INTERNATIONAL JOURNAL OF MECHANICS

Mathematical Model of Control Valves of

Three-Tank System
Petr Chalupa, Jakub Novák, Vladimír Bobál

 physical laws [4]. But there are usually a lot of unknown

Abstract—The paper is focused on development of constants and relations when performing analysis of a plant.
a mathematical model of valves of a hydraulic system. Laboratory The paper uses combination of both methods. Basic
three tank model (Amira DTS200) was investigated and relations are derived using mathematical physical analysis.
characteristics of its valves were measured. Consequently, the Values of model parameters are identified on the basis of real-
process of creating a mathematical model of the valves is described
in detail. The three tank system is a classical modeling task but this
time measurements. The goal of the work was to obtain a
paper focuses on nonlinearities which are present in real system and mathematical model of the valves of DTS200 Three-Tank
other differences between ideal mathematical model and real-time System [5] and to design the models in MATLAB-Simulink
system. Especially hysteresis represents a big difference between environment. The DTS200 laboratory equipment was
ideal mathematical model and real-time system. Even thought all developed by Amira Gmbh, Duisburg, Germany and serves as
valves the system is equipped with are of the same type, big a real-time model of different industrial systems concerning
differences were observed between their characteristics. The
approach to modeling of the system is not restricted to the particular
liquid transport.
system but can be used for many real-time hydraulic systems. The models of valves serve as a part in process of creating a
model of whole DTS200 system. The major reason for
Keywords—Control valve, Gray box, Hysteresis, Identification, creating the model of this laboratory equipment are big time
Modeling. constants of the plant and thus time consuming experiments. A
model, which represents the plant well, can considerably
I. INTRODUCTION reduce testing time of different control approaches. Then only
promising control strategies are applied to the real plant and
A LMOST all current control algorithms are based on a
model of a controlled plant [1]. Some information about
controlled plant is necessary for design of a controller with
verified.
A tank system with valves occurs often in industrial practice
satisfactory performance. A plant model can be also used to and was investigated by many researchers [6], [7]. Flow of
investigate properties and behavior of the modeled plant liquid through pipes is studied in [8].
without a risk of damage of violating technological constraints The paper is organized as follows. Section 2 presents the
of the real plant. There are two basic approaches of obtaining modeled system – Amira DTS200. Derivation of initial ideal
plant model: the black box approach and the first principles using first principles modeling is carried out in Section 3.
modeling. Section 4 and 5 presents characteristics and calibration water
The black box approach [2], [3] is based on analysis of level sensors and pumps respectively. Section 6 consists or
input and output signals of the plant. Usage the same results of measurements of valves.
identification algorithm for wide set of different controlled
plants is the main advantages of this approach. The knowledge II. THE DTS200 SYSTEM
of physical principle of controlled plant and solution of set of The photo of main part of Amira DTS200 system is shown
mathematical equation is not required. Main drawback of a in Fig. 1.
black box model persists in fact that it is generally valid only
for signals it was calculated from.
The first principle modeling provides general models valid
for wider range of plant inputs and states. The model is
created by analyzing the modeled plant and combining

Manuscript received June 29, 2010: This work was supported by the
Ministry of Education of the Czech Republic under grant 1M0567.
Petr Chalupa is with Tomas Bata University in Zlin, nám. T. G. Masaryka
5555, 760 01 Zlín, Czech Republic (phone: +420576035204; e-mail:
[email protected]).
Jabub Novák is with Tomas Bata University in Zlin, nám. T. G. Masaryka
5555, 760 01 Zlín, Czech Republic (e-mail: [email protected]).
Vladimír Bobál is with Tomas Bata University in Zlin, nám. T. G.
Masaryka 5555, 760 01 Zlín, Czech Republic (e-mail: [email protected]). Fig. 1. Amira DTS200 – three tank system

Issue 4, Volume 5, 2011 310

 INTERNATIONAL JOURNAL OF MECHANICS

III. INITIAL IDEAL MODEL

The system consists of three interconnected cylindrical This chapter is focused to derivation of mathematical model
tanks, two pumps, six valves, pipes, water reservoir in the of a valve. This derivation is based on ideal properties of
bottom, measurement of liquid levels and other elements. individual components.
Both pumps pump water from the bottom reservoir to the top The ideal flow of a liquid through a pipe can be derived
of the left and right tanks Valve positions are controlled and from Bernoulli and continuity equations for ideal liquid:
measured by electrical signals, which allow precision setting
of their position. 1 2
hg  v 
A simplified scheme of the system is shown in Fig. 2. The 2  q  Sv 2 g h (1)
pump P1 controls the inflow to tank T1 while the pump P2 q  SV v 
controls the liquid inflow to tank T2. There is no pump
connected to the middle tank Ts. The characteristic of the flow where ∆h is a difference between liquid levels on both sides of
between tank T1 and tank Ts can be affected by valve V1, the pipe (e.g. difference between levels of tanks that are
flow between tanks Ts and T2 can be affected by the valve V2 interconnected by the pipe), g is the standard gravity, v is the
and the outflow of the tank T2 can be affected by valve V3. liquid velocity and SV is the flow space of the pipe. The flow
The system also provides the capability of simulating leakage space SV is controlled by the valve position p.
from individual tanks by opening the valves V4, V5 and V6.
SV  p  SV max 0  v 1 (2)

where SVmax is the maximal flow area of the valve.

P1 P2 Since the flow through a valve depends only on the level
difference, the valve position and constants representing pipes
T1 Ts T2 and cylindrical tanks, the change of water level in tank T1 can
be written as follows:
dh1
h1  k1 h1  hs  sign  h1  hs   k4 h1 (3)
V1 hs V2 h2 V3 dt
The area of all three tanks is the same and is symbolized by
ST. The k is a parameter representing valve position
leakage (V4) leakage (V5) leakage (V6)
SV max 2 g
ki  pi i  1, 2,...,6 (4)
Fig. 2. Scheme of three tank system Amira DTS200 ST

Pumps are controlled by analogue signal in range from - and q represents inflow as change of water level in time:
10V to 10V. Heights of water level are measured by pressure
qi
sensors. Each valve is operated by two digital signals which qi  i  1, 2 (5)
control motor of particular valve. First signal orders to start ST
closing of the valve while the second signal is used for Similar equations can be derived for the other two tanks.
opening of the valve. If none of the signals is activated the The model obtained by using ideal properties and behaviour of
valve remains in its current position. Each valve also provides plant parts if further referred as “ideal model”. This model of
three output signals: analogue voltage signal correspond to the whole three tank system is successfully used in many control
current position of the valve and two informative logical system studies as a demonstration example [9], [10], [11].
signals which states that the valve is fully opened or fully
closed respectively. IV. CHARACTERISTICS OF THE VALVES
The overall number of inputs to the modeled plant DTS200
is 14: As stated in Section 2, each of plant’s 6 valves is driven by
• 2 analogues signals controlling the pumps, two dedicated logical signals. These signals are used for
• 12 digital signals (2 for each of the 6 valves) for opening starting valve’s motor in closing or opening direction
/ closing of the valves. respectively. If none signal is activated the valve remains in its
The plant provides 21 measurable outputs which can be current position. Activation of both signals at in a particular
used as a control feedback or for measurements of plant time represents an invalid state and valve motor is stopped.
characteristics: Each valve provides three output signals. The current valve
• 3 analogue signals representing level heights in the three position is determined by analogue signal. Higher values of
tanks, signal represent closed valve and lower values represent
• 6 analogues signals representing position of the valves, opened valve. The other two signals are logical and state that
• 12 logical signals (2 for each of the 6 valves) stating that valve is opened or closed respectively.
corresponding valve is fully opened / closed.

Issue 4, Volume 5, 2011 311

 INTERNATIONAL JOURNAL OF MECHANICS

A. Valve limits and speed draining of a tank to the reservoir situated below the tanks is
Process of opening all valves at once from fully closed state described by differential equation based on (3):
dh  t 
to fully opened state is presented in Fig. 3.
 k h  t  (7)
dt
full close => full open

Integrating in an appropriate time range leads to the

0.4
equation of time course of water level:

k2 2
0.2 h t    t  k h 0  t  h 0 (8)
4
valve pos [MU]

0 where h(0) is initial water level. An example of draining is

presented in Fig. 4. At the beginning of the experiment, the
-0.2 tank was full and all valves were closed, then valve V4 was
partially opened, its position was recorded and time course of
valve 1
-0.4
valve 2
water level height was measured.
valve 3
-0.6 valve 4 draining through valve V 4
valve 5 700
valve 6
-0.8
0 1 2 3 4 5 6 7 8 9 10 600
time [s]

Fig. 3. Closing all valves in full range 500

This process represents moving of valve position in full

400
h1 [mm]

range of its hard constraints. The vertical lines in left part of

Fig. 3 represent changes the “opened” signals of individual
300
valves. Before these signals drop down the valves are said to
be opened. The vertical lines in the right part of Fig. 3
200
represent the changes of “closed”. From these lines onward,
the valves are said to be closed.
100
It can be observed that the initial and final positions of the
valve as well as the positions corresponding to changes of
0
“opened” and “closed” signal differ. But all the valves are 0 20 40 60 80 100 120 140 160 180 200
moving at almost the same speed vvalve. time [s]

Fig. 4. Draining of tank T1 through valve V4

vvalve  0.175 MU s (6)

Valve positions corresponding to hard constraints and It is obvious that parabola depicted in Fig. 4 would continue
validity of “opened” and “closed” state are summarized in below zero contrary to (8). A term corresponding to the
Table I. vertical length of outflow pipe h0 was added to the model. The
vertical length h0 is depicted in Fig. 5.
Table I
Valve positions for important states
tank
full “closed” signal “opened”
Valve full open
closed change signal change
no. [MU]
[MU] [MU] [MU]
1 0.5199 0.4066 -0.6065 -0.6945 valve
h
2 0.4550 0.3880 -0.5719 -0.7493
3 0.5462 0.3996 -0.5423 -0.6501
4 0.4926 0.3594 -0.5751 -0.7151
5 0.4868 0.3296 -0.5157 -0.6718 reservoir
6 0.5242 0.4416 -0.5698 -0.6525 h0

B. Valve flow parameter for outflow valves

Valve flow parameters ki as appear in (3) were computed
from measurements of draining through individual valves
which are connected to outflow pipes (V3, V4, V5 and V6). The Fig. 5. Vertical length of outflow pipe (h0)

Issue 4, Volume 5, 2011 312

 INTERNATIONAL JOURNAL OF MECHANICS

Due to mechanical configuration of the plant, the value of This relation for one set of experiment on valve V4 is
h0 for outflow valves V3, V4, V5, and V6 cannot be measured presented in Fig. 7 where circles represent individual
directly. But it can be identified from draining course (Fig. 4). experiments. The characteristic is not strictly linear. It
To encapsulate h0 into model, equations (7) and (8) were contains saturation of fully closed and fully opened valve.
superseded: Transitions to saturation states are smooth.
Experiments with the same configuration were performed
dh  t 
 k h  t   h0 for all outflow valves. The results obtained from measurement
dt (9) of valve V6 are presented in Fig. 8. The upper graph depicts
k2 the course of k6 and values of corresponding h0 are shown in
h  t    t 2  k h  0   h0  t  h  0  the lower graph.
4
valve 6 - opening
A second order polynomial (parabola) was fitted to an 0.3
appropriate interval of draining data in least mean squares
sense. The MATLAB function polyfit was used for this task. 0.2
Parabola fitting is presented in Fig. 6.

k6
0.1
parabola fitting of draining through valve V 4
h1=0.0086*t 2-4.95*t+622
700 0
-0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5
h1
valve position V 6 [MU]
600 parabola
200
500
0
400 h06
h1 [mm]

-200
300

200 -400
-0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5
valve position V 6 [MU]
100
Fig. 8. Relation between valve position and k6 and h06
0
The values of h06 for the four rightmost experiments (i.e.
-100 valve position greater than 0.2 MU) are irrelevant because the
0 20 40 60 80 100 120 140 160 180 200
time [s] valve was closed and no outflow was observed. The remaining
Fig. 6. Parabola fitting to the draining course experiments resulted in almost the same values of h06 which
was expected as the values of h0 are physical values of the
Values of k and h0 can be easily obtained from polynomial Amira DTS200 model.
coefficient according to (9). Valve can be closed to different C. Valve flow parameter for interconnection valves
positions at the beginning of draining experiment and relation
Similar approach to obtaining values of k as presented in
between valve position and value of k can be achieved.
previous subsection can be used also for valves V1 and V2
valve opening which interconnects tanks T1 and Ts, and Ts and T2,
0.3
respectively. Flow from the full tank T1 to the empty tank Ts
was used to measure valve constant k1. The other valves were
0.25 closed during the experiment. According to (3), the flow can
be described by two differential equations:
0.2
dh1  t 
 k1 h1  t   hs  t   sign  h1  t   hs  t  
dt
(10)
0.15
dhs  t 
k4

 k1 h1  t   hs  t   sign  h1  t   hs  t  
dt
0.1
Since the value of h1 is always higher or equal to hs, the
term inside absolute values is always nonnegative. As the
0.05
water flow just from T1 to Ts and the geometry of both tanks is
the same, according to mass conservation law the sum of h1
0 and hs remain the same during the experiment. Then the
-0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4
position of valve V 4 [MU] course of draining T1 and filling Ts can be described by two
Fig. 7. Relation between valve position and k4
independent differential equations.

Issue 4, Volume 5, 2011 313

 INTERNATIONAL JOURNAL OF MECHANICS

dh1  t 
 k1 2h1  t   h A similar approach as presented for valve V1 was used to
dt
measure characteristics of valve V2. Resulting course of k2 is
dhs  t  (11)
 k1 h  2hs  t  depicted in Fig. 11.
dt valve 2 - opening
h  h1  t   hs  t  0.3

Solving these equations lead to time course described by 0.25

second order polynomial.

k12 2
h1  t    t  k1 2h1  0   h  t  h1  0 
0.2

2 (12)
k2
hs  t    1  t 2  k1 h  2hs  0   t  hs  0 
0.15

k2
2
An example of courses and corresponding parabolas are 0.1

depicted in Fig. 9.
0.05
parabola fitting of flow through valve V 1
h1=0.0209*t 2 -4.95*t +570
hs =-0.0212*t 2 +5.01*t -20
600 0
-0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4
h1 valve position [MU]
500 hs
Fig. 11. Relation between valve position and k2
parabola h1
400 parabola hs
D. Valve hysteresis
300 The experiments presented in previous subsections were
h [mm]

preformed for opening of a valve only. At the beginning, the

200
valve was fully closed and subsequently was partially opened
to a given position. In this section a problem of closing of a
100
valve is studied. Performed experiments are similar except
initial part. The experiment starts with full tank and closed
0
valve too, but then the valve was fully opened and then
-100
partially closed to the desired position. Therefore the same
0 20 40 60 80 100 120 140 160 valve position (value of analogue signal from a valve) was
time [s]
reached but from opposite direction.
Fig. 9. Parabola fitting to the flow course through V1
valve 2
Relation between position of valve V1 in MU and k1 is 0.3
shown in Fig. 10. closing
openig
valve 1 - opening 0.25
0.3

0.2
0.25

0.15
k2

0.2

0.1
0.15
k1

0.05
0.1

0.05 0
-0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4
valve position [MU]

0 Fig. 12. Hysteresis of valve V2

-0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5
valve position [MU]
Experiments unveiled a hysteresis present in all valves. The
Fig. 10. Relation between valve position and k1 characteristics for opening and for closing of valve V2 is

Issue 4, Volume 5, 2011 314

 INTERNATIONAL JOURNAL OF MECHANICS

presented in Fig. 12. The figure shows that hysteresis plays a Similar experiments were performed also for outflow valves
big role in the experiments. The value of position itself does V3, V4 V5, and V6. Corresponding figures contain courses of
not give sufficient information about current value of parameter k as well as values of h0 obtained from individual
parameter k2. For example, if the position is 0 MU the value of experiments.
k2 can be anywhere in range 0.03 to 0.13. Especially in case of Four sets of experiments and corresponding characteristics
using the valve as an actuator the hysteresis should be taken of valve V3 is shown in Fig. 15.
into account. Otherwise control process can easily become valve 3
0.3
unstable.
closing 1
Consequently, extensive set of experiments was performed 0.25 closing 2
to measure hysteresis characteristics. The resulting courses of openig 1
0.2 openig 2
valve parameters are shown of following figures.
Three sets of experiments for opening and another three sets 0.15

k3
for closing of valve V1 were performed. Results shown in Fig. 0.1
13 provided consistent data and the hysteresis was lower
0.05
comparing to Fig. 12.
valve 1 0
0.3 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4
valve position [MU]
closing 1
closing 2
0.25 closing 3
200
openig 1
openig 2
0.2 openig 3 150

h03
0.15 100
k1

closing 1
50 closing 2
0.1
openig 1
openig 2
0
0.05 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4
valve position [MU]
Fig. 15. Characteristics of valve V3
0
-0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4
valve position [MU] Twelve sets of experiments and corresponding
Fig. 13. Characteristics of valve V1 characteristics of valve V4 is shown in Fig. 16.
closing 1
valve 4 closing 2
Four sets of experiments for closing of the valve performed 0.3
closing 3
in various time concerning valve V2 are depicted in Fig. 14. 0.25
closing 4
Individual experiments correspond to stars in the figure. Three closing 5
0.2 closing 6
set of experiment are presented for closing of the valve where opening 1
individual experiments are represented by circles. 0.15 opening 2
k4

opening 3
valve 2 opening 4
0.3 0.1
opening 5
closing 1 opening 6
0.05
closing 2
0.25 closing 3 0
closing 4 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4
openig 1 valve position [MU]

0.2 openig 2
closing 1
openig 3
200 closing 2
closing 3
0.15 closing 4
k2

150 closing 5
closing 6
opening 1
0.1
h04

100 opening 2
opening 3
opening 4
0.05 50 opening 5
opening 6

0 0
-0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4
valve position [MU] valve position [MU]
Fig. 16. Characteristics of valve V4
Fig. 14. Characteristics of valve V2

Issue 4, Volume 5, 2011 315

 INTERNATIONAL JOURNAL OF MECHANICS

Five sets of experiments and corresponding characteristics valve V2. Experiments performed on valves V3 and V6 resulted
of valve V5 is shown in Fig. 17. in more consistent values of h0 when comparing with
valve 5 experiments performed on V4 and V5.
0.3
Since the shape of curves corresponding to opening and
closing 1
0.25 closing 2 closing of the valve is similar, an average difference between
opening 1 them in direction of position axis can be computed. This value
0.2 opening 2
can be used as measure of hysteresis. Values of hysteresis, h0
opening 3
0.15 as well as maximal value of k for each valve is presented in
k5

0.1
Table II.

0.05
Table II
0 Valve characteristics
-0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4
Valve hysteresis
valve position [MU] maximal k h0 [mm]
no. [MU]
1 0.0219 0.2180 -
200
closing 1
2 0.1783 0.2237 -
closing 2 3 0.0310 0.2601 126.6
150 opening 1 4 0.0426 0.2976 127.9
opening 2
opening 3 5 0.1307 0.2735 121.2
h05

100
6 0.0800 0.2688 97.1

50
E. Modeling of valve characteristics
0
The course of relation between valve position in MU and k
-0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4
valve position [MU]
is similar to step responses of dynamical system and therefore
it was modeled in similar way. Other types of approximation
Fig. 17. Characteristics of valve V5 functions, like sigmoids, were also tested, but did not achieve
better results. A model based on transfer of 4 th order aperiodic
Five sets of experiments and corresponding characteristics system produced satisfactory results. Thus relation between
of valve V6 is shown in Fig. 18. position and k was as follows:
valve 6
0.3  1 b b3  3b2 a  6ba 2  6a3 
closing 1 pos  pos0 : k  kmax 1  e a 
0.25 closing 2
 6 a3 
opening 1
0.2 opening 2 b  pos0  pos (13)
opening 3
0.15 pos  pos0 : k 0
k6

0.1
where pos is valve position in MU and parameters a and pos0
0.05 were obtained by nonlinear regression.
Values of parameters kmax for each valve as well as pos0 and
0
-0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 a both for opening and closing are summed in Table III.
valve position [MU]

Table III
200 Valve characteristics
closing 1
closing 2
Valve opening closing
kmax
150 opening 1 no. pos0 a pos0 a
opening 2 1 0.2180 0.2673 0.0818 0.2892 0.0825
opening 3
h06

100 2 0.2237 0.1522 0.0802 0.3305 0.0807

3 0.2601 0.2689 0.0813 0.2999 0.0841
50 4 0.2976 0.2317 0.0880 0.2743 0.0930
5 0.2735 0.1716 0.0705 0.3024 0.0764
0 6 0.2688 0.2397 0.0811 0.3197 0.0856
-0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4
valve position [MU]
The value of kmax is affected not only by valve itself but also
Fig. 18. Characteristics of valve V6 by connection pipes. Therefore values of kmax are lower for
interconnection valves V1 and V2 comparing to outflow valves
The hysteresis of valves V1, V3 and V4 is significantly V3, V4, V5, and V6.
smaller than the hysteresis of valves V5, V6 and especially

Issue 4, Volume 5, 2011 316

 INTERNATIONAL JOURNAL OF MECHANICS

Values of pos0 correspond to position of the curve in the It can be stated that the parameter k does not change its
valve position axis (x-axis). Therefore greater difference value till it reaches a border of hysteresis area, i.e. the curve of
between pos0 for opening and for closing was achieved for either opening or closing of the valve.
valves with greater hysteresis.
Values of parameter a correspond to the slope of the curves. V. CONCLUSION
Higher values represent steeper slope while lower values The paper presented a development of the model of valves
represent softer slope. of a hydraulic system. The Amira DTS200 three tank system
The regression for valve V2 is presented in Fig. 19. was considered but used techniques can be easily generalized
to wide set of hydraulic systems. The real system contains
0.25
data close
several nonlinearities which incorporate complexity to the
model close system. Total number of experiments concerning valves
0.2
data open reached 433 taking altogether more than 113 hours.
model open
Resulting model includes all major nonlinearities and can
be integrated into a Simulink model of whole three tank
0.15 system.
k2

REFERENCES
0.1 [1] V. Bobál, J. Böhm, J. Fessl & J. Macháček:. Digital Self-tuning
Controllers: Algorithms, Implementation and Applications, London:
Springer – Verlag, 2005.
[2] G. P. Liu, Nonlinear identification and control – A neural network
0.05 Approach, London:Springer - Verlag, 2001.
[3] L. Ljung, System identification : theory for the user, Upper Saddle
River, NJ : Prentice Hall PTR, 1999.
[4] D. M. Himmelblau, J. B. Riggs, Basic principles and calculations in
0
-0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 chemical engineering, Upper Saddle River, N.J. : Prentice Hall, 2004.
valve pos [MU] [5] Amira, DTS200 Laboratory Setup Three - Tank – System, Duisburg:
Amira GmbH, 2002.
Fig. 19. Model of parameter k for valve V2 [6] M. C. Popescu, G. Manolea, L. Perescu-Popescu, and A. Drighiciu,
“Implementation of New Solution Software for Three Tank System
Behavior of system inside hysteresis area was studied as Control” in Mathematical Methods and Applied Computing, vol. 1, N.
Mastorakis, M. Demiralp, I. Rudas, C. A. Bulucea, L. Rogozea, Eds.
well. This task was time consuming because a performed set Athens: World Scientific and Engineering Acad and Soc (WSEAS),
of experiments took more than 45 hours. Experiments were 2009, pp. 202-207.
performed on valve V2 to study behavior inside hysteresis area. [7] M. N. Noh, M. S. Najib, and N. Saadah, “Simulator of water tank level
control system using PID-controller” in Advanced Topics on Water
This valve was selected because its hysteresis is the greatest.
Resources, Hydraulics and Hydrology - Proceedings of the 3rd
Experiment results are depicted in Fig. 20. The valve was IASME/WSEAS International Conference On Water Resources,
partially opened to a given position and then slightly moved Hydraulics and Hydrology (WHH '08), J. Krope, I. Sakellaris Eds.
towards closed position. This resulted in 7 parts, each Athens: World Scientific and Engineering Acad and Soc (WSEAS),
2008, pp. 168-171.
corresponding to an initial opened position. Each of these [8] M. Salmanzadeh, S. Torf, “Analysis of Surge in Pipelines Systems by
parts contains 18 measurements for different closing. Characteristis Method”, International Journal of Mechanic, 2, vol. 5, pp.
83-90, 2011.
valve 2
0.25 [9] L.Li and D. Zhou, “Fast and robust fault diagnosis for a class of
closing nonlinear systems: detectability analysis”, Computers &Chemical
opening Engineering, 28, 2004, 2635-2646.
hystersis experiments [10] D. Henry and A. Zolghadri, “Norm-based design of robust FDI schemes
0.2 for uncertain system under feedback control: Comparison of two
approaches”, Control engineering practice, 14, 2006, 1081-1097.
[11] S. Blazic, I. Skrjanc & D. Matko, “Globally stable direct fuzzy model
reference adaptive control”, Fuzzy Sets and Systems, 2003, 3-33
0.15
k2

0.1

0.05

0
-0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4
valve position [MU]

Fig. 20. Behavior inside hysteresis area (towards closed position)

Issue 4, Volume 5, 2011 317