ADVANCED CONTROL OF AEROBIC INDUSTRIAL
WASTEWATER TREATMENT
Matei Vinatoru, Eugen Iancu, Gabriela Canureci and Camelia Maican
University of Craiova, Automation and Mechatronics Department
Keywords: Biological wastewater treatments, control systems, state estimators.
Abstract: The paper present the possibility of automatic control of the biological wastewater treatment station with
applications in Romanian Chemical Companies. In this paper are developed a mathematical model for
biological aeration basins and two automatic control systems (conventional control structure using three-
positional controllers or PLC and advanced control structure using state estimators) for wastewater
industrial purification stations.
1 INTRODUCTION
In the present world, environmental issues are a very
important topic. More and more countries and
international bodies are issuing stringent laws and
standards for environment protection. A major field
in environment protection is the industrial
wastewater treatment, geared toward protecting
world waters from pollution. Biological processes
are the ones most used in wastewater treatment
today. (Chen 2001, Peter, 2003). These processes,
used to remove both inorganic and organic products,
take place in wastewater treatment plants. The
wastewater is treated using complex chemical and
biological reactions, before being discharged in the
environment. The schematic operational block
diagram of such treatment plant is presented in
figure 1. The residual wastewater discharged by
industrial plants contains a lot of contamination
substances (organic and inorganic matters,
ammonium and nitric compounds), which shall be
eliminated before the water is discharged in
environment. Different treatment techniques are
used to eliminate those substances, as
physical/chemical treatment techniques and
treatment by microorganisms called biological
aerobe purification.
In the chemical treatment, certain chemicals are
added to the wastewater. These chemicals are
interacting with the contamination substances,
changing there structure and allowing their
elimination through mechanical processes (screen,
grit, filtration). In the same time, the pH of the
solution is brought to the neutral point. Most of
fertilizers such as nitrates can be removed this way.
Biological treatment processes are used to remove
the dissolved organic load from the water using
microorganisms. They use aerobic bacteria for the
decay of the organic matter. Aerobic bacteria must
be present, in order to perform the chemical
conversion of biological contaminants in other
substances that can be easily eliminated trough
simple mechanical processes.
Figure 1: Wastewater industrial purification.
Residual
water
Screen
Primary sludge
Grit
Returned activated
sludge
Excess
sludge to
treatment
Biological aeration basins
Treated
water
Primary
sludge
Rotating disk
Sands, Oils Removed solids
Activated sludge
165
Vinatoru M., Iancu E., Canureci G. and Maican C. (2007).
ADVANCED CONTROL OF AEROBIC INDUSTRIAL WASTEWATER TREATMENT.
In Proceedings of the Fourth International Conference on Informatics in Control, Automation and Robotics, pages 165-170
DOI: 10.5220/0001627401650170
Copyright
c
SciTePress
Figure 2: Aeration basins with activated sludge.
The contaminants are converted to carbon dioxide
and water as a result of the following reaction.
[]
.
Bacteria
moore
OHCOBacteriaO
load
Organic
222
++++
(1)
In this case the Biological Oxygen Demand (BOD)
defines the organic load. The biological treatment
processes take place in biological aeration basins.
Turbine aerators are used to aerate the water in order
to maintain the optimal oxygen concentration for
bacteria and also mix the water, keeping the media
homogenous. The industrial wastewater and primary
sludge (with bacteria) are mixed inside the aeration
basins, and bacteria consume the organic matter
resulting new bacteria, called activated sludge. The
activated sludge exists normally in the form of
flakes, which, besides live and dead biomass,
contains absorbed and stored, both organic and
mineral parts.
In order to control the content of the biomass, it
is necessary to control the oxygen dissolved in the
basin water, which can be achieved through the
control of the turbine aerators (level, flow, speed, or
number of running aerators). The turbine aerators
inside the aeration basins can be considered as a
distributed structure (see figure 2), and considering
bacteria’s growing and activity requirements, it is
necessary to design an advanced control system to
assure a high efficiency of the process (water
quality, cost reduction). This paper presents an
automatic control system for the biological
wastewater treatment in Romanian treatment
stations. The schematic diagram of the aeration
basins is presented in figure 2.
It is necessary to control the aeration process of
the wastewater in the biological treatment basins, in
order to obtain the optimal conditions for aerobic
bacteria growth and evolution (Peter, 2003). The
problem of oxygen concentration measurement in
solutions was already solved both in the country and
abroad (Vinatoru,1979, Vanrolleghem, 2003).
The aeration process can be controlled as
follows:
- through the control of the turbine immersion
relative to the water level in the basin;
- through the control of transit time of
wastewater in the aeration basin through the rising
or lowering of the control dam;
- through the number of running aerators in each
basin, which is the most efficient method for oxygen
concentration control;
- through variation of the rotational speed of the
aerators; this method cannot be applied in Romanian
installations, since the rotational speed cannot be
modified for the motors used.
2 THE MATHEMATICAL MODEL
OF THE AERATION BASINS
Considering the oxygen concentration control
possibilities, the aeration basins can be considered as
an oriented object with three inputs (w-rotational
speed of the motors driving the aerators, n-number
of aerators running and h-level of water in the basin
relative to the lowest position of the aerators) and
one output (c-oxygen concentration in the basin).
The oxygen concentration is a function of both time
and spatial coordinate along the water path from the
entrance in basin 6 to the exit over the dam in basin
number 1.
We will consider one basin and the mathematical
model for the dynamic regime is given by the
following equations:
-The equation for the water flow over the dam
in the aeration basin 1:
2/3
a
)lH.(g2.b.F α=
where α=(0,5 – 0,6)
-The equation for the aeration pump flow:
2/32
sp
h.)n/n.(.432,738F γ=
Residual water
Returned sludge
15
16
18
17 14
13
4
6
5
9
7
8
10
12
11
3
1
2
Aeration basins with activated sludge
Control
dam
To rotating dis
k
Turbine aerators 181
ICINCO 2007 - International Conference on Informatics in Control, Automation and Robotics
166
Figure 3: Dam measures.
-The quantity of oxygen used by microorganisms
in the time unit and volume unit:
)CC(k
dt
dC
sau
=
(2)
-The global mass balance equation can be
considered stationary since the water volume V
variations can be neglected inside the basins:
F
i1
+ F
i2
+ F
i3
- F
e
= 0 (3)
-The mass balance equation for the dissolved
oxygen:
)CC(VkC.F)CC(F.H
C.FC.FC.F
dt
dC
V
isauieirpi
3i3i2i2i1i1i
i
+
+++=
(4)
The liquid volume in the basin is given by the
formula:
V= A ( H
l
+ h )
For a particular case of treatment basins, the general
mathematical model described by equations 1 to 6
can be linearized around the stationary values and
simplified, resulting the schematic block diagram
presented in figure 4.
Figure 4: Block diagram of one basin.
1 to 6 can be linearized around the stationary values
and simplified, resulting the schematic block
diagram presented in figure 4.
The transfer functions are the following form:
i0i1
2
i2
i0i1
Hd1
asasa
bsb
)s(H
++
+
=
(5)
i0i1
2
i2
i2i1
i2
asasa
bsb
)s(H
++
+
=
(6)
i0i1
2
i2
i3i1
Ni1
asasa
bsb
)s(H
++
+
=
(7)
The interconnection between basins 1 to 6 is
done through the transfer flow between basins:
5,...2,1i,F.k)HH(g2A.F
pipi1iiiiei
=+μ=
where: ,D/L5,1/1
iiii
λ+=μ
(
)
,HH/(2186,0
3/1
1iii
++=λ
iiipiiiii
HSVkHHDA .),3...1(),(5,0
1
==
+
=
and the return flow is: F
ri
= N
i
*F
pi
Combining all 5 active basins, we obtain the
block diagram in figure 5.
3 AUTOMATIC CONTROL OF
THE AERATION BASINS
In order to determine the optimal structure of the
automatic control system for the oxygen
concentration in the biological treatment basins, the
mathematical model of the basins was developed,
using the technological diagram (figure 1) and the
available controls for the oxygen concentration. We
tried to get a better approximation of the real process
than the one used in (Vinatoru, 1979); therefore it
considered the influences at the border between two
basins.
Analysing the existing conditions in the aeration
basins, we can divide the installation in two big
sectors:
- in the first sector, containing basins 1,2 and 3,
the oxygen concentration control is done through the
number of running aerators N
1i
and through the
height of the dam H
d
(figure 4);
- in the second sector, containing basins 4 and 5,
the oxygen concentration control is done through the
number of running aerators N
2i
.
The high cost of oxygen sensors and especially
the high maintenance cost call for reduction of the
number of sensors used. From the analysis of the
biological water treatment basins at DOLJCHIM SA
Craiova, where the experiments were also made, we
determined that a minimum of two sensors are
necessary, one being mounted at the exit (measuring
concentration C
1
) and one in basin 4 (measuring
concentration C
4
). These measurements will be used
as output variables for the controlled process. The
right control strategy and structure to be used
depends of the financial capability of the company.
We studied three different control structures that can
be used for similar basins.
ΔC
i+1
Δ N
1i
Δh
Δ
C
i
H
2i
(s)
H
1Ni
(s)
H
1Hd
(s)
H
d
H
1D
(s)
d
H
h
ADVANCED CONTROL OF AEROBIC INDUSTRIAL WASTEWATER TREATMENT
167
Figure 5: Conventional control diagram.
- Conventional control structure using tri-
positional controllers, figure 5, which control the
starting and stopping of the aeration turbines
depending of the domain in which the oxygen
concentrations in basins 1 and 4 are located,
according with the algorithm presented in Table 1.
This solution was implemented for the biological
treatment basins. The control devices for the number
of aerators and the actuator for the dam can be easily
implemented and their control by the tri-positional
controller is done based on the limits C
min
and C
max
according with table 1, imposed by the particular
conditions inside the aeration basins.
Table 1: Control algorithm.
Control aerator basins 1,2 and 3
Concentration C
1
0 C
min
C
max
Aerator running All 1,3,5,7,9 3,5,7
Dam Upper
position
Upper
position
Lower
position
Control aerator basins 4 and 5
Concentration C
4
0 C
min
C
max
Aerator running All 11,13,15 11,15
From equations 1 to 6, considering the particular
conditions for the aeration basins, it results that the
entire process has a very slow dynamic regime, due
to the high volume of wastewater compared with the
transit time of the water in the basins. Therefore is
not necessary a continuous control of the oxygen
concentration. Moreover, the bacterial activity does
not require an exact oxygen concentration but
certain limits between which the activity is running
normally.
- Fuzzy control structure, where the fuzzifier
and the defuzzifier are following the control rules
presented in table 1.
- Control structure using state estimators
according with the diagram presented in figure 6.
For this structure, the current state in each basin
X1 = C
1
, X
2
= C
2
, X
3
= C
3
, X
4
= C
4
, X
5
= C
5
, is
estimated based on the measured output values Y
1
=
C
1
, Y
2
= C
4
.
Using the state variables and applying the
command synthesis principles developed by the
authors for the control of distributed parameter
systems (Vinatoru 1979), the command for the
aeration turbines will be generated by the Command
Synthesis block in figure 6.
4 CASE STUDY
4.1 Control with Two Tri-positional
Controllers
To show the performance and experimental results
behaviour of the control structures some
experiments have been carried out. The conventional
control structure using tri-positional controllers has
implemented to the DOLJ Chim SA Wastewater
Treatment Plant Craiova.
11
C
1
H
21
H
1N1
H
1Hd
H
d
25
13
12
C
2
H
32
H
1N2
C
3
H
43
H
1N3
C
4
H
54
H
1N4
C
i
F
i
C
5
H
i5
H
1N5
H
1Fi
Command Synthesis
24
Three-State Controller
2 0
1
1
0
X
1
*
X
4
*
Output Set-point Block Operator
Y
1
C
i
F
i
Y
2
Turbine Command Synthesis
2
Three-State Controller
Aeration basins
5 3 2 14
ICINCO 2007 - International Conference on Informatics in Control, Automation and Robotics
168
Figure 6: Advanced control diagram.
The experimental results in the various
conditions of residual water flow are presented in
figure 7. Aeration basins volume: 54.000 m
3
(9000
m
3
/basin); Internal recycle flow rate 2700 m
3
/hour;
Dissolved Oxygen concentration (DO): 6 mg/l basin
no. 1 and 4 mg/l in basin no. 4.
4.2 Control using a Sate Estimator
To sow the performance behaviour of advanced
controller structure, the proposed control strategy
has been implemented to the model of Wastewater
purification Plant, presented in figure 6. The transfer
functions from this structure are:
544332
2
21
173.971397
)14.80(66.0
)( HHH
ss
s
sH ===
++
+
=
5...1,
172.961344
)112.91(8.2
)(
2
1
=
++
+
= i
ss
s
sH
Ni
173.971397
)14.80(78.0
)(
2
5
++
+
=
ss
s
sH
i
According with the diagram presented in fig. 6
the state estimator has implemented in the form:
)t(wC)t(x)q(A
d
t
)t(x
ˆ
d
00
+=
(8)
where q is elements of the unknown vector
considered the tuning parameters of the observer.
The input w(t) is:
)t(y)t(y
ˆ
(Qg)t(u.c)t(w
m
T
+=
(9)
where g
T
is the weighting functions and QR
2
is
introduced for a better tuning of the observer
parameter in function of the difference between the
estimated output
)(
ˆ
ty and the real output y
m
(t)
(y
m1
=C
1
, y
m2
=C
2
). We have implemented the
command synthesis in function of each estimated
steady state variables
x
i
=C
i
(I=1…5), controlled by
inputs N
ij
(see diagram from fig. 6).
The results are presented in figure 7.
Figure 7: Experimental results I.
0 50 100 150 200 250 300
0
1
2
3
4
5
6
7
DO in basin 1
DO in basin 4
Controller outpu
t
Time [hours]
DO level [mg O
2
/l]
11
C
1
H
21
H
1N1
H
1Hd
H
d
25
13
12
C
2
H
32
H
1N2
C
3
H
43
H
1N3
C
4
H
54
H
1N4
C
i
F
i
C
5
H
i5
H
1N5
H
5Fi
Command Synthesis
24
State Estimator
X
1
X
3
X
4
X
2
X
5
X
1
X
3
X
4
X
2
X
5
*
State Set-point Block
Y
1
C
i
F
i
Y
2
ADVANCED CONTROL OF AEROBIC INDUSTRIAL WASTEWATER TREATMENT
169
Figure 8: Experimental results II.
5 CONCLUSIONS
The results obtained using the proposed control
structures to a Wastewater purification Plant are
satisfactory. It causes a better performance of the
plant because environmental law nearer to those
requires the level of purification obtained. Also, the
running costs have a notable reduction. The
conventional tri-positional control structure is in
implementation phase and we study the possibilities
for advanced control structure. The results obtained
till now establish the steps towards this objective.
REFERENCES
Chen W. C, Chang Ni-Bin, Shieh Wen K, Advanced
hybrid fuzzy-neural controller for industrial
wastewater treatment, Journal of environmental
engineering , ISSN 0733-9372 CODEN JOEEDU ,
2001, vol. 127, n
o
11, pp. 1048-1059
Chang Ni-Bin, Chen W. C, Shieh Wen K, Optimal
control of wastewater treatment plants via integrated
neural network and genetic algorithms, Civil
engineering and environmental systems (Civ. eng.
environ. syst.) ISSN 1028-6608,
2001, vol. 18, n
o
1, pp. 1-17
Demey D, Vanderhaegen V, Vanhooren H, Liessens J,
Van Eyck L, Vanrolleghem P. A, Hopkins L.,
Validation and implementation of model based control
strategies at an industrial wastewater treatment plant,
Water science and technology (Water sci. technol.)
ISSN 0273-1223, CODEN WSTED4 , 20011981,
vol. 44, n
o
2-3, pp. 145-153,
Henze M., Grady C.P.L., Gujer W., Marais G.V.R.,
(1987),
A general Model for Single-sludge Wastwater
Treatment Systems,
Water Resourses, Vol. 20, pp.
505-515.
Ng Wun Jern, (2006), Industrial wastewater treatment,
National University of Singapore , World Scientific
Publishing, ISBN 978-1-86094-580-9
Peter A. Vanrolleghem, Models in Advanced Wastewater
Treatment Plant Control (2003), Water Res. 2003
Jan;37(1):95-107, ELSEVIER
Urrutikoetxea A., Garcia de las Heras J.L., (1994),
Secondary Settling in Activated Sludge. Alab-scale
Dynamic Model of Thickening,
Proceeding of the 8
th
Forum for Applied Biotechnology, Gent Coupure
Links 653, B-9000, Gent Belgium.
Vinatoru M. , Niculescu E. ,Mihaiu M., (1979), The
automatic compensation with the temperature of the
oxyfen sensore –„3rd International Symposium on
Control System and Computer Sciens – Bucureşti , ,
vol.III.p.654-660
Yusof R., Omatu S., (1993),
A multivariable self-tuning
PID Controller,
Int. J. Control, Vol.57, No. 6, pp.
1387-1403.
APPENDIX
Notations and Symbols
V
i
– volume of basin i (i=1-5) [m
3
], F
ei
- output flow from
basin i ( F
6
=0) [m
3
/h], F
ia
– input wastewater flow in basin
i [m
3
/h] (F
la
=0), F
in
– input flow of activated sludge in
basin i [m
3
/h] (F
ln
=F
2n
=F
3n
=F
nn
=0), C
i
– oxygen
concentration in basin i [mgO
2
/l], C
in
- oxygen
concentration in flow F
in,
C
ia
– oxygen concentration in
flow F
ia
, C
a
– oxygen concentration at working
temperature, K
Au
– transfer coefficient in wastewater
F
ri
– recirculation flow of aeration pumps [m
3
/h]
C
ri
– oxygen concentration in flow F
ri
γ
-specific gravity of the medium inside the basin [N/m
3
]
n - nominal rotational speed of the pumps, n
a
– specific
rotational speed of the pumps, h
i
– immersion depth of the
aerator [m], F
pi
– pump flow [m
3
/s], H
i
–water level in the
basin i [m], L
i
–width of the separation wall [m], D
i
width of the transfer section between two basins, K
pi
ratio coefficient between pump flow F
pi
and exit flow
S
i
–area of horizontal section through the basin [m
2
], N
i
umber of running pumps in basin i,
m
c
–shape
coefficient, depending of the geometric shape of the
dam, b –dam width, H –the liquid level above the
dam, M
l
–coefficient of velocity, depending of the
access speed upstream of the dam, l
d
–dam height
[m].
0 50 100 150 200 250 300 350 400 450 500
0
1
2
3
4
5
6
7
8
DO [m
g
O
2
/l]
Time [hours]
ICINCO 2007 - International Conference on Informatics in Control, Automation and Robotics
170