- Email: [email protected]
Impact of solids residence time on biological nutrient removal performance of membrane bioreactor
water research 44 (2010) 3192–3202
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Impact of solids residence time on biological nutrient removal performance of membrane bioreactor Cagatayhan Bekir Ersu a, Say Kee Ong b,*, Ertan Arslankaya c, Yong-Woo Lee d a
Cukurova University, Environmental Engineering Department, Adana, Turkey Iowa State University, Civil, Construction and Environmental Engineering Department, 490 Town Engineering, Ames IA 50011, USA c Yildiz Technical University, Environmental Engineering Department, Istanbul, Turkey d Hanyang University, Department of Applied Chemistry, Seoul, Republic of Korea b
article info
abstract
Article history:
Impact of long solids residence times (SRTs) on nutrient removal was investigated using
Received 18 October 2009
a submerged plate-frame membrane bioreactor with anaerobic and anoxic tanks. The
Received in revised form
system was operated at 10, 25, 50 and 75 days SRTs with hydraulic retention times (HRTs) of
31 December 2009
2 h each for the anaerobic and anoxic tanks and 8 h for the oxic tank. Recirculation of oxic
Accepted 22 February 2010
tank mixed liquor into the anaerobic tank and permeate into the anoxic tank were fixed at
Available online 1 March 2010
100% each of the influent flow. For all SRTs, percent removals of soluble chemical oxygen demand were more than 93% and nitrification was more than 98.5% but total nitrogen
Keywords:
percent removal seemed to peak at 81% at 50 days SRT while total phosphorus (TP) percent
Nitrification
removal showed a deterioration from approximately 80% at 50 days SRT to 60% at 75 days
Denitrification
SRT. Before calibrating the Biowin model to the experimental data, a sensitivity analysis of
Phosphorus uptake
the model was conducted which indicated that heterotrophic anoxic yield, anaerobic
SRT
hydrolysis factors of heterotrophs, heterotrophic hydrolysis, oxic endogenous decay rate for
Modeling
heterotrophs and oxic endogenous decay rate of PAOs had the most impact on predicted effluent TP concentration. The final values of kinetic parameters obtained in the calibration seemed to imply that nitrogen and phosphorus removal increased with SRT due to an increase in anoxic and anaerobic hydrolysis factors up to 50 days SRT but beyond that removal of phosphorus deteriorated due to high oxic endogenous decay rates. This indirectly imply that the decrease in phosphorus removal at 75 days SRT may be due to an increase in lysis of microbial cells at high SRTs along with the low food/microorganisms ratio as a result of high suspended solids in the oxic tank. Several polynomial correlations relating the various calibrated kinetic parameters with SRTs were derived. The Biowin model and the kinetic parameters predicted by the polynomial correlations were verified and found to predict well the effluent water quality of the MBR at 35 days SRT. ª 2010 Elsevier Ltd. All rights reserved.
1.
Introduction
Membrane bioreactors (MBRs) have several operating advantages over conventional biological treatment processes
(Visvanathan et al., 2000). MBR can be operated efficiently at high-suspended solids concentrations, long solids residence times (SRTs) and short hydraulic retention times (HRTs) with excellent carbonaceous organic matter removal and
* Corresponding author. Tel.: þ1 515 294 3927; fax: þ1 515 294 8216. E-mail address: [email protected] (S.K. Ong). 0043-1354/$ – see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.02.036
water research 44 (2010) 3192–3202
nitrification without the concern of poor sludge settling since the membrane provides an excellent separation barrier (Stephenson et al., 2000). With stricter nutrient discharge requirements by regulatory agencies, MBR, with many advantages, can be modified to effectively remove nutrients (nitrogen and phosphorus) by using various anaerobic, anoxic, and oxic manipulations (Ersu et al., 2008; Abegglen et al., 2008). The high suspended solids concentrations provide the added benefits of minimizing the phosphorus content in the solids while the long SRTs allow growth of bacteria with slow growth rates such as nitrifying bacteria (Hasar et al., 2002). Hence, near complete nitrification is observed in MBRs provided oxygen and hydraulic retention requirements are met (Kishino et al., 1996; Delgado et al., 2002; Shin et al., 2004). In the case of phosphorus uptake, the phosphate accumulating organisms (PAOs) utilized their stored energy source under aerobic conditions (Ubukata and Takii, 1998). However, the long SRTs of MBRs may impact the luxurious uptake of phosphorus along with the survival of the PAOs due to the finite amount of stored substrate (Rosenberger et al., 2000). Abegglen et al. (2008) achieved 90% and 70% removal of total nitrogen (TN) and total phosphorus (TP) using an anaerobic/anoxic MBR treating household wastewater. Using a 6-stage MBR pilot plant operated at 23 days SRT, Fleischer et al. (2005) achieved TN concentrations of less than 3 mg/L and a TP concentration of less than 0.1 mg/L by a combination of biological and chemical (alum) treatment. Adam et al. (2003) studied the treatment of municipal wastewater using benchscale submerged MBR with anaerobic-anoxic-oxic configuration and obtained removals of 96.4% chemical oxygen demand (COD), 86.8% TN, 98.8% ammonia-nitrogen (NH3-N), and 99% TP for 15 days SRT and anaerobic, anoxic, and oxic hydraulic retention times of 3.7 h, 8.5 h, and 8.8 h, respectively. The same researchers also studied a similar MBR setup operated with a post-denitrification mode and obtained similar results with 95.1% COD, 89.6% TN, 99.9% NH3-N, and 99.2% TP. Lesjean et al. (2002) studied both configurations described by Adam et al. (2003) but at a pilot-scale with a total HRT of 18 h and an SRT of 26 days and found 81.9% TN and 99.3% TP removal in pre-denitrification mode as compared to 94.1% TN and 99.2% TP for post-denitrification mode. Rosenberger et al. (2002) applied an infinite SRT for a pilot-scale MBR (with an oxic tank only) and achieved 82% TN removal with 99.9% total kjeldahl nitrogen (TKN) removal for a municipal wastewater suggesting that long SRTs do not have an adverse effect on nitrogen removal. Similarly, Ogoshi and Suzuki (2000) treated synthetic wastewater in a pilot-scale MBR at over 400 days SRT and reported 95% removal for 5-day biochemical oxygen demand (BOD5), 80% removal for COD, and over 90% removal for TN. Urbain et al. (1998) reported COD, ammonia (NH3), and TP removals of 90%, 99%, and 50%, respectively for SRTs between 5 and 20 days. Based on the literature, it is possible to obtain TN and TP removals greater than 90%, respectively, for HRTs ranging from 6 to 18 h and typical SRTs of 10–25 days (Metcalf and Eddy, 2003). MBR systems with anaerobic, anoxic, and oxic manipulations have been shown to achieve high level biological nutrient removal but there are no studies on phosphorus removal for long SRTs over 25 days.
3193
The main objective of this study was to evaluate the impact of long SRTs (25 days and over) on biological nutrient removal performance of a submerged MBR with an integration of anaerobic, anoxic, and oxic tanks. Various kinetic parameters for biological nutrient removal over a range of SRTs were analyzed using the Biowin AS/AD model to investigate the important kinetic parameters that may be impacted due to the long SRTs of the MBRs. The study provides operating and kinetic information on nutrient removal for design and operation of MBRs systems with SRTs greater than 25 days. This study is probably the first to model an MBR with anaerobic, anoxic, and oxic tanks for nutrient removal with long SRTs.
2.
Materials and methods
2.1.
Experimental setup
Experiments were conducted using a bench-scale MBR configured with an anaerobic and anoxic tank as shown in Fig. 1. The synthetic wastewater (see Table 1) was fed into the bioreactor using a Masterflex Model 8800 pump (Hamden, CT). Sources of nitrogen were organic in nature (urea, nutrient broth and Isomil, a soy-based infant formula (Abbott Laboratories, Abbott Park, IL)) while sources of phosphorus were from both inorganic sources (biphosphate) and organic sources (nutrient broth and Isomil). The bioreactor was made of Plexiglass and had a total maximum operating volume of 16 L (4 L for the anaerobic, 4 L for anoxic and 8 L for oxic tanks) with an oxic tank dimension of 10 cm wide by 25 cm long and 50 cm deep. The cylindrical anaerobic and anoxic tank each has a diameter of 15 cm and a depth of 50 cm with variable operating volumes between 1 and 4 L by using various outlet ports located on the side of the tanks. To create anaerobic conditions, the anaerobic tank was sealed with a Plexiglass cover using six bolts and a rubber gasket and the contents were mixed gently with a variable-speed magnetic stirrer while the anoxic tank was opened to the atmosphere and was also gently mixed with a variable-speed magnetic stirrer. Wastewater flowed by gravity from the anaerobic tank to the anoxic tank and into the oxic tank. Air was introduced via air diffusers placed at the bottom of the oxic tank for microbial metabolism and to control membrane fouling. The air and liquid flow rates were measured using Gilmont ball flow meters (Barrington, IL). Experiments were conducted at room temperature at 22.5 1 C. For solid/liquid separation purposes, a double-sided plateframe cellulose membrane (Kubota Co., Osaka, Japan) with a nominal pore size of 0.2 mm and 0.15 m2 of filtration area was used. Constant permeate flux was maintained by a Cole– Parmer Model 7553-70 pump (Vernon Hills, IL) which was placed at the permeate outlet. Mixed liquor recirculation rates were controlled by another Masterflex Model 8800 pump (Hamden, CT). The plate-frame module was operated for 9 min with permeate fluxes ranging between 8 and 28 L/m2/h, followed by 1 min of idle period. No backwashing of the membrane was required. The operational cycles of filtration run and idle period were controlled using a ChronTrol microprocessing timer (ChronTrol Corp., San Diego, CA). The reactor was equipped with water level sensors to maintain
3194
water research 44 (2010) 3192–3202
Anaerobic (Slow mix) Anoxic (Complete mix)
C
Permeate PI
F
Level Sensors
D
E
Membrane
B
B
Air Mixed Liquor Recycle
G Permeate Recycle
G
Mixed Liquor Wastin g Permeate Recycle
Feed
A
Feed
A
Fig. 1 – Membrane bioreactor for biological nutrient removal with anaerobic and anoxic tanks (Sampling points: A – Synthetic wastewater feed; B – Anaerobic influent; C – Anaerobic effluent; D – Anoxic effluent; E – Mixed liquor; F – Permeate; G – Waste mixed liquor).
Table 1 – Ingredients and composition of the synthetic wastewater. Ingredient Calcium sulfate Magnesium sulfate Ferric chloride Sodium biphosphate Sodium bicarbonate Potassium chloride Urea Nutrient broth Sodium citrate Isomil (0.8% by vol.) Composition
Concentration (mg/L) 40 4 3 75 63 5 42 250 500 20 mL
Average concentration (mg/L) 95% CI pH 7.7 0.2 Chemical Oxygen Demand (COD) 555 9.1 soluble chemical oxygen demand (sCOD) 502 11.1 342 11 Biochemical Oxygen Demand (BOD5) Total nitrogen (TN) 43.3 1.9 Ammonia nitrogen (NH3-N) 22.7 1.8 0.2 0.1 Nitrite-nitrogen (NO 2 -N) 0.5 0.1 Nitrate-nitrogen (NO 3 -N) Organic nitrogen 20.0 1.5 Total phosphorus (TP) 13.6 0.9 Suspended solids (SS) 71.7 6.4
a constant volume in the bioreactor. Mixed liquor wasting was done manually on a daily basis to provide the desired SRT. The bioreactor was initially inoculated with a mixture of biomass from a bench-scale biological nutrient removal system (at 25 days SRT) and a nearby activated sludge municipal wastewater treatment plant. The oxic bioreactor was operated on an 8-hour oxic batch mode for about 5 cycles to acclimatize the microorganisms before the bioreactor was switched to continuous mode. The bench-scale MBR was operated under 100% recirculation of mixed liquor to the anaerobic tank and 100% recirculation of nitrate-rich permeate to the anoxic tank. The recirculation configuration was found to be the better of several recirculation configurations tested (Ersu et al., 2008). The feed water flow rate was maintained at 1 L/h giving hydraulic retention times (HRT) of 2, 2, and 8 h for the anaerobic, anoxic, and oxic tanks, respectively. The SRT in the oxic tank was initially maintained at 25 days with a biomass concentration of approximately 7500 mg/L in the oxic tank.
2.2.
Varying solids residence times
To study the impact of SRT on biological nutrient removal, the MBR was operated initially at 25 days SRT (see Ersu et al., 2008) followed by 50 days and 75 days SRT. The reactor was then operated at 10 days and 35 days SRT. For each change in SRT, the bioreactor was operated until steady state conditions were
3195
water research 44 (2010) 3192–3202
Fig. 2 – BioWin flow diagram of membrane bioreactor for biological nutrient removal.
achieved which ranged from three weeks for shorter SRTs to as much as eight weeks for longer SRTs. Steady state conditions were assumed to be achieved when the most recent three measurements of COD, TN and TP concentrations were within 10% (Ha et al., 2007; Ersu et al., 2008). Wastewater samples were collected and analyzed, on a regular basis (two to three times per week), from various sampling points as shown in Fig. 1. Anaerobic, anoxic, and oxic (permeate) effluents as well as influent wastewater were analyzed. Wastewater samples for analysis were obtained by opening the sampling valves and collecting a portion of the flow with a glass collection flask for half hour to obtain 1⁄2 -hour samples. Steady state results of the analyses presented here were the average of at least three samples collected at different times along with their 95% confidence intervals. Transmembrane pressure (TMP), pH, oxidation–reduction potential (ORP), and dissolved oxygen (DO) were monitored throughout the experimental study using dual-scale pressure gauge, pH/ORPmeter (Orion Model 260), and DO meter (Orion Model 830), respectively. NH3-N was measured using an ammonia meter (Orion Model 290A). Total suspended solids and volatile suspended solids were analyzed using Standard Methods 2540D and 2540E, respectively, (APHA et al., 2002) while COD and sCOD were analyzed using Hach Colorimetric Method 8000 (Hach Company, Loveland, CO). TN, TP and nitrate-nitrogen (NO 3 -N) were analyzed using Hach persulfate digestion Method 10071, Hach molybdovanadate Method 10127, and Hach cadmium reduction Method 8039, respectively (Hach Company, Loveland, CO).
Table 2 – Carbonaceous and nutrient fractions of the synthetic wastewater. Abbreviation Fbs
Fac Fxsp
Fus Fup Fna Fnox Fnus FupN
Fpo4 FupP
FZbh FZbm FZba
2.3. Kinetic modeling for biological nutrient removal in MBR using BioWin The Biowin AS/AD model (Envirosim Inc., Ontario, Canada) was calibrated using the experimental results of the MBR at different SRTs. To mimic the MBR, the Biowin AS/AD model simulates the MBR as a combination of an aerated tank and a sludge dewatering device with all the sludge returning back into the oxic reactor (see Fig. 2). Since the membrane of the
FZbp FZbpa FZbam FZbhm
Fraction Readily biodegradable (including acetate) [g COD/g of total COD] Acetate [g COD/g of readily biodegradable COD] Non-colloidal slowly biodegradable [g COD/g slowly degradable COD] Unbiodegradable soluble [g COD/g of total COD] Unbiodegradable particulate [g COD/g of total COD] Ammonia [g NH3-N/g TKN] Particulate organic nitrogen [g N/g Organic N] Soluble unbiodegradable TKN [g N/g TKN] N:COD ratio for unbiodegradable particulate COD [g N/g COD] Phosphate [g PO4-P/gTP] P:COD ratio for influent unbiodegradable particulate COD [g P/g COD] Non-PAO heterotrophs [g COD/g of total COD] Anoxic methanol utilizers [g COD/g of total COD] Autotrophs [g COD/g of total COD] PAOs [g COD/g of total COD] Propionic acetogens [g COD/g of total COD] Acetoclastic methanogens [g COD/g of total COD] H2-utilizing methanogens [g COD/g of total COD]
a Estimated using actual data.
Default Input 0.20
0.25a
0.15
0.15
0.75
0.70a
0.05
0.05
0.13
0.05a
0.66 0.50
0.50a 0.50
0.00
0.00
0.035
0.035
0.50 0.011
0.80a 0.011
1.0E-4
1.0E-4
1.0E-4
1.0E-4
1.0E-4
1.0E-4
1.0E-4 1.0E-4
1.0E-4 1.0E-4
1.0E-4
1.0E-4
1.0E-4
1.0E-4
3196
water research 44 (2010) 3192–3202
MBR acts almost like an impermeable barrier to the mixed liquor solids, use of a dewatering device to simulate the MBR would be adequate as the dewatering device ensures that all of the mixed liquor solids are retained within the oxic tank. The model assumed the following kinetic reactions: aerobic hydrolysis of slowly biodegradable substrate; anoxic hydrolysis of slowly biodegradable substrate; anaerobic hydrolysis of slowly biodegradable substrate; aerobic growth of heterotrophs on fermentable substrates; aerobic growth of heterotrophs on fermentation products; denitrification with fermentable substrates; denitrification with fermentation products; fermentation of complex soluble substrate to propionate, acetate, CO2; lysis of heterotrophs; storage of poly-hydroxy-alkanoates (PHA) by PAOs; aerobic storage of poly-P by PAOs; anoxic storage of poly-P by PAOs; aerobic growth of PAOs on PHA; anoxic growth of PAOs on PHA; aerobic growth of autotrophs; and lysis of autotrophs. Various carbonaceous and nutrient fractions needed by the model were estimated from the composition of the feed water. If the water quality parameters were not tested, the default values of the model were assumed (see Table 2). Impact of each kinetic parameter for autotrophs, heterotrophs, and PAOs in the Biowin AS/AD model on effluent TN and TP concentrations was assessed by using a sensitivity analysis starting with the default value of the kinetic parameters provided by the model and varying the default value by up to 50%. Based on the trends and impact of each kinetic parameter on effluent TN and TP concentrations, the kinetic parameters were selected accordingly and varied to calibrate the model against the total suspended solids (TSS) of the reactors followed by the effluent water quality by minimizing the sum of square errors. The calibrated model was then verified using the experimental data for a SRT of 35 days.
3.
Results and discussion
3.1.
Wastewater characteristics
BOD5 was approximately 70% of the total COD for the synthetic wastewater used while the soluble COD (sCOD) was about 85% of the total COD. The COD:TN:TP ratio was approximately 100:8:2. About 52% of the TN was due to NH3-N while the organic nitrogen was approximately 47% of the total nitrogen. For all SRTs, the ORP was maintained, on the average, below 340 mV in the anaerobic tank and 235 mV in the anoxic tank while ORPs in the oxic tank were approximately 100 mV (see Table 3). The TMPs were consistently kept below 15 kPa. In the oxic tank, the DO was kept above 2 mg/L while the DO in the anoxic tank was less than 0.3 mg/L. The pH in the anoxic and oxic tanks was between 7 and 8. Suspended solids concentrations in the tanks increased proportionally with SRT (see Table 3). TSS in the aerobic tank for 50 days and 75 days SRTs were approximately 2 and 2.5 times higher, respectively than the TSS for 25 days SRT.
3.2.
Impact of solids residence time on MBR performance
Fig. 3a presents the sCOD concentrations of the anaerobic effluent, anoxic effluent, and oxic effluent of the MBR for different SRTs (10, 25, 50, and 75 days). On the average, more than 93% of the sCOD was removed for all SRTs and the sCOD concentrations of the final effluent were between 26 and 32 mg/L with a mean value of 29 mg/L. There were no significant differences in the mean sCOD percent removals for the SRTs tested. Advantages of the MBR system as compared to conventional biological treatment systems include the ability
Table 3 – Operating wastewater conditions (average values D95% CI) in anaerobic, anoxic, and oxic tanks under varying SRTs. ORP (mV)
DOa (mg/L)
TSS (mg/L)
Anaerobic 10 25 50 75
2844 34 7350 99 13,897 439 18,825 50
6.8 6.9 6.8 7.0
0.2 0.2 0.1 0.2
23.0 23.0 22.0 23.0
0.5 0.5 0.5 0.5
369 346 364 391
13 23 15 14
0.0 0.0 0.0 0.0
0 0 0 0
Anoxic 10 25 50 75
2126 3505 7300 9530
7.0 7.3 7.2 7.3
0.2 0.2 0.2 0.2
23.0 23.0 23.0 23.0
0.5 0.5 0.5 0.5
232 236 267 291
13 8 6 4
0.1 0.3 0.1 0.1
0.1 0.1 0.1 0.1
Oxic 10 25 50 75
4546 74 7400 60 13265 560 18,275 310
7.2 7.5 7.3 7.1
0.2 0.2 0.2 0.1
23.0 23.0 22.8 22.0
0.2 0.5 0.5 0.5
3.7 2.9 2.2 2.0
0.1 0.3 0.2 0.2
a Dissolved oxygen.
29 120 170 148
pH
Temp. ( C)
SRT (days)
102 4 108 6 82 6 77 4
3197
water research 44 (2010) 3192–3202
mg sCOD/L
a
Anaerobic - Sampling Point C Anoxic - Sampling Point D Oxic - Sampling Point F
200 150
100
187 153
50
108
107
27
140
32
98
57
87
26
27
0
10 days
25 days
50 days
75 days
30
b
25
mg T N/L
20 15
10
21.4
17.7
5
10.1
21.1
16.8
13.5
20.9 16.2
9.3
15.8 8.4
8.3
0
10 days
mg NH 3-N/L
c
18 16 14 12 10 8 6 4 2 0
25 days
8.3
8.2
d
7.8
25 days
7.8
0.3
0.2
0.1
10 days
75 days
14.2
13.5
12.2
11.8
50 days
50 days
0.3
75 days
12
mg NO 3- -N/L
10 8 6
4
9.7 2.2
7.3
2.6
2
0.5
0.5
6.7
6.6 0.3
0.4
0.2
0.2
0
10 days
e
25 days
50 days
75 days
35 30
mg T P/L
25
20 15
10
30.8
15.6 3.4
5
11.3
8.7
3.5
14.4
10.1
14.3 2.8
5.4
10.4
0
10 days
25 days
50 days
75 days
Fig. 3 – Effluent concentrations of (a) sCOD, (b) total nitrogen, (c) ammonia-nitrogen, (d) nitrate-nitrogen, (e) total phosphorus for anaerobic, anoxic and oxic reactors and different SRTs (error bars – 95% CI of three or more samples taken at different times under steady state conditions).
to maintain high biomass concentrations resulting in sCOD removals equal or higher than conventional systems and a smaller reactor size. With the MLSS concentrations varying from 7500 to about 18,000 mg/L for 25–75 days SRT, the mass loadings for the aerobic tank were between 0.07 and 0.20 kg COD/kg MLSS/d which were lower than the permissible mass loading rates for conventional systems in the range of 0.3–0.6 kg COD/kg MLSS/d. This means that the MBR can take a higher influent COD load than a conventional system (Metcalf and Eddy, 2003; Stephenson et al., 2000). TN concentrations in the oxic effluent decreased with the SRT of the MBR as shown in Fig. 3b. Percent TN removals
were 65.7 3.4%, 77.9 1.0%, 80.6 0.3% and 81.0 0.3% for 10, 25, 50 and 75 days SRT, respectively. At 25 days, average NO 3 -N concentration was 0.5 mg/L for both anaerobic and anoxic tanks while, at 75 days, average NO 3 -N concentration was as low as 0.2 mg/L, demonstrating denitrification in anaerobic and anoxic tanks for SRTs greater than 10 days (see Fig. 3d). Average NO 3 -N concentrations in the oxic tank effluents decreased with longer SRT and reached an asymptotic value of approximately 6.6 mg/L for 50 days or more SRT. Nitrate concentrations constituted about 78–80% of the TN concentrations of the final effluent indicating there is still some room for improving denitrification by changing
3198
water research 44 (2010) 3192–3202
f
40 30 20 10 0 -10 -20 -30 -40 -50
TN TP
-60
-40
-20
0
20
40
30 20 10
0 -10 -20 -30
-40
60
-60
% Variation in Heterotrophic Anoxic Yield
-40
-20
0
20
40
40
60
0 -5 -10 -15 -60
h % Sensitivity in Effluent
% Sensitivity in Effluent
20
5
60
40 30 20 10 0 -10 -20 -30 -40
-40 -20 0 20 40 60 % Variation in Anaerobic Endogenous Decay Rate-PAO
15 10 5
0 -5
-10 -15 -20
-60
% Sensitivity in Effluent
0
10
% Variation in Oxic Endogenous Decay Rate-Heterotrophic
d
-20
g 15
30 20 10 0 -10 -20 -30 -40 -50 -60
c
-40
% Variation in Oxic Endogenous Decay Rate-PAO
% Sensitivity in Effluent
% Sensitivity in Effluent
b
% Sensitivity in Effluent
% Sensitivity in Effluent
a
-40
-20
0
20
40
60
-60
% Variation in Hydrolysis Rate (AS)-Heterotrophic
-40 -20 0 20 40 % Variation in Anoxic Growth Factor-PAO
60
15 10 5 0 -5 -10 -15 -60
-40
-20
0
20
40
60
% Variation in Anoxic Hydrolysis Factor-Heterotrophic
% Sensitivity in Effluent
e
80
60 40 20 0 -20
-40 -60 -100 -80 -60 -40 -20 0 20 40 60 80 100 % Variation in Anaerobic Hydrolysis Factor-Heterotrophic
Fig. 4 – Sensitivity analysis of various kinetic parameters on total nitrogen and total phosphorus effluent concentrations (a) Heterotrophic anoxic yield, (b) oxic endogenous decay rate – heterotrophic, (c) Hydrolysis rate – heterotrophic, (d) Anoxic hydrolysis factor – heterotrophic, (e) Anaerobic hydrolysis factor – heterotrophic, (f) Oxic endogenous decay rate – PAO, (g) Anaerobic endogenous decay rate – PAO, (h) Anoxic growth factor – PAO.
the recirculation flow rates. With 0.2–0.3 mg NH3-N/L in the final effluent, the MBR demonstrated excellent nitrification for all SRTs (see Fig. 3c) with more than 98% NH3-N removal. Final effluent phosphorus concentrations were as low as 2.8 mg/L for 50 days SRT as shown in Fig. 3e. Percent TP removals increased from 70.8 7.1% at 10 days SRT to 79.6 0.4% at 50 days SRT and then decreased to 60.1 0.42% for 75 days SRT. Deterioration of TP removal for 75 days SRT indicated that there might be a limitation in PAOs activity after a certain SRT and may also be due to lysis of cells at long SRTs.
For all SRTs, the higher TP concentrations in the anaerobic effluent (example, 30.8 mg/L for 10 day SRT) as compared to the feed water indicated that there was phosphorus release by PAOs as expected in the anaerobic tank.
3.3.
Modeling biological nutrient removal with BioWin
The impact of various kinetic parameters predicting the effluent TN and TP concentrations in the sensitivity analysis are presented in Fig. 4a–h. Of the various kinetic parameters,
– 8,590,304 56,850 5.4 8.5 5.8 6.7 8.1 7.7 0.3 0.3 0.3 8.3 10.4 8.1 a AA – anaerobic; AO – anoxic; 0 – oxic, mmax – maximum specific growth rate; kd oxic – oxic endogenous decay rate. b Sum of square errors. c Model assumes all nitrite converted to nitrate.
27 35 27 18,275 16,209 18,220 9530 8506 9715 18,825 17,016 18,965 – 0.10 0.26 – 0.95 1.4 – 0.62 0.72 – 3.20 2.80 – 0.90 1.20 Experimental (75 d) Default Calibrated
– 0.17 0.33
– 6,558,989 50,188 2.8 7.1 3.1 6.6 7.9 7.4 0.3 0.3 0.3 8.4 9.8 7.9 26 36 28 13,265 11,672 13,187 7300 6132 7498 13,897 12,267 13,967 – 0.10 0.20 – 0.95 1.4 – 0.62 0.65 – 3.20 2.80 – 0.90 1.20 Experimental (50 d) Default Calibrated
– 0.17 0.25
– 272,344 13,849 3.5 3.9 3.7 7.3 7.6 7.3 0.2 0.3 0.3 9.3 9.2 8.8 32 35 34 7400 6901 7296 3505 3635 3532 7350 7272 7398 – 0.10 0.12 – 0.95 1.4 – 0.62 0.55 – 3.20 2.80 – 0.90 1.20 Experimental (25 d) Default Calibrated
– 0.17 0.20
3.4 0.2 2.3 9.7 7.1 11.6 0.1 0.3 0.3 13.5 8.5 13.2 27 32 29 4546 4099 4526 2126 1796 1973 2844 2708 2976 – 0.10 0.07 – 0.95 1.4 – 0.62 0.53 – 3.20 2.80 – 0.17 0.17 – 0.90 1.20 Experimental (10 d) Default Calibrated
TN (mg/L) sCOD (mg/L) TSSAO (mg/L) TSSAAa (mg/L) kd oxic (d1) mmax (d1)
kd oxic (d1)
mmax (d1)
kd oxic (d1)
mmax (d1)
PAO Heterotrophs Autotrophs
Table 4 – Calibration for total suspended solids in MBR and effluent quality for varying SRTs.
TSS0 (mg/L)
Wastewater quality
NH3-N (mg/L)
NO-3-Nc (mg/L)
TP (mg/L)
– 327,344 41,319
S SSEb
water research 44 (2010) 3192–3202
3199
heterotrophic anoxic yield gave the largest change in the predicted effluent TN concentrations (see Fig. 4a) when its value was varied by 50% of the default value. Heterotrophic anoxic yield equal to 50% of the model default value of 0.54 resulted in approximately 15% variation in TN concentration. Available literature on MBR for biological nutrient removal reported lower heterotrophic anoxic yield between 0.2 and 0.4 as compared to default value of 0.54 of the Biowin AS/AD model (Macomber et al., 2005). The other kinetic parameter that had an impact on TN effluent concentration was heterotrophic oxic endogenous decay rate (see Fig. 4b). The other factor that has an impact on TN effluent concentration was anoxic hydrolysis factor – heterotrophic with approximately 7% change (see Fig. 4d) while the rest in Fig. 4 show very little change in TN effluent concentration. The five kinetic parameters that had the most impact on effluent TP were: heterotrophic anoxic yield (approximately between 30% and 42% change in TP for 50% change in default value) (Fig. 4a), anaerobic hydrolysis factors of heterotrophs (approximately between 40% and 60% change in TP) (Fig. 4e), heterotrophic hydrolysis (approximately between 30% and 35% change in TP) (Fig. 4c), oxic endogenous decay rate for heterotrophs (approximately between 25% and 40% change in TP) (Fig. 4b) and oxic endogenous decay rate of PAOs (approximately 30% change in TP) (Fig. 4f). It appeared that effluent TP concentrations were greatly impacted and were more sensitive to changes to the kinetic parameters than TN effluent concentrations. Using the default values as the starting point and based on the findings of the sensitivity analysis, TSS for all three reactors were first calibrated by adjusting (increase or decrease) the endogenous decay rates (kd) for autotrophs, heterotrophs, and PAOs, heterotrophic anoxic yield along with maximum specific growth rates (mmax) to minimize the sum of square errors. Once the TSS for all three reactors had been calibrated, the kinetic parameters were further adjusted to calibrate the effluent quality. From Table 4, the calibrated specific mmax for autotrophs, heterotrophs, or PAOs, did not change for the SRTs tested. The oxic endogenous decay rates for PAOs, however, were found to increase with SRT from 0.07 d1 for 10 days SRT to 0.26 d1 for 75 days SRT in comparison to the default value of 0.1 d1 (Metcalf and Eddy, 2003). Similarly, the calibrated autotrophic and heterotrophic oxic endogenous decay rates (d1) increased with SRTs as shown in Table 4 and plotted in Fig. 5a and b. Also plotted in Fig. 5 (d–f) are the other calibrated kinetic parameters: anoxic hydrolysis factor, anaerobic hydrolysis factor and fermentation rates. TSS, sCOD, TN, NH3-N, NO 3 -N, and TP concentrations of the calibrated model are presented in Table 4. The calibrated model typically overpredicted the effluent sCOD, NO 3 -N, and TP and underpredicted TN for all SRTs except for 10 days SRT. The calibrated kinetic parameters for the four SRTs were fitted with a second-order or third-order polynomial model as shown in Fig. 5. The equations can be used to estimate new kinetic parameters for other SRTs between 10 and 75 days only which in turn can be used to estimate the effluent quality concentrations. To verify the calibrated model, the model was used to predict the experimental data for the MBR system operating at 35 days SRT. Results are summarized in Table 5.
3200
water research 44 (2010) 3192–3202
Kd oxic Auto (d-1)
a
1 y = 2E-05x 2 + 0.0011x + 0.1595 R² = 0.9982
0.8 0.6
0.33
0.4
0.17
0.25
0.20
0.2 0
Kd oxi c hetero (d-1)
b
0
10
20
30
40
50
60
70
80
1.0 0.8
0.72
0.65
0.55
0.53
0.6
y = -7E-07x 2 + 0.003x + 0.4983 R² = 0.999
0.4
0.2 0.0
Kd oxic PAO (d-1)
c
0
10
20
30
40
50
60
70
80
1.0 0.8
y = -1E-05x 2 + 0.0039x + 0.0308 R² = 0.9997
0.6
0.26
0.4 0.2
0.20
0.12
0.07
0.0 Anox. Hydrolysis Factor
d
Ana. Hydrolysis Factor
e
Ferm entation Factor
f
0
10
20
30
40
50
60
70
80
1.0 0.8
0.70
0.65
0.6 0.28
0.28
0.4
y = -1E-05x 3 + 0.0012x 2 - 0.0322x + 0.4929 R² = 1
0.2 0.0 0
10
20
30
40
50
60
70
80
1.0 0.80
0.8
0.90 0.50
0.6
0.50
y = -7E-06x 3 + 0.0009x 2 - 0.0247x + 0.6635 R² = 1
0.4 0.2 0.0 0
10
20
30
40
5.0
60
70
80
0.42
0.37
3.6
4.0
50
0.47
3.0 y = -1E-05x 3 + 0.0015x 2 - 0.0462x + 4.0212 R² = 1
2.0 1.0 0.0 0
10
20
30
40
50
60
70
80
Solids Residence Time (days)
Fig. 5 – Plot of calibrated kinetic parameters at different SRTs and polynominal correlations: (a) oxic endogenous decay rates for autotrophs, (b) oxic endogenous decay rates for heterotrophs, (c) oxic endogenous decay rates for PAOs, (d) anoxic hydrolysis factor, (e) anaerobic hydrolysis factor, and (f) fermentation rates. (Numbers on plots indicate values at different SRTs).
The calibrated model using the estimated kinetic parameters at 35 days predicted the effluent characteristics very well when compared to the default values. In addition, the predicted results showed that use of a dewatering device and returning the solids back into the oxic tank appeared to be suitable for modeling an MBR.
Modeling of the experimental results indicates that the kinetic parameters of the membrane biological systems were a function of SRT and that the kinetic parameters should be corrected in the model to include this variation. The maximum specific growth rates did not change with SRT indicating that nutrient availability was similar between SRTs. However, oxic
water research 44 (2010) 3192–3202
Table 5 – Model verification showing experimental effluent concentrations (±95% CI) and predicted concentrations at 35 days SRT using calibrated kinetic parameters.
Experimental Default Predicted
sCOD (mg/L)
TN (mg/L)
NH3-N (mg/L)
NO 3 -N (mg/L)
TP (mg/L)
29 3 33 31
8.6 0.5 13.0 8.9
0.4 0.1 0.2 0.3
7.2 0.3 11.7 7.6
3.2 0.3 6.3 3.4
endogenous decay rates for autotrophs, heterotrophs, and PAOs increased with SRT indicating that the viable portion of cells to TSS would decrease. This may mean that longer SRT may result in higher TP concentration in the mixed liquor as a result of endogenous decay of the viable cells.
4.
Conclusion
The results of the MBR with anaerobic, anoxic, and oxic tanks for SRTs ranging from 10 days to 75 days showed that: sCOD were sufficiently removed with more than 93% removal along with excellent nitrification at more than 98%. as for nitrogen removal, the system performed well with TN percent removals increasing with longer SRTs but peaking at 81% removal for both 50 and 75 day day SRT. TP percent removal was highest for 50 days SRT at approximately 80% but decreased to 60% at 75 days SRT indicating a possible limiting operating condition for PAOs and the lysis of cells at this SRT which may also introduce phosphorus back into the mixed liquor although the actual cause is not yet known. The experimental results imply that operating an MBR for phosphorus removal at very long SRTs may have some limitations. the sensitivity analysis results of the Biowin AS/AD model indicated that the heterotrophic anoxic yield, anaerobic hydrolysis factors of heterotrophs, heterotrophic hydrolysis, oxic endogenous decay rate for heterotrophs, and oxic endogenous decay rate of PAOs had the most impact on predicted effluent TP concentrations. Calibration of Biowin AS/AD model with the experimental data revealed that oxic endogenous decay rates increased with SRT and the higher anoxic and anaerobic hydrolysis factors at higher SRTs compensated for the higher endogenous decay rates up to 50 days but then the conditions became limited due to high endogenous decay rates. Polynomial correlations for the various kinetic parameters (such as anoxic and anaerobic hydrolysis factors for heterotrophs and oxic endogenous decay rates) with SRTs were derived. Using the estimated kinetic parameters from various polynomial correlations derived from the calibration of the Biowin model, the predicted results showed that the model was able to predict well the experimental effluent sCOD, TN and TP concentrations at 35 days SRT. The experimental results provided information on the possible limitation of operating an MBR for phosphorus removal at long SRTs and the kinetic parameters obtained by the
3201
modeling efforts can be used to model nutrient removal for MBRs systems with SRTs greater than 25 days. However, further experimental and modeling efforts are required to build the body of kinetic information for MBR systems under different recirculation conditions and configurations, and different HRTs of the various anaerobic, anoxic and oxic reactors.
Acknowledgements Postdoctoral fellowship awarded to Ertan Arslankaya by The Scientific and Technical Research Council of Turkey (TUBITAK).
references
Abegglen, C.A., Ospelt, M., Siegrist, H., 2008. Biological nutrient removal in a small-scale MBR treating household wastewater. Water Research 42 (1–2), 338–346. Adam, C., Kraume, M., Gnirss, R., Lesjean, B., 2003. Membrane bioreactor configurations for enhanced biological phosphorus removal. Water Science and Technology: Water Supply 3 (5–6), 237–244. APHA, AWWA, WEF, 2002. Standard Methods for the Examination of Water and Wastewater, 20th ed. American Public Health Association/American Water Works Association/Water Environment Federation, Washington DC, USA. Delgado, S., Diaz, F., Villarroel, R., Vera, L., Diaz, R., Elmaleh, S., 2002. Influence of biologically treated wastewater quality on filtration through a hollow-fibre membrane. Desalination 146 (1–3), 459–462. Ersu, C.B., Ong, S.K., Arslankaya, E., Brown, P., 2008. Comparison of recirculation configurations for biological nutrient removal in a membrane bioreactor. Water Research 42 (6–7), 1651–1663. Fleischer, E.J., Broderick, T.A., Daigger, G.T., Fonseca, A.D., Holbrook, R.D., Murthy, S.N., 2005. Evaluation of membrane bioreactor process capabilities to meet stringent effluent nutrient discharge requirements. Water Environment Research 77 (2), 162–178. Ha, J.H., Ong, S.K., Rao, S., 2007. Nitrification and denitrification in partially aerated biological aerated filter (BAF) with dual size sand media. Water Science and Technology 55 (1–2), 9–17. Hasar, H., Kinaci, C., Unlu, A., Ipek, U., 2002. Role of intermittent aeration in domestic wastewater treatment by submerged membrane activated sludge system. Desalination 142 (3), 287–293. Kishino, H., Ishida, H., Iwabu, H., Nakano, I., 1996. Domestic wastewater reuse using a submerged membrane bioreactor. Desalination 106 (1–3), 115–119. Lesjean, B., Gnirss, R., Adam, C., 2002. Process configurations adapted to membrane bioreactors for enhanced biological phosphorus and nitrogen removal. Desalination 149, 217–224. Macomber, J., Cicek, N., Suidan, M.T., Davel, J., Ginestet, P., Audic, J.M., 2005. Biological kinetic data evaluation of an activated sludge system coupled with an ultrafiltration membrane. Journal of Environmental Engineering 131 (4), 579–586. Metcalf, Eddy, 2003. Wastewater Engineering, Treatment, Disposal and Reuse, fourth ed. McGraw Hill Inc., NY. Ogoshi, M., Suzuki, Y., 2000. Application of MBR to an easily installed municipal wastewater treatment plant. Water Science and Technology 41 (10–11), 287–293. Rosenberger, S., Kru¨ger, U., Witzig, R., Manz, W., Szewzyk, U., Kraume, M., 2000. Operation of different membrane bioreactors: experimental results and physiological state of
3202
water research 44 (2010) 3192–3202
the microorganisms. Water Science and Technology 41 (10–11), 269–277. Rosenberger, S., Kru¨ger, U., Witzig, R., Manz, W., Szewzyk, U., Kraume, M., 2002. Performance of a bioreactor with submerged membranes for oxic treatment of municipal wastewater. Water Research 36 (2), 413–420. Shin, H., Kang, S., Lee, C., Lim, J., 2004. Performance of a pilot scale membrane bioreactor coupled with SBR (SM-SBR) – experiences in seasonal temperature changes. Water Science and Technology: Water Supply 4 (1), 135–142. Stephenson, T., Judd, S., Jefferson, B., Brindle, K., 2000. Membrane Bioreactors for Wastewater Treatment. IWA Publishing, London, UK.
Ubukata, Y., Takii, S., 1998. Some physiological characteristics of a phosphate removing bacterium, Microlunatus phosphovorus, and a simplified isolation and identification method for phosphate-removing bacteria. Water Science and Technology 38 (1), 149–157. Urbain, V., Mobarry, B., De Silva, V., Stahl, D.A., Rittmann, B.E., Manem, J., 1998. Integration of performance, molecular biology and modeling to describe the activated mixed liquor process. Water Science and Technology 37 (4–5), 223–229. Visvanathan, C., Ben Aim, R., Parameshwaran, K., 2000. Membrane separation bioreactors for wastewater treatment. Critical Reviews in Environmental Science and Technology 30 (1), 1–48.
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Impact of solids residence time on biological nutrient removal performance of membrane bioreactor Cagatayhan Bekir Ersu a, Say Kee Ong b,*, Ertan Arslankaya c, Yong-Woo Lee d a
Cukurova University, Environmental Engineering Department, Adana, Turkey Iowa State University, Civil, Construction and Environmental Engineering Department, 490 Town Engineering, Ames IA 50011, USA c Yildiz Technical University, Environmental Engineering Department, Istanbul, Turkey d Hanyang University, Department of Applied Chemistry, Seoul, Republic of Korea b
article info
abstract
Article history:
Impact of long solids residence times (SRTs) on nutrient removal was investigated using
Received 18 October 2009
a submerged plate-frame membrane bioreactor with anaerobic and anoxic tanks. The
Received in revised form
system was operated at 10, 25, 50 and 75 days SRTs with hydraulic retention times (HRTs) of
31 December 2009
2 h each for the anaerobic and anoxic tanks and 8 h for the oxic tank. Recirculation of oxic
Accepted 22 February 2010
tank mixed liquor into the anaerobic tank and permeate into the anoxic tank were fixed at
Available online 1 March 2010
100% each of the influent flow. For all SRTs, percent removals of soluble chemical oxygen demand were more than 93% and nitrification was more than 98.5% but total nitrogen
Keywords:
percent removal seemed to peak at 81% at 50 days SRT while total phosphorus (TP) percent
Nitrification
removal showed a deterioration from approximately 80% at 50 days SRT to 60% at 75 days
Denitrification
SRT. Before calibrating the Biowin model to the experimental data, a sensitivity analysis of
Phosphorus uptake
the model was conducted which indicated that heterotrophic anoxic yield, anaerobic
SRT
hydrolysis factors of heterotrophs, heterotrophic hydrolysis, oxic endogenous decay rate for
Modeling
heterotrophs and oxic endogenous decay rate of PAOs had the most impact on predicted effluent TP concentration. The final values of kinetic parameters obtained in the calibration seemed to imply that nitrogen and phosphorus removal increased with SRT due to an increase in anoxic and anaerobic hydrolysis factors up to 50 days SRT but beyond that removal of phosphorus deteriorated due to high oxic endogenous decay rates. This indirectly imply that the decrease in phosphorus removal at 75 days SRT may be due to an increase in lysis of microbial cells at high SRTs along with the low food/microorganisms ratio as a result of high suspended solids in the oxic tank. Several polynomial correlations relating the various calibrated kinetic parameters with SRTs were derived. The Biowin model and the kinetic parameters predicted by the polynomial correlations were verified and found to predict well the effluent water quality of the MBR at 35 days SRT. ª 2010 Elsevier Ltd. All rights reserved.
1.
Introduction
Membrane bioreactors (MBRs) have several operating advantages over conventional biological treatment processes
(Visvanathan et al., 2000). MBR can be operated efficiently at high-suspended solids concentrations, long solids residence times (SRTs) and short hydraulic retention times (HRTs) with excellent carbonaceous organic matter removal and
* Corresponding author. Tel.: þ1 515 294 3927; fax: þ1 515 294 8216. E-mail address: [email protected] (S.K. Ong). 0043-1354/$ – see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.02.036
water research 44 (2010) 3192–3202
nitrification without the concern of poor sludge settling since the membrane provides an excellent separation barrier (Stephenson et al., 2000). With stricter nutrient discharge requirements by regulatory agencies, MBR, with many advantages, can be modified to effectively remove nutrients (nitrogen and phosphorus) by using various anaerobic, anoxic, and oxic manipulations (Ersu et al., 2008; Abegglen et al., 2008). The high suspended solids concentrations provide the added benefits of minimizing the phosphorus content in the solids while the long SRTs allow growth of bacteria with slow growth rates such as nitrifying bacteria (Hasar et al., 2002). Hence, near complete nitrification is observed in MBRs provided oxygen and hydraulic retention requirements are met (Kishino et al., 1996; Delgado et al., 2002; Shin et al., 2004). In the case of phosphorus uptake, the phosphate accumulating organisms (PAOs) utilized their stored energy source under aerobic conditions (Ubukata and Takii, 1998). However, the long SRTs of MBRs may impact the luxurious uptake of phosphorus along with the survival of the PAOs due to the finite amount of stored substrate (Rosenberger et al., 2000). Abegglen et al. (2008) achieved 90% and 70% removal of total nitrogen (TN) and total phosphorus (TP) using an anaerobic/anoxic MBR treating household wastewater. Using a 6-stage MBR pilot plant operated at 23 days SRT, Fleischer et al. (2005) achieved TN concentrations of less than 3 mg/L and a TP concentration of less than 0.1 mg/L by a combination of biological and chemical (alum) treatment. Adam et al. (2003) studied the treatment of municipal wastewater using benchscale submerged MBR with anaerobic-anoxic-oxic configuration and obtained removals of 96.4% chemical oxygen demand (COD), 86.8% TN, 98.8% ammonia-nitrogen (NH3-N), and 99% TP for 15 days SRT and anaerobic, anoxic, and oxic hydraulic retention times of 3.7 h, 8.5 h, and 8.8 h, respectively. The same researchers also studied a similar MBR setup operated with a post-denitrification mode and obtained similar results with 95.1% COD, 89.6% TN, 99.9% NH3-N, and 99.2% TP. Lesjean et al. (2002) studied both configurations described by Adam et al. (2003) but at a pilot-scale with a total HRT of 18 h and an SRT of 26 days and found 81.9% TN and 99.3% TP removal in pre-denitrification mode as compared to 94.1% TN and 99.2% TP for post-denitrification mode. Rosenberger et al. (2002) applied an infinite SRT for a pilot-scale MBR (with an oxic tank only) and achieved 82% TN removal with 99.9% total kjeldahl nitrogen (TKN) removal for a municipal wastewater suggesting that long SRTs do not have an adverse effect on nitrogen removal. Similarly, Ogoshi and Suzuki (2000) treated synthetic wastewater in a pilot-scale MBR at over 400 days SRT and reported 95% removal for 5-day biochemical oxygen demand (BOD5), 80% removal for COD, and over 90% removal for TN. Urbain et al. (1998) reported COD, ammonia (NH3), and TP removals of 90%, 99%, and 50%, respectively for SRTs between 5 and 20 days. Based on the literature, it is possible to obtain TN and TP removals greater than 90%, respectively, for HRTs ranging from 6 to 18 h and typical SRTs of 10–25 days (Metcalf and Eddy, 2003). MBR systems with anaerobic, anoxic, and oxic manipulations have been shown to achieve high level biological nutrient removal but there are no studies on phosphorus removal for long SRTs over 25 days.
3193
The main objective of this study was to evaluate the impact of long SRTs (25 days and over) on biological nutrient removal performance of a submerged MBR with an integration of anaerobic, anoxic, and oxic tanks. Various kinetic parameters for biological nutrient removal over a range of SRTs were analyzed using the Biowin AS/AD model to investigate the important kinetic parameters that may be impacted due to the long SRTs of the MBRs. The study provides operating and kinetic information on nutrient removal for design and operation of MBRs systems with SRTs greater than 25 days. This study is probably the first to model an MBR with anaerobic, anoxic, and oxic tanks for nutrient removal with long SRTs.
2.
Materials and methods
2.1.
Experimental setup
Experiments were conducted using a bench-scale MBR configured with an anaerobic and anoxic tank as shown in Fig. 1. The synthetic wastewater (see Table 1) was fed into the bioreactor using a Masterflex Model 8800 pump (Hamden, CT). Sources of nitrogen were organic in nature (urea, nutrient broth and Isomil, a soy-based infant formula (Abbott Laboratories, Abbott Park, IL)) while sources of phosphorus were from both inorganic sources (biphosphate) and organic sources (nutrient broth and Isomil). The bioreactor was made of Plexiglass and had a total maximum operating volume of 16 L (4 L for the anaerobic, 4 L for anoxic and 8 L for oxic tanks) with an oxic tank dimension of 10 cm wide by 25 cm long and 50 cm deep. The cylindrical anaerobic and anoxic tank each has a diameter of 15 cm and a depth of 50 cm with variable operating volumes between 1 and 4 L by using various outlet ports located on the side of the tanks. To create anaerobic conditions, the anaerobic tank was sealed with a Plexiglass cover using six bolts and a rubber gasket and the contents were mixed gently with a variable-speed magnetic stirrer while the anoxic tank was opened to the atmosphere and was also gently mixed with a variable-speed magnetic stirrer. Wastewater flowed by gravity from the anaerobic tank to the anoxic tank and into the oxic tank. Air was introduced via air diffusers placed at the bottom of the oxic tank for microbial metabolism and to control membrane fouling. The air and liquid flow rates were measured using Gilmont ball flow meters (Barrington, IL). Experiments were conducted at room temperature at 22.5 1 C. For solid/liquid separation purposes, a double-sided plateframe cellulose membrane (Kubota Co., Osaka, Japan) with a nominal pore size of 0.2 mm and 0.15 m2 of filtration area was used. Constant permeate flux was maintained by a Cole– Parmer Model 7553-70 pump (Vernon Hills, IL) which was placed at the permeate outlet. Mixed liquor recirculation rates were controlled by another Masterflex Model 8800 pump (Hamden, CT). The plate-frame module was operated for 9 min with permeate fluxes ranging between 8 and 28 L/m2/h, followed by 1 min of idle period. No backwashing of the membrane was required. The operational cycles of filtration run and idle period were controlled using a ChronTrol microprocessing timer (ChronTrol Corp., San Diego, CA). The reactor was equipped with water level sensors to maintain
3194
water research 44 (2010) 3192–3202
Anaerobic (Slow mix) Anoxic (Complete mix)
C
Permeate PI
F
Level Sensors
D
E
Membrane
B
B
Air Mixed Liquor Recycle
G Permeate Recycle
G
Mixed Liquor Wastin g Permeate Recycle
Feed
A
Feed
A
Fig. 1 – Membrane bioreactor for biological nutrient removal with anaerobic and anoxic tanks (Sampling points: A – Synthetic wastewater feed; B – Anaerobic influent; C – Anaerobic effluent; D – Anoxic effluent; E – Mixed liquor; F – Permeate; G – Waste mixed liquor).
Table 1 – Ingredients and composition of the synthetic wastewater. Ingredient Calcium sulfate Magnesium sulfate Ferric chloride Sodium biphosphate Sodium bicarbonate Potassium chloride Urea Nutrient broth Sodium citrate Isomil (0.8% by vol.) Composition
Concentration (mg/L) 40 4 3 75 63 5 42 250 500 20 mL
Average concentration (mg/L) 95% CI pH 7.7 0.2 Chemical Oxygen Demand (COD) 555 9.1 soluble chemical oxygen demand (sCOD) 502 11.1 342 11 Biochemical Oxygen Demand (BOD5) Total nitrogen (TN) 43.3 1.9 Ammonia nitrogen (NH3-N) 22.7 1.8 0.2 0.1 Nitrite-nitrogen (NO 2 -N) 0.5 0.1 Nitrate-nitrogen (NO 3 -N) Organic nitrogen 20.0 1.5 Total phosphorus (TP) 13.6 0.9 Suspended solids (SS) 71.7 6.4
a constant volume in the bioreactor. Mixed liquor wasting was done manually on a daily basis to provide the desired SRT. The bioreactor was initially inoculated with a mixture of biomass from a bench-scale biological nutrient removal system (at 25 days SRT) and a nearby activated sludge municipal wastewater treatment plant. The oxic bioreactor was operated on an 8-hour oxic batch mode for about 5 cycles to acclimatize the microorganisms before the bioreactor was switched to continuous mode. The bench-scale MBR was operated under 100% recirculation of mixed liquor to the anaerobic tank and 100% recirculation of nitrate-rich permeate to the anoxic tank. The recirculation configuration was found to be the better of several recirculation configurations tested (Ersu et al., 2008). The feed water flow rate was maintained at 1 L/h giving hydraulic retention times (HRT) of 2, 2, and 8 h for the anaerobic, anoxic, and oxic tanks, respectively. The SRT in the oxic tank was initially maintained at 25 days with a biomass concentration of approximately 7500 mg/L in the oxic tank.
2.2.
Varying solids residence times
To study the impact of SRT on biological nutrient removal, the MBR was operated initially at 25 days SRT (see Ersu et al., 2008) followed by 50 days and 75 days SRT. The reactor was then operated at 10 days and 35 days SRT. For each change in SRT, the bioreactor was operated until steady state conditions were
3195
water research 44 (2010) 3192–3202
Fig. 2 – BioWin flow diagram of membrane bioreactor for biological nutrient removal.
achieved which ranged from three weeks for shorter SRTs to as much as eight weeks for longer SRTs. Steady state conditions were assumed to be achieved when the most recent three measurements of COD, TN and TP concentrations were within 10% (Ha et al., 2007; Ersu et al., 2008). Wastewater samples were collected and analyzed, on a regular basis (two to three times per week), from various sampling points as shown in Fig. 1. Anaerobic, anoxic, and oxic (permeate) effluents as well as influent wastewater were analyzed. Wastewater samples for analysis were obtained by opening the sampling valves and collecting a portion of the flow with a glass collection flask for half hour to obtain 1⁄2 -hour samples. Steady state results of the analyses presented here were the average of at least three samples collected at different times along with their 95% confidence intervals. Transmembrane pressure (TMP), pH, oxidation–reduction potential (ORP), and dissolved oxygen (DO) were monitored throughout the experimental study using dual-scale pressure gauge, pH/ORPmeter (Orion Model 260), and DO meter (Orion Model 830), respectively. NH3-N was measured using an ammonia meter (Orion Model 290A). Total suspended solids and volatile suspended solids were analyzed using Standard Methods 2540D and 2540E, respectively, (APHA et al., 2002) while COD and sCOD were analyzed using Hach Colorimetric Method 8000 (Hach Company, Loveland, CO). TN, TP and nitrate-nitrogen (NO 3 -N) were analyzed using Hach persulfate digestion Method 10071, Hach molybdovanadate Method 10127, and Hach cadmium reduction Method 8039, respectively (Hach Company, Loveland, CO).
Table 2 – Carbonaceous and nutrient fractions of the synthetic wastewater. Abbreviation Fbs
Fac Fxsp
Fus Fup Fna Fnox Fnus FupN
Fpo4 FupP
FZbh FZbm FZba
2.3. Kinetic modeling for biological nutrient removal in MBR using BioWin The Biowin AS/AD model (Envirosim Inc., Ontario, Canada) was calibrated using the experimental results of the MBR at different SRTs. To mimic the MBR, the Biowin AS/AD model simulates the MBR as a combination of an aerated tank and a sludge dewatering device with all the sludge returning back into the oxic reactor (see Fig. 2). Since the membrane of the
FZbp FZbpa FZbam FZbhm
Fraction Readily biodegradable (including acetate) [g COD/g of total COD] Acetate [g COD/g of readily biodegradable COD] Non-colloidal slowly biodegradable [g COD/g slowly degradable COD] Unbiodegradable soluble [g COD/g of total COD] Unbiodegradable particulate [g COD/g of total COD] Ammonia [g NH3-N/g TKN] Particulate organic nitrogen [g N/g Organic N] Soluble unbiodegradable TKN [g N/g TKN] N:COD ratio for unbiodegradable particulate COD [g N/g COD] Phosphate [g PO4-P/gTP] P:COD ratio for influent unbiodegradable particulate COD [g P/g COD] Non-PAO heterotrophs [g COD/g of total COD] Anoxic methanol utilizers [g COD/g of total COD] Autotrophs [g COD/g of total COD] PAOs [g COD/g of total COD] Propionic acetogens [g COD/g of total COD] Acetoclastic methanogens [g COD/g of total COD] H2-utilizing methanogens [g COD/g of total COD]
a Estimated using actual data.
Default Input 0.20
0.25a
0.15
0.15
0.75
0.70a
0.05
0.05
0.13
0.05a
0.66 0.50
0.50a 0.50
0.00
0.00
0.035
0.035
0.50 0.011
0.80a 0.011
1.0E-4
1.0E-4
1.0E-4
1.0E-4
1.0E-4
1.0E-4
1.0E-4 1.0E-4
1.0E-4 1.0E-4
1.0E-4
1.0E-4
1.0E-4
1.0E-4
3196
water research 44 (2010) 3192–3202
MBR acts almost like an impermeable barrier to the mixed liquor solids, use of a dewatering device to simulate the MBR would be adequate as the dewatering device ensures that all of the mixed liquor solids are retained within the oxic tank. The model assumed the following kinetic reactions: aerobic hydrolysis of slowly biodegradable substrate; anoxic hydrolysis of slowly biodegradable substrate; anaerobic hydrolysis of slowly biodegradable substrate; aerobic growth of heterotrophs on fermentable substrates; aerobic growth of heterotrophs on fermentation products; denitrification with fermentable substrates; denitrification with fermentation products; fermentation of complex soluble substrate to propionate, acetate, CO2; lysis of heterotrophs; storage of poly-hydroxy-alkanoates (PHA) by PAOs; aerobic storage of poly-P by PAOs; anoxic storage of poly-P by PAOs; aerobic growth of PAOs on PHA; anoxic growth of PAOs on PHA; aerobic growth of autotrophs; and lysis of autotrophs. Various carbonaceous and nutrient fractions needed by the model were estimated from the composition of the feed water. If the water quality parameters were not tested, the default values of the model were assumed (see Table 2). Impact of each kinetic parameter for autotrophs, heterotrophs, and PAOs in the Biowin AS/AD model on effluent TN and TP concentrations was assessed by using a sensitivity analysis starting with the default value of the kinetic parameters provided by the model and varying the default value by up to 50%. Based on the trends and impact of each kinetic parameter on effluent TN and TP concentrations, the kinetic parameters were selected accordingly and varied to calibrate the model against the total suspended solids (TSS) of the reactors followed by the effluent water quality by minimizing the sum of square errors. The calibrated model was then verified using the experimental data for a SRT of 35 days.
3.
Results and discussion
3.1.
Wastewater characteristics
BOD5 was approximately 70% of the total COD for the synthetic wastewater used while the soluble COD (sCOD) was about 85% of the total COD. The COD:TN:TP ratio was approximately 100:8:2. About 52% of the TN was due to NH3-N while the organic nitrogen was approximately 47% of the total nitrogen. For all SRTs, the ORP was maintained, on the average, below 340 mV in the anaerobic tank and 235 mV in the anoxic tank while ORPs in the oxic tank were approximately 100 mV (see Table 3). The TMPs were consistently kept below 15 kPa. In the oxic tank, the DO was kept above 2 mg/L while the DO in the anoxic tank was less than 0.3 mg/L. The pH in the anoxic and oxic tanks was between 7 and 8. Suspended solids concentrations in the tanks increased proportionally with SRT (see Table 3). TSS in the aerobic tank for 50 days and 75 days SRTs were approximately 2 and 2.5 times higher, respectively than the TSS for 25 days SRT.
3.2.
Impact of solids residence time on MBR performance
Fig. 3a presents the sCOD concentrations of the anaerobic effluent, anoxic effluent, and oxic effluent of the MBR for different SRTs (10, 25, 50, and 75 days). On the average, more than 93% of the sCOD was removed for all SRTs and the sCOD concentrations of the final effluent were between 26 and 32 mg/L with a mean value of 29 mg/L. There were no significant differences in the mean sCOD percent removals for the SRTs tested. Advantages of the MBR system as compared to conventional biological treatment systems include the ability
Table 3 – Operating wastewater conditions (average values D95% CI) in anaerobic, anoxic, and oxic tanks under varying SRTs. ORP (mV)
DOa (mg/L)
TSS (mg/L)
Anaerobic 10 25 50 75
2844 34 7350 99 13,897 439 18,825 50
6.8 6.9 6.8 7.0
0.2 0.2 0.1 0.2
23.0 23.0 22.0 23.0
0.5 0.5 0.5 0.5
369 346 364 391
13 23 15 14
0.0 0.0 0.0 0.0
0 0 0 0
Anoxic 10 25 50 75
2126 3505 7300 9530
7.0 7.3 7.2 7.3
0.2 0.2 0.2 0.2
23.0 23.0 23.0 23.0
0.5 0.5 0.5 0.5
232 236 267 291
13 8 6 4
0.1 0.3 0.1 0.1
0.1 0.1 0.1 0.1
Oxic 10 25 50 75
4546 74 7400 60 13265 560 18,275 310
7.2 7.5 7.3 7.1
0.2 0.2 0.2 0.1
23.0 23.0 22.8 22.0
0.2 0.5 0.5 0.5
3.7 2.9 2.2 2.0
0.1 0.3 0.2 0.2
a Dissolved oxygen.
29 120 170 148
pH
Temp. ( C)
SRT (days)
102 4 108 6 82 6 77 4
3197
water research 44 (2010) 3192–3202
mg sCOD/L
a
Anaerobic - Sampling Point C Anoxic - Sampling Point D Oxic - Sampling Point F
200 150
100
187 153
50
108
107
27
140
32
98
57
87
26
27
0
10 days
25 days
50 days
75 days
30
b
25
mg T N/L
20 15
10
21.4
17.7
5
10.1
21.1
16.8
13.5
20.9 16.2
9.3
15.8 8.4
8.3
0
10 days
mg NH 3-N/L
c
18 16 14 12 10 8 6 4 2 0
25 days
8.3
8.2
d
7.8
25 days
7.8
0.3
0.2
0.1
10 days
75 days
14.2
13.5
12.2
11.8
50 days
50 days
0.3
75 days
12
mg NO 3- -N/L
10 8 6
4
9.7 2.2
7.3
2.6
2
0.5
0.5
6.7
6.6 0.3
0.4
0.2
0.2
0
10 days
e
25 days
50 days
75 days
35 30
mg T P/L
25
20 15
10
30.8
15.6 3.4
5
11.3
8.7
3.5
14.4
10.1
14.3 2.8
5.4
10.4
0
10 days
25 days
50 days
75 days
Fig. 3 – Effluent concentrations of (a) sCOD, (b) total nitrogen, (c) ammonia-nitrogen, (d) nitrate-nitrogen, (e) total phosphorus for anaerobic, anoxic and oxic reactors and different SRTs (error bars – 95% CI of three or more samples taken at different times under steady state conditions).
to maintain high biomass concentrations resulting in sCOD removals equal or higher than conventional systems and a smaller reactor size. With the MLSS concentrations varying from 7500 to about 18,000 mg/L for 25–75 days SRT, the mass loadings for the aerobic tank were between 0.07 and 0.20 kg COD/kg MLSS/d which were lower than the permissible mass loading rates for conventional systems in the range of 0.3–0.6 kg COD/kg MLSS/d. This means that the MBR can take a higher influent COD load than a conventional system (Metcalf and Eddy, 2003; Stephenson et al., 2000). TN concentrations in the oxic effluent decreased with the SRT of the MBR as shown in Fig. 3b. Percent TN removals
were 65.7 3.4%, 77.9 1.0%, 80.6 0.3% and 81.0 0.3% for 10, 25, 50 and 75 days SRT, respectively. At 25 days, average NO 3 -N concentration was 0.5 mg/L for both anaerobic and anoxic tanks while, at 75 days, average NO 3 -N concentration was as low as 0.2 mg/L, demonstrating denitrification in anaerobic and anoxic tanks for SRTs greater than 10 days (see Fig. 3d). Average NO 3 -N concentrations in the oxic tank effluents decreased with longer SRT and reached an asymptotic value of approximately 6.6 mg/L for 50 days or more SRT. Nitrate concentrations constituted about 78–80% of the TN concentrations of the final effluent indicating there is still some room for improving denitrification by changing
3198
water research 44 (2010) 3192–3202
f
40 30 20 10 0 -10 -20 -30 -40 -50
TN TP
-60
-40
-20
0
20
40
30 20 10
0 -10 -20 -30
-40
60
-60
% Variation in Heterotrophic Anoxic Yield
-40
-20
0
20
40
40
60
0 -5 -10 -15 -60
h % Sensitivity in Effluent
% Sensitivity in Effluent
20
5
60
40 30 20 10 0 -10 -20 -30 -40
-40 -20 0 20 40 60 % Variation in Anaerobic Endogenous Decay Rate-PAO
15 10 5
0 -5
-10 -15 -20
-60
% Sensitivity in Effluent
0
10
% Variation in Oxic Endogenous Decay Rate-Heterotrophic
d
-20
g 15
30 20 10 0 -10 -20 -30 -40 -50 -60
c
-40
% Variation in Oxic Endogenous Decay Rate-PAO
% Sensitivity in Effluent
% Sensitivity in Effluent
b
% Sensitivity in Effluent
% Sensitivity in Effluent
a
-40
-20
0
20
40
60
-60
% Variation in Hydrolysis Rate (AS)-Heterotrophic
-40 -20 0 20 40 % Variation in Anoxic Growth Factor-PAO
60
15 10 5 0 -5 -10 -15 -60
-40
-20
0
20
40
60
% Variation in Anoxic Hydrolysis Factor-Heterotrophic
% Sensitivity in Effluent
e
80
60 40 20 0 -20
-40 -60 -100 -80 -60 -40 -20 0 20 40 60 80 100 % Variation in Anaerobic Hydrolysis Factor-Heterotrophic
Fig. 4 – Sensitivity analysis of various kinetic parameters on total nitrogen and total phosphorus effluent concentrations (a) Heterotrophic anoxic yield, (b) oxic endogenous decay rate – heterotrophic, (c) Hydrolysis rate – heterotrophic, (d) Anoxic hydrolysis factor – heterotrophic, (e) Anaerobic hydrolysis factor – heterotrophic, (f) Oxic endogenous decay rate – PAO, (g) Anaerobic endogenous decay rate – PAO, (h) Anoxic growth factor – PAO.
the recirculation flow rates. With 0.2–0.3 mg NH3-N/L in the final effluent, the MBR demonstrated excellent nitrification for all SRTs (see Fig. 3c) with more than 98% NH3-N removal. Final effluent phosphorus concentrations were as low as 2.8 mg/L for 50 days SRT as shown in Fig. 3e. Percent TP removals increased from 70.8 7.1% at 10 days SRT to 79.6 0.4% at 50 days SRT and then decreased to 60.1 0.42% for 75 days SRT. Deterioration of TP removal for 75 days SRT indicated that there might be a limitation in PAOs activity after a certain SRT and may also be due to lysis of cells at long SRTs.
For all SRTs, the higher TP concentrations in the anaerobic effluent (example, 30.8 mg/L for 10 day SRT) as compared to the feed water indicated that there was phosphorus release by PAOs as expected in the anaerobic tank.
3.3.
Modeling biological nutrient removal with BioWin
The impact of various kinetic parameters predicting the effluent TN and TP concentrations in the sensitivity analysis are presented in Fig. 4a–h. Of the various kinetic parameters,
– 8,590,304 56,850 5.4 8.5 5.8 6.7 8.1 7.7 0.3 0.3 0.3 8.3 10.4 8.1 a AA – anaerobic; AO – anoxic; 0 – oxic, mmax – maximum specific growth rate; kd oxic – oxic endogenous decay rate. b Sum of square errors. c Model assumes all nitrite converted to nitrate.
27 35 27 18,275 16,209 18,220 9530 8506 9715 18,825 17,016 18,965 – 0.10 0.26 – 0.95 1.4 – 0.62 0.72 – 3.20 2.80 – 0.90 1.20 Experimental (75 d) Default Calibrated
– 0.17 0.33
– 6,558,989 50,188 2.8 7.1 3.1 6.6 7.9 7.4 0.3 0.3 0.3 8.4 9.8 7.9 26 36 28 13,265 11,672 13,187 7300 6132 7498 13,897 12,267 13,967 – 0.10 0.20 – 0.95 1.4 – 0.62 0.65 – 3.20 2.80 – 0.90 1.20 Experimental (50 d) Default Calibrated
– 0.17 0.25
– 272,344 13,849 3.5 3.9 3.7 7.3 7.6 7.3 0.2 0.3 0.3 9.3 9.2 8.8 32 35 34 7400 6901 7296 3505 3635 3532 7350 7272 7398 – 0.10 0.12 – 0.95 1.4 – 0.62 0.55 – 3.20 2.80 – 0.90 1.20 Experimental (25 d) Default Calibrated
– 0.17 0.20
3.4 0.2 2.3 9.7 7.1 11.6 0.1 0.3 0.3 13.5 8.5 13.2 27 32 29 4546 4099 4526 2126 1796 1973 2844 2708 2976 – 0.10 0.07 – 0.95 1.4 – 0.62 0.53 – 3.20 2.80 – 0.17 0.17 – 0.90 1.20 Experimental (10 d) Default Calibrated
TN (mg/L) sCOD (mg/L) TSSAO (mg/L) TSSAAa (mg/L) kd oxic (d1) mmax (d1)
kd oxic (d1)
mmax (d1)
kd oxic (d1)
mmax (d1)
PAO Heterotrophs Autotrophs
Table 4 – Calibration for total suspended solids in MBR and effluent quality for varying SRTs.
TSS0 (mg/L)
Wastewater quality
NH3-N (mg/L)
NO-3-Nc (mg/L)
TP (mg/L)
– 327,344 41,319
S SSEb
water research 44 (2010) 3192–3202
3199
heterotrophic anoxic yield gave the largest change in the predicted effluent TN concentrations (see Fig. 4a) when its value was varied by 50% of the default value. Heterotrophic anoxic yield equal to 50% of the model default value of 0.54 resulted in approximately 15% variation in TN concentration. Available literature on MBR for biological nutrient removal reported lower heterotrophic anoxic yield between 0.2 and 0.4 as compared to default value of 0.54 of the Biowin AS/AD model (Macomber et al., 2005). The other kinetic parameter that had an impact on TN effluent concentration was heterotrophic oxic endogenous decay rate (see Fig. 4b). The other factor that has an impact on TN effluent concentration was anoxic hydrolysis factor – heterotrophic with approximately 7% change (see Fig. 4d) while the rest in Fig. 4 show very little change in TN effluent concentration. The five kinetic parameters that had the most impact on effluent TP were: heterotrophic anoxic yield (approximately between 30% and 42% change in TP for 50% change in default value) (Fig. 4a), anaerobic hydrolysis factors of heterotrophs (approximately between 40% and 60% change in TP) (Fig. 4e), heterotrophic hydrolysis (approximately between 30% and 35% change in TP) (Fig. 4c), oxic endogenous decay rate for heterotrophs (approximately between 25% and 40% change in TP) (Fig. 4b) and oxic endogenous decay rate of PAOs (approximately 30% change in TP) (Fig. 4f). It appeared that effluent TP concentrations were greatly impacted and were more sensitive to changes to the kinetic parameters than TN effluent concentrations. Using the default values as the starting point and based on the findings of the sensitivity analysis, TSS for all three reactors were first calibrated by adjusting (increase or decrease) the endogenous decay rates (kd) for autotrophs, heterotrophs, and PAOs, heterotrophic anoxic yield along with maximum specific growth rates (mmax) to minimize the sum of square errors. Once the TSS for all three reactors had been calibrated, the kinetic parameters were further adjusted to calibrate the effluent quality. From Table 4, the calibrated specific mmax for autotrophs, heterotrophs, or PAOs, did not change for the SRTs tested. The oxic endogenous decay rates for PAOs, however, were found to increase with SRT from 0.07 d1 for 10 days SRT to 0.26 d1 for 75 days SRT in comparison to the default value of 0.1 d1 (Metcalf and Eddy, 2003). Similarly, the calibrated autotrophic and heterotrophic oxic endogenous decay rates (d1) increased with SRTs as shown in Table 4 and plotted in Fig. 5a and b. Also plotted in Fig. 5 (d–f) are the other calibrated kinetic parameters: anoxic hydrolysis factor, anaerobic hydrolysis factor and fermentation rates. TSS, sCOD, TN, NH3-N, NO 3 -N, and TP concentrations of the calibrated model are presented in Table 4. The calibrated model typically overpredicted the effluent sCOD, NO 3 -N, and TP and underpredicted TN for all SRTs except for 10 days SRT. The calibrated kinetic parameters for the four SRTs were fitted with a second-order or third-order polynomial model as shown in Fig. 5. The equations can be used to estimate new kinetic parameters for other SRTs between 10 and 75 days only which in turn can be used to estimate the effluent quality concentrations. To verify the calibrated model, the model was used to predict the experimental data for the MBR system operating at 35 days SRT. Results are summarized in Table 5.
3200
water research 44 (2010) 3192–3202
Kd oxic Auto (d-1)
a
1 y = 2E-05x 2 + 0.0011x + 0.1595 R² = 0.9982
0.8 0.6
0.33
0.4
0.17
0.25
0.20
0.2 0
Kd oxi c hetero (d-1)
b
0
10
20
30
40
50
60
70
80
1.0 0.8
0.72
0.65
0.55
0.53
0.6
y = -7E-07x 2 + 0.003x + 0.4983 R² = 0.999
0.4
0.2 0.0
Kd oxic PAO (d-1)
c
0
10
20
30
40
50
60
70
80
1.0 0.8
y = -1E-05x 2 + 0.0039x + 0.0308 R² = 0.9997
0.6
0.26
0.4 0.2
0.20
0.12
0.07
0.0 Anox. Hydrolysis Factor
d
Ana. Hydrolysis Factor
e
Ferm entation Factor
f
0
10
20
30
40
50
60
70
80
1.0 0.8
0.70
0.65
0.6 0.28
0.28
0.4
y = -1E-05x 3 + 0.0012x 2 - 0.0322x + 0.4929 R² = 1
0.2 0.0 0
10
20
30
40
50
60
70
80
1.0 0.80
0.8
0.90 0.50
0.6
0.50
y = -7E-06x 3 + 0.0009x 2 - 0.0247x + 0.6635 R² = 1
0.4 0.2 0.0 0
10
20
30
40
5.0
60
70
80
0.42
0.37
3.6
4.0
50
0.47
3.0 y = -1E-05x 3 + 0.0015x 2 - 0.0462x + 4.0212 R² = 1
2.0 1.0 0.0 0
10
20
30
40
50
60
70
80
Solids Residence Time (days)
Fig. 5 – Plot of calibrated kinetic parameters at different SRTs and polynominal correlations: (a) oxic endogenous decay rates for autotrophs, (b) oxic endogenous decay rates for heterotrophs, (c) oxic endogenous decay rates for PAOs, (d) anoxic hydrolysis factor, (e) anaerobic hydrolysis factor, and (f) fermentation rates. (Numbers on plots indicate values at different SRTs).
The calibrated model using the estimated kinetic parameters at 35 days predicted the effluent characteristics very well when compared to the default values. In addition, the predicted results showed that use of a dewatering device and returning the solids back into the oxic tank appeared to be suitable for modeling an MBR.
Modeling of the experimental results indicates that the kinetic parameters of the membrane biological systems were a function of SRT and that the kinetic parameters should be corrected in the model to include this variation. The maximum specific growth rates did not change with SRT indicating that nutrient availability was similar between SRTs. However, oxic
water research 44 (2010) 3192–3202
Table 5 – Model verification showing experimental effluent concentrations (±95% CI) and predicted concentrations at 35 days SRT using calibrated kinetic parameters.
Experimental Default Predicted
sCOD (mg/L)
TN (mg/L)
NH3-N (mg/L)
NO 3 -N (mg/L)
TP (mg/L)
29 3 33 31
8.6 0.5 13.0 8.9
0.4 0.1 0.2 0.3
7.2 0.3 11.7 7.6
3.2 0.3 6.3 3.4
endogenous decay rates for autotrophs, heterotrophs, and PAOs increased with SRT indicating that the viable portion of cells to TSS would decrease. This may mean that longer SRT may result in higher TP concentration in the mixed liquor as a result of endogenous decay of the viable cells.
4.
Conclusion
The results of the MBR with anaerobic, anoxic, and oxic tanks for SRTs ranging from 10 days to 75 days showed that: sCOD were sufficiently removed with more than 93% removal along with excellent nitrification at more than 98%. as for nitrogen removal, the system performed well with TN percent removals increasing with longer SRTs but peaking at 81% removal for both 50 and 75 day day SRT. TP percent removal was highest for 50 days SRT at approximately 80% but decreased to 60% at 75 days SRT indicating a possible limiting operating condition for PAOs and the lysis of cells at this SRT which may also introduce phosphorus back into the mixed liquor although the actual cause is not yet known. The experimental results imply that operating an MBR for phosphorus removal at very long SRTs may have some limitations. the sensitivity analysis results of the Biowin AS/AD model indicated that the heterotrophic anoxic yield, anaerobic hydrolysis factors of heterotrophs, heterotrophic hydrolysis, oxic endogenous decay rate for heterotrophs, and oxic endogenous decay rate of PAOs had the most impact on predicted effluent TP concentrations. Calibration of Biowin AS/AD model with the experimental data revealed that oxic endogenous decay rates increased with SRT and the higher anoxic and anaerobic hydrolysis factors at higher SRTs compensated for the higher endogenous decay rates up to 50 days but then the conditions became limited due to high endogenous decay rates. Polynomial correlations for the various kinetic parameters (such as anoxic and anaerobic hydrolysis factors for heterotrophs and oxic endogenous decay rates) with SRTs were derived. Using the estimated kinetic parameters from various polynomial correlations derived from the calibration of the Biowin model, the predicted results showed that the model was able to predict well the experimental effluent sCOD, TN and TP concentrations at 35 days SRT. The experimental results provided information on the possible limitation of operating an MBR for phosphorus removal at long SRTs and the kinetic parameters obtained by the
3201
modeling efforts can be used to model nutrient removal for MBRs systems with SRTs greater than 25 days. However, further experimental and modeling efforts are required to build the body of kinetic information for MBR systems under different recirculation conditions and configurations, and different HRTs of the various anaerobic, anoxic and oxic reactors.
Acknowledgements Postdoctoral fellowship awarded to Ertan Arslankaya by The Scientific and Technical Research Council of Turkey (TUBITAK).
references
Abegglen, C.A., Ospelt, M., Siegrist, H., 2008. Biological nutrient removal in a small-scale MBR treating household wastewater. Water Research 42 (1–2), 338–346. Adam, C., Kraume, M., Gnirss, R., Lesjean, B., 2003. Membrane bioreactor configurations for enhanced biological phosphorus removal. Water Science and Technology: Water Supply 3 (5–6), 237–244. APHA, AWWA, WEF, 2002. Standard Methods for the Examination of Water and Wastewater, 20th ed. American Public Health Association/American Water Works Association/Water Environment Federation, Washington DC, USA. Delgado, S., Diaz, F., Villarroel, R., Vera, L., Diaz, R., Elmaleh, S., 2002. Influence of biologically treated wastewater quality on filtration through a hollow-fibre membrane. Desalination 146 (1–3), 459–462. Ersu, C.B., Ong, S.K., Arslankaya, E., Brown, P., 2008. Comparison of recirculation configurations for biological nutrient removal in a membrane bioreactor. Water Research 42 (6–7), 1651–1663. Fleischer, E.J., Broderick, T.A., Daigger, G.T., Fonseca, A.D., Holbrook, R.D., Murthy, S.N., 2005. Evaluation of membrane bioreactor process capabilities to meet stringent effluent nutrient discharge requirements. Water Environment Research 77 (2), 162–178. Ha, J.H., Ong, S.K., Rao, S., 2007. Nitrification and denitrification in partially aerated biological aerated filter (BAF) with dual size sand media. Water Science and Technology 55 (1–2), 9–17. Hasar, H., Kinaci, C., Unlu, A., Ipek, U., 2002. Role of intermittent aeration in domestic wastewater treatment by submerged membrane activated sludge system. Desalination 142 (3), 287–293. Kishino, H., Ishida, H., Iwabu, H., Nakano, I., 1996. Domestic wastewater reuse using a submerged membrane bioreactor. Desalination 106 (1–3), 115–119. Lesjean, B., Gnirss, R., Adam, C., 2002. Process configurations adapted to membrane bioreactors for enhanced biological phosphorus and nitrogen removal. Desalination 149, 217–224. Macomber, J., Cicek, N., Suidan, M.T., Davel, J., Ginestet, P., Audic, J.M., 2005. Biological kinetic data evaluation of an activated sludge system coupled with an ultrafiltration membrane. Journal of Environmental Engineering 131 (4), 579–586. Metcalf, Eddy, 2003. Wastewater Engineering, Treatment, Disposal and Reuse, fourth ed. McGraw Hill Inc., NY. Ogoshi, M., Suzuki, Y., 2000. Application of MBR to an easily installed municipal wastewater treatment plant. Water Science and Technology 41 (10–11), 287–293. Rosenberger, S., Kru¨ger, U., Witzig, R., Manz, W., Szewzyk, U., Kraume, M., 2000. Operation of different membrane bioreactors: experimental results and physiological state of
3202
water research 44 (2010) 3192–3202
the microorganisms. Water Science and Technology 41 (10–11), 269–277. Rosenberger, S., Kru¨ger, U., Witzig, R., Manz, W., Szewzyk, U., Kraume, M., 2002. Performance of a bioreactor with submerged membranes for oxic treatment of municipal wastewater. Water Research 36 (2), 413–420. Shin, H., Kang, S., Lee, C., Lim, J., 2004. Performance of a pilot scale membrane bioreactor coupled with SBR (SM-SBR) – experiences in seasonal temperature changes. Water Science and Technology: Water Supply 4 (1), 135–142. Stephenson, T., Judd, S., Jefferson, B., Brindle, K., 2000. Membrane Bioreactors for Wastewater Treatment. IWA Publishing, London, UK.
Ubukata, Y., Takii, S., 1998. Some physiological characteristics of a phosphate removing bacterium, Microlunatus phosphovorus, and a simplified isolation and identification method for phosphate-removing bacteria. Water Science and Technology 38 (1), 149–157. Urbain, V., Mobarry, B., De Silva, V., Stahl, D.A., Rittmann, B.E., Manem, J., 1998. Integration of performance, molecular biology and modeling to describe the activated mixed liquor process. Water Science and Technology 37 (4–5), 223–229. Visvanathan, C., Ben Aim, R., Parameshwaran, K., 2000. Membrane separation bioreactors for wastewater treatment. Critical Reviews in Environmental Science and Technology 30 (1), 1–48.