- Email: [email protected]
Biomass growth and activity in a membrane bioreactor with complete sludge retention
ARTICLE IN PRESS
Water Research 38 (2004) 1799–1808
Biomass growth and activity in a membrane bioreactor with complete sludge retention Alfieri Pollice*, Giuseppe Laera, Massimo Blonda CNR Istituto di Ricerca Sulle Acque, Viale F. De Blasio, n. 5 (Zona Ind.), Bari 70123, Italy Received 31 July 2003; received in revised form 22 December 2003; accepted 5 January 2004
Abstract This work reports the main results of a bench scale membrane bioreactor operated for more than 100 days without sludge withdrawal and fed on real municipal wastewater. The experiments were oriented towards three main objectives. Firstly, the performance of the system was evaluated under two different volumetric loading rates (0.8 and 1 1.7 g COD Lreact. d 1). Secondly, biomass growth and accumulation of solids were assessed and a relationship between sludge concentration and volumetric loading rates was proposed. Thirdly, biomass activity was evaluated through respirometric tests, and endogenous and maximum respiration rates of heterotrophic and nitrifying bacteria were determined. The experimental campaign showed that these systems are easy to manage and very rapid to start-up. The SS concentrations under equilibrium conditions for both experimental periods were slightly lower than 10 times the volumetric loading rates, and the organic loading rates reached the same equilibrium value of 0.12 g COD g TSS 1 d 1. 1 Furthermore, under equilibrium conditions the system showed very limited sludge production (0.12 g VSS g CODrem ) and low biomass activity, although it readily responded to load variations. Further work is being carried out to evaluate the performance over the long term. r 2004 Elsevier Ltd. All rights reserved. Keywords: Membrane bioreactors; Complete sludge retention; Biomass activity; Municipal wastewater; Sludge yield
1. Introduction During the last decades, an increasing number of investigations have improved the knowledge of the membrane bioreactors and the interactions between biological wastewater treatment processes and membrane filtration. Research activities were mainly directed towards the optimization of process efficiency in terms of effluent quality, membrane types, plant configurations and operational practices [1,2]. Less attention has been devoted to the peculiarities of biological processes and microbial consortia developed through the application of membrane separation. Control of the bacterial environment through efficient solid/liquid separation and operation of the biological processes with high *Corresponding author. Tel.: +39-080-582-05-11; fax: +39080-531-33-65. E-mail address: [email protected] (A. Pollice).
biomass concentrations are likely to affect cell metabolism and limit the bacterial growth, and thus the sludge production [3–5]. By operating the processes at low organic loading rates (g COD g TSS 1 d 1) it is possible to divert the utilization of polluting compounds from biosynthesis to non-growth energy-demanding activities, favour the utilization of substrates for maintenance of bacterial vital functions, and limit the net growth [6,7]. This is the concept of maintenance energy, first introduced by Pirt in 1965, that can be considered as the amount of biochemical energy strictly necessary for endogenous respiration [8]. A direct consequence of the previous statements is that operating a membrane bioreactor without any sludge wastage should allow the system to reach a steady state at a given sludge concentration. Under these conditions, defined ‘‘complete sludge retention’’ by van Houten and Eikelboom, the biomass growth rate should balance the decay rate [7].
0043-1354/$ - see front matter r 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2004.01.016
ARTICLE IN PRESS A. Pollice et al. / Water Research 38 (2004) 1799–1808
1800
Early investigations on complete retention membrane bioreactors (crMBR) indicated that under constant F =M ratios the sludge concentration was almost constant [9,10]. A flattening of the sludge growth curve, suggesting equilibrium between endogenous respiration and new bacterial mass growth in MBRs with complete sludge retention was also observed [7,11,12]. Other researchers proposed the possibility of controlling the biomass concentration by regulating the influent organic load, also considering the substrate’s biodegradability [13–17]. The same authors suggested that operational strategies based on slow biomass turnover (i.e. limited sludge wastage) could be preferable to complete sludge retention when accumulation of inorganics such as heavy metals is likely to occur. Studies on the accumulation of inert material reported constant percentage of inorganics in the sludge, independent of its concentration [7]. Oxygen transfer limitations in MBRs operated with high biomass concentrations were reported by several authors [9,11,18]. The present work reports some results of an experimental campaign aimed at evaluating the efficiency of a laboratory scale crMBR fed on real municipal wastewater. The plant was operated for more than 100 days, during which the main operational parameters were monitored and controlled in order to let biological processes evolve towards equilibrium. The biomass was examined over time to evaluate the dependency of its growth patterns on process conditions. Moreover, the activity of heterotrophic and nitrifying organisms was regularly monitored by performing respirometric
tests, in order to investigate the variation of their viability.
2. Materials and methods 2.1. Experimental plant A scheme of the experimental set-up is reported in Fig. 1. The membrane bioreactor had 6 L operating volume and was mechanically stirred. A Zenon ZeeWeeds lab-scale hollow fibre membrane module (nominal surface 0.094 m2) was sank in the bioreactor and operated out-in. The permeate was extracted by imposing on the fibres a negative pressure, normally maintained well below the limiting value suggested by the manufacturer (0.7 bar). Operational cycles lasted 6 min and included extraction of the permeate (5.5 min) and backwash (0.5 min). Moreover, a constant air flow was pumped through the membrane module to prevent or reduce fouling and cake formation. The permeate flux was regulated depending on the influent COD concentration, in order to maintain a predetermined volumetric loading rate. The latter is defined as the amount of COD daily fed to the plant and 1 referred to the bioreactor’s volume (g COD Lreact. d 1). The system was continuously fed on pre-settled municipal wastewater screened at 1 mm, stored at 4 C and continuously mixed. The plant was started up without any biomass inoculum, and during the whole
AIR PUMP 2 pH-METER
AIR PUMP 1
MBR
MANOMETER CIP
OXYMETER
FEED TANK
TIMER
PROCESS PUMP LEVEL CONTROL
INFLUENT COD ANALYSIS
Fig. 1. Scheme of the experimental set-up.
ARTICLE IN PRESS A. Pollice et al. / Water Research 38 (2004) 1799–1808
experimental period no sludge was intentionally withdrawn from the plant except for measurements of suspended solids. Very limited amounts of sludge went lost when the module was removed from the reactor for cleaning and when the mixed liquor was temporarily sampled for respirometric determinations. These amounts were accounted for during the calculation of biomass production and the other sludge-related parameters. As a general rule of this investigation, biomass samples were always returned to the reactor after nondestructive determinations. A certain amount of biomass tended to stick to the parts of the reactor (walls, plastic pipes, etc.) close to the surface level. This fraction was daily removed and returned to the sludge bulk. The process was operated under continuous aeration in order to achieve biodegradation of the organic fractions and nitrification. Oxygen was provided through an air diffuser placed in the reactor. The air flowrate provided by this second air pump was regulated to maintain residual dissolved oxygen (DO) concentrations between 1 and 3 mg L 1. 2.2. Sampling and analyses The bench scale crMBR was continuously monitored for DO, temperature, pH, and transmembrane pressure (TMP). The influent municipal wastewater was sampled three times per week and analysed for total and volatile suspended solids (TSS and VSS), total COD, N-NH4, TKN, N-NO2, and N-NO3. The permeate was sampled daily and analysed for the same parameters. All analyses were performed according to standard methods [19]. The ion chromatographic method for the determination of nitrate and nitrite was modified adopting an UV detector (at 220 nm) instead of the conductivity detector. Biomass samples were withdrawn from the reactor twice per week and analysed for TSS and VSS according to standard methods [19]. One litre of the reactor’s content was withdrawn twice per week for respirometric tests and returned to the bioreactor afterwards. All respirometric tests were performed after aerating the sludge sample overnight
1801
in order to obtain endogenous respiration conditions. According to the definitions of the IAWQ Task Group on Respirometry, the respirometer adopted was a ‘‘static gas–static liquid’’ system (LSS) operated by measurement of the liquid phase DO concentrations [20]. A small air-tight respiration vessel (25 mL) was cyclically filled with activated sludge from the 1 L aerated vessel, then the oxygen consumption was measured and the mixed liquor was replaced for a new cycle. The fill-andstop sequence was composed of 50 s pumping to replace the vessel’s content and 100 s respiration measurement. In order to evaluate the respiration rates, substrates were added to the aerated vessel and DO and temperature were automatically recorded with a 5 s acquisition interval.
3. Results and discussion 3.1. Plant performance The bench scale plant was operated continuously for 115 days under constant volumetric loading rate, in order to monitor the system’s evolution over time in terms of performance, biomass growth and sludge features. Two experimental periods of 56 and 59 days were determined based on the two different volumetric 1 loading rates adopted (0.8 and 1.7 g COD Lreact. d 1, respectively). These were obtained by regulating the permeate flowrate according to the influent COD concentration. This is not a common operating procedure for MBRs (that are normally operated under constant flux) and it was adopted with the specific objective of monitoring the effects of constant load on the biology of the system. The average flowrate resulted 0.75 and 1.0 L h 1 during the first an the second period, respectively. The main operational parameters of the crMBR are shown in Table 1. The performance of the filtration unit is shown in Fig. 2. In the first period, the TMP tended to increase despite two chemical cleanings performed on days 10 and 43. Flux variations during the same period were
Table 1 Operational parameters of the bench scale membrane bioreactor Parameter
Period 1
2
1
Flux (L m h ) TMP (mbar) DO (mg L 1) Temperature ( C) PH
Period 2
Average
Min
Max
Average
Min
Max
8.0 360 3.0 24 6.9
3.1 150 0.7 20 6.3
12.4 700 7.1 27 8.0
11.4 160 2.0 20 6.9
7.2 30 0.2 17 5.3
12.9 450 3.9 24 7.4
ARTICLE IN PRESS A. Pollice et al. / Water Research 38 (2004) 1799–1808
1802 1.0
20 TMP
0.9
On-site jet rinsing with tap water
Flux
0.8
18 16 14
0.6
12
0.5
10
0.4
8
0.3
6
0.2
4
0.1
2
0.0 0
20
40
60
80
100
-2
TMP [bar]
-1
0.7
Flux [L m h ]
In situ chemical cleaning
0 120
days
Fig. 2. TMP and flux variations during the two experimental periods of 56 and 59 days.
80
100
80 60 50
60
40 40
30 Influent Ammonium Effluent Ammonium Effluent Nitrite Effluent Nitrate COD removal
20 10
20
0 0
20
40
60
80
COD removal [%]
-1
Inorganic nitrogen [mgN L ]
70
100
0 120
days
Fig. 3. Concentrations of influent ammonium, effluent ammonium, effluent nitrite, effluent nitrate and COD removal percentage during the whole experimental campaign.
mainly due to influent COD variability which was counteracted by flowrate regulation aimed at maintaining the volumetric loading rate constant. The initial cleaning strategy was to repeatedly backwash the fibres in situ (without removing the module from the bioreactor) with 10–15 mL of a 200 mg NaClO L 1 solution, when the TMP reached 0.4 bar. Then, it was observed that accurate on-site jet rinsing of the membrane module with moderately pressurized tap water resulted in better performance in terms of TMP recovery, and was then adopted as standard cleaning technique (days 56, 77, 94, and 108). This suggested that pressure increase was mainly caused by sludge accumulation and thickening in the spaces among the fibres, rather than fouling of the fibres’
surface. This feature was partly attributed to the limited effectiveness of aeration in moving and scouring the fibres of the lab-scale module (fibre length 0.2 m) with respect to a full scale one (fibre length 2.0 m). However, membrane cleaning techniques were only considered as a tool to maintain the whole process efficiency, and their investigation was beyond the scope of the present work. The plant was started up without any sludge inoculum, so that the system could select the most appropriate microbial environment. Biodegradation of the influent COD and complete nitrification were consistently obtained already in the first days of operation, indicating simple and rapid start-up of the membrane bioreactor. Fig. 3 shows the COD removal efficiency that averaged 87% and 94% in the first and
ARTICLE IN PRESS A. Pollice et al. / Water Research 38 (2004) 1799–1808
3.2. Biomass growth Accumulation of solids and biomass growth during the first period led, after 25 days, to an almost constant sludge concentration of 4.5–5.5 g TSS L 1, which was then maintained for 30 days, when the system was considered in equilibrium for the first tested volumetric loading rate (Fig. 4). During the second period, as soon as the volumetric loading rate was raised, the biomass concentration began to grow once again. In general, prompt response to load variations makes MBRs very
Table 2 Average performance of the membrane bioreactor in the two experimental periods Influent
1
TSS (mg L ) VSS (mg L 1) COD (mg L 1) NH4 (mg N L 1) NO3 (mg N L 1) TKN (mg N L 1)
Permeate
Period 1
Period 2
Period 1
Period 2
144 125 314 37 0 50
207 168 438 40 0 54
n.d. n.d. 41 0 40 1
n.d. n.d. 28 0 43 1
24
0.36
20
0.30
16
0.24
12
0.18
8
0.12
-1
-1
Sludge concentration [gTSS Lreact ]
n.d.=below the detection limit.
Sludge yield [gVSS gCOD ]
well suited for studying the biology of microrganisms under variable feed conditions, although comparisons with other systems are complicated by the peculiar biological interactions among bacterial strains due to retention of all microrganisms [21]. The sludge reached concentrations exceeding 15.0 g TSS L 1 in about 30 days, after which its growth rate decreased considerably. This was considered a new equilibrium, although the biomass concentration was not as stable as in the previous one. The equilibrium TSS concentrations for both periods were observed to be about 8 times the volumetric loading rate, in accordance with previous observations [9,10]. The observed sludge yield was calculated as the ratio between the VSS accumulated in the reactor and the cumulative COD removed. This calculation takes into account the sludge volumes extracted from the reactor for suspended solid determinations and those lost during on-site membrane cleaning, which affected the sludge concentration mainly in the second experimental period. A decreasing tendency of the sludge yield for increasing sludge concentration is shown in Fig. 4 and confirms previous observations [6,15]. When equilibrium biomass concentrations were approached, at the end of each experimental period, the sludge yield reached the same 1 value of 0.12 g VSS g CODrem corresponding to 0.17 g 1 COD g CODrem (using a factor of 1.42 g COD g VSS 1). This value is within the range of those reported for MBRs and lower than those observed in conventional activated sludge [1,22]. Fig. 5 reports the volumetric and organic loading rates over time. Also here, when the equilibrium biomass concentration was approched, the same organic loading rate of 0.12 g COD g TSS 1 d 1 (or 0.15 g COD g VSS 1 d 1) was reached independent of the volumetric loading rate.
second period, respectively. The ammonium oxidation performance over the whole experiment showed a typical nitrification start-up curve with initial N-NO2 production followed by complete nitrification that occurred only 10 days after the start-up of the plant. Table 2 shows the average features of the influent wastewater and the permeate during the two experimental periods.
Parameter
1803
4
0.06 Sludge concentration Sludge Yield
0 0
20
40
60
80
100
0.00 120
days
Fig. 4. Sludge concentration and observed sludge yield over time.
ARTICLE IN PRESS A. Pollice et al. / Water Research 38 (2004) 1799–1808
1804 2.0
1.0 Volumetric loading rate 0.8
1.2
0.6
0.8
0.4
0.4
0.2
0.0 0
20
40
60
80
100
Organic loading rate -1 -1 [gCOD gTSS d ]
Volumetric Loading Rate -1 -1 [gCOD Lreact d ]
Organic loading rate 1.6
0.0 120
days
Fig. 5. Variation of the volumetric loading rate and organic loading rate during the two experimental periods of 56 and 59 days.
The comparison between the volumetric loading rate and the F =M ratio over time suggests that the stability of the organic loading rate was not completely reached during the first period (despite the good stability of sludge concentration), and was only obtained in the last part of the second one. By definition, the ratio between the volumetric and organic loading rates under equilibrium conditions provides the steady-state biomass concentration. In the experiments presented here, for the two volumetric loading rates considered and for the same equilibrium organic loading rates (0.12 g COD g TSS 1 d 1), these theoretical concentra1 tions were 6.7 and 14.2 g TSS Lreact. , respectively. In other words, for the crMBRs the organic loading rate seems more accurate as an indicator of steady-state conditions than the biomass solids concentration. The ratio between the steady-state sludge concentration and the volumetric loading rate (i.e. the inverse of the organic loading rate) was 8.3 for our experiments, meaning that the steady-state sludge concentration was 8.3 times the volumetric loading rate, thus similar to the observed experimental value. Of course, the value of the above proposed ratio is specific for the system tested and depends on a number of factors, including operational conditions and feed characteristics. The VSS/TSS ratio was rather stable over the whole experiment, with average values of 80% in the first period and 79% in the second one. The accumulation of inert solids in the reactor was calculated by considering the difference between TSS and VSS in the sludge and the cumulative difference in the influent wastewater, i.e. the two non-volatile fractions. At the end of the experiment, the inert material in the sludge was about 27 g and the total non-volatile solids that had entered the reactor during the whole experiment were about 77 g
(Fig. 6). The influent contribution to the reactor’s inert concentration was observed to decrease very slowly over time. The average ratio between incoming inert material and biomass non-volatile fraction was 44% in the first period and 43% in the second one. Therefore, the ratio between the inert fraction entering with the sewage and the one resulting in the biomass was almost constant (Fig. 6). These results show that a relevant fraction of the incoming inert particulate material was not accumulated in the reactor, possibly due to hydrolysis or enzymatic solubilization producing compounds having molecular size compatible with permeation. 3.3. Respiration rates In the activated sludge process, oxygen consumption is directly associated with both substrate removal and biomass growth. Therefore, the oxygen respiration rate per unit volume per unit time (OUR) is widely recognized as an important parameter to monitor the biomass viability [20]. In order to monitor the viability of the biomass selected in the crMBR over time, endogenous and maximum respiration rates were evaluated. The endogenous respiration rate (rOend ) is defined as the oxygen consumption rate in the absence of substrate and includes consumption for bacterial growth-decay cycle, maintenance energy production and protozoa respiration. The maximum respiration rate (rOmax ) is defined as the oxygen consumption rate reached when all the individual substrates that can be oxidated by a heterogeneous microbial population are present in excess. This condition is not very likely to occur, but a respiration rate in the presence of an excess of a specific substrate or group of substrates can be measured [20].
ARTICLE IN PRESS A. Pollice et al. / Water Research 38 (2004) 1799–1808
1805
40
100
36 32
Sludge Inert [g]
28 24
60
20 16
40
12 8
Cumulative Influent Inert [g] VSS/TSS [%]
80
20 Sludge Inert Cumulative Influent Inert VSS / TSS
4 0 0
20
40
60 days
80
100
0 120
Fig. 6. Sludge inert material, cumulative influent inert material and VSS to TSS ratio over time.
120 Endogenous respiration rate Max acetate respiration rate Max ammonium respiration rate
-1
Respiration rate [mgO2 h ]
100
80
60
40
20
0 0
20
40
60
80
100
120
days
Fig. 7. Evolution of endogenous and maximum respiration rate with acetate and ammonium.
In this study, sodium acetate was selected as a carbonaceous substrate because it is readily biodegradable by most heterotrophic populations, and ammonium chloride was used to monitor the activity of autotrophic nitrifying populations (both ammonium and nitrite oxidizers). The evolution of endogenous and maximum respiration rates over the experimental campaign is shown in Fig. 7. The rOend and rOacetate showed the same trend, as expected considering that both parameters are referred to the heterotrophic biomass. During the transient step of each period, the respiration rate rose due to the increase of biomass concentration and viability. When the amount of substrate available approached the maintenance require-
ment, the respiration rate slowly decreased towards a new equilibrium, possibly due to sludge ageing. Although this equilibrium was not clearly visible in the first period, it is evident that by doubling the 1 volumetric loading rate (0.8–1.7 g COD Lreact. d 1) both the endogenous and the maximum respiration rates triplicated (9–30 mg O2 h 1 and 20–60 mg O2 h 1, respectively, Fig. 7). Consequently, a net improvement of the influent COD removal was observed (Fig. 3 and Table 2). A slightly different behaviour was shown by the autotrophic populations, probably due to their slower growth rate. Comparison between Figs. 7 and 4 shows that a slight increase of VSS concentration (which include the active
ARTICLE IN PRESS A. Pollice et al. / Water Research 38 (2004) 1799–1808
1806 20
Specific endogenous respiration rate Specific max acetate respiration rate Specific max ammonium respiration rate
18
Specific Respiration Rate -1 -1 [mgO2 gVSS h ]
16 14 12 10 8 6 4 2 0 0
20
40
60
80
100
120
days
Fig. 8. Evolution of specific endogenous and specific maximum respiration rate with acetate and ammonium.
biomass) during the two equilibrium periods corresponds to an almost constant respiration rate. In order to better understand how biomass growth and viability contribute to the evolution of the biomass activity, the specific respiration rates (ROend and ROmax ) were calculated (Fig. 8). Also in this case, ROend and ROacetate show similar trends, and it can be noted that at the end of each period the two specific respiration rates approached the same values (2 and 4 mg O2 g VSS 1 h 1, respectively) independent of the volumetric loading rate. Again, in the first period the autotrophic population showed a slightly different behaviour. Despite the low values of the specific respiration rates, the biomass showed good promptness in responding to organic loading rate increases. Moreover, when considering the high sludge concentrations characterizing these systems, the overall respiration rate results, comparable with those observed in traditional activated sludge plants, ensure high COD removal efficiencies and complete nitrification. It should be noted that the specific respiration rates and the organic loading rates had similar trends over time and both tended to approach a minimum equilibrium value (Figs 5 and 8). This suggests that, once the volumetric loading rate has been fixed, the sludge concentration freely evolves until the F =M ratio approaches a minimum value. It seems realistic to think that the minimum F/M ratio could depend on the specific endogenous respiration rate and on feed biodegradability. In this study, an equilibrium F =M ratio of 0.15 g COD g VSS 1 d 1 was found, i.e. only three times the equilibrium ROend (0.05 g O2 g VSS 1 d 1). Thus, the biochemical energy entering the
system has the same order of magnitude of the minimum mainteinance energy. This could also account for the 1 low observed sludge yield (0.12 g VSS g CODrem ).
4. Conclusions A bench scale membrane bioreactor was operated without any intentional sludge withdrawal and under constant volumetric loading rate in order to monitor the system’s evolution over time in terms of performance, biomass growth and sludge features. During the two experimental periods two different volumetric loading 1 rates were adopted (0.8 and 1.7 g COD Lreact. d 1). The plant was started up without any sludge inoculum, so that the system could select the most appropriate microbial environment. Biodegradation of the influent COD and complete nitrification were consistently obtained already in the first days of operation, confirming prompt and simple start-up of these systems. The SS concentrations under equilibrium conditions for both experimental periods were slightly lower than the steady-state value, calculated as the ratio between the imposed volumetric loading rate and the equilibrium organic loading rate reached by the system. This relationship between the steady-state sludge concentration and the volumetric loading rate was observed to be constant, and proposed to be typical of the system under given operating conditions and feed characteristics. Moreover, the VSS/TSS ratio was rather stable over the whole experiment. The mass balance of solid fractions showed that a relevant fraction of the incoming inert particulate
ARTICLE IN PRESS A. Pollice et al. / Water Research 38 (2004) 1799–1808
material was not accumulated in the reactor, possibly due to hydrolysis or enzymatic solubilization producing compounds having molecular size compatible with permeation. However, this aspect deserves further observations that will be carried out over long-term operation. In terms of activity, it was observed that by doubling 1 the volumetric loading rate (0.8–1.7 g COD Lreact. d 1) both the endogenous and the maximum heterotrophic respiration rates triplicated. Furthermore, the specific respiration rates and the organic loading rate had similar trends over time and both tended to approach a minimum equilibrium value (2 and 4 mg O2 g VSS 1 h 1 for ROend and ROacetate ; respectively, and 0.12 g COD g TSS 1 d 1 for the F =M ratio). These equilibrium values seem to be independent of the volumetric loading rate and this suggested that, once the volumetric loading rate has been fixed, the sludge concentration of such systems freely evolves until the F =M ratio approaches a minimum value, which appears to be typical of the operational conditions tested. These experimental observations may provide indications on some aspects of MBR design and management, such as forecasting steady-state biomass concentrations based on the plant’s volumetric loading rate and ability of these systems to readily cope with organic shockloads. The evaluation of the need and frequency of sludge withdrawal for limiting build-up of inert materials within the plant will have to be assessed over the long term.
Acknowledgements The present work is dedicated to the memory of prof. Alberto Rozzi, whose enthusiasm and dedication to water and environmental science and technologies showed the way to many others. The authors also acknowledge Zenon Environmental S.r.l. for providing the membrane module.
References [1] Stephenson T, Judd S, Jefferson B, Brindle K. Membrane bioreactors for wastewater treatment. London, UK: IWA Publishing; 2000. [2] Visvanathan C, Ben Aim R, Parameshwaran K. Membrane separation bioreactors for wastewater treatment. Crit Rev Environ Sci Technol 2000;30(1):1–48. [3] Arnot T, Howell J. Manipulation of cell physiology in membrane bioreactors. Proceedings of the MBR3—Third International Meeting on Membrane Bioreactors for Wastewater Treatment. Cranfield Univerisity, Cranfield, UK, 16 May 2001.
1807
[4] Chaize S, Huyard A. Membrane bioreactor on domestic wastewater treatment. Sludge production and modeling approach. Water Sci Technol (Kyoto) 1991;23: 1591–600. [5] Suwa Y, Suzuki T, Toyohara H, Yamagishi T, Urushigawa Y. Single-stage, single-sludge nitrogen removal by an activated sludge process with cross-flow filtration. Water Res 1992;26(9):1149–57. [6] Low EW, Chase HA. The effects of maintenance energy requirements on biomass production during wastewater treatment. Water Res 1999;33(3):847–53. [7] van Houten RT, Eikelboom DH High performance membrane bioreactors: A physiological approach. Proceedings of the MBR1—First International Meeting on Membrane Bioreactors for Wastewater Treatment Cranfield University, Cranfield, UK, 5 March 1997. [8] Pirt SJ. The maintenance energy of bacteria in growing cultures. Proc R Soc London 1965;163B:224–31. [9] Yamamoto K, Hiasa M, Mahmood T, Matsuo T. Direct solid-liquid separation using hollow fiber membrane in an activated sludge aeration tank. Water Sci Technol (Brighton) 1989;21:43–54. [10] Chiemchaisri C, Wong YK, Urase T, Yamamoto K. Organic stabilization and nitrogen removal in membrane separation bioreactor for domestic wastewater treatment. Water Sci Technol 1992;25(10):231–40. [11] Muller EB, Stouthamer AH, van Verseveld HW, Eikelboom DH. Aerobic domestic waste water treatment in a pilot plant with complete sludge retention by cross-flow filtration. Water Res 1995;29(4):1179–89. [12] Ben"ıtez J, Rodr"ıguez A, Malaver R. Stabilization and dewatering of wastewater using hollow fiber membranes. Water Res 1995;29(10):2281–6. [13] Rosenberger S, Kubin K, Kraume M. Operation of membrane bioreactors without sludge withdrawal. Proceedings of the Journees Internationales d’Etude du Cebedeau L’Eau et les Techniques Membranaires, Li"ege, 4–5 Avril 2000. [14] Rosenberger S, Witzig R, Manz W, Szewzyk U, Kraume M. Operation of different membrane bioreactors: experimental results and physiological state of the microorganisms. Water Sci Technol 2000;41(10–11):269–77. [15] Rosenberger S, Kruger . U, Witzig R, Manz W, Szewzyk U, Kraume M. Performance of a bioreactor with submerged membranes for aerobic treatment of municipal waste water. Water Res 2002;36:413–20. [16] Witzig R, Manz W, Rosenberger S, Kruger . U, Kraume M, Szewzyk U. Microbiological aspects of a bioreactor with submerged membranes for aerobic treatment of municipal wastewater. Water Res 2002;36:394–402. [17] Wagner J, Rosenwinkel K-H. Sludge production in membrane bioreactors under different conditions. Water Sci Technol 2000;41(10–11):251–8. [18] Lubbeke . S, Vogelpohl A, Dewjanin W. Wastewater treatment in a biological high-performance system with high biomass concentration. Water Res 1995;29(3): 793–802. [19] Eaton AD, Clesceri LS, Greenberg A, editors. Standard methods for the examination of water and wastewater. 19th ed.. Washington, DC: APHA, AWWA, WEF; 1995.
ARTICLE IN PRESS 1808
A. Pollice et al. / Water Research 38 (2004) 1799–1808
[20] Spanjers H, Vanrolleghem PA, Olsson G, Dold PL. Respirometry in control of the activated sludge process: principles. IAWQ Scientific and Technical Report no. 7. London, UK: IAWQ Publishing; 1998. [21] Smith S, Judd S, Stephenson T, Jefferson B. Membrane bioreactors—hybrid activated sludge or a new process?
Proceedings of the Fifth IMSTEC International Membrane Science and Technology Conference, UNSW, Sydney, Australia, 10–14 November 2003. [22] Metcalf A, Eddy A. Wastewater engineering: treatment, disposal and reuse. New York, SA: McGraw-Hill Int. Ed; 1991.
Water Research 38 (2004) 1799–1808
Biomass growth and activity in a membrane bioreactor with complete sludge retention Alfieri Pollice*, Giuseppe Laera, Massimo Blonda CNR Istituto di Ricerca Sulle Acque, Viale F. De Blasio, n. 5 (Zona Ind.), Bari 70123, Italy Received 31 July 2003; received in revised form 22 December 2003; accepted 5 January 2004
Abstract This work reports the main results of a bench scale membrane bioreactor operated for more than 100 days without sludge withdrawal and fed on real municipal wastewater. The experiments were oriented towards three main objectives. Firstly, the performance of the system was evaluated under two different volumetric loading rates (0.8 and 1 1.7 g COD Lreact. d 1). Secondly, biomass growth and accumulation of solids were assessed and a relationship between sludge concentration and volumetric loading rates was proposed. Thirdly, biomass activity was evaluated through respirometric tests, and endogenous and maximum respiration rates of heterotrophic and nitrifying bacteria were determined. The experimental campaign showed that these systems are easy to manage and very rapid to start-up. The SS concentrations under equilibrium conditions for both experimental periods were slightly lower than 10 times the volumetric loading rates, and the organic loading rates reached the same equilibrium value of 0.12 g COD g TSS 1 d 1. 1 Furthermore, under equilibrium conditions the system showed very limited sludge production (0.12 g VSS g CODrem ) and low biomass activity, although it readily responded to load variations. Further work is being carried out to evaluate the performance over the long term. r 2004 Elsevier Ltd. All rights reserved. Keywords: Membrane bioreactors; Complete sludge retention; Biomass activity; Municipal wastewater; Sludge yield
1. Introduction During the last decades, an increasing number of investigations have improved the knowledge of the membrane bioreactors and the interactions between biological wastewater treatment processes and membrane filtration. Research activities were mainly directed towards the optimization of process efficiency in terms of effluent quality, membrane types, plant configurations and operational practices [1,2]. Less attention has been devoted to the peculiarities of biological processes and microbial consortia developed through the application of membrane separation. Control of the bacterial environment through efficient solid/liquid separation and operation of the biological processes with high *Corresponding author. Tel.: +39-080-582-05-11; fax: +39080-531-33-65. E-mail address: [email protected] (A. Pollice).
biomass concentrations are likely to affect cell metabolism and limit the bacterial growth, and thus the sludge production [3–5]. By operating the processes at low organic loading rates (g COD g TSS 1 d 1) it is possible to divert the utilization of polluting compounds from biosynthesis to non-growth energy-demanding activities, favour the utilization of substrates for maintenance of bacterial vital functions, and limit the net growth [6,7]. This is the concept of maintenance energy, first introduced by Pirt in 1965, that can be considered as the amount of biochemical energy strictly necessary for endogenous respiration [8]. A direct consequence of the previous statements is that operating a membrane bioreactor without any sludge wastage should allow the system to reach a steady state at a given sludge concentration. Under these conditions, defined ‘‘complete sludge retention’’ by van Houten and Eikelboom, the biomass growth rate should balance the decay rate [7].
0043-1354/$ - see front matter r 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2004.01.016
ARTICLE IN PRESS A. Pollice et al. / Water Research 38 (2004) 1799–1808
1800
Early investigations on complete retention membrane bioreactors (crMBR) indicated that under constant F =M ratios the sludge concentration was almost constant [9,10]. A flattening of the sludge growth curve, suggesting equilibrium between endogenous respiration and new bacterial mass growth in MBRs with complete sludge retention was also observed [7,11,12]. Other researchers proposed the possibility of controlling the biomass concentration by regulating the influent organic load, also considering the substrate’s biodegradability [13–17]. The same authors suggested that operational strategies based on slow biomass turnover (i.e. limited sludge wastage) could be preferable to complete sludge retention when accumulation of inorganics such as heavy metals is likely to occur. Studies on the accumulation of inert material reported constant percentage of inorganics in the sludge, independent of its concentration [7]. Oxygen transfer limitations in MBRs operated with high biomass concentrations were reported by several authors [9,11,18]. The present work reports some results of an experimental campaign aimed at evaluating the efficiency of a laboratory scale crMBR fed on real municipal wastewater. The plant was operated for more than 100 days, during which the main operational parameters were monitored and controlled in order to let biological processes evolve towards equilibrium. The biomass was examined over time to evaluate the dependency of its growth patterns on process conditions. Moreover, the activity of heterotrophic and nitrifying organisms was regularly monitored by performing respirometric
tests, in order to investigate the variation of their viability.
2. Materials and methods 2.1. Experimental plant A scheme of the experimental set-up is reported in Fig. 1. The membrane bioreactor had 6 L operating volume and was mechanically stirred. A Zenon ZeeWeeds lab-scale hollow fibre membrane module (nominal surface 0.094 m2) was sank in the bioreactor and operated out-in. The permeate was extracted by imposing on the fibres a negative pressure, normally maintained well below the limiting value suggested by the manufacturer (0.7 bar). Operational cycles lasted 6 min and included extraction of the permeate (5.5 min) and backwash (0.5 min). Moreover, a constant air flow was pumped through the membrane module to prevent or reduce fouling and cake formation. The permeate flux was regulated depending on the influent COD concentration, in order to maintain a predetermined volumetric loading rate. The latter is defined as the amount of COD daily fed to the plant and 1 referred to the bioreactor’s volume (g COD Lreact. d 1). The system was continuously fed on pre-settled municipal wastewater screened at 1 mm, stored at 4 C and continuously mixed. The plant was started up without any biomass inoculum, and during the whole
AIR PUMP 2 pH-METER
AIR PUMP 1
MBR
MANOMETER CIP
OXYMETER
FEED TANK
TIMER
PROCESS PUMP LEVEL CONTROL
INFLUENT COD ANALYSIS
Fig. 1. Scheme of the experimental set-up.
ARTICLE IN PRESS A. Pollice et al. / Water Research 38 (2004) 1799–1808
experimental period no sludge was intentionally withdrawn from the plant except for measurements of suspended solids. Very limited amounts of sludge went lost when the module was removed from the reactor for cleaning and when the mixed liquor was temporarily sampled for respirometric determinations. These amounts were accounted for during the calculation of biomass production and the other sludge-related parameters. As a general rule of this investigation, biomass samples were always returned to the reactor after nondestructive determinations. A certain amount of biomass tended to stick to the parts of the reactor (walls, plastic pipes, etc.) close to the surface level. This fraction was daily removed and returned to the sludge bulk. The process was operated under continuous aeration in order to achieve biodegradation of the organic fractions and nitrification. Oxygen was provided through an air diffuser placed in the reactor. The air flowrate provided by this second air pump was regulated to maintain residual dissolved oxygen (DO) concentrations between 1 and 3 mg L 1. 2.2. Sampling and analyses The bench scale crMBR was continuously monitored for DO, temperature, pH, and transmembrane pressure (TMP). The influent municipal wastewater was sampled three times per week and analysed for total and volatile suspended solids (TSS and VSS), total COD, N-NH4, TKN, N-NO2, and N-NO3. The permeate was sampled daily and analysed for the same parameters. All analyses were performed according to standard methods [19]. The ion chromatographic method for the determination of nitrate and nitrite was modified adopting an UV detector (at 220 nm) instead of the conductivity detector. Biomass samples were withdrawn from the reactor twice per week and analysed for TSS and VSS according to standard methods [19]. One litre of the reactor’s content was withdrawn twice per week for respirometric tests and returned to the bioreactor afterwards. All respirometric tests were performed after aerating the sludge sample overnight
1801
in order to obtain endogenous respiration conditions. According to the definitions of the IAWQ Task Group on Respirometry, the respirometer adopted was a ‘‘static gas–static liquid’’ system (LSS) operated by measurement of the liquid phase DO concentrations [20]. A small air-tight respiration vessel (25 mL) was cyclically filled with activated sludge from the 1 L aerated vessel, then the oxygen consumption was measured and the mixed liquor was replaced for a new cycle. The fill-andstop sequence was composed of 50 s pumping to replace the vessel’s content and 100 s respiration measurement. In order to evaluate the respiration rates, substrates were added to the aerated vessel and DO and temperature were automatically recorded with a 5 s acquisition interval.
3. Results and discussion 3.1. Plant performance The bench scale plant was operated continuously for 115 days under constant volumetric loading rate, in order to monitor the system’s evolution over time in terms of performance, biomass growth and sludge features. Two experimental periods of 56 and 59 days were determined based on the two different volumetric 1 loading rates adopted (0.8 and 1.7 g COD Lreact. d 1, respectively). These were obtained by regulating the permeate flowrate according to the influent COD concentration. This is not a common operating procedure for MBRs (that are normally operated under constant flux) and it was adopted with the specific objective of monitoring the effects of constant load on the biology of the system. The average flowrate resulted 0.75 and 1.0 L h 1 during the first an the second period, respectively. The main operational parameters of the crMBR are shown in Table 1. The performance of the filtration unit is shown in Fig. 2. In the first period, the TMP tended to increase despite two chemical cleanings performed on days 10 and 43. Flux variations during the same period were
Table 1 Operational parameters of the bench scale membrane bioreactor Parameter
Period 1
2
1
Flux (L m h ) TMP (mbar) DO (mg L 1) Temperature ( C) PH
Period 2
Average
Min
Max
Average
Min
Max
8.0 360 3.0 24 6.9
3.1 150 0.7 20 6.3
12.4 700 7.1 27 8.0
11.4 160 2.0 20 6.9
7.2 30 0.2 17 5.3
12.9 450 3.9 24 7.4
ARTICLE IN PRESS A. Pollice et al. / Water Research 38 (2004) 1799–1808
1802 1.0
20 TMP
0.9
On-site jet rinsing with tap water
Flux
0.8
18 16 14
0.6
12
0.5
10
0.4
8
0.3
6
0.2
4
0.1
2
0.0 0
20
40
60
80
100
-2
TMP [bar]
-1
0.7
Flux [L m h ]
In situ chemical cleaning
0 120
days
Fig. 2. TMP and flux variations during the two experimental periods of 56 and 59 days.
80
100
80 60 50
60
40 40
30 Influent Ammonium Effluent Ammonium Effluent Nitrite Effluent Nitrate COD removal
20 10
20
0 0
20
40
60
80
COD removal [%]
-1
Inorganic nitrogen [mgN L ]
70
100
0 120
days
Fig. 3. Concentrations of influent ammonium, effluent ammonium, effluent nitrite, effluent nitrate and COD removal percentage during the whole experimental campaign.
mainly due to influent COD variability which was counteracted by flowrate regulation aimed at maintaining the volumetric loading rate constant. The initial cleaning strategy was to repeatedly backwash the fibres in situ (without removing the module from the bioreactor) with 10–15 mL of a 200 mg NaClO L 1 solution, when the TMP reached 0.4 bar. Then, it was observed that accurate on-site jet rinsing of the membrane module with moderately pressurized tap water resulted in better performance in terms of TMP recovery, and was then adopted as standard cleaning technique (days 56, 77, 94, and 108). This suggested that pressure increase was mainly caused by sludge accumulation and thickening in the spaces among the fibres, rather than fouling of the fibres’
surface. This feature was partly attributed to the limited effectiveness of aeration in moving and scouring the fibres of the lab-scale module (fibre length 0.2 m) with respect to a full scale one (fibre length 2.0 m). However, membrane cleaning techniques were only considered as a tool to maintain the whole process efficiency, and their investigation was beyond the scope of the present work. The plant was started up without any sludge inoculum, so that the system could select the most appropriate microbial environment. Biodegradation of the influent COD and complete nitrification were consistently obtained already in the first days of operation, indicating simple and rapid start-up of the membrane bioreactor. Fig. 3 shows the COD removal efficiency that averaged 87% and 94% in the first and
ARTICLE IN PRESS A. Pollice et al. / Water Research 38 (2004) 1799–1808
3.2. Biomass growth Accumulation of solids and biomass growth during the first period led, after 25 days, to an almost constant sludge concentration of 4.5–5.5 g TSS L 1, which was then maintained for 30 days, when the system was considered in equilibrium for the first tested volumetric loading rate (Fig. 4). During the second period, as soon as the volumetric loading rate was raised, the biomass concentration began to grow once again. In general, prompt response to load variations makes MBRs very
Table 2 Average performance of the membrane bioreactor in the two experimental periods Influent
1
TSS (mg L ) VSS (mg L 1) COD (mg L 1) NH4 (mg N L 1) NO3 (mg N L 1) TKN (mg N L 1)
Permeate
Period 1
Period 2
Period 1
Period 2
144 125 314 37 0 50
207 168 438 40 0 54
n.d. n.d. 41 0 40 1
n.d. n.d. 28 0 43 1
24
0.36
20
0.30
16
0.24
12
0.18
8
0.12
-1
-1
Sludge concentration [gTSS Lreact ]
n.d.=below the detection limit.
Sludge yield [gVSS gCOD ]
well suited for studying the biology of microrganisms under variable feed conditions, although comparisons with other systems are complicated by the peculiar biological interactions among bacterial strains due to retention of all microrganisms [21]. The sludge reached concentrations exceeding 15.0 g TSS L 1 in about 30 days, after which its growth rate decreased considerably. This was considered a new equilibrium, although the biomass concentration was not as stable as in the previous one. The equilibrium TSS concentrations for both periods were observed to be about 8 times the volumetric loading rate, in accordance with previous observations [9,10]. The observed sludge yield was calculated as the ratio between the VSS accumulated in the reactor and the cumulative COD removed. This calculation takes into account the sludge volumes extracted from the reactor for suspended solid determinations and those lost during on-site membrane cleaning, which affected the sludge concentration mainly in the second experimental period. A decreasing tendency of the sludge yield for increasing sludge concentration is shown in Fig. 4 and confirms previous observations [6,15]. When equilibrium biomass concentrations were approached, at the end of each experimental period, the sludge yield reached the same 1 value of 0.12 g VSS g CODrem corresponding to 0.17 g 1 COD g CODrem (using a factor of 1.42 g COD g VSS 1). This value is within the range of those reported for MBRs and lower than those observed in conventional activated sludge [1,22]. Fig. 5 reports the volumetric and organic loading rates over time. Also here, when the equilibrium biomass concentration was approched, the same organic loading rate of 0.12 g COD g TSS 1 d 1 (or 0.15 g COD g VSS 1 d 1) was reached independent of the volumetric loading rate.
second period, respectively. The ammonium oxidation performance over the whole experiment showed a typical nitrification start-up curve with initial N-NO2 production followed by complete nitrification that occurred only 10 days after the start-up of the plant. Table 2 shows the average features of the influent wastewater and the permeate during the two experimental periods.
Parameter
1803
4
0.06 Sludge concentration Sludge Yield
0 0
20
40
60
80
100
0.00 120
days
Fig. 4. Sludge concentration and observed sludge yield over time.
ARTICLE IN PRESS A. Pollice et al. / Water Research 38 (2004) 1799–1808
1804 2.0
1.0 Volumetric loading rate 0.8
1.2
0.6
0.8
0.4
0.4
0.2
0.0 0
20
40
60
80
100
Organic loading rate -1 -1 [gCOD gTSS d ]
Volumetric Loading Rate -1 -1 [gCOD Lreact d ]
Organic loading rate 1.6
0.0 120
days
Fig. 5. Variation of the volumetric loading rate and organic loading rate during the two experimental periods of 56 and 59 days.
The comparison between the volumetric loading rate and the F =M ratio over time suggests that the stability of the organic loading rate was not completely reached during the first period (despite the good stability of sludge concentration), and was only obtained in the last part of the second one. By definition, the ratio between the volumetric and organic loading rates under equilibrium conditions provides the steady-state biomass concentration. In the experiments presented here, for the two volumetric loading rates considered and for the same equilibrium organic loading rates (0.12 g COD g TSS 1 d 1), these theoretical concentra1 tions were 6.7 and 14.2 g TSS Lreact. , respectively. In other words, for the crMBRs the organic loading rate seems more accurate as an indicator of steady-state conditions than the biomass solids concentration. The ratio between the steady-state sludge concentration and the volumetric loading rate (i.e. the inverse of the organic loading rate) was 8.3 for our experiments, meaning that the steady-state sludge concentration was 8.3 times the volumetric loading rate, thus similar to the observed experimental value. Of course, the value of the above proposed ratio is specific for the system tested and depends on a number of factors, including operational conditions and feed characteristics. The VSS/TSS ratio was rather stable over the whole experiment, with average values of 80% in the first period and 79% in the second one. The accumulation of inert solids in the reactor was calculated by considering the difference between TSS and VSS in the sludge and the cumulative difference in the influent wastewater, i.e. the two non-volatile fractions. At the end of the experiment, the inert material in the sludge was about 27 g and the total non-volatile solids that had entered the reactor during the whole experiment were about 77 g
(Fig. 6). The influent contribution to the reactor’s inert concentration was observed to decrease very slowly over time. The average ratio between incoming inert material and biomass non-volatile fraction was 44% in the first period and 43% in the second one. Therefore, the ratio between the inert fraction entering with the sewage and the one resulting in the biomass was almost constant (Fig. 6). These results show that a relevant fraction of the incoming inert particulate material was not accumulated in the reactor, possibly due to hydrolysis or enzymatic solubilization producing compounds having molecular size compatible with permeation. 3.3. Respiration rates In the activated sludge process, oxygen consumption is directly associated with both substrate removal and biomass growth. Therefore, the oxygen respiration rate per unit volume per unit time (OUR) is widely recognized as an important parameter to monitor the biomass viability [20]. In order to monitor the viability of the biomass selected in the crMBR over time, endogenous and maximum respiration rates were evaluated. The endogenous respiration rate (rOend ) is defined as the oxygen consumption rate in the absence of substrate and includes consumption for bacterial growth-decay cycle, maintenance energy production and protozoa respiration. The maximum respiration rate (rOmax ) is defined as the oxygen consumption rate reached when all the individual substrates that can be oxidated by a heterogeneous microbial population are present in excess. This condition is not very likely to occur, but a respiration rate in the presence of an excess of a specific substrate or group of substrates can be measured [20].
ARTICLE IN PRESS A. Pollice et al. / Water Research 38 (2004) 1799–1808
1805
40
100
36 32
Sludge Inert [g]
28 24
60
20 16
40
12 8
Cumulative Influent Inert [g] VSS/TSS [%]
80
20 Sludge Inert Cumulative Influent Inert VSS / TSS
4 0 0
20
40
60 days
80
100
0 120
Fig. 6. Sludge inert material, cumulative influent inert material and VSS to TSS ratio over time.
120 Endogenous respiration rate Max acetate respiration rate Max ammonium respiration rate
-1
Respiration rate [mgO2 h ]
100
80
60
40
20
0 0
20
40
60
80
100
120
days
Fig. 7. Evolution of endogenous and maximum respiration rate with acetate and ammonium.
In this study, sodium acetate was selected as a carbonaceous substrate because it is readily biodegradable by most heterotrophic populations, and ammonium chloride was used to monitor the activity of autotrophic nitrifying populations (both ammonium and nitrite oxidizers). The evolution of endogenous and maximum respiration rates over the experimental campaign is shown in Fig. 7. The rOend and rOacetate showed the same trend, as expected considering that both parameters are referred to the heterotrophic biomass. During the transient step of each period, the respiration rate rose due to the increase of biomass concentration and viability. When the amount of substrate available approached the maintenance require-
ment, the respiration rate slowly decreased towards a new equilibrium, possibly due to sludge ageing. Although this equilibrium was not clearly visible in the first period, it is evident that by doubling the 1 volumetric loading rate (0.8–1.7 g COD Lreact. d 1) both the endogenous and the maximum respiration rates triplicated (9–30 mg O2 h 1 and 20–60 mg O2 h 1, respectively, Fig. 7). Consequently, a net improvement of the influent COD removal was observed (Fig. 3 and Table 2). A slightly different behaviour was shown by the autotrophic populations, probably due to their slower growth rate. Comparison between Figs. 7 and 4 shows that a slight increase of VSS concentration (which include the active
ARTICLE IN PRESS A. Pollice et al. / Water Research 38 (2004) 1799–1808
1806 20
Specific endogenous respiration rate Specific max acetate respiration rate Specific max ammonium respiration rate
18
Specific Respiration Rate -1 -1 [mgO2 gVSS h ]
16 14 12 10 8 6 4 2 0 0
20
40
60
80
100
120
days
Fig. 8. Evolution of specific endogenous and specific maximum respiration rate with acetate and ammonium.
biomass) during the two equilibrium periods corresponds to an almost constant respiration rate. In order to better understand how biomass growth and viability contribute to the evolution of the biomass activity, the specific respiration rates (ROend and ROmax ) were calculated (Fig. 8). Also in this case, ROend and ROacetate show similar trends, and it can be noted that at the end of each period the two specific respiration rates approached the same values (2 and 4 mg O2 g VSS 1 h 1, respectively) independent of the volumetric loading rate. Again, in the first period the autotrophic population showed a slightly different behaviour. Despite the low values of the specific respiration rates, the biomass showed good promptness in responding to organic loading rate increases. Moreover, when considering the high sludge concentrations characterizing these systems, the overall respiration rate results, comparable with those observed in traditional activated sludge plants, ensure high COD removal efficiencies and complete nitrification. It should be noted that the specific respiration rates and the organic loading rates had similar trends over time and both tended to approach a minimum equilibrium value (Figs 5 and 8). This suggests that, once the volumetric loading rate has been fixed, the sludge concentration freely evolves until the F =M ratio approaches a minimum value. It seems realistic to think that the minimum F/M ratio could depend on the specific endogenous respiration rate and on feed biodegradability. In this study, an equilibrium F =M ratio of 0.15 g COD g VSS 1 d 1 was found, i.e. only three times the equilibrium ROend (0.05 g O2 g VSS 1 d 1). Thus, the biochemical energy entering the
system has the same order of magnitude of the minimum mainteinance energy. This could also account for the 1 low observed sludge yield (0.12 g VSS g CODrem ).
4. Conclusions A bench scale membrane bioreactor was operated without any intentional sludge withdrawal and under constant volumetric loading rate in order to monitor the system’s evolution over time in terms of performance, biomass growth and sludge features. During the two experimental periods two different volumetric loading 1 rates were adopted (0.8 and 1.7 g COD Lreact. d 1). The plant was started up without any sludge inoculum, so that the system could select the most appropriate microbial environment. Biodegradation of the influent COD and complete nitrification were consistently obtained already in the first days of operation, confirming prompt and simple start-up of these systems. The SS concentrations under equilibrium conditions for both experimental periods were slightly lower than the steady-state value, calculated as the ratio between the imposed volumetric loading rate and the equilibrium organic loading rate reached by the system. This relationship between the steady-state sludge concentration and the volumetric loading rate was observed to be constant, and proposed to be typical of the system under given operating conditions and feed characteristics. Moreover, the VSS/TSS ratio was rather stable over the whole experiment. The mass balance of solid fractions showed that a relevant fraction of the incoming inert particulate
ARTICLE IN PRESS A. Pollice et al. / Water Research 38 (2004) 1799–1808
material was not accumulated in the reactor, possibly due to hydrolysis or enzymatic solubilization producing compounds having molecular size compatible with permeation. However, this aspect deserves further observations that will be carried out over long-term operation. In terms of activity, it was observed that by doubling 1 the volumetric loading rate (0.8–1.7 g COD Lreact. d 1) both the endogenous and the maximum heterotrophic respiration rates triplicated. Furthermore, the specific respiration rates and the organic loading rate had similar trends over time and both tended to approach a minimum equilibrium value (2 and 4 mg O2 g VSS 1 h 1 for ROend and ROacetate ; respectively, and 0.12 g COD g TSS 1 d 1 for the F =M ratio). These equilibrium values seem to be independent of the volumetric loading rate and this suggested that, once the volumetric loading rate has been fixed, the sludge concentration of such systems freely evolves until the F =M ratio approaches a minimum value, which appears to be typical of the operational conditions tested. These experimental observations may provide indications on some aspects of MBR design and management, such as forecasting steady-state biomass concentrations based on the plant’s volumetric loading rate and ability of these systems to readily cope with organic shockloads. The evaluation of the need and frequency of sludge withdrawal for limiting build-up of inert materials within the plant will have to be assessed over the long term.
Acknowledgements The present work is dedicated to the memory of prof. Alberto Rozzi, whose enthusiasm and dedication to water and environmental science and technologies showed the way to many others. The authors also acknowledge Zenon Environmental S.r.l. for providing the membrane module.
References [1] Stephenson T, Judd S, Jefferson B, Brindle K. Membrane bioreactors for wastewater treatment. London, UK: IWA Publishing; 2000. [2] Visvanathan C, Ben Aim R, Parameshwaran K. Membrane separation bioreactors for wastewater treatment. Crit Rev Environ Sci Technol 2000;30(1):1–48. [3] Arnot T, Howell J. Manipulation of cell physiology in membrane bioreactors. Proceedings of the MBR3—Third International Meeting on Membrane Bioreactors for Wastewater Treatment. Cranfield Univerisity, Cranfield, UK, 16 May 2001.
1807
[4] Chaize S, Huyard A. Membrane bioreactor on domestic wastewater treatment. Sludge production and modeling approach. Water Sci Technol (Kyoto) 1991;23: 1591–600. [5] Suwa Y, Suzuki T, Toyohara H, Yamagishi T, Urushigawa Y. Single-stage, single-sludge nitrogen removal by an activated sludge process with cross-flow filtration. Water Res 1992;26(9):1149–57. [6] Low EW, Chase HA. The effects of maintenance energy requirements on biomass production during wastewater treatment. Water Res 1999;33(3):847–53. [7] van Houten RT, Eikelboom DH High performance membrane bioreactors: A physiological approach. Proceedings of the MBR1—First International Meeting on Membrane Bioreactors for Wastewater Treatment Cranfield University, Cranfield, UK, 5 March 1997. [8] Pirt SJ. The maintenance energy of bacteria in growing cultures. Proc R Soc London 1965;163B:224–31. [9] Yamamoto K, Hiasa M, Mahmood T, Matsuo T. Direct solid-liquid separation using hollow fiber membrane in an activated sludge aeration tank. Water Sci Technol (Brighton) 1989;21:43–54. [10] Chiemchaisri C, Wong YK, Urase T, Yamamoto K. Organic stabilization and nitrogen removal in membrane separation bioreactor for domestic wastewater treatment. Water Sci Technol 1992;25(10):231–40. [11] Muller EB, Stouthamer AH, van Verseveld HW, Eikelboom DH. Aerobic domestic waste water treatment in a pilot plant with complete sludge retention by cross-flow filtration. Water Res 1995;29(4):1179–89. [12] Ben"ıtez J, Rodr"ıguez A, Malaver R. Stabilization and dewatering of wastewater using hollow fiber membranes. Water Res 1995;29(10):2281–6. [13] Rosenberger S, Kubin K, Kraume M. Operation of membrane bioreactors without sludge withdrawal. Proceedings of the Journees Internationales d’Etude du Cebedeau L’Eau et les Techniques Membranaires, Li"ege, 4–5 Avril 2000. [14] Rosenberger S, Witzig R, Manz W, Szewzyk U, Kraume M. Operation of different membrane bioreactors: experimental results and physiological state of the microorganisms. Water Sci Technol 2000;41(10–11):269–77. [15] Rosenberger S, Kruger . U, Witzig R, Manz W, Szewzyk U, Kraume M. Performance of a bioreactor with submerged membranes for aerobic treatment of municipal waste water. Water Res 2002;36:413–20. [16] Witzig R, Manz W, Rosenberger S, Kruger . U, Kraume M, Szewzyk U. Microbiological aspects of a bioreactor with submerged membranes for aerobic treatment of municipal wastewater. Water Res 2002;36:394–402. [17] Wagner J, Rosenwinkel K-H. Sludge production in membrane bioreactors under different conditions. Water Sci Technol 2000;41(10–11):251–8. [18] Lubbeke . S, Vogelpohl A, Dewjanin W. Wastewater treatment in a biological high-performance system with high biomass concentration. Water Res 1995;29(3): 793–802. [19] Eaton AD, Clesceri LS, Greenberg A, editors. Standard methods for the examination of water and wastewater. 19th ed.. Washington, DC: APHA, AWWA, WEF; 1995.
ARTICLE IN PRESS 1808
A. Pollice et al. / Water Research 38 (2004) 1799–1808
[20] Spanjers H, Vanrolleghem PA, Olsson G, Dold PL. Respirometry in control of the activated sludge process: principles. IAWQ Scientific and Technical Report no. 7. London, UK: IAWQ Publishing; 1998. [21] Smith S, Judd S, Stephenson T, Jefferson B. Membrane bioreactors—hybrid activated sludge or a new process?
Proceedings of the Fifth IMSTEC International Membrane Science and Technology Conference, UNSW, Sydney, Australia, 10–14 November 2003. [22] Metcalf A, Eddy A. Wastewater engineering: treatment, disposal and reuse. New York, SA: McGraw-Hill Int. Ed; 1991.