| |
337 Q IWA Publishing 2008 Water Science & Technology WST 58.2 2008
Optimisation of sanitary landfill leachate treatment
in a sequencing batch reactor
A. Spagni, S. Marsili-Libelli and M. C. Lavagnolo
ABSTRACT
A bench-scale SBR was operated for almost three years in an attempt to optimise the treatment A. Spagni
ENEA, Italian National Agency for New
of leachates generated in old landfill. The results of the first two years were used to design
Technologies, Energy and the Environment,
Environment Department, Water Resource
a monitoring and control system based on artificial intelligence concepts. Nitrogen removal was
Management Section, Via M. M. Sole 4,
40129 Bologna,
optimized via the nitrite shortcut. Nitrification and N removal were usually higher than 98%
Italy
and 90%, respectively, whereas COD (of the leachate) removal was approximately 30 40%. E-mail: [e-mail]
The monitoring and control system was demonstrated capable of optimizing process operation,
S. Marsili-Libelli
Department of Systems and Computers,
in terms of phase length and external COD addition, to the varying loading conditions. Using
University of Florence, Via S. Marta 3,
50139 Florence,
the control system developed, a significant improvement of the process was obtained: COD
Italy
and N load were increased (HRT decrease) and a significant decrease (approximately 34%) E-mail: [e-mail]
of the ratio of COD added to N leachate content was observed.
M. C. Lavagnolo
Department of Hydraulic, Maritime, Environmental
|
Key words denitritation, fuzzy control, nitritation, nitrite shortcut, nitrogen removal
and Geotechnical Engineering (IMAGE),
University of Padua, Via Loredan, Padova,
Italy
E-mail: [e-mail]
INTRODUCTION
Sanitary landfill leachate treatment is usually accomplished have been proposed for nitrogen removal optimization via
by multistage systems using chemical, physical and biologi- nitrite (reviewed by Villaverde 2004). Among these pro-
cal processes. Leachate generated in old landfills is a high- cesses, nitrite build-up may be sustained by optimizing
strength wastewater characterized by a low BOD/TKN phase duration in SBRs, switching nitritation process to
ratio. Therefore, biological nitrogen removal can be denitritation once the maximum nitrite concentration has
achieved only if an external biodegradable COD source is been reached (Abeling & Seyfried 1992; Fux et al. 2006).
provided for the denitrification process (Lema et al. 1988; Dissolved oxygen (DO), pH and oxidation-reduction
Kjeldsen et al. 2002). Among several technologies, sequen- potential (ORP) have been frequently used for monitoring
cing batch reactors (SBRs) have been demonstrated to be and control of batch reactors (Spagni et al. 2001). The
feasible for biological leachate treatment (EPA 1995). majority of studies using these process measurements have
Nitrogen removal from wastewaters is usually accom- been focussed on municipal wastewaters (e.g. Wareham
plished through nitrification and denitrification processes. et al. 1993; Battistoni et al. 2003), though a few have also
Instead of using the full nitrification/denitrification path, been carried out on industrial (Li et al. 2004) or agricultural
biological nitrogen removal via nitrite is a promising wastewaters (Ra et al. 1997; Cheng et al. 2000). In the last
alternative for the optimization of nitrogen removal, in few years, some applications of artificial intelligence, such
particular in the presence of a low biodegradable COD to as fuzzy logic (Marsili-Libelli 2006), have been reported for
TKN ratio. Nitrite pathway decreases the oxygen demand wastewater treatment monitoring and control.
and the carbon consumption up to 25% and 40%, In the present study a lab-scale SBR treating leachate
respectively. During the last decade, several processes from an old landfill was kept in operation for almost three
doi: 10.2166/wst.2008.399
| | |
338 A. Spagni et al. Optimisation of sanitary landfill leachate treatment in SBR Water Science & Technology WST 58.2 2008
years in order to optimise nitrogen removal and external
COD addition (used for denitrification). During the first
year the SBR was extensively monitored in order to
optimise nitrogen removal using conventional nitrification
and denitrification procedures. During the second year, the
plant was operated in order to accumulate nitrite in an
attempt to improve nitrogen removal (via nitritation and
denitritation process) and external COD addition. During
the third year a fuzzy control system (based on DO, pH and
ORP process signals) was applied to supervise the switching
sequence of the phases.
Figure 1 | Total (t) and filtered (f) COD in the influent (CODt_in, CODf_in) and effluent
(CODt_out, CODf_out).
EP the plant was operated by a control system based on
METHODS
fuzzy logic. As a result (during I and II EP), each sub-cycle
A lab-scale SBR, with a maximum working volume of 24 L,
was operated starting with an anoxic-anaerobic phase of 1.0
treating raw leachate originating from an old municipal
to 2.0 h hours followed by an oxic phase of 3.75 to 4.75
landfill, was operated for more than 900 days in a
hours (with a constant reaction of 5.75 hours). At onset of
thermostatic room at 20 ^ 0.58C. Initially, the SBR was
the anoxic-anaerobic phase (of each sub-cycle), leachate
operated with a  full -cycle of 24 hours divided in series of 4
(flow of 1.2 L/h) was added to the tank. In order to supply
sub-cycles of 5.75 hours, followed by one hour of settling.
biodegradable COD for denitrification, a concentrated
During the present study, operational conditions were
solution (20 g/L: 9.4 gCOD/L) of sodium acetate trihydrate
modified according to leachate characteristics. In particu-
was added during the anoxic-anaerobic phase (flow of
lar, due to the large variations registered in leachate
0.36 L/h during the I and II EP and 0.14 L/h during the III).
strength (Table 1, Figures 1 and 2a), the length of anoxic
The external COD was neglected in the calculation of the
and aerobic phases, feed load and the sludge age were
organic loading rate (OLR) and of the COD removal
modified in accordance with leachate concentration.
efficiency. The effluent was drawn during the last 3 minutes
During the I and II experimental period (EP) operational
of the settling phase to reach a minimum reactor volume
conditions were modified manually, whereas during the III
of 15 L. During the last minute of the fourth sub-cycle, a
small amount of mixed liquor was drawn in order to
Table 1 | Leachate characteristics
control the suspended solids concentration in the reactor:
Unit Mean Max Min
the mean solid retention time was approximately 25 days
TSS g/L 0.25 1.10 0.03 (with wide variations due to time-varying operational
VSS g/L 0.14 0.54 0.02 conditions).
TKN mgN/L 1,191 1,812 252 The plant was extensively monitored with analytical
NHþ mgN/L 1,061 1,540 167 measurements (according to Standard Methods 1998), and
4
pH  8.05 8.90 7.55 using pH, ORP and DO on-line signals. More details about
Alkalinity (to pH 4.3) meq/L 119 162 35
the monitoring methods and results and plant layout are
Conductivity (208C) mS/com 14.2 19.7 5.6
reported in Spagni et al. (2007).
Ptot mgP/L 5.7 9.5 2.1
Four experimental periods can be identified:
PO32 mgP/L 4.6 9.0 0.3
4
SU (start-up): the plant was seeded with sludge from a
BOD5 mg/L 301 1,000 30
municipal wastewater treatment plant.
CODt mg/L 1,759 3,060 528
I: the reactor was operated in order to optimize nitrogen
CODf mg/L 1,620 2,980 440
removal via nitrification and denitrification processes.
| | |
339 A. Spagni et al. Optimisation of sanitary landfill leachate treatment in SBR Water Science & Technology WST 58.2 2008
Figure 2 | TKN and ammonia concentration in the influent (a); ammonia, nitrite and nitrate in the effluent (b); and nitrification and nitrogen removal efficiency (c).
II: the reactor was operated in an attempt to optimize the decision chain, the phase termination signal activates
nitrogen removal and external COD addition via the relevant actuators, thus closing the control loop
nitritation and denitritation. During this experimental (the control algorithm is described in Marsili-Libelli et al.
2007). During the first part of the III EP (IIIa) the SBR
period the phases length (in particular the fill phase)
was operated with fixed timed phases (in order to
were manually modified (almost every day) according
stabilize the process to the new operation and verify
the analytical measurements and the behaviour of pH,
the algorithm) whereas during the second part (IIIb) the
ORP and DO signals (Spagni et al. 2007).
fuzzy supervisor took over the operation entirely,
III: a fuzzy supervisory system was introduced to identify
determining the duration of the anoxic and the oxic
and manage the correct switching sequence of the plant.
phase, and the addition of the external COD.
The control system performed the phase-end detection
and managed the on/off switching of the blower, mixer
The fuzzy inferential system used in the III EP was
and pumps (filling, acetate addition, sludge and effluent
developed in the LabView 7.1 platform (National Instru-
withdrawal). The monitoring and control system (III EP)
ments, Austin TX, USA) and provided both local and
is based on a number of successive operations on the
remote control through the Internet.
data. Upon acquisition, the data are validated and
Table 1 shows some of the characteristics of the raw
denoised using a wavelet filter, then numerical deri- sanitary landfill leachates used for the entire duration of the
vation is performed and a fuzzy inference algorithm is
study. The leachates were characterised by high nitrogen
used to detect the end of the current phase. At the end of
content with respect to COD and BOD, which is typical of
| | |
340 A. Spagni et al. Optimisation of sanitary landfill leachate treatment in SBR Water Science & Technology WST 58.2 2008
old landfills. Due to the low P concentration (relative to N treatment phosphorus limitation can inhibit biological
and COD), a concentrated solution of KH2PO4 was added processes and analytical measurements in the effluent
to the SBR to maintain phosphate concentration in the revealed a phosphate concentration lower than 0.1 mgP/L,
reactor effluent between 0.5 and 1.0 mgP/L. a solution of potassium phosphate was subsequently
added to the SBR maintaining the effluent phosphate
concentration higher than 0.5 mgP/L. After about 200 days
RESULTS AND DISCUSSION
of operation a malfunctioning of the pump of phosphorus
addition occurred and the phosphate concentration again
Leachate showed very high variation in COD and nitrogen
decreased below 0.1 mgP/L causing nitrification inhibition
concentration (Figures 1 and 2a). As a result, the COD
for the second time. Furthermore, to coincide with change of
removal efficiency varied widely, between 20 and 60% (the
the leachate, sometimes a slight (approximately 50%)
external COD added for denitrification was neglected in the
reduction in nitrification efficiency was observed. When
calculation), occasionally reaching negative values (below
nitrification inhibition was observed the loading rate was
zero). This has two different explanations. First, the very
decreased; with the temporary decrease of the load, nitrifica-
high variability in leachate COD concentration caused a
tion activity was recovered within a few days. During the II
sort of  memory in the reactor and, therefore, when the
EP, with daily manual adjustments of the load and length of
COD had a sudden decrease, the effluent concentration was
still influenced by the bulk liquid present in the reactor. This the SBR phases, nitrite built-up was observed but the process
effect is visible, for example, after approximately 300 day of was quite unstable (Figure 2b). The instability was demon-
operation when COD in the influent decreased from 3,060 strated by the high concentration not only of nitrite but also of
to 1,390 first and then to 715 in approximately forty days. ammonia and nitrate: in fact, the N removal efficiency (h)
The same also occurred between day 410 and 470. was also affected by the incorrect operation of the SBR
Secondly, a very high effluent COD concentration was and during EP II it was the lowest of the study (Figure 2c and
measured when sludge showed poor settling characteristics Table 2). The observed instability was mainly caused by
and wash out of suspended solids was observed (between the inability to correctly regulate the phase length by means
day 400 and 450). It is worth mentioning that the lowest of manual adjustment. On the contrary, N removal during
settling characteristic of the sludge was measured during the other experimental periods was higher (92 95%) than
the II EP. This could be explained by the occurrence of low during the II EP (Table 2). During EP IIIa the nitrification
DO concentration in the oxic phase; indeed, because DO and denitrification processes restarted showing good
limitation (among other operational parameters) seems to nitrification (98%) and nitrogen removal (95%) efficiency
facilitate the nitritation process (Garrido et al. 1997), (Figure 2c and Table 2). Immediately after the activation
dissolved oxygen was kept at low concentrations (between of the fuzzy controller (EP IIIb), nitrite built-up was
0.5 and 1 mg/L) during II EP, in an attempt to optimise observed and nitritation and denitritation processes
nitrite accumulation. occurred (Figure 2). Contrary to the EP II behaviour,
During EP III, COD removal was stable at an average nitrite built-up during EP IIIb was stable and nitrate
value of approximately 30%. It is not possible to claim concentration was usually below 1.0 mgN/L. The low
that the stability was caused by the control system ammonia concentration in the effluent (below 10.0 mgN/L)
because during this EP the influent COD concentration confirms the good nitrification efficiency (Table 2). The
showed the lowest variability of the entire three-year very high nitrite concentration (243 mgN/L) observed
period (Figure 1). approximately at day 900, was due to a break down of the
The plant exhibited a generally good nitrification, pump of acetate addition. It is noteworthy that the control
reaching levels of more than 99%, with the exception of a system was able to decrease the nitrite concentration in a few
few cases in which inhibition occurred (Figure 2). After days once the pump was repaired.
approximately 100 days of operation the SBR produced a The SBR was started up applying an OLR of approxi-
severe case of nitrification inhibition. Because in leachate mately 0.3 gCOD/(L p d) and a nitrogen load rate (NLR) of
| | |
341 A. Spagni et al. Optimisation of sanitary landfill leachate treatment in SBR Water Science & Technology WST 58.2 2008
Table 2 | Summary of the main operational results: data as average (standard deviation)
Experimental period (days, from to)
I (45 308) II (309 649) IIIa (650 734) IIIb (735 936)
h Nitrification (2) 0.98 (0.049) 0.95 (0.077) 0.98 (0.023) 0.99 (0.0025)
h N removal (2) 0.92 (0.12) 0.84 (0.20) 0.95 (0.032) 0.95 (0.050)
HRT (d) 7.22 (3.78) 5.63 (2.41) 7.85 (2.03) 5.80 (2.17)
Load COD [gCOD/(L p d)] 0.166 (0.065) 0.208 (0.093) 0.164 (0.053) 0.288 (0.085)
Load TKN [gN/(L p d)] 0.121 (0.048) 0.144 (0.065) 0.096 (0.030) 0.189 (0.057)
Load NH4-N [gN/(L p d)] 0.114 (0.041) 0.120 (0.064) 0.087 (0.028) 0.174 (0.055)
Hac/TKN (gCOD/gN) 4.68 (1.78) 3.87 (1.91) 4.15 (1.81) 2.73 (0.86)
Hac/NH4-N (gCOD/gN) 4.91 (1.83) 5.06 (2.95) 4.64 (1.96) 2.98 (0.99)
0.2 gN/(L p d), with a hydraulic retention time (HRT) of 8 750 mgN/L) and, therefore, acetate was added to improve
days (Figure 3). At these loads ammonia accumulated in the the denitrification process. During the entire experimental
reactor up to a concentration of approximately 450 mgN/L period both HRT and loading (COD and N) were very
(Figure 2). Therefore, the load was immediately decreased variable because of the wide variation of leachate charac-
causing an immediate increase in the nitrification efficiency. teristics. The high HRT measured in this SBR is typical for
Due to the low COD/N ratio of the leachate (Table 1), landfill leachate treatment. It is noteworthy that during EP
nitrate accumulated in the reactor (up to approximately II and EP IIIb the SBR showed (as average values) the
Figure 3 | HRT (a), COD (b), TKN and ammonia loading rate (c).
| | |
342 A. Spagni et al. Optimisation of sanitary landfill leachate treatment in SBR Water Science & Technology WST 58.2 2008
and 90%, respectively: COD removal was approximately
30 40% (as average value) due to the low biodegradability
of organic matter present in leachate from old landfills.
External COD was needed to accomplish the denitrification
process.
The study confirms the effectiveness of the nitrite path
for N removal optimisation in leachate treatment, in
particular when external COD has to be added to improve
the denitrification process. Due to the variations of the
leachate characteristics, a control system based on artificial
intelligence concepts was designed and engineered to
Figure 4 | Ratio of acetate added to TKN and ammonia concentration in the influent.
monitor and operate the SBR. With this control system a
lowest HRT and the highest loads (Table 2) demonstrating
significant improvement of the process was obtained: the
the effectiveness of using the  nitrite short-cut for nitrogen
COD and N load were increased (and HRT decreased).
removal optimization. Therefore, the best improvement in
Moreover, it is noteworthy that, using the control system, a
HRT and loading seems to be related to the application of
significant decrease (approximately 34%) in the ratio of the
the control system (Table 2).
amount of external COD added to N leachate content was
Figure 4 shows that applying the nitrite short-cut a large
also obtained.
saving in external COD addition can be obtained. In fact, the
Hac/TKN ratio during EP II was 17% lower than during EP
REFERENCES
I whereas that ratio in EP IIIb was 34% lower than during
EP IIIa. Similar results were obtained for Hac/NH4-N ratio
Abeling, U. & Seyfried, C. F. 1992 Anaerobic-aerobic treatment of
during EP IIIb and IIIa. On the contrary, the Hac/NH4-N
highstrength ammonium wastewater nitrogen removal via
nitrite. Water Sci. Technol. 26(5 6), 1007 1015.
ratio during EP II was similar (or a little higher) to that
Battistoni, P., De Angelis, A., Boccadoro, R. & Bolzonella, D. 2003
measured during EP I. Again, this finding could be
An automatic controlled alternate oxic anoxic process for
explained by the wide variation in leachate strength;
small municipal wastewater treatment plants. Ind. Eng. Chem.
moreover, during this experimental period the leachate Res. 42(3), 509 515.
Cheng, N., Lo, K. V. & Yip, K. H. 2000 Swine wastewater
presented the lowest strength and the highest TKN/NH4-N
treatment in a two stage sequencing batch reactor using real-
ratio (average values). The variability of the ratio of the
time control. J. Environ. Sci. Health B35(3), 379 398.
COD added to the nitrogen content of the leachate also
EPA Manual Ground Water and Leachate Treatment Systems 1995
U.S. Environmental Protection Agency, Washington, DC,
depends on the biodegradable organic matter in the
EPA/925/R-94/005. pp. 119.
leachate. It is of note that during the application of the
Fux, C., Velten, S., Carozzi, V., Solley, D. & Keller, J. 2006 Efficient
control system the Hac/TKN and Hac/NH4-N ratios
and stable nitritation and denitritation of ammonium-rich
showed the lowest standard deviations, compared to the
sludge liquor using an SBR with continuous loading. Water
Res. 40, 2765 2775.
same ratio during the other experimental periods (approxi-
Garrido, J. M., van Benthum, W. A. J., van Loosdrecht, M. C. M. &
mately 1/2). This finding seems to highlight the effectiveness
Heijnen, J. J. 1997 Influence of dissolved oxygen concentration
of the control system in adding an accurate amount of
on nitrite accumulation in a biofilm airlift suspension reactor.
external COD. Biotechnol. Bioeng. 53(2), 168 178.
Standard Methods for the Examination of Water and Wastewater
1998 20th edition, American Public Health Association/
American Water Works Association/Water Environment
CONCLUSIONS
Federation, Washington DC, USA.
Kjeldsen, P., Balzan, M. A., Rooker, A. P., Baun, A., Ledin, A. &
The SBR process proved itself as a suitable technology for
Christensen, T. H. 2002 Present and long-term composition of
biological treatment of leachates resulting from old landfills.
MSW landfill leachate, a review. Crit. Rev. Environ. Sci.
Nitrification and N removal were usually higher than 98% Technol. 32(4), 297 336.
| | |
343 A. Spagni et al. Optimisation of sanitary landfill leachate treatment in SBR Water Science & Technology WST 58.2 2008
Lema, J. M., Mendez, R. & Blazquez, R. 1988 Characteristics of Spagni, A., Buday, J., Ratini, P. & Bortone, G. 2001 Experimental
landfill leachates and alternatives for their treatment, a review. considerations on monitoring ORP, pH, conductivity
Water Air Soil Pollut. 40(3 4), 223 250. and dissolved oxygen in nitrogen and phosphorus
Li, Y. Z., Peng, C. Y., Peng, Y. Z. & Wang, P. 2004 Nitrogen biological removal processes. Water Sci. Technol. 43(11),
removal from pharmaceutical manufacturing wastewater via 197 204.
nitrite and the process optimization with on-line control. Spagni, A., Lavagnolo, M. C., Scarpa, C., Vendrame, P.,
Water Sci. Technol. 50(6), 25 30. Rizzo, A. & Luccarini, L. 2007 Nitrogen removal
Marsili-Libelli, S. 2006 Control of SBR switching by fuzzy pattern optimization in a sequencing batch reactor treating
recognition. Water Res. 40, 1095 1107. sanitary landfill leachate. J. Environ. Sci. Health A 42,
Marsili-Libelli, S., Spagni, A. & Susini, R. 2007 Intelligent 757 765.
monitoring system for long-term control of Sequencing Batch Villaverde, S. 2004 Recent developments on biological nutrient
Reactors. Proceedings of AutMoNet 2007 IWA Conference, removal processes for wastewater treatment. Rev. Environ. Sci.
Gent (B). 5 7 Sept. 2007. Bio/Technol. 3, 171 183.
Ra, C. S., Lo, K. V. & Mavinic, D. S. 1997 Swine wastewater Wareham, D. G., Hall, K. J. & Mavinic, D. S. 1993 Real-time
treatment by batch-mode 4-stage process, loading rate control control of wastewater treatment systems using ORP. Water Sci.
using ORP. Environ. Technol. 18(6), 615 622. Technol. 28(11 12), 273 282.