Documentos de Académico
Documentos de Profesional
Documentos de Cultura
Abstract
Proportional, Integral and derivative (PID) controllers are the most widely-used controller
in the chemical process industries because of their simplicity, robustness and successful
practical application. Many tuning methods have been proposed for PID controllers. Our
purpose in this study is comparison of these tuning methods for single input single output
(SISO) systems using computer simulation. Integral of the absolute value of the error
(IAE) has been used as the criterion for comparison. These tuning methods have been
implemented for first, second and third order systems with dead time and for two cases of
set point tracking and load rejection.
Key Words: PID Controller; Tuning Method; Set Point Tracking; Load Rejection
Introduction:
During the 1930s three mode controllers
with proportional, integral, and derivative
(PID) actions became commercially
available and gained widespread industrial
acceptance. These types of controllers are
still the most widely used controllers in
process industries. This succeed is a result
of many good features of this algorithm
such as simplicity, robustness and wide
applicability. Many various tuning
methods have been proposed from 1942 up
to now for gaining better and more
acceptable control system response based
on our desirable control objectives such as
percent of overshoot, integral of absolute
value of the error (IAE), settling time,
manipulated variable behavior and etc.
Some of these tuning methods have
considered only one of these objectives as
a criterion for their tuning algorithm and
some of them have developed their
algorithm by considering more than one of
the mentioned criterion. In this study we
-Ziegler-Nichols method
-Modified Ziegler-Nichols method
-Tyreus-Luyben method
-Damped oscillation method
Open loop methods are:
-Open loop Ziegler-Nichols method
-C-H-R method
-Cohen and Coon method
-Fertik method
-Ciancone-Marline method
-IMC method
-Minimum error criteria (IAE, ISE, ITAE)
method
Before proceeding with a brief discussion
of these methods it is important to note that
the non-interacting PID controller transfer
function is:
(1)
G c (s) = k c (1 + t I /s + t D .s)
Where kc= proportional gain
I= Integral time
D= derivative time
I
Pu/1.2
Pu/2
D
Pu/8
4
s +
Pu
(2)
G c (s) = 0.75k cu .Pu .
s
Thus the PID controller has a pole at the
origin and double zeros at s =-4/Pu.
The advantage of Z-N method is that it
does not require the process model.
0.2Kc
Pu/2
Pu/3
kc
kcu/3.2
kcu/3.2
I
2.2Pu
2.2Pu
d
(4)
G c ( s ) = 0.6 m
Km
S
Thus the PID Controller has a pole at the
D
Pu/6.3
1
d
PID
1.2 pn
.
Km d
2d
0.5d
Fertick Method:
This method uses a first order plus dead
time model for the process:
ke ds
(7)
Gm ( s) =
s + 1
then the Fertik controllability F , must be
calculated as:
T
d
(8)
F =
= d
d + T ps
Td = d
(9)
T ps = d +
Cohen-Coon Method:
In this method the process reaction curve
is obtained first, by an open loop test as
shown in Figure 1, and then the process
dynamics is approximated by a first order
plus dead time model, with following
parameters:
3
(5)
m = (t 2 t1 )
2
(6)
dm = 2 m
where
t1 = time at which C =0.283 C s
t1 = time at which C =0.632 C s
C = the plant output.
This method that proposed by Dr C. L.
Smith [15] provides a good approximation
to process reaction curve by first order plus
dead time model
After determining of three parameters of
km , m and d, the controller parameters
can be obtained, using Cohen-Coon [14]
relations given in Table 8. These relations
were developed empirically to provide
closed loop response with a decay ratio.
IAE= | e( t ) |.dt
0
(12)
ISE= e 2 ( t ).dt
0
(13)
ITAE= t | e( t ) | dt
0
(14)
ITSE= t .e 2 ( t ).dt
0
(15)
Simulation Study:
For simulation purpose the following
systems have been considered:
Gp(s) =
e 0.2 s
0.5s + 1
(16)
system
Input
Type
Set point
Changes
e 0.5 s
Gp(s) = 2
(17)
s + 1.8s + 1
2 e s
Gp(s) =
(18)
( 2 s + 1.) ( s + 1) 2
As can be seen, the second order system is
an under damped system. The simulation is
carried out, using MATLAB (version 6.1)
Load
Disturbances
First
Order
Second
Order
Third
Order
Damped
Oscillation
Modified
Z-N
(Some
Overshoot)
ISE
Z-N
(Closed
Loop)
ISE
Z-N
(Closed
Loop)
References:
1) Astrom K,J, T. Hagllund; PID controllers
Theory, Design and Tuning ,2nd edition,
Instrument Society of America,1994
2) Chen C.L., A Simple Method for Online
Identification and Controller Tuning, AIChe J
,35,2037 (1989)
3) Coughanowr D.R.; Process System
Analysis and Control,2nd edition McGrawHill,1991
4) Erickson K.T., J.L. Hedrick ; Plantwide
Process Control John Wiley & Sons,1999
5) Hang C.C., J.K. Astrom, W.K. Ho; Refine-
PID
0%
Kc
0.3
Km
0.6
Km
m
d
m
d
0.95 m
Km d
20%
__
__
4d
__
2.4d
0.42d
Kc
0.7 m
Km d
0.7 m
Km d
1.2 m
Km d
__
__
2.3d
__
2d
0.42d
0%
Kc
20%
0.3 m
Km d
PI
0.35 m
Km d
1.2 m
PID
0.6 m
Km d
Kc
0.7 m
Km d
__
__
0.6 m
Km d
0.5d
0.95 m
Km d
1.4 m
__
__
__
__
0.47d
kc
1 m
d
(1 +
)
Km d
3 m
1 m 9
d
( +
)
K m d 10 12 m
30 + 3d m / m
9 + 20d m / m
-
PD
1 m 5
d
( +
)
K m d 4 6 m
PID
1 m 4
d
( +
)
K m d 3 4 m
32 + 6d / m
13 + 8d / m
6 2d / m
22 + 3d / m
4
11 + 2d / m
G p ( s) =
Controller
type
K.Kc
PID
2 + d
2( + d)
PI
Improved PI
2 + d
2
k
e ds
s + 1
Recommended / d
( > 0.2 always)
d
2 + d
d
2( + d)
>0.25
__
__
>1.7
__
__
>1.7
d
2
d
2
a
Kc = 1
K
d
1 =
a 2
b1
K e ds
s +1
ISE
IAE
ITAE
1.495
1.435
1.357
-0.945
-0.921
-0.947
a2
1.101
0.878
0.842
b2
0.771
0.749
0.738
a3
0.560
0.482
0.381
b3
1.006
1.137
0.995
a1
b1
b2
D = a 3
b3
Table 11-Minimum error integral tuning formulas for set point changes
Process Model: G ( s ) =
Error Integral:
K e ds
s +1
IAE
ITAE
a1
1.086
0.965
b1
-0.869
-0.855
a2
0.740
0.796
b2
-0.130
-0.147
a3
0.348
0.308
b3
0.914
0.9292
a d 1
K c = 1
K
1 =
a 2 + b2 (d / )
d
D = a 3
b3
Table 12 IAE values for various tuning methods, Set Point Tracking
System
No
.
Method
Simulation Time
1
2
3
4
5
6
7
8
9
10
11
12
13
14
First
Order
System 1
Second
Order
System 2
Third
Order
System 3
10
0.47
20
2.25
30
4.26
0.71
1.95
7.53
0.71
1.4
0.44
0.47
0.59
0.51
2.3
4.87
2.13
3.56
2.2
5.06
9.46
7.6
5.7
12.22
9.29
8.77
0.5
2.51
4.28
0.69
0.67
0.72
0.68
0.6
3.39
2.29
2.26
2.12
2.16
11.9
4.36
6.35
5.58
4.41
Method
Simulation Time
Z-N (Closed Loop)
Modified Z-N
(some overshoot)
Modified Z-N
(no overshoot)
Tyreus-Luyben
Damped Oscillation
Fertik
Ciancone
Z-N(Open Loop)
C-H-R (0%
overshoot)
C-H-R (20%
overshoot)
Cohen-Coon
IAE
ISE
ITAE
IMC
First
Order
System
Second
Order
System
Third
Order
System
10
0.35
20
1.64
30
4.08
0.36
1.68
6.71
0.43
1.37
0.26
0.35
0.49
0.4
1.82
4.48
1.15
1.79
1.68
1.62
8.12
13.64
4.39
7.97
13.1
5.56
0.48
1.29
8.34
0.4
0.42
0.29
0.22
0.3
0.6
1.58
1.93
1.28
1.01
1.34
1.98
5.12
4.58
4.18
4.2
4.62
4.15
10
11
1.5
6
5
1
IAE
IAE
0.5
3
2
1
13
11
0
3
Mehtod Number*
11
13
Mehtod Number*
* Refer to Table 12
*Refer to Table 12
1.5
IAE
1
0.5
0
1
11
13 15
* Refer to Table 12
*Refer to Table 13
15
10
IAE
IAE
1
0
0
1
11
13
15
Mehtod Number*
11
13
15
Mehtod Number
*Refer to Table 13
* Refer to Table 13
12