International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 05 Issue: 02 | Feb-2018 www.irjet.net p-ISSN: 2395-0072

FUZZY GAIN SCHEDULED PI CONTROLLER FOR A TWO TANK

INTERACTING CONICAL TANK SYSTEM
J.Merlin(1), Dr.K.Prabhu(2)

M.E Control and Instrumentation, Kongu Engineering College, Erode.

(1)

(2)Associate Professor, Department of EIE, Kongu Engineering College, Erode.

---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract—This Paper presents the mathematical modeling researchers have done work on the conical tank process and
of two tank conical interacting system is designed. The adaptive control.
transfer function of TTICTS parameters and Proportional
Integral (PI) controller parameters are obtained for Anandanatarajan et al. [1] have proposed the performance of
implementing Gain Scheduled Adaptive Controller (GSAC) for PI controller at various operating regions in the conical tank.
TTICTS. The Ziegler-Nichols method is used to obtain the Bing Chen et al. [2] design of a nonlinear system has
optimal PI control settings. Then the tuning is done by using proposed the adaptive fuzzy control for a class of nonlinear
GSAC. The step response is obtained for both conventional systems with nonstrict-feedback structure by using fuzzy
PI&GSAC method. The response of GSAC shows less overshoot logic systems.
and quick settling time.
Jaya et al. [3] proposed the need of a new fuzzy gain
Keywords—Nonlinear process, Adaptive control, Gain scheduled PI controller for a two tank conical tank
scheduled Adaptive control, Ziegler-Nichols method, Two interacting process. It is then followed by the development of
Tank Conical Interacting System. a fuzzy gain scheduling scheme by PI controller for the
process. Marshina et al. [4] proposed the controller. Mohini
Narendra Naik et al [5] have proposed the adaptive fuzzy
1. INTRODUTION
control for a class of nonlinear systems with non-strict-
feedback structure by using fuzzy logic systems.
The control of fluid level in tanks and stream
between the tanks is a fundamental issue in process
Nagamani et al. [6] proposed a method for the new fuzzy
ventures. The procedure require the fluids to be pumped,
gain scheduled PI controller (FGSPI controller) for a two
put away in tanks and the pumped to another tank.
tank conical interacting level system. It is then followed by
Ordinarily the fluid will be prepared by synthetic or blending
the development of a fuzzy gain scheduling scheme by PI
treatment in the tanks, yet dependably the level of the liquid
controller for the process. Fuzzy gain scheduling scheme by
in the tanks must be controlled. The level is too high which
PI controller for the process. Fuzzy rules and reasoning are
may upset the reaction equilibria and cause damage to the
utilized to tune the PI controller parameters. Although
equipment, or result in spillage of hazardous material. If the
several methods have been developed, in most of them,
level is too low, it may have bad consequences for the
either some assumption is made or the nonlinear
sequential operations. Hence, control of liquid level is very
characteristics are linearised and then controller is designed.
important and common task in process industries.
The paper is organized as follows: In section II, the process
Conical tank finds wide application in process industries like
and hardware set-up are described. In section III, simulation
waste water treatment industries, sewage water treatment
studies and implementation of conventional controllers are
industries, concrete mixing industries, food processing
discussed. In section IV, adaptive control through Gain
industries and hydrometallurgical industries due to their
Scheduling method is discussed. Finally results and
shape leads to better drainage of solid mixtures, slurries and
conclusions are discussed in section V.
viscous liquid at the bottom of the tank. The control of the
cone shaped tank exhibits a testing issue because of its non-
linearity and always showing signs of change cross sectional 2 PROCESS DESCRIPTIONS:
territory through and through of the tank.
The TICTS is a benchmark problem due to its non-linearity
property. TICTS consist of two identical conical
Conventional PID Controllers are widely used in
tanks(TANK1 and TANK2) , two independent pumps that
industries since they are simple, robust & familiar to the field
deliver the liquid flows FIN1 and FIN2 to TANK1 and TANK2
Operator. Practical systems are not precisely linear but may
through the two control valves CV1 and CV2 respectively.
be as represented as linearized models around a nominal
These two tanks are interconnected at 31 cm of the tanks
operating point. The controller Parameters are tuned at that
through manual control valves MV12. MV1 and MV2 with
point does not reflect the real time system characteristics
valve coefficients β1 and β2 are the outlet valves for Tank1
due to variation in process parameters. The solution is
and Tank2 respectively.
controller parameters have to be done adjusted. Many

© 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 847
 International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 05 Issue: 02 | Feb-2018 www.irjet.net p-ISSN: 2395-0072

3 CONVENTIONAL PI CONTROL

Conventional PI Controllers are widely used in industries

and simple easy to understand the process initialization or
operation. PI controllers are designed for linear system
around six operating regions.

For each linearized region controller parameters are

optimally tuned using Ziegler Nichols method. The process
Fig 1 Experimental set up of TTICTS variable going through different operating region the
respective controller parameters table shows different
TABLE -1: Operating Parameters of TTICTS linearized region and controller parameters of TTICTS using
Z-N method.
Parameter Description Value
R Top radius of the conical 17.35 cm TABLE II Different Linearised Regions and Controller
tank Parameters for TTICTS
H Maximum height of TANK1 60 cm
& TANK 2 Operatin Flow rate Height Kp Ki
F &F Maximum Inflow to TANK1 129cm3/s g region (LPH) (h2)cm
IN1 IN2
& TANK 2 Region 1 0-150 10 70.08 0.13
β Valve co-efficient of MV 35cm3/s Region 2 150-250 20 80.4 0.11
1 1

β Valve co-efficient of MV 78.2cm2/s Region 3 250-320 30 40.5 0.7

2 2

β Valve co-efficient of MV 19.69cm2/s Region 4 320-390 35 150 0.14

12 12

The process variable is going through different operating

2.1 Representation of model region the respective controller parameters table II shows
the different linearised regions and Controller Parameters
A model is a mathematical representation of a system. Two for TTICTS.
Tanks conical interacting system is considered as measured
variable and FIN1 are considered as manipulated variables. 4 GAIN SCHEDULING ALGORITHM
The mathematical model of two tank conical interacting
system is given by the equations. Gain scheduling is an approach to control of non linear
system that uses a family of linear controller each of which
provides satisfactory control for a different operating point
dh1 FIN1 -β1 h1 -sign(h1 -h 2 )β12 h1 -h 2 of the system. It is a technique to increase the region of
= attraction to a range of possible operating points. One of the
dt R 2 h12
π most popular adaptive control technique is gain scheduling
H2 technique. It facilities process control, the gain and time
constants vary with the current value of the process variable.
dh 2 FIN2 -β 2 h 2 +sign(h1 -h 2 )β12 h1 -h 2
= 4.1 Principle
dt R 2h 22
π
H2 This method consists of two loops. Inner loop composed of
the process and the PI controller. Outer loop that adjusts the
Where controller parameters based on the operating conditions.

FIN1 =Inflow rate of the TANK1 (cm3/s)

FIN2 =Inflow rate of the TANK2 (cm3/s)
h1 =Liquid height in TANK1 (cm)
h2 =Liquid height in TANK2 (cm)
R =Top radius of the tank (cm)
H =Total height of the tank (cm)
β1 =Valve co-efficient of MV1 (cm2/s)
β2 =Valve co-efficient of MV2 (cm2/s)
β12 =Valve coefficient of MV12 (cm2/s) Fig- 3: Block diagram of GSAC for TTICTS

© 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 848
 International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 05 Issue: 02 | Feb-2018 www.irjet.net p-ISSN: 2395-0072

C1  , C2 
Gain scheduling can be regarded as mapping from process 1 1
2h h
parameters controller parameters. The scheduling variable
2h
used in the process is area of TANK1 A (h 1).The scheduling 10 20 20

R  , R 
variable has been determined at each operating condition, 1 1
the controller parameters are calculated at each operating
conditions by using some suitable set of equations. It relates 1 β C 2 β C
1 1 2 2
τ  A(h )R , τ  A(h )R
process parameters and controller parameters. The
controller is thus tuned for each operating condition. 1 1 1 2 2 2
Finally, the transfer function of Two Tank Conical Interacting
4.2 Transfer function of TTCIS System is given by,
h
2 
R
2

The model of TANK1 can be represented as,
τ τ s  [τ τ  A(h )R ]s  1
2
Rate of Accumulation=Inflow Rate- Outflow Rate. F 1 2 1 2 1 2
IN1
 Fin  Fout
dv
dt 4.3 RELATION BETWEEN TWO TANK CONICAL
INTERACTING SYSTEM PARAMETERS AND PI
Where,
 2
PARAMETERS
V r h.
3 The closed loop transfer function of unity feedback two

Since the ratio, 

r R tank conical interacting system with PI controller is derived
as,
h H

K p R 2 [1  τ i s]
Substitute equation we get,
 R 
2 C(s)
V
τ i τ1τ 2 s  τ i (τ1  τ 2  A(h1 )R 2 )s  (1  k p R 2 )τ i s  k p R 2
3
2
h R(s) 3 2
3 H
Diff.volume, we get, The characteristics equation of above transfer function is,

 A τ i τ τ s  τ (τ  τ  A(h )R )s 
2 2
dv R dh dh 3 2
h 1 2 i 1 2 1 2
2
(1  k R )τ s  K R  0
dt H dt dt
For tank 1, p 2 i p 2
A1 1  Fin  Fmin
dh The controller parameter can be written as,
A(h1 )(τ1  τ 2 )
Kp 
dt
For tank 2,
τ1τ 2
A1 1  Fmid  Fout τ τ τ
dh
dt i 1 2
From this,

K 
Equation is rewritten as, 1
dh1 Fin  β1 (h 2  h1 )

i τ
i
dt A1 The equations provide the relationship between two tank
Where, conical interacting system parameters and PI controller

A1  
2 parameters. And the equations are used to implement Gain
R 2
h scheduling Adaptive Controller for TTICTS.
2 1
β h
H

h 
TABLE III Tuned Parameters of Gain Scheduled Algorithm.
1 1
1 2h h
A2   h 2 h 
2 Operating Flow rate Height Kp Ki
R 2

10 20 region (LPH) (h2)cm
β
A(h )s 
2 1 h
H 1 1 Region 1 0-150 10 26.12 0.129
1 2h h Region 2 150-250 20 397.9 0.72
10 20 Region 3 250-320 30 128.4 0.018
Let us assume, Region 4 320-390 35 724.0 0.44

© 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 849
 International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 05 Issue: 02 | Feb-2018 www.irjet.net p-ISSN: 2395-0072

In TTICTS the scheduling variable A (h1) was calculated from 5.1 OBSERVATION:
the measurement of each of h1.Using this A (h1) the table III
shows the tuned parameters of gain scheduled algorithm. Table IV shows comparison results for PI controller and
Gain scheduled PI controller
5 RESULTS AND DISCUSSION
Operating PI controller Gain scheduled PI
The controller responses of conventional PI controller in region controller
operating region of two tank interacting conical tank system 1(set point= Rise time(sec)=5.76 Rise time(sec)=4.43
are simulated. 10cm) Settling Settling
time(sec)=28.5 time(sec)=3.65
For operating region1 of the set point=10cm Overshoot(%)=6.5 Overshoot(%)=5.5
2(set point= Rise time(sec)=9.75 Rise time(sec)=7.61
20cm) Settling Settling
time(sec)=30.2 time(sec)=24.2
Overshoot(%)=11.6 Overshoot(%)=9.54
3(set point= Rise time(sec)=5.76 Rise time(sec)=3.54
30cm) Settling Settling
time(sec)=10.75 time(sec)=7.54
Overshoot(%)=11.2 Overshoot(%)=8.46
4(setpoint=35 Rise time(sec)=6.79 Rise time(sec)=0.41
cm) Settling Settling
time(sec)=30.2 time(sec)=2.05
Overshoot(%)=8.5 Overshoot(%)=5.95

The table IV shows the settling time was very much reduced
Fig -3: Response of Two Tank Interacting Conical Tank when compared with conventional PI control scheme. In case
System for PI Controller of Gain scheduling adaptive algorithm, the response for the
process parameter variation fast without any steady state
From the Fig 3 it can be seen that if set point is given as error. This was achieved, in Gain scheduling adaptive
10cm for two tank interacting conical tank system with PI algorithm by measuring h1 and required set point thereby
Controller, the Settling time(sec)=28.5,Rise updating controller parameter. This leads to better
time(sec)=5.76,Overshoot(%)=6.5%. transients and steady state response.

The controller responses of gain scheduled PI controller in 6 CONCLUSION

operating region of TTICTS are simulated.
The conventional PI controller and gain scheduled PI
For the operating region 1 of the set point=10cm, controller for every operating region is designed and applied
to two tank conical interacting tank system. But under
situations where the process parameters vary the
conventional controllers does not provide satisfactory
results. By proper PI controller tuning of gain scheduled PI
controller provides the better results for every operating
region. The future work can be extended to tune the
controller parameters. It is based on switching conditions
plant operating conditions selected the controller gain
values. In order to improve the transient response with less
overshoot and quick settling time, the Fuzzy based gain
scheduled PI controller is used .

REFRENCES
Fig- 4: Response of two tank conical interacting system
with Gain scheduled PI Controller [1]Anandanatarajan.R, Limitations of a PI controller for a
first order Non-linear process with dead time , ISA
From the Fig 4 it can be seen that if set point is given as 10 transcations,pp185-199,April 2014.
cm for two tank interacting conical tank system with Gain
scheduled PI Controller, the Settling time [2]Bing Chen,Xiaoping P.Liu,Shuzhi Sam Ge and Chong Lin,
(sec)=3.65,Risetime(sec)=4.43,Overshot(%)=5.5% Adaptive Fuzzy Control of a Class of Nonlinear systems by
Fuzzy Approximation Approach , Institute of Electrical&

© 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 850
 International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 05 Issue: 02 | Feb-2018 www.irjet.net p-ISSN: 2395-0072

Electronic Engineers (IEEE Transcations ). Vol:20, No.6,pp. system ,Institute of Electrical & Electronics Engineers (IEEE
1012-1021,Dec-2012. Transcations ),Vol-18,No:6,pp.1070-1082,June 2010.

[3] Jaya.D, Fuzzy gain scheduled PI controller for a Two [ ]Zhang Huaguang and Lilong Cai, Nonlinear Adaptive
Tank Conical Interacting Level System International Control using the Integral and its application to CSTR
Journal of Engineering and Technology (IJET), Vol:6,No. 6, Systems , Institute of Electrical & Electronics Engineers
Jan 2015. (IEEE Transcations ),Vol-32,No:3,pp.367-372,June 2014.

[4]Marshiana.D & Thirusakthimurugan.p, Design of Zieger

Nichols Tuning Controller for a Non-linear system ,
International Conference on Computing and Control
Engineering (ICCCE 2012),12&13 April,2012.

[5]Mohini Narendra & Nicole Monterio, Design and

simulation of gain scheduled Adaptive Controller using PI
Controller for Conical Tank process , International Journal
for Innovative Research in Science & Technology(IJIRST)
Vol:2,Issue 04 | September 2015.

[6]Nagamani G.,Swathi Lakshmi &Mahalakshmi, Design of

Fuzzy gain scheduled PI controller for a Nonlinear SISO
Process , International Conference Engineering and
Technology(ICET),Jan 2015.

[ ]RaviV.R, & Thayagarajan.T, Adaptive decentralized PI

controller for two conical tank interacting level system ,
Journal for Science and Engineering,vol:39,Issue12,pp 8433-
8451,Dec-2014.

[ ]Ravi.V.R & Thayagarajan.T, Application of adaptive

controller technique to Interacting Non-Linear system ,
Institute of Electrical & Electronics Engineers
(IEEE),vol:45,No:28,2011.

[ ]RaviV.R & Thayagarajan.T, A Decentralized PID controller

for Interacting Non-Linear systems’, Emerging Trends in
Electrical and Computer Technology (ICETECT) ,
International Conference, pp.297-301,2011.

[10]Sowmya.P & Srivignesh.N, A Fuzzy Control Scheme for

Nonlinear process , International Conference On Advanced
Engineering Science and Management (ICAESMT), May 2012.

[11]Sukanya R, & Sivanandam Venkatesh, Design of

Controllers based on MPC for a Conical Tank System ,
International Conference On Advanced Engineering Science
and Management (ICAESMT),June 2015.

[ ]Warier & Sivanandam Venkatesh, Design of controllers

based on MPC for a conical Tank System , International
Conference on Advances In Engineering, Science and
Management (ICAESM),2012.

[ ]Yue Fu & Tianyou Chai, Self-Tuning Control With a Filter

and a Neural Compensator for a Class of Nonlinear Systems ,
Institute of Electrical& Electronics Engineers(IEEE
Transactions),Vol:24,No:5,pp.837-843,May 2013.

[ ]Yajun Zhang & Tianyou Chai, Self-Tuning Control With a

Filter and a Neural Compensator for a Class of Nonlinear

© 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 851