Icecs 2002 Cd-Rom Proceedings: Welcome To The
Icecs 2002 Cd-Rom Proceedings: Welcome To The
Icecs 2002 Cd-Rom Proceedings: Welcome To The
1
Osijek Water Authority, Osijek, Croatia, Boris.Azenic@vodovod.com,
University of Zagreb, Faculty of Electrical Engineering and Computing, Nedjeljko.Peric@fer.hr,
3
University of Osijek, Faculty of Electrical Engineering, Drazen.Sliskovic@etfos.hr.
ABSTRACT
The paper deals with the possibility of applying predictive
control for water level control in tanks of water supply plants.
The proposed control system includes ground water intake
plant, water treatment plant and water tanks. Since predictive
controlers are based on mathematical models of controlled
processes, the mathematical model of water supply plant has
been created for the purpose of design and verification of this
control algorithm. The price of electrical energy during the day
has also been taken into consideration for controller design. By
doing that, in addition to keeping the water level within
regulated limits, reduction of the costs of energy consumption
has been achieved. Properties of the designed control system
have been investigated by simulation on the mathematical
model of water supply plant.
Keywords: pipelines, water supply plant, mathematical model,
predictive control, level control, reduction of pumping costs.
1 INTRODUCTION
Water systems are complex systems that are very
important for our civilisation. Development of water supply
systems and the requirements for higher quality of water, lower
price and safer system operation result in the need for
systematic approach to water supply system control. On the
other hand, by applying information technology in water supply
systems in recent years, plant automation level has been
achieved that enables applications of advanced control
algorithms, which can include various intelligent control
algorithms. Application of advanced control algorithms makes
it possible to meet the requirements with optimal water supply
even if water resources, capacity of water treatment plant, and
cubic capacity of water tank are limited. The process in water
supply systems is affected by various disturbances the biggest
of which is water consumption in distribution network. That is
non-stationary and stochastic disturbance what additionally
makes it more difficult to design control system.
The analyses in the paper refer to water supply plant of the
Osijek Water Authority (seated in the city of Osijek, Croatia)
which consists of eighteen water wells, two water treatment
lines and five water tanks with total capacity of 10,000 m3.
Water production process in this plant is still being operated
manually by a team of operators.
2 MODELLING OF WATER SUPPLY PLANT
The mathematical model of a water supply plant is
structured according to its topology. Water supply plant consists
of three subsystems: water pumping and transport plant (ground
water intake plant), water treatment plant and water
accumulation plant, i.e. water tanks. On the other hand,
regardless of one of the mentioned subsystems or the entire
p j = R j q1j.852 ,
Rj =
1.07 10 5 l j
C 1j.852 d 4j .89
(1)
where:
( )
1
p j ,
Kj
Kj =
4 l j
d j
(2)
where:
Kj - water inertia coefficient [kg /m4],
- fluid density [kg/m3].
Non-linear differential equation that describes dynamical
behaviour of the fluid in non-elastic pipelines is [4]:
dq j
dt
1
Kj
p j R j q j q j
0.852
(3)
653
q
q
p = g (q, u , s ) = a + b
+ c
n
n 0
0
if n 0 and
(4)
0,
where:
0
q
n
7.0
h [m]
6.5
q [m3/s]
0 .4 2 0
6.0
5.5
0
c)
d)
300
tim e [s ]
m o d e l re s p o n s e
s y s te m re s p o n s e
p [Pa]
8 .0 E + 0 4
7 .5 E + 0 4
7 .0 E + 0 4
6 .5 E + 0 4
-1 0 0
100
200
300
tim e [s ]
10 12 14
time [h]
16
18
20
22
24
0 .3 8 0
200
a)
b)
100
m o d e l re s p o n s e
s y s te m re s p o n s e
0 .4 0 0
0 .3 6 0
-1 0 0
model response
system response
654
u
2
2
2
J (href , h, q ) = href (k + j ) h(k + j ) + [q (k + j 1)] ,
j = N1
j =1
Gip ( s ) =
1 e 20 s
,
1 + 1000s
1
G wt ( s ) =
,
3127 s
(7)
(8)
transfer function,
Gtp (s ) - water treatment plant transfer function,
q (k )
(6)
where:
1 e 5 s
,
1 + 150s
Gtp ( s ) =
(5)
where:
href (k ) - reference water level trajectory,
h(k )
655
where:
(9)
qmax
qmin
h [m]
6 .5
6
w a te r le v e l
5 .5
4 CONCLUSION
Based on the conducted analysis it can be concluded that
the designed GPC is able to meet the requirements of a
control system. The simulations show that the results
achieved after implementing GPC are better than those
achieved by manual control. When designing the control
system, beside adequate choice of GPC parameter values, it is
necessary also pay attention to water consumption prediction
calculation. Control quality depends considerable on the
accuracy of consumption prediction calculation.
In the next phase of our investigation we are going to
implement the control system based on GPC approach in the
real water supply plant.
22:00
20:00
18:00
16:00
14:00
12:00
10:00
08:00
04:00
02:00
00:00
06:00
r e fe r e n t w a te r le v e l
tr a je c to r y
5 REFERENCES
tim e [h ]
0 ,5 5
q[m3/s]
0 ,5
0 ,4 5
0 ,4
22:00
20:00
18:00
16:00
14:00
12:00
10:00
8:00
6:00
4:00
2:00
0:00
0 ,3 5
tim e [h ]
0 .4
0 .3
22:00
20:00
18:00
16:00
14:00
10:00
08:00
06:00
04:00
02:00
12:00
w a te r c o n s u m p tio n
w a te r c o n s u m p tio n p r e d ic tio n
0 .2
00:00
q [m3/s]
0 .5
tim e [h ]
656