You are on page 1of 6

2013 13th International Conference on Control, Automation and Systems (ICCAS 2013)

Oct. 20-23, 2013 in Kimdaejung Convention Center, Gwangju, Korea

IMC Based PI/D Controller Tuning Rule for Time Delay Processes
Mohammad Shamsuzzoha
Department of Chemical Engineering, King Fahd University of Petroleum and Minerals,
Dhahran, 31261, Saudi Arabia
(Tel : +966-3-860-7360; e-mail: mshams@kfupm.edu.sa)
Abstract: The motivation of this study is to obtain single tuning formula for different type of processes with enhanced
disturbance rejection performance. A simple tuning formula gives the consistently better performance as compared to
well-known SIMC method at the same degree of robustness for stable and integrating process. The performance of the
unstable process has been compared with other recently published method which also shows significant advantage in
the proposed method. For the selection of the closed-loop time constant, (c), a guideline is provided over a broad range
of time-delay/time-constant ratios on the basis of the peak of maximum uncertainty (Ms). Comparison of the IAE has
been conducted for wide range of / ratio for the first order time delay process. The proposed method gives better
performances for wide range of / ratio as compared to SIMC while Lee et al. shows poor disturbance rejection in lag
dominant process.
Keywords: PI/D controller, IMC method , stable process, unstable process

1. INTRODUCTION
Proportional integral (PI) controller has been the most
popular and widely used controllers in the process
industries because of their simplicity, robustness and
wide ranges of applicability with near-optimal
performance.
The stable and integrating processes are very
common in process industries in flow, level and
temperature loop. The open-loop unstable processes are
also encounter in chemical processing units and known
to be difficult to control, especially when there exists a
time delay, such as in the case of continuous stirred tank
reactors, polymerization reactors and bioreactors which
are inherently open-loop unstable by design.
On the basis of a survey of more than 11, 000
controllers in the process industries, Desborough and
Miller [1] reported that more than 97% of the regulatory
controllers utilize the PI algorithm. A recent survey of
Kano and Ogawa [2] shows that the ratios of
applications of different type of controller e.g., PI
control, conventional advanced control and model
predictive control is about 100:10:1. There is no perfect
alternative of the PID controller at least at the bottom
layer in the process industries. This was the clear
conclusion at the end of the IFAC Conference on
Advances in PID Control, held in Brescia (Italy) during
2830 March, 2012. Although the PI controller has only
two adjustable parameters, they are difficult to be tuned
properly in real process.
There are variety of controller tuning approach
reported in the literature and among that two types are
widely used for the controller tuning, one may use
open-loop or closed-loop plant tests. Most tuning
approaches are based on open-loop plant information;

typically the plants gain (k), time constant () and time


delay ().
The effectiveness of the internal model control (IMC)
design principle has made it attractive in the process
industries, where many attempts have been made to
exploit the IMC principle to design PI/D controllers for
both stable [3, 4, 5, 6, 7, 8] and unstable processes [9,
10, 11, 12]. The IMC-PID tuning rule has the advantage
of using only a single tuning parameter (c) to achieve a
clear trade-off between the closed-loop performance and
robustness.
Therefore, in this paper, a simple analytical method is
proposed for the design of the PI/D controller.
Overcoming the drawback of different tuning rules for
different types of processes, only single tuning rule is
capable to handle different types of processes with
performance improvement. A c guideline was
recommended
for
a
wide
range
of
time-delay/time-constant ratios (/). A simulation study
was performed to show the validity of the proposed
method at same Ms-value.

2. THEORY OF IMC-PI CONTROLLER


DESIGN
Figures 1-(a) and 1-(b) show the block diagrams of
the IMC control and equivalent classical feedback
control structures, respectively, where Gp is the process,
G p the process model, q the IMC controller, fr the
set-point filter, and Gc the equivalent feedback
controller.
For the nominal case (i.e., G p G p ), the set-point and
disturbance responses in the IMC control structure can
be simplified as:
y G p qf r r (1 G p q )G p d

(1)

According to the IMC parameterization [7], the


process model G p is factored into two parts:

G p p m p A

(2)

where pm is the portion of the model inverted by the


controller, pA is the portion of the model not inverted by
the controller and pA(0)=1. The noninvertible part
usually includes the dead time and/or right half plane
zeros and is chosen to be all-pass.
To get a superior response for unstable processes or
stable processes with poles near zero, the IMC
controller q should satisfy following conditions.
If the process Gp has unstable poles or poles near zero at
z1 , z2 ,, zm , then

controller. The ideal feedback controller that is


equivalent to the IMC controller can be expressed in
terms of the internal model G p and the IMC controller q
as:
Gc

q
1 G p q

(8)

Substituting Eqs. (2) and (5) into (8) gives the ideal
feedback controller:
pm1
Gc
1

(i) q should have zeros at z1 , z 2 , , zm

(im1 i s i 1)
( c s 1) r

p A im1 i s i 1

c s 1

(9)

(ii) 1 G p q should also have zeros at z1 , z2 ,, zm


Since the IMC controller q is designed as q pm1 f ,
the first condition is satisfied automatically. The second
condition can be fulfilled by designing the IMC filter (f)
as
f

s z1 , zm

d
u(t)

++

Process
y

Gp

setpoint
~

(3)

where c is an adjustable parameter which controls the


tradeoff between the performance and robustness; r is
selected to be large enough to make the IMC controller
(semi-)proper; i are determined by Eq. (4) to cancel the
poles near zero in Gp.
1 Gp q

IMC
controller

Process model

is 1
( c s 1) r
m
i 1

Disturbance
setpoint
filter
fr
+-

pA ( im1 i s i 1)
( c s 1) r
sz

Gp

-+

(a) The IMC structure

(4)

,, zm

Then, the IMC controller comes to be


q pm1

( im1 i s i 1)
( c s 1) r

(5)

Thus, the resulting setpoint and disturbance


responses are obtained as:
y
G p qf r p A
r

m
si
i 1 i

c s 1

1 G p q G p 1 p A
d

1
r

(6)

fr

1
Gp
c s 1
m
si
i 1 i

(7)

The numerator expression im1 i s i 1 in Eq. (6)


causes an excessive overshoot in the servo response,
which can be eliminated by introducing the setpoint
filter fr to compensate for the overshoot in the servo
response.
From the above design procedure, a stable, closed
loop response can be achieved by using the IMC

(b) Feedback control structure


Figure: 1 Block diagram of the IMC and classical
feedback control systems.
The resulting controller in Eq. (9) is physically
realizable but it does not have the standard PI/D form.
The desired form of the controller can be obtained by
using the proper approximation of the dead time term
for example Taylor series expansion. In this study, both
simplicity and approximation error due to dead time
term has been considered carefully during the PI/D
controller design.

3. CONTROLLER DESIGN OF FOPDT


PROCESS
First Order Plus Dead Time (FOPDT) process is
representative model and commonly used in the
chemical process industries.

4. SIMULATION STUDY

Ke - s
s 1

Gp

(10)

where K is the process gain, the time constant, and is


the time delay, the IMC filter structure selected is
f

s 1

c s 1

(11)

After utilizing the above design principle the ideal


feedback controller is given as
Gc

s 1 s 1
2
K c s 1 - e- s s 1

(12)

From Eq.(12), the resulting PI controller can be


obtained using Taylor series expansion, e s 1 s
and then simplification

Kc

K 2 c

; I

(13)

Furthermore, it is obvious that the remaining part of


the denominator in Eq.(12) contains the factor of the
process poles s 1 . It has been ignored because of its
little impact on the control performance, while keeping
the simple PI control structure.
The value of is selected so that it cancels out the
pole at s=-1/. From Eq.(4), this requires
1s 1 es s 12
and the value of is
c

s1 0

obtained as

1 1

(14)

The local detuning factor F is included in Eq.(13) and


the final form of the proposed tuning rule is given in
conclusion section.

Setpoint filter to enhance servo response


In the proposed controller design term (s+1) shows a
large overshoot for the step set-point change, it is
because the controller is design based on the disturbance
rejection. Therefore, a setpoint filter f r is suggested
to remove excessive overshoot and enhance the servo
response.

fr

cs 1
s 1

(15)

Tan et al. [10] also suggested this form of set-point


filter. Due to this type of lead-lag filter, resulting
response will be first order with the time constant of c
for the set-point change.

Although the simulation study is conducted on the


different types of processes, only few of them are
discussed below. The proposed tuning rule provides in
all cases acceptable controller settings with respect to
both performance and robustness. All results are without
local detuning (F=1). The closed-loop performance is
evaluated by introducing a unit step change in both the
set-point and load disturbance i.e, (ys=1 and d=1). The
brief overview of the performance and robustness
measure is mentioned here. To evaluate the robustness,
we compute the maximum closed-loop sensitivity,
defined as Ms =max 1/[1+g c(j)] . A small Ms-value
indicates that the control system has a large stability
margin.
To achieve the fair comparisons in the simulation
study, all controller have been tuned by adjusting c for
the same degree of robustness by fixing Ms. In the
simulation of the second order process, series form of
PID structure has been used with N=100 and no
differentiation of the setpoint (Skogestad [5]).
The results of 4 different processes are listed in Table
1. Figures 2-4 show comparison of the proposed method
with other methods like SIMC (Skogestad [5]), DCLR
(Lee et al. [4]), TL (Tyreus and Luyben [13]) and Yang
et al. [11]. In case of stable and integrating process
proposed method gives faster disturbance rejection and
has clear advantage over the DCLR and SIMC method.
The proposed method also works well in first and
second order unstable processes with dead time. The
results of examples C4 clearly shows that the proposed
method gives both smaller overshoot and faster
disturbance rejection while maintaining setpoint
performance for unstable process. From above analysis
it seems that the proposed method constantly gives
better closed-loop response for all types of processes at
same Ms-value compared with other methods.
Note: For the PI-controller design, DIP should be
approximated to the FOPDT process. In the present
simulation case C2 has been modeled as
20e 7.4 s 100 s 1 .
Figure 5 shows the manipulated variable (MV)
response for C1 as the representative case. The response
of the MV of the proposed method is comparable with
the SIMC and Lee et al.

Effect of local detuning factor F: Although the


figure is not shown in this paper the effect of local
detuning factor F has been investigated using a first
order process with time delay (Case C1). As expected,
using F=1.25 (here F>1) results in more robust
controller settings with Ms=1.43. The original setting
which represent F=1 gives Ms=1.60. As mentioned
earlier the closed-loop response can be speed-up by
selecting F=0.75 (here F<1) and then resulting
robustness will be reduced to Ms=2.05.

Effect of setpoint filter on servo response: The


proposed method is based on the disturbance rejection
so the large overshoot for the step setpoint change can
occur particularly for the unstable and integrating
processes. Therefore, lead-lag setpoint filter similar to
Tan et al. [10] is recommended to remove the overshoot
in setpoint response. To show the performance
improvement a first order unstable process with time
delay (Case C3) has been considered. The resulting set
point filter of the proposed study for C3 should be
f r 1.36s 1 8.25s 1 . Figure 6 shows the
closed-loop response of the proposed method for both
with and without setpoint filter where IAE is reduced
from 6.72 to 1.90. As expected the output response with
setpoint filter is fast without any overshoot.

Process Variable

1.25

Process Variable

1.5
1
0.5

(a)

Proposed method
Yang et al.
5

10

Time

15

20

0.5

Proposed method
Yang et al.

0.4

Process Variable

25

0.3
0.2
0.1

(b) 0
0

10

Time

15

20

25

Figure: 4 Responses of PID-control for high order


unstable process G p

e 0.5 s
(C4).
5s 1 2s 1 0.5s 1

For both setpoint and load disturbance of magnitude 1 at


t=0.

1
0.75
0.5

Proposed method
SIMC
Lee et al.

0.25

(a)

0
0

10

20

30

6
35

Proposed method
SIMC
Lee et al.

4
MV

Time
Proposed method
SIMC
Lee et al.

0.15
Process Variable

0
0

0.1

(a)

0.05

0
0

10
Time

0.5

15
Proposed method
SIMC
Lee et al.

(b)
10

Time

20

30

35

Figure: 2 Responses of PI-control of first-order process


e s 10s 1 (C1). For both setpoint and load

MV

0
0

20

-0.5
-1

(b)

disturbance of magnitude 1 at t=0.

-1.5
0

10
Time

15

20

Figure 5: MV plots of PI-control for first-order process


e s 10s 1 (C2). For both setpoint and load

(a)

disturbance of magnitude 1 at t=0.

0.8

Proposed method
SIMC
TL

0.4

0
0

40

80
Time

120

Process Variable

3.5
Proposed method without setpoint-filter (IAE=6.72, TV=11.98)
Proposed method with setpoint-filter (IAE=1.90, TV=2.26)

Proposed method
SIMC
TL

2.5

(b) 0
0

160

Process Variable

Process Variable

1.2

1.5

1
40

80
Time

120

160

Figure: 3 Responses of PI-control of integrating process


G p 0.2e 7.4 s s (C2). For both setpoint and load
disturbance of magnitude 1 at t=0.

0.5

0
0

Time

10

15

Figure 6: Effect of setpoint filter to remove the


overshoot from setpoint response: Setpoint responses of
PI-control of first-order unstable process with time
e 0.5 s
delay G p
(C3). For both setpoint and load
s 1
disturbance of magnitude 1 at t=0.

Table 1: PI/D controller setting for proposed method (F=1) and performance matrix.
Case

Process model

Methods

Resulting PI/PID-controller
c

C1

C2

C3
C4

e s
10 s 1
7.4 s

0.2e
s
0.5 s

e
s 1

0.5 s

5s 1 2s 1 0.5s 1

Ms

Kc

Setpoint
IAE (y)

Load
disturbance
IAE (y)

Proposed

2.46

1.60

4.57

4.85

3.1

1.06

SIMC

1.0

1.60

5.0

8.0

2.5

1.60

Lee et al.

1.0

1.60

5.12

10.25

2.17

2.0

Proposed

19.37 1.70

0.304

39.63

30.21

131.9

SIMC

7.4

1.70

0.338

59.2

28.8

174.5

TL

1.67

0.33

64.7

29.13

195

Proposed

1.36

6.0

1.646

8.25

6.72

5.01

Lee et al.
Proposed

1.4
2.9

6.0
2.2

1.668
3.22

8.67
9.50

6.77
6.37

5.20
2.95

Yang et al.

1.5

2.2

2.564

10.98

8.57

4.28

Note: For case C4: D =2.25 and 1.82 is used for the proposed and Yang et al., respectively.

5. DISCUSSIONS
c guideline for proposed tuning rule

In the proposed tuning rule, the closed-loop time


constant c controls the tradeoff between robustness and
performance of the control system. As c decreases, the
closed-loop response becomes faster and can become
unstable. On the other hand, as c increases, the
closed-loop response becomes sluggish and more stable.
A good tradeoff is obtained by choosing c to give
Ms-value in the range of 1.2~2.0 for stable process. An
analysis of the c selection has been conducted and plot
of Ms verses / for different value of c = , where
=1.0, 2.0 and 3.0 is shown in the Fig. 7. The figure
clearly shows that c = is not the proper choice because
for the lag dominant process it gives tight controller
setting. For c =3, it gives smooth and robust setting
because Ms lies between 1.61 to 1.25. A good tradeoff
between robustness and performance can be achieved
for c =2 where it will give Ms=2.0 for lag dominant
process and Ms=1.26 for delay dominant process.

4
c=
c=2
c=3
3

Ms
2

0
0

0.2

0.4

0.6

0.8

Figure: 7 c guidelines of first order stable process with


time delay based on the time delay in the process.

6. CONCLUSIONS
A simple analytical design method for PI/D controller
was proposed based on the IMC principle in order to
improve disturbance rejection performance. Another
important feature of the proposed methodology is that it
dealt stable, integrating and unstable process in a
unified way. As we have seen earlier a single tuning rule
gives satisfactory performance and robustness for all
cases.
In the resulting method c controls the tradeoff
between robustness and performance of the control
system. It is summarized as:
Kc

K 2 c F

1 1

I F;

D 2

where F is a local detuning factor. F=1 gives the


original proposed setting and to detune the response and
get more robustness one selects F>1. To speed-up the
closed-loop response in special cases where it appears
slow, one may select F<1. For the first order and
integrating process with time delay the resulting tuning
rule is PI where D=0. Several important representative
processes were considered in the simulation study in
order to demonstrate the advantage of the proposed
method. The design method was based on the
disturbance rejection and a setpoint filter was
recommended to eliminate the overshoot in setpoint
response. In particular, the proposed controller shows
excellent performance when the lag time dominates. A
guideline of c was also proposed for a wide range of /
ratio.
Acknowledgement: The author would like to
acknowledge the support provided by King Abdulaziz
City for Science and Technology (KACST) through the
KACST Annual Program at King Fahd University of
Petroleum & Minerals (KFUPM) for funding this work
through project number 613-11.

Works Cited
[1] L. D. Desborough and R. M. Miller, "Increasing
customer value of industrial control performance
monitoringHoneywells
experience,"
in
Chemical Process Control VI AIChE Symposium
Series , Tuscon, Arizona, Jan. 2001, 2002.
[2] M. Kano and M. Ogawa, "The state of art in
chemical process control in Japan: Good practice
and questionnaire survey," Journal of Process
Control, vol. 20, pp. 969-982, 2010.
[3] D. Seborg, T. Edgar and D. Mellichamp, Process
Dynamics and Control, New York: Wiley, 2004.
[4] Y. Lee, S. Park, M. Lee and C. Brosilow, "PID

controller tuning for desired closedloop responses


for SI/SO systems," AIChE, vol. 44, p. 106115,
1998.
[5] S. Skogestad, "Simple analytic rules for model
reduction and PID controller tuning," journal of
Process Control, vol. 13, p. 291309, 2003.
[6] M. Shamsuzzoha and S. Skogestad, "The setpoint
overshoot method: A simple and fast closed-loop
approach for PID tuning," Journal of Process
Control, vol. 20, p. 12201234, 2010.
[7] M. Shamsuzzoha and M. Lee, "IMCPID
controller design for improved disturbance
rejection of timedelayed processes," Ind Eng
Chem Res, vol. 46, p. 20772091, 2007.
[8] D. Rivera, M. Morari and S. Skogestad, "Internal
model control. 4. PID controller design," Ind Eng
Chem Process Des Dev, vol. 25, p. 252265, 1986.
[9] Y. Lee, J. Lee and S. Park, "PID controller tuning
for integrating and unstable processes with time
delay," Chemical Engineering Science , vol. 55, pp.
3481-3493, 2000.
[10] W. Tan, H. Marquez and T. Chen, "IMC design for
unstable processes with time delays," Journal of
Process Control , vol. 13, p. 203213, 2003.
[11] X. Yang, Q. Wang, C. Hang and C. Lin,
"IMC-based control system design for unstable
processes," Industrial and Engineering Chemistry
Research , vol. 41, p. 42884294, 2002.
[12] M. Shamsuzzoha and M. Lee, "Design of advanced
PID controller for enhanced disturbance rejection
of second order process with time delay," AIChE,
vol. 54, pp. 1526-1536, 2008.
[13] B. Tyreus and W. Luyben, "Tuning PI controllers
for integrator/dead time processes," Ind. Eng.
Chem. Res, p. 26252628, 1992.

You might also like