Professional Documents
Culture Documents
Abstract
Ammonium nitrogen concentration in leachate from old Italian landfills ranges from 0.5 to as high as 3 g L−1 . In this paper biological nitrogen
removal from leachate has been achieved by partial nitrification to nitrite in a pure-oxygen membrane bioreactor (PO-MBR) and by subsequent
denitrification in a moving-bed biofilm reactor (MBBR). When ammonium is biologically oxidized to nitrite, only 75% of the oxygen required
for full nitrification is needed. Moreover, denitrification can be performed by saving 30–40% of the carbon required. The process was carried out
by an MBR oxidation tank of 500 L equipped with an UF ceramic membrane followed by a 540-L post-denitrification tank filled with moving
plastic support media. The best operational conditions to achieve partial nitrification were analyzed. TKN loading rate was variable from 50 to
120 g TKN (kg TSS day)−1 with an influent ammonia concentration between 1000 and 1500 mg L−1 . When DO concentration in the MBR was
kept in the range 0.2–0.5 mg L−1 , 90% oxidation of ammonia to nitrite was achieved, with stable inhibition of nitrite-oxidizing bacteria even at
sludge retention time higher than 45 days.
© 2006 Elsevier B.V. All rights reserved.
Keywords: Pure oxygen membrane bioreactor (PO-MBR); Landfill leachate treatment; Carbon and nitrogen removal; Partial nitrification; Stable inhibition of
nitrite-oxidizing bacteria
0376-7388/$ – see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.memsci.2006.09.044
R. Canziani et al. / Journal of Membrane Science 286 (2006) 202–212 203
Table 1
Kinetic coefficients at T = 20 ◦ C
Coefficient Definition NH4 -oxididation NO2 -oxidation
Table 2
BOD5 , ammonium nitrogen (N-NH4 + ) and COD concentrations in raw leachate
COD (mg L−1 ) BOD5 (mg L−1 ) N-NH4 + (mg L−1 ) BOD5 /COD (◦ ) N-NH4 + /COD (◦ )
Number of data = 70 (from January 2001 to October 2003); (◦ ) average of the single ratios.
nium to nitrite as a pre-treatment of an Anammox process The pilot plant included a 500-L stainless steel tank for bio-
[29]. logical oxidation.
The aim of this research was to achieve complete conversion The membrane module was a tubular zirconium ceramic
of ammonium to nitrite in an MBR with high sludge retention MEMBRALOX® model P19-40 (Table 3).
time (SRT more than 45 days) and by controlling T, DO and pH. The membrane was operated at 1.5 × 105 Pa (1.5 bar). The
chemical cleaning procedure, which was performed only twice
2. Material and methods during the experimental period (days 306 and 406), was per-
formed according to the instructions provided by the producer.
2.1. Leachate Commercial membrane cleaners (Ultrasil, both acid and alka-
line solutions) were used. In the first washing the sequence was
The leachate used in the experimentation came from an old
as follows: (i) alkaline solution (NaOH, 1.5%), (ii) acid solution
municipal solid waste (MSW) sanitary landfill located in North-
(Ultrasil 0.5%) and (iii) NaClO solution (1.5%). Between each
ern Italy. Table 2 shows BOD5 , ammonium nitrogen (N-NH4 + )
washing step, the membrane was rinsed with tap water. In the
and COD concentrations in the raw leachate (in mg L−1 ). Con-
second washing, the sequence was different: at first NaClO (5%)
centrations are highly variable even eight years after closure.
was used, followed by an acid wash at pH 2 and a final alkaline
This variability, as well as the high variations in BOD5 /COD and
wash at pH 12.3. Again, tap water was used for rinsing after
N-NH4 + /COD ratios, is not due to biochemical processes inside
each washing step.
the mass of wastes, but to the collection systems of leachate:
Mixed liquor was pumped at a pressure of 4 × 105 Pa (4 bar)
in fact subsurface drainage is active only during wet weather
into the external ceramic membrane module at a flow rate of
periods, and deep wells are active during dry periods.
4.25 m3 h−1 , which corresponded to a cross-flow velocity of
The biodegradable fraction in the leachate fed to the pilot
4.94 m s−1 . Mixing was ensured by recirculating the retentate
plant was even lower, as it came from an anaerobic fixed-bed
biofilter, used as a pre-treatment prior to discharge into a munic-
ipal wastewater treatment plant (MWWTP). Table 3
Membrane technical data
2.2. Pilot plant
Parameter Value Unit
The flow-sheet of the pilot plant is reported in Fig. 1. Leachate Number of tubes 19 –
was stocked in a 1000 L tank and kept mixed by a recirculat- Tube diameter 4 mm
ing pump (Q = 1.8–4.8 m3 h−1 , Pmax = 4 × 105 Pa (4 bar)) and a Length 1020 mm
hydraulic ejector. The recirculating pump was also used to feed Pore size 50 nm
Total cross-sectional area 2.4 × 10−4 m2
the pilot plant, by opening an actuated solenoid valve for a fixed Filtering area 0.24 m2
time.
R. Canziani et al. / Journal of Membrane Science 286 (2006) 202–212 205
Table 4
Average characteristics of the influent during the experimental runs A, B and C (data in mg L−1 )
Run (days) DO TSS Total solids (TS) Total volatile solids (TVS) CODin N NH4, in + TKNin
flow from the ceramic membrane into the bioreactor. Pure liq- concentration in the leachate was near 200 mg L−1 and was mea-
uid oxygen was kept in a pressurized tank and was fed to the sured only occasionally. Total and volatile solids were measured
oxidation tank through a solenoid valve which was automatically regularly; inorganic solids were over 1 g L−1 , mostly dissolved.
actuated and controlled by an oxygen probe that was immersed Average phosphorus concentration in the leachate was always
in the oxidation tank; pH control in the oxidation-nitrification lower than required as nutrient for biomass growth. There-
tank was performed by dosing a sodium hydroxide solution. fore, phosphoric acid was continuously added to the influent, in
Before process start-up, it was observed that temperature order to maintain a concentration of phosphorus in the permeate
increased in the MBR tank, due to the energy dissipated by >5 mg P L−1 . The ratio BOD5 /COD in the influent was around
the recirculating pump; therefore a cooling hydraulic circuit 1:3. pH adjustment in the MBR by addition of a 0.2 M sodium
was installed, to maintain a constant temperature in the process hydroxide solution was required since alkalinity of the leachate
reactor. was not sufficient to buffer the acidity produced by nitrification.
The permeate came to an overflow tank and then was pumped Temperature of the mixed liquor was always kept between
by a peristaltic pump to the post-denitrification tank (540 L 30 and 37 ◦ C and pH between 7.8 and 8.2 in order to enhance
volume). Filtrate in excess was recirculated into the oxidation the growth of ammonium oxidizing bacteria.
tank. As much as 37.5% of the denitrification tank volume was During the first experimental run (run A, days 328–396), dis-
filled with floating plastic media that were used as supports for solved oxygen was kept low (0.2–0.5 mg L−1 ) and free ammonia
biofilm growth (ANOX KALDNES MBBR® carrier K1). Mix- was kept high (about 40 mg NH3 L−1 ) in order to study the
ing was ensured by a mechanical mixer. The external carbon inhibitory effect on nitrite-oxidizing biomass.
source for post-denitrification was provided by dosing a hydro- After day 396, run B, the oxygen concentration was increased
alcoholic mixture named BITOLEA® Liquifeed 25, a mixture to 1.5 mg L−1 and above in order to test the inhibitory effect on
of alcohols with 1–4 carbon atoms (methanol to butanol), with the biomass caused only by high free ammonia concentration.
methanol less than 20% and COD content of 415,000 mg O2 L−1 At day 406 an accidental biomass chlorination during mem-
(7.2–12 Lmixture added /kg Ntreated ). brane cleaning obliged to decrease for a few days the TKN sludge
loading rate to recover bacterial growth (run B1 before and run
2.3. Analytical methods B2 after chlorination).
After day 440, during the last experimental run (run C), DO
Determinations of TKN, nitrate, nitrite, COD, total sus- concentration was decreased again between 0.2 and 0.5 mg L−1
pended solids (TSS), total solids (TS) were performed according and free ammonia concentration was maintained at about
to Standard Methods, 20th Ed. [30]. Phosphate was determined 2–5 mg NH3 L−1 .
by spectrophotometry using Dr. Lange procedure and kit. Bac-
teria quantification and identification were performed by plate 3. Results and discussion
counts according to the Italian Standard Methods for Sludge [31]
and by optical microscopic observation [32,33]. 3.1. Start-up
2.4. Influent data and operating experimental conditions The process of partial nitrification was started up after 328
days of conventional full nitrification (ammonium to nitrate)
Influent data and operating parameters during the experimen- [34]. At the start-up of the new process, the MBR sludge con-
tal programme are listed in Tables 4 and 5. Suspended solids centration was as high as 10 g TSS L−1 .
Table 5
Average operating parameters
Run (days) Flow rate O2 Sludge loading rate MLTSS T (◦ C) pH Free NH3
(L h−1 ) (mg L−1 ) (mg L−1 ) (mg L−1 )
kg COD (kg TSS−1 day−1 ) g TKN (kg TSS−1 day−1 )
Fig. 2. COD specific loading rate (F/M, left axis) and removal efficiency (right
axis).
Fig. 5. (a) NO3 -N to NO2 -N ratio in the permeate. Dotted lines: changes in
dissolved oxygen control; grey area: accidental chlorination; (b) free ammo-
nia (right axis) and ammonium nitrogen (left axis) in the MBR after lowering
the dissolved oxygen to below 0.5 mg L−1 (day 440); (c) nitrate nitrogen vs.
dissolved oxygen in the MBR.
3.4.1. Heterotrophs
The decay constant of the heterotrophic bacteria (kd ) was Fig. 6. Regression of (TSS/TSScumulated ) vs. time during: (a) periods A, (b)
period B1 , (c1 ) period B2 , (c2 ) final period B2 ; during B2 , the effect of the
estimated by measuring the decrease of suspended solids in the chlorination is evident.
MBR after stopping the feed for 8 days. It resulted as low as
0.03 day−1 at a reactor temperature of 25 ◦ C and pH 7.9. In spite
suspended solids grown in the reactor and the corresponding
of applying the lowest correction coefficient for temperature
COD removed (Fig. 6). The actual yield coefficient (Y) has been
reported by [36], i.e. θ T = 1.03, the value of the decay rate at
calculated by taking into account the decay rate (kd ) as follows:
20 ◦ C (0.026 day−1 ) resulted well below the lower limit of the
range reported in the literature (0.06–0.2 day−1 , [36]).
The observed yield coefficient of heterotrophic bacteria (Yobs ) kd
Yobs · ν = Y · ν − kd or Y = Yobs +
has been calculated as the slope of the regression between the ν
208 R. Canziani et al. / Journal of Membrane Science 286 (2006) 202–212
Table 6
Observed (Yobs ) and net (Y) yield coefficients during the three experimental phases
Period (days) Yobs (g TSS T (◦ C) kd,T COD loading rate (kg COD COD removal (kg COD kg Y (g TSS g
g COD−1 ) (day−1 ) kg TSS−1 day−1 ) efficiency TSS−1 day−1 ) COD−1 )
Table 7
Theoretical and experimental growth rates of heterotrophic bacteria during the
three experimental periods
Period (days) μobs (day−1 ) μ (day−1 ) μcalc (day−1 )
Fig. 7. Regression of (TSS/TSScumulated ) vs. time during period C (low oxygen apply to period B, which was affected by an accidental chlori-
as in period A).
nation. It can be observed that low-oxygen conditions did not
affect negatively the growth of heterotrophs.
where “ν” is the daily COD removal rate per unit mass of sus-
pended solids (kg CODrem kg TSS−1 day−1 ). The observed and 3.4.2. Autotrophs
actual yield coefficients calculated for the three experimental Theoretical growth rates (μ) for autotrophic organisms have
periods are reported in Table 6. The VSS/TSS ratio of the mixed been estimated from Eqs. (1) to (4) by adopting a μMAX value of
liquor was fairly constant during the experimental runs (run A: 0.77 day−1 for ammonium oxidizers and 1.08 day−1 for nitrite
0.72 ± 0.05; run B: 0.68 ± 0.02; run C: 0.73 ± 0.05). oxidizers [21]. Results of these calculations are summarized in
The observed growth coefficient (μobs ) has been calculated Table 8 and show that the growth rate of ammonium oxidizers
during the three periods as the slope of the regression of is always higher than that of nitrite oxidizers, confirming the
(TSS/TSScumulated ) versus time (Figs. 6 and 7). experimental observations (e.g.: NO3 -N/NO2 -N ratio, Fig. 5a).
The observed growth rate includes the decay rate: In particular, low oxygen concentration increases the difference
μobs = μ − kd and, therefore: μ = μobs + kd . between the two growth rates, as it has also been observed by
The growth rate can also be calculated (μcalc ) from the actual others [29].
removal rate (ν) and growth yield (Y) as μcalc = Y · ν. In Table 7 As observed by Wyffels et al. [29], nitrate can be kept at low
values of μobs , μ and μcalc are reported. concentration in the reactor by lowering the dissolved oxygen
Theoretical and calculated estimations of growth rates agree concentration. Also, dissolved oxygen was shown to be more
very well either in period A or in period C, thus confirming the important than temperature and free ammonia as nitrate con-
consistency of the experimental trials. As the experimental peri- troller. This assumption is confirmed by the results obtained in
ods were short if compared with SRT, it is not possible to state the present work: in fact, Wyffels et al. performed partial nitri-
that steady-state conditions were achieved. However, removal tation as the preceding step of an ANAMMOX process and free
efficiency was stable and this allows to state that stable process ammonia concentration in their reactor was much higher than in
conditions were achieved (Fig. 4, period A). The same does not the experiments presented here.
Table 8
Theoretical growth rates (μ) for autotrophic organisms
Period (days) No. of data Ammonium oxidizers μNH (day−1 ) No. of data Nitrite oxidizers μNO (day−1 )
Fig. 9. (a) Left: Opercularia sp.; (b) Right: H. Hydrossys bridging between sludge flocs (1000×).
3.5. Microscopy observation and bacterial counts The results of the counts are shown in Table 9. What mat-
ters is that the total bacterial count did not show any substantial
Optical microscopic observation showed that sludge aggre- difference among the different experimental periods. On the con-
gates were generally smaller than 20 m with few zoogleal trary, the ratio of ammonium oxidizers to nitrite oxidizers passed
flocs around 50–100 m (Fig. 8). Sessile ciliate Protozoa (Oper- from around 1:6 (full nitrification) to around 2000–4000:1 after
cularia sp.) were always present at a concentration around switching to the partial nitrification mode of operation.
4 × 104 L−1 (Fig. 9a). These organisms are very resistant even The ratio of the two populations was estimated by MPN
in industrial wastewater treatment plants because of their strong (Table 9).
resistance to sub-optimal environmental conditions. Also, the ratio of the two populations has been calculated as
After the operation was switched from full to partial nitri- the ratio between their net growth, by assuming that the system
fication, the abundance of filamentous organisms, mainly H. was at steady state:
hydrossis, increased. Since sludge separation was operated by a
XNH YAOB (NH4 -N/(μAOB − kD ))
membrane, their presence did not affect the performance of the =
process. XNO YNOB (NO3 -N/(μNOB − kD ))
The evolution of the main bacterial populations in the MBR where μNOB were calculated according to formula (2b).
has been monitored with classical methods (CFU: colony form- Results are shown in Table 10 and Fig. 10:
ing units; MPN: most probable number) providing a result in The great discrepancy between the two series of results can be
terms of single bacterial units even if they could represent bac- explained by the error of the MPN method. This method, in fact,
terial aggregates, especially in the case of nitrifying organisms. can be affected by an error of one order of magnitude or more
As a matter of fact several clusters have been observed, as shown when applied to river water samples and even higher values can
in Fig. 8 above. be expected when this method is applied to samples of mixed
Table 9
Evolution of the bacterial counts over the experimental periods
Period Sample taken on day no. Total bacterial count (CFU mL−1 ) AOB (MPN mL−1 ) NOB (MPN mL−1 ) AOB/NOB
Legend: CFU: colony forming units; AOB: ammonium oxidizers; NOB: nitrite oxidizers.
210 R. Canziani et al. / Journal of Membrane Science 286 (2006) 202–212
Table 10
Calculated and measured ratios between AOB and NOB populations (KI = 20 mg L−1 )
Experimental period μAOB (day−1 ) μNOB (day−1 ) Ratio XAOB /XNOB Notes
Fig. 12. Specific flux related to total suspended solids (TSS) in the aerated tank.
Fig. 11. The regression with the highest R2 was found at an inhibition constant Fig. 13. Specific flux related to total suspended solids (TSS) in the aerated tank.
for free ammonia KI = 18 mg L−1 . Effect of chlorination after cleaning at day 406.
R. Canziani et al. / Journal of Membrane Science 286 (2006) 202–212 211
[11] J.J. Wu, C.-C. Wu, H.-W. Ma, C.-C. Chang, Treatment of landfill leachate ammonium oxidation (Anammox) in a pilot plant, J. Biotechnol. 99 (3)
by ozone-based advanced oxidation processes, Chemosphere 54 (7) (2004) (2002) 295–306.
997–1003. [26] G. Ruitz, D. Jeison, R. Chamy, Nitrification with high nitrite accumulation
[12] K.-U. Heyer, R. Stegmann, Leachate Management: Leachate Gen- for the treatment of wastewater with high ammonia concentration, Water
eration, Collection, Treatment and Costs, Ingenieurbüro für Abfall- Res. 37 (2003) 1371–1377.
wirtschaft, Prof. R. Stegmann und Partner, 2001, http://www.ifas- [27] H. Yoo, K. Ahn, H. Lee, K. Lee, Y. Kwak, K. Song, Nitrogen removal
hamburg.de/index engl.htm. from synthetic wastewater by simultaneus nitrification and denitrification
[13] G. Bressi, G. Favali, Use of MBR (Membrane BioReactor) in leachate (SND) via nitrite in an intermittently aerated reactor, Water Res. 33 (1)
treatment, in: Proceedings Sardinia 97, Sixth International Landfill Sym- (1999) 146–154.
posium, S. Margherita di Pula, Cagliari, Italy, 13–17 October, vol. II, 1997, [28] C. Hellinga, M.C.M. van Loosdrecht, J.J. Heijnen, The Sharon process for
pp. 261–274. nitrogen removal in ammonium-rich wastewater, in: Eleventh Forum for
[14] A.G. Vlissides, P.K. Karlis, G. Mahnken, Influence of various parameter on Applied Biotechnology, 1997, pp. 1743–1750.
the electrochemical treatment of landfill leachates, J. Appl. Electrochem. [29] S. Wyffels, S.W.H. Van Hulle, P. Boeckx, E.I.P. Volcke, O. Van Cleemput,
33 (2003) 155–159. P.A. Vanrolleghem, W. Verstraete, Modeling and simulation of oxygen-
[15] S.H. Lin, C.C. Chang, Treatment of landfill leachate by combined electro- limited partial nitritation in a membrane-assisted bioreactor, Biotechnol.
Fenton oxidation and sequencing batch reactor method, Water Res. 34 (17) Bioeng. 86 (5) (2004) 532–542.
(2000) 4243–4249. [30] Standard Methods for the Examination of Water and Wastewater, 20th ed.,
[16] W. Ahn, M. Kang, S. Yim, K. Choi, Advanced landfill leachate treat- 1998.
ment using an integrated membrane process, Desalination 149 (1–3) (2002) [31] IRSA CNR, metodi analitici per i fanghi, Quaderno IRSA no. 64, vol. 1,
109–114. 1983.
[17] I. Ozturk, M. Altinbas, I. Koyuncu, O. Arikan, C. Gomec-Yangin, [32] P. Madoni, Estimates of ciliated protozoa biomass in activated sludge and
Advanced physico-chemical treatment experiences on young municipal biofilm, Bioresour. Technol. 48 (1994) 245–249.
landfill leachates, Waste Manage. 23 (5) (2003) 441–446. [33] D. Jenkins, M.G. Richard, G.T. Daigger, Manual on Solving Activated
[18] J.M. Lema, R. Mendez, R. Blazquez, Characteristics of landfill leachates Sludge Bulking, Foaming, and Other Solids Separation Problems, third
and alternatives for their treatment: a review, Water Air Soil Pollut. 40 (3–4) ed., Lewis Publishers, London UK, 2003.
(1988) 223–250. [34] F. Malpei, R. Canziani, M. Garavaglia, S. Passarella, Leachate treatment
[19] J. Bohdziewicz, M. Bodzek, J. Gorska, Application of pressure-driven by a pilot scale ceramic membrane bioreactor, in: Proceedings of the
membrane techniques to biological treatment of landfill leachate, Process 2004 Water Institute of Southern Africa (WISA) Biennial Conference,
Biochem. 36 (2001) 641–646. 2–6 May, Cape Town, South Africa, 2004, ISBN 1-920-01728-3 (pro-
[20] D. Barnes, P.J. Bliss, Biological Control of Nitrogen in Wastewater Treat- duced by: Document Transformation Technologies, organised by Event
ment, E & FN Spon, London, England, 1983. Dynamics).
[21] U. Wiesmann, Biological nitrogen removal from wastewater, Adv. Bioch. [35] Y. Peng, X. Song, C. Peng, J. Li, Y. Chen, Biological nitrogen removal
Eng. 51 (1994) 113–154. in SBR bypassing nitrate generation accomplished by chlorination
[22] A.C. Anthonisen, R.C. Loehr, T.B.S. Prakasam, E.G. Srinath, Inhibition and aeration time control, Water Sci. Technol. 49 (5–6) (2004) 295–
of nitrification by ammonia and nitrous acid, J. Water Poll. Control 48 (5) 300.
(1976) 835–852. [36] Metcalf, Eddy, Wastewater Engineering Treatment and Reuse, fourth ed.,
[23] C. Hellinga, M.C.M. van Loosdrecht, J.J. Heijnen, The Sharon process for Mc Graw Hill, 2003.
nitrogen removal in ammonium rich waste water, in: Eleventh Forum for [37] C. Laspidou, B. Rittmann, A unified theory for extracellular polymeric
Applied Biotechnology, 1997, pp. 1743–1750. substances, soluble microbial products, and active and inert biomass, Water
[24] J.E. Alleman, Elevated nitrite occurrence in biological wastewater treat- Res. 36 (2002) 2711–2720.
ment systems, Water Sci. Technol. 17 (1984) 409–419. [38] Manem, Sanderson, Membrane Bioreactor in Water Treatment Membrane
[25] C. Fux, M. Boehler, F. Huber, I. Brunner, H. Siegrist, Biological treatment of Processes. AWWA. Lyonnaise de Eaux. Water Research Commission of
ammonium-rich wastewater by partial nitritation and subsequent anaerobic South Africa, McGraw-Hill, 1996 (Chapter 17).