See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/277949014

Process modeling and optimization of biological removal of carbon, nitrogen

and phosphorus from hospital wastewater in a continuous feeding &
intermittent discharge (CFID) bioreact...

Article  in  Korean Journal of Chemical Engineering · July 2015

DOI: 10.1007/s11814-014-0365-z

CITATIONS READS

15 179

6 authors, including:

Meghdad Pirsaheb Mitra Mohammadi

Kermanshah University of Medical Sciences kermanahah university of medical sciences
117 PUBLICATIONS   888 CITATIONS    16 PUBLICATIONS   48 CITATIONS   

SEE PROFILE SEE PROFILE

Amir Mohammad Mansouri Ali Akbar Zinatizadeh

Razi University Razi University
14 PUBLICATIONS   151 CITATIONS    166 PUBLICATIONS   3,256 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Using of an aerobic membrane bioreactor with activated carbon for landfill leachate treatment View project

Non-biodegradable pollutant removal from wastewater to produce hygienic water in a novel visible light immobilized nanophotocatalytic reactor and single integrated
photocatalytic-membrane process View project

All content following this page was uploaded by Amir Mohammad Mansouri on 02 January 2016.

The user has requested enhancement of the downloaded file.

Korean J. Chem. Eng., 32(7), 1340-1353 (2015) pISSN: 0256-1115
DOI: 10.1007/s11814-014-0365-z eISSN: 1975-7220
INVITED REVIEW PAPER INVITED REVIEW PAPER

Process modeling and optimization of biological removal of carbon, nitrogen

and phosphorus from hospital wastewater in a continuous feeding
& intermittent discharge (CFID) bioreactor
Meghdad Pirsaheb*,†, Mitra Mohamadi**, Amir Mohammad Mansouri*,***,†, Ali Akbar Lorestani Zinatizadeh****,
Sethupathi Sumathi*****, and Kiomars Sharafi******,*******

*Research Center for Environmental Determination of Health (RCEDH), Kermanshah University of Medical Sciences, Kermanshah, Iran
**Student Research Committee, Kermanshah University of Medical Science, Kermanshah, Iran
***Department of Analytical Chemistry, Faculty of Chemistry, Razi University, Kermanshah, Iran
****Water and Wastewater Research Center (WWRC), Department of Applied Chemistry,
Faculty of Chemistry, Razi University, Kermanshah, Iran
*****Department of Environmental Engineering, Faculty of Engineering and Green Technology,
UniversitiTunku Abdul Rahman, Kampar, Perak, Malaysia
******Department of Environmental Health Engineering, School of Public Health,
Kermanshah University of Medical Sciences, Kermanshah, Iran
*******Department of Environmental Health Engineering, School of Public Health,
Tehran University of Medical Sciences, Kermanshah, Iran
(Received 8 April 2014 • accepted 8 December 2014)

Abstract−We evaluated the feasibility and treatment performance of a continuous feeding and intermittent discharge
(CFID) bioreactor treating real hospital wastewater with the emphasis on simultaneous carbon, nitrogen and phospho-
rus (CNP) removal. The experiments were based on a central composite design (CCD) and analyzed by response sur-
face methodology (RSM). To analyze the process, three significant variables, aeration time (2-4 h), mixing time without
aeration (30-90 min) and MLSS concentration (2,000-6,000 mg/l), were studied. Results show that an increase in aera-
tion time increased the nitrogen and phosphorous removal efficiency. However, when the aeration time was more than
3 h, the efficiency of phosphorous removal was decreased due to insufficient acidification. A similar scenario was ob-
served when mixing time was increased for phosphorus and nitrogen removal efficiency. MLSS had a positive effect on
all the responses. Under optimal conditions, the concentrations of quality parameter in the influent in average were
recorded as 586 mg COD/l, 296 mg BOD5/l, 97 mgTN/l and 16.47 mg TP/l, which yields the following removal effi-
ciencies, 95.6%, 98.3%, 88.0% and 92.0%, respectively.
Keywords: Hospital Wastewater, Simultaneous Nutrient Removal, RSM

INTRODUCTION charging it into the receiving waters.

Hospital effluents are usually discharged to the urban sewer sys-
Hospital wastewater is one of the major sources of surface and tem where they mix with other effluents and finally reach the sew-
groundwater pollution. A hospital approximately generates about age treatment plant. The quality of hospital wastewater is relatively
400 to 1,200 liters of wastewater per day per bed [1]. Therefore, similar to municipal wastewater. However, conventional wastewa-
serious attention has been directed towards these significant vol- ter treatment processes such as activated sludge with suspended
umes of wastewater which is very complex and includes several growth could not treat hospital wastewater effectively [5,6]. Thus,
inorganic and organic components such as non-metabolized phar- alternative techniques such as integrated bioreactors (anoxic/oxic-
maceutical compounds, antibiotics, disinfectants, anesthetics, patho- membrane bioreactor (A/O-MBR), staged anaerobic-aerobic mem-
gens, radioactive elements, X-ray contrast agents and other persistent brane bioreactor (MBR) and integrated anaerobic-aerobic fixed-
and dangerous compounds [2,3]. Hospital wastewater contains film reactor (FFR)), activated carbon adsorption, advanced oxida-
high concentrations of chlorinated molecules and trace heavy metals tion processes (AOPs), and combinations of them, may be needed
such as mercury and silver. COD and BOD5 contents of the hospital to reach higher removals before the final disposal of the effluents.
wastewaters are about 850 and 600 mg/l, respectively [4]. There- Furthermore, it is necessary in order to reuse the wastewater for
fore, hospital wastewater must be collected and treated before dis- irrigation or ground water recharge.
Removal of organic matter, nitrogen (N) and phosphorus (P)
†
To whom correspondence should be addressed. compounds are often the main aims in designing different types of
E-mail: [email protected], [email protected] wastewater treatment systems. For example, biological wastewater
Copyright by The Korean Institute of Chemical Engineers. treatments such as anaerobic, anoxic and aerobic biological reac-

1340
 Process modeling and optimization of biological removal of CNP from hospital wastewater in a CFID bioreactor 1341

tors are usually used in combinations to provide an efficient treat- mization of a multi-factorial system was not used. Central com-
ment scheme for organics and nutrient removal [7,8]. Recently, sub- posite design (CCD) and response surface methodology (RSM)
stantial attention has been focused towards simultaneous carbon were applied to model and optimize the hospital wastewater biore-
and nutrient removal using compact high-rate directors in biore- actor treatment. RSM is a collection of statistical and mathematical
actors order to satisfy the strict constraints posted with respect to techniques useful for developing, improving and optimizing pro-
space, odor, view and bio-solids production [9]. The integrated cess [16,17]. The main advantage of RSM is that it could reduce
bioreactors, which are a combination of aerobic and anaerobic pro- the number of experimental trials needed to evaluate multiple param-
cesses in a single reactor, has been proven as a viable alternative eters and their interactions. Aeration time, mixing time (without
and could enhance overall degradation efficiency [9]. aeration) and MLSS concentration were identified as the factors or
Wen et al. [10] studied the treatment of hospital wastewater using variables. The interactions among the variables as well as their direct
a submerged membrane bioreactor. The results showed that the impacts on the nine process responses (chemical oxygen demand
removal efficiency for COD, NH4+-N, and turbidity was 80, 93 and (COD) removal, biological oxygen demand (BOD) removal, total
83%, respectively, with the average effluent quality of COD <25 nitrogen (TN) removal, total Kjeldahl nitrogen (TKN) removal,
mg/l, NH4+-N<1.5 mg/l and turbidity <3 NTU. Liu et al. [11] used organic nitrogen removal, effluent nitrite and nitrate concentrations,
a membrane bioreactor to treat hospital wastewater. The results total phosphorus (TP) removal and TSS removal) were discussed.
obtained from this work indicated that a membrane bioreactor
could remove more than 80% of the COD, BOD, TSS, and NH4+-N. MATERIALS AND METHODS
BOD concentration in the effluent varied from non-detectable level
to 20.6 mg/l. Sousa and Foresti stem composed of an UASB reactor 1. Wastewater Source and Characterization
followed by sequencing batch aerobic reactors (SBR) in treating Samples of hospital wastewater (HWW) were obtained from Imam
domestic sewage. The system performance was evaluated through Reza hospital in Kermanshah, Iran. Imam Reza Hospital is one of
a bench scale set-up comprised of a 4 liter volume UASB reactor the biggest hospitals in Kermanshah with a total of 650 beds and
followed by two SBRs of 3.6 liters each. The HRT of 4 h in UASB 1200 staff members. Imam Reza Hospital is located near the north
was maintained constant throughout the study, while the 4 h cycles of Kermanshah town with 1,000,0000 inhabitants. The effluent
in the following sequence of fill (0.10 h), reaction (1.9 h), sedimen- produced by the hospital is directly discharged into the combined
tation (1.6 h), discharge (0.25 h); idle (0.15 h) were maintained in sewage network, where it will be conveyed to the municipal waste-
SBR. The combined system removed about 85% of total nitrogen water treatment plant and co-treated with other urban wastewater.
through nitrification. The COD removal in UASB reactor was around Over a time-period of one month, 24 hours of composite waste-
86% while in SBR around 65% of the remaining; thus, combined water samples were taken daily. This was done to avoid large vari-
systems removed 95%. ations in concentration between the different departments. The
Greentech Company [13] has reported a case study in Dong Thap collected samples were stored in a cold room at 4 oC. The storage
General Hospital Vietnam. The study reported that a combina- technique had no observable effect on its composition. The char-
tion of the activated sludge system and biological contactor-ASBC acteristics of the HWW are shown in Table 1.
has been used to treat the hospital wastewater. The results of this 2. Analytical Methods
study showed 87.8, 71.2, 83.6 and 99.9% removal of COD, Total The concentrations of chemical oxygen demand (COD), total
N, Total P and Coliforms, respectively. Asadi et al. [14] reported a Kjeldahl nitrogen (TKN), nitrate, total nitrogen (TN), total phos-
study on simultaneous removal of carbon and nutrients from an phate (TP), N-organic nitrogen, MLSS and volatile suspended solids
industrial estate wastewater in a single up-flow aerobic/anoxic sludge (VSS) of the system were determined using standard methods of
bed (UAASB) bioreactor. The finding indicated that an optimum
condition for more than 80% removal of COD removal and 50%
of TKN can be achieved with HRT of 12 h and aeration time of Table 1. Characteristics of hospital wastewater
40-60 min/h. Sperling et al. 2011 [15] worked with a UASB - acti- Parameter Range
vated sludge system for the treatment of municipal wastewater. TCOD (mg/l) 450-654
The removal efficiencies for COD and TKN were reported 95 and BOD (mg/l) 220-345
85%, respectively. TSS (mg/l) 259-520
Therefore, considering the above-mentioned scenarios and studies,
TKN (mg/l) 81-120
we tested a continuous feeding and intermittent discharge (CFID)
NH4-N (mg/l) 18-41
bioreactor as a hybrid reactor to treat hospital wastewater with an
NO2-N (mg/l) 0.03-0.3
emphasis to remove carbon, nitrogen (N) and phosphorus (P) com-
NO3-N (mg/l) 0.08-0.36
pounds. Note that aerobic, anoxic and anaerobic conditions in this
N-organic (mg/l) 59-72
single reactor were provided by intermittent aeration. The possibility
to achieve high biomass concentration at a lower hydraulic reten- TN (mg/l) 81.1-120.7
tion time, without additional requirement of equipment to circulate TP (mg/l) 14-19
the mixed liquor between aerobic and anaerobic compartments, Turbidity (NTU) 50-71
will be an advantage of this reactor. pH 7.5-7.9
Furthermore, in this work conventional technique for the opti- Alkanity (mg/l) 376-509

Korean J. Chem. Eng.(Vol. 32, No. 7)

1342 M. Pirsaheb et al.

water and wastewater analysis [8]. For COD, a colorimetric tech- The HWW was continuously introduced into the bioreactor
nique with a closed reflux method was developed. Spectrophotom- from the bottom of the column. The treated effluent was occasion-
eter (DR 5000, Hach, Jenway, USA) at 600 nm wavelength was used ally discharged as supernatant at the end of the each run. The bio-
to measure the absorbance of COD samples. Total Kjeldahl nitro- reactor was inoculated under room temperature (20±2 oC) with
gen (TKN) was determined by TKN meter Gerhardt model (vap- activated sludge from an aeration tank of a municipal wastewater
odest10), whereas for the dissolved oxygen (DO) concentration treatment plant. The inoculums’ MLSS concentration was 6.4 g/l
DO probe (WTW DO CellOX 330, electro DO probe, Germany) and was diluted to the required initial concentrations.
was used. The pH meter model HANNA, pH211 was used to meas- The sequence of the bioreactor operation was controlled by a
ure the pH readings. Turbidity was measured by using a turbidit- pre-programmed timer. Each operation cycle consists of four phases.
meter model 2100p (Hach Co.). In the first phase, the reactor was aerated based on the designed
3. Bioreactor Configuration and Startup aeration time (2 to 4 h) at a constant aeration rate of 5 l/min. In
Fig. 1 shows the layout of the laboratory scale continuous feed- the second phase, aeration was stopped and mixing was started
ing and intermittent discharge (CFID) up-flow bioreactor. This reac- for about 30 to 90 min. The third phase was settling, which lasted
tor was used for biological nutrient removal from the HWW. The for 30 min and was fixed throughout the study. Finally, in the fourth
Plexiglas brand bioreactor column was designed and fabricated phase the effluent was withdrawn for 3 min. Wastewater feeding
with an internal diameter of 8 cm and a liquid height of 80 cm. The and withdrawal were done using peristaltic pumps and control
total volume of the column is 4 l. An automated control valve was valves. Since, biomass concentration is a variable in this study, the
mounted on the reactor column, at a height of 80 cm (25% of the biomass content of the reactor was maintained constantly by remov-
total volume) in order to achieve the intermittent discharge. Air ing surplus biomass after each cycle.
was introduced into the reactor at the bottom of the reactor with a In the first stage (bioreactor start-up), after adding the prepared
bubble air diffuser. The air flow rate and aeration time were con- inoculums, the bioreactor was operated in the batch mode with
trolled by an air flow-meter and a pre-programmed timer respec- the following conditions: cycle time of 6 h ( aeration time of 4 h,
tively. To distribute the feed uniformly in the reactor, an influent mixing time (without aeration) of 90 min and settling time of 30
liquid distributor was mounted at the base of the column. min). In this stage, HWW was used as feed with a COD concen-
tration of about 600 mg/l. This process was continued until a steady
state condition was achieved. In this condition the removal effi-
ciency for COD, TN and TP was around 88, 76 and 83%, respec-
tively. In the second stage, once the bioreactor reach the steady state
condition, the bioreactor was operated in the continuous mode ac-
cording to the three independent variables: aeration time (2-4 h),
mixing time (30-90 min) and MLSS concentration of (2,000-6,000
mg/l). The experimental runs were designed using Design Expert
Software (DOE) (Stat-Ease Inc., version 6.0) as described in Sec-
tion 2.4.
4. Experimental Design
The DOE software creates a platform to design the experimen-
tal runs. DOE removes systematic errors, estimates experimental
error and reduces the number of experiments to obtain the opti-
mum operating conditions. Moreover, it can be used to estimate
the relative significance of several affecting factors in the presence
of complex interactions. When a combination of several independent
variables and their interactions affects desired responses, response
surface methodology (RSM) is an effective tool for optimizing the
process. There are many classes of response surface designs such
as central composite design, Box-Behnken design, hybrid design
and three-level factorial design [18]. Among these methods the most
frequently used RSM design is central composite design (CCD).

Table 2. Experimental range and levels of the independent variables

Variables Ranges and levels
−1 0 1
Aeration time, h 2 3 4
Mixing (without aeration), min 30 60 90
Fig. 1. Experimental setup. MLSS, mg/l 2000 4000 6000

July, 2015
 Process modeling and optimization of biological removal of CNP from hospital wastewater in a CFID bioreactor 1343

Table 3. Experimental conditions and results

Variables Responses
Run COD BOD TKN TN N-organic Effluent Effluent TP TSS
Factor 1 Factor 2 Factor 3
rem. rem. rem. rem. rem. nitrate nitrite rem. rem.
Aeration time, Mixing time, MLSS,
% % % % % mg/l mg/l % %
h min mg/l
01 3 60 4000 70 75 58 50 52 5.26 3.45 70 70.3
02 4 60 4000 83 87 70 64.2 66 4.42 0.56 73 83.3
03 2 30 2000 30 34 20 17.5 16 1.24 1.38 26 48.6
04 4 30 2000 51 57 43 39.2 39 2 1.80 31 61.3
05 4 90 2000 57.6 63 44 37.7 37 3.24 3.20 54 74.3
06 3 60 2000 43.3 47 30 25 21 1.77 3.90 49 80
07 3 60 4000 69 74.1 56 48.8 51 6.3 2.70 69 72
08 3 90 4000 78 80 64 60 62.3 2.36 1.59 80 76.6
09 3 60 4000 71.8 76.2 60 51 54 5 2.60 71 71
10 4 30 6000 87.6 92 81 74 77 5.7 1.18 67 92
11 2 30 6000 70.3 76 59 54.6 55 2.78 1.97 58 74
12 2 60 4000 60.6 64 50 45 45 5.4 1.24 60 64
13 2 90 6000 77.3 82 64 58 59 5.7 0.87 80 77
14 3 60 4000 73 79 60 51 52 4.9 2.20 70 69.5
15 3 30 4000 63.6 68 55 42 48 3.47 7.44 60 63.6
16 3 60 4000 72 76 58.9 50.8 52.6 5.16 2.48 69 72
17 3 60 4000 69 75.1 59 51 52.7 5.2 2.53 71 71.7
18 2 90 2000 35 40 25 21 18 1.81 2.53 40 41
19 3 60 6000 82 89 79 73 75.6 5.1 1.04 81 91.3
20 4 90 6000 95 96 90 86.47 95.3 3.02 4.56 90 98.6

In the current study, CCD was used to study the three different respectively. All these coefficient variables are analyzed by multiple
factors: aeration time (A), mixing time (B) and MLSS concentra- regression analysis. Response contour plot will be generated using
tion (C). The region of exploration for the process was enclosed by DOE. Model terms are selected or neglected based on the probabil-
aeration time dosage (2-4 h), mixing time (30-90 min) and MLSS ity of error (P) value with 95% of confidence level. The results ob-
(2,000-600 mg/l) to evaluate nine different responses. The design tained from CCD were examined by analysis of variance (ANOVA).
consists of 2k factorial points augmented by 2k axial points and a Three-dimensional (3D) plots and their respective contour plots
center point, where k is the number of variables. The ranges or lev- were obtained based on the effect of the levels of the two factors
els of the parameters are shown in Table 2. The coded values of A, while other factors can be changed by default. Therefore, the results
B and C are set at three levels: 1 (minimum), 0 (central), +1 (max- of CCD can be presented in 3D presentations with contours. This
imum). These three levels were assessed based on the full face-cen- will help to study the simultaneous interaction of the three vari-
tered CCD experimental plan. Accordingly, 20 experiments were ables on the responses. The experimental conditions and results
employed in this study. Repetitions were conducted according to are depicted in Table 3.
the order of the runs designed by CCD as indicated in Table 3.
5. Mathematical Modeling RESULTS AND DISCUSSION
RSM involves screening and codification of the variables, math-
ematical-statistical treatment of data, and evaluation of the fitted 1. Statistical Analysis
model and determination of the optimal conditions. RSM, describes CCD was selected to find the relationship between the process
a model in the form of Eq. (1) to fit the experimental data and by responses and the variables. Table 4 shows a complete list of the
optimization, the coefficients for the model were calculated. The three independent variables (A, B and C) in the terms of coded
relationship between the responses, input and the quadratic equa- and actual units, and the experimental data obtained for the nine
tion model for predicting the optimal variables were identified using responses (Y1−Y9). 20 experimental runs were performed in accor-
the following: dance with Table 3 by CCD.
Table 4 illustrates the reduced models in terms of coded factors
Y=β0+βiXi+βjXj+βiiXi2+βjjXj2+βijXiXj+… (1)
with significant model terms and analysis of variance (ANOVA)
where, Y, i, j, β, X are process response, linear coefficient, quadratic results for the responses. Various responses were investigated with
coefficient, regression coefficient and coded independent variables, different degree polynomial models for data fitting (Table 4). To

Korean J. Chem. Eng.(Vol. 32, No. 7)

1344 M. Pirsaheb et al.

Table 4. ANOVA results for the equations of the Design Expert 6.0.6 for studied responses
Modified equations with Type of Adj. Pred. Adeq. Probability
Response R2 S.D CV PRESS P-value F-value
significant terms model R2 R2 precision for lack of fit
COD Y1 =+71.00+10.10A+4.04B Quadratic 0.99 0.99 0.98 70.242 1.92 02.86 101.48 <0.0001 025.58 0.3684
removal, % Y1 =+19.53C− 8.09C2
BOD Y2=+75.44+9.90A+3.40B Quadratic 0.99 0.99 0.98 62.300 1.92 02.68 096.79 <0.0001 22.6 0.3627
rem.% Y1 =+19.40C− 7.84C2 − 2.00AC
TN Y3 =+50.54+11.40A+5.14B Linear 0.99 0.97 0.96 59.850 2.80 05.54 232.91 <0.0001 252.70 0.0600
rem.% Y1 =+20.92C
TKN Y4 =+59.09+11.00A+2.90B Quadratic 0.99 0.98 0.97 62.500 2.24 03.98 148.18 <0.0044 008.39 0.1307
rem.% Y1 =+21.10C− 5.59C2
N-organic Y5 =+51.43+12.13A+3.66B Linear 0.95 0.94 0.91 37.100 4.69 09.11 675.35 <0.0001 105.30 0.0007
rem. Y1 =+23.09C
Effluent Y6 =+4.85+1.32C− 1.62B2 Quadratic 0.65 0.61 0.48 11.230 0.98 24.27 024.30 <0.0261 004.75 0.0500
nitrate
Effluent Y7 =1.7 Mean - - - - - - - - - -
nitrite
TP Y8 =+70.01+5.10A+10.20B Quadratic 0.98 0.97 0.95 43.190 2.78 04.40 245.76 <0.0001 032.14 0.4540
rem.% Y1 =+17.60C− 6.06A2− 7.56C2
TSS Y9 =+72.54+10.49A+12.77C Quadratic 0.92 0.89 0.79 21.730 4.42 06.13 747.24 <0.0110 006.36 0.0500
rem.% Y1 =− 10.71B2+9.84C2+3.02AB

quantify the curvature effects, the data from the experimental results presented in the following sections.
were fitted to higher degree polynomial equations: two factor interac- 2. Process Performance
tion (2FI), quadratic etc. The finalized model terms in the equations 2-1. Carbon Removal
are those remaining after the elimination of insignificant variables The mean concentration of BOD and COD of raw wastewater
and their interactions. of 70 hospitals in Iran, in terms of wastewater strength, has been
The values obtained from the ANOVA analysis determine the reported as 348 and 527 mg/l, respectively [22]. These readings are
rank of the significance’s degree. For each response, the F-value and close to the results of the present study. Table 1 shows the mean
P-value were computed to determine the significance of the model concentration of BOD5 and COD in the raw HWW as 296 and
terms. The greater the amount of F-value, the smaller will be the 586 mg/l (correspond to the BOD5/COD ratio of 0.5), respectively.
values of ‘Prob>F’. This will indicate that the corresponding mod- BOD5/COD ratio provides a good measurement of wastewater bio-
els and the individual coefficients are more significant [19]. degradability, whereby a BOD5/COD ratio greater than 0.4 is gen-
In Table 4, nine models (Y1−Y9) were developed with the fol- erally accepted as biodegradable [23]. From the literature it is shown
lowing F-values: 25.58, 25.6, 252.7, 8.39, 105.3, 12.33, 4.75, 32.14 that the BOD5/COD ratio for HWW is higher than 0.4. Hence, this
and 6.36. The probability values were very low (in the range of 0.0001- wastewater can be categorized as biodegradable wastewater [24].
0.0261). This implies that the terms were significant for all the mod- The effects of aeration time and MLSS at a constant mixing time
els. However, the lack-of-fit are insignificant (bigger than 0.05) for of 90 min (without aeration) on two responses, namely COD and
N-organic removal. BOD removal efficiency, are described by empirical models in the
The fit of the models were further verified by the correlation Table 4. The trend is shown in Fig. 2(a) and 2(b).
coefficients, R2, adjusted R2 and predicted R2 between the experi- Fig. 2(a) and 2(b) shows that the trend for COD and BOD re-
mental and the model predicted values. Table 4 proves that the cor- moval efficiency was the same as the aeration time and MLSS changed
relation coefficients, R2, adjusted R2 and predicted R2 are near to from 2 to 4 h and 2,000 to 6,000 mg/l, respectively. The COD and
each other and close to 1.0 except for effluent nitrate. BOD5 removal efficiency significantly increased with increase in
Adequate precision measures the signal to noise ratio. A ratio aeration time and MLSS. Mixing time (range 30-90 min) did not
greater than 4 is desirable. In all the cases, the value of adequate show any strong effect on the process because the changes were
precision was around 11.23-70.24. This value indicates adequate very small. However, the improvement of COD and BOD removal
model discrimination. Furthermore, low values of the standard efficiency was about 8% when the mixing time was increased from
deviation (SD) (1.8-4.69) and coefficient of variation (CV) (2.68- 30 to 90 min. Therefore, a mixing time of 60 min was chosen as
9.11%) indicated good precision and reliability of the experiments optimal condition.
as suggested by Khuri and Cornell [18] and Ahmad et al. [19]. Maximum COD (96.58%) and BOD (98.3%) removal efficiency
As mentioned, except for nitrate model, the other proposed mod- were observed at the aeration time of 4 h, MLSS of 6,000 mg/l and
els could be adequately used to describe the responses under a wide mixing time of 90 min. Meanwhile, the lowest predicted removal
range of operating conditions. Detailed analyses on the models are efficiency of COD (29.24%) and BOD (32.9%) was obtained at the

July, 2015
 Process modeling and optimization of biological removal of CNP from hospital wastewater in a CFID bioreactor 1345

greater than those with the highest MLSS. Fig. 2 shows that the
response increased upon increasing the aeration time at lower val-
ues of MLSS, and at higher MLSS, aeration time did not show any
significant effect on TCOD removal. This is due to sufficient amounts
of microorganisms at higher HRT, which makes the response rela-
tively independent of aeration time or HRT. As a result, when MLSS
was increased, lesser aeration time or HRT was needed. Mansouri
et al. [25] reported a similar effect of the cycle time and aeration
on the COD removal efficiency in the SBR reactor, so that a maxi-
mum COD removal efficiency was achieved 87.18% at the cycle
time and aeration time of 6.5 h and 50 min/h, respectively. A sequenc-
ing batch flexible fiber biofilm reactor was examined for the treat-
ment of dairy wastewater at three different OLR (0.4, 1.27 and 2.74
kg COD m−3 d−1) and 24 h aeration time by Abdulgader and cowork-
ers [26]. An inverse relationship between OLR and COD removal
efficiency was observed in this study. In another study, the interac-
tive effects of initial chemical oxygen demand (CODin), MLSS and
aeration time on the performance of a lab-scale sequencing batch
biofilm reactor (SBBR) treating a synthetic dairy wastewater were
investigated [27]. The results of this study indicated that as the aer-
ation time and MLSS increased the COD removal efficiency was
increased. Also, the reverse impact of the CODin on COD removal
was observed as the variable increased. Kargi and Konya [26] found
that stepwise increase in HRT from 5 to 15 h resulted about 40 per-
cent increase in COD removal and the efficiency remained almost
constant at larger HRT levels. Meng et al. [27] also reported simi-
lar results.
2-2. Nitrogen Removal
Biological nitrogen removal involves aerobic nitrification and
anoxic denitrification. The nitrification process contributes to the
transformation of ammonia to nitrite (NO2− ), and then to nitrate
(NO3− ). Therefore, in a conventional treatment process, nitrogen is
removed by using at least two separate reactors under different
environments (two kinds of bacteria: nitrifiers and denitrifiers)
[30]. The main drawbacks of conventional treatment processes are
their complicated operation and high cost due to the addition of
Fig. 2. 3D surface plots for carbon removal with respect to aeration external carbon source and recycling for the effluent denitrification
time and MLSS at constant value of mixing time (90 min);
(a) COD removal, (b) BOD removal.
process. Thus, in this study, the two conditions for nitrogen removal
were provided in a single reactor by implementing intermittent
aeration.
aeration time of 2 h, MLSS of 2,000 mg/l and mixing time of 30 The sum of organic-N, NH4+-N-N, NO2− -N and NO3−-N could
min. The influence of aeration time and MLSS on COD and BOD be regarded as TN. Thus, any change in the nitrogen compounds
removal efficiency was more significant than mixing time, which can be illustrated by TN removal efficiency. The response surface
had limited effects on both responses. plot for TN removal was made as a function of aeration time (A)
Note that the HRT in this study was in the range of 3 h (aera- and MLSS (C) of the system, while keeping the mixing time at 90
tion time (2 h): mixing time without aeration (30 min): settling time min (as the factor had no major effect on the model). Fig. 3(a) illus-
(30 min)) to 6 h (4 h (aeration time): 90 min (mixing time with- trates the effect of the two variables (A and C) on TN removal effi-
out aeration): 30 min (steeling time)). The range of HRT (3-6 h) ciency. As it is obvious from Fig. 3(a), the response increased with
studied corresponds to organic loading rate (OLR) and feed flow simultaneous increasing in aeration time and MLSS. The maximum
rate of 10.4-20.8 g COD/d and 16-32 l/d, respectively. As shown in predicted TN removal (88%) was achieved when aeration time and
Figs. 2 and 3, at the aeration time of 2 h (corresponding to HRT of MLSS were at the highest level (aeration time of 4 h and MLSS of
3 h) (aeration time=2 h, non aeration time with mixing=30 min 6,000 mg/l) with mixing time of 90 min. It is obvious from the 3D
and settling time=30 min), the COD, BOD and TN removal effi- surface and perturbation plots (Fig. 3(a) and (b)) that the MLSS
ciency are lower than other HRT, which was due to high organic plays a predominant and positive effect on the TN removal.
loading rate in this condition. This is because the high MLSS leads to the increase of the NO3−
The effect of aeration time at the lower values of the MLSS was and cell production as well as the favored condition for denitrifica-

Korean J. Chem. Eng.(Vol. 32, No. 7)

1346 M. Pirsaheb et al.

Fig. 3. (a) 3D surface plot and (b) Perturbation plot for TN removal efficiency with respect to aeration time and MLSS at a constant value of
mixing time (90 min).

Fig. 4. (a) 3D surface plot and (b) Perturbation plot for TKN removal efficiency with respect to aeration time and MLSS at a constant value
of mixing time (90 min).

tion resulted from high DO consumption rate, and consequently suming readily bCOD content of the fresh feed, which is not sig-
more nitrogen compounds removed. Therefore, increasing the MLSS nificant.
is more preferable than increasing aeration time or mixing time Fig. 4(a) represents the response surface plot of the quadratic
for the purpose of obtaining a high TN removal. Since settling time model for variation in TKN removal, as a function of aeration time
was constant in this study and samples were taken at the end of (A) and MLSS (C) with a constant mixing time (90 min). The per-
each cycle (aeration+non-aeration+settling) after settling, the results turbation plot (Fig. 4(b)) also shows the comparative effects of vari-
show the effect of all the phases (aeration+non-aeration) provided. ables on TKN removal efficiency. In Fig. 4(b), steep curvatures in
However, it is clear that the most of denitrification reaction (TNin- aeration time and MLSS curves show that the response of TKN
TNout) occurred in the anoxic condition provided in the non-aera- removal efficiency was very sensitive to these factors. With a simul-
tion phase. As the system is not mixed in the settling step, a small taneous increase in both variables, the TKN removal efficiency
volume of the biomass at the bottom of reactor was exposed with was increased, favoring the nitrification condition. In terms of inter-
the raw feed. So, the rate of settling step may only be related to the action effect of aeration time and MLSS, as shown in Fig. 3(a), it is
phosphate accumulating denitrifiers in producing PHBs by con- evident that the TN removal tends to reach the peak at the condi-

July, 2015
 Process modeling and optimization of biological removal of CNP from hospital wastewater in a CFID bioreactor 1347

Fig. 5. Nitrogen fractionation in (a) influent and (b) effluent under

different operational conditions.

tion of high aeration time (4 h) and MLSS (6,000 mg/l). By com-

paring the results obtained for TKN removal with TN removal (Fig.
3(a) and 4(a)), a similar trend in the responses was obtained, indi-
cating an appropriate proportion between nitrification and denitri-
fication processes. The maximum TKN removal was determined
to be 88.5% at aeration time, MLSS and mixing time 4 h, 6,000 mg/l
min and 90 min, respectively.
In similar work, the performance of continuous flow intermit-
tent decant type sequencing batch (CFID) reactor was investigated Fig. 6. 3D surface plots for (a) N-organic removal efficiency and (b)
in different HRTs (22, 8 and 6 h) and dissolved oxygen (DO) pat- effluent nitrate with respect to aeration time and MLSS at a
constant value of mixing time (90 min).
terns (0.5, 2.5-3.5 and 3.5-4.5 mg/l) by Khan et al. [31]. The opera-
tion under DO limiting conditions (0.5 mg/l) showed a suppression
of nitrification and a negligible removal of NH4-N disregarding that it is possible to derive a significant organic nitrogen fraction
the HRT. Under the operation of non-limiting DO conditions (2.5 from the metabolic products which is generated by the microbes
mg/l) high nitrification rate constants of greater than 0.89 h-1 were in the wastewater.
observed. The highest effluent quality was observed at the 8 h HRT The effect of variables on the organic nitrogen removal efficiencies
and 2.5-3.5 mg/l DO concentration. At this operational condition, are shown in Fig. 6(a) as a 3D plot. The organic nitrogen removal
the average BOD, TSS, ammonia nitrogen and fecal coliform re- increased significantly with simultaneous increasing aeration time
moval efficiencies were 83, 90, 74 and 99%, respectively. and MLSS. However, when MLSS was increased, the effect of aer-
As shown in Fig. 5(a)-(b), a major fraction of the total nitrogen ation time on the response was reduced. Also, from Fig. 6(a) the
in influent and effluent of HWW is organic nitrogen. Imam Reza maximum N-organic removal was achieved at the highest aeration
Hospital wastewater has about 65% of the organic nitrogen. To nitrify, time and MLSS of 4 h and 6,000 mg/l, respectively. At this condi-
organic-nitrogen must be converted to ammonia/ammonium. If it tion, maximum TN and TKN removal were also obtained. This
is not converted to ammonia/ammonium, then the organic-nitro- indicates a balance between nitrification and denitrification pro-
gen will pass through the treatment plant. The origin and compo- cesses. This proves that N-organic consumption was higher. An
sition of organic in HWW are unknown. However, it is known that interesting finding was also observed: it is expected that the N-organic
the organic comprises large portions of proteins, free and combined removal will decrease due to the domination of aerobic condition
amino acids, low molecular weight (LMW) aliphaticamines, urea per anaerobic condition; however, when the aeration time was in-
and amides [32]. Pehlivanoglu-Mantas and Sedlak [29] also reported creased, N-organic removal increased for all the HRTs tested. This

Korean J. Chem. Eng.(Vol. 32, No. 7)

1348 M. Pirsaheb et al.

Table 5. Mass balance model for TN removal in different runs except repeating ones
Factor Factor Factor
Influent parameters Effluent parameters
1 2 3
Aera- Nitrogen Nitrogen
Run Mixing NH4- NO2- NO3- N- NH4- NO2- NO3- N-
tion MLSS, TKN TN TKN TN removal by removal
time, N N N org. N N N org.
time, denitrificatin by sludge
h min mg/l mg/l mg/l mg/l mg/l mg/l mg/l mg/l mg/l mg/l mg/l mg/l mg/l g/d g/d
01 3 60 4000 106 41 0.19 0.27 65.00 106.46 44.52 13.32 3.45 5.26 31.20 53.23 1.68 0.31
02 4 60 4000 084 25 0.09 0.20 59.00 084.29 25.20 05.14 0.56 4.42 20.06 30.18 1.37 0.30
03 2 30 2000 094 35 0.12 0.21 59.00 094.33 75.20 25.64 1.38 1.24 49.56 77.82 1.20 0.19
04 4 30 2000 098 29 0.04 0.10 69.00 098.14 55.86 13.77 1.80 2.00 42.09 59.67 1.59 0.22
05 4 90 2000 100 40 0.03 0.20 60.00 100.23 56.00 18.20 3.20 3.24 37.80 62.45 1.75 0.14
06 3 60 2000 110 43 0.06 0.17 67.00 110.23 77.00 24.07 3.90 1.77 52.93 82.67 1.76 0.15
08 3 90 4000 096 39 0.12 0.16 56.72 096.28 34.56 13.18 0.15 2.36 21.38 38.51 1.54 0.30
10 4 30 6000 096 30 0.29 0.33 66.00 096.62 18.24 03.06 1.18 5.70 15.18 25.12 1.59 0.34
11 2 30 6000 103 28 0.26 0.24 75.00 103.50 42.23 08.48 1.97 2.78 33.75 46.99 1.36 0.47
12 2 60 4000 120 41 0.19 0.27 79.00 121.17 60.00 16.55 1.24 5.40 43.45 66.64 1.78 0.34
13 2 90 6000 107 38 0.18 0.20 69.00 107.38 38.52 10.23 0.87 5.70 28.29 45.10 1.61 0.41
15 3 30 4000 081 35 0.30 0.36 46.00 081.66 36.45 12.53 7.44 3.47 23.92 47.37 1.13 0.32
18 2 90 2000 103 41 0.17 0.08 62.00 103.25 77.25 26.41 2.53 1.81 50.84 81.57 1.56 0.16
19 3 60 6000 100 34 0.29 0.23 66.00 100.52 21.00 4.896 1.04 5.10 16.10 27.14 1.64 0.36
20 4 90 6000 095 23 0.10 0.14 72.00 086.47 9.5 6.116 0.36 3.02 3.384 8.647 1.48 0.33

could be due to the high nitrification rate in the high aeration time. dicted R2 value was negative (-0.12). A negative “Pred R-Squared”
Choi et al. [30] evaluated the performance of an intermittently aer- implies that overall mean can be a better predictor for this response.
ated membrane bioreactor (IAMBR) across several COD/N ratios. The mean of nitrite concentration in the influent was 1.7 mg/l. In
Their results showed that the increase of the COD loading rate led this study, NO2-N concentration was reported lower than 2.53 mg/l.
to a higher denitrification rate and better assimilation of organic The decrease of NO2-N in the bioreactor could be detected with
matter and nutrients. Asadi et al. [14] studied the treatment effi- an increase in aeration time and MLSS and a decrease in mixing
ciency of an industrial estate wastewater with low BOD5/COD ratio time. This scenario indicates that it is possible to have an alterna-
in an up-flow aerobic/anoxic sludge bed (UAASB) bioreactor, with tive occurrence of nitrification and denitrification.
an intermittent regime in aeration and discharge. In this study, simul- Table 5 shows the mass balances of nitrogen in different runs
taneous increase in the HRT and aeration time reduced TN removal obtained by measuring the amount of nitrogen in the influent and
efficiency. This is a result of insufficient carbon source and restric- effluent and estimating of the nitrogen content removed through
tion in anoxic conditions at the higher HRT and aeration times. the waste sludge discharge. The influent nitrogen concentration
2-3. Effluent Nitrite and Nitrate ranged from 81.1-121 mg/l and the major constituent was organic
Fig. 5(a) and 5(b) show the trend of nitrogen compounds in influ- nitrogen. The nitrogen in the effluent after one reaction cycle con-
ent and effluent of the bioreactors for all the experiments except for sisted of NH4-N; NOX−_N, organic nitrogen and nitrogen in the
the repeated runs. It can be concluded that the effluent nitrate and suspended sludge. A certain amount of sludge was removed daily
nitrite concentration are very low compared to organic and ammo- from the reactor to maintain constant SRT. The amount of nitro-
nium nitrogen in the raw wastewater. Fig. 6(b) shows the effect of gen removed in the waste sludge was assumed to be 0.12 mg N/
mixing time and MLSS on effluent nitrate. It was observed that an mg VSS, as suggested by Metcalf and Eddy [8]. The percentage of
increase in the MLSS caused an increase in the effluent NO3− con- influent nitrogen removed by denitrification and assimilation pro-
centration. This is due to an increase in oxidation potential that cess was in the range of 59-81%, and 6-21%, respectively. The results
favored nitrification process. Two opposite impacts of the mixing indicated that the CFID bioreactor exhibited a high performance
time on effluent nitrate were observed as the variable increased. Efflu- of TN and TKN removal, at aeration time of 4 h, non-aeration time
ent nitrate concentration was enhanced with mixing time increased of 1.5 h and MLSS of 6,000 mg/l. This could be attributed to a bal-
from 30 to 60 min. However, further increment in the mixing time ance of nitrification and denitrification processes in this condition.
from 60 to 90 min decreased the response. Note that the effluent Based on the calculation, the maximum N2 production in the sys-
NO3− concentration (<7.44 mg/l) was low in all the experiment, indi- tem was obtained in aeration time of 2 h and non-aeration time of
cating the appropriate proportion between nitrification and denitri- 1 h and MLSS of 200 mg/l, which was due to a high TN loading
fication processes. in this condition. The relatively short HRTs and SRTs (see Table 3)
For effluent nitrite, RSM did not propose any model and the pre- were not favorable for nitrification because the autotrophic nitrifi-

July, 2015
 Process modeling and optimization of biological removal of CNP from hospital wastewater in a CFID bioreactor 1349

Fig. 7. 3D surface plots for TP removal efficiency with respect to aeration time and MLSS at constant values of mixing time; (a) mixing
time=30 min (b) mixing time=60 min (c) mixing time=90 min, (d) perturbation plot.

ers are slow-growing bacteria and are washed out at low SRTs. was obtained at the lowest values of all the three variables. This proves
2-4. Total Phosphorus (TP) Removal that all three variables have significant effects on TP removal effi-
Based on the literature, an anaerobic/aerobic sequence is neces- ciency, as shown in Fig. 7(d), the perturbation plot.
sary to promote biological P removal. P was released in the anaer- Fig. 7(a)-(c) shows an increase in the aeration time from 3 to
obic stage followed by an excess of P uptake in the aerobic stage. P 4 h causes a decrease in the anaerobic condition, whereby at this
accumulation as poly-phosphates is bigger than the P release [14]. timing the phosphorus accumulating organisms (PAOs) accumu-
Thus, in this study, the aerobic/anoxic/anaerobic conditions for nitro- late poly hydro butyrate (PHB) from the volatile fatty acids (VFAs)
gen and phosphorus removal were provided in a single reactor by produced. In this process, COD and BOD (as the source of VFAs)
implementing intermittent aeration. Fig. 7 illustrates the overall TP require sufficient time for acidification [34]. Another reason for
removal efficiency (%) as a function of aeration time and MLSS the decrease in the phosphorous removal at high aeration time was
concentration at three different mixing times. Generally, the reactor due to the presence of nitrate, which inhibits the fermentation pro-
could achieve a high level of TP removal. Furthermore, an increase cesses which produces VFAs in the anaerobic zone. Studies have
in MLSS and aeration time from 2 h to 3 h increased the TP removal. shown that biomass subjected to using alternating anoxic and aer-
However, further increment in aeration time (3 h to 4 h) decreased obic conditions (activated sludge was cycled between anaerobic
the response. Maximum TP removal efficiency was about 92% under and aerobic phases) would promote the accumulation of PAOs [14,
the following conditions: aeration time of 3 h, 6,000 mg/l of MLSS 35]. Moreover, in this study, sludge recycle line was omitted, sug-
and a mixing time of 90 min. The lowest removal efficiency (23%) gesting that external anoxic (or anaerobic) and aerobic conditions

Korean J. Chem. Eng.(Vol. 32, No. 7)

1350 M. Pirsaheb et al.

have been alternated for the biomass. However, results showed that Accumulation=inflow-outflow-generation
PAOs existed in this single bioreactor. Thus, it can be concluded dC
------- V = QC0 − QC + rcV (2)
that there was an inner sludge recycle between the anoxic and oxic dt
zone, which induces PAOs’ accumulation. The liquid recycle flow
where, V is volume of the reactor (lit); Q is feed flow rate (l/d); C0
was mainly dependent on the extent of activated sludge mixing.
is influent COD concentration (mg/l); C is effluent COD concen-
The flow was driven by concentration differences between the anoxic
tration (mg/l) and rc is the reaction rate.
and oxic zones. Therefore, due to the possibility of inner recycle in
Assuming first-order removal kinetics (rc=−KC), Eq. (1) can be
the experiment, traditional denitrification could have been respon-
rearranged and written as follows:
sible for the efficient TN removal. These findings were in a good
agreement with those obtained by Asadi et al. [14]. Q
C' + BC = ----C0 (3)
2-5. Effect of DO Concentration on TN and TP Removal V
The dissolved oxygen (DO) concentration has significant impact
where and C'=dC/dt and B=k+Q/V
on the success of the nitrogen and phosphorus removal process.
To solve Eq. (3) both sides of the expression are multiplied by
DO plays a crucial role in nitrification, while it has an adverse impact
the integrating factor eBt:
on biological denitrification. The negative effects of high DO con-
centrations on the denitrification process depend on the amount Bt Q
(C' + BC)e = ---- C0e
Bt
(4)
and type of carbon source. As mentioned above, in this study, aer- V
obic and anoxic condition were provided by intermittent aeration.
The left-hand side of the above expression can be written as a dif-
During the aeration period, DO concentration in the reactor main-
ferential as follows:
tained in the range of 4-6 mg/l. Based on the obtained results, the
DO concentration was sufficient for nitrification as proven by high Q Bt
(Ce )' = ---- C0e
Bt
(5)
NO3− production and TKN removal. It is generally known that DO V
concentration above 1 mg/l is essential for complete dominated
nitrification; at lower DO levels, oxygen becomes a limiting factor Integration of Eq. (5) yields
for the nitrification process [36]. The high TKN removal obtained Bt Q C Bt
in this study could be explained by the relatively complete nitrifi- Ce = ---- ------0e + K (6)
VB
cation in the duration of aeration. In the non-aeration phase, the
DO concentration in the reactor was lower than 1 mg/l (0.1-1 mg/l) But when t=0 and C=C0 , K is equal to
where the denitrification process was progressing. As presented in QC
Table 4, TN and TKN removal efficiencies were stable and the aver- K = C0 − ---- ------0 (7)
VB
age values increased up to 88% for both responses. Influent nitrite
and nitrate levels were low, and generally nitrate in permeate was Substitution of K in the Eq. (6) and its simplification yields the fol-
lower than that in the influent. However, a slight increase in the lowing expression:
effluent and accumulation of nitrite in the reactor was found in QC −Bt −Bt
the high aeration time. It meant that the reactor offered relatively C = ---- ------0( 1− e ) + C0e (8)
VB
complete nitrification and denitrification.
Phosphorus can only be removed by its uptake into biomass, By solvng the Eq. (8) under steady-state condition (i.e., the accu-
which can be discharged from the system as surplus sludge. Thus, mulation rate is assumed to be zero (dc/dt=0)), we will have:
a biomass with high phosphorus content is desirable for biological C0 C0
phosphorus removal. Removal of phosphorus in wastewater is closely - = ----------------
C = ------------------ - (9)
⎛1+ k --- V-⎞ ( 1+ kτ)
dependent upon the phosphorus release in anaerobic conditions ⎝ Q⎠
and on the subsequent uptake process of the excess phosphorus,
including that contained in wastewater in aerobic conditions [25]. The CFID bioreactor without recycle, as shown in the Fig. 1, receives
In the present study, as the system is intermittently aerated, a micro wastewater with COD and TN concentration of 650 and 120 mg/l,
anaerobic environment seems to be provided in the biofloc formed respectively, in the optimal condition. In this condition, the flow
in the process. From the results obtained and in the cycle tests car- rate is 16 l/d and the reactor effluent COD, TN and VSS are 0.022,
ried out under normal operating conditions, denitrification was 0.014 and 0.18 g/l, respectively.
assumed to occur mainly through the nitrate pathway. Ideally, PAO The values of k for COD, TN and TP removal, in optimal con-
could be selected to simultaneously reduce nitrate and perform dition (aeration time=4 h, mixing time=90 min, settling time=30
anoxic dephosphatation. The low DO applied in the non-aeration min and MLSS=6,000 mg/l), were computed to be 3.87, 1.2 and
period favored the development of denitrifying PAOs, which en- 1.39 h−1, respectively, as shown in Table 6. The other results from
hanced phosphate removal coupled to denitrification (denitrifying mass balance analysis in the optimal condtion are summarized in
dephosphatation). Table 6. As shown in the table, VSS produced, COD removed, ob-
2-6. Mass Balance Analysis in the Optimal Condition served yield, oxygen used per unit of COD removed, TN removed
For a UAASFF reactor without biomass recycle, the rate of change and N2 produced from denitrification were calculated to be 2.88
in the substrate can be expressed as Eq. (2): gVSS/d, 10.045 g COD/d, 0.286 g VSS/g COD removed, 0.6 g O2/g

July, 2015
 Process modeling and optimization of biological removal of CNP from hospital wastewater in a CFID bioreactor 1351

Table 6. Kinetic parameters based on the mass balance model in the optimal condition
Yobs, Oxygen N2 produced
Effluent VSS COD TN
Variables (optimum Inffluent k, g VSS/g used per from
Response parameters, −1 produced, removed, removed,
condition) parameters h COD unit COD, denitrification,
mg/l gVSS/d g/d g/d
removed g O2/g COD g/d
CODin=650 mg/l CODin=22
HRT=6 h TN=120 mg/l TN=14
COD Aeration time=4 h 3.87 2.88 10.045 0.286 0.6 1.69 1.33
TP=19 mg/l TP=2.03
Mixing time=90 min Flow rate=16 l/d VSS=180
Settling time=30 min
TN - - 1.20 - - - - - -
MLSS=6000 mg/l
TP - - 1.39 - - - - - -

Table 7. The optimization criteria for chosen response

Response Limits Unit
COD removal >90 %
BOD removal >90 %
TN removal >80 %
TKN removal >80 %
N-organic removal >80 %
Effluent nitrate <60 mg/l
Effluent nitrite <30 mg/l
TP removal >90 %
TSS removal >90 %

Fig. 8. 3D surface plot for TSS removal efficiency with respect to

aeration time and MLSS at a constant value of mixing time
(90 min).

COD, 1.69 g TN/d and 1.33 g/d, respectively.

2-7. TSS Removal
Fig. 8 shows the removal of TSS. More than 98% (actual value)
of removal was achieved at the following conditions of reaction
time (4 h), MLSS (6,000 mg/l) and mixing time (90 min). Drastic
increase was observed in the TSS removal by increasing the MLSS
concentration. The removal efficiency of TSS was reported low at
minimum levels of reaction time and MLSS concentration and maxi-
mum level of mixing time. The high value of the biomass concen-
tration causes an increase in the sludge volume index (SVI). In this
study, high values of the SVI were found in the high values of the
biomass concentration because of decrease in dissolved oxygen
concentration and F/M ratio. Low-DO bulking is brought about
by filamentous bacteria such as Sphaerotilusnatans. They begin to
predominate when the dissolved oxygen concentration is not high Fig. 9. Overlay plots for the optimal regio.
enough to allow good oxygen penetration into the flock [37].
2-8. Process Optimization and Verification
Graphical optimization produces an overlay contour plot that adopted as shown in Table 7. The shaded area in the overlay plots
demonstrates the feasibility area for the responses. The optimum is the region that meets the proposed criteria. Fig. 9 shows the graphi-
region was identified based on nine critical responses, which were cal optimization as a function of aeration time and mixing time

Korean J. Chem. Eng.(Vol. 32, No. 7)

1352 M. Pirsaheb et al.

Table 8. Verification experiments at optimum conditions

Responses

Run Conditions COD BOD TN TKN N-organic Effluent Effluent TP TSS

removal Removal removal removal removal nitrate nitrite removal Removal
(%) (%) (%) (%) (%) (mg/l) (mg/l) (%) (%)
1 Experimental values Aeration time=4 h 96.58 98.3 88 88.5 90.3 4.8 1.7 89.28 94.9
Model response with Cl 95% Mixing time=90 min 92.8 95.4 83 85.6 89.2 5.1 1.64 86 90.6
Error MLSS=6,000 mg/l
Standard deviation −3.9 −2.9 −5.6 −3.2 −1.2 +6.25 −3.5 −3.6 −4.5

with a constant value of MLSS concentration (6,000 mg/l). The yel- 5. C. Lenz, G. Koellensperger, S. Hann, N. Weissenbacher, S. N.
low highlighted area satisfies the constraints, while the area that Mahnik and M. Fuerhacker, Chemosphere, 69, 1756 (2007).
does not meet the criteria is gray. The optimal region enclosed by 6. P. Verlicchi, A. Galletti, M. Petrovic and D. Barcelo, J. Hydrol., 389,
the aeration time (3.5-4 h) and mixing time (75-90 min) bound- 416 (2010).
ary is at the MLSS concentration of 6,000 mg/l. To verify the accu- 7. A. Akhbari, A. A. L. Zinatizadeh, P. Mohammadi, M. Irandoust
racy of the models, a point within the optimum region was chosen and Y. Mansouri, Chem. Eng. J., 168, 269 (2011).
(conditions shown by flags in Fig. 9). The bioreactors were oper- 8. Metcalf, Eddy, Wastewater Engineering: Treatment and Reuse, fourth
ated accordingly to compare the actual responses with the predicted Ed. McGraw-Hill, New York, USA (2003).
values. Table 8 presents the results of this experiment conducted 9. B. Tartakovsky, M. F. Manuel and S. R. Guiot, Biochem. Eng., 26,
within the optimum regions (Fig. 9). The accuracy of the optimum 72 (2005).
conditions, found for each response from the DOE experiments, 10. X. Wen, H. Ding, X. Huang and L. Ruopeng, Process Biochem., 39,
was tested using standard deviation. As a result, the experimental 1427 (2004).
findings were in close agreement with the prediction of the model. 11. Q. Liu, Y. Zhou, L. Chen and X. Zheng, Desalination, 250, 605
(2010).
CONCLUSION 12. J. T. Sousa and E. Foresti, Wat. Sci. Technol., 33(3), 73 (1996).
13. Greentech, Co. Ltd., Treatment of Hospital Wastewater Using Acti-
A continuous feeding up-flow bioreactor with an intermittent vated Sludge Combined With Biological Contactor, International
regime in aeration and discharge has been successfully designed, Sciences, 1(4), 259 (2005).
fabricated and operated for simultaneous removal of carbon, nitro- 14. F. Ghorbani, H. Younesi, S. M. Ghasempouri, A. A. Zinatizadeh,
gen and phosphorus (CNP) from HWW. It was found that, the bio- M. Amini and A. Daneshi, Chem. Eng. J., 145, 267 (2008).
reactors could achieve high CNP removal efficiency in a short dura- 15. M. von Sperling, V. H. Freire and C. A. L. Chernicharo,Wat. Sci.
tion. The RSM results demonstrated the effects of the operating Technol., 43(11), 323 (2001).
variables as well as their interactive effects on the responses. Experi- 16. R. L. Mason, R. F. Gunst and J. L. Hess, Statistical Design and Anal-
mental findings were in close agreement with the prediction of the ysis of Experiments, eighth applications to engineering and science,
model. Besides, the intermittent aeration process was the key fac- 2nd Ed., Wiley, New York (2003).
tor in achieving biological nutrients removal in a single bioreactor. 17. F. Shahrezaei, Y. Mansouri, A. A. L. Zinatizadeh and A. Akhbari,
This system could achieve high process performance in terms of Powder Technol., 221, 203 (2012).
COD (95%), BOD (98.3%), TN (88%), and TP (92%) removal with 18. A. I. Khuri and J. A. Cornell, Response surfaces: design and analy-
a shorter HRT. ses, Marcel Dekker, New York (1996).
19. A. L Ahmad and S. Bhatia, Environ. Sci. Technol., 39(8), 2828 (2005).
ACKNOWLEDGEMENT 20. N. M. Majlesi and A. Yazdanbakhsh, J. Environ. Health Sci. Eng.,
5(3), 211 (2008).
This letter resulted from Mitra Mohamadi’s thesis, major Envi- 21. A. R. Dincern, N. Karakaya, E. Gunes and Y. Gunes, Global Nest.,
ronmental Health Engineering, Kermanshah University of Medi- 10(1), 31 (2008).
cal Science, Kermanshah, Iran. 22. A. Mokhtari Azar, A. Ghadirpour Jelogir, G. NabiBidhendi, N.
Mehrdadi, N. Zaredar and M. Khalilzadeh Poshtegal, J. Agric. Food
REFERENCES Chem., 8(2), 1199 (2010).
23. A. M. Mansouri, Ali Akbar L. Zinatizadeh, M. Irandoust and A.
1. E. Emmanuel, Y. Perrodin, G. Keck and J. M. Blanchard, J. Haz- Akhbari, Korean J. Chem. Eng., 31(1), 88 (2014).
ard. Mater. A, 117, 1 (2005). 24. M. Abdulgader, Q. J. Yu, A. A. Zinatizadeh and P. Williams, Asia-
2. M. Carballa, F. Omil, J. M. Lema, M. Lompart, C. García-Jares, I. Pac. J. Chem. Eng., 4, 698 (2009).
Rodríguez, M. Gómez and T. Ternes, Water Res., 38, 2918 (2004). 25. A. A. L. Zinatizadeh, Y. Mansouri, A. akhbari and S. Pashaei, Chem.
3. C. Boillot, Sci. Total Environ., 403, 113 (2008). Ind. Chem. Eng. Q., 17(4), 485 (2011).
4. K. Kummerer and E. Helmers, Sci. Total Environ., 193, 179 (2006). 26. F. Kargi and I. Konya, J. Environ. Manag., 84, 20 (2007).

July, 2015
 Process modeling and optimization of biological removal of CNP from hospital wastewater in a CFID bioreactor 1353

27. F. Meng, S. R. Chae, A. Drews, M. Kraume, H. S. Shin and F. Yang, Kazmi, Bioprocess Biosyst. Eng., 36, 627 (2013).
Water Res., 43, 1489 (2009). 32. Y. L. Wang, S. L Yu, W. X. Shi, R. L. Baoa, Q. Zhao and X. T. Zuo,
28. Z. Fu, F. Yang, Y. An and Y. Xue, Biochem. Eng., 43, 191 (2009). Bioresour. Technol., 100, 3877 (2009).
29. E. Pehlivanoglu-Mantas and D. L. Sedlak, Crit. Rev. Environ. Sci. 33. A. Zafarzadeh, B. Bina, M. Nikaeen, H. Movahedian Attar and M.
Technol., 36, 261 (2006). Haji Khiadani, Iran J. Biotechnol., 9, 157 (2011).
30. C. Choi, J. Lee, K. Lee and M. Kim, Bioresour. Technol., 99, 5397 34. W. Janczukowicz, M. Szewczyk, M. Krzemieniewski and J. Pesta, J.
(2008). Pol. Environ. Stud., 10, 15 (2001).
31. A. A. Khan, R. Z. Gaur, V. Diamantis, B. Lew, I. Mehrotra and A. A.

Korean J. Chem. Eng.(Vol. 32, No. 7)

View publication stats