Enhanced biological phosphate removal by granular sludge in a ...

8 downloads 0 Views 212KB Size Report
Key words: enhanced biological phosphate removal, granular sludge, morphology, sequencing batch reactor, sludge volume index. Abstract. A laboratory-scale ...
Biotechnology Letters 25: 687–693, 2003. © 2003 Kluwer Academic Publishers. Printed in the Netherlands.

687

Enhanced biological phosphate removal by granular sludge in a sequencing batch reactor Ebru Dulekgurgen∗ , Suleyman Ovez, Nazik Artan & Derin Orhon Environmental Engineering Department, Faculty of Civil Engineering, Istanbul Technical University, Maslak 34469, Istanbul, Turkey ∗ Author for correspondence (Fax: +90 212 286 7913; E-mail: [email protected]) Received after revisions 4 March 2003; Accepted 5 March 2003

Key words: enhanced biological phosphate removal, granular sludge, morphology, sequencing batch reactor, sludge volume index Abstract A laboratory-scale sequencing batch reactor was started-up with flocculated biomass and operated primarily for enhanced biological phosphate removal. Ten weeks after the start-up, gradual formation of granular sludge was observed. The compact biomass structure allowed halving the settling time, the initial reactor volume, and doubling the influent COD concentration. Continued operation confirmed the possibility of maintaining a stable granular biomass with a sludge volume index less than 40 ml g−1 , while securing a removal efficiency of 95% for carbon, 99.6% for phosphate, and 71% for nitrogen. Microscopic observations revealed a morphological diversity.

Introduction Granulation of biomass, resulting in superior supernatant-biomass separation, high mixed liquor suspended solid (MLSS) concentrations, and ability to meet high loading rates, has been observed in many anaerobic processes, though the mechanism of this phenomenon remains to be obscure (Morgenroth et al. 1997, Beun et al. 1999, Ahn et al. 2000). In addition, some recent observations demonstrated that granulation is not a characteristic, exclusively restricted to anaerobic systems. Researchers reported aerobic and anoxic-aerobic granulation in sequencing batch reactors operated for carbon and/or nitrogen removal (Morgenroth et al. 1997, Beun 2001, Beun et al. 1999). In these studies, the operational parameters, suspected to favor granulation, were determined and the influence of settling time, shearing effect and substrate loading were investigated. Settling time was considered to be the main parameter for the selection of granules with high settling velocities and washout of flocs. Settling velocity values from 0.67 to 1.8 cm s−1 were reported for rapidly-settling granular sludge (Beun et al. 1999, Etterer & Wilderer 2001).

Information on aerobic, as well as anaerobic granulation continues to accumulate yet the use of granular sludge in enhanced biological phosphate removal (EBPR) systems has been reported only in a few studies (Zhu et al. 2002, Dulekgurgen et al. 2003). The prime objective of this study is to report biomass granulation in a laboratory-scale sequencing batch reactor (SBR) and to evaluate the EBPR efficiency of the granular sludge. For this purpose, changes in biomass characteristics and structure were monitored in parallel with the EBPR performance, during the first phase of the study. Prior to the second phase, some operational parameters, being settling time, initial reactor volume, hydraulic retention time, aeration capacity, influent COD concentration, were altered and the system was monitored for its EBPR performance. The ability to maintain a stable granular biomass was also investigated. Microscopic examinations were carried out to determine some microbiological features of the EBPR biomass.

688 Table 1. Influent composition, operational hydraulic conditions, and biomass data for the cycle-evaluation (I) at week 10a .

Table 2. Influent composition, operational hydraulic conditions, and biomass data for the cycle-evaluation (IV) at week 37a .

Influentb

Influentb

Operationc

COD (mg l−1 ) 244e V0 (l) 2.9 TP (mg l−1 ) 20 VF (l) 4.6 0.63 TKN (mg l−1 ) 34 V0 /VF SRT (d) 8 Ts (min) 30

Biomassd MLSS (mg l−1 ) MLVSS (mg l−1 ) MLVSS/MLSS SVI (ml g−1 ) %TP (w/w)

1672 1144 68% 114 20.1

a Deteriorated EBPR: P-removal efficiency of the flocculated bio-

mass was 40%. b TP: total phosphate (mg PO -P l−1 ), TKN: total Kjeldahl-N. 4 c V : initial volume, V : filling volume, T : settling time. s F 0 d %TP: total phosphate content of the biomass calculated as [(mg PO4 -P l−1 )mixedliquor – (mg PO4 -P l−1 )effluent ]/[mg MLVSS l−1 ]. e Theoretical COD of the influent was 300 mg l−1 .

Fig. 1. Operational modes of the reactor. Anaerobic mixing (M) lasted for 2 h. The reactor received influent for 1 h (F) at the beginning of anaerobiosis. After 3 h of aeration (A), the biomass was let settle for 30 min (S) and then the supernatant was discarded (D).

Materials and methods

Operationc

COD (mg l−1 ) 544e V0 (l) 1.8 TP (mg l−1 ) 20.8 VF (l) 2.1 0.86 TKN (mg l−1 ) 40 V0 /VF SRT (d) 10 Ts (min) 15

Biomassd MLSS (mg l−1 ) 4420 MLVSS (mg l−1 ) 3132 MLVSS/MLSS 71% 45 SVI (ml g−1 ) %TP 13.1

a Complete-EBPR: P-removal efficiency of the granular biomass

was 99.6%. b,c,d Same as those given in Table 1. e Theoretical COD of the influent was 640 mg l−1 .

plant. Average influent composition and operational hydraulic conditions during the first phase of the study are given below in Table 1. The SBR was fed with a synthetic wastewater, containing acetate as the sole carbon source. pH control was provided at 6.5–7.5. Prior to the second phase, initial and filling volumes were decreased to 1.8 and 2.1 l, respectively (week 20). Settling time was also decreased to 15 min (week 22) and the SRT was adjusted to 10 d. At the second phase (week 25), influent composition was the same, except that the amount of acetate was increased to maintain a theoretical of 640 mg COD l−1 (see Table 2).

Operation of the reactor The operational modes of the reactor are schematically described in Figure 1. The SBR was inoculated with flocculent activated sludge, taken from the aeration tank of a conventional wastewater treatment

Fig. 2. Concentration profiles of ortho-phosphate (PO4 -P; ()], soluble COD (), and oxidized nitrogen [NOX -N; ()] determined during cycle-evaluation (I) at week 10. Deteriorated EBPR: P-removal efficiency of the flocculated biomass was 40%.

Sampling and conventional parameters The performance of the reactor was monitored by daily measurements of COD, ortho-phosphate (PO4 P), and oxidized nitrogen (NOX -N). Analyses were performed according to the Standard Methods for the Examination of Water and Wastewater by APHA, except that COD measurements were carried out as described in the International Standard ISO 6060 (International Organization for Standardization 1986). Biomass characteristics were investigated via measuring sludge volume index (SVI), mixed liquor suspended solids (MLSS), and mixed liquor volatile suspended solids (MLVSS). Total phosphate content (%TP; w/w) of the biomass was also calculated. Bright-field microscopy was applied to evaluate the macro-structure of the EBPR biomass. To assess biomass characteristics with respect to the microscopic features of the EBPR phenomenon, staining procedures for the visualization of intracellular polyphosphate (poly-P) granules and lipophilic inclusions, mainly being poly-hydroxybutyrate (PHB), were performed according to Jenkins et al. (1993).

689 Results and discussion Start-up and normal operation After six weeks of a start-up period, the flocculent biomass used as the inoculum, acclimated to the operational conditions. SVI decreased below 100 ml g−1 , MLSS was 1500 mg l−1 , and MLVSS/MLSS was 73% (w/w). Average of 50, 64, and 3 mg PO4 -P l−1 , measured by the end of the filling, mixing, and aeration periods, respectively, suggested that the biomass started to perform EBPR. EBPR capacity continued to increase at the following two weeks, resulting in less than 0.5 mg PO4 -P l−1 in the effluent. Average SVI and MLSS were 70 ml g−1 and 2100 mg l−1 , respectively. Average %TP content of the sludge was 19.2% (w/w), leading to 60% MLVSS/MLSS (w/w). 10–16 mg NOX -N l−1 was measured at the effluent. This NOX -N was consumed totally during the first 20– 30 min of the subsequent cycle. Average of 88% COD removal efficiency was sustained during this period.

Meanwhile, oxygen utilization rate data were collected online via a Manotherm RA-1000 respirometer, recycling 850 ml mixed liquor, continuously. Recycling flow rate of the respirometer was high enough (397 ml min−1 ) to fracture all flocs. Further decrease in EBPR, enforced wasting 2/3 of the P-leaking biomass for future stable operation. Connecting the respirometer with a high recycling flow rate provided a major shearing effect, which disintegrated the flocs and subsequently, initiated granulation. Transient state with poor EBPR During the transient state of three weeks with poor EBPR, aeration was increased and effluent was discarded manually after a settling period of 15 min (for 10 d). After this period, SVI started to decrease, down to 30 ml g−1 , and the structure of the activated sludge flocs started to change from fine particles to small granules. Meanwhile, MLSS started to increase (avg. 2000 mg l−1 ) and %TP content of the sludge never exceeded 10% (w/w).

Foaming and deterioration in EBPR EBPR recovery and apparent granulation Although none of the operational conditions were altered at the following week, a significant foaming was observed during aeration, settling properties deteriorated, and SVI increased to 125 ml g−1 . EBPR capacity of the sludge decreased (11 mg PO4 -P l−1 in the effluent), despite the apparent P-release at the anaerobic phase (73 mg PO4 -P l−1 by the end of mixing). Similar system deteriorations and sudden Pleakage incidents were observed occasionally during the following 42 weeks (data not presented here). Such declines in EBPR capacity occurred right after a complete-EBPR performance, remaining stable for 4 to 12 weeks (see Figure 3, weeks 25 and 29). An additional observation was that when the influent COD concentration was increased from a theoretical of 300 mg l−1 (weeks 9–25) to 640 mg l−1 (after week 25), the biomass responded much rapidly to system deteriorations and the EBPR performance, as well as settling properties, improved in less than 5 d (see Figure 3). Cycle-evaluation and respirometric analyses A series of analyses were carried out at week 10, to characterize the biomass behavior during the subsequent periods of one cycle (Table 1, Figure 2).

Through the end of the transient period, the granules grew in size to a maximum diameter of 2.5–3 mm and dominated the biomass. Settling experiments revealed that 81% (v/v) of the sludge settled in 30 s. The average settling velocity of the sludge-bulk liquid interface was 1.2 cm s−1 , a value comparable to the ones given for granular sludge in the literature (Morgenroth et al. 1997, Beun 2001, Beun et al. 1999). Aside from the granules, most of the flocs settled in 10 min, though there were still pin-point flocs by the end of 30 min. Results of the bright-field microscopy revealed a rich activated sludge population with different types of organisms present (week 20). Granules appeared as distinct, dark particles surrounded by few Zooglea and fixed protozoa (see Figure 5, Panel K), like Vorticella campanula (adjacent to the flocs) and Carshesium spp. (on the granules). Almost no filaments were observed. Staining procedures (Jenkins et al. 1993) for intra-cellular poly-P and PHB visualization were also carried out. The images captured after staining for intracellular inclusions are given in Figure 5, panels A to D. Optimization of operational conditions Compact structure of the settled granular biomass enabled decreasing initial reactor volume to 1.8 l (week

690

Fig. 3. Performance and operational layout of the reactor. PO4 -Pi : influent ortho-phosphate concentration (theoretical), CODi : influent COD (theoretical), %TP: total phosphate content of the biomass, TS : settling time, V0 : initial reactor volume.

Fig. 4. Concentration profiles of ortho-phosphate [PO4 -P; ()], soluble COD (), and oxidized nitrogen [NOX -N; ()] during cycle-evaluation (IV) at week 37. Ammonia concentration in the effluent was also determined [NH4 -N; ()]. Complete-EBPR: P-removal efficiency of the granular biomass was 99.6%.

20). Consequently, the amount of NOX -N, being recycled to the anaerobic period and consuming carbon

source, decreased and the amount of acetate available for P-storage increased. As a result, partial-EBPR started to recover and the effluent PO4 -P decreased down to 1 mg l−1 in two weeks (Figure 3). At week 22, settling time was decreased (from 30 to 15 min), both for taking advantage of working with a rapidly-settling granular sludge with low SVI, and for further enhancing granulation as suggested in the literature (Morgenroth et al. 1997, Beun, 2001, Beun et al. 1999). Meanwhile, it became possible to work with increased MLVSS concentrations. Thus, influent COD value was doubled and feeding rate was halved to promote this increase in MLVSS, while keeping the organic loading constant (Table 2, Figure 3). Resulting hydraulic retention time was 11.2 h. Complete-EBPR and stable granular sludge Monitoring the performance of the SBR operated under these hydraulic and operational conditions demonstrated that MLVSS increased gradually to 3100– 3700 mg l−1 , and a complete-EBPR performance,

Fig. 5. Bright-field micrographs of the biomass samples. Panels A–D (week 20) and E–H (week 37): all with 1000× magnification. Panels A and C (week 20); E and G (week 37): samples were collected by the end of the anaerobic period and treated with Neisser-stain for poly-P visualization [dark blue-violet cells: poly-P(+), brown cells: poly-P(−)]. Panels B and D (week 20); F and H (week 37): Sudan Black B was used to stain PHB-inclusions in samples collected by the end of the aerobic period [blue–black cells: PHB(+), pink cells: PHB(−)]. Panel I: a typical mature granule (40×). Panel J: aggregating young granules and flocs (40×). Panel K: flocs hosting a stalked-ciliate colony (100×). Poly-phosphate accumulating organisms (PAO), tetrad-forming organisms (TFO), and coccoid-clusters (CC) are also marked on the micrographs.

691

692 resulting in less than 0.1 mg PO4 -P l−1 in the effluent, was achieved (Figure 3). SVI of the biomass was not more than 45 ml g−1 . Granules were always present throughout the study (Figure 5; panels I and J). As mentioned above, despite the occurrence of sudden P-leakage incidents at weeks 25 and 29, the granular biomass returned to the complete-EBPR mode in a couple of days (Figure 3). Another set of experiments was conducted at week 37 (Table 2), to characterize the behavior of the stable granular biomass during one cycle (see Figure 4). As can be seen from Figure 4, the COD and PO4 -P profiles demonstrated trends typical for EBPR. The PO4 -P value by the end of the anaerobic period was as high as 136 mg l−1 and almost all of this was consumed during the first 90 min of the aerobic period. Nitrification occurred to some extend, leading to 5 mg NH4 -N l−1 in the effluent. All of the NOX -N was removed within the first 10 min of the subsequent cycle during which some of the influent COD was consumed by the denitrifiers (Figure 4). SVI was 45 ml g−1 , indicating a good settling sludge, and the %TP content of the biomass was 13.1% (w/w) (Table 2). Results of the microscopic examinations at week 37 are presented in Figure 5; panels E to H. Poly-P and PHB-staining experiments revealed that the enriched biomass of the reactor, performing complete-EBPR and hosting aggregates of microorganisms with different shapes, was morphologically quite diverse. The rod-shaped cells were abundant in the biomass, though the presence of tetra- and/or sarcina-like cells, as well as dense coccoid-clusters, was noticeable. The rod-shaped cells were definitely poly-P negative (Figure 5, panel E) and mostly PHB positive (panel F) by the end of the anaerobic period. By the end of the aerobic period, these rods were not totally, but moderately Neisser(+), suggesting poly-P storage under the maximum capacity (Figure 5, panel G). Accordingly, they seemed to consume their PHB pools aerobically, to a limited extend (Figure 5, panel H). Since both the morphology and the response to the poly-P and PHB-staining reactions of these rods fit to the proposed characteristics of phosphateaccumulating organisms (PAO) given in the literature (Mino et al. 1998, Bond et al. 1999), they were concluded to be the conventional rod-shaped PAO present in the system. Interestingly, the coccoid-clusters and tetra/sarcinalike cells were strongly Neisser(+) only on their cell walls and did not posses Neisser(+) intracellular polyP granules, available for EBPR metabolism. Thus, it

was not possible to regard these Neisser(+) results (Figure 5, panels E and G) as an indication of the ability to accumulate and consume poly-P for EBPR, and to consider these tetrads and coccoid-clusters as PAO. Mino et al. (1998) also reported the observation of sarcina-like cells in an EBPR system. These cells stained Neisser(+) only on their cell walls and were easily distinguished from PAO containing strongly Neisser(+) granules inside their cells (Mino et al. 1998, and the references therein). The shape of the coccoid-clusters and especially tetra/sarcina-like cells (Figure 5, panels E, F, and G) confirmed the recently proposed morphology of glycogen-accumulating organisms (Liu et al. 1996, Mino et al. 1998, Bond et al. 1999). Yet, they did not give positive response to PHB-staining (end of the anaerobic period; panel F). Therefore, the tetrads were concluded to resemble glycogen-accumulating organisms (GAO) morphologically, but not functionally with respect to anaerobic PHB storage. Conclusively, the tetra/sarcina-like microorganisms observed in this study were referred as Tetrad Forming Organisms (after Tsai & Liu 2001), a name addressing morphology, rather than GAO, a name addressing function. Further interpretation of the morphological and functional characteristics of the observed microbial aggregates were detailed elsewhere (Dulekgurgen et al. 2003).

Conclusions This study substantiated the possibility of attaining granular sludge in an anaerobic/aerobic sequencing batch reactor with a complete-EBPR performance. This remark disproves the general belief that granulation is exclusively restricted to anaerobic processes and contributes to the emerging observations on granulation in EBPR systems. Operational flexibility of the SBR (ability to decrease settling time, initial reactor volume, etc.) played a key role to promote formation and maintenance of a compact granular biomass. Consequently, it became possible to work with increased MLVSS concentrations, and thus within a much smaller reactor volume, compared to conventional flocculated activated sludge systems. Microscopic observations, after staining the samples for poly-P and PHB, revealed that the EBPR biomass consisted of a microbial community diverse in terms of morphology and physiology. The rodshaped bacteria, concluded to be the PAO present in

693 the system, co-existed with tetra/sarcina-like cells and coccoid clusters, which resembled the GAO morphologically but not functionally with respect to anaerobic PHB storage. Future work should target the evaluation of the spatial distribution of the members of the EBPR microbial consortia within the granules. Assessment of the biomass with respect to the microscopic features of the EBPR phenomenon and phylogenetic identification of the microbial populations should be performed simultaneously to juxtapose the function and identity of the microorganisms observed in EBPR systems.

Acknowledgements This study was executed as a part of the research activities of the Environmental Biotechnology Center of the Scientific and Technical Research Council of Turkey. It was also supported by the Research and Development Fund of Istanbul Technical University.

References Ahn Y, Lee YJ, Kim HS, Park S (2000) Monitoring of specific activity of granular sludge by confocal laser scanning microscopy during start-up of thermophilic upflow anaerobic sludge blanket reactor. Biotechnol. Lett. 22: 1591–1596. Beun JJ (2001) PHB metabolism and N-removal in sequencing batch granular sludge reactors. Ph.D. Dissertation. Delft, The Netherlands: Technical University of Delft. ISBN 90-6464-2575.

Beun JJ, Hendriks A, van Loosdrecht MCM, Morgenroth E, Wilderer PA, Heijnen JJ (1999) Aerobic granulation in a sequencing batch reactor. Water Res. 33: 2283–2290. Bond PL, Erhart R, Wagner M, Keller J, Blackall L (1999) Identification of some of the major groups of bacteria in efficient and nonefficient biological phosphorus removal activated sludge systems. Appl. Environ. Microbiol. 65: 4077–4084. Dulekgurgen E, Yesiladali K, Ovez S, Tamerler C, Artan N, Orhon D (2003) Conventional morphological and functional evaluation of the microbial populations in a sequencing batch reactor performing EBPR. J. Environ. Sci. Health, Part A, A38(8) (in press). Etterer T, Wilderer PA (2001) Generation and properties of aerobic granular sludge. Water Sci. Tech. 43: 19–26. International Organization for Standardization (1986) In: Technical Committee ISO/TC 147, eds. International Standard ISO 6060: Water Quality – Determination of the Chemical Oxygen Demand, Reference No. ISO 6060-1986(E). Switzerland: ISO. Jenkins D, Richard MG, Daigger GT (1993) Manual on the Causes and Control of Activated Sludge Bulking and Foaming, 2nd edn. Michigan, USA: Lewis Publishers Inc. ISBN 0-87371-873-9. Liu WT, Mino T, Matsuo T, Nakamura K (1996) Glycogen accumulating population and its anaerobic substrate uptake in anaerobicaerobic activated sludge without biological phosphorus removal. Water Res. 30: 75–82. Mino T, van Loosdrecht MCM, Heijnen JJ (1998) Microbiology and biochemistry of the enhanced biological phosphate removal process. Water Res. 32: 3193–3207. Morgenroth E, Sherden T, van Loosdrecht MCM, Heijnen JJ, Wilderer PA (1997) Rapid communication: aerobic granulation in a sequencing batch reactor. Water Res. 31: 3191–3194. Tsai CS, Liu WT (2002) Phylogenetic and physiological diversity of tetrad-forming organisms in deteriorated biological phosphorus removal systems. Water Sci. Tech. 46: 179–184. Zhu J, Etterer T, Wilderer PA (2002) Structure and functions of granular activated sludge in SBR. In: Proceedings of the Enviro 2002 IWA 3rd World Water Congress, April 7–12, Melbourne, Australia, Paper No. e20908a, p. 272. ISBN 0 908255 53 5.