A TEXT BOOK OF CHEMICAL ENGINEERING PROCESS CONTROL

Fourth Edition

2017

By:Prof. Gurashi A. Gasmelseed

I
 COPYRIGHT

All rights reserved except as permitted by the author, no part of this book may be
reproduced or disturbed in any form or by any means, or stored in a data base or
retrieval system without the prior written permission of the author.

II
 About The Author
Gurashi A. Gasmelseed received his [Link]. in Chemical Engineering from University
of Khartoum. He worked for the leather industry corporation for seven years at the
White Nile Tannery. After earning his [Link]. in 1978 at the University of Khartoum,
Prof. Gurashi worked for the Chemical Engineering department, University of
Kuwait as a laboratory Engineer process control, mass transfer, heat transfer and
fluid mechanics. He was awarded his Ph.D in Chemical Engineering at the
University of Aston in Birmingham-UK in 1985. After earning his Ph.D Prof.
Gurashi worked as a director of research assessment together with teaching process
control. From 1990 to date he joined University of Juba and University of Science
and Technology for teaching mass transfer and heat transfer as well as process
dynamics and control.

Prof. Gurashi published over 20 technical papers and has authored three……. Books.
Prof. Gurashi has supervised over 50 graduate thesis 12 for PhD and more than 40
[Link]. He has devoted over 30 years to his profession as a teacher, researcher, author
and practicing engineer.

III
 Dedication
This book is dedicated to Dr. M. E. Abu-Goukh of the University of Khartoum,
Chemical Engineering department. Abu- Goukh is the Pioneer in process dynamics
and control and through his efforts proper process control fundamentals, design of
controllers, tuning and system stability are made available. Before Abu-Goukh,
process control courses are just on measurements and very naïve, he introduces an
integral course for underground and graduate students. His textbooks are now at
most of the Chemical Engineering Departments, they are clear with many illustrative
examples, in facts, we learnt much from him.

IV
 Table of Contents

Title Page

Copyright II

About The Author III

Dedication IV

Table of Contents V

Chapter One: Introduction

1.1 Introduction to process control 4

1.2 Principle of feedback control 6

1.3 Basic elements of a control loop 7

1.4 Transfer function 9

Chapter Two: Mathematical Tools

2.1 Linearization 10

2.2 Computer Simulation 15

2.3 The Laplace Transformation 27

2.4 Laplace Transform Using MATALB 56

Chapter Three: Mathematical Models and Simulations

V
3.1 Introduction 59

3.2 Principles of formations of mathematical models 62

3.3 Fundamental laws 63

3.4 Empirical Method 83

Chapter Four: Block Diagram and Block Diagram Rules

4.1 Block Diagrams and Block Diagram Connections 91

4.2 Block Diagram Rules 94

Chapter Five: Process Dynamics

5.1 Systems order 110

5.2 The degree of dampness 121

5.3 Sensors and transmitters 125

5.4 Final control or control valves elements 131

Chapter Six: Continuous Control Systems

6.1 Proportional Actions 136

6.2 Integral Action, I- Action 144

6.3 Derivative Action (D- Action) 150

6.4 Commercial Controller 151

6.5 Two Position Control 156

VI
Chapter seven: Controllers Tuning

7.1 Introduction 163

7.2 Continuous Cycling Method (Ziegler-Nichols Tuning) 163

7.3 Process Reaction Curve Method (Cohen-Coon Tuning) 165

7.6 Integral Performance Criteria 169

Chapter Eight: System Stability

8.1 Introduction 174

8.2 Application of Routh-Hurwitz method as a tuning method 186

8.3 The Relative Stability 189

8.4 An Alternative Approach For relative stability 198

8.5 The Root. Locus Analysis 202

8.6 MATLAB Plotting 210

Chapter Nine: Frequency Domain Dynamics

9.1 Frequency Response of Linear Processes 217

9.2 Basic Theorem 218

9.3 Bode Plot and System Stability 226

9.4 Computer plotting of Bode diagram 236

9.5 Gain and Phase Margins 238

VII
9.6 Nyquist plots 239

9.7 Nyquist stability criterion 241

9.8 Plotting Bode plot 243

9.9 Case studies 245

Chapter Ten: Interaction and Decoupling

10.1 Introduction 278

10.2 Interaction of control loop 279

10.3 Relative gain array and selection of loops 290

10.4 Selection of loops 294

10.5 Design of noninteracting control loops 308

Chapter Eleven: Control Systems With Multiple Loops

11.1 Introduction 315

11.2 Cascade control 315

11.3 Feed forward control 323

11.4 Dead time and dead time compensation 327

11.5 Override control 331

11.6 Ratio control 333

11.7 Adaptive control 334

VIII
11.8 Inferential control 337

11.9 Auctioneering Control 340

Chapter Twelve: Digital Control

12.1 Z-Transforms 342

12.2 From Continuous-Time to Discrete-Time Systems 356

12.3 Conversion of Continuous to Discrete-Time models 363

12.4 Discrete-time Response of Dynamic Systems 367

12.5 Discrete-time T. F. for discrete systems in series 372

12.6 Stability Analysis of Discrete-time Systems 395

12.7 Digital controller design 405

12.8 Conversion using c2d 410

12.9 Stability and Transient Response 411

12.10 Discrete Root Locus 416

12.11 SCADA Acquisition System 418

Chapter Thirteen: Fuzzy Logic Controller

13.1 Introduction 438

13.2 Benefits of Fuzzy Control 439

13.3 Designing a fuzzy logic control system 441

IX
13.4 A fuzzy logic control systems 441

13.5 Design Process 443

13.6 Fuzzification 444

References

References 446

X
Chapter One
Introduction
CHAPTER ONE

Introduction:

The purpose of this book is to make clear the importance of process control to
the control engineers. This book covers the conventional control theories as
well as digital control systems. The procedure to develop the mathematical
process models that help in the analysis, design and operation of control
systems are covered with simple clear examples. It explains how the basic
components of the control system work. Feedback controller transfer
functions are determined with emphasis on their role, advantages and
disadvantages. Deign and tuning of feedback controller is covered using
Routh Hurwitz; root locus, Bode and Nyquist plots. Advance control system,
system interaction and decoupling and dead time compensation are illustrated
and analyzed. Having studied the fundamental of the dynamic of the process
control systems MATLAB is used for determination and analysis of the
system. The adjustable parameters are thus determined by using Ziegler-
Nicolas techniques. Part one of this book present two mathematical tools that
are particularly useful for analyzing process dynamics and designing of
control systems. Laplace transforms and linearization’s, when combined
together allow us to gain insight into dynamic response of a wide variety of
processes and instruments. Laplace transform is a method of solution of linear
differential equations. Since differential equations that represent most
dynamic processes are nonlinear, linearization is required to approximate
nonlinear differential equations with linear ones that can be solved using
Laplace transform. In this part the block diagram rules and signals are
represented. The purpose is to determine the overall transfer function from the
block diagram using block diagrams rules. Block diagram algebra is not
recommended, but in some cases a combination of the two make things easier.

1
 INTRODUCTION

The responses and system orders are included; these response are related to
the parameters of the process transfer functions so that the important
characteristic of the responses can be inferred directly from the transfer
functions without having to reinve them each time.
In part two of this book, the type of conventional controllers and their
combination are presented and investigated, with emphasis on On-Off, two
positions controller, P, PI and PID actions. Controllers tuning with different
techniques is presented including reaction curve (Cohen-Coon), quarter decay
ratio, first-order with dead time approximation of PID-Controller and Ziegler-
Nicolas methods. Stability analysis with Routh-Hurwitz and root-locus are
presented with examples. Frequency domain dynamics, determination of
amplitude ratio, phase angle and Bode-plots using low and high frequency
asymptotes are illustrated with examples. MATLAB plots are used for Bode
diagram and consequently tuning using ultimate gain and periods obtained
from the cross-over frequency.
In part three advance control systems are presented, these include, Cascade
control, auctioneering, adaptive and inferential controls. In this part
introduction is made to computer-digital control in SI/SO and MI/MO using
multiplexers. The interface between continuous and discrete-time control is
presented and explained. The Z-transforms which is a method of solution of
difference equations is introduced with examples similar to those dealt with
Laplace transforms. This part covers transformation from continuous-time to
discrete-time response of dynamic systems, their responses with various types
of digital controllers, and disturbances. It included pulse transfer function,
stability regions in S-plane and in Z-plane. This part also covers the design of
digital feedback controllers.

2
CHAPTER ONE

It is hoped that this book will help the undergraduate and graduate students to
understand the principle of process control and its application in various type
of industries.
Regards
Gurashi Abdalla Gasmeelseed

3
 INTRODUCTION

1.1 Introduction to process control:

1. Control how old is it?

This went back to Routh who in 1922 developed the theory of the automatic
steering of ships.
2. Process dynamics:
The process dynamic of a process starts from rest or steady-state to the desired
final steady-state:

3. Industrial Applications:
i. Chemical and nuclear industries
ii. Power industry
iii. Metallurgical industries
iv. Automobile industry
v. Agricultural industry
vi. Transportation applications

4
CHAPTER ONE

vii. Military applications

viii. Space applications
ix. Domestic applications
x. Biological and Medical applications

The merits of automatic control over manual control, these are:

1. Freedom of the system from human limitations.

Efficiency of the human operator varies among individuals depending upon:

a) Health.
b) Environmental conditions.
c) Fatigues
Fatigue less concentration lower efficiency
d) Limited period of time, the automatic controller is capable of acting logically
for very long period of time.
e) Productively Gains:
Automation considerably improves productivity
Automation improvement of labor productivity
Skilled labor transferred to other jobs lower production cost.
f) Process efficiency will be set constant at the required levels.
Automation high process efficiency maximum
production.
g) Energy saving.
h) Safety.

5
 INTRODUCTION

Classification of Control:

Control

Manual Automatic

Feedback Forward

1.2 Principle of feedback Control:

6
CHAPTER ONE

1.3 Basic Elements of a Control loop:

1- The process (e.g. distillation column, heat exchanger, reactor etc.).

2- The measuring element (sensor) such as thermocouple, pH meter, Oriffice
plate, flow meter, chromatogram, electro photometer.
3- The controller.
4- The regulating element or final control element such as a valve.
5- The controlled or measured variable.
6- The manipulated variable.
7- The disturbances.
8- The load variable.

Symbol used:

Tc Temperature controller

Lc Level controller

Pc Pressure controller

Fc Flow controller

TT Temperature transmitter

7
 INTRODUCTION

Pneumatic line

Electric Line

Ti Tranducer that convert electrical

signal to pneumatic signal

8
CHAPTER ONE

1.4 Transfer Functions:

output
Transfer Function
input

Transfer function

y s
G s 1.1
x s

d: discrete-time signal

c: continues signal

9
 Chapter Two
Mathematical Tools
CHATER TWO

2.1 Linearization:
A number of techniques are available for the solution of linear problems, but
there are no general mathematical techniques for treating nonlinear problems
and no general analysis is therefore possible.
Linearization methods depend on the simple fact that any nonlinear system
(function) is approximately linear if considered over a sufficiently narrow
range.
Process control systems normally operate with several variables held normally
at constant operating levels. These variables are subject to some departure
from these levels due to disturbances, but if the system is well designed these
perturbations from the normal values are relatively small, and the dynamic
behavior of the system can be described reliably by linear differential
equations even though the basic process equations are nonlinear. Therefore,
systems can be adequately represented by linear relation.

Linear and nonlinear equations:

1/ 5 6 6 linear

2/ √ nonlinear

3/ 10 nonlinear

Method of Linearization:

The Taylor Series Expansion:

- Take a function which is nonlinear:

10
 MATHEMATICAL TOOLS

, , , , 2.1

- Partial differential of the original function:

, ,
, , , , · ·
1! 2!
, ,

, ,
· ·
1! 2!
, ,

, ,
· ·
1! 2!
, ,

,
·
1!
,

,
· 2.2
2!
,

- Retain the linear terms only:

, ,
, , . , , , · ·
1! 1!
, ,

,
· 2.3
1!
,

Example 2.1:

|
1!

Example 2.2:
Linearize the following nonlinear function:

11
CHATER TWO

, 3
Solution:

2 | 2 |

54 6 |
54 54 3
108 54
Example 2.3:
Linearize the following nonlinear function:
, 2
Solution:
5
2 5 2 2
32 80 2
32 80 160
128 80

Example 2.4:
Linearize the following eq.:
F k√h
The system is at steady state h
Solution:

F kh
The linearization:
∂
F kh kh h h
∂h

12
 MATHEMATICAL TOOLS

1
k h kh h h
2
1 1
k h k h h
2 h
Example 2.5:
Linearize the Arrhenius temp dependence of the specific reaction rate k which is
highly nonlinear:

, The steady state temp T

Solution:

T T T
T T

T T T T

T T T T
T

T T T
T
Example 2.6:
Linearize the following non-linear function:
f ,F F
Solution:
The linear function, the steady-state are: and F:
f f
,F ,F ,F F F
,F
F ,F

f f
,F ,F F F
,F
F ,F

13
CHATER TWO

Example 2.7:

Linearize the following equation:

The steady state veelocity

Solution:

Let: m , n

eq becomes

mh n

The first term in eq. (3) is linear.

we have to linearize the second term.
n

n n | n

n 2n

Example 2.8:
.
Linearize the following non linear eq.:
3
The steady-state value is h.
Solution:

.
∂ .
h h
3 ∂h 3

14
 MATHEMATICAL TOOLS

.
1.5 .
h h h
3 3

. .
h h h h
3 2
Example 2.9:
Linearize the following nonlinear equation: ,
, 2 , , 3
Solution:

, | , , , , , , ,

, , , ,

4 27 2 ,
2 3 ,
3
31 4 2 27 3
31 4 27 81
4 27 58
2.2 Computer simulation:
Digital simulation is a very useful tool for solving equations describing chemical
engineering systems.

Over the years number of digital simulation packages have been developed, the
simulation languages relieve scientists and engineers of knowing anything about
numerical integration. These packages make it easier for them to set up and solve
the problems.
Numerical integration methods:
1. Netwon-Raphson method:

Most popular convergence methods, general eq.:

15
CHATER TWO

f
T T 2.4
f

Where:

T new guess of temp

T old or pervious guess of temp

f value of f at T T

f derivative, at T T

Example 2.10:

Using Netwon-Raphson method determine the roots of the following using the direct
evaluation.

P 4 17.4 39 47 a

P 12 38.8 39 b

Solution:

Starting with a guess, 2

P 4 8 17.4 4 39 2 47 6.6 c

P 4 8 17.4 4 17.4 d

Substituting eq’s (c) , (d)into eq. (2.4)

First approximation:

6.6
2 2.3793
17.4
The second approximation is:

16
 MATHEMATICAL TOOLS

2.3793

P . 1.16763

P . 24.13318

1.16763
2.3793 2.3309
24.13318
The third approximation is:

P . 0.02526

P . 23.0818

0.02526
2.3709 2.3298
23.0818
The forth approximation is:

P . 23.05858

P . 9.89723

23.05858
2.3298 2.3298
9.89723

The iterative process does converge to 2.3298

Euler method:

This is the simplest possible numerical integration method.

Assume we wish to solve the following ODE:

Where: f(x, t) = general nonlinear function.

We need to know a known initial conditions of x, usually this is at time, t=o. If we

have a curve of x Vs t:

17
CHATER TWO

The initial slope of the tangent:

∆
Slope 2.5
∆T

If the∆T is very small the eq.(2.5) become

∆
Slope 2.6
∆T

From eq. (2.6):

∆ ∆T 2.7

∆ ∆T 2.8

Let ∆

18
 MATHEMATICAL TOOLS

If ∆T (The step size or the integration intrval) is small enough, this estimate of
will be very close to the current value.

To step out another ∆t to t t , we can estemate ∆ t from:

This initial slope at t 2∆t

∆ ∆
Slope 2.9
∆t ∆

∆ ∆ ∆t 2.10

Let ∆

∆t 2.11
∆

∆t

19
CHATER TWO

The general form of Euler method is:

∆t 2.12
∆

Example 2.11:
Use Euler method to solve the following eq:

With the following initial conditions: At t 0 , 0 , 1

The integration step size, ∆t 0.05

Solution:

The general form of Euler method is:

∆t 2.12

1
1

Let 0

Using eq’s (2.12) and (*):

1 0.05 0.0 1 0.0 0.05 0.05

Let 1

Using eq’s (2.12) and (*):

1 0.05 0.05 1 0.05 0.05 0.0975

20
 MATHEMATICAL TOOLS

1 0.05 0.0975 1 0.0975 0.05 0.1426

Time = 2.0

The following table will be obtained:

Time x dx
dt

0.0000 1.0000 1.0000

0.0500 0.0500 0.9500

0.1000 0.0975 0.9025

0.1500 0.1426 0.8574

. . .

. . .

. . .

2.000 0.8715 0.1285

The smaller the step size, the more accurate is the integration.

If the step size is larger, the solution will go unstable.

21
CHATER TWO

Interval halving:

This method depends on initial guess of the parameter that is used to calculate a
certain value through a mathematical model. The calculated value is then compared
with the given value, and if does not converge, the guessed parameter is increased
or decreased until convergence with a limit, say 10 .

Example 2.12:

Determination of the bubble point temp:

Data given:

The pressure, P.

The composition in the liquid phase, .

Antione eq. (math. model)

22
 MATHEMATICAL TOOLS

ln a

Raoult’s and Dalton’s laws

Let us have a binary mixture of component 1 and 2:

P P , P 1 x P

y yP, y 1 y P

P P P

Appling interval halving method:

1. Guess a temp. And get Antoine constants, A , A , B and B .

2. Calculate the vapors pressure using eq. (2.19).

3. Calculate the partial pressures using eq. (2.20).

4. Calculate the total pressure, P cal.

5. Compare with the given pressure, P: If (P P 10⁻⁶

The guessed temp is the bubble point, otherwise put: ₒ ∆

6. Go through the same procedure, each time compute (P P if this value is +ve
go in the same direction by adding each ∆T to Tₒ, until the sign changes to –ve, then
do the same calculation with (T ∆T), and calculate, do the comparison and if the
sign of (P P change to positive, change your temp to (T ∆T), and continue

23
CHATER TWO

until the sign changes, then make the temp to be ( T ∆T ) and so on until
convergence.

Optimization

− Optimum conditions
− One variable
Example 2.13:
The total cost of insulation depends on the thickness of insulation

CT a b d

Where:
a, b, c and d constants
CT Total variable cost
x insulation thickness
Solution:

24
 MATHEMATICAL TOOLS

CT
0

If 64 16

64
2
16

For max and min:

CT 2 2 64
2
16

optimum point 0 ve

optimum value occurs at a min. but if the second derivative is

ve the optimum value occurs to max

Example 2.14:
Determine the optimum value for the two independent variables
11,900
CT 2.33 1.86 10 a

Determine the values of x and y which will give the least total cost

Solution:

CT 11,900
2.33 0

25
CHATER TWO

CT 11,900
1.86 0

At the optimum point:

11,900
2.33 0 b

11,900
1.86 0 c

Solving eq’s b and c simultaneously: 16 20

By substitution of x and y on eq’s (a): CT 121.6

26
 MATHEMATICAL TOOLS

2.3 The Laplace transformation

Definition:

The Laplace transformation of a function f(t) denoted by L[F(T)] is defined

according to the following equation:

LF T F s 2.13

Where:

The symbol S is a complex quantity of the form:

Re s , Real

Im Imaginary

√ 1 or √ 1

Transformation of important functions:

Example 2.15:

Unit step functions:

- The unit step function is defined by:

U( ) = 1

- The unit step function represented by:

0
1 0

27
CHATER TWO

- Unit step:

- Apply the law of transformation:

LF T F s 1 0

Example 2.16:

Determine the Laplace transformation of a constant function, f(t) = k

Solution:

Applying the transformation law:

LF T F s 0

28
 MATHEMATICAL TOOLS

Example 2.17:

Determine the Laplace transformation of unit step function delayed by a unit of time,
which is of the following function:

f t 1 a

Solution:

Applying the transformation law:

Lf t 1

e 0

Example 2.18:

Determine the Laplace transformation of the following function:

F t

Solution:

Applying the law of transformation:

29
CHATER TWO

Lf t

Example 2.19:

Determine the Laplace transformation of the sin function:

F t sin

Solution:

Using the trigonometric fact that:

sin

Applying the law of transformation:

1
Lf t L sin It
2

1
Lf t
2

I II

- Taking term I:

30
 MATHEMATICAL TOOLS

0—

Taking term II:

1
II
2

1 1
·
2 jω s

1 1 1 1
0—
2 jω s 2 jω s

1 1 1 1
2 s jω 2 jω s

1 1 2
2 2

Example 2.20:

Determine the Laplace transformations of a Cos function:

F t cos

31
CHATER TWO

Solution:

Applying the transformation law:

Lf t cos

- Applying the trigonometric fact:

1
cos
2

1
cos
2

1
2

I II
Taking term I:

1
I
2

1 1
·
2 s jω

1 1 1 1
0
2 s jω 2 s jω

- Taking term II:

32
 MATHEMATICAL TOOLS

1
II
2

1 1
·
2 jω s

1 1 1 1
0
2 jω s 2 jω s

1 1 1
cos
2 s jω jω s

1 1 2
2 2

Example 2.21:

Determine the Laplace transformation of a ramp function:

Solution:

What is the ramp function; it is that function which is represented graphically from
the above equation:

33
CHATER TWO

- Applying the transformation law:

Lf t . a

- This equation can only be solved by integration by parts:

Integration by parts in general application:

. | b

- Applying equation (a) to equation (b):

Let t = u

Let:

34
 MATHEMATICAL TOOLS

Putting equations (c) and (d) into equation (b):

1 1
. .

1
0

1 1 1
·

Example 2.22:

Develop Laplace transformation of an impulse function, also referred to as delta

function, or heavy side. It is in a form of long spike with very infinitism width. It
has a unit area. It is L.T. of a differential of step function.

Solution:

35
CHATER TWO

- This means that the height is very long compared with the area.
- Represent the unit step function by:

1 a
∞
As at ∞ approach zero, we may just play a trick to realize a
solution.

- Differentiating equation (a)

0 b

c
∞

- Taking the Laplace transform of equation (c)

0 lim d
∞

36
 MATHEMATICAL TOOLS

Dividing equation (d) by :

1
lim 1
1

Laplace transformation of Derivatives:

- Laplace transformation of a derivative function:

F t 2.14

Is given by the following rule:

F t L 0 2.15

Where f(s) = L[f(t)]

Example 2.23:

Find the Laplace transform for the function:

F t

Solution:

By the transform rule:

L 0

- by comparison f t

37
CHATER TWO

L ·

f s

f 0 1

2] Theorem 2:

L 0 0

Or

L 0 0

3] Theorem 3:

L 0 0

38
 MATHEMATICAL TOOLS

Example 2.24:

Use the differential theorems to prove the following L.T:

i- Lt

ii- L

iii- L sin

Solution (i):

Applying the derivative theorem:

L 0

F(t) = t

F(0) = 0

L 0

1 0

Solution (ii):

Let:

F t

39
CHATER TWO

F(0) = 1

Substituting these into the transformation equation:

L 1

L 1

L 1

L 1

Solution (iii):

Applying theorem 2:

L 0 0

Let:

f t sin

cos

sin

F(0) = 0

Substituting these into equation (*):

L sin sin 0

40
 MATHEMATICAL TOOLS

Rearranging:

L sin sin

L sin

L sin

Example 2.25:

Solve the following equation:

5 6 6

The initial conditions ́ 2, 2

Solution:

- Taking the first term:

L 0 0

2 2

- Taking the second term:

L 5 5 5 0

5 10

- Taking the third term:

6 6

- Taking L.T. of the term on the right:

41
CHATER TWO

6
L6

2 2 5 10 6f s

5 6 2 12

2 12 6 2 12 6
5 6 3 2
Taking the initial conditions:

y 0 0 0

5 6

5 6

Example 2.26:

Determine L.T. of the following equation:

The initial value 0 2, 0 0

Solution:

- Taking the first term:

L 0 0

42
 MATHEMATICAL TOOLS

2 0 2

- Taking the second term:

L 0

- Taking the term on the right:

L .

2 2

2 2

Laplace transformation of an integral function

F t 2.16

L 2.17

43
CHATER TWO

Integrating by parts:

In general:

. |

Let:

218

2.19

Let :

2.20

2.21

The operation of integration is equivalent to division by S in the Laplace domain,

using zero initial conditions.

e.g.

Inversion of Laplace Transformations

Heavy side Expansion:

44
 MATHEMATICAL TOOLS

- This is to invert from L-domain to the time-domain,

- Assume Laplace Transfer of a function is given by:

2.22

Where:

Polynomial in S of order m

Polynomial in S of order n

Procedure for the inversion by partial fraction expansion:

1- Expansion:

2.23

2- Determine the values of the constants , , ,

3- Determine the invert of every partial fraction, then the unknown function i.e.

2.24

Where:

the inverse of L.T.

Example 2.27:

Find the inverse of the follow equation:

45
CHATER TWO

Solution:

- Fractional expansion

b
1 1 2

c
Determine the constants , :

To determine :

6
d
1 1 2 1 1 2

- Multiply both sides of equation (4) by 1 and set 1 0 1

1 1 1

6
1 2

- Putting s = 1 into equation (e):

1 1 6
1 1 1 2

6
3
2

46
 MATHEMATICAL TOOLS

- To determine :

Multiply both sides of equation (d) by (s+1) and set s+1 = 0 1

6
1 1 1
1 1 2 1 2

6
f
1 2

Putting s = -1 into equation (f):

- To determine :

Multiply both sides of equation (d) by 2 and set 2 0 2

2 2 2

, then putting into equation (c):

Example 2.28:

Find the inversion of the following function

47
CHATER TWO

1
a
2 5

Solution:

Factorization of by inspection:

2 5 1 2 1 2

Equation (a) becomes:

1
b
1 2 1 2 1 2 1 2

It is known that the

L c

Let:
1 2
And
1 2

f
1 2 1 2

Putting the values of a and b from equations (d) and (e) into equation (g):

48
 MATHEMATICAL TOOLS

Determine the values of and

The initial and Final Value Theorem:

- For initial value theorem:

0.0 Unit of time (h, min, s)

- For final value theorem:

Example 2.29:

The initial and final values of the f(t) can be obtained without first inverting the
Laplace transform, F(s). This is done by using the initial and final- value theorems.

6 3
F s
1 4 5

49
CHATER TWO

Solution:

- Initial value:

lim lim . a

6 3
lim . lim
1 4 5

6 18
lim b
10 29 20

Dividing the num and den of equation (b) by :

0

= lim 6
0 0 0

The initial value of f(t) =

lim 6

- The final value theorem:

lim lim .

6 3 6 18
lim lim 0.0
1 4 5 10 29 20

The final value of f(t) = lim f t 0.0

50
 MATHEMATICAL TOOLS

Worked examples:

Example 2.30:

Solve the following differential eq’s in L.T.

Solution:

From L.T. table:

1
L

1
L
4

1
4

1
4

1
1 4

Example 2.31:

3 2 8

Initial conditions: 0

51
CHATER TWO

Solution:

From L.T. table:

1
L t

2
L 2t

Lk

8
L8

2 8 2 8
3

2 8
3

2 8
3

Example 2.32:

Invert the following eq. from L.T. domain to time domain using partial fraction.

1
1

Solution:

1
a
1 1

Multiply both sides of eq. (a) by 1

52
 MATHEMATICAL TOOLS

1 1 1 1

1 1 1 1 b
1

Set S = 0:

1 0 B 0

B 1

Multiply both sides of eq. (a) by (s-1):

eq a becomes:

1
1 1

Set 1 0 , 1

1 0 0

1
1 1

By trial and error:

Set S = -1:

1
A B
2 2

1 2A 2B C

1 2A 2 1

53
CHATER TWO

2A 2

A 1

eq a becomes:

1 1 1 2
1 1

Using L.T. table:

1 t 2e

Example 2.33:

Use partial fraction to find the inverse of the following function:

Solution:

Making some trial and error it can be shown that:

Multiply eq. (b) by :

Set 0 ,

54
 MATHEMATICAL TOOLS

A 0 d
2

Putting eq’s (c) and (d) into eq. (b):

1
2

From L.T. table:

1
sin wt
2

Short Table of L.T.

Function in time domain Laplace transform

Unit impulse: 1 1

Unit step 1

Ramp: f t at

f t t n!
s 1

Exponential f t e 1

f t te 1

f t sin ω
s

55
CHATER TWO

f t cos s
s

f t 1 e

f t sin e ω
2

2.4 Laplace transform using MATLAB:

MATLAB format:
>> syms t
>> laplace(f)
where:
 laplace(f): is the Laplace transform of the function F with default
independent variable t.
 syms: shortcut for constructing symbolic objects.
Example 2.34:

Determine Laplace transform of an exponential function f(t) = e –at

Solution:
>> syms a t
>> f = exp(-a*t)
>> laplace (f)
ans =
1/ (a + s)
Example 2.35:

Determine Laplace transform of the sinusoidal function : f(t) =sin (kt)

56
 MATHEMATICAL TOOLS

Solution:
>> syms k t
>> laplace(sin(k*t))
ans =
k/(k^2 + s^2)
Inverse of Laplace transform:

MATLAB format:
>> syms s
>> ilaplace(L)
where:
ilaplace(L) is the inverse Laplace transform of the function L with default
independent variable s.
Example 2.36:
Determine the inverse of the function: Y(s) = 5/
Solving ordinary differential equations:

Solution:
MALAB format:
>> dsolve('equation', 'initial condition1', ' initial condition2', ………….. )
Where:
dsolve is the symbolic solution of ordinary differential equations.
Example 2.37:

Solve the following ODE:

4 5 2 2 0, 0, 0

57
CHATER TWO

Solution:
>> syms x
>>dsolve('D3x+4*D2x+5*Dx+2*x=2','x(0)=0',
'Dx(0)=0','D2x(0)=0')
ans :
1 - (2*t)/exp(t) - 1/exp(2*t)

58
 Chapter Three
Mathematical Models and Simulations
 MATHEMATICAL MODELS AND SIMULATIONS

3.1 Introduction:

This is proved to be useful for design & optimization of physical, chemical and
biological processes. The use of simulation has expanded rapidly during the previous
three decades because of the availability of high speed computers and computer
workstations. All engineering computations will shortly be computerized and
engineers need to understand the principles behind software and to effectively use
the software to solve pertinent problems, engineer should develop the following
skills:
1. A sound understanding of engineering fundamentals. The engineer should be
familiar with the physical system and its mechanisms in order to be able to
simulate a real process and evaluate the simulation.
2. The engineer has to be able to develop a set of mathematical relations that are
adequately describing the process behavior.
3. Rapid solution to simulation problems must be obtained. The engineer must
be evaluate and use correctly the available commercial software packages.
Benefits of process industry simulation:

1. Economic desirability:

It is cheaper to use simulation techniques incorporating fundamental lab. Data

in the math. Model then building and using pilot plant data.

2. It is a convenient way to process disturbance upon operation. A comparison

between alternatives can be easily analyzed and assessed.

General strategy of process simulation this is represented in the following block

diagram:

59
CHATER THREE

1. Definition of the problem.

2. Math. Modeling of the process.
3. Equation organization.
4. Solution and compilation.
5. Validification and interpretation of the results.
1) Definition of the problem:
The following must be raised and answered:
 What do I want to find out?
 What are the important consequences of study?
 Why this job be done?
 What engineering effort should be required?
 How long should the job take?
2) Math. Modeling of the process:
 Assumptions.
 Data collection and Lab. experiments whenever necessary to determine
unknown parameters.
 Appropriate relations of the process.
3) Equation organization:
Arranging the eq’s in an information –flow diagram.
4) Compilation:

60
 MATHEMATICAL MODELS AND SIMULATIONS

For obtaining solution to process simulation problem, the engineer has several levels
of compilation ranging from solution by inspection to analytical& high -speed
computer solution. Numerical methods for the solution of sets of algebraic eq. ode
and partial differential eq’s (pde) are modeled.
A practically useful and well documented if the software’s are available Additional
software package such as Matlab are also now available which have excellent
graphical capabilities and ease the programming of certain problems.
5) Interpretation of the results:
The real payoff of the simulation of a process is in intelligent interpretation of the
results by the engineer.
At this point, the engineer must ascertain whether the model is valid representation
of the actual data or whether it needs revision.
Limitation of the process simulation:
The engineer should aware of the following:
1. Lack of good data and knowledge of the process mechanism.
2. Solution of the model which might pose a problem, a model that cannot be
solved is useless.
3. The danger of under looking the assumptions.
Usefulness of process simulation:

Computer simulation is playing important role in the solution of chemical,

biological, energy and environmental problems. Mathematical model simulation can
be used for:

1. Research and development (R& D).

2. Design, exploring the sizing and arrangement of processing equipment for
dynamic performance; studying the interactions of various parts of the

61
CHATER THREE

process; simulating start-up, shutdown and emergency situations and

procedures.
3. Plant operations, troubleshooting, requirements for expansion ….etc.

3.2 Principles of formations of mathematical models:

1. Basis: The bases for mathematical model are the fundamental physical and
chemical laws, such as laws of conservation of mass, energy and momentum.
2. Assumptions: This depends on engineering scientific judgment such taking
constant density, specific heat …. etc. An extremely rigorous model that
includes every phenomenon down to microscopic details will be difficult to
solve and is useless. An engineering compromise between rigorous and
simple assumptions to get an answer that is good enough is always required;
this has been called optimum sloppiness. it involves making as many
simplifying assumptions which are reasonable without throwing out the baby
with the bath water.

3. Mathematical consistency of the model:

Once all the eq’s have been developed, make sure that the number of
variables equals to the number of equations. That is the so-called degree of
freedom of the system must be zero:

Degree of freedom (DF) = 0

The units of all terms of the must be consistent i.e. use (SI) units i.e. kg, m, s.

4. Solution of the model:

The available solution techniques and tools must always be kept in mind as the
mathematical model is developed. An eq. without any way to solve is useless.
5. Verification of the model

62
 MATHEMATICAL MODELS AND SIMULATIONS

It must be proved that the model describes the real-world situation. At the design
stage this sometime cannot be done because the plant has not yet been built, data can
be collected from similar plants. A pilot plant can also be built in which some
experimental data can be obtained.
3.3 Fundamental laws
1. Continuity equations (Mass balance eq’s ):
Total continuity eq. (mass balance):

The principle of the conservation of mass when applied to a dynamic system says:

Mass flow into mass flow out of time rate of change of

3.1
the system the system mass inside the system

The right hand side of eq. (3.1) will either be a partial derivative or ordinary

derivative with the mass inside the system with respect to the independent

variable t.

Example 3.1:

63
CHATER THREE

Assumption constant density:

Example 3.2:

Assumptions:
1. Constant diameter cylindrical pipe.
2. Flow is laminar i.e. plug flow.
3. No radial gradients in velocity or properties, but axial gradients can exist.
4. Density and velocity can change as the fluid flows in the axial or z- direction.
5. There are two independent variables, time t, and axial position z. density and
velocity are functions of both t and z i.e. , , , .
Apply the continuity eq. to a system consists of small slice shown in the fig. above,
is now a microscopic one:
- The differential element (slice) has a thickness dz and x-sectional area equal to x-
sectional area of the pipe, A . .

64
 MATHEMATICAL MODELS AND SIMULATIONS

Eq’s development:

Time rate of mass inside the system:

kg/sec

Mass flowing into the system through the boundary at z:

kg/sec

Mass flowing out of the system through the boundary at ( ):

The last expression for the flow at ( ) may be thought of as a Taylor series
expansion of a function f (z) - around z. The value of the function at a spot away
from z is:

∂F ∂ dz
F z dz f z dz
∂z ∂z 2!

If is small the series can be truncated after the first derivative term:

Let:

, in Taylor series,

f dz

Substituting these terms into eq. (a), the continuity eq.

Canceling out and assuming constant

65
CHATER THREE

Component continuity eq. (Component balances):

Let the number of component 1, 2, 3 …

The component continuity eq. for the ’ component of the system
Flow of mass of j Flow of the mass of j
component into component out of
the system the system
Rate of formation Time rate of change
of moles of j component of j component 3.2
from chemical reaction inside the system
The units are moles of j per unit time.

Example 3.3:

Consider the following CSTR:

66
 MATHEMATICAL MODELS AND SIMULATIONS

Component A reacts irreversibly at specific reaction rate to form a product

component B:
Let:

The initial concentration of component A

The concentration of component A inside the reactor and exit steam

The reaction rate is simple first-order

Rate of consumption of A from the reaction

- Flow of A into the system

- - Flow of A out of the system
- Rate of consumption of A
(the –ve sign means A is consumed)

The rate of change of A inside the tank

Note that we have used ordinary derivative since time, t, is the only independent
variable in this lumped system. The right hand side of the eq. is the dynamic part of
the system.
Example 3.4:
A microscopic system shown below is used for the following reaction:

Reactant A goes to B at the specific reaction rate, but B can react to form
component C at specific reaction rate; develop the mathematical model for the
system.

67
CHATER THREE

Assuming first-order reactions, the component continuity eq. for components A, B

& C are:

Example 3.5:

A microscopic system shown below is used for the following reaction: . As

slice of material moves down the length of the reactor the concentration of reactor
, decreases as is consumed. The density , velocity and concentration
can all vary with time and axial position z, determine the mathematical model.
Solution:
- Assume plug flow conditions.
- No radial gradients in velocity and concentration, as well as density.
- The concentration of A in the feed at 0 is
,0
- The concentration of A in the reactor at is defined at L

68
 MATHEMATICAL MODELS AND SIMULATIONS

, ( )

Appling the component continuity eq. for reactor A to a small differential slice of
width, as shown above.
The diffusive flux of , is given by Fick’s law of diffusion:

Where: diffusion coefficient, the – ve sign is to show that diffusion takes place
from high to low concentration.
Molar flow of A into the boundary at z = bulk flow and diffusion flow

.
Molar flow of A out of the slice boundary at

Rate of formation of A inside the system

Time of change of A inside the system

69
CHATER THREE

Substituting to continuity eq.:

But

This is a math. model of a CSTR.

Example 3.6:
Develop a mathematical model for an ideal Binary distillation column shown below:

70
 MATHEMATICAL MODELS AND SIMULATIONS

Assumptions:

1. Vapor hold-up on each tray will be neglected.

2. The molar heat of vaporization of both binary A and B are approximately
equal, this means that 1 moles of condensing vapor release enough heat to
vaporize 1 mole of liquid.
3. The heat losses from the columns to the surrounding are assumed to be
negligible.
4. The relative volatility, of the two components remains constant through the
column.
5. Each tray is assumed to be 100% efficient i.e. vapor leaving each tray is in
equilibrium with liquid leaving the tray.

71
CHATER THREE

The boil-up rate in the rectifying section and stripping sections are considered to
be equal
and therefore there is no need for energy balance.
Vapor-liquid equilibrium relationship:

1 1

Where = relative volatility.

6. The dynamic of the condenser and reboiler are neglected.
7. Molar flow rate of the liquid leaving each tray is related to the liquid hold-up by
Francis weir eq.:
f …… 1,2, … . . , … …

Let us now develop the state q′ that describe the dynamic model of a distillation
column. The fundamental quantities are the total mass of components.
1. Feed tray ( ):

72
 MATHEMATICAL MODELS AND SIMULATIONS

Total mass balance:

=
Component balance, A:

1. Top tray N:

Total mass balance:

Component, A, balance:

73
CHATER THREE

2. Bottom tray ( 1)

Total mass balance:

Component, A, balance:

1. tray ( 1, 2, … . 1 and i f)

74
 MATHEMATICAL MODELS AND SIMULATIONS

Total mass balance:

Component, A, balance:

2. Reflux drum:

75
CHATER THREE

Total mass balance:

Component balance:

3. Column base:

76
 MATHEMATICAL MODELS AND SIMULATIONS

Component, A, balance:

All the eq’s above are state eq’s and describe the dynamic behavior of the
distillation column.

The state variables of the model are:

Liquid hold-up: , , ….., ,…… ,….. RD and B

Liquid compositions: , , ….., ,…… ,….. RD and B

′
To complete modeling of the column in addition to the state , we need the
following relationship:

1. y f( …… 1, 2 … f … …

2. Hydraulic relationship (francis wer formula):

f … 1, 2 … f … …

′
When all the modeling is sloved, we will find how the flow rates and
compositions of the two product streams (top and bottom) change various input
variables.

Energy equation:

The first law of thermodynamic put forward the principle of conservation of energy,
written for a general open system it says:

77
CHATER THREE

Flow of internal, Flow of internal,

Heat added
kinetic and potential kinetic and potential
to the system
energy into the system energy out of the system
Time rate of change
Work done by the
of internal, kinetic
system on the 3.3
and potential energy
surrounding PV
inside the system

Example 3.7:

A CSTR is cooled by a cooling coil inside the tank to remove the exothermic heat
of reaction (j/mole of A reacted).

The following fig shows the operation:

Develop a mathematical model for the process.

Solution:

- Use the normal convention :

78
 MATHEMATICAL MODELS AND SIMULATIONS

- is –ve for exoshermic reaction

- is +ve for endothermic reaction
- The rate of heat generation due to chemical reaction is the rate of consumption
of A times :

The rate of heat removed

The energy balance using eq. (3.3):

3.4

In the system shown above:

- There is no shaft work, 0

- The velocity is not a very high, 0
- The elevation of inlet and outlet are the same , 0

eq. (3.4) reduces to:

An alternative easy method:

Time rate of change of energy inside the tank

Rate of Energy into system

79
CHATER THREE

Rate of energy out of the CSTR

Example 3.8:

Develop a mathematical model of a water heating system consisting of 3 tanks in

series with hold-up volumes of 0.35, 0.50 and 0.75 respectively. The temp of the
feed water, which flows through the tanks at a rate of 2.7 / is measured on the
third tank. The system contains a signal heater in the first tank only, 4.1888
/ , 1000 / .

Solution:

The physical diagram of the system:

80
 MATHEMATICAL MODELS AND SIMULATIONS

Appling the law of conservation of energy:

#1

. Hold-up mass

Dividing eq. (b) by in all through:

Let:

Time constant = Residence time

Take L.T of eq. (c)

1
1

1 1
1 1

81
CHATER THREE

0.75 4.1888 10 3140.1 /

. 0.35 1000
466.67
0.75

. . .

Tank #2

Dividing eq. (d) by in all through and rearrange:

Putting

1
e
1

Tank #3

Dividing eq. (f) by in all through and rearrange:

Putting

Putting eq. (e) into eq. (g)

82
 MATHEMATICAL MODELS AND SIMULATIONS

1 1
h
1 1

Putting eq. (d) into eq. (h):

1 1 1 1
1 1 1 1

1 1 1
i
1 1 1

Given the following data:

, , (To be calculated)

= given

= given

=given

Then we invert eq. (i) into time domain and solve for and tray time (t), at s-s ,t ∞

3.4 Empirical Method:

To develop an empirical steady-state or dynamic relation between two variables such

as process input, x, and output, y, it is instmetive first:

1. To plot the data y Vs. x for steady –state conditions

83
CHATER THREE

2. To plot the data y and x Vs. time ,t , for transient response data

3. Once the data are plotted, you look for the best curve that fits the points that
is referred to as curve fitting:

84
 MATHEMATICAL MODELS AND SIMULATIONS

Linear regression:

85
CHATER THREE

Get the model eq.:

Slope of the fitting line

The model eq. is:

3.5

Let represent the actual experimental data; then the difference between
calculated and experimental data are:

3.6

Is referred to as error or residual between the model prediction, putting eq.(3.5)

into eq.(3.5)

3.7

Since there is no more than one possibility for fitting the data, the common
acceptable method of determining the best fit is to calculate the values of and
that minimize the sum of the squares of the error , ,namely:

, 3.8

Note that the value of and are known while and are to be found so as to
minimize , the sum of square of the errors.

86
 MATHEMATICAL MODELS AND SIMULATIONS

The estimates of and found for specific data set are referred to or designated
as:

And

For a linear model and r data points, values of and that minimize eq. (3.8) are
found by setting the derivatives of with respect to and equal to zero. Since
there is a quadratic function, this approach leads to two linear in two unknowns
and . This least square estimation approach, also called linear regression can be
extended to more general models:

1. Any number of input variables

2. Single input-single output(SISO)
The analytical solution:

0 2
3.9

r
1 2
3.10

Where:

x , x 3.11

y , xy 3.12

87
CHATER THREE

Example 3.9:

An experiment has been run to determine the steady-state power delivered by a gas
turbine driven generator as a function of fuel rate.

The following data are obtained:

Fuel flow rate power generated (MW)

x y

1.0 20

2.3 4.4

2.9 5.4

4.0 7.5

4.9 9.1

Find the best linear model

Solution:

A linear model can be represented by:

3.5

First determine and using eq’s (3.9), (3.10), (3.11), and (3.12)

1 2.3 2.9 4.0 4.9 15.1

x 54.71

88
 MATHEMATICAL MODELS AND SIMULATIONS

y 28.4

102.37

7.977
0.175
45.54

83.01
1.823
45.54

Putting these values into eq. (3.5)

0.175 1.823

Eq. (3.6) is an empirical model obtained by linear regression:

To compare the data predicted (calculated) by the mathematical model a gains the
experimental data we apply eq. (3.6) putting the experiment value

Point (1):

0.175 1.823 1.0 1.998

Point (2):

0.175 1.823 2.3 4.368

Point (3):

0.175 1.823 2.9 5.461

Point (4):

0.175 1.823 4.0 7.467

Point (5):

89
CHATER THREE

0.175 1.823 4.9 9.107

Comparison of the results:

Experimental model calculate

, , exp %

1.0 2.0 1.998 0.1%

2.3 4.4 4.368 0.73%

2.9 5.4 5.461 1.13%

4.0 7.5 7.466 0.45%

4.9 9.1 9.107 0.08%

The empirical model is vilified with very good agreement

90
 Chapter Four
Block Diagram and Block Diagram Rules
 BLOCK DIAGRAM AND BLOCK DIAGRAM RULES

4.1 Block Diagrams and Block Diagram Connections:

1. Distribution Point:

2. Summing point:

3. In and Out signals:

Transfer function

91
CHAPTER FOUR

Transfer function
I

4. Summing Point of two elements:

4.1

4.2

4.3

4.4

5. Branching connection:

92
 BLOCK DIAGRAM AND BLOCK DIAGRAM RULES

4.5

4.6

4.7

4.8

6. Series connection:

4.9

4.10

Putting equation (4.10) into equation (4.9):

4.11

7. Feedback connection:

93
CHAPTER FOUR

Algebraic Method:

- Starting from the end is equal in all directions

4.12

4.13

4.14

- Back substitution:

4.15

4.16

4.17
1

4.2 Block Diagram Rules:

4.18
1

Multiplication of the forward transfer functions

Multiplication of the loop transfer functions

Positive for – ve feedback and –ve for positive feedback.

94
 BLOCK DIAGRAM AND BLOCK DIAGRAM RULES

Overall transfer function =

Example 4.1:

Find the overall transfer function of the multiloop shown below:

Solution:

Applying the block diagram rules:

The overall transfer function:

∑
a
,

Transfer function of the forward path

Transfer function of loop k and k = 1 refers to the innermost loop as shown

in the fig.

95
CHAPTER FOUR

Putting equations (b), (c) and (d) into equation (a):

Example 4.2:

Reduce the block diagram shown below to a single block diagram with an overall
transfer function :

96
 BLOCK DIAGRAM AND BLOCK DIAGRAM RULES

(Figure 1)

Solution:

In this case we apply a combination of block diagram algebra and block diagram
rules:

Taking loop 1:

– a

a E b

b E c

c E d

Putting equation (b), (c) and (d) into equation (a):

M E

(Figure 2)

Taking loop (2):

e
1 1

97
CHAPTER FOUR

(Figure 3)

Putting Figures (2) and (3) into (Figure 1):

(Figure 4)

Applying the block diagram rules to (Figure 4):

· · 1

1 1 · 1 ·
1

98
 BLOCK DIAGRAM AND BLOCK DIAGRAM RULES

1
1

1
1
·
1 1 ·
1

Example 4.3:

Determine the overall transfer function for the fig. shown below:

99
CHAPTER FOUR

(Figure 1)

(Figure 1) can be represented as follows:

- These are parallel paths:

100
 BLOCK DIAGRAM AND BLOCK DIAGRAM RULES

The overall transfer function of parallel paths (feed forward-loop) is given by:

Example 4.4:

Determine the overall transfer function for the following figure:

(Figure 1)

Solution:
(Figure 1) can be represented in an equivalent block diagram:

101
CHAPTER FOUR

(Figure 2)

- (Figure 2) consists of a single feedback loop with two parallel. Feed forward
paths.
- Applying the block diagram rules:

∑
a
1

Putting equations (2), (3), and (4) into equation (1):

e
1

102
 BLOCK DIAGRAM AND BLOCK DIAGRAM RULES

Example 4.5:

a) Determine the overall transfer function of the following block diagram:

b) Determine the Laplace transform of the output, for the following input
signals:
i- Unit step Ų
ii- Impulse function,
iii- Sin function, sin

Solution:

a)

∑
, a
1 ∑ ,

Equation (a) represents the general multi-path, multi-loop rule which considers m
path that pass through the n loops together with “P” direct feed forward paths.

103
CHAPTER FOUR

, d

, e

, f

Substituting these values into equation (a):

g
1 1

b)

i. Unit step change in the output:

1 h

1
i

Putting equation (i) into equation (g):

1 1
j
1

ii. Impulse function input:

104
 BLOCK DIAGRAM AND BLOCK DIAGRAM RULES

iii. Input sin function:

= sin

Example 4.6:

For the block diagram shown in the figure below, determine:

The overall transfer function using the block diagram rules.

- The original block diagram consist of :

i- The major outer loop: loop 1

105
CHAPTER FOUR

ii- The inner loop: loop 2

iii- The inner most loop: loop 3

Solution:

- Eliminate the inner most loop i.e. loop 3:

0.5 0.5
1 a
1 3 1 3

0.5 0.25
1 0.5
1 3 1 3

0.25 1 3 0.25
1 1
1 3 1 3

0.5 0.5 0.5

1 3 1 3 0.25 1 3 0.25 3 1.25
1 3

- Replace loop (3) by it is transfer function, we will get the following figure:
- The overall transfer function is:

106
 BLOCK DIAGRAM AND BLOCK DIAGRAM RULES

b
1 ∑ ,

1 0.3
0.3 c
0.2 1 0.2 1

1 0.5 0.25
0.5 d
0.2 1 3 1.25 0.2 1 3 1.25

For loop 1:

1 0.3
, 0.3 e
0.2 1 0.2 1

107
CHAPTER FOUR

- Substituting equations (c),(d),(e) into equation (b):

I
0.3
0.2 1
0.25 0.3
1
0.2 1 3 1.25 0.2 1

II

Untying envelope II

0.2 1 3 1.25 0.25 0.3 3 1.25

0.2 1 3 1.25

.
. . . . . .
. .

III

0.3 3 1.25
2
0.2 1 3 1.25 0.25 0.3 3 1.25

IV

From envelope III:

0.3 3 1.25 0.9 0.375

0.2 1

+ 3 1.25

108
 BLOCK DIAGRAM AND BLOCK DIAGRAM RULES

3 0.6 3

3 0.6 3

+ 1.25 0.25 1.25

3 1.85 3.25 1.25

0.3 3 1.25 0.9 0.375

3 1.85 0.9 3.25 0.375

. .
. . . . .

109
 Chapter Five
Process Dynamics
CHAPTER FIVE

5.1 Systems order:

If we have a linear differential equation:

5.1

1. Zero-order systems:

0 5.2

Putting equation (5.2) into equation (5.1):

5.3

Taking L.T. equation (5.3) becomes:

Let:

Process gain

5.4

In block diagram form:

110
 PROCESS DYNAMICS

Example of zero-order systems: Valve of T.F., thermocouple, transducer . . . . etc.

2. First-order systems:

1 5.5

Putting equation (5.5) into equation (5.1):

5.6

Dividing equation (5.6) by :

5.7

Let:

, where is the time constant. The time constant is the time required for the

process to adjust it self after a change.

5.8

Taking Laplace transform of equation (5.8):

The process T.F.:

1
5.9
1

111
CHAPTER FIVE

In block diagram form:

Example of first-order systems: T.F of a level in a tank, T.F of a heating tank . . .

etc.

Example 5.1:

Consider a thermometer to be located in a flowing stream of fluid for which the

temperature varies with time. Determine the response of time variation of
thermometer reading upon a change in . Where:

Surrounding temperature

y = thermometer reading

Solution:

112
 PROCESS DYNAMICS

Assumptions:

1- Resistance due to the glass and mercury is neglected.

2- Uniform temperature throughout the mercury.
3- Glass wall does not expand.

Apply the energy balance equation:

Input rate-output rate = time rate of accumulation or depletion

Hat variable or deviation variable:

Constant

0.0

represents the dynamic deviation,

Heat input , h = heat transfer coefficient

113
CHAPTER FIVE

Heat output A = heat transfer area, ́ output temp

Accumulation

m a

m hA b

Dividing equation (b) by :

Let:

Taking L.T. of equation (3):

1
d
1

In block diagram form:

114
 PROCESS DYNAMICS

Forcing function:

Unit step function:

Putting equation (e) into equation (d):

1
f
1

Inverting equation (f):

1 g

Putting equation (g) into the following equation:

1 h

Preparing the following table:

115
CHAPTER FIVE

0.0 0.0

1.0 ∞

Plot these data in the following figure:

When = 1

0.632

1- The initial slope = 1

116
 PROCESS DYNAMICS

Example 5.2:

A Thermometer having a time constant of 0.1 min, is at steady-state temperature of

[Link] At time t = 0, the Thermometer bath maintained at 100oC. Determine the time
needed for the Thermometer to read 98oC.

Solution:

98 90 8

The size of the step change

100 90 10

10 1

10

8 10 1 .

0.8 1

1 0.8 0.2

-10t =ln 0.2 0.161

Example 5.3:

Consider the system shown in the figure:

117
CHAPTER FIVE

Assumptions:

1- Laminar flow
2- Constant density

Rate of accumulation

Let:

118
 PROCESS DYNAMICS

, time constant

Taking a unit step change in the inflow:

1
1
Making the inversion of the last equation:

3. Second-order systems

2 5.10

Putting this into equation (1):

5.11

Dividing by :

119
CHAPTER FIVE

5.12

Let:

where time constant, and

where damping coefficient

2 5.13

Taking L.T.:

2 1 5.14

5.15
2 1

Putting a unit step function in the input:

1
5.16
2 1

120
 PROCESS DYNAMICS

Equation (5.16) can be transformed from Laplace domain to the time domain
according to the degree of the dampness i.e. the numerical value of . There are three
different values of , these are:

1, 1, 1

We take the characteristic equation of equation (5.15), which is defined as the

denominator of the response in Laplace-domain equated to zero:

2 1 0 5.17

Equation (5.17) is a quadratic equation whose solution is:

√ 4
, 5.18
2

2 4 4
2
2
1
2
1
1 5.19

1
1 5.20

5.2 The degree of dampness:

1. 1 The system is critically damped putting this in equation (5.19)

and (5.20)

121
CHAPTER FIVE

Two equal real roots

2. 1 system is overdamped
In this case 1 real +ve value

We have two distinct real roots

3. 1 system is underdamped

In this case 1 is a complex value and we have two different imaginary

roots.

Response of a second-order systems at the three values of the damping

coefficient:

122
 PROCESS DYNAMICS

1 1:

System is critically damped i.e. fastest response without overshoot.

2 1:

System is overdamped, slow and sluggish, hardly reaches the final value.

3 1:

System is underdamped, overshoots the set point value, oscillates round the set point,
decaying and finally settled.

123
CHAPTER FIVE

Terms used to describe an underdamped second-order response:

1. Overshoot :

exp
1

2. Decay ratio:

2
exp
1

3. Rise time:

the time required for the response to first reach its ultimate value.

4. Response time:

Or settling time or Recovery time

124
 PROCESS DYNAMICS

time for the response to be within 5 of its ultimate value.

5. Period of oscillation: Time for a complete cycle

2
1

5.3 Sensors and transmitters:

The sensor produces a signal, mechanical or electric related to the process variable
it measures. The transmitter in turn converts this signal into a signal that can be
transmitted. Thus the purpose of the sensor-transmitter combination is to generate a
signal. The transmitter output is related to the process variable linearly, this is the
case for example with pressure, level and temperature transmitters.

There are three terms related to the sensor-transmitter combinations:

i. The range of the instrument, which:

Low value high value

125
CHAPTER FIVE

Example is to measure pressure between 20 50 psig, the range of the sensor-

transmitter combination 20 to 50.

ii. The span of the instrument is: 50 – 20 30 psig

The low value 20 psig is often referred to as zero of the instrument, for example
the zero is 20 psig.

iii. The transfer function is usually first-order:

/ 1 , , 1.0

Industrial sensors:

1. Pressure sensors:

Bourdon tube:

One end is sealed, and the other end is connected to the pressure source. The tubing
tends to straighten as pressure is applied, and if the pressure is released the tube tends
to turn back.

126
 PROCESS DYNAMICS

The amount of straightening is proportional to the applied pressure and calibrated

accordingly and thus produces a signal.

. .

2. Flow sensors:

Orifice meter:

Pressure drop 1 2 , change into a signal and calibrated.

DS
T. F. e

pressure drop across the orifice.

127
CHAPTER FIVE

Volumetric flow rate.

Area of orifice.

Orifice coefficient.

Liquid density.

Dimensionless ratio of orifice diameter (d) to the diameter of the pipe.

3. Level sensors:

Three important level sensors are:

i. Differential pressure
ii. Float
iii. Air-bubbler

DS
T. F. e

Differential pressure sensor:

DS
T. F. e

128
 PROCESS DYNAMICS

The float sensor:

Defects the change in buoyant force on a body immersed in the liquid. The force is
then converted into a signal.

DS
T. F. e

The bubbler sensor:

This is a hydrostatic pressure sensor. It consists of an air or inert gas pipe immersed
in a liquid. The air or gas flows through the pipe is regulated to produce a continuous
stream of bubbles. The pressure requires to produce the continuous stream is bubbles
is the measure of the liquid level.

The pressure adjusted to produce a continuous stream of bubble is changed into a

signal.

DS
T. F. e

4. Temperature sensor

Thermocouples:

129
CHAPTER FIVE

Two dissimilar wires produce a signal which is calibrated to temperature.

DS
T. F. e

5. Composition:

Electrophometers:

The strength of the deflection is changed into signal.

Chromatogram:

The vapor pressure is changed into a signal.

DS
T. F. e

130
 PROCESS DYNAMICS

5.4 Final control or control valves elements:

Air-to-close valve – fails open

Air-to-open valve – fails closed

131
CHAPTER FIVE

Example 5.5:

Selection of the action of the control valve:

The flash drum separates vapor and liquid.

1. Should be A-O and fails closed so that there will be no steam, and the liquid
will not boil dry and burns the coil.
2. Should be A-O, fails closed, this will not give any chance for the column to
be completely empty, the operator will have chance to stop the feed rate.
3. (3)Should be Air-to-closed, it fails open so that there will be no pressure
build-up and burst in tubes or the flash drum.

The selection of the fail position of control valves is part of the procedure known
as (HAZAN), hazard analysis. Team of engineers performs such as a procedure
at process design stage.

132
 PROCESS DYNAMICS

Identification of T.F. of a pneumatic valve:

The position of the stem is determined by the balance of all forces acting on it.

Force exerted by the compressed air at the top of the diaphragm; where:

Signal that closed the valve

Area of the diaphragm

This force acting downward

Force exerted by the spring attached to the stem, where:

Stem position moved

Hooke’s constant

C / Friction force exerted upward and resulting from the closed contact of
the stem with valve packing; where C is the friction coefficient between stem and
packing.

133
CHAPTER FIVE

or b

Dividing eq. (b) by K:

Let:

2 d

Taking Laplace transform of eq. (d) at zero conditions:

2 e

f
2 1 2 1

eq. (f) is the transfer function of a pneumatic valve.

But usually, , eq. (c) becomes:

Let:

134
 PROCESS DYNAMICS

Taking Laplace transform of eq. (h):

j
1 1

A pneumatic valve can be taken as first-order lag.

The gain of the valve can be obtained:

Where:

pm= gallon per minute

135
 Chapter Six
Continuous Control Systems
 CONTINUOUS CONTROL SYSTEMS

6.1 Proportional Actions:

R = Forcing function or set point

B = Measured Value

e(t) = is the error or controller input

m(t) = Controller output

P-Action Equation:

6.1

6.2

Taking L.T. of eq. (6.2):

6.3

/ 6.4

136
CHAPTER SIX

The P-Controller transfer function is:

is referred to as the controller gain

The Proportional Band (PB):

1
. 100

Offset:

Offset is the deviation from steady- state upon the introduction of forcing function
i.e. in set point or in load variable and denoted by .

137
 CONTINUOUS CONTROL SYSTEMS

Example 6.1:

By considering the proportional control of the first-order system shown in the figure

a) Determine the offset for a step change with magnitude A in the desired value
b) Determine the offset upon a step change of magnitude α in the load variable.
Solution:

138
CHAPTER SIX

The overall transfer function:

a
1
As the feedback signal is negative eq. (a):

b
1

c
1

1.0 d
1
1
1 1 e
1 1
Putting equations (c) and (e) into eq. (b):

1
/ f
1 1 1

Let:

g
1
and

h
1

Putting eq's (g) and (h) into eq. (f):

i
1
Making step change in the set point:

139
 CONTINUOUS CONTROL SYSTEMS

Putting eq. (j) into eq. (i):

k
1

Offset:

Є l

lim

lim m
1

Є 1 n

But

o
1
Putting eq. (o) into eq. (n):
1
1 p
1 1 1
1
Є q

b) For a step change in the load variable:

Overall T.F:

140
CHAPTER SIX

a
1

b
1

c
1
1
1 1 d
1 1
Putting eq's (c) and (d) into eq. (a):

1
e
1 1 1
1

Let:

Eq. (5) becomes:

f
1

For step change α:

141
 CONTINUOUS CONTROL SYSTEMS

Putting eq. (g) into eq. (f):

1
h
1
The offset:
Є i

For any change in the load variable or disturbance, u, the controlled variable, C(t),
should not be affected if the system behave ideally , that is:
0 j

lim k

Putting the value of C(s) from eq. (h) into eq. (k) and substitute 0:

1
lim l
1

1
Offset:

Є 0.0 m
1 1

1
Є n

Example 6.2:

Effect of proportional gain ( ) on second – order system:

Solution:

142
CHAPTER SIX

The overall transfer function:

1
a
1 1
Where:
overall gain
The characteristic equation is:

1 1 0

1 1 0 b

The standard characteristic equation of a second- order system:

² ² 2ξ 1 0 c
Where ξ = damping coefficient

By comparing eq. (c) and (b):

d
1 1

143
 CONTINUOUS CONTROL SYSTEMS

ξ e
1 1

1
f

1
ξ g

Effect of on damping coefficient:

Є h

1
ξ i

6.2 Integral Action, I- Action:

The integral action is defined as:

6.4

The proportionality constant

Where:

144
CHAPTER SIX

Proportional gain

Integral time

Eq. (6.4) becomes:

6.5

Taking L.T of equation (6.5):

6.6

6.7

Step change in deviation variable:

6.8

6.9

6.10

Putting eq. (6.10) into eq. (6.8)

1
6.11
²

145
 CONTINUOUS CONTROL SYSTEMS

Inverting eq. (6.11) into time domain:

6.12

eq. (6.12) is an equation of straight line with slope =

It is observed that:

0 ∞

is known as reset time

Integral time has the capability of completely eliminating the offset or the residual
error.

146
CHAPTER SIX

Example 6.3:

Determine the offset for an integral control action used as in the following figure
upon a step change in the load variable:

Solution:

a
1

b
1

1.0 c
1

1
1 d
1

Putting eq's (b) and (d) into eq. (a):

147
 CONTINUOUS CONTROL SYSTEMS

/ e
1

The overall T.F

The characteristic equation:

0 g

1 0

The standard characteristic equation of a second-order system is:

² ² 2 1 0 h

Comparing eq. (g) with (h):

2 i

1 1 √
2 2 2

i.e less stable and more under damping.

148
CHAPTER SIX

Offset investigation.

From eq. (f) which is:

Eq. (6) becomes:

For upset in load variable, 0.0

lim .

lim . 00

Є 0.0 – 0.0 0.0

:. Offset is completely eliminated, on the expense of stability

Integral action increases the order of the system by one degree as seen from eq. (f).

149
 CONTINUOUS CONTROL SYSTEMS

6.3 Derivative Action (D- Action):

6.12

6.13

Taking L.T of equation (6.14):

6.14

Derivative Action Response:

Take a step change in the input:

0 6.15

Putting eq. (6.15) into eq. (6.13):

0 6.16

150
CHAPTER SIX

This means the D- action gives the same output m(t), whether e(t) = 0.0 or a constant
D-action alone is incapable for excreting control and a combined PD, PID must be
used.
6.4 Commercial Controller:

These are a combination of two or three modes together. They are usually in form
of:

- P – Only
- PI
- PD
- PID
The method of combination is to combine the relevant mode in a simple controller
such as:
1- Proportional (P), T.F, .

2- Proportional Integral (PI), , T.F, 1

3- Proportional- Derivative (PD), T.F, 1

4- Proportional – Integral – Derivative (PID), T.F,

1
1

Definition:
Derivative time is the time taken by the P-action to produce the initial step of the
D- action.

151
 CONTINUOUS CONTROL SYSTEMS

The derivative action anticipates the proportional action by an immediate step before
the proportional response starts, it shifts P- action a head by a time.

Example 6.4:
Discuss the effect of derivative time by considering the following block diagram
upon a step change in the load variable.

Solution:

152
CHAPTER SIX

The overall T.F:

a
1

b
1 1

1 c
1 1

Where:

The characteristic equation is:

Dividing eq. (e) all through by 1 :

1 f
1 1
The standard Characteristic equation is:

2 1 0 g

Comparing eq. (g) with eq. (f):

h
1

153
 CONTINUOUS CONTROL SYSTEMS

1
i
2 1

For PD-action:

:. Addition of D-action increases the damping coefficient, ξ and therefore provides

greater stability. This allows higher values of to be used within an optimum value
of ξ of 0.2 to 0.3, this reduces the offset and peak deviation.

The main advantage of adding D-action is that it will permit increase in the gain to
reduce the offset to a value that integral control which eliminates the offset is not
required.

Offset:

Taking a unit step change in the load variable

1
k

Putting eq. (k) into eq. (d):

The final value:

lim . m
1

154
CHAPTER SIX

For load variable:

0.0

Є N
1

But,

Є
1

Which is always 1

Addition of D- action keeps the offset to be a fraction.

155
 CONTINUOUS CONTROL SYSTEMS

6.5 Two Position Control:

Basic Requirements:

When the level reaches x, the valve will shift to the higher flow and the level starts
to rise.

When the level reaches y, the valve will shift to the lower level and the level at y
will fall until the level at x and so on.

156
CHAPTER SIX

System dynamics:

1- Rising level:
Flow rate = which is

6.17

2- Falling level :
Flow rate = which is

2 0
6.18

The system is represented by:

Response of level in case of two- position control for the system shown above.
Mass balance:

157
 CONTINUOUS CONTROL SYSTEMS

Generally
[input rate] – [output rate] = Time rate of accumulation or depletion

6.19

Or

Taking L.T of eq. (6.19)

Let (capacitance)

6.20

Oscillation Period, P:
P Rising period Falling Period 6.21

Distance
Period
Velocity
Distance

Velocity

6.22

158
CHAPTER SIX

But

1 0

:. Eq. (6.22) becomes:

1 1
6.22
1 0 0 2 1 0 0 2
Definition of the fig:

Example 6.5:
The level in a tank of 0.75m cross –sectional area is controlled by a two
position controller with a differential zone of 70 mm at a desired value of 1.25
m. The tank is discharged through a constant underflow of 30 L/s and it has
two inflows:
= 36 L/s
= 24 L/s
the system has a delay time of 2.5s:

159
 CONTINUOUS CONTROL SYSTEMS

a) Sketch a complete physical diagram for the system.

b) Plot the response of the level, h against time and determine the period of
oscillation, p in seconds.
Solution:

Data given:

- Delay time, 2.5

- A=0.75m²
- v=1.25 m
- Original differential zone, 1.285 1.215 70
= 36 L/s
- = 24 L/s
- Let the level rising = α

160
CHAPTER SIX

- Let the level falling = β

0.036 0.030
8 10 m/s
0.75
0.030 0.024
8 10 m/s
0.75

Due the time of lag (the effect of dead time) the system will behave as shown
below:

From the fig above:

∆
∆

∆ 8 2.5 20 mm
∆
∆

∆ 8 2.5 20 mm

161
 CONTINUOUS CONTROL SYSTEMS

and are the actual and lower values of the levels (h) in the presence of dead
time .
x and y are the values of the level without dead time , L=0
the actual differential zone , is evaluated as :
∆ ∆
70 20 20 110 mm
From the fig:

∆ ∆
The period of oscillation
1 1

1 1
110 27.5
8 8

162
 Chapter Seven
Controllers Tuning
CHAPTER SEVEN

7.1 Introduction:

How do we choose the values of controller parameters , and ?

They must be chosen to ensure that the response of the controlled variable remains
stable and returns to its steady-state value (disturbance rejection), or move to a new
desired value (set point tracking), quickly. However the action of controller tends to
introduce oscillations.

7.2 Continuous Cycling Method (Ziegler-Nichols Tuning):

In the continuous cycling method, the system (process with controller) is brought
to the edge of instability under Proportional control only. Suitable values of
parameters can then be determined from proportional gain found at that
condition. The procedure is as follows:
1. Close the feedback loop.
2. Turn on proportional action only (equivalent to setting ∞, 0) for
a controller).
3. Increase the controller gain, , until the process starts to oscillate.
Continue and slowly increase the gain until the cycles continue with
constant amplitude.
4. Note the period of these cycles (distance in time between two peaks)
and the value of at which they were obtained (called );
5. Determine the controller settings according to the tuning (Ziegler-Nichols)
rules in table 1.
The settings obtained by this method are good initial estimates but are not optimal
and some retuning may be necessary. Note that this method operates the system at
the brink of instability and if the controller gain is chosen too high during the
tuning procedure, the system will become unstable. The method is therefore not

163
 CONTROLLERS TUNING

recommended for processes in which instability may lead to dangerous situations

(e.g. runway for a reactor).

Start at a low gain; :

2 5

10 4

40

164
CHAPTER SEVEN

Table 7.1: continuous cycling (Ziegler-Nichols) tuning parameters

P 0.5 - -
PI 0.45 -
.

PID 0.6

7.3 Process Reaction Curve Method (Cohen-Coon Tuning):

For some processes, it may be difficult or hazardous to operate with continuous
cycling, even for short periods. The process reaction curve method obtains setings
based on the open-loop response and thereby avoids the potential problem of closed-
loop instability. The procedure is as follows:
i. Disconnect the control loop between the controller and the manipulated
variable (valve);
ii. Make a step change in the manipulated variable (valve opening);
iii. Record the response of the process (should be similar to figure (7.1));
iv. Determine the controller setting according to the tuning (Cohen-Coon)
rules in table 7.2. The parameter is defined as

165
 CONTROLLERS TUNING

∆
, where change in input at steady state
∆

Where ∆ is the final change in the output (the controlled variable) and ∆ is the
initial change in the input (the manipulated variable). The tangent line is drawn as a
tangent to the point of inflection. The dead time and process time are as shown
in figure (7.1).

166
CHAPTER SEVEN

The settings obtained by this method are good initial estimates but are not optimal
and some returning may be necessary as for the Continuous Cycling method.

Table 7.2 Process reaction curve (Cohen-Coon) tuning parameters

P ⁄ 1 ⁄3 - -

PI ⁄ 0.9 ⁄12 ⁄ 30 3 ⁄ 9 20 ⁄ -

PID ⁄ 1.33 ⁄4 ⁄ 32 6 ⁄ , 13 8 ⁄ 4 ⁄ 11 2 ⁄

Simple performance criteria:

The simple performance criteria are based on some characteristic features of the
closed-loop response of the system. The most often quoted are:

167
 CONTROLLERS TUNING

[Link]
[Link] time
[Link] time
[Link] ratio

- Any one of the above characteristics can be used by the designer as basis
criteria for selecting the controller and the values of its adjusted parameters
i.e. ( , ,
From all the above, the decay ratio has been the most popular by practicing
engineers. Experience has shown that a quarter decay ratio is reasonable trade-off
between a fast rise time and a reasonable setting time:

1
Decay Ratio
4

- Take same step change in the set point at P, PI, PID, select the lowest decay
ratio:

168
CHAPTER SEVEN

1
4

7.4 Integral Performance Criteria:

These are based on the entire response of the system.

1- Integral of the square of the error (ISE):

The error

7.1

Where:

Set point

Measured value

169
 CONTROLLERS TUNING

ISE 7.2

2- Integral of the absolute value of the error:

IAE 7.3

3- Integral of the time-weighed absolute error:

IATE 7.4

Procedure:

Select any of (ISE, IAE or ITAE) according to the following guide lines:

i. To suppress large error use ISE

ii. To suppress small error 1.0 , use IAE
iii. To suppress errors that persist for long times use ITAE.

170
CHAPTER SEVEN

Design Relations that Minimize the ITAE:

These relations are based on the first-order plus time delay model of the following
equation:

Note that the optimum controller settings are different depending on whether
step disturbance are made on set point or load variable. Design relations of similar
form have been developed for ISE and IAE. The following table shows the controller
design relation on the ITAE performance index and a first-order plus time delay
Model.

Type of input Type of controller Mode A B

Load PI P 0.859 -0.977
I 0.674 -0.680
Load PID P 1.357 -0.947
I 0.842 -0.738
D 0.381 0.995
Set point PI P 0.586 -0.916
I 1.03 -0.165
Set point PID P 0.965 -0.85
I 0.796 -0.1465
D 0.308 -0.929

171
 CONTROLLERS TUNING

Design Relation:

Where:

Dead time

time constant

For proportional mode

For Integral mode

-For derivative mode

The letter (b) on top of (1.03 , 0.796 , 0.165 , -0.1465 indicate the following:

For the set point changes, the design relation for integral mode is:

Example 7.1:

Use the integral error approach to obtain alternative controller settings for a process
with T.F.:

10
a
2 1

Use PI-Controller with load changes

Solution:

Equation (a) can be put in the following form:

172
CHAPTER SEVEN

The design relation is

For P-Controller:

A = 0.859

B = - 0.977

K = 10

.
0.859 1.69

1.69

.
0.169

For Integral action:

.
0.674 1.08

1.85
.

173
 Chapter Eight
System Stability
 SYSTEM STABILITY

8.1 Introduction:
Routh – Hurwitz Method
The stability of a linear system is determined from the system
characteristic equation.
Consider the following overall transfer function:
1
1
8.1

The characteristic equation is the den of the overall transfer function equated
to zero.
0 8.2
Procedure:
1. First lock at the chr. eq. If there is any negative, the system is unstable,
no further calculations.
2. If all the terms in the chr. eq. are positive, then proceed as follows:
- Number of rows = n+1
- Develop the following array.
Table (1) construction of Rooch array
Row Coefficient
1
2
3 ……
4 ……
1

174
CHAPTER EIGHT

-After arrangement of the coefficients do the following calculations:

Continue in this way down to row

Stability Analysis:
Take the first column of the array

Rule (1):
If there is no change of sign the system is stable, if there is change of
sign the system is unstable. The number of sign changes give the number of
roots of the chr. eq that lie on the RhP (right half plune or S. plane. on Argand
Diagram.

175
 SYSTEM STABILITY

Rule (2):
If the coefficient of the last row is, zero the chr. eq has one root or the origin.
Rule (3):
If the coefficient of the last two rows are zeros, then the chr. eq has a
double roots at the origin.
Example 8.1:
Determine the stability of the system with the following chr. eq.
5 15 10 a
Solution:
Dividing eq. (a) by 5.
3 2 0
Putting into Rooch Array:
Number of rows = 5
1 2
2 0
3 0
0 0
0 0

176
CHAPTER EIGHT

Since the coefficient of the last two rows are zeros, the chr. eq will have two
roots at the origin and the system is critically stable.

Eq. (a) can be put into the following form:

1 2 0
To locate the roots of the chr. eq.
0, , 0, i. e. will lie on the origin
1 0 , 1
2 0 , 2
This means that , will lie on LHP and , on the origin and the system
is critically stable.
Critical stability:

If some of the roots lie on the LHP and others on imaginary axis, the
system is critically stable.

Example 8.2:

3 4 12 0 a

177
 SYSTEM STABILITY

Solution:
Putting eq. (a) in Routh array”
- Number of rows = 4
Table (8.1) matrix (b)
Rows Coefficient
1 1 4
2 3 12
3 0
4 0

3 4 1 12
0
3

To determine , put

where is vey small digital number, say 10 . ThereFore

matrix (b) becomes:

Table (8.2) matrix (c)

Rows Coefficient
1 1 4
2 3 12
3 0
4 0

Completing the array in matrix (c)

178
CHAPTER EIGHT

12
12

Matrix (c) becomes:

Table (8.3) matrix (d)

Rows Coefficient
1 1 4
2 3 12
3 0 0
4 12 0
Taking the first Column:

1
3
0
12

As row n= 0, then row 3 in the first.

Column = 0 the system is critically stable.

To determine the root of the chr. eq. take row (n-1) and equate it to zero.

3 12 0

4 0 4

, 2

These can be located on Argand diagram as shown below:

179
 SYSTEM STABILITY

- The system is critically stable.

Example 8.3:

Given the following chr. eq.:

5 10 10 5 3 0

Using Routh Array, determine whether, the system is stable or unstable.

Solution:

Number of rows = 5+1= 6

Row Coefficient
1 1 10 5
2 5 10 3
3 0
4 0
5 0
6 0 0

180
CHAPTER EIGHT

Example of calculation

5 10 1 10
8
5
5 5 1 3
4.4
5
8 10 5 4.4
7.25
58
7.25 4.4 8 3
1.09
7.25
Taking the first column

1
5 all positive, The system is stable
8 (i.e. no change of sign), all the roots
7.25
of the chr. eq. Lie on the LHP)
1.09
3

Example 8.4:

Given the Following chr. eq.:

2 3 7 4 4 0

Determine whether the system in stable or unstable:

Solution:

Number of rows = 5 1 6

Row Coefficient
1 1 3 4
2 2 7 4

181
 SYSTEM STABILITY

3 0
4 0
5 0
6 0 0
Putting the coefficients of the chr. eq in Routh Array.

1 3 4
2 7 4
0.5 2 0
15 4 0
2.13 0 0
4 0 0

2 3 1 7
0.5
2
5 4 1 4
2
2
0.5 7 2 2
15
0.5
0.5 4 2 0
4
0.5
15 2 0.5 4
2.13
15
0
Taking the first column of the array:
1
2 There is one change of sign, from
0.5 positive to negative, and from negative
15 to positive. the system is unstable, and
2.13 the roots Lie on the RHP
4

182
CHAPTER EIGHT

Example 8.5:

Use Routh criterion to determine the stability of the system which has the
following chr. eq.

4 6 6 9 3 1 0

Solution:

Putting the coefficients of the chr. eq. in Routh Array:

- Number of rows = 6
Row Coefficient
1 4 6 3
2 6 9 1
3 0
4 0
5 0
6 0 0
6 6 4 9
0
6
6 3 4 1 7
6 4
0
Routh Array becomes:

Row Coefficient
1 4 6 3
2 6 9 1

183
 SYSTEM STABILITY

3 7
0 0
3

4
5
6

As the entry of the third row equal to Zero proceed as follows:

Let and substitute in the chr. eq.

the chr. eq. become.

1 1 1 1 1
4 6 6 9 3 1 6

Multiply all through by :

4 6 6 9 3 0

Or

3 9 6 6 4 0

Then put in Routh Array, and proceed normally

Number of rows = 5 1 6

Row Coefficient
1 1 9 6
2 3 6 4
3 14
7 0
3

184
CHAPTER EIGHT

5 7
0 0
6 3
4 0 0

You will find that, There is one change of sign and the system is usable, two
roots Lie on RHP. Other solution, you can put 0 and proceed to complete
the 6 rows.

Example 8.6:

Determine the stability of a system with the following chr. eq.

4 4 4 0 a

Solution:

Number of rows 7 1 8

Routh Array:

Row Coefficient
1 1 1 4 4
2 1 1 4 4

3 0 0 0 0
4
5
6

185
 SYSTEM STABILITY

To complete the number of rows, proceed as follows:

Make an auxiliary eq. from the proceeding row:

4 4 0 b

6 4 8 c

3 2 4 0 d

Replace the zeros by the coefficients of the derivative of the auxiliary eq;
The Array becomes:

Row Coefficient
1 1 1 4 4
2 1 1 4 4

3 3 2 4 0
4 1/3 8/3 4 0
5 22 32 0 0
6 2.182 4 0 0
7 4 0 0 0
There are negatives and the system in unstable, the number of roots that lie
on RHP are two.

8.2 Application of Routh-Hurwitz method as a tuning method:

Routh-Hurwitz method determines whether the system is:
i. Stable, unstable, critically stable
In earlier days when the computer and software are not available, Routh
method is the possible solution to determine the conditions of stability of the

186
CHAPTER EIGHT

system; the method specifies the position of the roots of the characteristic
equation on Argand diagram which clearly tells the condition of stability. On
the other hand the solution of the linear differential equation in Laplace
transform gives exactly the values of stability, which is faster and accurate.
The question is can we consider that Routh criterion for stability is obsolete?
and that it can be superseded by computer solution for the roots of the
characteristic equation.

If Routh criterion is superseded by MATLAB software for the solution of the

roots of the characteristic equation and determination of the conditions of
stability, Routh method is still useful for controller tuning.

Example 8.7:
Use Routh criterion to determine the adjustable parameters for tuning a PID-
controller:
2 3 1 a
Solution:
Putting eq. (a) in Routh-array:
No. of roots: 4 1 5
Routh-array:
Rows coefficient
1 3 1
2 1 0
2.5 1 0
0.5 2
0 0
2.5
1 0 0

187
 SYSTEM STABILITY

For the system to be critically stable:

0.5 2 /2.5 0

0.5 2 0

2 0.5

0.25

This is the ultimate gain: 0.25

To get the ultimate period:

Set:

Putting the value of in eq. (a) using the value obtained for :

2 3 1.25 0 c

Taking the imaginary part of eq. (c) and equate it to zero:

2 0
2
2 1
1/2 0.71 rad/sec
The ultimate period: 2 / 8.845 sec
Use Z-N to tune the PID- controller:
Z-N for PID-controller:
06 , 0.6 0.25 0.15

188
CHAPTER EIGHT

/2, 8.845/2 4.4225 sec

/8, 8.845/8 1.1056 sec

8.3 The Relative Stability:

The verification of stability by Routh criterion provides only partial answer to

the equation of stability. This criterion ascertains the absolute equation
stability of a system by determining if any of the roots of the characteristic
equation lie in right-half of the S-plane. However, if the system satisfies the
Routh criterion and absolute stable, it is desirable to determine the “The
Relative Stability”; that is, it is necessary to investigate the relative damping
of each root of the characteristic equation. Relative Stability is represented by
real part of each root. The more negative stability of each root is clearly
necessary.

Since the relative stability of system is dictated by the location of the root of
the characteristic equation, a first approach using an S-plane formulation is to
extend the Routh criterion to ascertain relative stability.

This can by simply accomplished by utilizing a change of variable, which

shifts the S-plane axis in order to utilize the Routh criterion.

To determine how many zeros (roots) of the characteristic equation lie in right
of some vertical line, a distance | | from the imaginary axis (i.e. to determine
the numbers of zeroes that have a real part greater than where " " may be
positive or negative depending upon whether the new axis the - plane).
Transform the characteristic equation by substituting ( for (S), and
apply Routh’s criterion as described before. The number of changes sign in

189
 SYSTEM STABILITY

the first column for this new equation is equal to the number of zeros which
are located to the right of the vertical line through ( ).

For example the zeros of the polynomial, P(S), are shown in figure.
Construction of the Routhian array would show that there are no roots of
characteristic equation in right half of the S-plane. If the imaginary axis is
shifted to the left by the amount | |there are four roots to the right of the new
axis. This information is available from the Routh criterion when it is applied
after shifting the axis to the new -plane.

The shifting of the S-plane axis in order to ascertain the relative stability of a
system is a useful approach, particularly for high-order systems with several
pairs of closed-loop complex conjugate roots.

Shifting the imaginary axis of the origin to the left by the amount | | requires
a change of variable:

190
CHAPTER EIGHT

; where 0

i.e. substituting ( for (S) in polynomial P(s) in order to get the new
polynomial .

Example 8.8:

Determine the number of the roots that lie to the right of the imaginary axis
of Argand diagram, and then determine the number of the roots that have a
real part 4, for the following [Link].

3 4 6 a

Solution:

(1) Using Routh array:

No. of rows =5

Row coefficient

1 1 4

2 3 6

3 2 0

4 6 0

5 0 0

Since all roots in first column are positive with only one root lies on the
origin, the system is critically stable.

191
 SYSTEM STABILITY

Thus Routh criterion provides only the absolute stability i.e. stable, unstable
or critically stable.

If it is required to determine the relative damping of the system represent by

the above eq. relative stability is to be applied.

(2) To go from Routh to relative stability put:

b
Let 4
4 c

Putting eq. (c) into eq. (a)

13 64 136 104 0 d

(3) Put the new eq. i.e. eq (d) into Routh array:
Row coefficient

1 1 64 104

2 13 136 0

3 53.384 104 0

4 112.676 0 0

5 104 0 0

(4) Shift the imaginary axis by ( as follows:

192
CHAPTER EIGHT

From Routh array after transformation to relative stability there are 4 changes
of sign in the first column there are 4 roots in the RH of the -plane i.e. the
shifted plane.

These roots are 4 0 , note that Routh array will not give any
information of the value of relevant roots, but the relative stability determines
the region in which the roots are located.

193
 SYSTEM STABILITY

Example 8.9:

Determine the number of roots that lie on or to the right of the imaginary axis,
and then determine the number of roots have a real part 4 for the
following [Link]:

F(s) = 2 6 2 5

Solution:

The Routh array:

Rows Coefficient

1 1 6 5

2 2 2 0

3 5 5 0

4 0 0 0

5 5 0 0

Since there is no change of sign over the zero, then there are two roots on the
imaginary axis and the system is critically stable.

For the vertical line through (-4), substituted

4 2 4 6 4 2 4 5 0 a

Expansion of eq. (a)

194
CHAPTER EIGHT

14 78 206 221 0 b
Routh array of eq. (6)
Row coefficient

1 1 78 221

2 14 206 0

3 63.289 221 0

4 157.11 0 0

5 221 0 0

There are 4 changes in the first column of the array, there are 4 roots on the
right half of the -plane where the imaginary axis is shifted to (-4).
Therefore the original polynomial, F(S) =0, has 4 roots to the right of the
vertical line through (-4), or it has 4roots that have a real part -4.
Example (8.10):
Repeat example (8.9) considering a vertical line through (-1) as a test line.
Solution:
1
1 2 1 6 1 2 1 5 0 a
Expand eq. (a):
2 6 8 8 0 b
Routh array of eq. (b)

195
 SYSTEM STABILITY

Rows Coefficient

1 1 6 8

2 2 8 0

3 2 8 0

4 0 0 0

5 8 0 0

From the first column of the array we notice:

1. That there are two changes of sign and therefore two roots of the
original eq. lie to the right of the (-1) line.
2. Since there is no change of sign over zero, it follows that there are two
roots on the vertical line through (-1).
These roots are:
2 6 2 5 0
Set
2 6 2 5 0
Taking the imaginary part:
2 2 0
2 2 0
2 0
1
1

196
CHAPTER EIGHT

At the origin 0

By inspection

1 2 1 2
2 6 2 5

1 2 0

1 2 1 2

197
 SYSTEM STABILITY

1 2 1 2

On Argand diagram:

We determine the region of the location of the roots of chr. eq.

8.4 An Alternative Approach For relative stability:

The procedure can be applied as shown previously by Putting , but

the calculations become tedious as the degree of the polynomial increases.

Simpler producer is discussed below:

Example 8.11:
Determine the number of roots in the following regions:
a) 0 -2
b) 0 - 0.5

198
CHAPTER EIGHT

For the following polynomial:

P 4 42 202 599 1209 1639 1315
450 0 a
Solution:
Routh array:

Row Coefficient
1 4 202 1209 1315
2 42 599 1639 450

3 144.952 1052.95 1272.143 0

4 293.920 1270.395 450 0
5 426.387 1050.217 0 0
6 546.452 450 0 0
7 699.09 0 0 0
8 450 0 0 0

The system is stable; all the roots lie on LHP.

To determine the regions of the roots that lie in the regions (0 to -2) and
(0 to -05):
Construct a new eq. setting:
For S i.e. (S-2) and (S- 0.5):
This involves factorization eq.:

2
2
2

199
 SYSTEM STABILITY

2
2
2
This is a tedious method and time consuming we need to use synthetic
division, the [Link]. Will be:
0 b
Putting eq. (i) into the following format using synthetic division:

From the synthetic successive division of the original polynomial is:

4 14 34 21 17 7 13 0 c

200
CHAPTER EIGHT

Putting eq. (c) into Routh array:

Row Coefficient
1 4 34 17 13
2 14 21 7 0

3 28 15 13 0
4 13.5 13.5 0 0
5 13 13 0 0
6 0 0 0 0
7 0 0 0 0
8 0 0 0 0

1. There are 3 sign changes in the first column, which indicates that 3 roots
lie to right of the shifted axis through x= -2
2. All the elements of the last 2 row are zeros, i.e. there are 3 roots on the
shifted axis, two complex roots together with a third at the origin.
3. The two complex roots can be determined using the auxiliary eq. since
all elements of the sixth row ( are zero, the auxiliary eq. is
obtained from the 5 row or the 4 row ( :
13 13 0
0
1 0
0
,
The transformed poly-nominal, of eq. (c) is reduced to:
4 14 34 21 17 7 13

201
 SYSTEM STABILITY

8.5 The Root. Locus Analysis:

- The root- locus are the plots, in complex plane, of the roots of the OLTF.
- They are very useful to determine the stability of closed-loop system as the
gain changes.
Example 8.12:

- The process T.F. is:

a
1 1

1,

The closed-loop chr. eq. is:

1 1 0 b
1 1

1 0

c
1 1

=k

0
1 1

The roots of the chr. eq

1 0

1 1
1 0

202
CHAPTER EIGHT

As K increases from zero the roots of the chr. eq. of the closed loop is given by:

1 1 0

1 0 d

- The above Eq. is a quadratic eq. whose roots are:

4 1
. e
2

Using the above information we can construct the root locus of the system as
follows:

1. The beginning of the root –Locus correspond to 0 and is given by the

points , 0 and ,0

203
 SYSTEM STABILITY

Examples on the Root –Locus Method:

Evan Rules:

Example 8.13:

Sketch the root –Locus plot for the closed –loop system with the following
charlatanistic equation,

1 3 6 0 a

Solution:

1. Step (1) get the OLTF:

1 3 6 0

Dividing eq. (a) all through by 1 3 :

6
1 0 b
1 3

The closed – Loop charlatanistic eq. is:

1 0 c

Comparing eq. (c) with eq. (b):

6
0 d
1 3

2. Plot poles and zero of the open-loop charlatanistic eq. on s-plane:

6
0 e
1 3

204
CHAPTER EIGHT

1, 3

Mark zeros with 0.

3. Gross over points:

This is an imaginary axis crossing, this will get the critical or ultimate value
of the gain , it will also gives the cross-over frequency. Any gain will
let the roots of the open-loop charlatanistic eq Lie on the RHP indicating that the
system is unstable.

Taking the closed- loop charlatanistic eq.:

1 3 6 0

4 3 6 0

4 3 6 0

4. Putting this in Routh array:

205
 SYSTEM STABILITY

1 3 6
4 0
3 6 0

5. Element of the first column:

1
4
3 6

If is not negative, then it is obvious that all elements of the first column are
+ ve regardless of the value of K and three is no cross-over frequency.

The system is uncondionally stable, it is sable for all values of ., all the roots of
the open-loop charlatanistic eq. Lie on the LHP.

Example 8.14:

Sketch the root –Locus plot for the following systems:

1. 2 3 2 1 0 a

2. 1 2 1 2 0 b

Solution:

Dividing eq. (a) by 2 3 :

2 1
1 0 c
2 3

The charlatanistic eq. of the closed –loop is:

1 0 d

2 1
0
2 3

206
CHAPTER EIGHT

The poles are:

2 0, 2

3 0, 3

The zeros:

1 0, 1

Sketching the root –locus plot:

Cross – over points:

1 2S 1 K S 2 0

2 3 S 1 4K S 4K 0

2 1 4
3 4
0
0

207
 SYSTEM STABILITY

3 K 1 4K 8K
3 K

4K

Rourh array is:

2 1 4
3 4
3 1 4 8
0
3
4 0

Taking row 1

3 1 4 8
0
3

4 5K 3 0

The system is stable for all values of K, form K 0 to K ∞, there is no

crossover frequency.

b) 1 2S 1 K S 2 =0

2
1 0
1 2 1

2 2
1 2 1 1 2 1

The roots an:

0, 0

208
CHAPTER EIGHT

1 0, 1

2 1 0, 0.5

3 0, 3

The cross-over value:

The original chr. eq can be expanded to:

2 3 K S 1 4K S 4K 0 a

Routh array:

2 1 4
3 4
0
0

3 K 1 4K 8K
3 K

The system is critically stable when 0, n 1 row.

209
 SYSTEM STABILITY

3 1 4 8
0
3

3 1 4 8 0

4 5 0

b √ 4
,
2

√25 48
5 0.625 0.599
8

The system become critically stable only when the gain K has a complex value, for
all real values of K the system is stable.

8.6 MATLAB Plotting:

The chr. eq. is 100 1 0

1
1 0
100

1
100

MATLAB format

1 0 1;

1 0 , 1 0 100 ;

210
CHAPTER EIGHT

Plotting Root locus using MATLAB:

MATLAB format:

>> num= [ ];

>> den=[ ];

>> sys= tf(num,den)

>> rlocus(sys)

where:

rlocus(sys): calculates and plots the root locus of the open-loop SISO model sys.

tf: computes the transfer function

num: The numerator of the transfer function

den: The denominator of the transfer function

211
 SYSTEM STABILITY

Example 8.15:

Sketch the root locus for the system with the following OLTF:

1
1 3

Solution:

>> num= [1];

>>den=conv([1 1],[1 3]);

>>sys=tf(num,den);

>>rlocus(sys)

212
CHAPTER EIGHT

Since root locus plot lies completely on the LHP of the complex plane, system is
stable.

Example 8.16:

Sketch the root locus for the system with the following OLTF:

1
3 2 2

Solution:

>> num= [1];

>>a=conv([1 0],[1 3]);

>>den=conv(a,[1 2 2]);

>>sys=tf(num,den);

>>rlocus(sys)

( conv(u,v) is used to multiply two polynomials u and v.)

213
 SYSTEM STABILITY

Since part of root locus plot lies on the RHP of the complex plane,

The system is unstable.

Constructing Routh-Hurwitz Array using MATLAB:

Example 8.22:

For the given F(s) 2 2 6 6 , check the stability of the

system.

Solution:

214
CHAPTER EIGHT

function RSC=routh(poly,epsilon)

poly=[1 1 2 2 6 6];

disp('The Routh array formulated when Row of zeros occur');

dim=size(poly);

coe=dim(2);

RSC=sym(zeros(coe,ceil(coe/2)));

for i=1:coe,

RSC(2-rem(i,2),ceil(i/2))=poly(i);

end

r=coe-2;

index=zeros(r,1);

for i=1:r,

index(r-1+1)=ceil(I,2);

end

for i=3:coe,

if(all(RSC(i-1,:)==0)),

a=coe-i+2;

b=ceil(a/2)-rem(a,2)+1;

t1=RSC(i-2,1:b);

215
 SYSTEM STABILITY

t2=a:-2:0;

RSC(i-1,1:b)=t1.*t2;186

end

for j=1:index(i-2),

RSC(i,j)=-det([RSC(i-2,1) RSC(i-2,j+1);RSC(i-1,1) RSC(i-1,j+1)])/RSC(i-1,1)

end

end

Output: The Routh array formulated when Row of zeros occur

ans =

[ 1, 2, 6]

[ 1, 2, 6]

[ 4, 4, 0]

[ 1, 6, 0]

[ -20, 0, 0]

[ 6, 0, 0]

216
 Chapter Nine
Frequency Domain Dynamics
CHAPTER NINE

9.1 Frequency Response of Linear Processes:

When a linear system is subjected to a sinusoidal input, its ultimate response will
also be a sustained sinusoidal wave. In frequency response we are interested in
determining how the features of the output sinusoidal wave (amplitude ratio and
phase shift) change with the frequency of the input sinusoidal.

The frequency response of the process is defined as steady-state behavior of the

system when forced by a sinusoidal input.

Suppose that the input is m (t) which is introduced as a sine wave to the process:

9.1

Where: = frequency in radian / time

= Amplitude

The output will be

9.2

The output will lag or lead the input, and the phase angle will be negative if the
output lags the input and it will be positive if the output leads the input. The angle
is defined as the angular difference between the input and the output. If is the
angular difference between the input and output:

2 In radians 9.3

360 In degrees 9.4

Where T is the time per unit cycle.

217
 FREQUENCY DOMAIN DYNAMICS

9.2 Basic Theorem:

The frequency response of a system can be found by substituting for s:

s=

The amplitude ratio is:

218
CHAPTER NINE

√ 9.5

The phase angle or phase shift is:

1
9.6

Example 9.1:

Find the frequency response of a first-order process with the following transfer
function:

a
1
Solution:

Set s=

Substituting in equation (a):

b
1

1 1
. 1–
1 1 1

1
c
1 1

and
1 1

219
 FREQUENCY DOMAIN DYNAMICS

d
1

The phase angle:

tan

1
tan ·
1
1

tan e

Example 9.2:

Find the frequency response of a pure capacitive processes or a pure integrator:

Solution:

Set:

220
CHAPTER NINE

0 b

1
tan tan c
0

tan ∞ 90°

Example 9.3:

Find the amplitude ratio and phase angle of N-noniteracting capacities in series with
the following transfer function:

· a
1 1 1

Solution:

Set:

Recall each transfer function as follows:

221
 FREQUENCY DOMAIN DYNAMICS

The amplitude ratio and phase angle for these first-order processes are:

, ,
1 1 1

tan , tan , tan

The overall AR:

b
1 1 1

The overall :

tan tan tan c

Example 9.4:

Find the frequency response of a second-order system:

a
2 1

Solution:

Set:

2 1

2 1

222
CHAPTER NINE

1 2

Let: 1

tan tan

223
 FREQUENCY DOMAIN DYNAMICS

Putting the value of A and B:

b
1 2

2
tan c
1

Example 9.5:

Find the frequency response of a pure dead time transfer function:

Where D dead time

Solution:

Set:

It is known from trigonometry that:

cos sin b

cos

sin

cos sin √1 1

sin
tan tan

224
CHAPTER NINE

θ tan tan

Example 9.6:

Find the amplitude ratio and phase angle for P, PI, PID Controllers:

i. P –action,

ii. PI-action, 1 )

iii. PID-action, 1 )

Solution:

i. Set:

0
tan tan tan 0 0°

ii. Set:
1 1 )

√ 1

225
 FREQUENCY DOMAIN DYNAMICS

K
tan tan tan

iii. Set:
1 )

1
1 1

1
1

tan tan

9.3 Bode Plot and System Stability:

The Bode diagram in honour of [Link] gives a convenient method to represent

the frequency response characteristics of a system. It represents the amplitude ratio
and phase angle of the response of the system as the function of the frequency. It
shows the variation of the logarithm of the amplitude ratios with the frequency and
the variation of the phase shift with the frequency. To cover a large range of
frequencies, the log scale is used for the frequency.

226
CHAPTER NINE

Example 9.7:

Use low and high frequency asymptotes to plot Bode diagram for the first-order lag:

Solution:

Set:

tan

For simplicity let =1

.
1

Low frequency asymptote:

0 or 0

log 0.5 log 1 0.5 log 1 0

High frequency asymptote:

.
1

As ∞

227
 FREQUENCY DOMAIN DYNAMICS

log 1 log

This is an equation of straight line with slope 1

If 1

log = log 0

The straight line passes through the points ( = 1, AR = 1)

Bode phase angle:

0.0 0.0

∞ 90°

System Stability:

When you analyze the plot, if the phase angle does not reach 180°, the system is
stable.

Example 9.8:

Plot Bode diagram for the second-order system

228
CHAPTER NINE

2 1

Solution:

Set:

The amplitude ratio

1 2

The phase angle

tan

Low frequency asymptote

High frequency asymptote

As ∞, will dominate:

Log log 2 log

This is an equation of straight line of slope = -2 and it cuts the y-axis at , and

229
 FREQUENCY DOMAIN DYNAMICS

τω =1

If 1

log 2 log

The phase angle:

0.0 0.0

180°

∞ 360

230
CHAPTER NINE

At the phase angle of -180° the frequency is equal to the cross over frequency and
AR = 1

From AR =1, and you can obtain, the ultimate gain , and ultimate period

Taking the OLTF:

Then you can tune your controller using Z-N.

231
 FREQUENCY DOMAIN DYNAMICS

Bode diagram of two systems in series:

Example 9.9:

The overall T.F is:

1 6
·
2 1 5 1

These are first-order processes with:

, tan

Setting

·
√ √

log log 6 log log

√ √

The overall phase shift is:

tan 2 tan 5

232
CHAPTER NINE

Example 9.10:

Plot Bode diagram and find the limit of stability and the adjustable parameters
( , , ) using Ziegler-Nicolas Criterion.

233
 FREQUENCY DOMAIN DYNAMICS

Solution:

The overall open loop transfer function is:

∏ of OLTF

∑ of OLTF

All the functions of denominator are first-orders:

tan

· · ·
√ √ √ √

tan 60 tan 30 tan 20 tan 15

log AR log K log log log

√ √ √

log
√

Plot:

AR °

0.0

∞ 90

234
CHAPTER NINE

To get the ultimate gain and ultimate period:

- Extend a horizontal line from (-180°) to cut the phase angle curve at the
point .
- Drop a vertical line from point (a) to cut the x-axis at point (b).
- The reading at point (b) on the x-axis is the cross-over frequency.

At point -180° and the system is critically stable and,

AR = 1

Putting AR =1, the value of you can obtain the ultimate gain:

at AR = 1 and

235
 FREQUENCY DOMAIN DYNAMICS

In this example:

0.0373

168.45
.

5.25

Z-N tuning parameters

Type of controller

P 2.628 - -

PI 2.365 140.125 -

PID 3.154 84.225 21.056

9.4 Computer plotting of Bode diagram:

Example 9.11:

Take the pervious OLTF, use MATLAB to plot Bode diagram and obtain the
ultimate gain and period.

Solution:

OLTF
60 1 30 1 20 1 15 1

Open the MATLAB for plotting:

» num=[1];

236
CHAPTER NINE

» den=conv(conv([60 1],[30 1]),conv([20 1],[15 1]));

» sys=tf(num,den);
» bode(sys),grid;

Fig 4.4: Bode diagram for Temperature Control System

The results of MATLAB are:

At 180° , 0.0373

5.25

Calculate:

168.45
.

Use Z-N to get the adjustable parameters.

237
 FREQUENCY DOMAIN DYNAMICS

9.5 Gain and Phase Margins:

Gain Margin:

The gain margin, GM is a performance specification associated with the frequency

response technique. The gain margin represents the factor by which the total loop
gain must increase to just make the system unstable. The controller gain that yields
a desired margin is calculated by:

Where K is the product of all gains (overall gain) of all other elements in the
loop, or the gain margin:

In more specific form:

Where = ultimate gain at 180° and at which AR = 1

Phase Margin:

The phase margin is the difference between (-180°) and the phase angle at the
frequency for which AR = 1:

180° |

238
CHAPTER NINE

Example 9.12:

The ultimate gain of a controller is determined to be, = 12.8 and the gain margin,
GM = 2 use these data to tune a proportional controller.

Solution:

. .
2 6.4

9.6 Nyquist plots:

A Nyquist plot also called a polar plot of G-plane plot is a plot of amplitude ratio
versus phase angle (polar plot) as the frequency is varied from zero to infinity.

Example 9.13:

Plot the Nyquist diagram of a first-order system

a
1

Solution:

Set

, tan b

When 0

tan 0 0°

239
 FREQUENCY DOMAIN DYNAMICS

1
When

, tan 1 45°
√

Prepare the following table

AR

0.0 0

1 √2 45°

∞ 0 90°

Plot AR versus :

240
CHAPTER NINE

9.7 Nyquist stability criterion:

Rule:

If a polar plot of the total OLTF of the system wraps a round (-1,0) point in the s-
plane as the frequency goes from zero to infinity, the system is closed-loop
unstable.

The following polar plots show that system A is closed-loop unstable where as
system B is closed-loop stable.

- Curve A warps point 1,0 it is unstable

- Curve B does not warp point 1,0 it is stable.

Example 9.14:

Consider the transfer function of three CSTR:

241
 FREQUENCY DOMAIN DYNAMICS

Use a proportion controller to plot the Nyquist plot and determine the stability of the
system.

Solution:

OLTF

The characteristic equation is:

1 0

The roots lie on LHP and the system is open-loop stable and there is no need to
plot Nyquist diagram.

It must be observed that Nyquist plot is a unique characteristic of the system (i.e it
is like a finger print).

242
CHAPTER NINE

9.8 Plotting Bode plot:

Matlab format:

>> num= [ ];

>> den=[ ];

>> sys=tf(num,den)

>> bode(sys)

where:

bode(sys): plots the Bode diagram of the OLTF.

Example 9.15:

Sketch the Bode plot for:

.

>> num=[2];

>> a=conv([1 0],[1 1]);

>> den=conv(a,[1 0.2]);

>> bode(num,den)

243
FREQUENCY DOMAIN DYNAMICS

244
CHAPTER NINE

Example 9.16:

Draw a Nyquist plot for the following open loop transfer function:
1
0.5 1

>> num=[1];

>> den=[1 0.5 1];

>> nyquist(num,den)

9.9 Case studies:

• PH-control of waste water before and after treatment.

• Level and B.O.D. control in constructed wetland, produced water in oil field

• Recycling control of chrome in tannery waste water.

245
 FREQUENCY DOMAIN DYNAMICS

• Control of biodegradation in constructed wetland in tanneries.

• Treatment and recycle of sugar-cane waste-water and control.

• Treatment of solid-waste in Khartoum state, biogas and fertilizer from solid-

waste.

Case study 9.1:

Control of pH in waste-water
Control strategy:

Transfer function identification

• Loop1:

2
0.15 1

0.8 1
1.5 1 0.05 1

246
CHAPTER NINE

The block diagram:

The overall transfer function:

2 0.8 1.6
. .
0.15 1 1.5 1 0.15 1 1.5 1

1.6 1
. .
0.15 1 1.5 1 0.05 1
1.6
0.15 1 1.5 1 0.05 1

The characteristic equation: 1 0

0.15 1 1.5 1 0.05 1 1.6 0

0.01125 0.3075 1.7 1 1.6 0

247
 FREQUENCY DOMAIN DYNAMICS

Using Routh-Hurwitz method:

Putting the characteristic equ. into Routh array:

(After multiplying the ch. equ. by 10) :

0.01125 0.3075 1.7 1 1.6 0

Multiply by 10:

Rows Coefficients

0.1125 17

3.075 10 16

16.63 0.59 0

10 16 0

To get the ultimate gain, let:

16.63 0.59 0 28.2

Using direct substitution method:

• Set in the characteristic equation:

0.1125 3.075 17 10 16 0

Taking the imaginary part:

0.1125 17 0

0.1125 17

248
CHAPTER NINE

17
12.3 rad/sec
0.1125

Substitute the value of cross over frequency. In the real part

3.075 12.3 10 16 0

28.5

The ultimate period:

2 2
0.51 sec
12.3

Using Root-locus and MATLAB

• Get the Open Loop Transfer Function :

16
OLTF
0.1125 3.075 17 10

MATLAP format:

>> num=[16];

>> den=[0.1125 3.075 17 10];

>> rlocus(num,den)

249
 FREQUENCY DOMAIN DYNAMICS

From the figure:

28.2 12.3 rad/sec

2 2
0.51 sec
12.3

Using Bode plot and MATLAB:

16
OLTF
0.1125 3.075 17 10

MATLAP format:

>> num=[16];

250
CHAPTER NINE

>> den=[0.1125 3.075 17 10];

>> bode(num,den)

• From the figure at -180˚ and the phase angle curve read the cross-over
frequency.

12.2 rad/sec
2 2
0.515 sec
12.2

• To determine the Amplitude ratio value AR should be converted from decibels

to absolute.

• This can be done using one of the following:

251
 FREQUENCY DOMAIN DYNAMICS

1. Double click on bode diagram window.

A ‘property editor’ window will appear. Select ‘units’ option and then change
AR from (db) to (abs). Then find the value of AR at from the new curve.

2. Use the formula: Reading in db = 20 log AR

From the figure:

0.0367

1 1
27.25
0.0367

252
CHAPTER NINE

Ziegler-Nichols tuning

• The following is Z-N tuning table:

Type of controller

P 0.5 - -

PI 0.45 -
1.2

PID 0.6
2 8

28.5 27.25 28.9

, 28.21
3 3

0.51 0.51 0.515

, 0.511
3 3

The adjustable parameters are:

P-Controller:

14.105

PI-Controller:

, 0.511
12.69 , 0.426
1.2 1.2

PID-Controller:

16.92

253
 FREQUENCY DOMAIN DYNAMICS

, 0.511
0.256
2 2

, 0.511
0.064
8 8

The response of the system after tuning:

For P-controller:

The closed-loop transfer function:

1.6
0.15 1 1.5 1
1.6
1
0.15 1 1.5 1 0.05 1

1.6 0.05 1
0.01125 0.3075 17 1 1.6

14.105

1.6 0.05 1
0.01125 0.3075 17 1 1.6

1.128 1 22.57
0.01125 0.3075 1.7 23.57

MATLAB is used to plot the step response from the closed loop transfer
function for P, PI, PID:

MATLAB format:

>> num=[1.128 22.57];

>> den=[0.01125 0.3075 1.7 23.57];

254
CHAPTER NINE

>> step(num,den)

• To determine the characteristics of this system: Press the right-click button

and from the menu appeared select the characteristic choice which contain
the following: peak amplitude (overshoot),rise time, steady state value and
settling time. The period of oscillation is determined from the figure.

255
 FREQUENCY DOMAIN DYNAMICS

Case study 9.2:

Constructed wetland

• Procedure and waste-water represent the largest volume waste stream in

production of crude oil. Treatment and disposal of such large volume is of
great concern to the environment. Waste is of great concern to the
environment waste-water from the oil industry contain aromatic
hydrodynamic such as benzene and xylene which are high soluble, neurotoxin
and carcinogenic. Constructed wetland is a method of treatment of waste oil
contents in water produced from oil field. The method is found to be efficient
and may reach greater than 90% biodegradation if it is properly controlled.

256
CHAPTER NINE

Control strategy

Cascade control of level and B.O.D in constructed wetland

Transfer functions Identification

The secondary loop (Slave loop):

3
0.2 1

1
0.5
3 1 1

The primary loop (Master loop):

0.8
0.5
4 1 1

257
 FREQUENCY DOMAIN DYNAMICS

The block diagram of process:

For the secondary loop:

3 1
.
0.2 1 3 1 1

1.5
0.2 1 3 1 1

The closed-loop transfer function:

3
0.2 1 3 1 1 1.5

The characteristic equation: 1 0

1.5
1 0
0.2 1 3 1 1
0.2 1 3 1 1 1.5 0

0.6 3.8 4.25 1 1.5 0

258
CHAPTER NINE

Using Routh-Hurwitz method

Rows Coefficients

0.6 4.2

3.8 1 1.5

4.04 0.237 0

1 1.5 0

To get the ultimate gain, let:

4.04 0.237 0

17.05

• Using direct substitution to get the ultimate period :

Set in the characteristic equation:

0.6 3.8 4.2 1 1.5 0

Taking the imaginary part:

0.6 4.2 0

4.2
2.64 rad/sec
0.6

The ultimate period, :

259
 FREQUENCY DOMAIN DYNAMICS

2 2
2.38 sec
2.64

Comparison between the gain obtained from Routh and direct substitution:

Taking the real part:

3.8 1.0 1.5 0
3.8 2.64 1.0 1.5 0

16.99 (Using Routh-Hurwitz method: 17)

The methods are in agreement.

Using Root-locus and MATLAB:

The Open Loop Transfer Function :

1.5
OLTF
0.2 1 3 1 1

MATLAP format:

>> num=[1.5];

>> den= conv([0.2 1],conv([3 1],[1 1]));

>> rlocus(num,den)

260
CHAPTER NINE

From the figure:

16.7 2.62 rad/sec

2 2
2.39 sec
2.62

Bode plot

MATLAP format:

>> num=[1.5];

>> den= conv([0.2 1],conv([3 1],[1 1]));

261
 FREQUENCY DOMAIN DYNAMICS

>> bode(num,den)

2.63 rad/sec

2 2
2.39 sec
2.63

262
CHAPTER NINE

From the figure:

0.0609

1 1
16.4
0.0609

Ziegler-Nichols tuning:

k , k , k , k , 16.99 17 16.7 16.4

k , 16.77
4 4

263
 FREQUENCY DOMAIN DYNAMICS

Using Z-N tuning controller, the adjustable parameters for proportional

controller:

, 0.5 , 0.5 16.77 8.385

The primary loop:

The reduced block diagram:

3 0.8
. .
0.2 1 3 1 1 1.5 4 1

3 0.4
. .
0.2 1 3 1 1 1.5 4 1

The characteristic equation of the closed-loop of the primary loops: 1 0

3 0.4
1 . 0
0.2 1 3 1 1 1.5 4 1

0.2 1 3 1 1 1.5 4 1 1.2 0

8.385

264
CHAPTER NINE

0.2 1 3 1 1 1.5 4 1 10.062 0

Expand eq.:

2.4 15.8 20.6 58.512 13.578 10.062 0

This eq. can be solved by Routh or by direct substitution method using the same
procedure.

Root Locus method:

The open loop transfer function:

1.2 1
OLTF .
0.2 1 3 1 1 1.5 4 1

10.062
OLTF
2.4 15.8 20.6 58.512 13.578 10.062

MATLAP format:

>> num=[10.062];

>> den= [2.4 15.8 20.6 58.512 13.578 10.062];

>> rlocus(num,den)

265
 FREQUENCY DOMAIN DYNAMICS

From the figure:

2.83 0.847rad/sec

2 2
7.41 sec
2.62

Bode plot:

1.2 1
OLTF .
0.2 1 3 1 1 1.5 4 1

10.062
OLTF
2.4 15.8 20.6 58.512 13.578 10.062

MATLAP format:

>> num=[10.062];

266
CHAPTER NINE

>> den= [2.4 15.8 20.6 58.512 13.578 10.062];

>> bode(num,den)

0.842rad/sec
2 2
7.458 sec
2.62

267
 FREQUENCY DOMAIN DYNAMICS

0.429

1 1
2.33
0.429

The remainder of the procedure is left for the student to solve using the same
procedure.

Case study 9.3:

Environmental pollution monitoring and control in tannery

268
CHAPTER NINE

Treatment technique:

1. Physical treatment (screening)

Impurities such as hair, dirt, dung and flashing are screened.
2. Equalization and sedimentation
The waste from tannery is not uniform and if fluctuates with wide variations,
with both acidic and alkaline characteristics. Hence to even out an
equalization tank of one day capacity may be installed.
Removal of hydrogen sulfide
The sulfide content and consequently H2S may be reduced by aeration.
3. Chemical treatment
Tannery waste can be treated with chemical coagulants such as alum. , CO2 ,
H2SO4 , ferric chloride and lime.
4. Biological treatment:
a. Anaerobic digestion
Degradation of organic matter brought about by the action of micro-
organisms in the absence of air and free oxygen, under controlled
conditions.
b. Anaerobic lagoon
The anaerobic lagoon method of treatment employs the principles of
anaerobic digestion with no gas collection
c. Aerobic treatment
i. Oxidation pond
This is an earthen shallow pond utilizing the sun light and aerobes
for treating organic wastes, detention period is 40 days.
ii. Activated sludge process.

269
 FREQUENCY DOMAIN DYNAMICS

Tricking filtration

The trickling filter is an aerobic biological treatment process. It is a porous bed

made of solid particles of stone or PVC on the surface of it a fixed growth micro-
organism is contained. A microbial film develops on the surface of the bed media
and removes the organic materials from waste water. As this is an aerobic biological
system, oxygen is required for the rapid metabolism of the removal organic matter.
The large surface area of the filter media permits to rapid transfer of O2 by molecular
diffusion from the void space into the liquid layer.

In this process organically polluted waste-water is made to come in contact with

aerobic micro-organisms present in the form of flocculent biological mass terms as
activated sludge.

Aerated lagoon

The aerated lagoon consists of an earthen tank; the aeration is accomplished with
diffused air. Detention time varied from 3 to 5 days.

8. Oxidation ditch

This is an extended aeration system with low organic loadings and high mixed
liquor suspended solids concentration of the order of 400 mg/l .

Treatment of tannery waste containing chrome

• Treatment of tannery waste containing chromium is toxic for micro-organism,

hence cannot be treated by aerobic or anaerobic biological process. For this
purpose chromium recycling is very important.

270
CHAPTER NINE

Control strategy

Transfer functions Identification:

3
0.2 1

3
0.1
3 1 4 1

The primary loop (Master loop):

1.5
0.3 1

0.85 1
4.5 1 2 0.2 1

271
 FREQUENCY DOMAIN DYNAMICS

The block diagram of loop1:

0.9
OLTF
0.2 1 3 1 1 0.1 1

The characteristic equation of the closed-loop of the primary loops: 1 0

0.9
1 0
0.2 1 3 1 1 0.1 1

0.2 1 3 1 1 0.1 1 0.9 0

0.06 0.98 4.22 4.3 1 0.9 0

272
CHAPTER NINE

Putting the characteristic eq. in Routh array:

Rows Coefficients

0.06 4.22 1 0.9

0.98 4.3 0

3.95 1 0.9 0

16 0.882 0 0
3.95

1 0.9 0 0

To determine the ultimate gain value, let:

16 0.882
0
3.95

18.14

Using direct substitution to get the ultimate period:

Let in the characteristic equation:

0.06 0.98 4.22 4.3 1 0.9 0

0.06 0.98 4.22 4.3 1 0.9 0

Taking the imaginary part:

0.98 4.3 0

273
 FREQUENCY DOMAIN DYNAMICS

4.3
2.09 rad/sec
0.98

The ultimate period, :

2 2
3.0 sec
2.09

Using Root-locus and MATLAB:

The open loop transfer function:

0.9
OLTF
0.2 1 3 1 1 0.1 1

0.9
OLTF
0.06 0.98 4.22 4.3 1 0.9

MATLAP format:

>> num=[0.9];

>> den= [0.06 0.98 4.22 4.3 1.9];

>> rlocus(num,den)

274
CHAPTER NINE

From the figure:

18.1 2.14 rad/sec

2 2
2.39 sec
2.14

Bode plot

MATLAP format:

>> num=[0.9];

>> den= [0.06 0.98 4.22 4.3 1.9];

275
 FREQUENCY DOMAIN DYNAMICS

>> rlocus(num,den)

From the figure:

2.09 rad/sec

2 2
3 sec
2.14

276
CHAPTER NINE

From the figure:

0.0.0598

1 1
16.722
0.0598

The remainder of the procedure is left for the student to solve using the same
procedure.

277
 Chapter Ten
Interaction and Decoupling
 INTERACTION AND DECOUPLING OF CONTROL LOOPS

10. Interaction and Decoupling of Control Loops:

10.1 Introduction:

Two characteristics should be clear concerning the design of control systems for
processes with multiple inputs and multiple outputs (M/I) and (M/O):

1. A control system is composed of several interacting control loops.

2. The number of feasible, alternative configurations of control loops is very
large, (n! , n 4 No. of control loops 4 3 2 1 24 .

For example, to control the operation of a flash drum we need a configuration of

four loops which must be selected from 24 possible such loop configuration.

278
CHAPTER TEN

These two characteristics dictate the content of this chapter. In particular, we will
study:

1. The interaction among the control loops of a MIMO processes.

2. The relative-gain array method, which determines how the controlled and
manipulated variables should be coupled to yield control loops with minimal
interaction.
3. The design of special control systems with noninteracting loops.

10.2 Interaction of Control Loops:

Consider a process with two controlled outputs and two manipulated inputs. The
input-output relationships are given by

279
 INTERACTION AND DECOUPLING OF CONTROL LOOPS

10.1
10.2

, , and are the four transfer functions relating the two

outputs to the two inputs. Equations (10.1) and (10.2) indicate that a change in
or will affect both controlled outputs.

Let us form two control loops by coupling with and with as show in
Figure 10.1b. To simplify the presentation, we have assumed that the transfer
functions of the measuring devices and final control elements in both loops are equal
to 1 if and are the transfer functions of the two controllers, the values
of the manipulations are given by:

.SP 10.3

.SP 10.4

To understand the nature of interaction between two control loops and how it arises,
we will study the effects of input changes on the outputs when 1. one loop is closed
and other is open, 2. both loops are closed.

1. One loop closed: Assume that loop1 is closed and loop 2 is open. Also
assume that constant [i.e., 0,

280
CHAPTER TEN

281
 INTERACTION AND DECOUPLING OF CONTROL LOOPS

make a change in the set point . . After substituting eq. (10.3) into eqs. (10.1)
and (10.2) we take

. 10.5
1

. 10.6
1

It is clear, then, that any change in the set point . will affect not only the behavior
of the controlled output , but also the uncontrolled output . The dashed lines in
Figure 10.2a indicate schematically the routes through which . affects the two
outputs. Similar conclusions are drawn if we consider loop 1 open and loop 2 closed.
The situation becomes more complex when both loops are closed.

2. Both loops closed (10.2b): Initially, the process is at steady state with both
outputs at their desired values. Consider a change in the set point .

only, and keep the set point of loop 2 the same (i.e., . 0 . then the
following things happen:
(a) The controller of loop 1 will change the value of in such a way as to
bring the output to new set point value. This is the direct effect of
on through loop1, and is shown schematically by the dashed line in
(10.2b).
(b) The control action of will not only attempt to bring to the new set
point, but will also disturb from its steady-state value. Then the controller
of loop 2 attempt to compensate for the variations in by changing
appropriately the value of the manipulated variable . But a change in
, in turn, affects output . This is an indirect effect of on through

282
CHAPTER TEN

loop 2, and is shown schematically by the dashed line in (10.2b). It

constitutes the essence of interaction between two control loops.

An analogous conclusion as to the loop interaction will be drawn if we consider a

change in . while keeping . the same. In summary, we can make the
following statement, which describes the interaction between two control loops:

The regulatory action of a control loop deregulates the output of another loop (in
the same process), which in turn takes control action to compensate for the variations
in its controlled output, disturbing at the same time the output of the first loop.

Having completed a qualitative presentation, let us now examine the quantitative

ramifications of the interaction between two control loops. Substitute (10.3) and
(10.4) into eqs. (10.1) and (10.2) and take

1 . . 10.7

1 . . 10.8

Solve eqs. (10.7) and (10.8) with respect to the controlled outputs and and take
the following closed –loop input-output relationships:

. . 10.9

. . 10.10

Where:

283
 INTERACTION AND DECOUPLING OF CONTROL LOOPS

and

1 1 1.11

It observed that:

1. Equations (10.9) and (10.10) describe the response of outputs and

when both loops are closed.
2. When 0, there is no interaction between the two control loops.

The closed-loop outputs are given by the following equations:

. .
1 1

The closed-loop stability of the two noninteracting loops depends on the roots of
their characteristic equations. Thus if the roots of the two equations have negative
real parts, the two noninteracting loops are stable.

1 0, 1 0 10.12

3. The stability of the closed-loop outputs of two interacting loops is

determined by the roots of the characteristic equation
1 1 0 10.13

Thus if the roots of eq. (10.13) have negative real parts, the two interacting loops are
stable.

284
CHAPTER TEN

4. Suppose that the two feedback controllers and are tuned separately
(i.e., keeping the loop under tuning closed, and the other open). Then we
cannot guarantee stability for the overall control system, where both loops
are closed. The reason is simple: Tuning each loop separately, we force the
roots of the characteristic equations (10.12) for the individual loops to
acquire negative real parts. But the roots of these equations are different from
the roots of the characteristic equation (10.13) which determine the stability
of the overall system with both loops closed.
5. Normally, we tune the two controllers in such a way that the roots of all eqs.
(10.12) and (10.13) have negative real parts. Such tuning guarantees stability
when both loops are closed [roots of eq. (10.13)], or only one is closed while
the other is open [roots of eq. (10.12)], due to a hardware failure.
6. The previous discussion indicates that the interaction between control loops
is a significant factor and affects in a very profound manner the “goodness”
of a control system. For this reason, a control designer attempt to couple the
manipulated variables with the controlled outputs in such a way as to
minimize the interaction of the resulting control loops. If strong interactions
persist for any possible pairing, the control engineer will design a special
control system which eliminates the interaction (decoupling and loops.

Example 10.1: Interaction of Control Loops in a Stirred Tank Heater

A stirred tank heater. Figure 3 shows the two control loops: loop 1 controls the
liquid level by manipulating the effluent flow rate, and loop 2 regulates the
temperature by manipulating the stream flow rate. Let us see how the two loops
interact:

1. When the inlet flow rate (load) or the desired value of liquid level (set point)
change, loop 1 attempt to compensate for the changes by manipulating the

285
 INTERACTION AND DECOUPLING OF CONTROL LOOPS

value of the effluent flow rate. This, in turn, will disturb the temperature of
the liquid in the tank and loop 2 will compensate by adjusting appropriately

the value of the stream flow rate.

2. If, on the other hand, the temperature of the inlet stream (load) or the desired
value of the temperature (set point) changes, loop 2 will adjust the steam flow
rate to compensate for the changes. This will leave the liquid level
undisturbed.

Thus we notice that loop 1 affects loop 2, but not vice versa. In other words, the
interaction is in a single direction.

286
CHAPTER TEN

Example 10.2: Interaction of Control Loop in a Stirred Tank Reactor

In the CSTR of Figure 10.4, the temperature is controlled by the flow of coolant
in the jacket while the effluent concentration and temperature are at their desired
values.

1. Consider a change in the inlet concentration (load) or the desired

effluent concentration (set point). Loop 1 will compensate for these
changes by manipulating the feed flow rate. However, this change in
the feed rate also disturbs the reactor temperature away from the desired

value. Then loop 2 attempt to compensate for the change in temperature

by varying the coolant flow rate, which in turn affects the effluent
concentration.
2. On the other hand, an attempt to compensate for the changes in feed
temperature (load) or the desired set point of reactor temperature, it also

287
 INTERACTION AND DECOUPLING OF CONTROL LOOPS

causes the effluent concentration to vary. Then loop1 attempts to

compensate for the change in effluent concentration by varying the feed
rate, which in turn disturbs the reactor temperature.

It is clear from the above that loop1 interacts with loop 2 in both directions (unlike
the loops of the stirred tank heater, which interact in a single direction).

Example 10.3: Tuning the Controllers of Two Interacting Loops

Assume that the input-output relationships of a process with two controlled outputs
and two manipulated inputs are given by:

1 5
0.1 1 1
1 2
0.5 1 0.4 1

Solution:

From two loops by coupling with and with . The closed-loop input-
output relationships are given by eqs. (10.9) and (10.10), where

1 5 1 2
, , ,
0.5 1 1 0.5 1 0.4 1

Let the two controllers be simple proportional controllers with

1. Tuning each loop separately: The characteristic equation of loop 1 (when loop
2 is open) is given by

1 1 0
0.1 1
and yield the closed-loop pole

288
CHAPTER TEN

10 1 0
Therefore, when loop 2 is open, loop1 is stable for any value of gain .
Similarly, the closed-loop pole for loop 2 when loop 1 is open is given by
2.5 1 0

and consequently, loop 2 is stable for any value of when loop 1 is open.
2. Tuning with both loops closed: When both loops are closed, the characteristic
equation is given by eq. (1.13) and for this example takes the following form:
2 5 1
1 1 0
0.1 1 0.4 1 1 0.5 1

or

0.02 0.1 3.1 2 1.29 1.3 0.8

2 3.2 1.9 0.5 1 2 3
0 10.14

According to the first test of the Routh-Hurwitz criterion for stability eq. (10.14) has
at least one root with positive real part if any of its coefficients are negative. Thus
the closed-loop behavior of the process is unstable if and take on such
values that make the last term of eq. (10.14) negative [all other terms in eq. (10.14)
are always positive]:

1 2 3 0 10.15

Inequality (10.15) places restrictions on the value that and can take, to
render a stable performance when both loops are closed. This is in direct contrast to
our earlier result [see (10.1) above], whereby all values of and were
acceptable if each loop were tuned separately.

289
 INTERACTION AND DECOUPLING OF CONTROL LOOPS

Note: The allowable range of values that and which render stable
responses when both loops are closed, can be found by applying the second test of
the Routh-Hurwitz criterion. The reader is encouraged to complete this example and
find the range of values of and , which render stable response.

10.3 Relative-Gain Array and the Selection of loops:

We recognized that for a process with N controlled outputs and N manipulated

variables there are ! different ways to form the control loops. Which one is the
best? One way to answer this question is to consider the interactions among the loops
for all ! loop configurations and select the one where the interactions is minimal.
The relative-gain array provides exactly such a methodology, whereby we select
pairs of input and output variables in order to minimize the amount of interaction
among the resulting loops. It was first proposed by Bristol and today is a very
popular tool for the selection of control loops. Let us now study the logic of the
method and present some examples describing its usage.

Definitions

Consider a process with two outputs and two inputs. Then do the following two
experiment:

1. Assume that remains constant (Figure 5a). Introduce a step change in the
input of magnitude ∆ and record the new steady-state value of output
. Let ∆ be the change from the previous steady-state. It is clear that it has
been caused only by the change in . The open-loop static gain between
and when is kept constant is given by:
∆
∆

290
CHAPTER TEN

2. In addition to the static gain computed above, there is another open-loop gain
between and when varies by a feedback loop controlling the other
output, (Figure 5b). Thus, introducing a step change ∆ we record a
change ∆ in the steady-state value of . In general, ∆ will be different
from ∆ for the following reason: The input change ∆ affects not only
but also . Then the control loop attempts to keep constant by
varying , which in turn affects the steady-state value of . Therefore
∆ is compound result of the effects from and . Let the new open-
loop gain between and , when is kept constant by the control loop,

291
 INTERACTION AND DECOUPLING OF CONTROL LOOPS

be given by:
∆
∆

The ratio of the two open-loop gains computed above defines the relative gain, ,
between output and input :

∆ /∆
10.16
∆ /∆

The relative gain provides a useful measure of interaction. In particular:

1. If 0 , then does not respond to and should not be used to

control .

292
CHAPTER TEN

2. If 1 , then does not affect and the control loop between

and does not interact with the loop of and . In this case we have
completely decoupled loops.
3. If 0 1 , then an interaction exists and as varies it affects the
steady-state value of . The smaller value of , the larger the interaction
becomes.
4. If 0 , then causes a strong effect on and in the opposite
direction from that caused by . In this case, the interaction effect is very
dangerous.

In a similar manner as above, we can define the following three remaining

relative gains between the two inputs and two outputs of the process we have
been considering:

∆ /∆
relative gain between y and m
∆ /∆

∆ /∆
relative gain between y and m
∆ /∆

∆ /∆
relative gain between y and m
∆ /∆

The values of these gains can also be used as measure of interaction for the
corresponding cases, in a similar way as it was done above for .

293
 INTERACTION AND DECOUPLING OF CONTROL LOOPS

10.4 Selection of loops:

For a process with two outputs there are two different loop configurations, shown
in figure 10.2. Let us see how we can use the relative gains to select the
configuration with minimum interaction between the loops.

Arrange the four relative gains , , and into a matrix form,

which is known as the relative-gain array.

It can be shown that the sum of the relative gains in any row or column of the array
is equal to 1. Thus

1 1
and 10.17
1 1

Therefore, we need to know only one of the four relative gains, while the other three
can be easily computed. For example, if 0.75, then 0.25 and
0.75.

Depending on the value of , we can distinguish the following different

situations:

1. Then the relative-gain array is

1 0
0 1

and it is obvious that we can have two noninteracting loops: coupled with and
coupled with (Figure 10.2a).

0. Then the relative-gain array is given by

294
CHAPTER TEN

0 1
1 0

The l’s in the off-diagonal elements indicate that we can form two noninteracting
control loops by coupling with and with (Figure 10.2b).

0.5. Then

0.5 0.5
0.5 0.5

and the amount of interaction between the two loops is the same in both
configurations of Figure 10.2. In other word, it does not matter how we couple inputs
and outputs; the degree of interaction remains the same.

0 0.5, say 0.25. Then

0.25 0.25
0.75 0.75

The two larger numbers (i.e. 0.75) indicate the recommended coupling with smaller
amount of interaction. Thus we couple with and with (Figure 10.2b).

005 1, say 0.8. Then

0.8 0.2
0.2 0.8

and the recommended couple is the opposite of the previous case:

couple with and with (Figure 10.2a).

1 . Then 1 and 1 1 . Situations with

relative gains outside the range 0 to1 create difficult control problems. Let us see
why.

295
 INTERACTION AND DECOUPLING OF CONTROL LOOPS

1. Suppose that you couple with and with . The corresponding

relative gains, and are larger than 1. Then from the definitions of the
relative gains, we conclude that:
∆ ∆ ∆ ∆
and
∆ ∆ ∆ ∆

In other words, the response of the outputs is held back by the interaction from
the other loop, and the larger the value of the relative gains above unity, the
larger the “holding back” effect will be. Thus we need larger values for the
controller gains.
2. If you couple with and with , the corresponding values for
and are negative. In this case, the interaction will take the controlled
outputs in the opposite direction from that desired by the control effort and
control will be lost altogether. Therefore, never form loops by coupling to
outputs with negative relative gains.

We can summarize all the forgoing observations with the following rule for
selecting the control loops:

Select the control loops by pairing the controlled outputs with the
manipulated variables in such a way that the relative gains are positive
and as close as possible to unity.

1 for 1, 2, … , summation by rows

The loop selection rule remains the same.

Example 10.4: Select the Loops Using the Relative-Gain Array

Consider a process with the following input-output relationships:

296
CHAPTER TEN

1 5
a
1 0.1 1

0.2 0.8
b
0.5 1 1

Let us compute the relative gains:

Solution:

1. Make a unit step change in (i.e., 1/ ) while keeping (i.e.,

0). Then from eq. (a) we take
1 1
1
Recall the final-value theorem (section 7.5) and find the resulting new steady
state in :
1
, lim lim 1
1
Therefore, ∆ /∆ 1/1 1.
2. Keep constant under control by varying . Introduce a unit step in .
Since must remain constant (i.e., 0), eq. (b) will tell us by how much
change;
0.2 1
0.8 0.5 1
Substitute this value in eq. (a) and find
1 1 0.2 1
1 0.1 1 0.8 0.5 1

Then, the resulting new steady state for is given by:

1 1 1 0.2 1 1
, lim lim
1 0.1 1 0.8 0.5 1
1.25

297
 INTERACTION AND DECOUPLING OF CONTROL LOOPS

Therefore, ∆ /∆ 1.25/1 1.25 and

∆ /∆ 1
0.8
∆ /∆ 1.25

Using eqs. (10.17), we find 0.2 and 0.8 . It is easy no to

conclude that we should pair with and with to form loops with
minimum interaction. It should be noted that had we selected the loops differently
(i.e., coupled with and with ), the interaction of the loops would have
been four times larger (i.e., 0.8/0.2 4).

Example 10.5: Selecting the Loops in a Mixing process

Two streams with flow rates and and compositions (mole percent)
80% and 20% are mixed in a vessel (Figure 10.6a). We would like to form
two control loops to regulate the product composition x and flow rate F. Let F
and be the two controlled outputs, while and are the two
available manipulated variables. There are two possible control configurations with
different pairings between the inputs and outputs, and they are shown in Figure 10.6b
and c. Which one should be preferred?

Solution:

The steady-state mass balances yield

298
CHAPTER TEN

299
 INTERACTION AND DECOUPLING OF CONTROL LOOPS

Remarks:

1. The relative gains provide a measure of interaction based on steady-state

considerations. Therefore, the rule given above for the selection of loops does
not guarantee that the dynamic inter-action between the loops will be also
minimal.
2. The relative-gain array is square matrix, which implies that the number of
manipulated variables is equal to the number of controlled outputs. Now,
suppose that we have a process with two outputs and three possible
manipulations, , and .There are three possible pairs of manipulated
variables: ( , ), ( , ), and ( , ). Therefore, we can form three
different relative-gain array:

300
CHAPTER TEN

and we need to examine all of them before we can select the set of two loops with
minimal interaction. (Note: In general, , ,etc.)

3. There are two ways to obtain the relative gains of a process: a computational
approach using a steady state input-output model for the process and an
experimental approach. When a steady-state model is available, we can
obtain the numerator and denominator of the relative gain [see eq. (10.16)] in
a very simple manner. This way we can express the relative gains in terms of
the controlled and manipulated variables themselves, which enables us to
evaluate the interaction across a range of operating conditions.
4. For an existing process we can evaluate the relative gains experimentally by
performing the following two experiments:
(a) Experiment 1 (all loops open): Keeping all loops open, make a small step
change ∆ in , keeping constant. Record the changes in the steady-
state values of and (i.e., ∆ and ∆ ). Then compute
∆ ∆
and
∆ ∆

Return the system to the initial steady state and repeat the same experiment
by varying by ∆ . Record the changes ∆ and ∆ and compute
∆ ∆
and
∆ ∆

(b) Experiment 2(one loop closed); Make a small change ∆ in while

keeping constant by feedback control through . Record the change ∆
in the steady state of and compute the gain
∆
∆

301
 INTERACTION AND DECOUPLING OF CONTROL LOOPS

Repeat the same experiment, but now keep constant through a control loop
with . Record the change ∆ and compute the gain
∆
∆

Similarly, we can compute the following two gains:

∆ ∆
and
∆ ∆

Taking the ratios of the corresponding gains in experiments 1 and 2, we can

compute the relative gains , , and .[Note: Remember that do not need
to compute all relative gains, since they are related by eq. (10.17).]

5. The definitions of the relative gains and their use in selecting the control
loops are not limited to systems with two inputs and two outputs. The
extension to general processes is straightforward. Thus the relative gain
between an output and a manipulation is defined by:
∆ /∆
∆ /∆
The subscript denotes constant values for all manipulations except
(i.e., all loops open), while subscript indicates that all outputs except are
kept constant by the control loops (i.e., all loops are closed). Similarly, the
relative-gain array is given by:

...
...
...................
...

The Entries of satisfy the following two properties:

302
CHAPTER TEN

1 for 1, 2, … , summation by columns

(Note: We have neglected the energy balance because the temperature of the
product stream is not our operating requirement.) The desired steady state for
operational purposes is

mol
200 and 60% by moles
hr

With these values we find the following steady-state solution of eqs. (a) and (b):

133.4 and 66.6

To compute the relative gain between and do the following:

1. Change by one unit (i.e., 133.4 ) while holding 66.6 (the

same). Solve eqs. (a) and (b) for and and find the following new steady
states:

201 and 0.6012

Therefore,

Δ 1 Δ 0.0012
1 0.0012
Δ 1 Δ 1

2. Change by one unit (i.e., 133.4) while holding 60% constant.

Solve eqs. (a) and (b) and find

201.67 and 67.27

Therefore,

Δ 1.67
1.67
Δ 1

303
 INTERACTION AND DECOUPLING OF CONTROL LOOPS

Consequently, the relative gain between and is

∆ /∆ 1
0.6
∆ /∆ 1.67

It follows easily that the complete relative-gain array is

0.6 0.4
0.4 0.6

We can draw two main conclusions:

1. The two loops with minimum interaction are formed when we couple with
and with (Figure 10.6b).
2. Although the interaction between the two selected loops is smaller than that
of the alternative configuration (Figure 10.6c), it is still significant. Thus any
control action to regulate will seriously disturb , and vice versa.

Example 10.6:
Consider the transfer function model:

F R
2 1.5
xd
10 1 1
1.5 2
T
1 10 1
Use the RGA approach to determine the recommended controller pairing based on
steady state considerations.

Manual Solution:
The corresponding steady state gain matrix is:

304
CHAPTER TEN

2 1.5
1.5 2

1
1 12 21
11 22

1
1.5 1.5 0.64
1 2 2

0.64 0.36
Λ
0.36 0.64

The recommended pairings are:

y Vs m

y Vs m

Solution using MATLAB:

1. Enter the steady state gain matrix (k) to the command window:

>> k = [-2 1.5; 1.5 2];

2. Determine the inverse of the steady state gain matrix (b):

>> b=inv(k)

b=

0.3200 0.2400
0.2400 0.3200

3. Obtain the transpose of the inverse of the steady state gain matrix (c):

>> c=b'

c=

305
 INTERACTION AND DECOUPLING OF CONTROL LOOPS

0.3200 0.2400
0.2400 0.3200

4. Multiply each element in the transpose matrix by the corresponding element

in the steady state gain matrix, this yields the relative gain array:

>> RGA= c.*k

RGA =

0.6400 0.3600
0.3600 0.6400

The recommended pairings are:

y Vs m

y Vs m

Example 10.7:
Based on the following steady state gain matrix,

0.167 0.117 0.083

0.033 0.067 0.067
1 1 1
Use the RGA approach to determine the recommended controller pairing.

Solution using MATLAB:

1. Enter the steady state gain matrix (k) to the command window:

>> k=[0.167 -0.117 0.083;-0.033 0.067 -0.067; 1 1 1];

2. Determine the inverse of the steady state gain matrix (b):

>> b=inv(k)

306
CHAPTER TEN

b=
7.4214 11.0767 0.1262
1.8830 4.6522 0.4680
5.5383 15.7288 0.40583
3. Obtain the transpose of the inverse of the steady state gain matrix (c) :

>> c=b'

c=

7.4214 1.8830 5.5383

11.0767 4.6522 15.7288
0.1262 0.4680 0.40583

4. Multiply each element in the transpose matrix by the corresponding element

in the steady state gain matrix, this yields the relative gain array:

>> RGA= c.*k

RGA =

1.2394 0.2203 0.4597

0.3655 0.3117 1.0538
0.1262 0.4680 0.40583

The recommended pairings are:

y Vs m

y Vs m

y Vs m

307
 INTERACTION AND DECOUPLING OF CONTROL LOOPS

10.5 Design of Noninteracting Control Loops:

The relative-gain array indicates how the inputs should be with the outputs to
form loops with the smaller amount of interaction. But the persisting interaction,
although it is the smallest possible, may not be small enough. In such a case, the two
control loops still affect each other’s operation very seriously, and the overall control
system is characterized as unacceptable.

When the designer is confronted with two strongly interacting loops, he or she
introduces in the control system special new elements called decouplers. The
purpose of decouplers is to cancel the interaction effects between the two loops and
thus render two noninteracting control loops. Let us now study how we can design
the decouplers for a process with two strongly interacting loops.

Consider the process whose input-output relationships are given by eqs. (10.1)
and (10.2). From two control loops by coupling with and with (see
Figure 10.1b).

Assume that initially both outputs are at the desired set point values. Suppose that
a disturbance or set point change causes the controller of loop 2 to vary the value of
. This will create an undesired disturbance for loop 1 and will cause to deviate
from the desired value. However, we could change by such an amount as to
cancel the interaction effect from . But the question arises: How much should we
change ?

From eq. (10.1) we find that in order to keep constant (i.e., 0), should
change by the following amount:

308
CHAPTER TEN

10.18

which uses the value of as input and provides as output the amount by which we
should change , in order to cancel the effect of on . This dynamic element
called decoupler and when it is installed in the control system (Figure 10.7a) it
cancels any effect that loop 2 might have on loop 1, but not vice versa.

To eliminate the interaction from loop 1 to loop 2, we can follow the same
reasoning as above and we find that the transfer function of the second decoupler is
given by:

10.19

309
INTERACTION AND DECOUPLING OF CONTROL LOOPS

310
CHAPTER TEN

The block diagram of the process with two feedback control loops and two
deccouplers is given in Figure 10.7b.

From the block diagram of Figure 10.7b it is easy to develop the following two
closed-loop input-output relationships:

/
. 10.20
1 /

/
. 10.21
1 /

The last two equations demonstrate complete decoupling of the two loops since the
controlled variable of each loop depends only on its own set point of the other loop.
Figure 10.7c shows the net block diagram of the two noninteracting loops describe
by eqs. (10.20) and (10.21). It is completely equivalent to that Figure 10.7b.

Remarks:

1. Two interacting control loops are perfectly decoupled only when the process
is perfectly known, because only in this case are the transfer functions ,
, and known exactly. Since this requirement is rarely specified in
practice, the decouplers, offer only partial decoupling, with some weak
interaction still persisting between the two loops.
2. As we have mentioned repeatedly, chemical processes are mostly nonlinear
and nonstaionary (i.e., the values of their parameter change). Therefore, even
if the decoupling is initially perfect, as the desired operating conditions
change, the decoupling deteriorates. One solution to this problem is to use
adaptive decouplers. Thus as the process changes, we estimate the new
transfer function , , and and compute new decouplers. How to

311
 INTERACTION AND DECOUPLING OF CONTROL LOOPS

design new adaptive decouplers is quite complex and goes beyond the scope
of this text.
3. Perfect or very good decoupling allows independent tuning of each controller
without risking the stability of the overall system.
4. A close examination of Figure 10.7b reveals that for all practical purposes the
decouplers are essentially feed forward control elements. Thus decouplers
measures the changes in and takes appropriate action to cancel the effect
that would have on before it has been felt by .
5. If the decouplers are designed using steady-state models for the process, we
talk of steady-state or static decoupling. Equations (10.18) and (10.19)
provides the design dynamic decouplers. It should be emphasized that of
severely interacting loops, static decoupling is better than no decoupling at
all.
6. For a general process with two inputs and two outputs, we need two
decouplers to produce noninteracting loops. Whenever we use only one
decoupler .despite the fact that two are needed, we talk about partial or one
way decoupling. Such systems allow the interaction to travel in one direction.
Figure 10.7a shows a partial decoupling of the loops. Thus disturbance
entering loop 2 cannot disturb loop 1 due to decoupler . On the other hand,
disturbances originating in loop 1 may enter loop 2 but cannot be returned.

Example 10.8: Partial Decoupling

Let us return to the mixing process we studied in example 5. Suppose that the
operating requirements allow small variations in the product flow rate F, while
dictating very tight control on the concentration x, leaving the simple feedback
loop to regulate the value of the product flow rate, F.

312
CHAPTER TEN

Solution:

Assuming that x is kept at the desired value of 0.6, eq. (b) yields

0.8 0.2

or

The last equation describes the necessary steady-state decoupler which cancels
any effects that the flow control loop might have on the composition control loop.

Example 10.9: Physically Unrealizable Decouplers

Consider a process whose input-output relationships are given by:

. .
0.5
1 2 1
.
2 1
0.5 1 1

Solution:

From two control loops by coupling with and with . Then the transfer
functions of the two decouplers are given by eqs. (10.18) and (10.19):

1 .
2
0.5 1

1 .
2
0.5 1

313
 INTERACTION AND DECOUPLING OF CONTROL LOOPS

.
Decoupler is physically unrealizable because the term implies that we
need a future value of , which is not available at the present time now.

314
 Chapter Eleven
Control Systems with Multiple Loops
CHAPTER ELEVEN

11. Control systems with multiple loops

11.1 Introduction:

The feedback control configuration involves one measurement (output) and one
manipulated variable in a single loop. There are other simple control configurations
which may be used.

If there are more than one measurement and one manipulated variable or one
measurement and more than one manipulated variables in such case the control
systems with multiple loops may arise. Typical example is the cascade control.

11.2 Cascade control:

In a cascade control configuration we have one manipulated variable and more than
one measurement.

Cascade control for a jacket CSTR (Continuous Stirred Tank Reactor):

Single Loop

315
CONTROL SYSTEMS WITH MULTIPLE LOOPS

Single Loop

316
CHAPTER ELEVEN

Reduction of the block diagram and determination of the overall T.F:

From Figure 11.2b:

First reduce the inner loop:

11.1
1

1 1

The block diagram will be as follows:

317
 CONTROL SYSTEMS WITH MULTIPLE LOOPS

The characteristic equation is:

1 0

1 0

The overall T.F with change in set point can be obtained and analyzed.

Example 11.1:

Consider a process with the following . for its primary and secondary
processes:

,
. .

Use a PI-Controller with = 1 min, and normal feedback control, determines:

a. Which process is faster

b. Determine the ultimate frequency ( ), the ultimate gain and the ultimate
period .

318
CHAPTER ELEVEN

Solution:

1 1
OLTF 1 a
0.1 1 0.5 1 1

a) The secondary process is faster ( = 0.1) than the primary process.

319
 CONTROL SYSTEMS WITH MULTIPLE LOOPS

b) The cross-over frequency can be found from the equation when we set the
total phase lag 180°

Taking each term in equation (a) and get the phase angle (all are first-order except
term (1 )

For a first-order system:

b
1

tan c

For 1

Put:

1 1
1 1 d

tan tan

But 1 min (given)

320
CHAPTER ELEVEN

1
1 e

tan f

Getting the total phase angle at the cross-over frequency:

tan tan 0.1 tan 0.5 tan

180°

By trial and error

4.45 rad / min

1 · · ·
. . √

The ultimate gain is:

4.45

1 1 · · ·
. . . . . .

11.88

11.88 , 1.39 min

.

Another solution : Use bode plot to and

321
 CONTROL SYSTEMS WITH MULTIPLE LOOPS

Example 11.2:

Show the procedure how to tune the cascade control system using the following
diagram:

Solution:

1. Get the OLTF for the secondary-loop using P-action

1
OLTF a
0.1 1

Proportional-action

From eq. (a)

θ tan tan 0.1

322
CHAPTER ELEVEN

For the cross-over frequency

180° tan 0.1

Plotting-Bode, there is no ( 180° no cross-over frequency

Large gain can be used to make the response faster without crossing over the zone
of stability.

2. Get the overall OLTF:

OLTF b
1

Determine for the overall OLTF using eq. (b)

3. Select a value of for the secondary-loop. From the AR of the overall

OLTF determine the value of =

Use Z-N and , to get the adjusted parameters.

You can get and from bode plot.

2.3 Feed forward Control

1. Feedback control systems:

323
 CONTROL SYSTEMS WITH MULTIPLE LOOPS

It measures the controlled variable, sends a signal to the controller to make the
appropriate correction to track the desired value (S.P.).

2. Feed forward control:

324
CHAPTER ELEVEN

It measures the inlet stream and adjusts appropriately the manipulated variable so
that the manipulated variable will track the set point value before it enters the
process.

Example 11.3:

Heat- Exchanger

Objective: To heat cold water to a specific temperature

Feed forward- feedback control:

325
 CONTROL SYSTEMS WITH MULTIPLE LOOPS

- The feedback loop measures the temperature of the controlled variable and
sends a signal to the controller; the controller output will track the s.p. On the
other hand the feed forward-loop measures the temperature of the feed before
it enters the tank heater and sends a signal to controller 2. Controller2 will
track the set point and sends a signal to a summing point where it will be added
with the feedback signal, the resulting signal will adjust the flow rate of the
heating media so as to keep the outlet temperature as near as possible to both
set points.

326
CHAPTER ELEVEN

11.4 Dead time and dead time compensation:

All the dynamic components of the closed-loop may have significant time delays
in their response, thus:

1. Process may have transportation of fluids over long distance.

Transportation lags (dead time)

2. The measuring element may require long period of time for sampling and
analysis such as the gas chromatograph.
3. The final control element may need some time to develop the actuating signal.
4. A controller whether analog or digital may need time to take the proper control
action.

327
 CONTROL SYSTEMS WITH MULTIPLE LOOPS

Dead time compensation

Consider the OLTF:

11.2
0.5 1

If ‘ 0.01 min

- The cross-over frequency, 160 rad/ min

- The ultimate gain, 80.01

1. If is increased to , 0.1 min

Then 17 rad/ min

8.56
We notice that the increase in dead time has introduced significant additional phase
lag, which reduces the crossover frequency and the maximum allowable gain. In
other words, the increase in dead time has made the closed-loop response more
sensitive to periodic disturbances and has brought the system closer to the brink of
instability.

2. A further increase in dead time (say, 1.0) yields a crossover frequency

2.3 rad/ min and 1.52, the same trend as above is observed.

328
CHAPTER ELEVEN

Dead time compensation:

Assume that, (all dead time is caused by the process).

- For simplicity, 1
Then the open-loop response:

- To eliminate the dead time we would like to have an open-loop feedback

signal that carries current and not delayed information.
The following block diagram will cancel the dead time; an intelligent device
can perform the task:

329
 CONTROL SYSTEMS WITH MULTIPLE LOOPS

11.3

a 1 11.4

b a 11.5

b 11.6

b 11.7

Equation (11.7) completely removes the dead time term or function.

330
CHAPTER ELEVEN

The feedback control configuration involves one measurement (input) and one
(output) in a single loop. But there may be more than one measurement and one
manipulated variable or one measurement and more than one manipulated variables.

11.5 Override control:

It is a control situation to change from normal control action and attempt to prevent
a process variable from exceeding an allowable upper or lower limit. This can be
achieved through the use of special type of switches. The High Selector Switch
(HSS) is used whenever a variable should not exceed an upper limit, and Low
Selector Switch (LSS) is employed to prevent a process variable from exceeding a
lower limit.

Override control or selective control as it is sometimes called is a form of

multivariable control in which a manipulated variable can be set at any point in time
by one of a number of different controlled variables.

Example 11.4:

The base level in a distillation column is normally controlled by withdrawal of the

residue. A temperature in the stripping selection is held by steam to the reboiler.
Situation may arise where the base level continues to drop even with the valve of the
bottom flow is fully closed, this is due to the fact that the boil rate is greater than the
liquid condensation rate which means that the rate of the steam to the reboiler is very
high. The following figure shows the physical diagram:

331
 CONTROL SYSTEMS WITH MULTIPLE LOOPS

Solution:

If it is required that the low level should not go below 20% of the set point level, this
means that when the level is above 20%, the level controller will work normally
within the set point, at the same time the signal from the level transmitter will always
be reported to LSS. If the level goes below 20%, the LSS switch will override the
level controller and reduce the steam rate so that the column will not boil dry. When
the level goes above 20% the level controller will then work normally and so on.

332
CHAPTER ELEVEN

Block diagram of override control system

11.6 Ratio control:

Ratio control is special type of feed forward control where two disturbances (loads)
are measured and held in constant ratio to each other. It is mostly used to control the
ratio of flow rates of two streams. The stream whose flow rate is not under control
is referred to as wild stream, while the stream under controlled is termed
manipulative stream, the following figure shows the physical block diagram of a
ratio control system:

333
 CONTROL SYSTEMS WITH MULTIPLE LOOPS

11.7 Adaptive control:

Adaptive control system is the system that can adjust its adjustable parameters
automatically in such a way as to compensate for variation in disturbances of the
process it controls.

Examples are decay of catalysts activity in a reactor and overall heat transfer
coefficient in a heat-exchanger due to scales and fouling. These changes lead to a
deterioration of the performance of the process, which was designed and tuned using
some nominal values from Z-N, to cope with these changes, this may need adaptation
of the adjustable parameters continuously and automatically as the changes may
require. This may need adaptive tuning of the controller using any one of the
following performance criteria:

i- - decay ratio

334
CHAPTER ELEVEN

ii- Integral square of the error

iii- Gain and phase margin (Bode plot)

For example, variation in process parameters may lead to decay ratios higher than ,

in such case the adaptive mechanism will adjust the controller parameters in such a
way that the process response gives a decay ratio of .

Gain scheduling adaptive Control:

The following feedback control-loop illustrates the mechanism of scheduling

adaptive control:

Where:

1
1 11.8

11.9
1

335
 CONTROL SYSTEMS WITH MULTIPLE LOOPS

11.10
1

11.11
1

OLTF 11.12

The overall open-loop gain

11.13

In this case the control valve may exhibits a nonlinear character, and in such a case
its gain will depend on the current steady-state.

Suppose it is required to keep the overall gain of the system constant.

The following must be observed:

- The gain and are known exactly

- can be calculated

2.14

Equation (11.14) yields the adapation mechanism of this simple gain scheduling
adaptive controller. The following figure shows the control structure:

336
CHAPTER ELEVEN

11.8 Inferential Control:

Often, the controlled variable cannot be measured directly, therefore we cannot use
feedback. If the disturbance that make the control upset can be measured and that an
adequate process model is made available, we could use feed forward control to keep
the unmeasured output at its desired value or set point.

337
 CONTROL SYSTEMS WITH MULTIPLE LOOPS

If neither the disturbance nor the controlled variables can not be measured,

inferential control must be used.

Consider the following block diagram:

338
CHAPTER ELEVEN

11.15

11.16

Similarly

11.17

From equation (11.17) we can solve for which is the unmeasured disturbance:

11.18

Substituting equation (11.18) into equation (11.16) to find the following relation:

11.19

Equation (11.19) provides the required estimator that relates the unmeasured
controlled variable to the measured quantities likes m and z.

The following figure shows the structure of the resulting control system:

339
 CONTROL SYSTEMS WITH MULTIPLE LOOPS

Note that the estimated value of the unmeasured output plays the same role as the
regular measured output.

11.9 Auctioneering Control:

This type of control select among several similar measurements taking the one with
the highest value and feed it to the controller which acts accordingly.

Example of such type of control is the regeneration of a catalyst. The catalyst in the
catalytic reactor undergoes deactivation as the reaction proceeds due to
carbonaceous deposit on it. It can be regenerated by burning off these deposits with
air or oxygen. To avoid destruction of the catalyst due to high temperature during
the combustion of the deposits, the following auctioneering system is designed:

340
CHAPTER ELEVEN

341
Chapter Twelve
Digital Control
 DIGITAL CONTROL

12.1 Z-Transforms

The use of Z-Transforms offers a very simple an elegant method for solving linear
difference equations that result from the conversion of continuous-to discrete time
models.

Z-Transforms play the same role for discrete-time systems as that played by Laplace
transforms for dynamic analysis and design of continuous open or closed loops
systems.

Definition:

Consider a continuous function (signal) sampled at uniform intervals of period T. let

the sequence of sampled values be:

0 , , 2 …………… 12.1

The Z-Transform of the sequence of sampled values is defined by the equation:

0 , , 2 ∑ 12.2

The Z-Transform of the function y(t) is:

∑ 12.3

From now on:

or will be used for Z-transform of Z-transform of y(t) or any other variable.

Example 12.1:

Find the Z-transform of a unit step function:

F(t) = 1, Z( where =1

342
CHAPTER TWELVE

Solution:

∑ a

Taking sampled data at n=0, 1, 2, 3

1 c

For| | 1, eq. (c) is a finite difference series that converges and gives:

Example 12.2:

Find the Z-transform of an exponential function:

Solution:

Taking n = 0, 1, 2, 3

343
 DIGITAL CONTROL

Again if | | 1

Example 12.3:

Find the Z-Transform of a ramp function

Solution:

Taking n = 0, 1, 2, 3

0 2 3

1 2 3

As 1 2 3

Provided that | | 1

344
CHAPTER TWELVE

Example 12.4:

Find the Z-transform of the sin function:

Solution:

sin a

But:

1
b
2

1
c
2

From the previous solution, eq. (*):

The Z-transform of (

Comparing equation (*) and (c):

1,

1 1 1
2 1 1

1 1 1
2 1 1

345
 DIGITAL CONTROL

1
2 1

sin
1

sin
1 2 cos

Example 12.5:

Determine the Z-transform of a dead time element

Solution:

Eq. (a) is the same as:

Let D kT

346
CHAPTER TWELVE

Let

Putting eq. (d) into eq. (c)

∑ e

n=0 x = -k

Eq. (e) becomes

I
But term I is just the Z-transform of

Linearity of Z-Transforms:

The z-transform is a linear function. thus if and are constant parameters, then:

Z[ =∑

- Final value theorem:

This allows us to compute the final value of a function from its z-transform.
It is known in L.T.:

347
 DIGITAL CONTROL

t ∞ and 0

Final value = lim .

In Z-transform:

For final value theorem z=1

lim lim 1 ]

- The initial value theorem:

t 0, s ∞

Z= ∞

lim lim 1 ]

Z-Transforms of an integral

Consider the following integral:

- Approximate the integral using trapezoidal form of integration as follows:

348
CHAPTER TWELVE

Area under linear line:

– 1

Area = 1

349
 DIGITAL CONTROL

Z-Transform of a derivative:

slope of the tangent

1
Slope
1

Example 12.6:

Obtain the Z-transform of the following function:

Solution:

Using partial functions:

350
CHAPTER TWELVE

1
a
1 1

Multiply both sides of eq. (a) by S and set

0 b

1
1 1

1 0 1 c

Multiply both sides of equation (a) by (S+1) and set S+1= 0, 1

1
1

1 0 1 d

Putting equation (c) and (d) into equation (a)

1 1
f
1

From the table of Z-transform:

1 1
1

351
 DIGITAL CONTROL

1
1

Inversion of Z- Transform:

This is to invert from Z-domain to time-domain, it is symbolized as follows and the

time-domain inverted is discrete:

, , ,………….. 12.4

The following must be observed:

1. The inverse Z-transform yields the values of function at the sampling instants
only.
2. The inverse Z-transform does not help to determine the sampling period T.

Partial fraction expansion:

1. 12.5

Where:

= zero’s of order m

= Poles of order n

2. Expand the

3. 12.6

4. Compute the values of , ,………………

5. Determine the inverse Z-transform for every partial fraction i.e.

352
CHAPTER TWELVE

………… 12.7

12.8

Example 12.7:

Determine the inverse Z-transform by partial fraction of the following Z-transformed

function:

4 3

Solution:

- Convert into a ratio of polynomial by multiplying num and den by

a
1 4 3

By factorization:

b
1 4 3 1 1 3

c
1 1 3 1 1 3

Multiply equation (c) all through by (1 and put (1 =0

1 1
1 1 3 1 3

353
 DIGITAL CONTROL

0
1 3

1 1
1 3 2

Similarly multiply equation (c) all through by (1-3 ) and set (1-3 )=0

1
3

From table of Z-transform (table p--)

is the Z-transform of a step function of magnitude = (

And:

may have the general function which is the Z-transform of an

exponential with =3 and K = 2.

354
CHAPTER TWELVE

Properties of z-Transforms:

Like the Laplace transforms, z-transforms possess certain properties that we will
find very useful in dealing with discrete-time systems. Let us examine these
properties.

Table 12.1 Z-Transforms of various functions

Function in time domain Laplace transform z-Transform

Unit impulse: (t) =1 1 1

Unit step

Ramp:

!
lim 1

Exponential:

355
 DIGITAL CONTROL

12.2 From Continuous-Time to Discrete-Time Systems:

The overall transfer functions using analog system are differential equations inverted
to Laplace-domain which can effectively be used for analysis and design.

The introduction of a digital computer in the process control introduces the need for
handling information on a discrete-time basis at particular time instant. The
following must be observed:

1- Continuous signals must be converted to discrete-time before they enter the

computer. Discrete-time command produced by the computer must be
converted to continuous signals before they can actuate the final control
element.
2- The continuous models such differential equations in time or Laplace-
domains are not convenient to analyze the dynamic behavior of loops with
computer control, discrete-time models are needed.

Sampling Continuous Signals:

- The sampler is a physical switch, which stays closed for a short finite time,
∆t.

356
CHAPTER TWELVE

How to convert continuous signals and models to discrete-time equivalent and vice
versa?

Sampling Continuous Signals

Consider a line carrying a measurement signal, y, which varies continuously in time

as shown.

The line is interrupted by a switch, called a sampler, which closes every T seconds
and remains closed for infinitesimally short period, a time point.

357
 DIGITAL CONTROL

T = Sampling time

When T =1 Second or 1/100 Second

Example 12.8:

Sampling the response of a first-order system with and without dead time?

Solution:

a) Without dead time:

358
CHAPTER TWELVE

At 1

y = 0.632

To be well within the process dynamics of a first-order system take the sampling
time T = 0.1

b) With dead time:

359
 DIGITAL CONTROL

To be well within the dynamic of the process take the sampling time:

0.1

Or 0. 1

Take whichever is smaller.

Example 12.9:

Determine the sampling time, T for an overdamped system.

Solution:

Overdamped system can be represented by a first-order system with dead time.

0.1

0.1

Take whichever is smaller

Example 12.10:

Sampling the oscillating response of a system?

Solution:

This is a case of underdamped system open or closed-loop.

Take steady-state response of a linear system and introduce a sinusoidal input

change. The response after the step change will be as follows:

360
CHAPTER TWELVE

Sampling an oscillating signal more than two time per cycle, (seconds)
2

The expression for the sequence of impulses coming out of the impulse sampler in
Laplace domain is:

This is equivalent to the definition of Laplace Transfer which is:

Comparing equation (b) with equation (a):

361
 DIGITAL CONTROL

Results of Zero-order and first-order Hold elements:

1] Discrete-time signal slowly varying with time:

Signals are changing over a large period of time with almost a constant slope of the
changing signal:

This permits a successful linear extrapolation.

2] Rapidly changing signal of the response:

362
CHAPTER TWELVE

- Long sampling period between two successive time values to average the
response. This period should be shortened as far as possible for improvement.

The transfer functions of hold elements:

1- Zero-order Hold element

2- First-order Hold element:

1 1

12.3 Conversion of Continuous to Discrete-Time models:

Example 12.11:

Convert the transfer function of a PID. Controller to a discrete-time transfer

function:

Solution:

The T.F of a PID controller is:

1
a

The controller action in P-mode for discrete-time (nT) is:

363
 DIGITAL CONTROL

The controller action of the integral-action for discrete-time is approximated

by numerical integration:

For derivative action

We need a numerical evaluation of the derivative, , as shown in the following

figure:

From the figure:

1
1

1
e

Approximately the derivative:

364
CHAPTER TWELVE

1 f

- Consequently the control action of a digital PID Controller is determined by

the following discrete time model:

Eq. (g) is referred to as difference equation due to use of finite difference of the
approximation of integral and derivatives.

Example 12.12:

Determine the discrete time model of a linear first-order process described by:

Solution:

Approximate the derivative by first-order differences

At a given time instant:

- For a Linear first-order system:

365
 DIGITAL CONTROL

Using eq. (a):

1 d

Eq. (d) is a difference equation which is in discrete-time model.

Example 12.13:

Determine the discrete-time model of a second-order process:

2 a

Solution:

We already know that:

For second-order derivative, we have:

366
CHAPTER TWELVE

1
2 c

Replacing the derivatives in eq. (a) by their approximation we will get:

2 2 d

2 1 2 1 f

Eq. (f) represents the discrete model of a second-order process.

12.4 Discrete-time Response of Dynamic Systems:

- Discrete-Digital Control Loop (DDC):

- In continuous analog systems, got the overall transfer in Laplace or time-domain

and we make our analysis of the system such as stability, controller settings and
design.

367
 DIGITAL CONTROL

- The above methodology cannot be used for dynamic analysis of digital control-
loops as these processes are discrete element (Digital-Control algorithm) and
discrete time signals.

These are two primary distinct components whose responses are very important
in the dynamic analysis of control systems

1- Digital-Control algorithm:

This is a discrete element of the DDC-Loop with discrete-time input and output
signal.

2- Process with hold element:

- These are continuous elements of the DDC Loop.

368
CHAPTER TWELVE

- The relationship between and can be expressed in Laplace domain

as they are continuous.
- But how can we describe the sampled value response to discrete-time
changes in the controller output ?

Once we know how to handle separately the response of these systems, we should
closed the loop and examine the dynamic response of the closed-loop digital
control systems.

Response of Discrete Dynamic Systems:

- Discrete is a system whose input and output are discrete-time signals. Such as
a digital control algorithm and digital filter.

- The input-output relationship is given by a discrete-time model or by a difference

equation such as a difference equation of the PID controller which is:

12.9

- A Digital filter is small program residing in the memory of a process control

computer which is used to filter out process or measurement noise from the
sampled response of a process.

369
 DIGITAL CONTROL

- The input and output signals of discrete system are related through a linear
differences equation of the general form:

12.10

, , ,… , And , ,…, = constants

Let Z-transform of the output’s discrete-time values.

̂ Z-transform of the input’s discrete-time values.

Z-transforming of eq. (12.10):

̂ ̂ ̂ ̂ ̂
̂ 12.11

- The transfer function in z-domain =

̂
12.12
̂ 1

is referred to as T.F. of the discrete system. It is noticed that is completely

parallel to the overall T.F. in Laplace domain denoted by .

Example 12.14:

Transform the PID Controller T.F. from continuous to discrete T.F.

Solution:

As developed previously for PID controller:

370
CHAPTER TWELVE

For n = n-1

Subtract eq. (b) from eq. (a) we get:

2 c

Take z-transform of both sides of equation (c)

̂ 1 ̂ ̂ 2 ̂ ̂

1 ̂ 1 ̂ ̂ 1 2 d

1 e

Eq. (e) is the discrete T.F. of the PID Controller and referred as velocity PID control
algorithm.

Example 12.15:

A first-order filter has an input/output relationship:

1 a

Where:

371
 DIGITAL CONTROL

Discrete-time input to the filtering algorithm i.e. measurement signal

corrupted with process or measurement noise.

Discrete-time output from the filter i.e. the signal without noise.

Solution:

The Z-transform of eq. (a) is:

1 b

The T.F. is:

This is the discrete T.F. of a first-order digital filter.

12.5 Discrete-time . . ′ for discrete systems in series:

The overall T.F. is:

12.13

For discrete system with multiple input/outputs, we define the discrete T.F. matrix
as follows:

372
CHAPTER TWELVE

……
……
……

For 2x2 systems:

In Block diagram form:

̂ ̂ 12.14

̂ ̂ 12.15

Example 12.15:

373
 DIGITAL CONTROL

Analysis of the sampled response of pure integrator:

Solution:

The signal from the D/A convertor to the valve and process is discrete, it is must
be changed into continuous by the hold element such as follows:

Taking a zero hold of T.F.:

1
a

The T.F. of the process

The relation between the output and the input:

· c

· 1

374
CHAPTER TWELVE

1 d

From Z-T-Label

e
1

Making a unit step change:

1
·
1 1

f
1

The final value after step change:

lim 1 · ∞

This means that the system reaches no steady-state under a unit step change.

375
 DIGITAL CONTROL

Example 12.16:

Analysis of the sampled data response of first-order process:

Solution:

1
· a
1

376
CHAPTER TWELVE

1 · b
1

Factorization:

c
1 1

Multiply by S and set 0

·
1

· d
1

Multiply by 1 and set 1 0

1 1 1

Eq. (c) becomes:

1 1

f
1 1

From Z-table:

g
1

377
 DIGITAL CONTROL

1
· h
1 1

From Z-table:

in eq. (f)

1
· i
1
1

H 1

1
1 j
1
1

1
1 1
1 k
1 1

Making a unit step change:

378
CHAPTER TWELVE

1
1
1
1 1 1
·
1
1 1

At steady state

1
1 1
lim 1
1 1

Example 12.17:

Make the appropriate digital analysis for two noninteracting first-order processes in
series

379
 DIGITAL CONTROL

a
1 1
Solution:

Using eq. (a) and the hold T.F.:

1
· b
1 1

1
1 c
1 1

1 1 1
1 d
1 1

380
CHAPTER TWELVE

Or

1 f

Taking a sequence of unit step ( ):

1
g
1

To find the S.S value of the process output:

lim lim 1

lim 1

0 0

lim

Example 12.18:

Make an appropriate analysis of second-order process:

a
2 1

Solution:

381
 DIGITAL CONTROL

The T.F. of the hold element is:

1
b

Putting equation (a) into the following form:

c
́ 1 ́ 1

The root of eq. (c) i.e. poles

́ 1 0 ́

́ 1 0 ́

Both real roots

- If the process is second-order; we may have:

1 Critically damped

1 Overdamped

1 Underdamped

- For each case we have different expression:

1
d
́ 1 ́ 1

- Consequently different pulse T. F.

382
CHAPTER TWELVE

2 1 1

e
1 1 2 1

Fractional expansion of eq. (e):

Example 12.19:

Determine the offset for a first-order process under proportion digital control:

T.F. of the process:

, Load T.F.

Assume zero=order hold the pulse T.F. of the hold element is:

The P-Controller T.F.,

Solution:

383
 DIGITAL CONTROL

1
a
1
For P-Controller:

b
The discrete-time Closed-Loop response of a direct digital control loop
(DDC) is:

H
c
1 H 1 H

Putting the above expression into eq. (c):

384
CHAPTER TWELVE

1 1
d
1 1
1 1
1 1

a) Step change in set point (zero problem):

Eq. (d) becomes:

1
e
1 1

Where: b

- Make a unit step change in the set point:

1
f
1

Putting eq. (f) into eq. (e):

1 1
g
1 1 1

The offset:

1 (The magnitude of the step change)

385
 DIGITAL CONTROL

1 1
lim 1
1 1 1
1 1
1 1 1 1

1
1 h
1 1

Example 12.20:

A sequence of impulses enters a continuous process of a transfer function:

Determine:

a) The response if the impulses enter the process without passing through the hold
element upon a unit step change.
b) The response if the impulses enter the process with the hold element upon a unit
step change.

Solution:

a)

386
CHAPTER TWELVE

1
1

1
1

1
1
1 1

387
 DIGITAL CONTROL

b) Process transfer function is:

The hold element T.F. is:

H 1

H 1

= 1

388
CHAPTER TWELVE

The final value

lim lim 1

389
 DIGITAL CONTROL

Example 12.21:

Two noninteracting first-order Lags in series have the process T.F.:

Determine the offset upon a unit step change in the controller output.

Solution:

390
CHAPTER TWELVE

H 1
̂ 1 1

1
1 1

By partial fraction and transforming to Z-domain

H
̂

1 1
1
1
1
1

Taking a unit step change in input:

̂ 1

lim 1 =1

Offset = ̂

1 1

391
 DIGITAL CONTROL

Example 3.22:

A closed-Loop response of a first-order process under proportional digital control

has the following T. F., :

Determine the offset:

a) Upon a unit step change in the set point

b) Upon a unit step change in the load variable.

Solution:

A/D

392
CHAPTER TWELVE

a) Step change in the set point:

H ·

The overall transfer function:

Taking a Unit step change in the set point i.e.

393
 DIGITAL CONTROL

Let

The steady-state of the closed-Loop response i.e.

lim 1

lim 1 ·

The offset = 1

b) For the step change in the load variable (Regulator) problem

Taking a unit step change in the load variable:

394
CHAPTER TWELVE

The Z-transform of after making factorization by partial fractions and

rearrangement:

Let

12.6 Stability Analysis of Discrete-time Systems:

A discrete-time system is considered stable if for every bounded input, it produces a

bounded output, regardless of its initial state.

Consider a discrete-time system with a discrete transfer function given by:

395
 DIGITAL CONTROL

12.16
̂ 1

Using partial fraction expansion:

̂ 12.17

Where:

, ,………, are the roots (poles) of the characteristic equation which is:

1 12.18

For terms with in the bracket; if:

1 Where 1, 2, … ,

The function is bounded when the root lies inside or on a unit circle in the
complex plane as shown in the following figure:

396
CHAPTER TWELVE

Stability in the Z-plane:

The stability of any system is determined by the location of the roots of its
characteristic equation of its transfer function. The characteristic equation of the
continuous system is a polynomial in the complex variable S. If all the roots of this
polynomial lie in the LHP of the S-plane, the system is stable.

The stability of a sampled-data system is determined by the location f the roots of a

characteristic equation that is polynomial in the complex variable Z.

The region of stability in the Z-plane can be found directly from the region of
stability in the s- plane using the basic relationship between the complex variable S
and Z:

397
 DIGITAL CONTROL

12.19

Let the real part of S be and the imaginary of S be =

12.20

Putting eq. (12.20) into eq. (12.19):

12.21

The absolute value of | | 12.22

i. When is negative, it follows that:

| | 1

398
CHAPTER TWELVE

ii. When is positive, it follows that:

| | 1

The LHP of the S-plane maps into the inside of the unit circle in the Z-plane as
shown in the following figures:

Example 12.23:

Determine the conditions of stability of the OLTF with a zero hold and proportional
digital controller and first order process:

399
 DIGITAL CONTROL

Solution:

The open-loop pulse T.F. is:

1 1
H a
1

Let

1
H b

The open-loop characteristic equation is:

But b = which is 1

The root of the open-loop characteristic equation is b which lies inside the unit
circle and the system is open-loop stable.

The closed-loop characteristic eq. of this system is:

1 0 c

1+ 0

1 0

400
CHAPTER TWELVE

There is a single root. It lies on the real axis in the Z-plane and its location depends
on the value of the controller gain, .

Example 12.24:

a) Analysis the stability of a digital control system with the following specifications:
1) The process transfer function is:

2) A zero-hold transfer function is:

3) The transfer function of a digital PI-Controller is:

Where:

T = Sampling period

Proportional gain

Integral time constant

b) Use :

0.10, 1.0, T 1

And determine the roots of the characteristic equation, comment on the stability.

c) Keep:

401
 DIGITAL CONTROL

0.10, 1.0, but change the sampling period, T to be = 50 comment on the

system stability.

d) Keep:

T 1, 0.01, 1.00, Comment on system stability.

Solution:

a) The pulse T.F. of the combination hold is:

1 10
H a
0.1 1 2 1

10 50 1000
1 95 95 b
10 0.5

10 50 1000
H 1 95 95 c
1 1 1 .

The characteristic equation of the closed-loop T.F. is:

1 0 d

H ·

The closed-loop characteristic eq. is:

1 H 0 , where 1

402
CHAPTER TWELVE

1 1 .
1

0 e

The closed loop characteristic equation and the value of the sample period T in
addition to , , , therefore we have to be careful in selecting the sampling
period, T to be very small.

b) Let:

0.10, 1.0, T 1

Putting these values in eq. (e):

1 0.89 0.316 0.321 0

Solving for Z =

0.162, 0.365 0.775, 0.365 0.775

Since the magnitude of all roots is 1, it is concluded that the digital control-loop
is stable.

c) If:

0.10, 1.0, 50

It is found that one the poles moves outside the unit circle and the system is unstable,
this shows clear the destabilizing effect of the sampling period.

d) Keeping:

T = 1 and 1 with 0.01, it is found that the loop becomes unstable when
1.00 and 0.01 .

403
 DIGITAL CONTROL

Example 12.25:

Considering the OLTF to be:

.
.

Determine if the system is closed-loop stable or unstable.

Solution:

The closed-loop characteristic equation is:

1 0

.
1 0
.

1 0.368 10 1 0.368 0

4.95 0.368 0

√
,

b = 4.95, a = 1, c = 0.368

0.076, 4.876

As one root of the characteristic equation has an absolute value 1 the system is
unstable.

For characteristic equation in polynomial Z in higher order, it is not easy to solve for
Z, in such case, put:

and use Routh Array to determine the stability of the system.

404
CHAPTER TWELVE

12.7 Digital controller design:

Introduction:

The figure bellows shows the typical continuous feedback system that we have been
considering so far in the tutorial. Almost all of the continuous controllers can be built
using analog electronics.

The continuous controller, enclosed in the shadow square, can be replaced by a

digital controller, shown below, that performs same control task as the continuous
controller. The basic difference between these controllers is that the digital system
operates on discrete signals (or samples of the sensed signal) rather than on
continuous signals.

405
 DIGITAL CONTROL

Difference types of signals in the above digital schematic can be represented by the
following plots.

406
CHAPTER TWELVE

The purpose of this digital control tutorial is to show you how to use MATLAB to
work with discrete functions either in transfer function or state-space form to design
digital control systems.

Zero-Hold Equivalence:

In the above schematic of the digital control system, we see that the digital control
system contains both discrete and the continuous portions. When designing a digital
control system, we need to find the discrete equivalent of the continuous portion so
that we only need to deal with discrete functions.

For this technique, we will consider the following portion of the digital control
system and rearrange as follows.

407
 DIGITAL CONTROL

The clock connected to the D/A and A/D converters supplies a pulse every T seconds
and each D/A and A/D sends a signal only when the pulse arrives. The purpose of
having this pulse is to require that have only samples u(k) to work on and
produce only samples of output y(k); thus, can be realized as a discrete
function.

The philosophy of the design is the following. We want to find a discrete function
so that for a piecewise constant input to the continuous system H(s), the
sampled output of the continuous system equals to discrete output. Suppose the
signal u(k) and holding it to produce a continuous signal . the sketch below
shows that is held constant at u(k) over the interval to 1 T. This
operation of holding constant over the sampling time is called zero-order hold.

408
CHAPTER TWELVE

The zero-order held signal goes through and A/D to produce the output
u(k) that will be the piecewise same signal as if the discrete signal u(k) goes through

to produce discrete output u(k).

Now we will redraw the schematic, placing in place of the continuous

portion.

409
 DIGITAL CONTROL

By placing , we can design digital control systems dealing with only discrete
functions.

Note: There are certain cases where the discrete response does not match the
continuous response due to a hold circuit implemented in digital control systems.
For information, see lagging effect associated with the hold.

12.8 Conversion using c2d:

There is a MATLAB function called c2d that converts a given continuous system
(either in transfer function or state-space form) to a discrete system using the zero-
order hold operation explained above. The basic command for this in MATLAB is
The sampling time (Ts in sec/sample) should be smaller than 1/(30*BW), where BW
is the closed-loop bandwidth frequency.

Example 12.26: Mass-Spring-Damper:

Transfer Function

Suppose you have the following continuous transfer function

Solution:

410
CHAPTER TWELVE

Assuming the closed-loop bandwidth frequency is greater than 1 rad/sec, we will

choose the sampling time (Ts) equal to 1/100 [Link], create a new m-file and
enter the following commands

M 1;

b 10;

k 20;

s tf ‘s’ ;

sys 1/ M*s^2 b*s k ;

Ts 1/100

sys_d c2d sys, Ts, ‘zoh’

sys_d

4.837e‐05 z 4687e‐05

z^2 ‐ 1.903 z 0.9048

Sample time: 0.01 seconds

Discrete‐time transfer function.

12.9 Stability and Transient Response:

For continuous systems, we know that certain behaviors results from different pole
locations in the s-plane. For instance, a system is unstable when any pole is located
to the right of the imaginary axis. For discrete systems, we can analyze the system

411
 DIGITAL CONTROL

behaviors from different pole locations in the z-plane. The characteristics in the z-
plane can be related to those in the s-plane by the expression

12.23

 T = Sampling time (sec/sample)

 s = Location in the s-plane
 z = Location in the z-plane

The Figure below shows the mapping of lines of constant damping ratio (zeta) and
natural frequency (W ) from the s-plane using the expression shown above.

412
CHAPTER TWELVE

If you noticed in the z-plane, the stability boundary is no longer imaginary axis, but
is the unit circle and unstable when any pole is located outside.

For analyzing the transient response from pole locations in the z-plane, the following
three equations used in continuous system designs are still applicable.

4.6
12.24

1.8
12.25

ln % /100
12.26
ln % /100

Where,

 Damping ratio
 W = Natural frequency (rad/sec)
 Ts Settling time
 Mp = Maximum overshoot

Important: The natural frequency W in z-plane has the unit of rad/sample, but
when you use the equations shown above, the W must be in the unit rad/sec.

Suppose we have following discrete transfer function

1
0.3 0.5

Create an new m-file and enter the following commands. Running this m-file in the
command window gives you the following plot with the lines of constant damping
ratio and natural frequency.

413
 DIGITAL CONTROL

numDz 1;

denDz 1 ‐0.3 0.5 ;

sys tf numDz , denDz , ‐1 ; % the ‐1 indicates that the sample time is

undetermined

pzamp sys

axis ‐1 1 ‐1 1

zgrid

414
CHAPTER TWELVE

From this plot, we see poles are located approximately at the natural frequency of
(rad/sample) and the damping ratio of 025. Assuming that we have a sampling

time of 1/20 sec (which leads to 28.2 rad/sec) and using three equations
shown above, we can determine that this system should have the rise time of 0.06
sec, a settling time of 0.65 sec and a maximum overshoot of 45% (0.45 more than
steady-state value). Let’s obtain the step response and see if these are correct. Add
the following commands to the above m- file and rerun it in the command window.
You get the following step response.

sys tf numDz , denDz , 1/20 ;

step sys ,2.5 ;

415
 DIGITAL CONTROL

As you can see from the plot, the rise time, settling time and overshoot came out to
be what we expected. This shows how you can use the locations of poles and the
above three equations to analyze the transient response of the system.

12.10 Discrete Root Locus:

The root-locus is the locus of points where roots of characteristic equation can be
found as a single gain is varied from zero to infinity. The characteristic equation of
an unity feedback system is

1 0 12.27

Where is the compensator implemented in the digital controller and

is the plant transfer function in z.

The mechanics of drawing the root-loci are exactly the same in the z-plane as in the
s-plane. Recall from the continuous root –locus tutorial, we used the MATLAB
function called sgrid to find the root-locus region that gives an acceptable gain (K).
For the discrete root-locus analysis, we will use the function zgrid that has the same
characteristics as sgrid. The command zgrid (zeta, ) draws lines of constant
damping ratio (zeta) and natural frequency ( ).

Suppose we have following discrete transfer function

0.3
1.6 0.7

and the requirement are a damping ratio greater than 0.6 and a natural frequency
greater than 0.4 rad/sample (these can be found from design requirements, sampling
time (sec/sample) and three shown in the previous section). The following

416
CHAPTER TWELVE

commands draw the root-locus with the lines of constant damping ratio and natural
frequency. Create a new m-file and enter the following commands. Running this m-
file should give you the following root-locus plot.

numDz 1 ‐0.3 ;

denDz 1 ‐1.6 0.7 ;

sys tf numDz , denDz , ‐1 ;

axis ‐1 1 ‐1 1

zeta 0.4;

Wn 0.3;

zgrid zeta, Wn

417
 DIGITAL CONTROL

From this plot, you should realized the system is stable because all plots are located
inside the unit circle. Also, you see two dotted lines of constant damping ratio and
natural frequency. The natural frequency is greater than 0.3 outside the constant-
line, and the damping ratio is greater than 0.4 inside the constant-zeta line. In this
example, we do have the root-locus drawn in the desired region. Therefore, a gain
(K) chosen from one of the loci in the desired region should give you the response
that satisfies design requirements.

12.11 SCADA Acquisition System:

A data aquisition system is used for measurement and processing of plant signal
data before it is displayed on the operator desk or permanently recorded.

A block diagram of a Pc-(computer) based data acquisition is shown below:

418
CHAPTER TWELVE

- Block diagram of data acquisition system

1. Consists of individual transducer (sensor) for measurement of physical
plant parameters, such as temperature, pressure, flow rate, composition.

2. After measurement, the transducer is fed to the signal conditioning device

to bring the signal level up to a sufficient value to make it useful for
conversion, processing, indicating and recording.

A signal conditioner is used to amplify, modify or select certain signals.

3. The output of the signal conditioner is fed to the multiplexing device for
multiplexing, the multiplexed data are converted to digital converter
(A/D), then to the computer for further processing, computation, storage
….. etc. the final and process data can be displayed on display panel or
recorded.

Data aquisition system is generally, a process of collecting input signal in analog or

digital form, as rapidly, as accurately and economically as far as possible.

Definitions:

1. Backup systems:

Backup systems are used to allow the system to safely shutdown, allow for limited
production when the control system fails, and allow for continued plant operation.

2. Redundancy:

Redundancy is normally static in that a backup device does nothing until it is

automatically switch-on because of failure in the primary device.

419
 DIGITAL CONTROL

Redundancy is generally incorporated either by supplying additional elements on a

one-for one basis or by one for many systems.

3. Algorithms:

An algorithm is the description of the mechanization of a particular equation by the

digital computer.

Algorithms are set of several equations that are used in process control field for
implementation of P, PI or PID.

4. Programmable logic controllers (PLCs):

Programmable logic controllers (PLCs) are microprocessors that perform discrete or

continuous control functions in processing plants and factories.

5. Modem:

A modem is a device used to transmit data between computers and workstations.

Modem transforms data from a digital device to analog form suitable for
transmission over the lines.

6. Field-buses:

Field-buses are used to link sensors and actuators to DCS Systems and PLC’s so that
the information they supply can flow into plant information system.

7. Robustness:

The discrepancy between a system and its mathematical model may lead to violation
in closed-loop instability. We say the system is robust if the design performs
satisfactory under variations in the dynamics of the plant.

420
CHAPTER TWELVE

8. Direct Digital Control (DDC):

Typical DAC Configuration

9. Stochastic inputs:

A process typically experiences a continued stream of small and large disturbances,

so that the process is never at an exact steady-state. A process that is subjected to
such random upsets is termed stochastic system.

421
 DIGITAL CONTROL

SCADA System (supervisory Control and Data Aquisition):

SCADA (supervisory Control and Data Aquisition)

Unlike the situation of a direct digital control we may use the computer to change
only the set points or values of adjustable parameters for the local controllers. The
resulting system is known as supervisory control.

422
CHAPTER TWELVE

423
 DIGITAL CONTROL

Assignment:

Example 1:

Develop a discrete time response of a digital PID-controller; whose continuous

output is:

Solution:

A digital controller uses error value at particular time instant:

, ,

It produces control commands at discrete time points:

, , ,

Discrete – time control actions are:

P action

I action

D action

Control action of a PID is:

424
CHAPTER TWELVE

Example 2:

Develop a discrete-time model of first-order process:

Solution:

Approximate the derivative by a first-order difference:

Putting eq. (b) in eq. (a)

425
 DIGITAL CONTROL

Multiply both sides of eq. (d) by T

Divided by

1 h

Eq. (h) is a difference eq.

Example 3:

Develop a discrete-time model of a second-order process:

2 a

Solution:

1
c

1
2 d

426
CHAPTER TWELVE

Putting eq. (b) and eq. (d) into eq. (a)

2
2 e

2 1 2 1 f

Eq. (f) represent a discrete-time model of a second-order process, it is a difference

eq.

Example 4:

Develop a discrete transfer function of a digital PID control algorithm

Solution:

Consider the control command at sampling instants and n-1:

Subtract eq. (b) from eq. (c):

2 d

Take the Z-transform of both sides of eq. (d):

427
 DIGITAL CONTROL

2 e

1 1 2 f

The overall T.F is:

1 1 2

Dividing by and 1

1 1 2
1 h
1 1

Eq. (h) yields discrete T.F of the PID control algorithm.

Example 5:

Develop a discrete time response of a first order digital filter. The input-output
relation of a digital filter is:

428
CHAPTER TWELVE

Discrete time input to the filtering algorithm i.e. the measurement signal
corrupted with process or measurement noise.

Discrete time output from the filter which is free from noise.

a = A constant called filtering parameter.

Solution:

or

1 1

1 1

Eq. (*) yields discrete T.F of a first order digital filter.

Example 6:

Develop a sampled value response of the following:

1-

2- A first order process:

429
 DIGITAL CONTROL

Solution:

The signal coming to the process should be continuous; a hold element must be
added

The pulse T.F is:

But

eq. (*) becomes:

From table

Z-transform putting

430
CHAPTER TWELVE

1
1 1

Make a unit step change in the input

Z-transform of

1
1 1 1

The final value theorem indicates that:

lim lim lim 1

, 0 1

lim 1 lim 1 ∞
1

431
 DIGITAL CONTROL

The system reaches no steady-state continuous integration.

-A first-order process with a hold element:

The pulse function, T.F:

1
1
1 1

From table

1
1 1

432
CHAPTER TWELVE

Or

1
1
1

1 1
, 1

1
1
1 1

1
1
1 1

Consider a sequence of inputs (nT) corresponding to a unit step:

1
1

The overall T.F:

1
1
1

433
 DIGITAL CONTROL

1 1
1
1
1 1

The steady-state value:

lim lim lim 1

1
lim 1
1 1

lim

Example 7:

Solution:

1
1 1

434
CHAPTER TWELVE

1 1
1 a
1 1

1
1
1

1
1
from the table of Z Transform:
1 1

1
1 1

1
1
b
1
1 1

Put (b) in (a):

1 1
1
1 1 1

1
1
1
1
1
1

1
let
1 1

435
 DIGITAL CONTROL

1
1 1

1
c
1 1

Eq. (c) can be rearranged to (Dominator) 33

1 1

1
1 1

1
1

1
1 1

1 1
1 1 1

, 1 ∞ 1

lim 1

lim 1 .

436
CHAPTER TWELVE

1 1
lim
1 1 1 1

1
1 1

1
1
1 1

437
 Chapter Thirteen
Fuzzy Logic Controller
 FUZZY LOGIC CONTROLLER

13. Fuzzy Logic Control:

13.1 Introduction:

There has been a rapid growth in the number and variety of applications of fuzzy
logic:

 Consumer Products, e.g., cameras, camcorders, washing machines,

microwave ovens.
 Industrial Process Control.
 Medical Instrumentation.

Fuzzy Logic is synonymous with the theory of fuzzy sets, a theory which relates to
classes of objects with unsharp boundaries in which membership is a matter of
degree.

The basic concept underlying fuzzy logic is that of a linguistic variable, i.e., a
variable whose values are words rather than numbers.

Much of fuzzy logic may be viewed as a methodology of computing with words

rather than numbers. Although words are inherently less precise than numbers, there
use in closer to human intuition. Furthermore, computing with words exploits the
tolerance for imprecision and thereby lowers the cost solution.

Another basic concept in fuzzy logic is that of a fuzzy if- than rule, or simply, fuzzy
rule, and the calculus of fuzzy rules.

Fuzzy logic does a good job of trading off between significance and precision-
something that humans have been managing for a very long time.

Fuzzy logic is both old and new! The modern and methodical science of fuzzy logic
is still young, yet the concepts of fuzzy logic reach right down to our bones.

438
CHAPTER THERTEN

Fuzzy logic is conventional way to map an input space to an output space, with
emphasis on the word convenient.

13.2 Benefits of Fuzzy Control:

 Suitable for complex III-defined systems where mathematical modeling is

difficult or plant is too abstract e.g. Complex chemical plant, cement-kiln, etc.
 Able to design along linguistic lines – usage of rules based on experience.
 Better performance than conventional PID controller.
 Simple to design.
 Fuzzy logic is conceptually easy to understand. It’s natural!
 Fuzzy logic is flexible. It is easy to change without starting from scratch.
 Fuzzy logic is tolerant of imprecise data. This is built into the process, rather
than added onto the end.
 Fuzzy logic can model nonlinear functions of arbitrary complexity.
 Fuzzy logic can be built on top of the experience of experts. It lets you rely
on the experience of people who already understand the systems.
 Fuzzy logic can be blended with conventional control techniques. Fuzzy
systems do not necessarily replace conventional control methods. In many
cases, fuzzy systems augment them and simplify their implementation.

439
 FUZZY LOGIC CONTROLLER

 Fuzzy logic is based on natural language, that which is used by ordinary

people on a daily basis and which is convenient and efficient. The basis for
fuzzy logic is the basis for human communication. as a result, fuzzy logic is
easy to use.

440
CHAPTER THERTEN

13.3 Designing a fuzzy logic control system:

13.4 A fuzzy logic control systems:

441
 FUZZY LOGIC CONTROLLER

The fuzzy controller has four main components:

 The rule-base holds the knowledge, in the form of a set of rules, of how best
to control the system.
 The inference mechanism evaluates which control rules are relevant at the
current time and then decides what the input to the plant should be.
 The fuzzification interface simply modifies the inputs so that they can be
interpreted and compared to the rules in the rule base.

If the output from the defuzzifier is not a control action for a process, then the system
is a fuzzy logic decision support system.

The fuzzy controller itself is normally a two-input and a single output component. It
usually a MISO system.

442
CHAPTER THERTEN

Component of the Fuzzy Logic Controller:

13.5 Design Process

 Identify controller Inputs and outputs as the fuzzy variables.

 Break up inputs and outputs into several fuzzy sets and label them according
to the problem to be solved and set up the fuzzy variables on the appropriate
universes of discourse (Fuzzification).
 Configure/develop rules to solve the problem.
 Choose inference encoding procedure.
 Choose a defuzzification strategy.
 Tune the adjustable parameters.

A fuzzy logic controller usually has 2 inputs and 1 output for a SISO plant.

The 2 inputs are used for accessing the current conditions of the process such that
the necessary action can be taken by providing the correct signal (which is its
output).

443
 FUZZY LOGIC CONTROLLER

The inputs to the fuzzy controller is usually the error (e) which measures the system
performance and the rate at which the error changes ( ) and the output is the change
of the control signal ( ).

A fuzzifier then transforms the crisp values of e and De into corresponding fuzzy
values (usually there are several fuzzy values of e and De).

From the rule base, the fuzzy values of e and De determine which rules are to be
fired through an inferencing algorithm.

Several values of Du will then be obtained and a defuzzification mechanism will

then transform these values into one crisp value.

The actual control signal is obtained by adding Du to the past value of u which is
send to the plant.

13.6 Fuzzification:

 Involves the conversion of the input/output signals into a number of fuzzy

represented values (fuzzy sets).
 Choose an appropriate membership function to represent each fuzzy set.
 Label the fuzzy sets on appropriately such that they reflect the problem to be
solved.

444
CHAPTER THERTEN

 Set up the fuzzy sets on appropriate universes of discourse.

 Adjust / tune the widths and centerpoints of membership functions
judiciously.

Some analysis of fuzzy control:

Usually fuzzy sets are overlapped by about 25%.

 It can be observed that in fuzzy logic control the transition from one set to
another provides a smooth transition from one control action to another.
 If there is no overlapping in the fuzzy sets then the control action would
resemble bivalent control (transition from one error region to another is rather
abrupt).
 On the other hand if there is too much overlap in the fuzzy sets, there would
be a lot of fuzziness and this blurs the distinction in the control action.

445
References
 REFERENCES

References:

Abu-Gouk, M. E “ Controlling Techniques and System Stability, University of

Khartoum Press, 2003.

Dale, E, Edgar, T.F and Duncan, A.M “ Process Dynamic and Control. John Willey
and Sons, New York, 1998.

Luyben, L. L” Process Modeling, Simulation and Control for Chemical

Engineering”. McGraw Hill Publishing Company, New York, 2007.

Stephan Opoulus, G “ Chemical Process Control An Introduction to Theory and

Practice” Prentice Hall, India, 1994.

446