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

net/publication/340932223

SIMULATION AND PERFORMANCE OF LIQUID LEVEL CONTROLLERS FOR

LINEAR TANK

Article in Jurnal Teknologi · April 2020

DOI: 10.11113/jt.v82.14245

CITATIONS READS
6 3,365

3 authors, including:

Qahtan Adnan Amer Talal

Tikrit University Tikrit University
9 PUBLICATIONS 29 CITATIONS 23 PUBLICATIONS 194 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Qahtan Adnan on 30 May 2020.

The user has requested enhancement of the downloaded file.

 Jurnal Full Paper
Teknologi
SIMULATION AND PERFORMANCE OF LIQUID Article history
Received
LEVEL CONTROLLERS FOR LINEAR TANK 19 September 2019
Received in revised form
Qahtan A. Mahmooda*, Amer T. Nawafa, Shaho A. 22 March 2020
Accepted
Mohamedalib
12 April 2020
aTikrit Published online
University, College of Petroleum & Minerals Engineering,
28 April 2020
Chemical Engineering Department, IRAQ
bTikrit University, College of Petroleum & Minerals Engineering, *Corresponding author
Petroleum and Gas Engineering Department, IRAQ Qahtan.adnan@tu.edu.iq

Graphical abstract Abstract

Level control of liquid in a tank or any similar container is widely used in applications
such as chemical and oil industrial processes. Control the level at desired value is
very important. This paper studies the performance of P, PI, and PID controllers in
controlling the level of a liquid. Mass balance is used to find mathematical model
of water tank level. Ziegler-Nichol (Z-N) and Cohen-Coon (C-C) tuning methods are
used to evaluate parameters of the controllers. The error indices such as Integral
Absolute Error (IAE) and Integral Squared Error (ISE) are used to compare between
performances of the controllers. MATLAB is used to test the control system
performance and compare the results with real values. Both simulation and
experimental results show that liquid level system can be controlled effectively by
using Z-N tuning method. The result shows that the PI controller gives better
performance in comparison with P and PID controller.

Keywords: Liquid level, Proportional Integral Derivative (PID) Control, Ziegler- Nichol
(Z-N) and Cohen-Coon (C-C), MATLAB/SIMULINK

© 2020 Penerbit UTM Press. All rights reserved

1.0 INTRODUCTION controllers are investigated to control the liquid level

in a tank by adjusting the inlet flowrate of the water
Controlling of liquid level in many processes is very to the tank. There are many ways to tuning PID
important. The refinery process requires the crude oil controllers. In this paper the most famous of these
to be pumped, stored in tanks then pumped to methods, Ziegler-Nichol (Z-N) and Cohen-Coon (C-
distillation column and stored in another tank. The C) were used to find controller parameters. The
level of the fluid in the tanks must be always process simulation is carried out using MATLAB and
controlled within the desired variables. Level control the results are compared with results from real
is widely used in process industries and wastewater system.
treatment industries [1]. Simulation is used in initial
system design to optimize controller gains, as well as
in Model Based Design. Real-time operation of 2.0 METHODOLOGY
continuous simulation is used for operator training
and off-line controller tuning. The software has to 2.1 Modeling of Liquid Level
solve the mass and energy balance to find a stable
operating point. Process simulation is widely used to The mathematical modeling of liquid level in the
find optimal conditions for an examined process [2]. tank is obtained using Mass balance. The scheme of
In this paper the performance of P, PI and PID the tank system is shown in Figure 1. Water is flowing

82:3 (2020) 75–82 | www.jurnalteknologi.utm.my | eISSN 2180–3722 |DOI: https://doi.org/10.11113/jt.v82.14245|

76 Qahtan A. Mahmood et al. / Jurnal Teknologi (Sciences & Engineering) 82:3 (2020) 75–82

to the tank at flowrate qin (lit/hr.) and the outlet flow

is qo (lit/hr.). The cross-section area of tank is (A cm2).
The height of the water level in the tank is
represented by h (cm) which is controlled by Taken Laplace for both sides of equation (9)
adjusting the flowrate of the pump. Assuming the
density of the inlet and outlet flowrate is constant
and the tank has a uniform cross-sectional area [3].

Table 1 Tank level parameters

Parameters Value Units

Area of tank (A) 240.4 cm2
Height (H) 7 cm
R 0.472 sec/cm2

The final transfer function of tank level is

evaluated by using the values of the parameters
Figure 1 scheme of tank system
shown in Table 1 as follows:

Amaterial balance around the SISO tank gives [3].

(Mass in) – (Mass out) = (accumulation of mass in
tank)
2.2 PID Controller

PID control composed of three types of controllers

(P, I and D) as shown in Figure 2 [4, 5].

The output volumetric flowrate of the tank is

assumed follows nonlinear relationships (square root
of height).

Where c is constant
Substitute equation (3) in equation (2) we get

Figure 2 Schematic of PID Controller

The inlet flowrate qin is a function of time. The outlet
qo is modeled as a nonlinear function of the liquid
level.
2.2.1 Proportional Control

√ √ The proportional controller is used to reduce the

√ error between the process output (measure value
√ MV) and the set point (SP), but cannot eliminate it.
√ [4, 5].
[√ ]
The P-controller has one adjustable parameter, (the
Subtracting equation (7) from equation (6) and controller gain)[5, 6].
rearrangement we get.
2.2.2 Integral Control

√ Integral control ultimately drives the error to zero by

the proportional control [4, 7]
√
77 Qahtan A. Mahmood et al. / Jurnal Teknologi (Sciences & Engineering) 82:3 (2020) 75–82

Table 2 Ziegler-Nichols setting based on the continuous

∫ cycling method

Kc τI τD
The PI-controller has two adjustable parameters, Ziegler-
(Proportional (integral (derivative
(the gain) and (the integral time) [4, 8]. Nichols
gain) time) time)
P-
2.2.3 Derivative Control Kcu/2 --- ---
controller
PI-
Kcu/2.2 Pu/1.2 ---
The derivative controller acts upon the derivative of controller
the error, so it is most active when the error is PID-
Kcu/1.7 Pu/2 Pu/8
changing rapidly. The derivative controller is worked controller
to decrease oscillation in output process[4, 9].

Cohen-Coon

The Cohen and Coon tuning method is applied in

The PID- controller has three adjustable parameters, the graphical construction as shown in Figure 4. (C-
(the gain),(the integral time) and (the derivative C) method reduces the process reaction curve to
time)[4, 7]. first-order with the transport lag model given by
equation (11) [4, 13-15].
2.2.4 Tunning of the Controller

Ziegler-Nichols

The Ziegler-Nichole (Z-N) setting has been widely

used as a benchmark for evaluating different tuning
methods and control strategies [5, 6]. Ziegler-Nichole
(Z-N) is applied in a closed loop system for tuning a
PID controller as follows [5, 10-12]:
1. Reduce the (integral time) and (derivative
time) to zero and using only P-controller.
2. Increase Kp untill oscillations occur at critical
value (Kp=Kcu)
3. Evaluate (ultimate gain Kcu) (ultimate period
Pu sec) as shown in Figure 3.

Figure 4 Typical process reaction curve showing graphical

construction

The experimental procedure for (C-C) tuning

method is quite simple and the control loop shown
in Figure 5. When the process reaches to steady-
state, the controller is changed to a manual mode.
Then (3 to 5 %) step change in the controller output
is introduced. The response of the system is called
the process reaction curve [5, 16].
Figure 3 Experimental determination of the ultimate gain The parameters of PID-controller tuning by using
Kcu (Cohen – Coon) open loop response are shown in
Table 3.

The PID-controller parameters are now specified

as showns in Table 2.
78 Qahtan A. Mahmood et al. / Jurnal Teknologi (Sciences & Engineering) 82:3 (2020) 75–82

receiver to the tank. The flowrate is measured by

using a rotameter on the discharge of the pump.
The outlet flowrate is adjusted by using a hand valve
between the tank and the receiver to circulate the
water. The level transmitter is used to measure the
level inside the tank and send a signal to the PID
controller which is used to control the pump speed.

Figure 5 Control loop for process reaction curve method

Table 3 PID controller parameters by (Cohen-Coon) tuning

Controller

P [ ]

PI [ ]

PID [ ]
Figure 7 Experimental tank level control

3.0 RESULTS AND DISCUSSION

2.3 Simulation
3.1 Modeling of Liquid Level
The simulation for the tank process system was
created by using MATLAB Simulink software, and the A step change (25 to 30%) in the input flowrate to
block diagram is shown in Figure 6 [17-20]. The the process is done and the output response of the
system simulation response is analyzed for change in system is observed as shown in Figure 8. This
the set point of the liquid level in the tank from (3 to procedure is called the (process reaction curve).
6%) and responses recorded at a different controller
and different controller tuning methods.

13

∆L
Level cm

11
level

SLOPE
7
0 10 20 30 40 50 60 70 80 90 100

Figure 6 Simulink Block Diagram for Process Tank Figure 8 Tank level reaction curve for Cohen-Coon tuning

2.4 Experimental Work The dynamic model for liquid level is described
by a (first order with dead time), as shown in
A five-liter cylindrical vessel made from glasses is Equation (11).
used as a tank as shown in Figure 7. The pump is
used to flowrate the water (0 to 200 lit/hr.) from the
79 Qahtan A. Mahmood et al. / Jurnal Teknologi (Sciences & Engineering) 82:3 (2020) 75–82

The parameters of analytical model, Equation (10) compare with (C-C) tuning method. Figures 13 and
and experimental model, Equation (11) are 14 show response of PID controller to step change in
calculated and summarized in Table 4. set point of liquid level from (3 to 6 %( and we see
the PID controller have oscillation with (Z-N) and (C-
Table 4 Parameters of transfer function of liquid level C) tuning method.

Parameters Analytical Experimental

R sec/cm2 0.472 0.407
τ second 113.4 110
second 0.5 3.5

From the results, we see small differences in

values between analytical and experimental model
for the liquid level.

3.2 (Ziegler–Nichols) and (Cohen–Coon) Methods

(Z-N) Is used for closed loop response which made

the system oscillate, and the values of ultimate gain
and ultimate period are found to be as shown in
Table 5. Figure 9 Simulation response of P-controller by using (Z-N)
tuning
Table 5 Parameters (Ziegler–Nichols) closed loop response

(Ultimate gain Kcu) (Ultimate period Pu )(sec)

12 10

Cohen and Coon used the approximate model

as shown in equation (11) and estimate the values
(R, and ) as shown in Table 6.

Table 6 Parameters of (Cohen – Coon) approximate model

R [sec/cm2)] (sec) (sec)

0.407 110 3.5

The values of PID-controller parameters

calculated by applying both (Z-N) and (C-C) tuning Figure 10 Simulation response of P-controller by using (C-C)
methods are summarized in Table 7. tuning

Table 7 Parameters of PID controller

Ziegler- Nichols Cohen-Coon

Controller
Kc sec Kc sec
P 6 --- --- 78 --- ---
PI 5.4 8.33 --- 69 7 ---
PID 7.2 5 1.25 98 8.5 1.27

3.3 Simulation Results

Simulation results of (Z-N) and (C-C) methods are

discussed in this section. Responses of P controller to
step change in set point of liquid level from (3 to 6 %(
has been shown in Figures 9 and 10. From the
figures, we see the P controller with (Z-N) tuning is
Figure 11 Simulation response of PI-controller by using (Z-N)
better than (C-C) tuning method. Similarity, response
tuning
of PI controller to step change in set point of liquid
level from (3 to 6 %( has been shown in Figures 11
and 12. From the figures, we see the PI controller
with (Z-N) tuning is better and have less oscillation in
80 Qahtan A. Mahmood et al. / Jurnal Teknologi (Sciences & Engineering) 82:3 (2020) 75–82

of P, PI and PID controller to step-change in set point

of liquid level from )3 to 6 %( have been observed
and applied in real-time. Figures 15, 16 and 17 show
the response of controllers in real-time. From the
figures, we found that the PI- controller has lower
overshoot and minimum rise time.

Level cm
z-n
5
Figure 12 Simulation response of PI-controller by using (C-C) set point
tuning
4 c-c

3
0 20 40 60 80 time sec
Figure 15 Real time response of P-controller by using (Z-N)
and (C-C) tuning

6
Level cm

5 z-n
set point
Figure 13 Simulation response of PID-controller by using (Z- 4
N) tuning c-c
3
0 20 40 60 80 time sec
Figure 16 Real time response of PI-controller by using (Z-N)
and (C-C) tuning

6
Level cm

z-n
5
set point
Figure 14 Simulation response of PID-controller by using (C- 4 c-c
C) tuning

3
3.4 Performance Analysis of System 0 20 40 60 80 time sec
In this section, P, PI and PID controller have been Figure 17 Real time response of PID-controller by using (Z-N)
used to control the liquid level efficiently. The three and (C-C) tuning
parameters of the PID controller have been
adjusted by applying (Z-N) and (C–C) tuning
methods as summarized in Table 7. The performance
81 Qahtan A. Mahmood et al. / Jurnal Teknologi (Sciences & Engineering) 82:3 (2020) 75–82

The comparison between the Rise time, Settling (Z-N) method gives best response for the controlling
time, and percentage overshoot for the controllers of liquid level.
P, PI, and PID is shown in Table 8. It is noted that the
PI-controller provides more satisfactory performance Nomenclature
with reference to rise time, settling time and
percentage overshoot % in comparison to the P and qi : Water flow inlet (cm3/sec)
PID controller. qo : Water flow outlet (cm3/sec)
V : Volume of water in the tank (cm3)
Table 8 Comparison of Performance Response between (Z- A : Area of tank (cm2)
N) and (C-C) Methods H : Height of water in tank (cm)
: Output signal from controller, psig.
Tuning (Ziegler-Nichols) (Cohen-Coon)
R : Process gain (sec/cm2)
rules Method Method
Kc : Proportional gain
P PI PID P PI PID
Rise time
Ku : Ultimate gain.
17 16 18 19 18 25 Pu : Ultimate period (minutes per cycle)
(sec)
Settling e : Error (set point)–(measured variable).
˃80 65 ˃80 ˃80 ˃80 ˃80
time (sec)
Overshoot
20% 16% 19% 19% 18% ---- Greek Letters
% (cm)
 : Water density (g/cm3)
τI :Integral time (minute)
The integral absolute error (IAE) and integral τD : Derivative time (minute)
square error (ISE) are often used to evaluate the : Time delay (minute)
response of control system at different controller τ : Time constant (minute)
tuning methods as shown in Table 9. From the table,
it is observed that PI-controller with (Z-N) and (C-C) Abbreviations
tuning has the lowest value of (IAE) and (ISE).
P Proportional
Table 9 Error Analysis for Different Tuning Methods
PI : Proportional-Integral
PID : Proportional-Integral-Derivative
P- Controller PI- Controller PID- Controller IAE : Integral Absolute Error
Method
IAE ISE IAE ISE IAE ISE ISE : Integral Square Error
Z-N 29.560 46.724 29.285 46.477 29.90 52.386
C-C 31.195 56.450 30.368 48.970 39.950 68.386
Reference

4.0 CONCLUSION [1] Luyben, L. 1989. Process Modeling, Simulation, and

Control for Chemical Engineer. New York: McGraw Hill,
Inc.
In this study, PID Controller is implemented to control [2] Rhodes, L. 1996. The Process Simulation Revolution:
the liquid level in tank within the desire values. (Z-N) Thermo physical Property Needs and Concerns. Journal
and (C-C) tuning methods were used to find of Chemical and Engineering Data. 41: 947-950.
controlling variables that were actually tested in the [3] Donald, C., and Stevan, L. 2009. Process Systems
Analysis and Control. New York: McGraw Hill, Inc.
real time to find the efficiency of the controllers. The
[4] Ogunnaike, A. and Ray, H. 1994 Process Dynamics,
mathematical model of liquid level in the tank has Modeling and Control. New York: Oxford University
been derived depending on the material balance Press, Inc.
and the transfer function was first order with a delay [5] Qahtan, A. M., Amer, T. N., Maha, E., Layth T. and
time. Omar, D. 2018. PID Temperature Control of
Demineralized Water Tank. International Conference
on Materials Engineering and Science, IOP Publishing.
454(2018): 012031
[6] Dale, S., Thomas, E. and Duncan, M. 2011. Process
Dynamics and Control. John Wiley and Sons, Inc.
[7] Marlin, E. 2000. Process Control: Designing Processes
MATLAB is used to test the controller’s and Control Systems for Dynamic Performance.
performance and then compare the results with McGraw-Hill.
those presented in the real time. From simulation [8] Brian, R. and Ben, B. 2006. Process Dynamics and
results, it was found that PI controller has best Control, Modeling for Control and Prediction.
Netherlands: John Wiley and Sons Ltd.
performance and giving lower overshoot and less [9] Tyreus, D. and Luyben, L. 1992. Tuning of PI Controllers
setting time. (IAE) and (ISE) values are calculated for for Integrator/Dead Time Processes. Ind. Eng. Chem.
tow tuning methods, and it is observed that (Z-N) Res. 31(11): 2625-2628.
method give good performance for PI-controller. [10] Beza, G. 2016. Water Level Controlling System Using Pid
Hence, we concluded that PI controller tuning with Controller. International Journal of Applied Engineering
Research. 11(23): 11223-11227.
 82 Qahtan A. Mahmood et al. / Jurnal Teknologi (Sciences & Engineering) 82:3 (2020) 75–82

[11] Vassilios, T., Sanjo, P., Matthew, S. and Truong, D. 2013. [15] Astrom, K.J. and Hagglund, T. (1995). PID controllers:
Modeling, Simulation and Control of Temperature and theory, design and tuning, Instrument Society of
Level in a Multivariable Water Tank Process. America.
International Journal of Engineering, Research and [16] Power, H. M. and Simpson, R. J. 1978. Introduction to
Development. 8(6): 01-11 Dynamics and Control. McGraw-Hill.
[12] Mostafa, F. and Aboubaker, G. 2015. PID Controller [17] Araki, M. 2008. PID Control. Encyclopedia of Life
Design for Two Tanks Liquid Level Control System Using Support Systems (EOLSS). 2(1): 26.
Matlab. International Journal of Application or [18] Shinskey, F. G. 1996. Process Control Systems-
Innovation in Engineering and Management. 4(5). Application, Design and Tuning. 4th Edition. McGraw-
[13] Yuli, W. 2018. Study of Double-holding Water Tank Hill Inc.
Liquid Level Control Base on Neural Networks PID [19] Ashish, T. 2002. Modern Control Design with MATLAB
Control. Advances in Intelligent Systems Research and Simulink. England: John Wiley and Sons.
(AISR). 151. [20] Leigh, J. R. 1987. Applied Control Theory. Revised 2nd
[14] Philips, C. L. and Harbor, R. D. 1991. Feedback Control edition. Peter Peregrinus Ltd.
Systems. 2nd edition. Prentice-Hall.

View publication stats