A novel process maximizing energy conservation potential of biological treatment: Super fast membrane bioreactor

Author’s Accepted Manuscript A novel process maximizing energy conservation potential of biological treatment: Super fast membrane bioreactor Seval Sözen, Senem Teksoy-Başaran, İpek Ergal, Cansu Karaca, Buşra Allı, Cansın Razbonyalı, Emine Ubay-Çokgör, Derin Orhon www.elsevier.com/locate/memsci

PII: DOI: Reference:

S0376-7388(17)32379-7 http://dx.doi.org/10.1016/j.memsci.2017.09.029 MEMSCI15567

To appear in: Journal of Membrane Science Received date: 18 August 2017 Accepted date: 9 September 2017 Cite this article as: Seval Sözen, Senem Teksoy-Başaran, İpek Ergal, Cansu Karaca, Buşra Allı, Cansın Razbonyalı, Emine Ubay-Çokgör and Derin Orhon, A novel process maximizing energy conservation potential of biological treatment: Super fast membrane bioreactor, Journal of Membrane Science, http://dx.doi.org/10.1016/j.memsci.2017.09.029 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

A novel process maximizing energy conservation potential of biological treatment: Super fast membrane bioreactor Seval Sözen1,2*, Senem Teksoy-Başaran1, İpek Ergal3, Cansu Karaca1, Buşra Allı2, Cansın Razbonyalı1, Emine Ubay-Çokgör1, Derin Orhon1,2 1

Faculty of Civil Engineering, Environmental Engineering Department, Istanbul Technical University, 34469 Maslak, Istanbul, Turkey 2 ENVIS Energy and Environmental Systems Research and Development Ltd., ITU ARI Technocity, Maslak, 34469 Istanbul, Turkey 3 Faculty of Science and Letters, Molecular Biology and Genetics Department, Istanbul Technical University, 34469 Maslak, Istanbul, Turkey *Corresponding Author, Mailing address: Istanbul Technical University, Civil Engineering Faculty, Department of Environmental Engineering, 34469 Maslak, Istanbul, Turkey. Phone: +90 212 2856544, Fax: +90 212 2856587 e-mail: [email protected] Abstract The study evaluated the microbial behavior of a submerged super fast membrane reactor (SFMBR) with settled sewage and a soluble synthetic substrate (peptone mixture) exhibiting similar biodegradation characteristics with sewage. A laboratory scale SFMBR was run at very low sludge retention times of 0.5-2.0 d and a hydraulic retention time of 8 h. Effluent COD was always lower than the soluble COD in the reactor and remained in the range 14-28 mg/L for the peptone mixture and 36-44 mg/L for settled sewage. Significant characteristics of SFMBR performance were assessed by particle size distribution analysis; respirometry; modeling defining related process kinetics; and molecular analysis, which revealed changes in microbial community composition under different operating conditions. Model simulation was also performed for raw sewage for revealing COD fractionation in the permeate and biomass in the reactor for the same operating conditions. The results indicated, as predicted, partial removal of the particulate slowly biodegradable COD fraction while soluble biodegradable COD components were almost totally removed. Modeling also highlighted energy conservation and recovery as the major feature of SFMBR, which was assessed to vary between 54-77%, a range significantly higher than what can be achieved with different activated sludge alternatives. Graphical Abstract

Energy Conservation (%) Extended Aeration… Conventional… Super Fast… Super Fast…

28% 34% 40% 45% 54% 63% 77%

Keywords: Super fast membrane bioreactor; settled sewage and peptone mixture; respirometry; particle size distribution analysis; modeling

1.

INTRODUCTION

Ingenious research during the early periods helped a great deal to understand microbial processes involved in substrate utilization in activated sludge systems 1. However, accumulated knowledge has not been properly translated into process design and technology, which still relies on gravity settling for separating treated effluent from the microbial cultures. However, they clearly indicated that empirically defined values for the sludge retention time (SRT) and hydraulic retention time, (HRT) resulted in oversized biological reactors: The major fraction of the activated sludge reactor was used to provide an additional volume for establishing a biomass structure with good biomass settling conditions. Conventional activated sludge (CAS) design also imposes upper limits for the level of biomass that may be kept in the reactor. In this respect, the membrane bioreactor (MBR) should be considered as a major break-through for the traditional activated sludge technology, as it replaced gravity settling by membrane filtration, and eliminated the constraints on SRT and biomass concentration for ensuring efficient settling of biomass 2, 3. The potential of MBR was initially interpreted as the ability to maintain markedly higher biomass concentrations at equally much higher SRT levels 4, 5, 6, 7. It was soon recognized that the total solids retention potential in MBR process offered an equally promising alternative for a stable system operation at substantially lower sludge retention times and biomass levels compared with conventional activated sludge process. A few studies were conducted at low SRTs in the range of 0.25-5 d with equally low hydraulic retention times of 3.0-6.0 h 8, 9. These studies, all reporting satisfactory performance with 95% COD removal, have not been recognized enough for practical applications. In this context, the Environmental Biotechnology Group at Istanbul Technical University developed the concept of the super fast membrane bioreactor process (SFMBR) operated in the extremely low SRT range of 0.5-2.0 d. This novel MBR application could best be understood and justified based on (i) COD fractionation and biodegradation of domestic sewage, and (ii) optimization of energy recovery in biological treatment. It is now well known that the COD content of domestic sewage mostly varies in the range of 450-600 mg/L 10; the particulate COD remains around 65% of the total COD, i.e. the COD fraction that is larger than 450 nm, an index used for defining suspended solids and biomass. Biodegradation studies reported that about 85% of the total COD is biodegradable, with a readily biodegradable fraction of around 10% (40-60 mg COD/L); the remaining part is defined as slowly biodegradable, undergoing a hydrolysis phase before microbial utilization 11, 12, 13. This basic information provide support for the SFMBR application, where the soluble COD components would be utilized, while the particulate COD would be partially and physically removed by entrapment and adsorption onto biomass, this way optimizing the energy recovery from sewage. Related research involved a sequence of studies, which started by exploring the fate of readily biodegradable substrate, the most critical COD fraction to test the performance limits of SFMBR.

The first study involved a lab-scale side-stream SFMBR operated at the SRT range of 0.5-2.0 d, using a synthetic substrate mixture, approximating the readily biodegradable COD fraction in domestic sewage 14. Substrate feeding was adjusted to 200 mg COD/L conservatively higher than the readily biodegradable fraction in sewage. The same study was repeated with acetate feeding of 250 mg COD/L 15. The potential of SFMBR system was later tested and confirmed with a lab-scale submerged system fed with the same synthetic substrate and acetate concentrations, operated at the same SRT range 16. In the next phase, the performance of the side-stream SFMBR was tested with a complex substrate using 250 mg COD/L of starch as a slowly biodegradable substrate. The following outlines the significant results commonly observed in these studies; (i) As given in Table 2 later in the text, the effluent (permeate) COD remained always below 20 mg COD/L, under all operating conditions. (ii) The total soluble COD was always higher than the permeate COD, and (iii) The effective filtration size was always significantly lower than the nominal pore size of the membrane. In this context, SFMBR was finally tested with real wastewater, settled sewage in this case - together with a complex synthetic substrate, peptone mixture. The study particularly focused on process performance; lower generation of soluble microbial products; process kinetics by model calibration; and changes in the structure of the microbial community under different operating conditions at different SRTs. Model evaluation was extended to cover and highlight significantly higher energy conservation potential as the major feature of SFMBR as compared to different activated sludge alternatives. 2.

MATERIALS & METHODS

2.1.

Experimental Approach

The experiments primarily focused on process performance and fate of the peptone/meat extract mixture (hereafter peptone mixture) and settled sewage separately used as sole organic carbon feed for the operation of SFMBR. Settled sewage was selected mainly to visualize real-case application of the novel process. Sewage was settled before being used as substrate, because process performance does not relate to particulate components, which accumulate in the reactor. Peptone mixture was adopted as a synthetic substrate due to its similarities with sewage based on COD fractionation and biodegradation characteristics 17, 18. A lab-scale submerged MBR unit was operated at steady-state at three different SRTs of 0.5, 1.0 and 2.0 d consecutively with two selected substrates. The hydraulic retention time was adjusted to 8.0 h due to restrictions of the experimental set-up. The influent COD concentrations were regulated to 200 mg/L and 250 mg/L for the peptone mixture and settled sludge respectively, to compare with previous studies carried out with SFMBR. A fill and draw reactor adjusted to an SRT of 2.0 d was run with the two substrates to obtain the initial biomass seed for the experiments. Steady-state conditions were checked and confirmed by daily measurements of COD and volatile suspended solids (VSS) in the reactor.

The study included all available, state-of-the-art experimental methods, serving the objective and scope of the study: Aside from monitoring reactor operation at steady state, it incorporated (i) particle size distribution (PSD) analysis, (ii) respirometric measurements for the assessment of relevant oxygen uptake rate (OUR) profiles; (iii) determination of kinetic and stoichiometric data by model calibration of the OUR profiles; (iv) molecular biology tools to assess the behavior and changes in the composition of microbial community and also (v) model assessment of the energy recovery capacity of SFMBR and other activated sludge modifications, all generating related data and serving a specific purpose in the study. Experimental program was designed and implemented in the following sequence: (i) Performance assessment of SFMBR operation under different conditions, emphasizing dissimilarity between the permeate COD and soluble COD in the reactor. (ii) Particle size distribution analysis revealing the role of membrane filtration and cake formation on retained COD in the reactor. (iii) Respirometric analyses and model calibration of OUR profiles for the numerical evaluation of process kinetics under different conditions. (iv) Molecular evaluation for detecting changes in the composition of the microbial culture under different operating conditions for possible explanation of corresponding process kinetics. Experimental data were compared with modeling outputs simulating expected removal performance of SFMBR operation under different conditions, in terms of major COD fractions for the tested substrates, i.e. expected COD fractionation of permeate COD and particulate COD components in the reactor. Finally, model evaluation was also performed to determine the energy recovery potential of SFMBR, when used for treating raw sewage in a real life scenario. 2.2.

Reactor Setup

The study was conducted by running a lab-scale submerged MBR that has been also used for different previous studies 16, 19, 20. The system consisted of a Plexiglas reactor with an operating volume of 3 L. The system was equipped with hallow fiber Zee Weed*1 (GE) membrane module with a nominal pore size of 0.04 μm and total membrane surface area of 0.1 m2. The technical properties of Zee Weed*1 ultra filtration membrane are listed in Table 1.

Table 1. Technical properties of Zee Weed*1 ultra filtration membrane Module Type Nominal Membrane Surface Area Module Dimensions Height Diameter Membrane Properties Material Nominal Pore Size Surface Properties Fiber Diameter Flow Path Operating Specifications TMP Range Max. Operating Temperature Operating pH Range Cleaning Specifications Max. Cleaning Temperature Cleaning pH Range Max. Cl2 Concentration

Property 1 ft2 (0.1 m2) 175 mm 56 mm PVDF 0.04 μm Non-ionic & hydrophilic 1.9 mm OD/ 0.8 mm ID Outside-in -55 to 55 kPa (-8 to 8 psi) 40°C (104°F) 5.0-9.5 40°C (104°F) 2.0-10.5 1.000 ppm

In order to control and measure the fundamental operational variables such as dissolved oxygen (DO), trans membrane pressure (TMP), pH and temperature, submerged MBR was automatically controlled with a Programmable Logic Controller (PLC) as shown in Figure 1. The equipment and the operating conditions of the lab scale submerged MBR unit were described in detail by Sözen et al. 16.

Figure 1. Process flow scheme of the lab scale MBR Two different substrates, domestic wastewater and peptone mixture, were tested to investigate the treatment efficiency of complex substrates in submerged MBR systems. The peptone mixture was prepared by dissolving 16 g peptone, 11 g meat extract, 3 g urea, 0.7 g NaCl, 0.4 g CaCl2.2H2O, 0.2 g MgSO4.7H2O and 2.8 g K2HPO4 in 1 L distilled water. The synthetic sewage was prepared daily and fed to

the reactor without causing any change in operation temperature and conditions. The domestic wastewater was taken from ISKI Baltalimanı Wastewater Treatment Plant, in Istanbul, nearby the laboratory facility. The SFMBR operation with settled sewage lasted about two months. In this period SFMBR was fed from a pool of settled sewage inside the laboratory. Weekly, the pool was replenished with fresh settled sewage. The COD variation in the fresh portion was always below ± 8%. The effect of this variation on the COD level in the pool remained always below ± 5%. The temperature of the feed was never changed (between 18oC and 22oC). 2.3.

Particle size distribution analysis

Particle size distribution analyses were conducted with the methodology defined by Dülekgürgen et al. 21, using a sequential filtration/ultrafiltration cell. The experiments were run in a 400 ml Millipore Amicon 8400, USA stirred cell with an effective membrane area of 41.8 m2 at a batch mode 22. The filtration procedure, sequence and consecutive filtration norms used in the experiments were described in detail by Dülekgürgen et al. 21. COD and DOC measurements were conducted on the collected permeates and a correlation was created between PSD and DOC measurements, which was converted to COD for each filtration/ultrafiltration step. 2.4.

Respirometric Analysis

The oxygen uptake rate (OUR) in the biochemical processes, is observed as a timely change in the dissolved oxygen concentration. It is an overall process rate incorporating the collective effect of all oxygen/energy consuming reactions. OUR profiles were obtained by using the respirometric method defined by Ekama et al. 23 with a Manotherm RA-1000 The adopted procedure and the implemented testing conditions were described in detail elsewhere 17, 18. 2.5.

Model structure

System kinetics was investigated using a model that would numerically interpret biodegradation of different COD fractions in the peptone mixture and settled sewage and convert it into meaningful and applicable kinetic and stoichiometric parameters. For this purpose, a model having the basic template of ASM 1 24 was adopted, modified to include: (i) Endogenous respiration as used in many modeling exercises 6; (ii) dual hydrolysis concept differentiating readily and slowly hydrolysable fractions 25 and (iii) generation of microbial products through endogenous respiration related processes 26. Consequently, the model included seven model components, namely readily biodegradable COD, SS; readily hydrolysable COD, SH (SH1 for peptone mixture); slowly hydrolysable COD (SH2 for peptone mixture and XS for settled sewage); soluble microbial products, SP; heterotrophic active biomass, XH; particulate microbial products, XP and dissolved oxygen, SO, which is the basic component in respirometric evaluations. Accordingly, the model was based on four processes; microbial growth on SS; hydrolysis of SH1 or SH; hydrolysis of SH2 or XS

and endogenous respiration of XH. Table 4 outlines related process kinetics and stoichiometry in the generally accepted matrix format reflecting all stoichiometric numerical relations between model component and processes. The matrix indicates that formation of soluble microbial products, a process of outmost importance for membrane system, is kinetically defined as follows: (1) where, fES is the soluble fraction released from endogenous biomass and bH is the endogenous decay coefficient. The methodology defined in the AQUASIM program 27 was used in model calibration. The iterative simulation of the OUR profiles, is explained in detail by Pala Ozkok et al. 28. The results of model calibration were checked by means of identifiability analysis: In fact, related literature has reported evidence that OUR profiles yield identifiable model coefficients used for the kinetics of organic carbon removal, as in this study. Conclusive results have been presented by Insel et al. 13: Basically, OUR is a biodegradation fingerprint and different parts of the OUR profile exhibit variable sensitivity for different model coefficients; so, each coefficient is determined in the corresponding OUR region, where the profile is most sensitive. Recent studies also give details of the parameter identifiability procedure in similar evaluations. The procedure basically involves the use of the UNCSIM module proposed by Brun et al. 29. 2.6

Molecular Analysis

Changes in the composition of the microbial culture for SFMBR operation under different conditions, i.e. the effect of substrate and SRT, were evaluated applying denaturing gradient gel electrophoresis (DGGE) to identify significant indicators revealing variations in the community structure. The studies included DNA extraction, polymerase chain reaction (PRC) amplification, denaturing gradient gel electrophoresis, before analyzing DGGE configuration. DGGE also involved similarity analysis based on clustering of the band patterns. Detailed information on related analytical methodology can be found elsewhere 14, 16.

Table 4. Matrix representation of the selected model structure Compo nents→ SI

XI

SS

SH

Process

X

SO

S

X

S

P

P

XH

Rate Equations

es↓ Growth

1

of XH Hydrolys is of SH1

1

̂

⁄

-1

⁄

(or SH) Hydrolys is of SH2

1

⁄

-1

⁄

(or XS)

Decay

-1

of XH

Paramet ers

C O D

C O D

C O D

C O D

C O D

O2

C O D

C O D

cel l C O D

3. EXPERIMENTAL RESULTS 3.1. System performance The performance of SFMBR was evaluated by simultaneously monitoring both the effluent/permeate COD, SE and the soluble COD, ST contained in the reactor volume under different steady-state operating conditions. The operation was implemented for the two different substrates, peptone mixture and settled sewage. System performance was tested for each substrate when the SFMBR system was continuously run at three different SRTs of 0.5, 1.0 and 2.0 d. The monitoring covered a period of not less than two weeks during steady-state conditions largely exceeding the selected SRT. The peptone mixture was selected as a substrate, mainly because it is totally soluble and exhibits the same characteristics as sewage in terms of COD fractions, i.e. readily biodegradable COD, SS; readily hydrolysable COD, SH1; slowly hydrolysable COD, SH2 and no inert fractions. The kinetic parameters related to its biodegradation

were assessed in many similar studies 18, 28, 30. Settled sewage was selected as the second substrate since the study is mainly focused on the fate of soluble organics as the system performance is concerned. The influent concentration of the peptone mixture was adjusted to 200 mg COD/L and settled sewage to around 250 mg/L to remain in agreement with previous studies on the subject 19, 31. During the steady-state periods of SFMBR operation with peptone mixture, biomass remained stable around 100 mg VSS/L; 194 mg VSS/L and 330 mg VSS/L at sludge retention times of 0.5, 1.0 and 2.0 d respectively; For the settled sewage, the same biomass values were measured as 166 mg VSS/L; 307 mg VSS/L and 482 mg VSS/L. As outlined in Table 2, when the peptone mixture was used as substrate, the average effluent COD; SE was stabilized in limited range of 14-28 mg COD/L in all experimental conditions. For settled sewage, a higher SE range was observed between 36-44 mg COD/L, mainly because the permeate stream also included the initial soluble inert COD, SI, which bypasses the membrane. A significant feature of the monitoring program was the magnitude of the average soluble COD, ST, simultaneously measured in the reactor volume, which was always recorded significantly higher than the corresponding effluent COD. It should be noted that the The table included a total of 12 COD values, five around 50 mg COD/L of above, five around 30 mg COD/L of above and only two around 15 mg COD/L. As elaborated in detail elsewhere 32, it was statistically proven that the laboratory facilities used in the studies could measure CODs around 30 mg/L with 3% accuracy and around 1015 mg/L with 10-15% accuracy.

Table 2. SFMBR performance under different operation conditions Substrate Feeding 200 mg/L peptone mixture 250 ± 12 mg/L settled sewage

SRT (d) 2.0 1.0 0.5 2.0 1.0 0.5

Soluble COD in MBR ST (mg/L) 56 ± 4.6 35 ± 2.9 33 ± 2.3 65 ± 7.2 53 ± 4.0 49 ± 3.2

Permeate COD SE (mg/L) 14 ± 1.0 15 ± 1.0 28 ± 1.8 36 ± 2.0 39 ± 2.2 44 ± 2.6

3.2. Particle Size Distribution Analysis Particle size distribution (PSD) analysis was conducted to clarify the significant difference observed between the magnitude of soluble COD in the reactor and the effluent/permeate COD. For this purpose, representative samples were taken during the periods of steady-state operation of SFMBR fed with settled sewage, sustained at different SRTs of 2.0 and 1.0 d. The samples were filtered through 450 nm for separating the biomass in the reactor from what is conventionally called as soluble COD, ST. The filtrates were subjected to PSD analysis, which were carried out by

measuring the dissolved organic carbon (DOC), in successive filtration steps, mainly because the collected volumes from each filtration were to small for accurate COD measurements. Consequently, the PSD analysis was started first, by measuring COD of the filtered sample, i.e. the total soluble COD, ST, and the corresponding DOC to establish the COD/DOC ratio; it was continued with DOC measurements using the same ratio: The DOC of the sample representing SFMBR operation at SRT of 2.0 d was 16.1 mg/L and ST was measured as 57 mg COD/L corresponding to a COD/DOC ratio of 3.55. This ratio is reported in the range of 3.0-3.64 in few studies involving simultaneous COD and DOC measurements 20, 33, 34, 35. Table 3 outlines the cumulative DOC values measured after each filtration step, defining the DOC content of the sample below the corresponding filter size. The same table also gives the differential DOC between two consecutive filter size fractions. Values in the table indicate three size fractions with higher DOC values as compared to others, one between 45-220 nm, the other between 8-13 nm and the third below 2 nm. The different DOC fractions were also converted to corresponding COD levels using the initial COD/DOC ratio.

Table 3. Particle size distribution of DOC in the SFMBR bulk liquid under different operating conditions Separation Technique

Particle Size (nm)

Filtration HV Filter 450 GV Filter 220 Ultrafiltration 100kDa 13 30kDa 8 10kDa 5 3kDa 3 1kDa 2

Cumulative DOC (mg/L) SRT SRT 2.0 d 1.0 d 16.1 12.4

13.3 12.1

10.1 8.7 5.6 4.1 3.5

11.0 8.8 5.7 4.3 3.8

Size Category (nm)

Incremental DOC (mg/L) SRT SRT 2.0 d 1.0 d

Incremental COD (mg/L) SRT SRT 2.0 d 1.0 d

220-450

3.7

1.2

13.1

4.3

13-220 8-13 5-8 3-5 2-3 <2

2.3 1.4 3.1 1.5 0.6 3.5

1.1 2.2 3.1 1.4 0.5 3.8

8.1 5.0 11.0 5.3 2.1 12.4

3.9 7.8 11.0 5.0 1.8 13.5

Table 3 includes a similar PSD analysis connected on the sample characterizing SFMBR operation at SRT of 1.0 d. The initial DOC and ST values were 13.3 and 47 mg COD/L respectively giving a slightly lower COD/DOC ratio of 3.53. The

incremental DOC exhibits a slightly different profile with higher fractions between 5-8 nm and again below 2 nm. At this point it should be remembered that the SFMBR unit has a membrane with filter size of 40 nm. In conventional activated sludge system coupled with a settling tank, the soluble COD in the reactor is bound to be the same as the effluent soluble COD. However, during the days where the samples were processed, the permeate COD was 35 mg/L (9.85 mg DOC/L) for the SRT of 2.0 d compared to an ST value of 57 mg/L in the reactor: Similarly, the ST value of 47 mg/L for the SRT of 1.0 d needs to be compared with the corresponding permeate COD of 38 mg/L (10.7 mg DOC/L). These results agree well with similar significant COD differences between reactor soluble COD and permeate COD observed during continuous operation of SFMBR under different conditions. The PSD analysis serves as a useful tool to uncover and clarify these observations: Figure 2 gives the cumulative PSD profiles of DOC for the two samples showing that the effective filter size of the membrane was reduced below 13 nm i.e. between 8-13 nm, during SFMBR operation at SRTs of 2.0 d and 1.0 d. As also reported in previous studies 36, 37, 38, these observations are easily explained by the formation of an additional cake filter layer on the membrane, which entraps soluble organics much smaller than its actual pore size. This way, organics larger than the effective filter size of the membrane remain entrapped in the reactor and accumulate at a similar rate as biomass as a function of the ratio between the SRT and the hydraulic retention time.

20

DOC (mg/L)

15

10

5

0 450

220

13

8 Particle Size (nm)

5

3

2

5

3

2

(a) SRT of 2.0 d 14 12 DOC (mg/L)

10 8 6 4 2 0 450

220

13

8 Particle Size (nm)

(b) SRT of 1.0 d Figure 2. Cumulative size distribution of DOC in the SFMBR bulk liquid for settled sewage for (a) SRT of 2.0 d; (b) SRT of 1 3.3. System kinetics Model calibration was carried out against three different OUR profiles, two for the peptone mixture (SRT of 1.0 and 0.5 d samples) and one for the settled sewage (SRT of 0.5 d sample). The other OUR sets could not be used for calibration, due inconsistent nature of the corresponding respirometric data. It should be underlined that the procedure did not involve arbitrary selection of model coefficients. The calibration exercise yielded best fits with the experimental OUR profiles as shown in Figures 3 and 4, enabling to define the valid values for model coefficients defining process kinetics and stoichiometry. The initial COD value of the peptone mixture used in the respirometric experiment was selected higher than the level used for reactor operation, to better interpret the OUR profiles.

90 80

OUR (mg O2/L.hr)

70 60 50 40 30 20 10 0 0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

Time (d) Data

Model

(a) SRT of 0.5 d 70

OUR (mg O2/L.hr)

60 50 40 30 20 10 0 0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Time (d) Data

Model

(b) SRT of 1.0 d Figure 3. Model calibration of the OUR profiles for the peptone mixture (a) SRT of 0.5 d; (b) SRT of 1.0 d

45

OUR (mg O2/L.hr)

40 35 30 25 20 15 10 5 0 0

0.05

0.1

0.15

0.2

0.25

0.3

Time (d) Data

Model

Figure 4. Model calibration of the OUR profile for settled sludge – SRT of 0.5 d Identified values of parameters related to each model component for the two substrates are outlined in Table 5. For both substrates, the table also includes results of other studies conducted at a SRT of 10 d, using the same peptone mixture and sewage as the sole organic substrate 30, 39. First, results indicate that different processes cannot be defined with the same model coefficient for different SRTs, an observation that confirms the validity of variable kinetics with respect to culture history. In fact, the growth kinetics indicated a slower microbial growth for higher SRTs, where the maximum specific growth rate, ̂ decreased from 8.7/d at SRT of 0.5 d to 7.7/d at SRT of 1.0 d and finally to 6.8/d at SRT of 10 d. A supporting change was also observed in the magnitude of the half saturation coefficient, K S changing from 13 mg/L to 24 mg/L. Similarly, ̂ determined as 6/d for the settled sludge at SRT of 0.5 d is significantly higher than the range of 3.5-4.2/d estimated for treatment plants operated at much higher SRTs for the treatment domestic sludge in the same city 39. Second, the endogenous decay coefficient, bH was determined as 0.3/d for all samples, an unusually high level compared to the 0.10-0.20/d bracket commonly adopted for most wastewaters. It is well known that maintenance energy finds its expression in endogenous respiration in activated sludge models; consequently, a b H value of 0.3/d reflects high maintenance energy requirement of the microbial community sustained at extremely low SRTs, presumably due to faster cellular reactions to maintain the necessary substrate/material gradients across the membrane and macromolecular turnover 40. A similar result was also observed as the response of microbial community to antibiotic inhibition 41, 42. Model calibration also defines respective COD fractionation for substrates used in respirometric analysis as state variables. As given in Table 5, it is noteworthy to mention that the relative magnitudes of different COD fractions remain basically the same, regardless of different process kinetics and microbial communities sustained at different SRTs.

Table 5. Results of model calibration for the biodegradation of peptone mixture and settle sewage Model Parameters & State Variables Maximum growth rate for XH Half saturation constant for growth of XH Endogenous decay rate for XH Maximum hydrolysis rate for SH (SH1) Hydrolysis half saturation constant for SH Maximum hydrolysis rate for XS (SH2) Hydrolysis half saturation constant for XS Yield coefficient for XH Fraction of biomass converted to SP Fraction of biomass converted to XP State Variables Initial active heterotrophic biomass Active biomass ratio Initial total COD Initial biodegradable COD Initial readily biodegradable COD Initial readily hydrolizable COD Initial slowly hydrolizable COD

Unit

̂

Peptone Mixture SRT (d) 10 0.5 1.0 Katipoğlu This This Yazan et Study Study al. 30

Settled Sewage SRT (d) 0.5 10 This Study

Okutman et al. 39

1/d

6.8

8.7

7.7

6.0

3.5

KS

mg COD/L

24

16

13

10

5.0

bH

1/d

0.1

0.3

0.3

0.3

0.2

khs

1/d

6.3

7.6

3.5

3.1

2.8

KX

g COD/ g COD

0.2

0.12

0.044

0.2

0.12

khx

1/d

0.5

2.8

1.45

1.2

1.35

KXX

g COD/ g COD

0.01

0.07

0.009

0.5

0.14

YH

g COD/ g COD

0.6

0.58

0.58

0.58

0.64

fES

-

0.05

0.05

0.05

0.05

0.05

fEX

-

0.15

0.15

0.15

0.15

0.15

XH1

mg COD/L

850

335

430

300

fa

%

84

80

82

67

CT1

mg COD/L

12

505

205

250

CS1

mg COD/L

506

505

205

225

SS1

mg COD/L

43

48

20

38

SH1

mg COD/L

254

283

114

72

XS1

mg COD/L

158

174

71

115

406

3.4. Molecular Analysis Molecular analyses based on PRC-DGGE methodology were carried out to detect changes in the structure of the microbial communities sustained during SFMBR operation at different SRTs. These changes, if any, would relate to and explain parallel changes in the process kinetics. It is well known that systems operating at high SRT levels would enrich a slow growing microbial community capable of consuming macro-molecules such as polysaccharides, proteins, etc. 4. However, much fever experimental evidence exists on the effect of short SRTs on community structure: A few studies reported higher microbial diversity as SRT increased from 34.5 to 10 d 43, 44. Conversely, Saikaly and Oerther 45 provided theoretical proof that species diversity was augmented at SRTs in the range of 2.28-5.56 d, as compared to longer SRTs. Saikaly et al. 46 showed that sequencing batch reactor operation at SRT of 2.0 d suggested higher microbial diversity compared to SRT of 8.0 d.

100

80

60

Molecular community analyses were conducted on six biomass samples taken from the reactor operated at steady-state, three when peptone mixture was used as substrate (P0.5, P1 and P2) and the others where SFMBR was fed with settled sewage (S0.5, S1 and S2). Several (more than five) replicates of DGGE profiles were evaluated and exhibited high reproducibility. Figure 5 illustrates a representative dendogram. The similarity (Dice) coefficient, based on clustering of the profiles according to pair Dice (Tol 1.0%-1.0%) (H>0.0% S>0.0%) [0.0%-100.0%] wise similarities, was used as the major parameter for determining changes in the cevre peptone cevre peptone microbial community structure induced by SFMBR operation at different SRTs and substrates.

P0.5 100 P2 P1 P4

50.0

100

P2 P6

55.2

36.4

100

S2 T6

37.5

50.0

60.6

100

S0.5 41.2 T2

47.4

45.7

52.6

100

S1 T4

50.0

54.1

65.0

71.4

50.0

Figure 5. DGGE profiles of the samples with different SRT values, clustering of the profiles according to pairwise similarities was obtained using Dice's coefficient and UPGMA (Scale bar represents % similarity) The band patterns in the DGGE profiles in Figure 5, when properly processed, provide significant numerical indicators on the effect of substrate and SRT: First, although peptone mixture and sewage exhibit similar characteristics related to COD fractionation and biodegradation kinetics of different fractions, they sustain microbial communities with significant different composition: In fact, the dissimilarities of the

100

biomass with peptone mixture and sewage was 58.8% at SRT of 0.5 d, 50% at SRT of 1.0 d and 39.4% at SRT of 2.0 d, which clearly shows that dissimilarity increases for lower SRTs. Second, compatible dissimilarities were detected for the same substrate when the SRT of SFMBR operation was changed. As outlined in Table 6 dissimilarities range between 44.8% and 63.6% for the peptone mixture and 28.6% and 47.4% for the settled sludge, but they do not follow a distinct pattern: Higher dissimilarities were calculated as 63.6% between SRTs of 1.0 and 2.0 d for the peptone mixture and as 47.4% between SRT of 0.5 and 2.0 d for settled sewage. Table 6. Dissimilarities in the composition of microbial community for different SRTs Samples Peptone SRT 0.5 d vs 1 d SRT 1 d vs 2 d SRT 0.5 d vs 2 d Settled Sewage SRT 0.5 d vs 1 d SRT 1 d vs 2 d SRT 0.5 d vs 2 d

Dissimilarity (%) 50.0 63.6 44.8 28.6 35.0 47.4

Evaluation of results also yielded the Shannon diversity index, which accounts both for the diversity and abundance offering more evidence about community structure than species richness, i.e. the number of detected different species. Table 7 outlines the relationship between SRT and diversity for each substrate, which indicates no pattern with respect to SRT, as the diversity index for SRT 1.0 d was 1.26±0.004 and 1.31±0.005 for peptone mixture and sewage respectively, higher than indexes characterizing other STRs. The same was noticed for species richness, also indicating the effect of substrate, where more bands were detected with sewage compared to peptone mixture. Sewage has a more complex composition with a much wider array of different organic compounds as compared to the peptone mixture, which makes it more suitable for sustaining a microbial culture with higher diversity and richness 47. Table 7. Shannon bacterial diversity index, and species richness for microbial cultures sustained in SFMBR under different conditions Sample P0.5 (Peptone SRT 0.5 d) P1 (Peptone SRT 1 d) P2 (Peptone SRT 2 d) S0.5 (Settled sewage SRT 0.5 d) S1 (Settled sewage SRT 1 d) S2 (Settled sewage SRT 2 d)

Shannon Diversity Index (H) 1.129 1.264 0.979 1.271 1.315 1.227

Species Richness (S) 18 20 10 20 22 18

4. DISCUSSION 4.1. System performance The performance of SFMBR in the removal of the peptone mixture and settled sewage in this study should be evaluated within the framework of the performance profile observed in previous SFMBR studies of the Environmental Biotechnology group at ITU, conducted with different substrates, namely synthetic readily biodegradable mixture, acetate and starch, using both side-stream and submerged reactors operated under different conditions. Table 8 outlines major parameters related to these studies. Significant observations derived from the table may be outlined as follows: (i) Substrate feeding was adjusted to the narrow range of 200250 mg COD/L, mainly to remain compatible with the soluble portion of domestic sewage and most industrial wastewaters. (ii) The effluent/permeate COD, S E, was always below 20 mg COD/L and within the range of 12-19 mg COD/L for various substrates and different SRTs of 0.5-2.0 d. (iii) Soluble COD in the reactor, ST, was always higher than SE, due to cake formation on the membrane operating at an effective filtration size much lower than its nominal pore size. This mechanism was well ascertained as in this study, by means of particle size distribution (PSD) analysis. (iv) Retained portion of COD in the reactor, STR, was calculated to remain in the range of 3.3-7.0 mg COD/L in this study, which accumulated in the reactor the same way as biomass. (v) Even at high loading at 1000 mg COD/L, SE remained at 21 mg COD/L for acetate and between 45-56 mg COD/L when the substrate in the feed was switched to synthetic readily biodegradable mixture. (vi) S E was mostly, if not entirely, composed of soluble microbial products, SP, clearly demonstrated by acetate experiments. The results of this study, which are also included in Table 8, exhibited similar features; the permeate COD, SE of SFMBR operation with peptone mixture also remained below 20 mg COD/L, except for SRT of 0.5 d, which yielded 28 mg COD/L. SE of SFMBR operation with settled sewage was in the range of 36-44 mg COD/L, due to initial soluble inert COD content of settled sewage, which bypasses the membrane. In short, the permeate flow of SFMBR operated with settled sewage contained both the initial inert (SI1) and the soluble residual microbial products (SP).

Table 8. Performance profile of SFMBR operated under different conditions

MBR Type

Side-stream

Side-stream

Submerged

Submerged

Submerged

Substrate Feeding

200 mg/L synthetic readily biodegredable mixture

200 mg/L acetate 255 mg/L synthetic 235 mg/L biodegradable mixture 1000 mg/L synthetic readily biodegradable mixture 250 mg/L acetate 1000 mg/L acetate

Side-stream

250 mg/L starch

Submerged

200 mg/L peptone mixture

Submerged

250 mg/L settled sewage

*

SRT (d)

HRT (h)

2.0 1.0 0.5

2.0

2.0 1.0 0.5

1.0

2.0 1.0 0.5 2.0 1.0 0.5 2.0 1.0

0.5

1.0

8.0

0.5 2.0 1.0

8.0

0.5 1.0 1.0 2.0 1.0 0.5 2.0 1.0 0.5 2.0 1.0 0.5

8.0 1.0

8.0

8.0

Soluble COD in MBR ST (mg/L) 67 55 35

12 17 19

Retained COD in MBR STR* (mg/L) 2.30 3.20 0.20

78 72 43

11 14 17

2.80 2.40 2.20

77 56 47 73 58 49 37 36

22 25 25 12 15 17 20 10

0.57 0.65 0.92 1.27 1.79 2.67 2.83 8.67

23

15

5.33

100 65

56 45

7.33 6.67

59

50

6.00

39 60 78 43 40 56 35 33 65 53 49

20 21 14 15 18 14 15 28 36 39 44

6.33 13.0 1.33 1.17 1.83 7.0 6.7 3.3 4.8 4.6 3.3

Permeate COD SE (mg/L)

References

12

13

14

28

33 30

This study

This study

⁄

4.2. Variable kinetics due to microbial community change It should be noted that, activated sludge models, such as the one adopted for this study, while accounting for substantial refinements defined for substrate (COD) fractions with different biodegradation kinetics, operated with a single model component for biomass for organic carbon removal, the active heterotrophic biomass, XH in this case. Obviously, this is a gross simplification since XH represents a given microbial community composition for selected operating conditions; these conditions and especially SRT determines the culture history for XH. Studies indicate that process kinetics observed under different culture history (SRT) also exhibit significant changes 28. It is important to determine the factors leading to variable process kinetics, which may be related to (i) different metabolic functions of the same microbial structure, or (ii) changes inflicted on the composition of the microbial

culture. It is true that in a multi-component microbial community, such as activated sludge, process kinetics relate to the overall response of the microbial culture, which is bound to be the major indication of the kinetic characteristics of the predominant components in the microbial culture. This way, a shift in the predominant microbial groups could be the main reason for observing variable kinetics 48. The results of the molecular analysis of this study provide a solid support to this argument, as they yield significant dissimilarities of microbial community composition sustained for the same substrate, even for the narrow change of SRT between 0.52.0 d, namely, 45 to 63% for the peptone mixture and 28 to 47% for settled sewage. Respective values of diversity index (H) and species richness (S) listed in Table 7 also offer additional support for variable kinetics derived from respirometric analyses. Kinetic parameters listed in Table 5 always indicate faster growth kinetics for lower SRT values, confirming selection of species equipped with a metabolic machinery allowing faster growth: These species establish themselves as the dominant fractions of the microbial community under selected operating conditions and impose their kinetic characteristics to SFMBR operation. Consequently, the system exhibits variable kinetics as a function of operating conditions. 4.3. Performance modeling At this stage, steady-state performance of SFMBR operation under different conditions was simulated using the same model structure. The simulation was conducted for the actual influent COD levels of 200 mg COD/L for the peptone mixture and 250 mg COD/L for the settled sewage, based on kinetic parameters and relative magnitude of COD fractions as outlined in Table 5; model evaluation was also carried out for SRT of 4.0 d, characterizing high rate activated sludge, for the purpose of comparison. Results of the simulation for the peptone mixture are illustrated in Figure 6, both for soluble and particulate COD components. The simulation yielded permeate total COD levels of 34, 24 and 21 mg COD/L for SRTs of 0.5, 1.0 and 2.0 d, respectively. Simulation results for settled sewage indicated, as shown in Figure 7, permeate total COD levels of 42, 35, 33 mg COD/L, similarly for SRTs of 0.5, 1.0 and 2.0 d, respectively. As previously mentioned, the difference between peptone mixture and settled sewage is due to the initial inert soluble COD of approximately 18 mg COD/L, which by passes the membrane. It should be noted that, (i) the predicted results by model simulation agree well with the measured permeate COD values given in Table 2. The differences are quite within the analytical precision of COD determinations in this low range; (ii) the experimentally observed results stay systematically lower than model predictions. This issue may be further explored by checking how the generation of soluble microbial products, SP, is expressed in the model structure: In fact, the mass balance for SP accounting for the rate expression (1) may be formulated as follows:

(2) and, (3) where,

⁄ ; VR, the reactor volume and Q, the flow rate.

Using the following relationship: (4) Substituting (

) in expression (3): (5)

In SFMBR systems requiring high maintenance energy, bH was actually determined as 0.3/d, higher compared to conventional activated sludge system. Similarly, the endogenous fraction of biomass fES may be greater than what was accepted in model simulation, resulting in slightly higher SP values and effluent COD levels in experimental operation. As in previous parts of the study, membrane fouling was not experienced during SFMBR operation under different conditions. Similarly, model simulation supported by mass balance indicates significantly lower SP generation, mostly below 3-5 mg COD/L, as compared to MBR operation at high SRT levels. This should be considered as valuable virtue of SFMBR systems for alleviating expected fouling problems.

Concentration (mg COD/L)

40 35 30 25 20 15 10 5 0 0.5

1

1.5

ST

SS

2

2.5 SRT (d)

SH1

3

SI

SP

3.5

4

SH2

(a) effluent/permeate COD components

Concentration (mg COD/L)

800 700 600 500 400 300 200 100 0 0.5

1

1.5

2

XT

2.5 SRT (d) XH

3

3.5

4

XP

(b) particular COD components Figure 6. Simulation of the steady-state performance of SFMBR fed with peptone mixture, in terms of (a) effluent/permeate COD components; (b) particular COD components in the reactor volume

Concentration (mg COD/L)

45 40 35 30 25 20 15 10 5 0 0.5

1

1.5

ST

2

2.5 SRT (d)

SS

SH

3

SI

3.5

4

3.5

4

SP

(a) effluent/permeate COD components

Concentration (mg COD/L)

1200 1000 800 600 400 200 0 0.5

1

1.5

XT

2

XH

2.5 SRT (d) XS

XP

3

XI

(b) particular COD components Figure 7. Simulation of the steady-state performance of SFMBR fed with settled sewage, in terms of (a) effluent/permeate COD components; (b) particular COD components in the reactor volume (XT = total biomass as cell COD/L) 4.4. Energy Conservation Evaluation of experimental results should also be extended to cover energy conservation potential of SFMBR, aside from effective system performance. It is well known that the calorific value of the organic content of sewage is around 3300 kcal/kg COD 49, 50, i.e. the sewage equivalent of 6 to 10 persons/d. Conventional biological treatment is basically focused on consuming this energy by COD removal, at the expense of additional energy input for aeration. Conventionally, a small fraction of this energy is conserved in the biomass generated through biological removal of

organic matter. SFMBR operated at extremely low SRT levels has the unique potential of maximizing recoverable energy in biomass. This potential is also explored by model simulation of process performance for the treatment of raw sewage. Model simulation used the appropriate characteristics of sewage, adopted the model coefficients derived in the study and implemented the same model to simulate the entire range of MBRs and activated sludge modifications to calculate sludge and energy recovery. In fact, characteristics and COD fractionation of raw sewage were selected to remain compatible with that of settled sewage used in the experiments, as given in Table 9. Model simulation adopted the kinetic parameters presented in Table 5, providing experimental support for the results and representing Istanbul sewage 39. Results of model simulation are illustrated in Figure 8: As predicted, they essentially showed partial removal of the particulate slowly biodegradable COD fraction, while soluble biodegradable COD components were almost totally removed.

Table 9. COD fractionation of raw sewage adopted for model simulation 39, 51 COD Fractions Total COD, CT Total biodegradable COD, CS Readily biodegradable COS, SS Readily hydrolysable COD, SH Slowly hydrolysable COD, XS Soluble inert COD, SI Particulate inert COD, XI

Magnitude mg/L % 425 100 365 86 40 9.4 70 16.5 255 60 18 4.2 42 10

Concentration (mg COD/L)

30 25 20 15 10 5 0 0.5

1

1.5

ST

2

2.5 SRT (d)

SS

SH

3

SI

3.5

4

3.5

4

SP

(a) effluent/permeate COD components

Concentration (mg COD/L)

2500 2000 1500 1000 500 0 0.5

1

1.5

XT

2

XH

2.5 SRT (d) XS

XP

3

XI

(b) particular COD components Figure 8. Simulation of the steady-state performance of SFMBR fed with raw sewage, in terms of (a) effluent/permeate COD components; (b) particular COD components in the reactor volume Evaluations for energy conservation were not limited to SFMBR, but also covered high rate, conventional, nutrient removal and extended aeration activated sludge systems for benchmarking: They were based on calculating production rates of respective biomass (particulate COD) components per unit flow rate or, in simpler terms, concentrations of these fractions expressed in terms of removal rate of influent biodegradable COD concentration. In fact, basic mass balance allows to express: (6)

(7) and, (8) or, (9) where, Q is flow rate, YNH, the net heterotrophic yield, fEX, particulate residue of endogenous biomass. pXS fraction is calculated by modeling, and (10) The energy conservation potential of the system may then be calculated as: (11) Table 10 outlines achievable energy conservation of SFMBR and different activated sludge alternatives together with the fate of soluble and particulate COD fractions used in the relevant calculations. It clearly shows the ability of SFMBR to conserve influent energy by 54-77%. The level of energy conservation of 63% achievable when SFMBR is operated at an SRT level of 1.0 d is calculated to be 1.4 times higher than the high rate activated sludge process; 1.6 times higher than the conventional activated sludge process and 2.3 times higher than the extended aeration process.

Table 10. Fate of soluble and particulate COD components and energy recovery in different activated sludge alternatives SS+ SH (mg/ L)

pXS (m g/L )

pXI (m g/L )

pXP (m g/L )

ΔCS * (mg /L)

YNH (mg COD/mg COD)

pXH (m g/L )

pXT* * (mg /L)

Energy Conserv ation (%)

6.8

179

42

0.7

180

0.59

106

328

77

3.9

53

42

3.3

308

0.55

169

267

63

3.1

13

42

6.0

349

0.48

167

228

54

4. 0

2.7

5

42

10

357

0.35

122

195

45

8. 0

-

-

42

15

365

0.25

91

148

40

13 .0

-

-

42

17

365

0.18

66

125

34

25 .0

-

-

42

20

365

0.11

40

102

28

Treatment Mode

S RT (d)

Super Fast Membrane Bioreactor (SFMBR)

0. 5 1. 0 2. 0

High Rate Activated Sludge Conventional Activated Sludge Nutrient Removal Activated Sludge Extended Aeration Activated Sludge *ΔCS: Biodegradable COD removed **pXT=pXH+pXS+ pXI+pXP

It is also important to note that energy conservation from biological treatment is a hot topic now. A few recent studies investigated the potential of energy recovery, i.e. energy harvesting from excess sludge, in different activated sludge modifications, such as contact stabilization, completely mixed reactors and plug flow schemes, all operated at extremely low sludge ages with conventional gravity settling 52, 53, 54. The results, while highly debatable in terms of basic mass balances and related calculations presented, should be evaluated with the following essential facts: (i) Sludge generation potential, i.e., recoverable excess particulate COD, is the same and depends only on the SRT for all extremely high rate activated sludge modifications. (ii) In systems depending on gravity settling, the energy conservation potential heavily relies upon the appropriate conditioning of the biomass for effective settling and the removal performance of the system for particulate and even colloidal/soluble COD. Unfortunately, high rate activated sludge systems run the risk of not capturing effectively recoverable COD due to poor biomass settling, as can be observed in the above-mentioned studies. Therefore, the only reliable and

sustainable option for maximizing energy recovery remains to be an MBR, or more precisely an SFMBR system as described in this study, which secures total capture of particulate and colloidal COD. Furthermore, Yağcı et al. 55 reported that a chemically enhanced membrane system with a sequence of ultrafiltration and nanofiltration provided almost complete COD recovery without the need for biological treatment. So, membrane is the only viable option!.. 5. CONCLUSION The paper reported the concluding phase of a comprehensive study on super fast membrane bioreactor. In this phase, system performance was experimentally tested with real wastewater, settled sewage, together with a soluble synthetic substrate, peptone mixture, which basically involves similar COD fractionation and biodegradation characteristics with sewage. Performance assessment utilized the latest available experimental methods for bioresearch, respirometry; size distribution analysis; molecular analysis; process modeling, aside from conventional measurements and analyses. Different issues concerning the results obtained were discussed in detail in the evaluation of results section above. Therefore, only the noteworthy aspects of SFMBR features will be mentioned here as “take home” messages. SFMBR was defined primarily to challenge the traditional concept of membrane bioreactor (MBR), operated at long SRTs. The key issue related to the novel biological treatment scheme is membrane and membrane filtration. SFMBR could not even be conceived without membrane filtration and complete retention and control of biomass, simply because system operation at extremely low SRTs of <2.0 d would not be possible with gravity settling. Membrane filtration not only contained biomass but the effective filtration size was improved beyond the nominal pore size of the membrane by cake formation and entrapment of larger organic particles on the cake. The second important attribute of SFMBR is the changes it inflicts on the composition of the microbial community that favors faster growers, imposing their kinetic fingerprint to the system and insuring faster utilization of available substrate. This observation is also significant as it confirms the validity of variable process kinetics, previously reported in similar studies on system kinetics, even within the narrow SRT range selected for SFMBR operation. Low generation of soluble microbial products is also another important attribute that contributes to the observed effective COD removal. Experimental justification of the SFMBR flow scheme essentially focused on two main issues (i) effective COD removal to ensure effluent reuse, and (ii) energy conservation. In fact, model simulations based upon raw sewage characteristics essentially showed partial removal of the particulate slowly biodegradable COD fraction, which accumulated in the reactor volume together with biomass, while soluble biodegradable COD components were almost totally removed. This way, the results of the study also highlighted energy conservation as the major feature of SFMBR, which was assessed to vary between 54-77%, a range 1.5-2.5 times higher than what can be achieved with different activated sludge alternatives.

ACKNOWLEDGMENT This study was conducted as part of a research & development project funded by The Scientific & Technological Research Council of Turkey (TUBITAK) with the project number of 109Y261. TUBITAK is kindly acknowledged for its financial support. REFERENCES 1 D. Orhon, Evolution of the activated sludge process: The first 50 years, J Chem Technol Biotechnol. 90 (2015) 608-640. 2 S.J. Judd, A review of fouling of membrane bioreactors in sewage treatment, Water Sci Technol. 49 (2004) 229-235. 3 N. Ren, Z. Chen, A. Wang, D. Hu, Removal of organic pollutants and analysis of MLSS–COD removal relationship at different HRTs in a submerged membrane bioreactor, Int Biodeterior Biodegradation. 55 (2005) 279-284. 4 A. Massé, M. Spérandio, C. Cabassud, Comparison of sludge characteristics and performance of a submerged membrane bioreactor and an activated sludge process at high solids retention time, Water Res. 40 (2006) 2405-2415. 5 G. Laera, A. Pollice, D. Saturno, C. Giordano, R. Sandulli, Influence of sludge retention time on biomass characteristics and cleaning requirements in a membrane bioreactor for municipal wastewater treatment, Desalination. 206 (2009) 104-110. 6 S.M. Hocaoglu, G. Insel, E. Ubay-Cokgor, D. Orhon, Effect of sludge age on simultaneous nitrification and denitrification in membrane bioreactor, Bioresource Technol. 102 (2011) 6665–6672. 7 P. Le-Clech, V. Chen, T.A. Fane, Fouling in membrane bioreactors used in wastewater treatment, J Membrane Sci. 284 (2006) 17-53. 8 H.Y. Ng, S.W. Hermanowicz, Membrane bioreactor operation at short solid retention times: Performance and biomass characteristics, Water Res. 39 (2005) 981–992. 9 W.F. Harper, M. Bernhardt, C. Newfield, Membrane bioreactor biomass characteristics and microbial yield at very low mean cell residence time, J Water SA. 32 (2006) 193–198. 10 M.N. Pons, H. Spanjers, D. Baetens, O. Nowak, S. Gillot, J. Nouwen, N. Schuttinga, Wastewater characteristics in Europe-A survey. European Water Management Online. 4(10) (2004).

11 M. Henze, Characterization of wastewater for modelling of activated sludge processes, Water Sci Technol. 25 (1992) 1-15. 12 D. Orhon, E. Ubay-Çokgör, S. Sözen, Experimental basis for the hydrolysis of slowly biodegradable substrate in different wastewaters, Water Sci Technol. 39 (1999) 87-95. 13 G. Insel, D. Orhon, P.A. Vanrolleghem, Identification and modelling of aerobic hydrolysis–application of optimal experimental design, J Chem Technol Biotechnol. 78 (2003) 437-445. 14 S. Teksoy Basaran, M. Aysel, H. Kurt, I. Ergal, M. Kumru, A. Akarsubaşı, S. Sözen, D. Orhon, Removal of readily biodegradable substrate in super fast membrane bioreactor, J Membrane Sci. 423 (2012) 477-486. 15 S. Teksoy Basaran, M. Aysel, H. Kurt, I. Ergal, A. Akarsubaşı, N. Yağcı, S. Doğruel, E. Ubay-Çokgör, B. Keskinler, S. Sözen, D. Orhon, Kinetic characterization of acetate utilization and response of microbial population in super fast membrane bioreactor, J Membrane Sci. 455 (2014) 392-404. 16 S. Sözen, I. Pala-Ozkok, M. Aysel, S. Teksoy Başaran, A. Akarsubaşı, I. Ergal, H. Kurt, E. Ubay-Çokgör, D. Orhon, Performance and microbial behavior of submerged membrane bioreactor at extremely low sludge ages, Desalin Water Treat. 56:4 (2015) 862-874. 17 G. Insel, O. Karahan, S. Ozdemir, I. Pala, T. Katipoglu, E. Ubay-Cokgor, D. Orhon, Unified basis for the respirometric evaluation of inhibition for activated sludge, J Environ Sci Heal A. 41 (2006) 1763-1780. 18 I. Pala Ozkok, T. Katipoğlu Yazan, E. Ubay-Cokgor, G. Insel, I. Talinli, D. Orhon, Respirometric assessment of substrate binding by antibiotics in peptone biodegradation, J Environ Sci Heal A. 46 (2011) 1588-1597. 19 D. Orhon, S. Sözen, S. Teksoy Basaran, B. Alli, Super fast membrane bioreactor—transition to extremely low sludge ages for waste recycle and reuse with energy conservation, Desalin Water Treat. 57 (2016) 21160-21172. 20 S. Sözen, E. Ubay-Çokgör, S. Teksoy Başaran, M. Aysel, A. Akarsubaşı, I. Ergal, H. Kurt, I. Pala-Ozkok, D. Orhon, Effect of high loading on substrate utilization kinetics and microbial community structure in super fast submerged membrane bioreactor, Bioresource Technol. 159 (2014) 118-127. 21 E. D lekg rgen, S. Doğruel, . Karahan, D. Orhon, Size distribution of wastewater COD fractions as an index for biodegradability, Water Res. 40 (2006) 273–282. 22 Millipore Corporation, 75 Wiggins Ave, Bedford, MA 01730, United States.

23 G.A. Ekama, P.L. Dold, G.v.R. Marais, Procedures for determining influent COD fractions and the maximum specific growth rate of heterotrophs in activated sludge systems, Water Sci Technol. 18 (1986) 91–114. 24 M. Henze, C.P.L.Jr. Grady, W. Gujer, G.V.R. Marais, T. Matsuo, Activated sludge model No. 1, IAWQ Scientific and Technical Report No. 1, London, UK, 1987. 25 D. Orhon, E. Ubay-Çokgör, S. Sözen, Dual hydrolysis model of the slowly biodegradable substrate in activated sludge systems, Biotechnol Tech. 12 (1998) 737-741. 26 D. Orhon, N. Artan, Y. Cimşit, The concept of soluble residual product formation in the modelling of activated sludge, Water Sci Technol. 21 (1989) 339-350. 27 P. Reichert, J. Ruchti, W. Simon, AQUASIM 2.0, Swiss Federal Institute for Environmental Science and Technology (EAWAG), CH-8600, Duebendorf, Switzerland, 1998. 28 I. Pala-Ozkok, A. Rehman, N. Yagci, E. Ubay-Cokgor, D. Jonas, D. Orhon, Characteristics of mixed microbial culture at different sludge ages: effect on variable kinetics for substrate utilization, Bioresource Technol. 126 (2012) 274-282. 29 R. Brun, P. Reichert, H.R. Künsch, Practical identifiability analysis of large environmental simulation models, Water Resour Res. 37(4) (2001) 1015-1030. 30 T. Katipoglu Yazan, I. Pala-Ozkok, E. Ubay-Cokgor, D. Orhon, Acute impact of tetracycline and erythromycin on the storage mechanism of polyhydroxyalkanoates, Biochem Eng J. 91 (2014) 283-289. 31 S. Sözen, S. Teksoy Başaran, A. Akarsubaşı, I. Ergal, G. Insel, C. Karaca, D. Orhon, Toward a novel membrane process for organic carbon removal—fate of slowly biodegradable substrate in super fast membrane bioreactor, Environ Sci Pollut R. 23 (2016) 16230-16240. 32 K. Noyan, B. Allı, D. Okutman Taş, S. Sözen, D. Orhon, Relationship between COD particle size distribution, COD fractionation and biodegradation characteristics in domestic sewage, J Chem Technol Biotechnol, 92 (2017) 2142–2149. doi:10.1002/jctb.5223 33 M.F. Dignac, P. Ginestet, D. Rybacki, A. Bruchet, V. Urbain, P. Scribe, Fate of wastewater organic pollution during activated sludge treatment: nature of residual organic matter, Water Res. 34 (2000) 4185-4194. 34 E. Pehlivanoglu-Mantas, D. Okutman Tas, G. Insel, E. Aydin, D.C. Ozturk, T. Olmez, E. Gorgun, E. Ubay-Cokgor, D. Orhon, Evaluation of municipal and industrial wastewater treatment sludge stabilization in Istanbul, CLEAN–Soil, Air, Water. 35 (2007) 558-564.

35 N. Yagci, M. Konuk, S. Sozen, S. Meriç, D. Orhon, Chemically enhanced membrane process–towards a novel sewage treatment concept to potentially replace biological processes, Desalin Water Treat. 57 (2016) 16238-16249. 36 S.M. Hocaoglu, D. Orhon, Fate of soluble residual organics in membrane bioreactor, J Membrane Sci. 364 (2010) 65-74. 37 S.M. Hocaoglu, D. Orhon, Fate of proteins and carbohydrates in membrane bioreactor operated at high sludge age, J Environ Sci Heal A. 45 (2010) 1101-1108. 38 K. Bani-Melhem, Z. Al-Qodah, M. Al-Shannag, A. Qasaimeh, M.R. Qtaishat, M. Alkasrawi, On the performance of real grey water treatment using a submerged membrane bioreactor system, J Membrane Sci. 476 (2015) 40-49. 39 D. Okutman Tas, O. Karahan, G. Insel, S. Ovez, D. Orhon, H. Spanjers, Biodegradability and denitrification potential of settleable chemical oxygen demand in domestic wastewater, Water Environ Res. 81 (2009) 715- 727. 40 G. Stephanopoulos, A.A. Aristidou, J. Nielsen, Metabolic engineering: principles and methodologies, Academic press, San Diego, CA, 1998. 41 Y. Zhang, I.P. Adams, C. Ratledge, Malic enzyme: the controlling activity for lipid production? Overexpression of malic enzyme in Mucor circinelloides leads to a 2.5fold increase in lipid accumulation, Microbiology. 153 (2007) 2013-2025. 42 I. Pala-Ozkok, A. Rehman, G. Kor-Bicakci, A. Ural, M.B. Schilhabel, E. UbayCokgor, D. Jonas, D. Orhon, Effect of sludge age on population dynamics and acetate utilization kinetics under aerobic conditions, Bioresource Technol. 143 (2013) 68-75. 43 T.T. Wee, H.Y. Ng, S.L. Ong, Effect of mean cell residence time on the performance and microbial diversity of pre-denitrification submerged membrane bioreactors, Chemosphere. 70 (2008) 387-396. 44 L. Duan, I. Moreno-Andrade, C.L. Huang, S. Xia, S.W. Hermanowicz, Effects of short solids retention time on microbial community in a membrane bioreactor. Bioresource Technol. 100 (2009) 3489-3496. 45 P.E. Saikaly, D.B. Oerther, Bacterial competition in activated sludge: theoretical analysis of varying solids retention times on diversity, Microbial Ecol. 48 (2004) 274284. 46 P.E. Saikaly, P.G. Stroot, D.B. Oerther, Use of 16S rRNA gene terminal restriction fragment analysis to assess the impact of solids retention time on the bacterial diversity of activated sludge, Appl Environ Microb. 71 (2005) 5814-5822.

47 C. Bertoldo, G. Antranikian, Starch-hydrolyzing enzymes from thermophilic archaea and bacteria, Curr Opin Chem Biol. 6 (2002) 151-160. 48 I. Pala Ozkok, A. Rehman, E. Ubay-Cokgor, D. Jonas, D. Orhon, Pyrosequencing reveals the inhibitory impact of chronic exposure to erythromycin on activated sludge bacterial community structure, Biochem Eng J. 90 (2014) 195-205. 49 P.L. McCarty, Energetic and Bacterial Growth, 5th Rudolph Research Conference, Rutgers. The State University, New Brunswick, New Jersey, 1972. 50 K. Svardal, H. Kroiss, Energy requirements for wastewater treatment, Water Sci Technol. 64 (2011) 1355-1361. 51 D. Okutman, S. Ovez, D. Orhon, Hydrolysis of settleable substrate in domestic sewage, Biotechnol Lett. 23 (2001) 1907-1914. 52 A. Rahman, F.A. Meerburg, S. Ravadagundhi, B. Wett, J. Jimenez, C. Bott, A. Al-Omari, R. Riffat, S. Murthy, H. De Clippeleir, Bioflocculation management through high-rate contact-stabilization: A promising technology to recover organic carbon from low-strength wastewater, Water Res. 104 (2016) 485-496. 53 J. Jimenez, M. Miller, C. Bott, S. Murthy, H. De Clippeleir, B. Wett, High-rate activated sludge system for carbon management–Evaluation of crucial process mechanisms and design parameters, Water Res. 87 (2015) 476-482. 54 F.A. Meerburg, N. Boon, T. Van Winckel, K.T. Pauwels, S.E. Vlaeminck, Live fast, die young: optimizing retention times in high-rate contact stabilization for maximal recovery of organics from wastewater, Environ Sci Technol. 50 (2016) 97819790. 55 N. Yagci, M. Konuk, S. Sözen, S. Meriç, D. Orhon, Chemically enhanced membrane process–towards a novel sewage treatment concept to potentially replace biological processes, Desalin Water Treat. 57 (2016) 16238-16249.

Highlights 

Almost totally removal of soluble COD components with super fast MBR



Improvement of the effective filtration size by cake formation (8-13 nm)



Faster substrate utilization due to favoring faster growers



Low generation of soluble microbial products in super fast MBR



Energy conservation and recovery between 54-77%