See discussions, stats, and author profiles for this publication at: [Link]

net/publication/366324310

Tuning of PID Controller for Unstable Processes

Conference Paper · December 2022

CITATIONS READS
0 202

1 author:

Dilip N Pawar
Bharati Vidyapeeth Deemed University
15 PUBLICATIONS 30 CITATIONS

SEE PROFILE

All content following this page was uploaded by Dilip N Pawar on 16 December 2022.

The user has requested enhancement of the downloaded file.

 CONTROL INSTRUMENTATION SYSTEM CONFERENCE (CISCON2005)

Tuning of PID Controller for Unstable

Processes
D.N. Pawar, Assistant Professor, BVCOE, Navi Mumbai.

classical PID feedback controller can be

Abstract— This paper deals with the tuning of PI accomplished with a technique by considering the
and PID controllers for the unstable processes due D, I, and P – Controllers as compensators in
to transportation lag. The methods based on the cascade and adding them sequentially to the process
application of Padé approximation to describe the [4], [8]. This technique is suggested by Anderson
transfer function of the delay. The over all transfer N. A. [4]. The analysis and simulation of the
function of the system is obtained by convoluting problems under study is done by using MATLAB
the transfer functions. The systems chosen are [7].
integrating and unstable. The common The methods for stability analysis of linear
characteristic of integrating and unstable systems is systems are failing when any of the blocks on the
that they are non-minimum phase. Therefore their open loop contains a dead time or transportation
open-loop frequency response, with the controller in lag. This is because the dead time introduces an
cascade, has a phase maximum. For these systems exponential function of the Laplace transform
frequency domain design technique such as Bode
variable in the characteristic equation. This means
Plot is applied to obtain cascade PI and PID
the equation is non-rational. So to convert the non-
controller. The PI and PID controller action to
rational into rational one, a Padé approximation is
unstable processes due to dead time is studied.
Index Terms — Integrating Processes, Liquid Flow used [5], [6].
Control, Padé Approximation, PID,
A. PADÉ APPROXIMATION:
I. INTRODUCTION
The most common non-rational transfer function
In the literature of control theory different
model encountered is that resulting from a delay in
approaches have been found for tuning methods for
the system as
integrating and unstable processes. Internal Model
L [f(t-T)] = e-Ts F(s)
Control (IMC) used by Morari and Zafiriou for PID
The function e-Ts is clearly transcendental and this
tuning in case of parameter uncertainty. Modified
changes the fundamental character of the system if
Smith predictors were used by De Paor and Egan
treated rigorously. There is an often-useful
and Astrom, to cope with unstable and integrating
approximation of the transfer function, which is
processes with long time delay [2], [3]. The design
rational and hence results in the addition of finite
of PID controller for first order system with long
states to the differential equation.
delay was presented by [8], where algorithm is
The general idea is to find a rational function
implemented in the program is used here to find the
approximation to a given function, in the case e-Ts
controller parameters of unstable processes. An
and replace the function with the rational
open-loop frequency response design of the
approximation. One commonly used approximation
is Padé Approximation and derived as –
Manuscript received on August 2, 2005 for review and selection of this
paper is intimated through e-mail on August 23, 2005 for presentation. This
work was supported in part by the Department of Instrumentation, BVCOE,
e-Ts/2 1 – Ts/2 + T2s2/8 + …
-Ts
Navi Mumbai, and permitted to send this paper for CISCON – 2005. e = -----= --------------------------- -Æ 1
D. N. Pawar is with the Bharati Vidyapeeth College of Engineering, Navi eTs/2 1+ Ts/2 + T2s2/8 +…
Mumbai, and working as Assistant Professor in Instrumentation Department
Since 1999. (Author phone: (O) 022 2746 1074, Ext. 115; fax: 022 2757
3196; e-mail: d_n_pawar@ [Link]).

1
 CONTROL INSTRUMENTATION SYSTEM CONFERENCE (CISCON2005)

It ensured that there is no static error, and then a

B. PROPERTIES OF PADÉ APPROXIMATION: controller with an integrator must be used. So, PI
controller is designed for this process.
a) One of the nice property of the Padé
Approximation of a delay is that the poles are A. PROPORTIONAL + INTEGRAL (PI)
all in the left half – plane, however, the zeros CONTROLLER
are all in the right half – plane. This does reflect
the fact that delay is often destabilizing in feed The transfer function of the PI controller is given by
back loop.
b) The transfer functions are All Pass, i.e. the
magnitude of the transfer function is 1 for all K c (Ti s + 1)
frequencies; |T (jw)| = 1. Gc ( s) = −−−−−−−−−− >3
Ti s
c) The transfer functions are non-minimum phase,
i.e. they have zeros in the right half – plane.
d) As the order of the approximation is increased,
and the open-loop transfer function of the system is
they approximate the low frequency phase
given by –
characteristic with increasing accuracy.
e) It results in a finite dimensional model, which
may hide fundamental behavior of partial G ( s ) = G c ( s )G p ( s )
differential equations.
f) Padé Approximation of a stable transfer K c K p (Ti s + 1)e −Ts
= −−−−−−−− > 4
function is not always stable. Ti s 2 (Ti s + 1)

II. INTEGRATING PROCESSES Typical open-loop frequency responses

Consider the transfer function of a second order G(jw) with Kp = 1, and T= 0.2 sec. is initially
process with time delay is given by considered in the algorithm and plotted by using
bode command.
K p e −Ts
G p ( s) = − − − − − − − − − −− > 2
s (T1 s + 1)
From Figure 1 it is seen that there is no static error
to a set point change i.e. Kp, with a proportional
controller.

Figure 2: Bode Plot of Process

(Equation 3)
Similarly the frequency response of the process
represented by Equation 4 is also shown in
Figure 1: Step Response of Process for
following Figure 3.
Kp = 1, 2, and 3

2
 CONTROL INSTRUMENTATION SYSTEM CONFERENCE (CISCON2005)

In this process, PID controller is electronic, and rest

of the components is pneumatic. The process is
shown in Figure 4
The open-loop transfer function of the process is
given below –

Y ( s) 1.6e −0.25 s
Gc ( s) = = − −− > 5
M ( s ) (1 + 1.1s )(1 + 0.21s )

The time constants 1.1 min and 0.21 min are

associated with the column and delay channel
respectively. The transport lag is 0.25 min is
Figure 3: Bode Plot of Process associated with delay channel.
(Equation 4)
PID Compensation Technique
III. LIQUID FLOW CONTROL
PID Controller Process
One of the commonly used examples in process
control laboratories is considered here. Water is
feed through the tank via precisor-driven control
valve and delay channel to a Perspex column from
the bottom of which the water flows through an
orifice type flow meter to a sump tank.
A PID controller monitors the column outflow rate Figure 5: Block Diagram
i.e.-controlled variable, compares this with a set
point and drives the valve i.e. manipulated variable, PID Controller Form:
according to the difference.
 1 
Control Valve G c ( s ) = k c (1 + Td s )1 +  − − − −− > 6
From Reservoir
 Ti s 
Tank
The addition of the control modes to a process can
Delay Channel
be done as per the guidelines suggested by
Anderson, N. A.
L
Precisor A. DESIGN SPECIFICATIONS OF DERIVATIVE
MODE
m(t) M(s)
Set
PID First adding the Derivative Mode by aligning the
Controller
Point
y(t) G PD ( s ) = 1 + Td s
Flow Meter
G PD ( jw) = 1 + jwTd
Y(s)

therefore,
To Sump Tank tan −1 (wTd ) = 46.6 0
Figure 4: Liquid Flow Control on solving we get,
wTd = 1.05747

3
 CONTROL INSTRUMENTATION SYSTEM CONFERENCE (CISCON2005)

46.60 phase point of the controller and the -1800 C. ALGORITHM FOR PID CONTROLLER
phase point of the open-loop process Bode curve. DESIGN BY USING BODE PLOT TECHNIQUE
Step - I: Find Transfer Function of Transportation
Lag

Step - II: Find Transfer Function of

Uncompensated System with Transportation Lag
and obtaining the frequency response specifications
such as gain margin, phase margin, gain cross-over
frequency, and phase cross-over frequency.

Step - III: Now applying the Criterion for addition

of Derivative Mode as -
ωTd=1.05747;
Therefore, Td = 1.05747/ω [here ω = wcg =Phase
Crossover freq. of Uncomp. System]
Figure 6: Bode Plot of Derivative Mode
Step - IV: Find Transfer Function of Process +
B. DESIGN SPECIFICATIONS OF INTEGRAL
Derivative Mode and obtaining the frequency
MODE
response specifications such as gain margin, phase
1 1 + Ti s margin, gain crossover frequency, and phase
G PI ( s ) = 1 + =
Ti s Ti s crossover frequency.
1 + jwTi
G PI ( jw) = Step - V: Now applying Criterion for Reset Mode
jwTi Design as
therefore, ωTi =8.97575;
tan −1 ( wTi ) − 90 0 = −6.38 0 Therefore, Integral Time Ti = 8.97575/ω [here w
=wcgpd=Phase Crossover freq. Of Process +D+ I ]
on solving, we get, Step - VI: Find Transfer Function of Process +
wTi = 8.97575 Integral Mode and obtaining the frequency response
Adding the Integral (Reset) Mode with the –6.380 specifications such as gain margin, phase margin,
phase of the reset curve aligned the –173.620 phase gain cross-over frequency, and phase cross-over
point of the process - plus - derivative curve. frequency.

Step - VII:To find the Proportional Gain kc: To

achieve 6 dB gain margin. A Proportional Band is
required to lower the gain curve by 6 dB
i.e.20log10(kc) = 6 – Gain Margin(GM),
[here GM=GMpi]
Therefore, find antilog of [6-(GMpi)/20] and gives
the value of kc, Where,
GMpi = Gain Margin of Process + D + I

Step - VIII: Find Transfer Function of PID

Controller using the proportional gain in step VII .

Figure 7: Bode Plot of Integral Mode

4
 CONTROL INSTRUMENTATION SYSTEM CONFERENCE (CISCON2005)

Step - IX: TF of Overall System i.e. Process + PID Time Response Specifications of Uncompensated
Controller, this is obtained by convoluting Process:
numerator/denominator of uncompensated and Peak time = 1.45875
numerator/denominator of PID as obtained in step Percent overshoot = 14.5153
VIII, and obtaining the frequency response Rise time = 0.558132
specifications such as gain margin, phase margin, Settling time = 2.79066
gain cross-over frequency, and phase cross-over Time response Specifications of Compensated
frequency. Process:
Peak time = 0.767114
IV. RESULT ANALYSIS Percent overshoot = 20.9245
Rise time = 0.260039
The PI controller parameters for Equation 4 found Settling time = 1.69025
from the algorithm7 are as follows
Ti = 2.6199
Kc = 0.28706
The PID controller parameters for Equation 6 are as
follows:
Td = 0.24995
Ti = 1.2501
Kc = 1.9573
The process under consideration described by
Equation 5, consists of series of first order
elements, therefore, the critical frequency would
increased 1.5 fold to three folds by having the
controller contribute 460 phase lead at the critical
frequency. This corresponds to wTd =1. The
Figure 9: Bode Plot of Uncompensated System
frequency response is shown in Figure 6. The
Frequency Response Specifications of
integral mode parameters are decided by the gain.
Uncompensated System
Increasing the nominal gain Kc displaces the gain
Gain Margin = 11.574
curves upward and does not affect the phase angle,
Phase Margin = 97.199
so presenting only the curve for Kc = 1 would have
been sufficient. The frequency response for any Phase Crossover Frequency = 4.2309
values of gain and reset time is to use the curve Gain Crossover Frequency = 1.1717
shown in Figure 7 and set wTi = 1.

Figure 10: Bode Plot of Compensated System

Figure 8: Step Response of Uncompensated &
Compensated Process

5
 CONTROL INSTRUMENTATION SYSTEM CONFERENCE (CISCON2005)

Frequency Response Specifications of [7] B. Shahian., M. Hassul., “Control Systems

Compensated Process: Design Using MATLAB”, Prentice Hall, Englewood
Gain Margin of Process = 6 dB Cliffs, New Jersey, 1993.
Phase Margin of Process = 51.60 [8] D. N. Pawar, “Design of PID Controller for
Phase Crossover Frequency = 6.73 System with Transportation Lag”, National
Gain Crossover Frequency = 3.15 Conference (SPCCN01 – 2005), VIT, Pune, Tata
7 Algorithm steps 1, 2, 3, and 5 of the Section III – McGraw – Hill, New Delhi, pp. 219 – 224.
C are to be used with required specifications of the
problem.
Authors Profile: Mr. D. N. Pawar is an Assistant Professor in
Instrumentation Department of BVCOE, Navi Mumbai. He
V. CONCLUSION has passed his B. E. (Electronics) and M. E. (Electrical –
Control Systems) from Shivaji University, Kolhapur. He is
basically an academician and has 14 years of teaching
The frequency response method presented in this experience. He has published Technical papers in National
paper is found simple tuning method for the Conferences and also has to his credit books that he wrote on
integrating and unstable processes considered in “Electronic Measurement and Instruments” for Second Year
this paper (Equations 3 and 5). The algorithm Diploma Course in Instrumentation. He has reviewed a book
developed for these problems with suggestions on Microcontrollers published by Tata McGraw – Hill, New
Delhi. He has been a Life Member of ISTE and presently
given in [4], is applied to find the parameters of PI acted as a Secretary of ISTE Chapter of BVCOE, Navi
and PID controller respectively. The tuning method Mumbai. He is Member of Syllabus making committee for B.
gave good responses for integrating process. E. Instrumentation of University of Mumbai. He has attended
many meetings, conferences, STTP, Workshops, Training, etc.
ACKNOWLEDGMENT at the college and university level. He also acted as judge for
evaluation of Project presentation contest at BVIT, Navi
Mumbai.
The Author wishes to thank Management Council
of Bharati Vidyapeeth, Pune for permitting to send
this paper and Department of Instrumentation,
BVCOE, Navi Mumbai for providing computer
facilities for the preparation of this paper.

REFERENCES:

[1] Ė. Poulin, A. Pomerleau, “PID tuning for

integrating and unstable processes”, IEE Proc. –
Control Theory Appli. Vol. 143, No.5, September
1996. pp.429 – 435.
[2] Karl J. Åström, Tore Hägglund, “PID Control”,
Control Handbook, CRC Press – 1996, pp. 198 –
209.
[3] Tore Hägglund, Karl Åström, “Automatic
Tuning of PID Controllers”, Control Handbook,
CRC Press – 1996, pp. 817 – 826.
[4] Eronin Umez – Eronin, “System Dynamics &
Control”, Thomson Books/Cole, 1999.
[5] B. Friedland, “Advanced Control System
Design”, Prentice Hall International Editions,1996.
[6] I. J. Nagrath, M. Gopal, “Control Systems
Engineering”, 2nd Edition, Wiley Eastern Ltd.,
1982.

6
View publication stats