Content uploaded by Eleni Vaiopoulou
Author content
All content in this area was uploaded by Eleni Vaiopoulou on Mar 28, 2018
Content may be subject to copyright.
Technical Note
A modified UCT method for biological nutrient removal: Configuration
and performance
E. Vaiopoulou
*
, A. Aivasidis
1
Department of Environmental Engineering, Democritus University of Thrace, Vas. Sofias 12, 67100 Xanthi, Greece
article info
Article history:
Received 31 October 2007
Received in revised form 16 April 2008
Accepted 16 April 2008
Available online 2 June 2008
Keywords:
UCT
Wastewater
Nutrients
Phosphorus
Step feeding
BNRAS
abstract
A pilot-scale prototype activated sludge system is presented, which combines both, the idea of University
of Cape Town (UCT) concept and the step denitrification cascade for removal of carbon, nitrogen and
phosphorus. The experimental set-up consists of an anaerobic selector and stepwise feeding in subse-
quent three identical pairs of anoxic and oxic tanks. Raw wastewater with influent flow rates ranging
between 48 and 168 l d
1
was fed to the unit at hydraulic residence times (HRTs) of 5–18 h and was dis-
tributed at percentages of 60/25/15%, 40/30/30% and 25/40/35% to the anaerobic selector, 2nd and 3rd
anoxic tanks, respectively (influent flow distribution before the anaerobic selector). The results for the
entire experimental period showed high removal efficiencies of organic matter of 89% as total chemical
oxygen demand removal and 95% removal for biochemical oxygen demand, 90% removal of total Kjeldahl
nitrogen and total nitrogen removal through denitrification of 73%, mean phosphorus removal of 67%, as
well as excellent settleability. The highest removal efficiency and the optimum performance were
recorded at an HRT of about 9 h and influent flow rate of 96 l d
1
, in which 60% is distributed to the anaer-
obic selector, 25% to the second anoxic tank and 15% to the last anoxic tank. Consequently, the plant con-
figuration enhanced removal efficiency, optimized performance, saved energy, formed good settling
sludge and provided operational assurance.
Ó2008 Elsevier Ltd. All rights reserved.
1. Introduction
Biological nutrient removal activated sludge (BNRAS) systems
remove carbon, nitrogen and phosphorus by biological means with
low costs and less waste sludge production (Metcalf and Eddy,
2003). One of the most commonly applied BNRAS methods for ur-
ban wastewater treatment relies on the University of Cape Town
(UCT) concept. The UCT process was designed to minimize the ef-
fect of nitrate to the anaerobic contact zone, which is crucial for
maintaining truly anaerobic conditions and thus, allowing biologi-
cal phosphorus release (Ekama and Wentzel, 1999; Metcalf and
Eddy, 2003; Vaiopoulou et al., 2007a). In fact, the higher the phos-
phorus concentration released in the anaerobic tank, the higher is
the phosphorus concentration taken up under aerobic conditions.
From a biological aspect, the anaerobic selector not only contrib-
utes to phosphorus removal, but also forms heavy large flocs that
enhance settleability (Metcalf and Eddy, 2003). Experiments and
full-scale applications have shown that this pre-denitrification–
nitrification process, which includes an anaerobic selector and
internal recycle, is efficient for removal of carbon, nitrogen and
phosphorus (Schlegel, 1992; Ekama and Wentzel, 1999; Metcalf
and Eddy, 2003). The modified UCT method upgraded the process;
however, an internal recycling of nitrified liquor to an anoxic tank
is still needed (Metcalf and Eddy, 2003).
Modifications in such BNRAS systems have proved that step
feeding is an attractive process to eliminate internal recycling
and sludge recirculation, as well as to optimize organic carbon
for denitrification, resulting in energy savings (Schlegel, 1992;
Görgün et al., 1996; Metcalf and Eddy, 2003; Pai et al., 2004; Vai-
opoulou et al., 2007a). On the other hand, multiple stage cascades
optimize removal efficiency with minimum reactor volume. A
three-stage cascade is expected to yield optimum effluent quality,
which is proved by model simulation results (Görgün et al., 1996).
Experiments in a two denitrification stage UCT system with step
feeding and influent flow distribution by 50% to the anaerobic tank
and to the second anoxic tank showed high removal efficiencies
(Schlegel, 1992). A similar attempt was undertaken by Pai et al.
(2004) resulting in an economically efficient removal of organics
and nutrients with a highly clarified effluent. However, the effect
of the step feeding process was minimized, whereas the plant
was fed with a synthetic wastewater, which can not simulate real
wastewater.
On this basis, the UCT approach could be enhanced by combin-
ing multiple stages of anoxic and oxic zones with the step feeding
process. Herein, a modified UCT method is proposed, which con-
0045-6535/$ - see front matter Ó2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.chemosphere.2008.04.044
*Corresponding author. Tel.: +30 6973 805513; fax: +30 25410 79376.
E-mail addresses: vaiop@env.duth.gr (E. Vaiopoulou), spiadou@env.duth.gr
(A. Aivasidis).
1
Tel.: +30 25410 79375; fax: +30 25410 79376.
Chemosphere 72 (2008) 1062–1068
Contents lists available at ScienceDirect
Chemosphere
journal homepage: www.elsevier.com/locate/chemosphere
sists of an initial anaerobic selector followed by a cascade of three
identical pairs of tanks. Each pair comprises of an anoxic and aer-
obic tank for denitrification and nitrification, respectively. This
method may lead to high performance and energy saving due to
its configuration, which includes a three-stage cascade, step feed-
ing, sludge recycle minimization and exclusion of the primary sed-
imentation. In particular, the three-stage cascade is expected to
enhance removal efficiency. Organics and nutrients could be main-
tained for phosphorus removal and denitrification by primary sed-
imentation exclusion and by application of the step feeding
strategy. Feeding ratios of the influent flow rate to the anaerobic
selector and anoxic tanks were calculated so that the carbon to
nitrogen ratio is theoretically sufficient for denitrification.
Objective of this work is to present a novel pilot-scale activated
sludge system for biological removal of carbon, nitrogen and phos-
phorus, combining the idea of the UCT process with the step feed-
ing approach, and also to confirm the theoretically expected design
advantages.
2. Materials and methods
2.1. Plant description
The pilot plant has total operational volume of 44 l and consists
of an anaerobic selector (ANAER) of 3.5 l followed by a cascade of
three identical pairs of anoxic/aerobic bioreactors (Fig. 1). Denitri-
fication takes place in the three anoxic bioreactors (DNi,i= 1–3),
which have an operational volume of 3.4 l each, whereas nitrifica-
tion and carbon oxidation is accomplished in the three aerobic bio-
reactors (AEi) with an operational volume of 7.3 l each. Finally, a
sedimentation tank (ST) of 8.3 l is used for sludge separation and
recycling into the anaerobic selector via the first anoxic tank. A
more detailed description of the plant and some preliminary re-
sults of its performance are given in (Vaiopoulou et al., 2007a).
The pilot-scale unit has been in operation for about two years.
The plant was fed with raw municipal wastewater from the com-
bined sewer system of Xanthi city in Greece. Particularly, tanks
of 1 m
3
full of raw wastewater were delivered once or twice per
week depending on influent volume needs. Influent average con-
centrations of biochemical oxygen demand (BOD) and chemical
oxygen demand (COD) were 320 ± 112 and 510 ± 147 mg l
1
,
respectively. Influent PO
3
4
–P and TP concentration spanned be-
tween 2 and 9 mg PO
3
4
–P l
1
and 2–15 mg TP l
1
. The percentage
of orthophosphates content in total phosphorus was in average
63 ± 53%. Total nitrogen (TN) influent concentration was typically
around 55 ± 7 mg l
1
. With the exception of some low phosphorus
concentrations, which is on average at about 4 ± 2 mg l
1
, organic
substrate and nitrogen extent can be regarded as normal for a
Greek municipal wastewater. According to a study of the Greek
sewage system, BOD and COD concentrations range between 320
and 580 and 460 and 645 mg l
1
, respectively, orthophosphates
span between 0.03 and 10.5 mg PO
3
4
–P l
1
, whereas TKN concen-
tration is typically fluctuating between 75 and 86 mg l
1
(Melidis
et al., 2005). The minor differences in values are attributed to the
fact that in our case wastewater was stored, whereas Melidis
et al. (2005) worked in the field.
Influent wastewater rate (Q
F
) is distributed to the anaerobic
selector with a flow rate of Q
ANAER
, and to the second and third an-
oxic zone with flow rates of Q
DN2
and Q
DN3
, respectively (Fig. 1).
Incoming percentages are calculated according to the ratio of the
tank influent flow rate to the total influent flow rate (e.g. Q
DN2
/
Q
F
). Influent flow distribution to the anaerobic selector, to the sec-
ond and to the third anoxic tank (Q
ANAER
/Q
DN2
/Q
DN3
) was set to 60/
25/15%, 40/30/30% and 25/40/35%, respectively. Experiment runs
were performed by changing either the influent flow rate or the
feed distribution ratio. Each run refers to a specific set of these
parameters under steady-state conditions, e.g the run 4(60/25/
15%) stands for Q
F
=4lh
1
and feeding ratio of 60/25/15%. Once
the operational conditions were set, the pilot plant was run until
it reached steady-state conditions. After steady-state conditions
were confirmed, sampled, analyzed and finally set of operational
conditions for the next run were taking place. Sampling and ana-
lyzing for each run were repeated at least three times to assure
steady-state conditions establishment, results reproduction and
repeatability. Initially, a specific hydraulic residence time (HRT)
was applied and then all three feeding ratios were examined. Q
F
into the activated sludge system was increased from 48 to
168 l d
1
in order to obtain an HRT from 18 to 5 h, respectively.
Sludge recycling rate (Q
R
) was set equal to influent flow rate
(Q
R
=Q
F
) for each run, whereas the overall sludge retention time
(SRT) was kept constant at 10 d by controlling sludge wastage flow
rate.
Table 1 presents operational plant parameters for each run. The
main disadvantage of feeding with real wastewater is the concen-
tration fluctuations.
2.2. Water quality monitoring
Under steady-state conditions, samples were collected from the
influent of the unit and the effluents of each tank, as well as from
the separated sludge in the clarifier. The samples were analyzed
following standard procedures for the determination of BOD,
COD, ammonia, TKN, nitrite, nitrate, total phosphorus, orthophos-
phates, VSS (volatile suspended solids) and SVI (sludge volume in-
dex; APHA, 1998). A high pressure liquid chromatography (Alltech,
USA) was used for anion determination. Wastewater temperature,
pH and DO were monitored on a daily basis. Typical values for pH
ranged between 7.4 and 7.6, whereas temperature was kept con-
stant at 20 °C. The dissolved oxygen (DO) in the aerated reactors
was maintained above 2 mg l
1
. Sludge samples were also fre-
quently taken and observed within one hour under bright field
and phase contrast microscopy.
Fig. 1. Experimental set-up of the plant.
E. Vaiopoulou, A. Aivasidis / Chemosphere 72 (2008) 1062–1068 1063
3. Results
Results of two year operation showed constantly high removal
efficiencies of organic matter of 89 ± 4.7% as total COD and
95 ± 3.6% removal for BOD on average. Fig. 2 depicts the influent
and effluent concentrations for each steady-state run including re-
moval efficiency. Standard deviations of mean values in Fig. 2 are
shown in Table 2. Results for each pollution parameter, which
are presented in charts, were divided by vertical lines in three sec-
tions corresponding to the three feeding ratios (60/25/15%, 40/30/
Table 1
Operational plant conditions, F/M and loading rates for organic, nitrogenous and phosphorus substrates for each run (VSS of the anoxic and aeration tanks were used for F/M
calculation)
Run number Q
F
(l h
1
); AN/DN2/DN3 (%Q
F
) Loading (g d
1
) Sludge loading rate (F/M)
(kg
BOD
kg
1
VSS
d
1
)
HRT (
s
d)
COD BOD NH
þ
4
–N PO
3
4
–P
1 2 (60/25/15) 22.8 9.6 1.5 0.42 0.09 17.80
2 2 (40/30/30) 11.2 9.6 1.9 0.44 0.12 17.80
3 2 (25/40/35) 32.6 11.5 1.4 0.19 0.15 17.80
4 3 (60/25/15) 37.4 21.6 3.4 0.54 0.16 11.87
5 3 (40/30/30) 34.2 27.4 3.0 0.50 0.33 11.87
6 3 (25/40/35) 29.3 17.3 2.5 0.32 0.12 11.87
7 4 (60/25/15) 65.7 36.5 4.6 0.35 0.33 8.90
8 4 (40/30/30) 44.6 17.3 4.1 0.22 0.15 8.90
9 4 (25/40/35) 39.3 37.4 5.2 0.26 0.39 8.90
10 5 (60/25/15) 79.8 62.4 4.5 0.72 0.52 7.12
11 5 (40/30/30) 85.5 60.0 4.5 0.68 0.42 7.12
12 5 (25/40/35) 91.0 31.2 4.2 0.50 0.27 7.12
13 6 (60/25/15) 69.5 77.8 5.3 0.58 0.82 5.93
14 7 (60/25/15) 91.9 53.8 5.8 0.74 0.56 5.09
Fig. 2. Influent and effluent concentration as well as removal efficiency of (A) BOD, (B) COD, (C) total Kjedahl nitrogen (TKN), (D) total nitrogen (TN) and (E) phosphate–
phosphorus (PO
3
4
–P).
1064 E. Vaiopoulou, A. Aivasidis / Chemosphere 72 (2008) 1062–1068
30% and 25/40/35%). Total COD removal ranged between 80% and
98% and BOD removal between 87% and 98%, receiving constantly
the highest values at the ratio of 60/25/15%. A slight drop in BOD
and COD removal was observed at low HRTs (high influent flow
rate), which is attributed both to the high hydraulic and organic
substrate loading rate. COD removal fluctuated around 88% in feed-
ing schemes of 40/30/30% and 25/40/35%, whereas it remained
constant above 90% during at the ratio of 60/25/15%. The influent
total COD concentrations with a mean value of 540 ± 147 mg l
1
spanned between 230 and 760 mg l
1
, whereas effluent concentra-
tions with a mean value of 55 ± 23 mg l
1
fluctuated between 15
and 95 mg l
1
. Accordingly, mean influent concentration of BOD
was 330 ± 123 mg l
1
with a range of 180–540 mg l
1
, whereas
effluent concentrations ranged between 5 and 35 mg l
1
with an
average value of 18 ± 11 mg l
1
.
Complete nitrification was accomplished in all steady-state
experiments with the exception of low HRT runs. Mean removal
of NH
þ
4
–N and TKN was 95 ± 6% and 90 ± 7%, respectively, for the
entire experimental period. In fact, NH
þ
4
–N removal, and thus nitri-
fication, was even higher, since more NH
þ
4
–N was produced from
organic nitrogen in the plant in addition to the influent NH
þ
4
–N.
Mean influent and effluent NH
þ
4
–N concentrations attained values
of 40 ± 6.7 and 2 ± 2.2 mg l
1
, respectively, whereas mean influent
and effluent TKN concentrations received values of 56 ± 7 and
6 ± 3.5 mg l
1
, respectively. On the other hand, denitrification effi-
ciency fluctuated fairly at the feeding ratios of 40/30/30% and 25/
40/35%, whereas at the ratio of 60/25/15% was constantly high
(Fig. 2C). TN influent concentrations ranged between 40 and
60 mg l
1
with a mean value of 52 ± 7 mg l
1
, whereas mean
Table 2
Mean influent and effluent concentration and removal efficiency as well as standard deviation values shown in brackets for each steady-state run
Run number Influent (mg l
1
) Effluent (mg l
1
) Efficiency
COD BOD TN TKN PO
3
4
–P COD BOD TN TKN PO
3
4
–P COD BOD TN TKN PO
3
4
–P
1 475 (96) 200 (35) 51.6 (5.3) 51.5 (4.8) 8.7 (2.2) 44 (6) 10 (4) 10.5 (0.5) 4.2 (0.6) 3.1 (0.6) 91 (3.1) 95 (1.2) 80 (1.1) 92 (2) 64 (3.4)
2 233 (45) 180 (23) 47.5 (3.8) 47.4 (4.3) 9.2 (0.4) 44 (3) 5 (3) 19.2 (9.2) 3.9 (0.2) 5.3 (2.2) 81 (2.4) 97 (1.3) 60 (22.8) 92 (0.4) 42 (22)
3 679 (92) 240 (70) 44 (2.2) 44 (2.2) 4 (2.5) 39 (0) 10 (4) 17 (4.3) 4 (0.5) 2.9 (2.7) 94 (0.8) 96 (0.5) 61 (7.8) 90 (0.6) 27 (45)
4 520 (25) 300 (40) 69 (23.4) 68.9 (1.2) 7.5 (0.8) 15 (11) 10 (2) 13.5 (12.7) 4.5 (2.2) 2.5 (0) 97 (2) 97 (0.2) 80 (13.7) 93 (3) 67 (3.2)
5 475 (277) 380 (190) 52.1 (16.5) 52 (17) 5.4 (2.9) 40 (31) 6 (3) 13.4 (0.1) 4.1 (0.4) 3.9 (1.2) 92 (1.7) 98 (1.8) 74 (9.1) 92 (2.1) 27 (14)
6 407 (40) 240 (45) 57 (2.5) 56.9 (3) 4.4 (2.5) 68 (8) 5 (3) 14.8 (1.2) 2.8 (0.7) 1.9 (0.7) 83 (3.5) 98 (1.1) 74 (1) 95 (1.4) 56 (10)
7 684 (164) 380 (37) 68.9 (14.4) 68.8 (14) 3.6 (2.2) 43 (14) 6 (7) 11.5 (1) 4.4 (2.5) 0.2 (2) 94 (4.7) 98 (1.5) 83 (2.2) 94 (2.4) 93 (29)
8 465 (69) 180 (46) 55.1 (11.1) 55.1 (11) 2.3 (1.7) 58 (19) 15 (5) 17.6 (2.5) 3.8 (3.4) 0.7 (0.8) 88 (2.2) 92 (2.2) 58 (1.9) 93 (8.4) 67 (8)
9 409 (80) 390 (65) 58.6 (3.5) 58.5 (3) 2.7 (4) 38 (9) 22 (8) 13.6 (9.1) 7.8 (1.5) 1.4 (2.4) 91 (4.3) 94 (2.7) 77 (17) 87 (2.8) 50 (6)
10 665 (87) 520 (191) 56.1 (6.4) 56 (6) 4.3 (1.6) 48 (49) 35 (34) 11.3 (4.6) 3.8 (2.1) 1.9 (0.9) 93 (8.8) 93 (17.7) 80 (5.6) 93 (2.9) 56 (26)
11 713 (181) 500 (186) 60.7 (2.7) 60.6 (2) 5.6 (3.7) 86 (25) 20 (14) 11.3 (8.1) 5.8 (0.4) 1.6 (1.4) 88 (9.3) 96 (6.9) 81 (12.1) 90 (1) 71 (12)
12 758 (248) 260 (60) 57.1 (2.5) 57 (2) 4.2 (1.8) 89 (25) 35 (5) 15.1 (7.6) 5.3 (1.7) 1.9 (0.9) 88 (1.5) 87 (4.7) 74 (14.4) 91 (3.3) 54 (18)
13 482 (122) 540 (46) 50.8 (6.9) 50.7 (7) 4 (2.2) 60 (6) 20 (10) 8.1 (1.1) 5.6 (2.2) 0.3 (0.2) 88 (4.6) 96 (1.5) 84 (1.1) 89 (2.8) 92 (12)
14 547 (64) 320 (181) 53.8 (3.1) 53,7 (3) 4.4 (2.2) 96 (21) 35 (15) 17 (5.8) 17 (7.2) 0.5 (0.9) 82 (2.6) 89 (4) 68 (9.6) 68 (12.6) 89 (9)
Table 3
Effluent wastewater characteristics in each bioreactor of the plant for the three
feeding ratios at the optimum HRT of 9 h (standard deviation values in brackets)
Bioreactors Effluent concentration (mg l
1
)
60/25/15% 40/30/30% 25/40/35%
ANAER COD 116 (4) 63 (33) 48 (8)
NH
þ
4
–N 30.7 (5.8) 21.8 (9.4) 20.4 (4.5)
TKN 36.5 (4.9) 26.6 (8.5) 23.9 (4)
PO
3
4
–P 4.4 (1.1) 3.2 (0.1) 8.7 (1.1)
NO
3
–N 1 (0.6) 0.8 (0.1) 0.1 (0)
DN1 COD 52 (11) 87 (1) 73 (3)
NH
þ
4
–N 18.4 (4.4) 13.5 (4.9) 13.6 (2.6)
TKN 30.8 (10.9) 20 (3.3) 24.4 (4)
PO
3
4
–P 6.5 (0.3) 3.2 (0.2) 1.6 (0.3)
NO
3
–N 0.4 (0.5) 2 (0.2) 0.2 (0.1)
AE1 COD 49 (10) 63 (1) 59 (4)
NH
þ
4
–N 5.6 (2.3) 2.2 (0.5) 0.8 (0.3)
TKN 16 (10.5) 17.1 (10) 15.7 (2.1)
PO
3
4
–P 2.9 (2) 1.2 (0.2) 0.6 (0.2)
NO
3
–N 7.7 (0.5) 7.9 (0.1) 25.8 (4)
DN2 COD 57 (13) 29 (25) 71 (9)
NH
þ
4
–N 11.8 (3.7) 9.5 (1.9) 12.3 (2.5)
TKN 15.1 (5.2) 11.5 (3) 16.9 (2.5)
PO
3
4
–P 1.1 (2) 2.4 (0.6) 3.4 (0.5)
NO
3
–N 0.2 (0.5) 6.7 (1.4) 0.3 (0.1)
AE2 COD 48 (4) 53 (6) 39 (4)
NH
þ
4
–N 5.8 (3.2) 1.9 (1.2) 1.1 (0.5)
TKN 7.8 (4.1) 3.4 (1.5) 8 (2)
PO
3
4
–P 1 (0.9) 1.7 (0.4) 0.2 (0.1)
NO
3
–N 7.8 (2) 8.5 (1.5) 8.3 (0.3)
DN3 COD 49 (13) 54 (16) 64 (3)
NH
þ
4
–N 9.9 (4.8) 3.9 (0) 12.3 (2.4)
TKN 12.9 (5.6) 12.3 (6.6) 16.5 (1.5)
PO
3
4
–P 1.1 (1.4) 1.3 (0.3) 1.7 (0.3)
NO
3
–N 5.7 (1.5) 6 (1.5) 1 (0.1)
AE3 COD 38 (4) 19 (20) 59 (3)
NH
þ
4
–N 1.7 (0.3) 2 (0) 2.7 (0.9)
TKN 6.6 (1.7) 4 (3.6) 7.3 (0.3)
PO
3
4
–P 0.4 (0.7) 1 (0.3) 0.6 (0.3)
NO
3
–N 6.5 (1.8) 8.9 (1.5) 5.3 (1.5)
E. Vaiopoulou, A. Aivasidis / Chemosphere 72 (2008) 1062–1068 1065
effluent concentration was 14 ± 3 mg l
1
fluctuating between 8 and
19 mg l
1
for the entire experimental period. Low removal effi-
ciency and high effluent concentrations emerged from the high
NO
3
–N effluent concentrations, which in some cases exceeded
10 mg l
1
in the feeding ratios of 40/30/30 and 25/40/35. This re-
sult is most probably attributed to the unit configuration, which
ended with an aerobic tank, where nitrification was processed
and nitrate was produced. However, aerobic conditions as the final
step are essential for phosphate accumulation. Another explana-
tion could be the accumulation of nitrate from previous aerobic
stages on its way to the effluent (Table 3).
The phosphate removal efficiency ranged intensively between
27% and 96% with a mean value of 67 ± 21% for the whole experi-
mental period (Fig. 2E), which was attributed to the influent con-
centration fluctuation. A recent study stressed the significance of
stable operating conditions and influent concentrations for suc-
cessful phosphorus removal (Mullkerins et al., 2004). Influent
phosphate concentrations spanned between 2 and 9 mg l
1
with
an average value of 5(2.1) mg l
1
, whereas mean effluent phos-
phate concentration was less than 2 ± 1.4 mg l
1
with a range of
0.2–5.3 mg l
1
for the whole experimental period. The lowest re-
moval efficiency was recorded at the highest HRT. Constantly high
removal efficiencies occurred at HRTs of 5–9 h at the feeding ratio
of 60/25/15, during which the phosphate loading rate exceeded
10 g m
3
d
1
. Total phosphorus removal efficiency followed the
same profile with a mean value of 53 ± 19% ranging between 30%
and 79%. Mean influent total phosphorus concentration was about
8 ± 3.7 with a range of 2–15 m l
1
, whereas mean effluent concen-
tration was 3.4 ± 2 fluctuating between 0.9 and 7.5 mg l
1
.
Throughout the experimental period, long, robust and compact
flocs comprised of a large species diversity were developed. Mean
VSS concentration decreased from 3.3 ± 0.06 g l
1
in DN1 to
2.5 ± 0.14 g l
1
in AE3 due to the sludge recycle entering DN1 tank,
due to the dilution effect and the higher HRTs in the initial tanks, as
a result of the step feeding process. HRT and feeding ratio hardly
had any effect on VSS values. The same profile stood for SVI, which
decreased averagely from 160 ± 37 in AE1 and 130 ± 51 in AE2 to
100 ± 40 ml g
1
in AE3 and thus, in the clarifier. These SVI values
were characterized as ideal (Jenkins et al., 1993) and showed that
the unit configuration actually enhanced sludge settleability. How-
ever, sludge characteristics were enhanced for the feeding ratio of
60/25/15% with a mean value of 94 ± 35 ml g
1
and gradually dete-
riorating at the feeding ratio of 40/30/30% with a mean value of
102 ± 36–129 ± 49 ml g
1
at the ratio of 25/40/35%.
Fig. 3. Removal efficiency for each feeding ratio.
1066 E. Vaiopoulou, A. Aivasidis / Chemosphere 72 (2008) 1062–1068
It is quite clear that the feeding ratio of 60/25/15% resulted to
the most efficient performance, since all removal efficiencies were
constantly higher in comparison to the other two ratios (Fig. 3).
That means that increasing influent flow rate in the initial stages
of the plant increases removal efficiency. Additionally, the SVI re-
mained lower than 150 ml g
1
, implying ideal sludge settleability.
Since BOD removal represents the biodegradable part of organic
pollution, NH
þ
4
–N removal stands for plant nitrification capability,
PO
3
4
removal for desphosphatation capability and TN removal for
plant denitrification capability, the mean removal efficiency calcu-
lated by using these data could be an indicator of the plant effi-
ciency. SVI should also be taken into account as a settleability
indicator. Comparing these data, the most efficient unit perfor-
mance was recorded at HRT of 9 h and feeding ratio of 60/25/15%
to the anaerobic selector, 2nd and 3rd anoxic tank, respectively.
This optimum steady-state run will be stated as 4(60/25/15%) for
brevity reasons.
Focusing on the performance of unit’s each bioreactor for the
optimum HRT of 9 h (Table 3), the anaerobic selector attains
the higher effluent concentrations, expect orthophosphates, in
the feeding ratio of 60/25/15%, which is attributed to the higher
influent flow percentage. In any case, DO was measured at 0 mg l
1
in the anaerobic selector, whereas nitrate was intruding due to the
flow rate from the first anoxic tank resulting to low phosphorus
release in the anaerobic selector. Therefore, phosphorus release is
the lowest in the anaerobic selector for the 60/25/15% ratio,
although its removal in the unit is the highest (Table 2). In addi-
tion, semi-batch experiments showed that specific anaerobic phos-
phorus release rate was determined at 0.03 g PO
3
4
–P g
1
VSS d
1
(Vaiopoulou and Aivasidis, 2007). Anoxic tanks contain higher
concentrations of pollution parameters in comparison to the aero-
bic bioreactors, since they receive the influent raw wastewater
flow rate. Denitrification is accomplished more efficiently in the
first anoxic tank, which is obvious from the lower nitrate concen-
trations. The feeding ratios of 60/25/15% and 40/30/30% tend to
accumulate more nitrate along the treatment train in comparison
with the 25/40/35% ratio, although TN removal in the plant is the
highest in the 60/25/15% ratio (Table 2). On the other hand, nitrifi-
cation is highly efficient in any ratio.
4. Discussion
Schlegel (1992) suggested multiple stage units when influent
carbon to nitrogen ratio (BOD/TKN) is above 5, which attained
the influent raw wastewater fed to our plant. Table 4 compares
similar BNRAS unit configurations and their performance. The pro-
posed method achieves a high quality effluent with the minimum
SRT and HRT as well as without any extra addition of volatile fatty
acids in the anaerobic tank. Moreover, internal recycle is unneces-
sary, whereas recycle flow rate is relatively low at 100% of the
influent wastewater flow rate. In any case, model simulations show
that a two-stage system is insufficient and requires high SRTs
(Lesouef et al., 1992; Görgün et al., 1996). As far as feeding with
synthetic wastewater is concerned (Pai et al., 2004), some real
wastewater composition parameters, such as COD, phosphorus,
pH etc., are of crucial importance for the effective biological phos-
phorus removal (Mullkerins et al., 2004). Therefore, they can not be
precisely simulated and thus, lead to bulking sludge and system
failure (Vaiopoulou et al., 2007b). In comparison to the European
Economic Community (1991) discharge standards, which refer to
125 mg COD l
1
, 25 mg BOD l
1
, 10–15 mg TN l
1
and 1–2 mg
PO
3
4
–P l
1
, our results are below the obligatory limits (Table 2).
The feed distribution ratio affected the unit’s each bioreactor
performance. Higher influent flow percentages in the anaerobic
selector lead to higher effluent concentrations. Occasional nitrate
intrusion to the anaerobic selector reduced phosphorus release,
but was unable to affect unit’s removal. Introduction of raw waste-
water to anoxic tanks increased concentrations of pollution param-
eters, which were further removed. Kayser et al. (1992) showed
that nitrogen removal was enhanced by influent flow rate reduc-
tion to the last stages of the cascade rather than by increasing
the recycle flow rate. In that case, maximum phosphorus release
was recorded in the first anoxic tank. Kayser et al. (1992) claimed
that this tank plays a crucial role in phosphorus removal. This con-
clusion was confirmed by our results, since removal efficiency was
maximized for the feeding ratio of 60/25/15, whereas under these
operational conditions maximum phosphate release took place to
the first anoxic tank. Nitrification remained unaffected by the
feeding ratio and was successful in all aeration tanks, whereas
Table 4
Differences among BNRAS unit configurations, removal efficiencies and effluent parameters
Schlegel (1992) Lesouef et al. (1992) Kayser et al. (1992) Pai et al. (2004) Proposed system
Configuration
Primary sedimentation Yes Yes No No No
Anaerobic tank Yes No No Yes Yes
Denitrification stages 2 3 3 3 3
Step feeding ratio (%) 50/50 30/40/40 40/40/20 70/20/10 60/25/15
Sludge digester Yes No No No No
Type of wastewater Real Hypothetically real Real Synthetic Real
VFA feed to the anaerobic tank Yes (digester effluent) – No Yes (synthetic wastewater) No
Recycle ratio (%) No data 90 150 50 100
SRT (d) No data 17 No data 10 10
HRT (h)
Total HRT No data 10 22.5 9.3 8.9
Anaerobic tank 2 – – 2.73 0.55
Anoxic tanks No data 2 38.64 2.97 1.46
Aerobic tanks No data 8 12.71 5.23 3.13
Removal (%)
COD No data No data 93 87 94
TN 87 82.5 92 75 83
PO
3
4
–P 70–80 No data 81 94 93
Effluent concentration (mg l
1
)
Soluble COD 23 No data 65 28 30
NO
3
–N 4.6 4.4 5 8 6.9
PO
3
4
–P 1.8 No data 2.8 0.3 0.2
E. Vaiopoulou, A. Aivasidis / Chemosphere 72 (2008) 1062–1068 1067
denitrification is accomplished more efficiently in the first anoxic
tank. The experimental practice and foremost the inconsistency
of the influent C/N ratio were responsible for the limited denitrifi-
cation activity, and thus efficiency. This is attributed to the main-
tenance of a large quantity of wastewater, which, even under low
temperature, inevitably leads to organics biodegradation under
storage (4–5 d). However, under full-scale application of the pre-
sented plant, where influent wastewater flow will be constant
and the C/N ratio constantly high, denitrification efficiency could
reach higher values in the second and third anoxic tank.
On the basis of optimization without extra costs, this modifica-
tion of the UCT method is necessary, since the plant configuration
omitted nitrified liquor recycle, reduced sludge recycle and volume
requirements, eliminated nitrate/oxygen intrusion to anaerobic
conditions, optimized organic carbon usage for denitrification
and enhanced effluent water characteristics. In particular, step
feeding configuration removed nitrogen by minimizing internal re-
cycle and optimizing carbon usage for denitrification (Kayser et al.,
1992; Schlegel, 1992; Görgün et al., 1996; Metcalf and Eddy, 2003).
The cascade enhanced effluent water quality with the minimum
reactor volume, whereas provided operational safety, since what-
ever is left untreated, may be treated in the next stage (Kayser
et al., 1992; Schlegel, 1992).
Furthermore, the residence time distribution (RTD) of the pre-
sented plant (data not shown here) played a beneficial role regard-
ing its enhanced response to the existing first order kinetics, due to
the limited axial dispersion. RTD experiments showed that the
Bodenstein number ranged from 5 to 6 (dispersion model),
whereas the cascade number has been determined between 2.5
and 3 (tanks in series model). From a kinetic point of view, the
plant behaved operationally as if a cascade of three continuous
flow stirred tank reactors and consequently, as if a plug flow reac-
tor (PFR). The favorable plant characteristics, which were responsi-
ble for high removal efficiencies and good sludge settleability, were
attributed to this PFR behavior.
5. Conclusions
The beneficial characteristics of the proposed optimum plant
configuration are plentiful. The initial anaerobic tank helps select
phosphate accumulating organisms, according to the enhanced
biological phosphorus removal technique, as well as floc-formers
over filamentous bacteria. The three denitrification cascade yields
to high removal efficiency and also offers operational safety. Influ-
ent flow rate is distributed to anaerobic or anoxic tanks (step feed-
ing process) to provide organic substrate for phosphate release and
denitrification, as well as to transform these tanks into selectors of
floc-formers over filaments. Furthermore, the primary sedimenta-
tion tank is excluded, which means that organic substrate is main-
tained for biological phosphate removal and denitrification.
Internal nitrified liquor recycle is also totally omitted, whereas
sludge recycle is introduced to the anaerobic selector via the 1st
anoxic tank. This would result in (1) nitrate reduction by denitrifi-
cation and thus, anaerobic conditions may remain undisturbed, (2)
sludge recycle flow minimization and (3) continuous biomass sup-
ply to the anaerobic selector. Finally, the limited axial dispersion in
the cascade (plug flow) due to the narrow spectrum of the RTD is
ideal for first order kinetics, which in combination to the anaerobic
selector may also suppress filamentous bacteria.
The highest removal efficiencies, the most favorable sludge
characteristics and the lowest effluent concentrations were ob-
served at HRT of 9 h and feed distribution ratio of 60% to the anaer-
obic selector, 25% to the second and 15% to the third anoxic tank. In
particular, COD and BOD removal received values of 94% and 98%,
respectively, whereas phosphate removal of 93%. Nitrogen removal
was recorded at 83% as TN, 94% as TKN and 99% as NH
þ
4
–N. SVI was
satisfactory at 140 ml g
1
.
References
APHA, 1998. Standard Methods for the Examination of Water and Wastewater.
American Public Health Association, Washington, DC, USA.
European Economic Community, 1991. Urban Wastewater Treatment, Directive 91/
271/EEC.
Ekama, G.A., Wentzel, M.C., 1999. Difficulties and developments in biological
nutrient removal technology and modelling. Water Sci. Technol. 39 (6), 1–11.
Görgün, E., Artan, N., Orhon, D., Sözen, S., 1996. Evaluation of nitrogen removal
by step feeding in large treatment plants. Water Sci. Technol. 34 (1–2),
253–260.
Jenkins, D., Richard, M.G., Daigger, G.T., 1993. Manual on the Causes and Control of
Activated Sludge Bulking and Foaming. Lewis Publishers, Washington, USA.
Kayser, R., Stobbe, G., Werner, M., 1992. Operational results of the Wolfsburg
wastewater treatment plant. Water Sci. Technol. 25 (6), 203–209.
Lesouef, A., Payraudeau, M., Rogalla, F., Kleiber, B., 1992. Optimising nitrogen
removal reactor configurations by on-line calibration of the IAWPRC activated
sludge model. Water Sci. Technol. 25 (6), 105–123.
Metcalf and Eddy, 2003. Wastewater Engineering: Treatment and Reuse. McGraw
Hill, New York City, USA.
Melidis, P., Siakir, A., Vaiopoulou, E., Aivasidis, A., 2005. Fate of wastewater
composition in a sewerage system. In: 9th Conference on Environmental
Science and Technology, Rhodes, Greece, September 1–3, 2005.
Mullkerins, D., Dobson, A.D.W., Colleran, E., 2004. Parameters affecting biological
phosphate removal from wastewaters. Environ. Int. 30, 249–259.
Pai, T.Y., Tsai, Y.P., Chou, Y.J., Chang, H.Y., Leu, H.G., Ouyang, C.F., 2004. Microbial
kinetic analysis of three different types of EBNR process. Chemosphere 55, 109–
118.
Schlegel, S., 1992. Operational results of waste water treatment plants with
biological N and P elimination. Water Sci. Technol. 25 (4–5), 241–247.
Vaiopoulou, E., Melidis, P., Aivasidis, A., 2007a. An activated sludge treatment plant
for integrated removal of carbon, nitrogen and phosphorus. Desalination 211,
192–199.
Vaiopoulou, E., Melidis, P., Aivasidis, A., 2007b. Growth of filamentous bacteria in an
enhanced biological phosphorus removal system. Desalination 213, 288–296.
Vaiopoulou, E., Aivasidis, A., 2007. A semi-batch (on-line) method for biokinetics
determination in an enhanced biological phosphorus removal system. In:
European Congress of Chemical Engineering (ECCE-6), Copenhagen, Denmark,
September 16–20, 2007.
1068 E. Vaiopoulou, A. Aivasidis / Chemosphere 72 (2008) 1062–1068
