover, the high sludge age during period A (around 20 days during phases A1 and A2, around 16 days during phase A3) could also account for the low measured emissions, because in these conditions complete nitrification is expected, with low N2O production (Noda et al., 2003). In fact, higher sludge ages are known to cause a significant variation in the microbial community, favoring the retention of microorganisms with low maximum specific growth rates (such as nitrifying bacteria) (Yuan et al., 2008). Furthermore, as ammonia oxidizing bacteria (AOB) and nitrite oxidizing bacteria (NOB) are characterized by different maximum specific growth rates, sludge retention time (SRT) determines also the relative abundance of species belonging to the two groups (Zhu et al., 2008). Nevertheless, the effect of the DO set-point during period A dominated over the one of sludge age reduction, as the lowest emissions were recorded in period A3, which was characterized by a lower sludge age (16 days), but a higher DO (3 mg/l).
w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 3 5 6 3 e3 5 7 3
Fig. 3 e Daily fraction of the total influent nitrogen emitted as N2O from the first oxidation tank during the two phases of period B. Filled markers refer to normal operation of the plant, while hollow markers refer to the results obtained during stress tests (ammonia addition or DO reduction). If emissions from the second oxidation tank are taken into account, global emissions increase and can be even doubled at high influent nitrogen loads.
The influence of the DO set-point can be due to the fact that both AOB and NOB change their metabolic activity as a function of oxygen concentration, thereby affecting both the rates of nitrification and aerobic denitrification, together with N2O emission (Colliver and Stephenson, 2000). Compared with AOB, NOB require higher DO concentrations, due to the lower
3567
oxygen affinity (different DO half-saturation value for oxygen); therefore, AOB dominate over NOB at low DO concentrations, which results in nitrite accumulation and in the occurrence of partial nitrification (being DO values higher than 2 mg/l required for complete nitrification) (Zhu et al., 2008). Furthermore, at low DO, some processes allow for simultaneous nitrogen oxidation and reduction, e.g., simultaneous nitrification and denitrification, autotrophic nitrogen removal over nitrite or nitrifier denitrification (i.e., denitrification by AOB), leading to higher N2O emissions (Wrage et al., 2001; Blackburne et al., 2008). During period B we reduced sludge age to less than 15 days in order to better assess its impact (Fig. 3). The fraction of nitrogen emitted as N2O from the first oxidation tank was on average 0.20 0.11& in phase B1 (sludge age around 15 days) and 0.56 0.30& in phase B2 (sludge age around 10e12 days) including week-end days and holidays and 0.11 0.10& and 0.31 0.18& discarding these values, being the average daily influent TN similar in the two phases (750 g and 747 g, respectively). As apparent, at constant conditions, the reduction of sludge age gave rise to an increase in N2O emissions. It is also worth considering that the values shown in Fig. 3 for period B refer only to emissions from the first oxidation tank. However, as both ammonia and nitrite were not completely oxidized in the first stage, N2O emissions were expected also from the second tank, even if to a lower extent due to lower ammonia concentrations. In order to estimate such contribution, we have measured emissions from both tanks for a week during phase B2, switching measurements from one reactor to the other one every half an hour by means of a temporized valve. The ratio between the emission from the
Fig. 4 e Daily variability of N2O emissions over two consecutive days with the correlation with ammonia influent trends (the first day is a holiday, while the second one is a working day): (a) influent ammonia concentration (continuous line) [mgN/l] and load (dots) [gN/h]; (b) dissolved nitrogen species in the oxidation tank measured with sensors (ammonia in continuous line, nitrite in dotted line and nitrate in dash-dot line) [mgN/l]; (c) off-gas N2O concentration (continuous line) [ppm] and load (dots) [10L2 gN/h].
3568
w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 3 5 6 3 e3 5 7 3
second oxidation tank and the one from the first oxidation tank ranged from about 5% at a low influent nitrogen load to more than 90% for a daily nitrogen load higher than 800 g. Therefore, global emissions at the highest influent nitrogen loads can be even doubled with respect to the values shown in Fig. 3. In general, it is possible to observe an increase in the fraction of influent nitrogen emitted as N2O at higher influent daily nitrogen loads, considering data of the same phase (Figs. 2 and 3). Hence, influent nitrogen load is another key factor that affects nitrous oxide emissions from the nitrification stage of a wastewater treatment plant. However, emissions for similar influent nitrogen loads in the same operational conditions (i.e., in the same phase) are quite scattered. Such result can be ascribed to other factors than the actual nitrogen load, like organic carbon concentration or the history of nitrogen load during previous days (that is to say if similar, higher or lower influent concentrations were observed in the days before). More precisely, higher influent COD can increase the competition for oxygen between heterotrophs and nitrifiers, thereby reducing ammonia removal rate and favoring ammonia and nitrite accumulation in the tank. Batch tests with the sludge taken from the oxidation tank of the pilot plant have shown that
higher N2O emissions were recorded during ammonia oxidation at lower DO concentrations and correspondingly lower nitrogen conversion rates (Wunderlin et al., 2012); moreover, if organic carbon is available in excess, simultaneous nitrification/denitrification can occur, bringing about a further increase in emissions (Wunderlin et al., 2012). As to nitrogen load history, we have observed that normal loads after some days of low load (due for example to rainy periods) resulted in higher emissions if compared to periods of almost constant load. This can be explained in terms of population dynamics, as activity and distribution of the different bacterial strains is related to the available substrate. The ratio of the feed biodegradable organic carbon to the nitrogen available for nitrification is a critical factor in the performance of nitrogen removal systems, as a high C/N ratio would retard the growth of nitrifying bacteria and especially nitrite oxidizers (Lin et al., 2009). It is interesting to compare our findings with the ones of Czepiel et al. (1995), which have been proposed as a reference value (IPCC, 2006). Even if no information about the influent nitrogen load is present in that study, their emission factor of 3.2 gN2O per population equivalent per year can be expressed as 0.35& of influent nitrogen if a wastewater nitrogen load of 16 g PE1 d1 is assumed (Foley et al., 2010).
Fig. 5 e Ammonia addition stress tests in period A: (a) phase A1 (sludge age 22 days, DO [ 1.5 mg/l); (b) phase A2 (sludge age 20 days, DO [ 2.0 mg/l); (c) phase A3 (sludge age 16 days, DO [ 3.0 mg/l). The shock was achieved by instantaneously adding 85 g of N-NH4 at 10:10 a.m., while the daily influent TN load was about 600 g. The first line shows influent ammonia concentration (continuous line) and load (dots), excluding ammonia added directly in the tank; the second line presents dissolved nitrogen species in the oxidation tank (ammonia with circles and continuous line, nitrite with stars and dotted line and nitrate with squares and dash-dot line) measured on grab samples (lines have not chemical meaning, but are drawn for a clearer visualization of trends); in the third row off-gas N2O concentration (continuous line) and load (dots) are represented.
w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 3 5 6 3 e3 5 7 3
Hence, comparable or even lower average emissions were observed in our case. However, in a study conducted on a German wastewater treatment plant even much lower emissions were reported (0.01& of the total nitrogen received) (Su¨mer et al., 1995). The analysis of N2O emission normalized to the influent flow reveals that values lower than 10e20 mgN-N2O/m3 were measured during period A and B, below the average of about 20 mgN-N2O/m3 reported by Czepiel et al. (1995). As observed by Ahn et al. (2010), such a normalization, even if usually adopted for U.S. EPA or IPCC emission factors, appears to be inadequate as a result of different global water-use patterns. In fact, the lower values referred to the influent flow in our experimental campaign can be due to the rather low nitrogen concentration in the influent (average value lower than 30 mgN/l) against typical values recorded in full-scale wastewater treatment plants (40e70 mgN/l for medium and high strength untreated domestic wastewater) (Metcalf & Eddy Inc., 2003). The nitrogen load computed for our plant is around 8e9 g PE1 d1, about a half of the value of 16 g PE1 d1 assumed for developed countries (Foley et al., 2010; Kampschreur et al., 2009), but in agreement with the value
3569
of TKN of 10 g PE1 d1 suggested in German guidelines about the dimensioning of wastewater treatment plants (DWA, 2000). Another reason behind the difference in values between this study and the one of Czepiel et al. (1995) can be the variability of nitrous oxide emissions over the day. Actually, nitrous oxide emissions are not constant but exhibit a typical pattern. During period A, they were quite negligible for most of the day and presented a bell-like shape in the morning in the hours of maximum nitrogen load in the plant (between 9:00 and 12:00 during working days, between 10:00 and 14:00 during week-ends and holidays). During period B, similar trends were identified; however, emissions lasted for longer periods and a second peak was often observed in the late afternoon, corresponding to a second peak in ammonia concentration normally present in the influent. As an example, the trends recorded over two consecutive days in phase B2 (the first day is a holiday, while the second one is a working day) are depicted in Fig. 4. As can be seen, as soon as the ammonia influent reaches the peak in terms of concentration and load (computed multiplying point by point the ammonia concentration by the influent flow rate)
Fig. 6 e Ammonia addition stress tests in period B: (a) phase B1 (sludge age 15 days, DO [ 2.0 mg/l); (b) phase B2 (sludge age 10e12 days, DO [ 2.0 mg/l). The shock was achieved by instantaneously adding 85 g of N-NH4 at 10:10 a.m., while the daily influent TN load was about 850 g. The first line shows influent ammonia concentration (continuous line) and load (dots), excluding ammonia added directly in the tank; the second line presents dissolved nitrogen species in the oxidation tank (ammonia with circles and continuous line, nitrite with stars and dotted line and nitrate with squares and dash-dot line) measured on grab samples (lines have not chemical meaning, but are drawn for a clearer visualization of trends); in the third row off-gas N2O concentration (continuous line) and load (dots) are represented.
3570
w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 3 5 6 3 e3 5 7 3
(Fig. 4a), ammonia and nitrite start to accumulate in the oxidation tank (Fig. 4b). Simultaneously, N2O emission takes place, with the rate of emission increasing up to a maximum value when ammonia in the oxidation tank peaks (Fig. 4c). Depending on the amount of nitrite produced, N2O emission starts to decrease if no more nitrite is present (as in the second plotted day) or maintains a constant value until nitrite concentration peaks (as in the first plotted day). Nevertheless, when both ammonia and nitrite are removed, N2O emission ceases. A similar profile was detected in the first aerobic zone of the second pass of a full-scale step-feed BNR process, with a close correlation between mixed-liquor ammonia and nitrite peak concentrations and N2O gas concentrations (Ahn et al., 2010). The significant variations in emissions over a day and a week detected in our study indicate that great attention should be paid to this aspect when planning a monitoring campaign at a full-scale wastewater treatment plant, especially for plants devoid of an equalizing tank and, hence, subject to flow and load variability. In order to avoid systematic errors in the average estimates, measures should be extended over longer time frames to encompass both the daily and weekly variability, as sampling campaigns concentrated in few hours would neglect the variations of emissions over the day.
3.3.
Stress tests: ammonia addition
Fig. 5 compares the results obtained with the ammonia addition tests in the different phases of period A, while Fig. 6 reports the data referred to period B. Similar trends were observed in dissolved nitrogen species in all the tests. The reduction of ammonia concentration in the tank was partially due to transport as an effect of the influent and recycle flow; hence, not all the ammonia was really converted but it was partially lost in the effluent (period A) or nitrified in the second nitrification tank (period B), as demonstrated by ammonia concentration trends in effluent or in the second nitrification tank (data not shown). No high nitrite accumulation was observed in any of the three phases of period A, while some nitrite build-up was measured in period B (up to 1.7 mgN/l in phase B1, Fig. 6a). A minor increase in nitrate concentration was recorded as well. Even if N2O emission was already present to some extent, a sharp increase in N2O concentration and load was recorded soon after the ammonia addition, until a maximum value was reached after about 20e30 min. Afterward, emissions leveled out at the maximum for about 2.5e3.0 h in phases A1 and A2 (Fig. 5a and b) and less than 2 h in phases A3, B1 and B2 (Figs. 5c and 6). Then, they decreased again to the values observed
Fig. 7 e DO reduction stress tests from 9:30 a.m. to 11:30 a.m. in period A: (a) from 1.5 mg/l (normal DO set-point) to 0.5 mg/l in phase A1 (sludge age 22 days); (b) from 1.5 mg/l (normal DO set-point) to 1.0 mg/l in phase A1 (sludge age 22 days); (c) from 3.0 mg/l (normal DO set-point) to 1.0 mg/l in phase A3 (sludge age 16 days). The first line shows influent ammonia concentration (continuous line) and load (dots); the second line presents dissolved nitrogen species in the oxidation tank (ammonia with circles and continuous line, nitrite with stars and dotted line and nitrate with squares and dash-dot line) measured on grab samples (lines have no chemical meaning, but are drawn for a clearer visualization of trends); in the third row off-gas N2O concentration (continuous line) and load (dots) are represented.
w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 3 5 6 3 e3 5 7 3
before the ammonia addition and became negligible after a period ranging from less than 3 h in phases A3 and B2 to 4e5 h in the other phases. Interesting evaluations can be done observing the trend of the cumulative emissions of N2O (Supplementary data, Figure S5). Our results point out how dilution due to hydraulic conditions plays a major role in the attenuation of the ammonia peak and, consequently, in N2O emission. Keeping the same hydraulics (phases A1 and A2, phases A3 and B1), the extent of the N2O peak was related to the DO setpoint in the tank, with a higher DO value allowing lower N2O emissions. It can be inferred that a plant operated at higher DO concentrations is able to better afford ammonia overloading, with lower N2O emissions. The results obtained in our study can be compared with those of Burgess et al. (2002), who tested ammonia shocks with increasing amounts of ammonia in the range 0.48e3.30 mgNH4-N gTSS1 in a lab-scale activated sludge reactor. In our study, the shock loading normalized to MLSS is about 2.7e2.8 mgNH4-N gTSS1 for period A, thus within the range tested by Burgess et al. (2002), while the values in period B (5.0e7.2 mgNH4-N gTSS1) are much higher.
3571
Considering their results, a value around 10& should have been achieved in our tests in period A, higher than the values around 1.5e5.0& of total nitrogen obtained. Such a difference can be due to the different plant conditions in the study by Burgess et al. (2002) (higher hydraulic retention times, lower MLSS, different oxygen control in the tank). However, if we refer the nitrous oxide emitted only to the added ammonia (instead of considering the total nitrogen load including that of the influent), we get values in the range 6e20&, closer or higher than those of Burgess et al. (2002).
3.4.
Stress tests: DO reduction
Figs. 7 and 8 show the results obtained for DO reduction tests during period A and B, respectively. No nitrite accumulation was observed in any of the tests. However, some loss in the nitrification capacity could be recognized in the reduction of nitrate concentrations and in a slight accumulation of ammonia (even up to 3 mgN/l in the test at DO ¼ 0.5 mg/l in phase A1, as shown in Fig. 7a). A different trend is visible during phase B1 (Fig. 8a), with no loss in nitrification (no accumulation of ammonia, nitrite
Fig. 8 e DO reduction stress tests from 9:30 a.m. to 11:30 a.m. in period B: (a) from 2.0 mg/l (normal DO set-point) to 1.0 mg/l in phase B1 (sludge age 15 days); (b) from 2.0 mg/l (normal DO set-point) to 1.0 mg/l in phase B2 (sludge age 10e12 days). The first line shows influent ammonia concentration (continuous line) and load (dots); the second line presents dissolved nitrogen species in the oxidation tank (ammonia with circles and continuous line, nitrite with stars and dotted line and nitrate with squares and dash-dot line) measured on grab samples (lines have no chemical meaning, but are drawn for a clearer visualization of trends); in the third row off-gas N2O concentration (continuous line) and load (dots) are represented.
3572
w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 3 5 6 3 e3 5 7 3
build-up up to 1 mgN/l and increase in nitrate concentration). Compared to the ammonia addition tests (Figs. 5 and 6), also in DO reduction tests N2O emission started to increase soon after the stress application (at 9:30 a.m.), but with a more gradual slope. In the tests conducted during phase A1 and in the one in phase B2 (Figs. 7a,b and 8a), N2O emission persisted for several hours (even more than 6 h), while it became quite negligible in phases A3 and B1 less than an hour after the restoration of the normal DO value (Figs. 7c and 8a). A two-bell trend was observed in phase A1 in the test at DO ¼ 0.5 mg/l (Fig. 7a), with N2O emissions increasing again after the restoration of the normal set-point of 1.5 mg/l to values higher than those registered during the DO reduction. Such behavior can be justified by the fact that limited nitrification occurred at DO ¼ 0.5 mg/l, leading to accumulation of ammonia, which was further degraded when DO was increased again, resulting in N2O production. Moreover, another possible explanation is that the air flow necessary to maintain a DO of 0.5 mg/l was not enough to strip the N2O produced, which was then emitted when the air flow rate increased again to re-establish the higher DO concentration. A similar increase in N2O off-gas concentrations was observed by Burgess et al. (2002a), who tested an hour oxygen depletion, starting from initial DO values of 2.5 mg/l and 2.9 mg/l (with peak values of N2O observed 1 h after DO fell to <1.0 mg/l, followed by a decrease to pre-test levels after further 3.0e3.5 h). As demonstrated by the analysis of the cumulative emission of N2O (Supplementary data, Figure S6), lower emissions were achieved when the plant was normally operated at higher DO values (phase A3 with respect to phase A1). However, sludge age seems to play an important role, as the highest emissions during phase B2 were concomitant with the lowest tested sludge age (10e12 days).
4.
Increased nitrogen loads result in higher N2O emissions. However, as nitrous oxide emissions are correlated with the influent ammonia peak, lower emissions are to be expected if good strategies are adopted to achieve constant ammonia loads (for example, through influent equalization or anaerobic supernatant recycle in the plant in the period of lower influent ammonia load). Stress conditions (ammonia overloading, DO reduction) increase N2O emissions. However, when the system is normally operated at higher DO values, the plant is able to better withstand disturbance, resulting in lower emissions in terms of both peak concentration and duration. Higher recycle rates (sludge recycle and internal recirculation) contribute to the reduction of N2O emissions, because they attenuate the peak of ammonia and dilute the concentration of all the intermediates of nitrification/ denitrification.
Acknowledgments This study was financially supported by the Swiss Federal Office for the Environment (FOEN), the canton of Bern (AWA), the canton of Basel-Landschaft (AIB), the canton of St. Gallen (AFU), the canton of Thurgau (AfU), the canton of Zurich (AWEL), TBF þ Partner AG Consulting Engineers, WWTP region Bern, WWTP ProRheno, WWTP REAL Luzern, WWTP Worblental, WWTP Zurich-Werdho¨lzli, Abwasserverband Altenrhein, Abwasserverband Morgental, Entsorgung St. Gallen, Zentralschweizer Umweltdirektorenkonferenz (ZUDK) and Eawag.
Appendix A. Supplementary material Supplementary data related to this article can be found online at doi:10.1016/j.watres.2012.03.067.
Conclusions
Nitrous oxide emissions exhibit a typical pattern during the day, being quite negligible for most of the time and presenting a bell-shape form in the morning during the hours in which the maximum nitrogen load in the plant is recorded. Weekly variability is also present, with higher emissions during week-end. Attention should be paid to these trends to avoid systematic errors during full-scale monitoring. DO set-point in the oxidation tank plays a crucial role in N2O emissions as it influences ammonia removal rates. In order to prevent N2O emissions, the first aerobic tank in a wastewater treatment plant should not be operated at DO values lower than 2.0 mg/l. Low sludge ages (10e12 days) resulted in higher emissions from the first oxidation tank, even if average absolute values still remained quite low (mostly <1& of the total nitrogen in the influent). However, higher global emissions (even doubled) were obtained when considering also emissions from the second nitrification tank in the three-tank plant configuration.
references
APHA, AWWA, WEF, 1998. Standard Methods for the Examination of Water and Wastewater, 20th ed. American Public Health Association, American Water Works Association, Water Environment Federation, Washington, DC, USA, 1220 pp. Ahn, J.H., Kim, S., Park, H., Katehis, D., Pagilla, K., Chandran, K., 2010. Spatial and temporal variability in atmospheric nitrous oxide generation and emission from full-scale biological nitrogen removal and non-BNR processes. Water Environment Research 82 (12), 2362e2372. Benckiser, G., Eilts, R., Linn, A., Lorch, H.J., Su¨mer, E., Weiske, A., Wenzho¨fer, F., 1996. N2O emissions from different cropping systems and from aerated, nitrifying and denitrifying tanks of a municipal waste water treatment plant. Biology and Fertility of Soils 23, 257e265. Blackburne, R., Yuan, Z., Keller, J., 2008. Demonstration of nitrogen removal via nitrite in a sequencing batch reactor treating domestic wastewater. Water Research 42, 2166e2176. Burgess, J.E., Colliver, B.B., Stuetz, R.M., Stephenson, T., 2002. Dinitrogen oxide production by a mixed culture of nitrifying bacteria during ammonia shock loading and aeration failure.
w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 3 5 6 3 e3 5 7 3
Journal of Industrial Microbiology and Biotechnology 29, 309e313. Burgess, J.E., Stuetz, R.M., Morton, S., Stephenson, T., 2002a. Dinitrogen oxide detection for process failure early warning systems. Water Science and Technology 45 (4e5), 247e254. Ce´bron, A., Garnier, J., Billen, G., 2005. Nitrous oxide production and nitrification kinetics by natural bacterial communities of the lower Seine river (France). Aquatic Microbial Ecology 41, 25e38. Colliver, B.B., Stephenson, T., 2000. Production of nitrogen oxide and dinitrogen oxide by autotrophic nitrifiers. Biotechnology Advances 18, 219e232. Czepiel, P., Crill, P., Harriss, R., 1995. Nitrous oxide emissions from municipal wastewater treatment. Environmental Science and Technology 29, 2352e2356. DWA (German Association for Water, Wastewater and Waste), 2000. Dimensioning of Single-Stage Activated Sludge Plants ATV-DVWK-A 131E. DWA, Hennef, Germany, 54 pp. Foley, J., Lant, P., 2008. Fugitive Greenhouse Gas Emissions from Wastewater Systems: WSAA Literature Review No. 01. Water Services Association of Australia, Melbourne. Foley, J., de Haas, D., Yuan, Z., Lant, P., 2010. Nitrous oxide generation in full-scale biological nutrient removal wastewater treatment plants. Water Research 44, 831e844. Hu, Z., Zhang, J., Li, S., Xie, H., Wang, J., Zhang, T., Li, Y., Zhang, H., 2010. Effect of aeration rate on the emission of N2O in anoxicaerobic sequencing batch reactors (A/O SBRs). Journal of Bioscience and Bioengineering 109 (5), 487e491. IPCC, 2001. In: Houghton, J.T., Ding, Y., Griggs, D.J., Noguer, M., van der Linden, P.J., Dai, X., Maskell, K., Johnson, C.A. (Eds.), Climate Change 2001: The Scientific Basis. Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change Published for the Intergovernmental Panel on Climate Change. Cambridge University Press, United Kingdom and New York, NY, USA, p. 881. IPCC, 2006. In: Eggleston, H.S., Buendia, L., Miwa, K., Ngara, T., Tanabe, K. (Eds.), 2006 IPCC Guidelines for National Greenhouse Gas Inventories. IGES, Japan. Kampschreur, M.J., Temmink, H., Kleerebezem, R., Jetten, M.S.M., van Loosdrecht, M.C.M., 2009. Nitrous oxide emission during wastewater treatment. Water Research 43, 4093e4103.
3573
Kimochi, Y., Inamori, Y., Mizuochi, M., Xu, K.Q., Matsumura, M., 1998. Nitrogen removal and N2O emission in a full-scale domestic wastewater treatment plant with intermittent aeration. Journal of Fermentation and Bioengineering 86 (2), 202e206. Lin, Y.M., Tay, J.H., Liu, Y., Hung, Y.-T., 2009. Biological nitrification and denitrification processes. In: Wang, L.K., et al. (Eds.), Handbook of Environmental Engineering. Biological Treatment Processes, vol. 8. The Humana Press, Totowa, NJ, pp. 539e588. Ch. 13. Noda, N., Kaneko, N., Mikami, M., Kimochi, Y., Tsuneda, S., Hirata, A., Mizuochi, M., Inamori, Y., 2003. Effects of SRT and DO on N2O reductase activity in an anoxic-oxic activated sludge system. Water Science and Technology 48 (11e12), 363e370. Metcalf & Eddy Inc., 2003. Wastewater Engineering: Treatment and Reuse, fourth ed. McGraw-Hill, Boston, USA, 1848 pp. Ravishankara, A.R., Daniel, J.S., Portmann, R.W., 2009. Nitrous oxide (N2O): the dominant ozone-depleting substance emitted in the 21st century. Science 326, 123e125. Su¨mer, E., Weiske, A., Benckiser, G., Ottow, J.C.G., 1995. Influence of environmental conditions on the amount of N2O released from activated sludge in a domestic waste water treatment plant. Experientia 51, 419e422. Tallec, G., Garnier, J., Billen, G., Gousailles, M., 2006. Nitrous oxide emissions from secondary activated sludge in nitrifying conditions of urban wastewater treatment plants: effect of oxygenation level. Water Research 40, 2972e2980. Wrage, N., Velthof, G.L., van Beusichem, M.L., Oenema, O., 2001. Role of nitrifier denitrification in the production of nitrous oxide. Soil Biology and Biochemistry 33, 1723e1732. Wunderlin, P., Mohn, J., Joss, A., Emmenegger, L., Siegrist, H., 2012. Mechanisms of N2O production in biological wastewater treatment under nitrifying and denitrifying conditions. Water Research 46, 1027e1037. Yuan, Z., Oehmen, A., Peng, Y., Ma, Y., Keller, J., 2008. Sludge population optimisation in biological nutrient removal wastewater treatment systems through on-line process control: a re/view. Reviews in Environmental Science and Biotechnology 7, 243e254. Zhu, G., Peng, Y., Li, B., Guo, J., Yang, Q., Wang, S., 2008. Biological removal of nitrogen from wastewater. Reviews of Environmental Contamination and Toxicology 192, 159e195.
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Nitrous oxide emissions from the oxidation tank of a pilot activated sludge plant Adriana Maria Lotito a,b,*, Pascal Wunderlin a, Adriano Joss a, Marco Kipf a, Hansruedi Siegrist a a b
Eawag, Swiss Federal Institute of Aquatic Science and Technology, U¨berlandstrasse 133, P.O. Box 611, 8600 Du¨bendorf, Switzerland Department of Water Engineering and Chemistry, Politecnico di Bari, via Orabona 4, 70125 Bari, Italy
article info
abstract
Article history:
This study discusses the results of the continuous monitoring of nitrous oxide emissions
Received 3 February 2012
from the oxidation tank of a pilot conventional wastewater treatment plant. Nitrous oxide
Received in revised form
emissions from biological processes for nitrogen removal in wastewater treatment plants
27 March 2012
have drawn great attention over the last years, due to the high greenhouse effect. However,
Accepted 31 March 2012
even if several studies have been carried out to quantify nitrous oxide emission rates from
Available online 9 April 2012
different types of treatment, quite wide ranges have been reported. Only grab samples or continuous measurements over limited periods were considered in previous studies, which
Keywords:
can account for the wide variability of the obtained results. Through continuous moni-
Nitrous oxide emissions
toring over several months, our work tries to fill this gap of knowledge and get a deeper
Activated sludge
insight into nitrous oxide daily and weekly emission dynamics. Moreover, the influence of
Sludge age
some operating conditions (sludge age, dissolved oxygen concentration in the oxidation
Dissolved oxygen
tank, nitrogen load) was studied to determine good practices for wastewater treatment
Nitrogen load
plant operation aiming at the reduction of nitrous oxide emissions. The dissolved oxygen
Stress conditions
set-point is shown to play a major role in nitrous oxide emissions. Low sludge ages and high nitrogen loads are responsible for higher emissions as well. An interesting pattern has been observed, with quite negligible emissions during most of the day and a peak with a bell-like shape in the morning in the hours of maximum nitrogen load in the plant, correlated to the ammonia and nitrite peaks in the tank. ª 2012 Elsevier Ltd. All rights reserved.
1.
Introduction
The enforcement of more restrictive regulations concerning nitrogen discharge from wastewater treatment plants has led to the implementation of processes for nitrogen removal, using both innovative treatment technologies (e.g., moving bed, biofilters, supernatant treatment with nitritation/ anammox) and sensor control for optimal operation of conventional activated sludge plants in order to achieve
maximum nitrification capacity and biological nitrogen removal (BNR). Nitrous oxide (N2O) is one of the intermediates produced in both nitrification and denitrification processes and it can be released into the atmosphere, especially when incomplete nitrification or denitrification take place. As its greenhouse effect estimated over a time horizon of 100 years is about 300 times the one of CO2 and it is really persistent (lifetime of 114 years) (IPCC, 2001), the analysis of its emission from
* Corresponding author. Department of Water Engineering and Chemistry, Politecnico di Bari, via Orabona 4, 70125 Bari, Italy. Tel.: þ39 340 3005056. E-mail address: [email protected] (A.M. Lotito). 0043-1354/$ e see front matter ª 2012 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2012.03.067
3564
w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 3 5 6 3 e3 5 7 3
wastewater treatment plants is of paramount importance. Moreover, N2O is currently the single most important ozonedepleting gas and is expected to remain the largest throughout the 21st century (Ravishankara et al., 2009). Several attempts to quantify nitrous oxide emissions from wastewater treatment plants have been carried out over the past years: different treatment schemes have been scrutinized, ranging from conventional activated sludge plants to sequencing batch reactors to modified bioreactors for enhanced nitrogen removal (extended aeration, modified Ludzack Ettinger, four stage Bardenpho, etc.) (Czepiel et al., 1995; Su¨mer et al., 1995; Benckiser et al., 1996; Kimochi et al., 1998; Foley et al., 2010; Hu et al., 2010). However, wide emission ranges have been reported so far, also due to the different methods used for N2O capture and measurement and to the adopted normalization (emissions with reference to influent nitrogen load, population equivalent, influent flow rate, etc.), which makes results not easily comparable. Furthermore, different results have been obtained in lab-scale analyses or full-scale measurements in wastewater treatment plants. On the grounds of the results of Czepiel et al. (1995), the Intergovernmental Panel on Climate Change proposed in its Guidelines of 2006 to use an annual default emission factor of 0.0032 kgN2O per population equivalent (IPCC, 2006), which corresponds to about 0.35& of influent nitrogen when assuming a wastewater nitrogen load of 16 g per population equivalent per day (Foley et al., 2010). Nevertheless, in a review of numerous studies, a higher median value of 10& of influent nitrogen, within a range of values from 0.3& to 30& (10e90th percentiles), has been computed (Foley and Lant, 2008). Such huge variability is also confirmed in another review by Kampschreur et al. (2009), both in lab-scale and full-scale studies. None of previous studies refers to continuous N2O monitoring over a long time lapse. Grab samples or continuous online measurements on limited periods were always used for full-scale analyses, while longer on-line measurements were adopted only for batch tests or continuous processes with synthetic wastewaters (Kampschreur et al., 2009). For example, Czepiel et al. (1995) measured nitrous oxide fluxes at roughly weekly intervals at approximately the same time of the day for a period of about 4 months. Foley et al. (2010) performed four more intensive sampling rounds of 2e4 h in the morning and in the afternoon of two consecutive days for each of the plants considered, taking grab samples. Ahn et al. (2010) examined the diurnal variability in N2O emissions using continuous measurements over a 24-h period, but over just one day. On the contrary, this paper proposes the results of the continuous monitoring of N2O emissions from the oxidation tank of a conventional pilot wastewater treatment plant over several months in order to obtain a more extensive and complete data-set about N2O emission in terms of both total emission during a day and temporal variability. Two plant configurations have been considered (Fig. 1): first, a two-stage biological treatment (pre-denitrification and nitrification) and, second, a three-stage process (pre-denitrification, first oxidation, second oxidation). As both process design and operating conditions are likely to influence the magnitude and variability of N2O generation (Foley et al., 2010), some operational parameters (sludge age, dissolved oxygen, nitrogen load), supposed to be fundamental in N2O production according to
Fig. 1 e Pilot plant sketch. Wastewater was directly pumped from the sewer system and reached the biological compartment after preliminary treatments and primary settling. Two plant configurations were tested: a) two-tank configuration (activated sludge system with predenitrification); b) three-tank configuration (predenitrification followed by two oxidation tanks).
previous studies, have been varied to assess their influence. In fact, sub-optimal growth conditions favor N2O production in the nitrification stage: more specifically, both low concentrations of dissolved oxygen and short sludge retention time were found to promote nitrous oxide build-up by nitrifying activate sludge; furthermore, N2O production proved to be proportional to both nitrogen and organic loads in fully aerated activated sludge plants (Colliver and Stephenson, 2000). In addition to monitoring normal plant operation, specific tests have been performed to study the behavior of the system under stressed conditions (ammonia overloading, lowering of dissolved oxygen), in order to understand which factors play a major role in nitrous oxide emissions.
2.
Material and methods
2.1.
Pilot plant
The pilot plant under examination perfectly reproduces a fullscale plant (Fig. 1). Municipal wastewater is directly pumped
3565
w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 3 5 6 3 e3 5 7 3
from the combined sewer system of the city of Du¨bendorf (Switzerland) and undergoes pre-treatments for the removal of grit and sand before going into the primary settler. The biological compartment is made up of two or three tanks (depending on the plant configuration adopted), each with a volume of 7.5 m3. After the biological compartment, a secondary settler removes solids before discharge. A sludge recycle from the secondary settler to the denitrification tank, set to be proportional to the influent flow through a sludge recycle factor, maintains sludge concentration in the tanks and brings back oxidized nitrogen in the denitrification tank, from where excess sludge is removed. When a two-tank biological compartment configuration was adopted, the first tank was used for pre-denitrification, while the second one (oxidation tank) was aerated in order to oxidize both organic carbon and ammonia (Fig. 1a). Aeration was controlled to keep a constant concentration of dissolved oxygen (DO) in the reactor. In the three-tank configuration (Fig. 1b), the first tank was still used for pre-denitrification, while both the others were aerated. A different DO set-point was fixed for the two tanks, with a higher value in the first oxidation tank and a lower in the second one. An internal recycle of 1.5 m3/h was added from the second oxidation tank into the pre-denitrification one. The influent flow rate into the biological compartment was variable, with changes every 2 h to simulate the daily flow fluctuations typical in sewer systems. Two different daily average values of about 1.0 and 1.2 m3/h were adopted (in particular, the maximum flow rate was set to be 1.3 or 1.6 m3/h in the period 8:00e10:00 a.m., while the minimum one was 0.6 or 0.7 m3/h in the period 4:00e6:00 a.m.). These influent flow rates corresponded to about 80 or 100 population equivalents (PEs) (120 gCOD PE1 d1 or 0.3 m3 PE1 d1).
2.2.
Analytical methods
The plant was equipped with various analytical instruments, including sensors for ammonia in the primary settler (Ion Selective Electrode Endress&Hauser) and DO in the biological tanks (Optical LDO, Hach Lange; Optical LDO, Endress&Hauser). All equipment was connected to a central programmable logic controller (PLC) linked to a supervisory control and data acquisition (SCADA) system for direct access and data storage, with data taken every 2 min, except for DO and air flux (data taken every 10 s). Laboratory analyses were carried out three times per week on 24-h flow proportional composite samples of influent and effluent and on grab samples from the tanks to determine total nitrogen TN (commercial photochemical test kits, Hach Lange), ammonia (Foss FIAstar flow injection 5000 analyzer), nitrites and nitrates (ion chromatography; 761 compact IC, Metrohm). Total and soluble chemical oxygen demand (COD) were measured using commercial photochemical test kits (Hach Lange), while total suspended solids (TSS) in the influent and in the effluent and mixed liquor suspended solids (MLSS) were measured according to standard methods (APHA, 1998). Off-gas concentrations in terms of O2 and CO2 (Rosemount Analytical Binos 100 2M Dual-Channel Gas Analyzer) and
N2O (Rosemount Analytical X-Stream X2) were continuously measured from the oxidation tank, which was perfectly sealed and regularly checked to avoid any loss of gas (see Supplementary data for accuracy of N2O measurements, Figures S1eS3). Data taken every 20 s were logged for further elaborations. Gas analyzers were calibrated three times per week with a two point calibration, using N2 for zero point calibration, pure gas with known concentration for N2O (200 ppm) and CO2 (2%), and ambient air for O2. Off-gas N2O measurements were processed to compute the mass of nitrogen emitted as N2O by multiplying them by the air flow rate (see Supplementary data, Figure S4). At standard conditions, 1 ppm N2O corresponds to 1:249 mgN2 O N=m3air . Sludge age was computed dividing the total amount of solids present in the biological tanks by the amount of solids discharged every day through effluent and excess sludge removal; hence, it was referred to the biological compartment, excluding the sludge contained in the secondary settler.
2.3.
Operational schedule
In order to establish steady-state equilibrium, the plant was operated for several weeks at constant sludge age before the beginning of off-gas measurements. During the first part of the monitoring campaign (two-tank configuration; period A), covering about two months, high sludge ages were tested (higher than 15 days). Different operational conditions were tested, varying some parameters as reported in Table 1 (excess sludge removal, sludge age, DO set-point). Therefore, three phases can be distinguished in period A. In the second part of the monitoring campaign (period B, three-tank configuration) sludge ages were reduced. Before starting experiments, the plant was operated for more than two months to be sure to have steady-state conditions in the new configuration. Nevertheless, N2O emissions were continuously measured even during the transitional period.
Table 1 e Operational conditions applied in period A (two-tank configuration) and B (three-tank configuration). Parameters
Period A A1
A2
Period B A3
B1
B2
Duration [d] 14 8 28 14 28 1.0 1.0 1.0 1.0 1.2 Average influent flow rate [m3/h] Return sludge 2 2 3 2.5 2.5 flow ratio MLSS [g/l] 4.1 4.1 4.3 2.3 1.6 0.6 0.6 0.75 1.15 1.80 Excess sludge [m3/d] Sludge retention 22 20 16 15 10e12 time (SRT) [d] DO set-point in 1.5 2.0 3.0 2.0 2.0 oxidation 1 [mg/l] DO set-point in e e e 0.5 0.5 oxidation 2 [mg/l] Stress tests Ammonia addition Tested Tested Tested Tested Tested DO ¼ 0.5 mg/l Tested e e e e DO ¼ 1.0 mg/l Tested e Tested Tested Tested
3566
w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 3 5 6 3 e3 5 7 3
Moreover, some stress tests were run as indicated in Table 1, consisting in the instantaneous addition of ammonia (in form of ammonium bicarbonate) or in the reduction of the DO set-point for 2 h in the oxidation tank (in period B, such tests were performed on the first oxidation tank). Unfortunately, due to bulking problems in the secondary settler, it was necessary to stop phase A2 before carrying out the other planned stress tests. The measurements were stopped for about one week in order to change the operational conditions and allow the plant to recover between phases A2 and A3. The oxidation tank was spiked with ammonia in every phase in order to observe the behavior of the system overloaded with ammonia: 480 g of (NH4)HCO3 were added instantaneously to increase the ammonia concentration by about 11.5 mgN/l. The salt was added at 10:10 a.m., in the period of maximum nitrogen influent load. Grab samples were taken from the oxidation tank to monitor nitrogen conversion. Dissolved oxygen reduction tests consisted in lowering the DO set-point from the normal value to 1.0 mg/l in the period 9:30e11:30 a.m., which corresponded to the phase of maximum nitrogen load to the tank. This value was chosen as it has been indicated in previous studies as the value that maximizes N2O emission (Ce´bron et al., 2005; Tallec et al., 2006). Only during phase A1 a second test with a DO setpoint of 0.5 mg/l was performed. Like in ammonia addition tests, grab samples were taken to follow nitrogen patterns.
3.
Results and discussion
3.1.
Nitrification performance of the plant
Influent TN values in the range between 12.9 and 43.3 mgN/l were measured during the whole monitoring campaign (mean value of 26.5 5.7 mgN/l), with average ammonia around 17.9 4.7 mgN/l and nitrites and nitrates lower than 1.0 mgN/ l. Average influent temperature was 14.9 1.4 C in period A and 19.3 0.8 C in period B, with pH being 8.0 0.4 and 7.4 0.5 for the two periods, respectively. During period A, the plant exhibited complete nitrification, with both ammonia and nitrite concentrations in the effluent below 0.5 mgN/l, and progressive reduction of nitrates from 5 mgN/l to less than 2 mgN/l. In period B, complete nitrification was achieved as well. However, ammonia was not completely removed in the first oxidation tank, especially during the morning peak in influent concentration, when some ammonia was removed in the second oxidation tank.
3.2.
Nitrous oxide emissions during normal operation
Nitrous oxide emissions from the oxidation tank during normal operation in period A (two-tank configuration) were always rather low (Fig. 2). In fact, the amount of nitrogen emitted per day from the oxidation tank as N2O (related to the total nitrogen load entering the biological compartments) was in most of the cases lower than 0.40&. Examining the trends for every week, N2O emissions generally increased during the week and reached the highest value on Sundays, with a peak around 1.15& during phase A1.
Fig. 2 e Daily fraction of the total influent nitrogen emitted as N2O during the three phases of period A. Filled markers refer to normal operation of the plant, while hollow markers refer to the results obtained during stress tests (ammonia addition or DO reduction).
Differences were observed among the three phases in which period A was divided, due to the different set-points of DO in the oxidation tank. As can be seen in Fig. 2, higher fractions of nitrogen were emitted as N2O with lower DO values (i.e., in phase A1). Discarding the results obtained with stress tests, average values of 0.41 0.32& during phase A1, 0.13 0.11& during phase A2 and 0.12 0.06& during phase A3 were measured. However, the small variation between phases A2 and A3 is due to the absence in phase A2 of data from week-end days or holidays (during which higher loads were normally observed) due to the previously mentioned bulking problems: average values calculated without such days amount to 0.29 0.16& for phase A1 and 0.04 0.02& for phase A3, respectively. The low recorded emissions can be explained as the outcome of the operation of the plant at constant DO. In the study by Ahn et al. (2010) about different plant configurations, it was observed that, in general, wider ranges of operating DO concentrations in the aerobic zones correspond to higher fractions of nitrogen emitted as nitrous oxide. Moreover, the high sludge age during period A (around 20 days during phases A1 and A2, around 16 days during phase A3) could also account for the low measured emissions, because in these conditions complete nitrification is expected, with low N2O production (Noda et al., 2003). In fact, higher sludge ages are known to cause a significant variation in the microbial community, favoring the retention of microorganisms with low maximum specific growth rates (such as nitrifying bacteria) (Yuan et al., 2008). Furthermore, as ammonia oxidizing bacteria (AOB) and nitrite oxidizing bacteria (NOB) are characterized by different maximum specific growth rates, sludge retention time (SRT) determines also the relative abundance of species belonging to the two groups (Zhu et al., 2008). Nevertheless, the effect of the DO set-point during period A dominated over the one of sludge age reduction, as the lowest emissions were recorded in period A3, which was characterized by a lower sludge age (16 days), but a higher DO (3 mg/l).
w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 3 5 6 3 e3 5 7 3
Fig. 3 e Daily fraction of the total influent nitrogen emitted as N2O from the first oxidation tank during the two phases of period B. Filled markers refer to normal operation of the plant, while hollow markers refer to the results obtained during stress tests (ammonia addition or DO reduction). If emissions from the second oxidation tank are taken into account, global emissions increase and can be even doubled at high influent nitrogen loads.
The influence of the DO set-point can be due to the fact that both AOB and NOB change their metabolic activity as a function of oxygen concentration, thereby affecting both the rates of nitrification and aerobic denitrification, together with N2O emission (Colliver and Stephenson, 2000). Compared with AOB, NOB require higher DO concentrations, due to the lower
3567
oxygen affinity (different DO half-saturation value for oxygen); therefore, AOB dominate over NOB at low DO concentrations, which results in nitrite accumulation and in the occurrence of partial nitrification (being DO values higher than 2 mg/l required for complete nitrification) (Zhu et al., 2008). Furthermore, at low DO, some processes allow for simultaneous nitrogen oxidation and reduction, e.g., simultaneous nitrification and denitrification, autotrophic nitrogen removal over nitrite or nitrifier denitrification (i.e., denitrification by AOB), leading to higher N2O emissions (Wrage et al., 2001; Blackburne et al., 2008). During period B we reduced sludge age to less than 15 days in order to better assess its impact (Fig. 3). The fraction of nitrogen emitted as N2O from the first oxidation tank was on average 0.20 0.11& in phase B1 (sludge age around 15 days) and 0.56 0.30& in phase B2 (sludge age around 10e12 days) including week-end days and holidays and 0.11 0.10& and 0.31 0.18& discarding these values, being the average daily influent TN similar in the two phases (750 g and 747 g, respectively). As apparent, at constant conditions, the reduction of sludge age gave rise to an increase in N2O emissions. It is also worth considering that the values shown in Fig. 3 for period B refer only to emissions from the first oxidation tank. However, as both ammonia and nitrite were not completely oxidized in the first stage, N2O emissions were expected also from the second tank, even if to a lower extent due to lower ammonia concentrations. In order to estimate such contribution, we have measured emissions from both tanks for a week during phase B2, switching measurements from one reactor to the other one every half an hour by means of a temporized valve. The ratio between the emission from the
Fig. 4 e Daily variability of N2O emissions over two consecutive days with the correlation with ammonia influent trends (the first day is a holiday, while the second one is a working day): (a) influent ammonia concentration (continuous line) [mgN/l] and load (dots) [gN/h]; (b) dissolved nitrogen species in the oxidation tank measured with sensors (ammonia in continuous line, nitrite in dotted line and nitrate in dash-dot line) [mgN/l]; (c) off-gas N2O concentration (continuous line) [ppm] and load (dots) [10L2 gN/h].
3568
w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 3 5 6 3 e3 5 7 3
second oxidation tank and the one from the first oxidation tank ranged from about 5% at a low influent nitrogen load to more than 90% for a daily nitrogen load higher than 800 g. Therefore, global emissions at the highest influent nitrogen loads can be even doubled with respect to the values shown in Fig. 3. In general, it is possible to observe an increase in the fraction of influent nitrogen emitted as N2O at higher influent daily nitrogen loads, considering data of the same phase (Figs. 2 and 3). Hence, influent nitrogen load is another key factor that affects nitrous oxide emissions from the nitrification stage of a wastewater treatment plant. However, emissions for similar influent nitrogen loads in the same operational conditions (i.e., in the same phase) are quite scattered. Such result can be ascribed to other factors than the actual nitrogen load, like organic carbon concentration or the history of nitrogen load during previous days (that is to say if similar, higher or lower influent concentrations were observed in the days before). More precisely, higher influent COD can increase the competition for oxygen between heterotrophs and nitrifiers, thereby reducing ammonia removal rate and favoring ammonia and nitrite accumulation in the tank. Batch tests with the sludge taken from the oxidation tank of the pilot plant have shown that
higher N2O emissions were recorded during ammonia oxidation at lower DO concentrations and correspondingly lower nitrogen conversion rates (Wunderlin et al., 2012); moreover, if organic carbon is available in excess, simultaneous nitrification/denitrification can occur, bringing about a further increase in emissions (Wunderlin et al., 2012). As to nitrogen load history, we have observed that normal loads after some days of low load (due for example to rainy periods) resulted in higher emissions if compared to periods of almost constant load. This can be explained in terms of population dynamics, as activity and distribution of the different bacterial strains is related to the available substrate. The ratio of the feed biodegradable organic carbon to the nitrogen available for nitrification is a critical factor in the performance of nitrogen removal systems, as a high C/N ratio would retard the growth of nitrifying bacteria and especially nitrite oxidizers (Lin et al., 2009). It is interesting to compare our findings with the ones of Czepiel et al. (1995), which have been proposed as a reference value (IPCC, 2006). Even if no information about the influent nitrogen load is present in that study, their emission factor of 3.2 gN2O per population equivalent per year can be expressed as 0.35& of influent nitrogen if a wastewater nitrogen load of 16 g PE1 d1 is assumed (Foley et al., 2010).
Fig. 5 e Ammonia addition stress tests in period A: (a) phase A1 (sludge age 22 days, DO [ 1.5 mg/l); (b) phase A2 (sludge age 20 days, DO [ 2.0 mg/l); (c) phase A3 (sludge age 16 days, DO [ 3.0 mg/l). The shock was achieved by instantaneously adding 85 g of N-NH4 at 10:10 a.m., while the daily influent TN load was about 600 g. The first line shows influent ammonia concentration (continuous line) and load (dots), excluding ammonia added directly in the tank; the second line presents dissolved nitrogen species in the oxidation tank (ammonia with circles and continuous line, nitrite with stars and dotted line and nitrate with squares and dash-dot line) measured on grab samples (lines have not chemical meaning, but are drawn for a clearer visualization of trends); in the third row off-gas N2O concentration (continuous line) and load (dots) are represented.
w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 3 5 6 3 e3 5 7 3
Hence, comparable or even lower average emissions were observed in our case. However, in a study conducted on a German wastewater treatment plant even much lower emissions were reported (0.01& of the total nitrogen received) (Su¨mer et al., 1995). The analysis of N2O emission normalized to the influent flow reveals that values lower than 10e20 mgN-N2O/m3 were measured during period A and B, below the average of about 20 mgN-N2O/m3 reported by Czepiel et al. (1995). As observed by Ahn et al. (2010), such a normalization, even if usually adopted for U.S. EPA or IPCC emission factors, appears to be inadequate as a result of different global water-use patterns. In fact, the lower values referred to the influent flow in our experimental campaign can be due to the rather low nitrogen concentration in the influent (average value lower than 30 mgN/l) against typical values recorded in full-scale wastewater treatment plants (40e70 mgN/l for medium and high strength untreated domestic wastewater) (Metcalf & Eddy Inc., 2003). The nitrogen load computed for our plant is around 8e9 g PE1 d1, about a half of the value of 16 g PE1 d1 assumed for developed countries (Foley et al., 2010; Kampschreur et al., 2009), but in agreement with the value
3569
of TKN of 10 g PE1 d1 suggested in German guidelines about the dimensioning of wastewater treatment plants (DWA, 2000). Another reason behind the difference in values between this study and the one of Czepiel et al. (1995) can be the variability of nitrous oxide emissions over the day. Actually, nitrous oxide emissions are not constant but exhibit a typical pattern. During period A, they were quite negligible for most of the day and presented a bell-like shape in the morning in the hours of maximum nitrogen load in the plant (between 9:00 and 12:00 during working days, between 10:00 and 14:00 during week-ends and holidays). During period B, similar trends were identified; however, emissions lasted for longer periods and a second peak was often observed in the late afternoon, corresponding to a second peak in ammonia concentration normally present in the influent. As an example, the trends recorded over two consecutive days in phase B2 (the first day is a holiday, while the second one is a working day) are depicted in Fig. 4. As can be seen, as soon as the ammonia influent reaches the peak in terms of concentration and load (computed multiplying point by point the ammonia concentration by the influent flow rate)
Fig. 6 e Ammonia addition stress tests in period B: (a) phase B1 (sludge age 15 days, DO [ 2.0 mg/l); (b) phase B2 (sludge age 10e12 days, DO [ 2.0 mg/l). The shock was achieved by instantaneously adding 85 g of N-NH4 at 10:10 a.m., while the daily influent TN load was about 850 g. The first line shows influent ammonia concentration (continuous line) and load (dots), excluding ammonia added directly in the tank; the second line presents dissolved nitrogen species in the oxidation tank (ammonia with circles and continuous line, nitrite with stars and dotted line and nitrate with squares and dash-dot line) measured on grab samples (lines have not chemical meaning, but are drawn for a clearer visualization of trends); in the third row off-gas N2O concentration (continuous line) and load (dots) are represented.
3570
w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 3 5 6 3 e3 5 7 3
(Fig. 4a), ammonia and nitrite start to accumulate in the oxidation tank (Fig. 4b). Simultaneously, N2O emission takes place, with the rate of emission increasing up to a maximum value when ammonia in the oxidation tank peaks (Fig. 4c). Depending on the amount of nitrite produced, N2O emission starts to decrease if no more nitrite is present (as in the second plotted day) or maintains a constant value until nitrite concentration peaks (as in the first plotted day). Nevertheless, when both ammonia and nitrite are removed, N2O emission ceases. A similar profile was detected in the first aerobic zone of the second pass of a full-scale step-feed BNR process, with a close correlation between mixed-liquor ammonia and nitrite peak concentrations and N2O gas concentrations (Ahn et al., 2010). The significant variations in emissions over a day and a week detected in our study indicate that great attention should be paid to this aspect when planning a monitoring campaign at a full-scale wastewater treatment plant, especially for plants devoid of an equalizing tank and, hence, subject to flow and load variability. In order to avoid systematic errors in the average estimates, measures should be extended over longer time frames to encompass both the daily and weekly variability, as sampling campaigns concentrated in few hours would neglect the variations of emissions over the day.
3.3.
Stress tests: ammonia addition
Fig. 5 compares the results obtained with the ammonia addition tests in the different phases of period A, while Fig. 6 reports the data referred to period B. Similar trends were observed in dissolved nitrogen species in all the tests. The reduction of ammonia concentration in the tank was partially due to transport as an effect of the influent and recycle flow; hence, not all the ammonia was really converted but it was partially lost in the effluent (period A) or nitrified in the second nitrification tank (period B), as demonstrated by ammonia concentration trends in effluent or in the second nitrification tank (data not shown). No high nitrite accumulation was observed in any of the three phases of period A, while some nitrite build-up was measured in period B (up to 1.7 mgN/l in phase B1, Fig. 6a). A minor increase in nitrate concentration was recorded as well. Even if N2O emission was already present to some extent, a sharp increase in N2O concentration and load was recorded soon after the ammonia addition, until a maximum value was reached after about 20e30 min. Afterward, emissions leveled out at the maximum for about 2.5e3.0 h in phases A1 and A2 (Fig. 5a and b) and less than 2 h in phases A3, B1 and B2 (Figs. 5c and 6). Then, they decreased again to the values observed
Fig. 7 e DO reduction stress tests from 9:30 a.m. to 11:30 a.m. in period A: (a) from 1.5 mg/l (normal DO set-point) to 0.5 mg/l in phase A1 (sludge age 22 days); (b) from 1.5 mg/l (normal DO set-point) to 1.0 mg/l in phase A1 (sludge age 22 days); (c) from 3.0 mg/l (normal DO set-point) to 1.0 mg/l in phase A3 (sludge age 16 days). The first line shows influent ammonia concentration (continuous line) and load (dots); the second line presents dissolved nitrogen species in the oxidation tank (ammonia with circles and continuous line, nitrite with stars and dotted line and nitrate with squares and dash-dot line) measured on grab samples (lines have no chemical meaning, but are drawn for a clearer visualization of trends); in the third row off-gas N2O concentration (continuous line) and load (dots) are represented.
w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 3 5 6 3 e3 5 7 3
before the ammonia addition and became negligible after a period ranging from less than 3 h in phases A3 and B2 to 4e5 h in the other phases. Interesting evaluations can be done observing the trend of the cumulative emissions of N2O (Supplementary data, Figure S5). Our results point out how dilution due to hydraulic conditions plays a major role in the attenuation of the ammonia peak and, consequently, in N2O emission. Keeping the same hydraulics (phases A1 and A2, phases A3 and B1), the extent of the N2O peak was related to the DO setpoint in the tank, with a higher DO value allowing lower N2O emissions. It can be inferred that a plant operated at higher DO concentrations is able to better afford ammonia overloading, with lower N2O emissions. The results obtained in our study can be compared with those of Burgess et al. (2002), who tested ammonia shocks with increasing amounts of ammonia in the range 0.48e3.30 mgNH4-N gTSS1 in a lab-scale activated sludge reactor. In our study, the shock loading normalized to MLSS is about 2.7e2.8 mgNH4-N gTSS1 for period A, thus within the range tested by Burgess et al. (2002), while the values in period B (5.0e7.2 mgNH4-N gTSS1) are much higher.
3571
Considering their results, a value around 10& should have been achieved in our tests in period A, higher than the values around 1.5e5.0& of total nitrogen obtained. Such a difference can be due to the different plant conditions in the study by Burgess et al. (2002) (higher hydraulic retention times, lower MLSS, different oxygen control in the tank). However, if we refer the nitrous oxide emitted only to the added ammonia (instead of considering the total nitrogen load including that of the influent), we get values in the range 6e20&, closer or higher than those of Burgess et al. (2002).
3.4.
Stress tests: DO reduction
Figs. 7 and 8 show the results obtained for DO reduction tests during period A and B, respectively. No nitrite accumulation was observed in any of the tests. However, some loss in the nitrification capacity could be recognized in the reduction of nitrate concentrations and in a slight accumulation of ammonia (even up to 3 mgN/l in the test at DO ¼ 0.5 mg/l in phase A1, as shown in Fig. 7a). A different trend is visible during phase B1 (Fig. 8a), with no loss in nitrification (no accumulation of ammonia, nitrite
Fig. 8 e DO reduction stress tests from 9:30 a.m. to 11:30 a.m. in period B: (a) from 2.0 mg/l (normal DO set-point) to 1.0 mg/l in phase B1 (sludge age 15 days); (b) from 2.0 mg/l (normal DO set-point) to 1.0 mg/l in phase B2 (sludge age 10e12 days). The first line shows influent ammonia concentration (continuous line) and load (dots); the second line presents dissolved nitrogen species in the oxidation tank (ammonia with circles and continuous line, nitrite with stars and dotted line and nitrate with squares and dash-dot line) measured on grab samples (lines have no chemical meaning, but are drawn for a clearer visualization of trends); in the third row off-gas N2O concentration (continuous line) and load (dots) are represented.
3572
w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 3 5 6 3 e3 5 7 3
build-up up to 1 mgN/l and increase in nitrate concentration). Compared to the ammonia addition tests (Figs. 5 and 6), also in DO reduction tests N2O emission started to increase soon after the stress application (at 9:30 a.m.), but with a more gradual slope. In the tests conducted during phase A1 and in the one in phase B2 (Figs. 7a,b and 8a), N2O emission persisted for several hours (even more than 6 h), while it became quite negligible in phases A3 and B1 less than an hour after the restoration of the normal DO value (Figs. 7c and 8a). A two-bell trend was observed in phase A1 in the test at DO ¼ 0.5 mg/l (Fig. 7a), with N2O emissions increasing again after the restoration of the normal set-point of 1.5 mg/l to values higher than those registered during the DO reduction. Such behavior can be justified by the fact that limited nitrification occurred at DO ¼ 0.5 mg/l, leading to accumulation of ammonia, which was further degraded when DO was increased again, resulting in N2O production. Moreover, another possible explanation is that the air flow necessary to maintain a DO of 0.5 mg/l was not enough to strip the N2O produced, which was then emitted when the air flow rate increased again to re-establish the higher DO concentration. A similar increase in N2O off-gas concentrations was observed by Burgess et al. (2002a), who tested an hour oxygen depletion, starting from initial DO values of 2.5 mg/l and 2.9 mg/l (with peak values of N2O observed 1 h after DO fell to <1.0 mg/l, followed by a decrease to pre-test levels after further 3.0e3.5 h). As demonstrated by the analysis of the cumulative emission of N2O (Supplementary data, Figure S6), lower emissions were achieved when the plant was normally operated at higher DO values (phase A3 with respect to phase A1). However, sludge age seems to play an important role, as the highest emissions during phase B2 were concomitant with the lowest tested sludge age (10e12 days).
4.
Increased nitrogen loads result in higher N2O emissions. However, as nitrous oxide emissions are correlated with the influent ammonia peak, lower emissions are to be expected if good strategies are adopted to achieve constant ammonia loads (for example, through influent equalization or anaerobic supernatant recycle in the plant in the period of lower influent ammonia load). Stress conditions (ammonia overloading, DO reduction) increase N2O emissions. However, when the system is normally operated at higher DO values, the plant is able to better withstand disturbance, resulting in lower emissions in terms of both peak concentration and duration. Higher recycle rates (sludge recycle and internal recirculation) contribute to the reduction of N2O emissions, because they attenuate the peak of ammonia and dilute the concentration of all the intermediates of nitrification/ denitrification.
Acknowledgments This study was financially supported by the Swiss Federal Office for the Environment (FOEN), the canton of Bern (AWA), the canton of Basel-Landschaft (AIB), the canton of St. Gallen (AFU), the canton of Thurgau (AfU), the canton of Zurich (AWEL), TBF þ Partner AG Consulting Engineers, WWTP region Bern, WWTP ProRheno, WWTP REAL Luzern, WWTP Worblental, WWTP Zurich-Werdho¨lzli, Abwasserverband Altenrhein, Abwasserverband Morgental, Entsorgung St. Gallen, Zentralschweizer Umweltdirektorenkonferenz (ZUDK) and Eawag.
Appendix A. Supplementary material Supplementary data related to this article can be found online at doi:10.1016/j.watres.2012.03.067.
Conclusions
Nitrous oxide emissions exhibit a typical pattern during the day, being quite negligible for most of the time and presenting a bell-shape form in the morning during the hours in which the maximum nitrogen load in the plant is recorded. Weekly variability is also present, with higher emissions during week-end. Attention should be paid to these trends to avoid systematic errors during full-scale monitoring. DO set-point in the oxidation tank plays a crucial role in N2O emissions as it influences ammonia removal rates. In order to prevent N2O emissions, the first aerobic tank in a wastewater treatment plant should not be operated at DO values lower than 2.0 mg/l. Low sludge ages (10e12 days) resulted in higher emissions from the first oxidation tank, even if average absolute values still remained quite low (mostly <1& of the total nitrogen in the influent). However, higher global emissions (even doubled) were obtained when considering also emissions from the second nitrification tank in the three-tank plant configuration.
references
APHA, AWWA, WEF, 1998. Standard Methods for the Examination of Water and Wastewater, 20th ed. American Public Health Association, American Water Works Association, Water Environment Federation, Washington, DC, USA, 1220 pp. Ahn, J.H., Kim, S., Park, H., Katehis, D., Pagilla, K., Chandran, K., 2010. Spatial and temporal variability in atmospheric nitrous oxide generation and emission from full-scale biological nitrogen removal and non-BNR processes. Water Environment Research 82 (12), 2362e2372. Benckiser, G., Eilts, R., Linn, A., Lorch, H.J., Su¨mer, E., Weiske, A., Wenzho¨fer, F., 1996. N2O emissions from different cropping systems and from aerated, nitrifying and denitrifying tanks of a municipal waste water treatment plant. Biology and Fertility of Soils 23, 257e265. Blackburne, R., Yuan, Z., Keller, J., 2008. Demonstration of nitrogen removal via nitrite in a sequencing batch reactor treating domestic wastewater. Water Research 42, 2166e2176. Burgess, J.E., Colliver, B.B., Stuetz, R.M., Stephenson, T., 2002. Dinitrogen oxide production by a mixed culture of nitrifying bacteria during ammonia shock loading and aeration failure.
w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 3 5 6 3 e3 5 7 3
Journal of Industrial Microbiology and Biotechnology 29, 309e313. Burgess, J.E., Stuetz, R.M., Morton, S., Stephenson, T., 2002a. Dinitrogen oxide detection for process failure early warning systems. Water Science and Technology 45 (4e5), 247e254. Ce´bron, A., Garnier, J., Billen, G., 2005. Nitrous oxide production and nitrification kinetics by natural bacterial communities of the lower Seine river (France). Aquatic Microbial Ecology 41, 25e38. Colliver, B.B., Stephenson, T., 2000. Production of nitrogen oxide and dinitrogen oxide by autotrophic nitrifiers. Biotechnology Advances 18, 219e232. Czepiel, P., Crill, P., Harriss, R., 1995. Nitrous oxide emissions from municipal wastewater treatment. Environmental Science and Technology 29, 2352e2356. DWA (German Association for Water, Wastewater and Waste), 2000. Dimensioning of Single-Stage Activated Sludge Plants ATV-DVWK-A 131E. DWA, Hennef, Germany, 54 pp. Foley, J., Lant, P., 2008. Fugitive Greenhouse Gas Emissions from Wastewater Systems: WSAA Literature Review No. 01. Water Services Association of Australia, Melbourne. Foley, J., de Haas, D., Yuan, Z., Lant, P., 2010. Nitrous oxide generation in full-scale biological nutrient removal wastewater treatment plants. Water Research 44, 831e844. Hu, Z., Zhang, J., Li, S., Xie, H., Wang, J., Zhang, T., Li, Y., Zhang, H., 2010. Effect of aeration rate on the emission of N2O in anoxicaerobic sequencing batch reactors (A/O SBRs). Journal of Bioscience and Bioengineering 109 (5), 487e491. IPCC, 2001. In: Houghton, J.T., Ding, Y., Griggs, D.J., Noguer, M., van der Linden, P.J., Dai, X., Maskell, K., Johnson, C.A. (Eds.), Climate Change 2001: The Scientific Basis. Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change Published for the Intergovernmental Panel on Climate Change. Cambridge University Press, United Kingdom and New York, NY, USA, p. 881. IPCC, 2006. In: Eggleston, H.S., Buendia, L., Miwa, K., Ngara, T., Tanabe, K. (Eds.), 2006 IPCC Guidelines for National Greenhouse Gas Inventories. IGES, Japan. Kampschreur, M.J., Temmink, H., Kleerebezem, R., Jetten, M.S.M., van Loosdrecht, M.C.M., 2009. Nitrous oxide emission during wastewater treatment. Water Research 43, 4093e4103.
3573
Kimochi, Y., Inamori, Y., Mizuochi, M., Xu, K.Q., Matsumura, M., 1998. Nitrogen removal and N2O emission in a full-scale domestic wastewater treatment plant with intermittent aeration. Journal of Fermentation and Bioengineering 86 (2), 202e206. Lin, Y.M., Tay, J.H., Liu, Y., Hung, Y.-T., 2009. Biological nitrification and denitrification processes. In: Wang, L.K., et al. (Eds.), Handbook of Environmental Engineering. Biological Treatment Processes, vol. 8. The Humana Press, Totowa, NJ, pp. 539e588. Ch. 13. Noda, N., Kaneko, N., Mikami, M., Kimochi, Y., Tsuneda, S., Hirata, A., Mizuochi, M., Inamori, Y., 2003. Effects of SRT and DO on N2O reductase activity in an anoxic-oxic activated sludge system. Water Science and Technology 48 (11e12), 363e370. Metcalf & Eddy Inc., 2003. Wastewater Engineering: Treatment and Reuse, fourth ed. McGraw-Hill, Boston, USA, 1848 pp. Ravishankara, A.R., Daniel, J.S., Portmann, R.W., 2009. Nitrous oxide (N2O): the dominant ozone-depleting substance emitted in the 21st century. Science 326, 123e125. Su¨mer, E., Weiske, A., Benckiser, G., Ottow, J.C.G., 1995. Influence of environmental conditions on the amount of N2O released from activated sludge in a domestic waste water treatment plant. Experientia 51, 419e422. Tallec, G., Garnier, J., Billen, G., Gousailles, M., 2006. Nitrous oxide emissions from secondary activated sludge in nitrifying conditions of urban wastewater treatment plants: effect of oxygenation level. Water Research 40, 2972e2980. Wrage, N., Velthof, G.L., van Beusichem, M.L., Oenema, O., 2001. Role of nitrifier denitrification in the production of nitrous oxide. Soil Biology and Biochemistry 33, 1723e1732. Wunderlin, P., Mohn, J., Joss, A., Emmenegger, L., Siegrist, H., 2012. Mechanisms of N2O production in biological wastewater treatment under nitrifying and denitrifying conditions. Water Research 46, 1027e1037. Yuan, Z., Oehmen, A., Peng, Y., Ma, Y., Keller, J., 2008. Sludge population optimisation in biological nutrient removal wastewater treatment systems through on-line process control: a re/view. Reviews in Environmental Science and Biotechnology 7, 243e254. Zhu, G., Peng, Y., Li, B., Guo, J., Yang, Q., Wang, S., 2008. Biological removal of nitrogen from wastewater. Reviews of Environmental Contamination and Toxicology 192, 159e195.