Download as pdf or txt
Download as pdf or txt
You are on page 1of 56

Chapter 4

Feedback Controller Design

1 @VLC-2015 Process Control


Contents
 The standard feedback control algorithms (also called
control laws) are presented in this chapter, with emphasis on
control algorithms that are widely used in the process
industries.
 Proportional-Integral-Derivative (PID) control and on-off
control are predominant types of feedback control.
 Consequently, features and options for PID controllers are
discussed in detail. A digital PID control algorithms are also
mentioned to explain clearly how they are implemented in
real application.

2 @VLC-2015 Process Control


4.1 PID Controller
 Input and Output of a Controller
Error
SP Controller Output
Controller
MV
PV

 Controller decides “what to do” based on the error between


the desired value (SP) and process value (PV)
 Intuitively, if the error is large, make large change in MV
and if the error is small, make small change in MV
(MV) ∝ 𝑲𝒄 (Error)
 The sign of Kc has to be determined to the direction of
reducing error
3 Process Control
@VLC-2015
4.1.1 P Control
 Proportional Control
𝑝 𝑡 = 𝑝 + 𝐾𝑐 𝑒(𝑡) Error = SP - PV (4.1)

Proportional Gain
Steady-state (bias) value
Controller output

 When the PV gets larger, controller should increase the p(t)


in order to reduce the error, then choose negative Kc value
Direct acting mode (e.g., coolant control)
 When the PV gets smaller, controller should increase the p(t),
then choose positive Kc value
Reverse acting mode (e.g., flow control)
4 Process Control
@VLC-2015
4.1.1 P Control
 Some controllers have a proportional band setting instead of
a controller gain. The proportional band PB (%) is defined
100%
𝑃𝐵 ≜ (4.3)
𝐾𝑐
 The ideal proportional controller Eq. (4.1) does not include
physical limits on the controller output. A more realistic
representation is show as follow
𝑝
𝑝𝑚𝑎𝑥
𝑝

𝑝𝑚𝑖𝑛
5 0 𝑒 Process Control
@VLC-2015
4.1.1 P Control
 An inherent disadvantage of proportional only control is that
a steady state error (offset) occurs after a change in SP or
DV
 For control applications where offset can be tolerated,
proportional only control is attractive because of its
simplicity

6 Process Control
@VLC-2015
4.1.2 Integral Control
𝑡
1
𝑝 𝑡 =𝑝+ 𝑒 𝜏 𝑑𝜏 (4.4)
𝜏𝐼 0

Integral time (adjustable parameter)


 Integral control is widely used because it provides an
important practical advantage, the elimination of offset
 The action is not immediate until the integral becomes
significant. Also, the integral mode tends the system to be
more oscillatory, even unstable
 In practice, integral control action is normally used in
conjunction with proportional control as the proportional-
integral (PI) controller
7 Process Control
@VLC-2015
4.1.2 Integral Control
𝑡
1
𝑝 𝑡 = 𝑝 + 𝐾𝑐 𝑒 𝑡 + 𝑒 𝜏 𝑑𝜏 (4.5)
𝜏𝐼 0

 The corresponding transfer function for the PI controller


𝑃′ (𝑠) 1
= 𝐾𝑐 1 + (4.6)
𝐸(𝑠) 𝜏𝐼 𝑠
 Advantages: fast action, eliminating offset
 Disadvantages: Oscillatory, more parameter to tune
 Reset Windup
Anti-Windup

8 Process Control
@VLC-2015
4.1.2 Derivative Control
𝑑(𝑒)
𝑝 𝑡 = 𝑝 + 𝜏𝐷 (4.7)
𝑑𝑡
Derivative time (adjustable parameter)
 By providing anticipatory control action, the derivative
mode tends to stabilize the controlled process
 Derivative control action also tends to improve the dynamic
response of the controlled variable by decreasing the
process settling time
 The most disadvantage of derivative control is the
sensitivity of the control calculations to high-frequency
noise in the measurement. Therefore, it is usually used with
the derivative filter
9 Process Control
@VLC-2015
4.1.2 Derivative Control
 In practice, derivative control is never used alone; it is
always used in conjunction with proportional or
proportional-integral control
 For example, an ideal PD controller has the transfer function
𝑃′ (𝑠)
= 𝐾𝑐 1 + 𝜏𝐷 𝑠 (4.8)
𝐸(𝑠)
 Then, the PD controller with derivative filter
𝑃′ (𝑠) 𝜏𝐷 𝑠 (4.9)
= 𝐾𝑐 1 +
𝐸(𝑠) 𝛼𝜏𝐷 𝑠 + 1
where, the constant 𝛼 typically has a value between 0.05 and
0.2, with 0.1 being a common choice.
10 Process Control
@VLC-2015
4.1.3 PID Control
Now we consider the combination of the proportional, integral
and derivative control modes as a PID controller.
Unfortunately, many variations of PID control are used in
practice. Next, we consider the three most common forms
 Parallel Form of PID Control
1 𝑡 𝑑𝑒(𝑡)
𝑝 𝑡 = 𝑝 + 𝐾𝑐 𝑒 𝑡 + 𝑒 𝜏 𝑑𝜏 + 𝜏𝐷 (4.10)
𝜏𝐼 0 𝑑𝑡
 The corresponding transfer function
I
𝑃′ (𝑠) 1 𝐸(𝑠) 𝑃′ (𝑠)
= 𝐾𝑐 1 + + 𝜏𝐷 𝑠 1
𝐾𝑐
𝐸(𝑠) 𝜏𝐼 𝑠 𝜏𝐼 𝑠

(4.11) 𝜏𝐷 𝑠
11
@VLC-2015
4.1.3 PID Control
 Series Form of PID Control
𝑃′ (𝑠) 1 𝜏𝐷 𝑠 + 1 (4.12)
= 𝐾𝑐 1 +
𝐸(𝑠) 𝜏𝐼 𝑠 𝛼𝜏𝐷 𝑠 + 1

𝐸(𝑠) 1 𝑃′ (𝑠)
𝐾𝑐 1 +𝜏 𝑠 𝜏𝐷 𝑠 + 1
𝐼

 Expanded Form of PID Control


𝑡
𝑑𝑒(𝑡) (4.13)
𝑝 𝑡 = 𝑝 + 𝐾𝑐 𝑒 𝑡 + 𝐾𝐼 𝑒 𝜏 𝑑𝜏 + 𝐾𝐷
0 𝑑𝑡

 Elimination of Derivative and Proportional Kick


12
@VLC-2015
4.1.3 PID Control
 Elimination of Derivative and Proportional Kick
 One disadvantage of the previous PID controllers is that a
sudden change in set point will cause the derivative term
momentarily to become very large and thus provide a
derivative kick to the final control element.
 Eq. (4.10) can be rewritten to eliminate Derivative kick
1 𝑡 𝑑𝑦𝑚 (𝑡)
𝑝 𝑡 = 𝑝 + 𝐾𝑐 𝑒 𝑡 + 𝑒 𝜏 𝑑𝜏 − 𝜏𝐷 (4.14)
𝜏𝐼 0 𝑑𝑡
 This method of eliminating derivative kick is a standard
feature in most commercial controllers. For a series form
PID controller, it can be implemented quite easily by
13 placing the PD element in the feedback path
@VLC-2015
4.1.3 PID Control

𝑌𝑠𝑝 (𝑠) 𝐸(𝑠) 𝜏𝐼 𝑠 + 1 𝑃′ (𝑠)


𝐾𝑐
𝜏𝐼 𝑠 Block diagram of the
series form of PID
𝜏𝐷 𝑠 + 1 control that eliminate
derivative kick
𝑌𝑚 (𝑠)

 A more flexible PID control algorithm can be obtained by


weighting the set point in the proportional term, as well as
in the derivative term. This modification eliminates the
proportional kick that also occurs after a step change in set
point
1 𝑡 𝑑𝑒𝐷 (𝑡)
𝑝 𝑡 = 𝑝 + 𝐾𝑐 𝑒𝑃 𝑡 + 𝑒 𝜏 𝑑𝜏 + 𝜏𝐷 (4.15)
14 𝜏𝐼 0 𝑑𝑡
@VLC-2015
4.1.3 PID Control
Where, 𝑒𝑃 𝑡 ≜ 𝛽𝑦𝑠𝑝 𝑡 − 𝑦𝑚 𝑡
e 𝑡 ≜ 𝑦𝑠𝑝 𝑡 − 𝑦𝑚 𝑡 (4.16)
𝑒𝐷 𝑡 ≜ 𝛾𝑦𝑠𝑝 𝑡 − 𝑦𝑚 𝑡

15
@VLC-2015
4.2 Closed-Loop Control
4.2.1 Closed Loop Transfer Function
D

𝐺𝐷

SP PV
𝐺𝐶 𝐺𝑃

Where, GC: Controller transfer function


GP: Plant transfer function
GD: transfer function of disturbance
N: measurement noise
PV: Process value
SP: Set point (desired value)
16
@VLC-2015
4.2.1 Closed-Loop Transfer Function
PV  GP GC [ SP  PV  N )]  GD D
[1  GP GC ]PV  GPGC [ SP  N ]  GD D
GP GC GG GD
 PV ( s )  SP  P C N  D (4.17)
1  GP GC 1  GP GC 1  GPGC
Example:
w1,x1 w2, x2=1

The approximate dynamic


pt
model of this process was
I/P developed in Eq. (2.33)
p
x xm
AT AC
w, x sp
17
@VLC-2015
4.2.1 Closed-Loop Transfer Function
w1s 1
X (s) X 1 (s)  W2 ( s )
V  s  w1s V  s  w1s
1 Kp
G p1  , Gp2 
 s 1  s 1
 Process Block Diagram
Mass
X 1 (s) fraction
Mass G p1
(0÷1) (0÷1)
fraction
(0÷1) X ( s )
Mass Mass
W2 ( s ) fraction
Gp2 fraction
[Kg/s] (wmin ÷ wmax) (0÷1)

18
@VLC-2015
4.2.1 Closed-Loop Transfer Function
 Sensor Transmitter (AT)
The dynamic model of the sensor transmitter can be
approximated by a first order transfer function:
X m (s) Km X (s) X m (s)
Gm ( s )   Gm [mA]
X ( s)  m s  1 Mass fraction
(0÷1) (4  20mA)

The time constant of the instrument is much smaller than


that of the whole system, it is useful to set 𝜏𝑚 = 0
 Controller (AC)
Suppose that a proportional integral controller is used.
P( s) 1
Gc ( s )   K c (1  )
E (s) Is
19 e(t )  sp (t )  xm (t )
@VLC-2015
4.2.1 Closed-Loop Transfer Function
SP ( s ) SP( s ) E (s) P( s)
Km [mA] [mA]
Gc [mA]
Mass fraction
(0÷1) (4  20) (4  20)
X m (s)
mA
(4  20)
 I/P Transducer (I/P)

Pt ( s ) P( s) Pt ( s )
 K IP [mA] K IP
P( s) [ psi / bar ]
(4  20) ( pmin  pmax )

 Control valve
Pt ( s ) W2 ( s )
W2 ( s ) Kv Gv
Gv ( s )  
Pt ( s )  v s  1
[ psi / bar ] [kg / s ]
( pmin  pmax ) ( wmin  wmax )
20
@VLC-2015
4.2.1 Closed-Loop Transfer Function
X 1 (s)
𝐺𝑝1
[mf]
[mf]
X 1 (s)
Pt ( s ) Pt ( s ) Pt ( s ) W2 ( s) X (s) X 1 (s)
𝐾𝑚 [mA] 𝐺𝑐 𝐾
[mA] 𝐼𝑃 [psi] 𝐺𝑣 [kg/s] 𝐺𝑝2
[mass [mA] [mf] [mf]
fraction]
E( s )
𝐺𝑚
From Eq. (4.17) we have the closed-loop transfer function:

X sp (s) (4.18)

The above block diagrams considered so far have been


specially developed for the stirred-tank blending system. The
21 more general block diagram will be as follows:
4.2.1 Closed-Loop Transfer Function
Y
𝐺𝑑
E
P U D D Y E Y
𝐾𝑚 𝐺𝑐 𝐺𝑣 𝐺𝑝

P 𝐺𝑚

K mGc Gv G p Gd
Y Ysp  D (4.19)
1  Gc Gv G p Gm 1  GcGvG p Gm
 Set-Point Changes: servomechanism problem
We assume for this case that no disturbance change occurs and
thus D = 0. The closed-loop transfer function:
22
VLC-2015
4.2.1 Closed-Loop Transfer Function
Y( s ) K mGc Gv G p
G1 ( s )   (4.20)
Ysp ( s ) 1  Gc Gv G p Gm
 Disturbance Changes: regulator problem
Y( s ) Gd
G2 ( s )   (4.21)
D( s ) 1  Gc Gv G p Gm
Both closed-loop transfer functions (4.20) & (4.21) have the
same denominator
A(s)  1  Gc Gv G p Gm  1  GOL ( s ) (4.22)
where, 𝐺𝑂𝐿 (𝑠): open-loop transfer function
Then, we have the general transfer function:
K mGc Gv G p Gd
Y  Ysp  D (4.23)
23
VLC-2015
1  GOL 1  GOL
4.2.1 Closed-Loop Transfer Function
Thus, the response to the simultaneous disturbance variable
and set point changes is merely the sum of individual
responses. This result is the consequence of the Superposition
Principle for linear systems.
 General expression for Feedback control systems
Z f
 (4.24)
Zi 1   e
where Z, Zi: output and input variable

𝑓 :product of the transfer functions in the forward


path from Zi to Z;

𝑒 : product of every transfer function in the feedback


24
@VLC-2015 loop.
4.2.1 Closed-Loop Transfer Function
 Proportional Control and Set Point changes 𝐷=0
Using the general transfer function, Eq. (4.23), with these
transfer functions of the components:
Gc ( s )  K c
Gm ( s )  K m (4.25)

Gv ( s )  K v
Then, Y (s) K m K c K v K p ( s  1)

Ysp ( s ) 1  K c K v K m K p ( s  1)

Y (s) K1
  (4.26)
Ysp ( s )  1s  1
25 @VLC-2015 Process Control
4.2.1 Closed-Loop Transfer Function
𝐾𝑂𝐿 𝜏
where, 𝐾1 = ;𝜏 = 𝜏1 < 𝜏: faster response
𝐾𝑂𝐿 +1 1 𝐾𝑂𝐿 +1

and the open loop gain: 𝐾𝑂𝐿 = 𝐾𝑚 𝐾𝑐 𝐾𝑣 𝐾𝑝


 Y (t )  K1Ysp (1  e  t /1 ) (4.27)

Note that a steady state error (offset) exits because the


new steady state value is 𝐾1 ∆𝑌𝑠𝑝 instead of the desired
value of ∆𝑌𝑠𝑝
The offset is defined as:
offset  Ysp  Y ()
Ysp
offset  Ysp  K1Ysp  (4.28)
K OL  1
26 @VLC-2015 Process Control
4.2.1 Closed-Loop Transfer Function
 Proportional Control and Disturbance Changes
Y (s) K d / ( s  1)

Yd ( s ) 1  K OL / ( s  1)
Y (s) K2
  (4.29)
Yd ( s )  1s  1
𝐾𝑑
where, 𝐾2 =
𝐾𝑂𝐿 +1

The closed loop response to a step change in disturbance of


magnitude ∆𝑌𝑑 is given by
 Y (t )  K 2 Yd (1  e  t /1 )
K Y
offset  0  K 2 Yd   d d (4.30)
27 @VLC-2015 K OL  1 Process Control
4.2.1 Closed-Loop Transfer Function
 PI Control and Disturbance Changes
1
For PI control, 𝐺𝑐 𝑠 = 𝐾𝑐 (1 + ). The closed loop transfer
𝜏𝐼 𝑠
function for disturbance changes can then be derived from
(4.23)
Y (s) K d / ( s  1)

Yd ( s ) 1  K (1  1 ) / ( s  1)
OL
Is
Y (s) K d I s
 
Yd ( s )  I s ( s  1)  K OL ( I s  1)

Y (s) K3 s
  2 2 (4.31)
Yd ( s )  3 s  23 3 s  1
28 @VLC-2015 Process Control
4.2.1 Closed-Loop Transfer Function
𝐾𝑑 𝜏𝐼 𝜏𝐼 𝜏
where, 𝐾3 = ; 𝜏3 =
𝐾𝑂𝐿 𝐾𝑂𝐿
(4.32)
1  1  K OL  I
3   
2  K OL  

∆𝑌𝑑
For step change in disturbance, 𝑌𝑑 𝑠 = , Eq. (4.31)
𝑠
becomes:
K 3 Yd
 Y (s)  2 2 (4.33)
 3 s  23 3 s  1
It is obvious from Eq. (4.33) that 𝑌 ∞ = 0. Thus, the
addition of integral action eliminates offset for a step change
in disturbance
29 Process Control
@VLC-2015
4.2.2 Stability of closed loop control systems
An important consequence of feedback control is that it can
cause oscillatory responses
This figure illustrates the effect
of controller gains on closed-
loop response to a unit step
change in set point.

Definition of Stability: An unconstrained linear systems is


said to be stable if the output response is bounded for all
bounded inputs. Otherwise it is said to be unstable
30 Routh Stability Criterion Process Control
@VLC-2015
4.2.2 Stability of closed loop control systems
Example 1: Consider a feedback control system with the
following transfer function: 1
1 
Gc ( s)  K c Gv ( s )  G ( s )
s 1
m
2s  1
1
G p ( s )  Gd ( s ) 
5s  1
Find the values of controller gain 𝐾𝑐 that make the feedback
control system stable?
Example 2: Consider a feedback control system with the
following transfer function:
4e  s
Gc ( s)  K c Gv ( s )  2 Gm ( s )  0.25 G p ( s ) 
5s  1
Find the values of controller gain 𝐾𝑐 that make the feedback
control system stable?
31 Process Control
@VLC-2015
4.3 PID Controller Design
 Performance criteria for closed-loop systems
Ideally, we would like the closed-loop system to satisfy the
following performance criteria:
1. The closed-loop system must be stable

2. The effects of disturbances are minimized, providing good


disturbance rejection
3. Rapid, smooth responses to set point changes are obtained,
that is, good set-point tracking.
4. Steady state error (offset) is eliminated.
5. Excessive control action is avoided.

6. The control system is robust.


32 Process Control
@VLC-2015
4.3 PID Controller Design

33 Process
@VLC-2015 Control
4.3 PID Controller Design
PID controller settings can be determined by a number of
alternative techniques:
1. Direct Synthesis (DS) method

2. Internal Model Control (IMC)

3. Controller tuning relations

4. Frequency response techniques


5. Computer simulation

6. On-line controller tuning

34 Process
@VLC-2015 Control
4.3 PID Controller Design
4.3.1 Direct Synthesis Method
In the DS method, the controller design is based on a process
model and a desired closed-loop transfer function.
The closed-loop transfer function for set-point changes is
derived from Eq. (4.20)
Y K mGc Gv G p

Ysp 1  Gc Gv G p Gm
For simplicity, let 𝐺 = 𝐺𝑣 𝐺𝑝 𝐺𝑚 and assume that 𝐺𝑚 = 𝐾𝑚
Y Gc G
  (4.34)
Ysp 1  Gc G
not known
1  Y Ysp  (4.35)
 Gc    a priori
35 G  1 Y Y  Process
unknown  sp  Control
4.3 PID Controller Design
A practical design equation can be derived by replacing the
unknown 𝐺 by 𝐺 , and 𝑌 𝑌𝑠𝑝 by a desired closed-loop transfer
function 𝑌 𝑌𝑠𝑝 𝒅
1  Y Ysp d 
 Gc    (4.36)
G 1  Y Ysp  
 d 

Ideally, we would like to have 𝑌 𝑌𝑠𝑝 𝑑 = 1 so that the


controlled variable (CV) tracks set-point changes
instantaneously without any error perfect control. That is
impossible
For process without time delays, the first-order model is a
more reasonable choice:
 Y  1
   (4.37)
36
@VLC-2015 Y
 sp  d  c s  1 Process
Control
4.3 PID Controller Design
The controller design equation becomes:
1 1
 Gc  (4.38)
G cs
If the process transfer function contains a known time delay 𝜃,
a reasonable choice for the desired closed loop transfer
function is:  Y  e  s
   (4.39)
 Ysp  d  c s  1
1 e  s
 Gc  (4.40)
G cs 1 e  s

Applying Taylor series expansion:


1 e  s
 Gc  (4.41) Process
37
@VLC-2015 G ( c   ) s Control
4.3 PID Controller Design
 FOPTD Model
Ke  s
G (s)  (4.42)
 s 1
Replacing Eq. (4.42) into Eq. (4.41), we have:
  1 
 Gc   
 s 
1 (4.43)
K ( c   ) 
Compare Eq. (4.43) with the transfer function of PI controller,
Eq. (4.6), we have the following controller settings

Kc 
K ( c   ) (4.44)
I 
38 Process
@VLC-2015 Control
4.3 PID Controller Design
 SOPTD Model
Ke  s
G (s)  (4.45)
( 1s  1)( 2 s  1)

Substitution into Eq. (4.41) and rearrangement gives a PID


controller in parallel form
 1  (4.46)
Gc  K c  1   Ds 
 Is 
where, 1 1   2
Kc 
K c 
 I  1   2
(4.47)

D  1 2
39
@VLC-2015
1   2 Process
Control
4.3 PID Controller Design
Homework: Use DS design method to calculate PID
controller settings for the process
2e  s
G
(10 s  1)(5s  1)
Consider three values of the desired closed loop time constant:
𝜏𝑐 = 1, 3, 10. Evaluate the controllers for unit step changes in
both the set point and the disturbance, assuming that 𝐺𝑑 = 𝐺.
Repeat the evaluation for two cases:
a. The process model is perfect 𝐺 = 𝐺
b. The model gain is incorrect 𝐾 = 0.9 instead of the actual
value 𝐾 = 2. 0.9e  s
G Process
40
@VLC-2015 (10 s  1)(5s  1) Control
4.3 PID Controller Design
4.3.2 Internal Model Control (IMC)
A more comprehensive model-based design method,
Internal Model Control (IMC), was developed by Morari
and coworkers (Garcia and Morari, 1982; Rivera et al.,
1986)
The IMC method, like the DS method, is based on an
assumed process model and leads to analytical
expressions for the controller settings.
These two design methods are closely related and
produce identical controllers if the design parameters are
specified in a consistent manner.
41 Process
@VLC-2015 Control
4.3 PID Controller Design
4.3.2 Internal Model Control (IMC)
The IMC method is based on the simplified block
diagram shown in the figure. A process model 𝑮 and the
controller output U are used to calculate the model
response, 𝒀
The block diagrams for the conventional feedback control
and IMC are compared in the below figures
D
𝒀𝒔𝒑 E U Y
𝑮𝑪 G

42
Classical feedback controller Process
@VLC-2015 Control
4.3 PID Controller Design
4.3.2 Internal Model Control (IMC)
IMC controller D
𝒀𝒔𝒑 E U Y
𝑮𝑪 G

𝒀
𝑮
𝒀−𝒀

In general, 𝐘 ≠ 𝒀 due to modeling errors (𝐆 ≠ 𝑮 ) and


unknown disturbances (𝑫 ≠ 0) that are not accounted for
in the model
It can be proved that the two block diagrams are identical
if the controller 𝑮𝒄 and 𝑮𝒄 satisfy the relation
43 Process
@VLC-2015 Control
4.3 PID Controller Design
4.3.2 Internal Model Control (IMC)
Gc
Gc  (4.48)
1  Gc G
Thus, any IMC controller 𝑮𝒄 is equivalent to a standard
feedback controller 𝑮𝒄 , and vice versa.
The following closed-loop relation for IMC can be
derived from the above block diagram:
Gc G 1  GcG
Y Ysp  D (4.49)
1  Gc (G  G ) 1  Gc (G  G )
For the special case of a perfect model, 𝑮 = 𝑮, eq. (4.49)
reduces to Process
44
@VLC-2015 Control
4.3 PID Controller Design
4.3.2 Internal Model Control (IMC)

Y  GcGYsp  (1  GcG ) D (4.50)


The IMC controller is designed in two steps:
Step 1. The process model is factored as:
G  G G (4.51)

Where 𝑮+ contains any time delays and right-half plane


zeros

In addition, 𝑮+ is required to have a steady-state gain


equal to one.
45 Process
@VLC-2015 Control
4.3 PID Controller Design
4.3.2 Internal Model Control (IMC)
Step 2. The controller is specified as
1 (4.52)
Gc  f
G
Where 𝒇 is a low-pass filter with a steady-state gain of
one. It typically has the form:
1
f  (4.53)
 c s  1
r

𝝉𝒄 is the desired closed-loop time constant (DS method).


The usual choice of r is one.

46 Process
@VLC-2015 Control
4.3 PID Controller Design
4.3.2 Internal Model Control (IMC)
For the ideal situation where process model is perfect
(𝑮 = 𝑮) , substituting eq. (4.52) into (4.50) give the
closed loop expression
Y  G fYsp  (1  fG ) D (4.54)

Thus, the closed-loop transfer function for SP changes is


Y
 G f (4.55)
Ysp
Example:
Use the IMC method to design two controllers for the
FOPTD model. Consider two approximations for time
delay term:
47 Process
@VLC-2015 Control
4.3 PID Controller Design
4.3.2 Internal Model Control (IMC)
a. Taylor series approximation
b. Pade 1/1 approximation
Solution:
a. Taylor approximation (first-order)
e  s 1 s
Ke  s K (1   s )
G (s)  (4.56)
 s 1  s 1
Factor this model as 𝑮 = 𝑮+ 𝑮− , where
K
G  1   s G  (4.57)Process
48
@VLC-2015  s 1 Control
4.3 PID Controller Design
4.3.2 Internal Model Control (IMC)
Setting 𝑟 = 1, and using eq. (4.52) gives
 s 1
Gc ( s )  (4.58)
K ( c s  1)
The equivalent controller 𝑮𝒄 can be obtained from (4.48)
 s 1
Gc ( s ) 
K ( c   ) s
  1 (4.59)
Gc ( s )   1  
K ( c   )   s 
Eq. (4.59) is equivalent to the PI controller with

Kc  I  (4.60)Process
49
@VLC-2015 K ( c   ) Control
4.3 PID Controller Design
4.3.2 Internal Model Control (IMC)
b. Pade 1/1 approximation
 s 1  0.5 s
e
1  0.5 s
Ke  s K (1  0.5 s )
G (s)  (4.61)
 s 1 ( s  1)(1  0.5 s )
Factor this model as 𝑮 = 𝑮+ 𝑮− , where

G  1  0.5 s (4.62)

K
G  (4.63)
( s  1)(1  0.5 s)
50 Process
@VLC-2015 Control
4.3 PID Controller Design
4.3.2 Internal Model Control (IMC)
Setting 𝑟 = 1, and using eq. (4.52) gives
(1  0.5 s )( s  1) (4.64)
Gc ( s ) 
K ( c s  1)
The equivalent controller 𝑮𝒄 can be obtained from (4.48)
(1  0.5 s )( s  1)
Gc ( s )  (4.65)
K ( c  0.5 ) s
Eq. (4.65) is equivalent to the PID controller with
1 2    
Kc  I   D  (4.66)
K 2 c   2 2  
51 Process
@VLC-2015 Control
4.3 PID Controller Design
4.3.2 Internal Model Control (IMC)
 Selection of 𝝉𝒄
 The choice of design parameter 𝝉𝒄 is a key decision in
both the DS and IMC design methods
 In general, increasing 𝝉𝒄 procedures a more
conservative controller because 𝐾𝑐 decreases while 𝜏𝐼
increases.
 Several IMC guidelines for 𝝉𝒄 have been published for
the FOPTD .

1. c  0.8 and  c  0.1 (Rivera et al., 1986)

2.    c   (Chien and Fruehauf, 1990)

52
3.  c   (Skogestad, 2003) Process
@VLC-2015 Control
4.3 PID Controller Design
4.3.3 Online Controller Tuning
We make a few general observations:
1. Controller tuning inevitably involves a tradeoff
between performance and robustness
2. Controller settings do not have to be precisely
determined
3. For most plants, it is not feasible to manually tune
each controller.
4. Diagnostic techniques for monitoring control system
performance are available. Process
53
@VLC-2015 Control
4.3 PID Controller Design
 Continuous Cycling Method (Ziegler Nichols)
Step 1: After process has reached steady state, eliminate the
integral and derivative control action.
Step 2: Set 𝐾𝑐 equal to a small value (eg., 0.5) and place the
controller in the automatic mode.
Step 3: Introduce a small, momentary set point change so that
the CV moves away for the set point. Gradually increase 𝐾𝑐 in
small increments until continuous cycling occurs.
Step 4: Calculate PID controller settings using Ziegler-
Nichols tuning relations or Tyreus-Luyben settings.
Step 5: Evaluate the Z-N controller settings by introducing a
54 Process
small set point change and observe the closed loop response Control
4.3 PID Controller Design

Example: 2e s
G
(10s 1)(5s 1)
Compare PID controllers with the following settings:
a. Z-N settings
b. TL settings
55 Process
c. DS method with 𝜏𝑐 =3 Control
4.3 PID Controller Design
The ultimate gain and ultimate period are determined by trial
and error to be 𝐾𝑐𝑢 = 7.88 and 𝑃𝑢 = 11.66. The calculated
PID controller settings are:

56 Process
@VLC-2015 Control

You might also like