Automatic Control For Mechanical Engineers
Automatic Control For Mechanical Engineers
Automatic Control For Mechanical Engineers
M. Galal RABIE
Professor of Mechanical Engineering
This book is published by the author. The hardcopy is for free distribution to the
author's students.
PREFACE
Automatic control plays an important role in the advance of engineering and science.
It is of extreme importance in most of the engineering fields; such as the aerospace
engineering, chemical engineering, robotic systems, automotive and mobile
equipment engineering as well as manufacturing and industrial processes. Automatic
control provides the means of understanding the problems of stability and precision
of dynamic systems. Actually most engineers must have good understanding of this
field.
The majority of textbooks on automatic control are most appropriated for electrical
engineers. The main problem in designing and analyzing a control loop for non-
electrical systems normally arises when deducing adequate mathematical model for
the system. Generally, the components cannot readily be represented by simple
discrete ideal elements. The classical approach based on the transfer function and
associated techniques of analysis is more easily comprehended and related to
practice by the beginners than is the modern control theory. Therefore this text is
prepared for the mechanical engineering students. It deals with the basics of linear
control theory. The text includes simple examples enabling applicants to understand
the problems of dynamic systems accuracy and stability. The text includes examples
and exercises that facilitate the comprehension of the control theory, especially for
the mechanical engineering students. The text is arranged in ten chapters dealing
with the following topics:
1. An introduction giving the basic definitions and methods of system
representation, Chapter 1.
2. A revision of selected topics from mathematics, Chapter 2.
3. Deduction of the transfer functions using mathematical models, block
diagrams and signal flow graphs, Chapters 3, 4 & 5.
4. Analysis of the transient and frequency responses of the system and how
does the response vary with the form of transfer function and the input
excitation. The text discusses also how a transfer function can be determined
by practical testing of a system, Chapters 6 & 7.
5. Analysis of the accuracy and stability of the feedback system, Chapter 8.
6. Root locus analysis, Chapter 9.
7. Improvement of the system stability by introducing and designing different
types of compensators, mainly the P, PI & PID controllers, Chapter 10.
I am indebted to my colleague Prof. Dr. Gamal Ahmed El-Sheikh, for his objective
comments on the whole text.
THE AUTHOR
i
M Galal RABIE, Automatic Control for Mechanical Engineers
THE AUTHOR
COVER DESIGN
The author is indebted to the cover designer, Rehab RABIE, MSc.
ii
M Galal RABIE, Automatic Control for Mechanical Engineers
CONTENTS
iii
M Galal RABIE, Automatic Control for Mechanical Engineers
iv
M Galal RABIE, Automatic Control for Mechanical Engineers
6.5 EXERCISE 85
v
M Galal RABIE, Automatic Control for Mechanical Engineers
REFERENCES 163
INDEX 165
vi
M Galal RABIE, Automatic Control for Mechanical Engineers
1 INTRODUCTION
1.1 INTRODUCTION
The system can be defined as being that part of the universe in which interest lays. It
is a combination of components acting together to perform a specific function. A
system may be an aircraft, a jet engine, the engine fuel pump or even the pump
controller. A component is a single functioning unit of a system. A single component
of a large assembly may be looked at as a system. The system should be defined
carefully and separated from the environment by means of an imaginary boundary. It
interacts with the environment by means of three types of signals, as shown by in
Fig.1.1.
System inputs are the controllable signals passing from the environment to the
system.
1
M Galal RABIE, Automatic Control for Mechanical Engineers
System outputs are signals passing from the system to the environment.
Linear system: A system is said to be linear if and only if it obeys the superposition
principles. This requires that if the separate application of the independent inputs I1(t)
and I2(t) produces outputs O1(t) and O2(t) respectively, then the application of an
input I(t) = I1(t) + I2(t) produces an output O(t) = O1(t) + O2(t), Fig.1.2. No real system
component is completely linear, but sometimes the range of operation is such that
the linearity can be assumed.
Static system is a system whose outputs depend on the present value of inputs.
Dynamic system is a system whose variables are time dependent and where the
output depends on the history of its inputs. Actually all of the systems are dynamic.
But, in some operating conditions, a dynamic system can be treated as a static one
for simplicity.
2
M Galal RABIE, Automatic Control for Mechanical Engineers
The objective of system control is to force the controlled system to produce certain
required outputs. It is necessary to adjust / select / design the values of one or more
of the inputs to obtain the required output.
There exist two classes of control systems; open loop control system and closed loop
(feedback) control system.
On the basis of knowledge about the system and of past experience, a prediction is
made of what input should be applied to produce the required output. Thus the
objective of an open loop control system is to achieve the required output by utilizing
an actuating device to control the process directly without measuring the system
output, Fig.1.3. Such control is frequently unsatisfactory because of the effect of
unexpected disturbances, which can lead to the deviation of the output from the
required value. However, the open loop control can be satisfactory and cost effective
for disturbance free applications.
3
M Galal RABIE, Automatic Control for Mechanical Engineers
Figure 1.4 shows an example of an open loop system where a hydraulic cylinder 1 is
controlled by means of a directional control valve 2. When the spool of the valve is
displaced by distance e, it connects the pressure line to the upper piston chamber A
and connects the lower piston chamber B to the drain line T. The throttling area, oil
flow rate and piston speed are proportional to the displacement e. Therefore, the
piston speed can be controlled via the valve displacement e. This is a typical
example of open loop control system. By adjusting the input e(t) the system produces
a corresponding output, piston speed for example. But, for the same input, if the
operating pressure, load or internal leakages change, the system output changes,
while keeping the input unchanged.
4
M Galal RABIE, Automatic Control for Mechanical Engineers
Another typical example of open loop system is the simple speed control of piston
engine. The engine speed is controlled by means of the fuel throttle valve, Fig.1.5. By
displacing the accelerator upwards, the throttle opens and the flow rate of fuel to
engine increases. Consequently, the engine speed is increased. But the actual value
of output speed n is not measured and not compared with the required value. If the
operating conditions are not changed, the speed n is kept constant, corresponding to
input y. But the disturbance, engine load for example, affect the system output, even
while keeping the input unchanged.
The system output is measured and compared with the desired value, Fig.1.6. The
system attempts continuously to reduce the difference (error) between the two
signals. The measured output value is called feedback (FB) signal. Sometimes, the
loop is closed through the human being as in the case of motor vehicle control.
Generally, the feedback reduces the effect of noise and disturbance on the system
performance.
Closing a loop makes the system more accurate, but it can make the system
oscillatory or even unstable. Therefore it is necessary to stress more on the analytical
approach for system analysis, to reach the required system accuracy and stability
and reduce cost and time.
The initial position of the feedback linkage is shown by the centerline connecting
points 4 & 5, Fig. 1.7(a). When an input displacement x is applied, the feedback
linkage first pivots about its connection with the piston rod, point 5, Fig.1.7(b). The
piston is hold temporarily in its position, shown by the dashed line connecting points
5
M Galal RABIE, Automatic Control for Mechanical Engineers
5 & 6. The spool is thus displaced by distance e. It allows the pressurized oil to flow
to the upper side of cylinder, A. The piston starts to move downwards by distance y
which causes the feedback linkage to pivot about the point 6, Fig.1.7(c). The piston
motion continues until the spool returns back to the original, closing position. This
position is reached when the piston displacement y causes a spool displacement e
equal to that caused by the input x, but opposite in direction. The total displacement
of the spool becomes zero. The final position of the feedback linkage is indicated by
the line connecting points 6 & 7.
(a) (b)
(c)
6
M Galal RABIE, Automatic Control for Mechanical Engineers
Another example of the closed loop control system is the speed control of some
internal combustion engines (ICE). Figure 1.8 shows a typical speed control system
of a piston engine 6. The position of the throttle lever sets the desired speed of the
engine. The speed control is shown at certain reference operating mode. The
positive direction of motion of the system parts is indicated by the arrows.
7
M Galal RABIE, Automatic Control for Mechanical Engineers
below its reference value, the centrifugal force of the flyweights decreases,
decreasing the force exerted on the bottom of the spring. This causes the spring seat
to move downwards (displacement x), which in turn moves the spool downwards
(displacement e). The pressurized fluid then flows to the lower piston chamber B,
which causes an increase of the displacement y and thus opens the fuel flow control
valve 5 wider. By supplying more fuel, the speed of the engine will increase until the
equilibrium is reached again. As the throttle lever is moved to a higher speed setting,
the spring upper-seat moves downwards by distance z. Consequently, the fuel
control valve opens to increase the fuel flow rate. Then, the engine speed is
increased.
8
M Galal RABIE, Automatic Control for Mechanical Engineers
In determining a reasonably simple model, one must decide which physical variables
and relationships are negligible and which are crucial to the accuracy of the model.
Neglecting the effect of some parameters and assuming linear relations instead of
nonlinear ones lead to simpler model. But, if these decisions are not justified, the
results obtained from this model will not agree with the real system behavior.
Moreover, there is no guarantee for the controller to be applicable.
The following equations describe the car suspension system drawn schematically in
Fig.1.9.
d2 x
f qp m (1.1)
dt 2
dx
p fr and qk x (1.2)
dt
The transfer function is written for linear system with, for simplicity, zero initial
conditions. It is defined as the ratio of the Laplace transform of the system output to
that of the input. The dynamics of car suspension, Fig.1.9, is described by the
following Differential equation.
d2 x dx
m 2
fr k x f (1.3)
dt dt
The system input is the driving force f and the output is the displacement x. Then, by
applying Laplace transform, for zero initial conditions, the following transfer function
can be obtained.
X(s) 1
G(s) (1.4)
F(s) ms fr s k
2
9
M Galal RABIE, Automatic Control for Mechanical Engineers
The block diagram is a pictorial representation of the system clarifying the inter-
relation between its different parts. This diagram can be used for system analysis. In
this case, it must be supplied by a quantitative description of the relations between
system variables in the form of appropriate mathematical expressions; usually
transfer functions. Figure 1.10 shows a block diagram describing the quarter-car
illustrated by Fig.1.9.
10
M Galal RABIE, Automatic Control for Mechanical Engineers
d2 x dx
m fr k x f (1.5)
dt 2 dt
Define,
dx
x1 (1.6)
dt
then,
dx 1
m fr x 1 k x f (1.7)
dt
That is, the second order differential equation is replaced by the following two first
order differential equations.
dx 1 1
( f fr x 1 k x) (1.8)
dt m
dx
x1 (1.9)
dt
Or, in a matrix form:
x 0 1 x 0
f (1.10)
k
fr x 1
x1 1
m m m
11
M Galal RABIE, Automatic Control for Mechanical Engineers
When the system input changes, the system output varies depending on the nature
of the input signal, the system design and the disturbance signals. In the absence of
disturbance signals and for known system structure, the system output depends only
on the input signals. The most commonly used system input forcing functions are:
1. Transient disturbances such as step, ramp or impulse functions.
2. Sinusoidal signals.
3. Statistical signals of random nature.
4. Discrete signals.
This text deals with the first two types of input signals for the time and frequency
domains analysis.
1.6 EXERCISE
2. Define the different types of signals connecting a system with the environment.
Give some examples of real systems and define the different types of signals
for each system.
3. Discuss briefly the open and closed loop control systems; give some examples
of each type.
4. Define the static and dynamic systems giving some examples for each type.
12
M Galal RABIE, Automatic Control for Mechanical Engineers
2 MATHEMATICAL
TOPICS
2.1 INTRODUCTION
One of the problems facing engineers is to deduce the equations describing the
operation of components and systems. Once the equations are determined, they can
be solved to find the response of the system to various inputs relevant to the way the
system will be used.
Linear differential equations can be solved using the formal classical methods.
However, engineers are more comfortable with an automated technique such as
Laplace transform. This chapter is directed to the revision of selected mathematical
topics needed for the analysis of control systems performance.
This part deals with linear differential equations (DE) of constant coefficients. The
linear differential equation includes neither variables raised to a power different than
unity nor a product of variables.
dx
x8 (2.1)
dt
d3 x d2 x dx
3
2 2
8 3 x 10 (2.2)
dt dt dt
d2 x dx
2
x 4t x 10 (2.3
dt dt
d2 x dx
2 2 x 8 (2.4)
dt dt
13
M Galal RABIE, Automatic Control for Mechanical Engineers
The order of differential equations is the highest time derivative present. For the
solution of a differential equation, a number of initial conditions are necessary, equal
to the order of the DE.
The solution of a differential equation yields the system response which consists of
two parts; steady state response and transient response.
If the variable of the differential equation is x, the solution can be written as:
x xt xs (2.5)
x t x s 2( x t x s ) 8
The right hand side of the differential equation is the input or forcing function.
Assuming that, after long time, the transient part of the response has decayed while
the steady state response part remains; xt = 0, then,
x s 2x s 8 To be solved for the steady state response,
and x t 2x t 0 To be solved for the transient response.
Since xs is time independent, then, x s 0 and hence xs = 4
x t / x t 2 , The integration of this equation yields:
ln( x t ) 2t c
x t e 2t c Ae 2t
Since x( t ) x t x s ,
14
M Galal RABIE, Automatic Control for Mechanical Engineers
then, x 4 Ae 2t
dx
Fig.2.1 Plot of the solution of DE 2x 8
dt
d2 x dx
m fr k x f
dt 2 dt
d2 x dx
2
3 2x 6
dt dt Fig.2.2 Scheme of a quarter-
car
15
M Galal RABIE, Automatic Control for Mechanical Engineers
dx
and x(0) 1, (0 ) 0
dt
d2 x t dx t d2 x s dx
2
3 2 xt 2 3 s 2 xs 6
dt dt dt dt
After long time, the transient part of solution decays; xt=0. The time derivatives of the
steady state part are zeros, then:
dx s d2 x s
0, 0 and xs = 3
dt dt 2
d2 x t dx t
2
3 2 xt 0
dt dt
s 2 e st 3se st 2e st 0
Since e st 0 , then s 2 3s 2 0
The last equation is called the characteristic equation. It can be obtained directly from
the homogeneous equation by using the following substitution rule.
dx t d2 x t
x t 1, s, 2
s 2 , … etc.
dt dt
s = -1 or s = -2
Then, x t A 1e t A 2 e 2t
16
M Galal RABIE, Automatic Control for Mechanical Engineers
and x x s x t 3 A 1e t A 2 e 2t
The constants A1 and A2 are calculated considering the following initial conditions;
dx
x(0) 1 and (0 ) 0 .
dt
A1 + A2 = -2 and A1 + 2 A2 = 0
Then A1 = - 4 and A2 = 2
x 3 4e t 2e 2t
This method becomes very complicated as the order of the differential equation
increases. Therefore, it would be better to use an algebraic method that combines
the natural response, forced response and determination of the coefficients in one
procedure. The Laplace transform method does just that.
X(s) L x( t ) = x(t )e
st
dt (2.6)
0
0 for t 0
x( t )
a for t 0
17
M Galal RABIE, Automatic Control for Mechanical Engineers
e st a
X(s) ae dt a st
0
s 0
s
k
X(s) x( t )e dt ke
st at
e dt k e ( a s ) t dt
st
0 0 0
sa
x( t ) cos(t ) e it e it / 2 where = constant
X(s) x( t )e st dt
0
1 ( si) t
1 st it
it
X(s) e ( e e )dt e dt e ( si) t dt
20 2 0 0
s
Or X(s)
s 2
2
The inverse Laplace transform is an integral operator, which transforms from the
Laplace domain to the time domain. It is given by the following expression.
1 1 iR
x( t ) Lim X(s)e st ds (2.7)
2i R 1
iR
Actually, all functions in the time domain have direct Laplace transform but some of
the functions in the Laplace domain have no inverse Laplace transform.
Some of the Laplace transforms properties used in this text are summarized as
follows:
18
M Galal RABIE, Automatic Control for Mechanical Engineers
3. L e at f ( t ) F(s a) (2.10)
4. Initial value problem f (0 ) Lim f ( t ) lim sF(s) (2.11)
t 0 s
df ( t )
6. L sF(s) f (0 ) (2.13)
dt
d2 f ( t )
df ( t )
L 2 s 2F(s) sf (0 ) f (0 ) ; where f ( t ) (2.14)
dt dt
dn f ( t ) n 1 n 2 df
2
n 3 d f dn1f
L n s F(s) s f (0 ) s
n
(0 ) s (0 ) n1 (0 ) (2.15)
dt dt dt 2 dt
7.
L f (t)dt 1s F(s) f (t)dt 0
(2.16)
Example 2.6 Find L x( t ) , if x( t ) e 2t
1
X(s) and x(0 ) 1
s2
s 2
L x( t ) sX(s) x(0 ) 1
s2 s2
1
Example 2.7 Find x (t) after very long time if X(s)
s2
s
x( ) lim x( t ) lim sX(s) lim 0
t s 0 s 0 s 2
19
M Galal RABIE, Automatic Control for Mechanical Engineers
m
P(s)
Given X(s) , where Q(s) (s a i ) with real distinct ai and the degree of
Q(s) i 1
ki
X(s) i1 Where k i lim s ai X(s)
m
(2.17)
s ai s ai
1
Example 2.8 Find x(t) given X(s)
s(s 2)
The inverse Laplace transform of this expression can be obtained after performing a
partial fraction expansion as follows.
1 k k
1 2
s(s 2) s s 2
1 1
X(s) 0.5
s s 2
x( t ) 0.5 1 e 2t
s5
Example 2.9 Find v( t ) , given: V(s)
s(s 1)(s 2)
k1 k 2 k
V ( s) 3
s s 1 s 2
20
M Galal RABIE, Automatic Control for Mechanical Engineers
k 2 lim (s 1) V(s) 4
s 1
k 3 lim ( s 2 ) V ( s ) 1 .5
s 2
Case 2: Repeated real roots of characteristic equation (i.e. Repeated real roots of
the polynomial of the denominator).
m
P(s)
Given X(s) , where Q(s) (s a i ) with real ai and the degree of P(s) is less
Q(s) i 1
than that of Q(s), the root (-a1) is repeated with multiplicity n, then:
k n i
1
i!
di
1 ds
n
lim i s a1 X(s) ; where i 0,..., n 1 (2.18)
s a
s4
Example 2.10 Find the partial fractions of Y(s)
(s 1)(s 2) 2
s4 k1 k k3
Y( s ) 2
(s 1)(s 2) 2 s 1 s 2 ( s 2) 2
k 1 lim (s 1)Y(s) 3
s 1
k 3 lim ( s 2 ) 2 Y ( s ) 2
s 2
d s 4 (s 1) (s 4)
k 2 lim
d
(s 2) 2 Y(s) lim lim 3
s 2 ds
s2 ds s 1 s2 (s 1) 2
3 3 2
Y( s )
s 1 s 2 ( s 2) 2
21
M Galal RABIE, Automatic Control for Mechanical Engineers
The value of k2 is found by equating the coefficients of s in the nominator of the two
equation sides;
1 = 12 + 3k 2 - 2 , then k 2 -3
8
Example 2.11: Find x(t) given: X(s)
2
s(s 2s 2)
8
X(s)
s[( s 1) 2 1]
Find the inverse Laplace transform, x(t), using the Laplace transform tables.
k k e at sin(t )
x(t)= , tan 1( / a)
2 2
a a 2 2
To solve the differential equations using Laplace transform, the following steps are
followed;
EXAMPLE 2.12 Solve x 3 x 2x 6 for x(0)=1 and x(0) 0
L [ x ( t )] s 2 X(s) sx(0) x(0) s 2 X(s) s
22
M Galal RABIE, Automatic Control for Mechanical Engineers
L [ x( t )] sX(s) x(0) sX(s) 1
6
L [ 6]
s
s 2
X(s) s 3sX(s) 1 2 X(s)
6
s
X(s) s 2 3s 2 6
s
3s
s 2 3s 6
s
s 2 3s 6 s 2 3s 6
Step 2: X(s)
s[s 2 3s 2] s(s 1)(s 2)
k1 k 2 k
Step 3: X(s) 3 where k 1 3, k 2 4 and k 3 2
s s 1 s 2
Step 4: x( t ) 3 4e t 2e 2t
Example 2.13 Solve x 2 x x 8 for x(0) 0 and x(0) 0
Step 1: s 2
2s 1 X(s) 8
s
8
Step 2: X(s)
s(s 1) 2
k1 k2 k
Step 3: X(s) 3 where k1 k 2 k 3 8
s (s 1) 2 s 1
Step 4:
x( t ) 8 1 te t e t
Example 2.14 Solve y 4 y 8 for y(0) 0 and y(0) 0
Step 1: s 2
4 Y( s )
8
s
23
M Galal RABIE, Automatic Control for Mechanical Engineers
8
Step 2: Y( s ) 2
s(s 4)
k1 k 2s k 3
Step 3: Y( s ) , where k1=2, k2=-2 & k3=0
s s2 4
2 2s
Y( s ) 2
s s 4
The following are some of the important relations of complex variables, needed for
further analysis.
z re i (2.20)
z z (2.21)
n
z n z n (2.22)
where
e i e i
cos (2.27)
2
e i e i
sin (2.28)
2i
x x2 x3 x 4 xn
e x 1 ... ... (2.29)
1! 2! 3! 4! n!
24
M Galal RABIE, Automatic Control for Mechanical Engineers
ks
8. k cos t
s 2 2
k
9. 2 2 k e at sin t
( s a)
k ( s a)
10. k e at cos t
( s a) 2 2
k k k e at sin(t )
11. 2 2
, tan 1( / a)
2 2
s[( s a) ] a a 2 2
2.6 EXERCISE
a) x( t ) 1 2t b) x( t ) 4(1 e 2t )
e) x( t ) 2e 4 t sin 4t
25
M Galal RABIE, Automatic Control for Mechanical Engineers
3s 1 6s 2 10s 2
a) b)
s(s 1) s 3 3s 2 2s
5s 6 1
c) d)
(s 2)(s 3) 4s(s 4)
3s s
e) F(s) 2
f) F(s) 2
s 9 s 2s 1
s
g) F(s)
s 2 2s 2
4. df ( t )
Find F(s) for the following differential equations for f (0) 2 and (0 ) 1 .
dt
df ( t )
a) 3 4f ( t ) 5 where f (0 ) 0
dt
d2 f (t ) df ( t )
b) 2 3 4f ( t ) e 5 t 6 cos 7t
2 dt
dt
d2 y( t ) dy( t ) dy
b) 4 3 y( t ) 6 where y(0) 0 & (0 ) 0
dt 2 dt dt
d2 y dy
c) 2
9y 0 where y(0) 0 & (0 ) 3
dt dt
d2 x dx dx
d) 2
2 10 x 10 where x(0) 0 & (0 ) 0
dt dt dt
26
M Galal RABIE, Automatic Control for Mechanical Engineers
6. 100
Find y(t) if Y(s)
s{(s 5) 2 25}
dx 2
a) L where X(s) & x(0) 2
dt s
d2 x 5 dx
b) L 2 where X(s) , x(0) 5 and (0) 5
dt s 1 dt
3s 2 s 8
E(s)
5s 3 4s 2 2s
a) Find the initial error value e(0).
b) Find the final (steady state) value of error e( ).
m 1000 kg
27
M Galal RABIE, Automatic Control for Mechanical Engineers
28
M Galal RABIE, Automatic Control for Mechanical Engineers
3 TRANSFER
FUNCTIONS
The static and dynamic behavior of a linear, single-input single-output system (SISO)
can be described by the transfer function. The transfer function (TF) is defined as the
Laplace transform of the system output divided by that of its input. It is systematically
deducible for linear system of zero initial conditions. If the initial conditions are
different from zero, then, for simplicity of mathematical manipulation, assume zero
initial conditions by simple transformation of axes and then consider their real values
during solution.
Conventionally, the symbol G(s) is used for the transfer function, but if the element
appears in the feedback path, the symbol H(s) is used.
Y(s) P(s)
G(s) (3.1)
X(s) Q(s)
m n
P(s) b i si and Q(s) a i si (3.2)
i 0 i0
For real systems, the order of P(s) should not exceed that of Q(s); n m .
Example 3.1 Consider the linear system, of input x(t) and output y(t), described by
the following differential equation of zero initial conditions.
d4 y d3 y d2 y dy d2 x dx
a4 a 3 3 a 2 2 a1 a 0 y b 2 2 b1 b0 x (3.3)
dt 4 dt dt dt dt dt
The transfer function (TF) can be deduced by applying Laplace transform as follows.
29
M Galal RABIE, Automatic Control for Mechanical Engineers
a s
4
4
a 3 s 3 a 2 s 2 a1s a 0 Y(s) b 2 s 2 b1s b 0 X(s) (3.4)
Y( s ) b 2 s 2 b1s b 0
Then, G(s) (3.5)
X(s) a 4 s 4 a 3 s 3 a 2 s 2 a1s a 0
a 4 s 4 a 3 s 3 a 2 s 2 a1s a 0 0 (3.6)
Matrix transfer function: For a multi-input multi-output system (MIMO), the system
inputs and outputs can be related by a transfer matrix, Eq. 3.7. The transfer functions
relating the system inputs and outputs can be directly obtained from this matrix,
where X and Y are the input and output vectors respectively.
Example 3.2 In the case of a hydraulic line, Fig.3.1, the inlet pressure Po and the
volumetric flow rate Qo are related to the outlet parameters, PL and QL by a transfer
matrix:
30
M Galal RABIE, Automatic Control for Mechanical Engineers
C = line capacitance.
I = line inertia.
The line end can be closed, QL=0, or open to atmosphere, PL=0. For closed end line,
QL=0, the transfer functions describing the line are:
PL 1 PL 1
( s) and
Po ICs RCs 1
2
Q o Cs
While for open end line, PL=0, the line transfer functions are:
QL 1 QL
( s) and 1
Po Is R Qo
Gain of transfer function is the proportionality coefficient relating the input and
output in the steady state, i.e. as t or s 0.
The proportional element has a transfer function of the form G(s)=K , where K is
constant. The simple coil spring is a typical proportional element. The simple
equation describing the coil spring can be deduced considering the following
assumptions:
1. The force is not so large that the spring coils touch each other and then the
spring stiffness k is constant.
31
M Galal RABIE, Automatic Control for Mechanical Engineers
f ( t ) kx( t ) (3.8)
X(s) 1
G(s) K ; where K is a constant (3.10)
F(s) k
The integrating element has a transfer function of the form G(s)=k/s , where k is
constant. The following are some examples of integrating members.
Figure 3.3 shows a scheme of a hydraulic cylinder. Neglecting the inertia of moving
parts, the effect of oil compressibility and wall elasticity, the relation between the
system input (flow rate q) and output (displacement x) can be deduced as follows.
dx
q( t ) A , (3.11)
dt
where A is the piston rod-side area.
32
M Galal RABIE, Automatic Control for Mechanical Engineers
X(s) 1 k 1
G(s) ; k (3.13)
Q(s) As s A
This is the transfer function of an integrating element. The output depends on the
time integral of the input, or:
x( t ) k q( t )dt (3.14)
Figure 3.4 shows a typical example of open loop system. The hydraulic cylinder 1 is
controlled by means of a directional control valve 2. When the spool of the valve is
displaced by distance e, it connects the pressure line to the upper piston chamber A
and connects the drain line to the lower piston chamber B. The throttling area and oil
flow rate are proportional to displacement e. Therefore, for an input e(t), a
proportional output piston speed (dy/dt) is produced. However, the output changes,
for the same input, if the operating pressure, load or internal leakage are changed.
The transfer function of this system can be deduced considering the following
assumptions:
1. No internal leakage and the oil is incompressible.
33
M Galal RABIE, Automatic Control for Mechanical Engineers
At(e) = ω e (3.16)
q( t ) k e( t ) (3.17)
dy dy
q( t ) A or q( t ) / A , where A is the piston area (3.18)
dt dt
Q(s)
Q(s) k E(s) and sY(s) (3.19)
A
Then
Y Y Q 1 K
G(s) k or G(s) (3.20)
E Q E As s
where K=k/A.
dy 1 k
Actually; q( t ) A
dt
,then: y
A
q( t ) dt
A
e( t ) dt K e( t ) dt (3.21)
Equation 3.21 indicates that the output y(t) is directly proportional to the time integral
of the input e(t) dt . Therefore the shown valve controlled actuator is said to be an
integrating element.
34
M Galal RABIE, Automatic Control for Mechanical Engineers
The hydraulic servo actuator shown in Fig.3.5, is actuated by a low power input
signal x(t) and produces a corresponding sufficiently high power output signal y(t).
The construction and operation of this system are explained in sec.1.3.2.
The following transfer function was deduced considering the discussion of valve
controlled hydraulic cylinder, section 3.2.2.2, page 33:
Y K
G(s) (3.22)
E s
b a
e x y or e 0.5 ( x y ) for a=b (3.23)
ab ab
E 0. 5 ( X Y ) (3.24)
35
M Galal RABIE, Automatic Control for Mechanical Engineers
The hydraulic servo-actuator is described mathematically by Eqs. 3.22 and 3.24. The
block diagram shown by Fig. 3.6 was developed using these equations and the
system transfer function is:
Y 0 .5 K 1
( s) (3.25)
X s 0. 5 K T s 1
This is a unity gain first order transfer function of time constant T=2/K.
The following is the general form of transfer function of first order element, where T is
the time constant and k1 is the gain
k1
G(s) (3.26)
Ts 1
The basic elements of the RLC circuits are described by the following expressions:
Resistance R v Ri (3.27)
di
Inductance L v L (3.28)
dt
dv
Capacitance C iC (3.29)
dt
The given R-C circuit, Fig.3.7, can be described by the following equations.
v i (t) v o (t) dv
i C o (3.30)
R dt
36
M Galal RABIE, Automatic Control for Mechanical Engineers
dv o ( t )
v i ( t ) v o ( t ) RC (3.31)
dt
Assuming zero initial conditions, the application of Laplace transform to Eq. 3.31,
yields:
Vo (s) 1 1
G(s) (3.33)
Vi (s) RCs 1 Ts 1
k
The second order element has a transfer function of the form: G(s) .
as bs 1
2
The quarter car model shown by Fig. 3.8, is a typical example of the second order
element. Assuming ideal linear spring and damper and neglecting the spring inertia,
the following equation can be deduced.
d2 x dx
m 2
fr k x f (3.34)
dt dt
37
M Galal RABIE, Automatic Control for Mechanical Engineers
Assuming zero initial conditions, then the application of Laplace transform yields:
X(s) 1
G(s) (3.36)
F(s) ms 2 fr s k
The general form of this equation is the second order transfer function, given by:
k1 k 1 n2
G(s) 2 2
s 2 s 2n s n2 (3.37)
s 1
n2 n
3.3 EXERCISE
38
M Galal RABIE, Automatic Control for Mechanical Engineers
4. Y
Derive the transfer function G(s)
(s) of the
F
given system and give the mathematical
expressions for the system gain, natural
frequency and damping ratio.
7. X(s)
Derive the transfer function, G(s) , of each of the given systems.
F(s)
(a) (b)
39
M Galal RABIE, Automatic Control for Mechanical Engineers
G(s) =X1(s)/F(s)
9. A process plant consists of two tanks with cross-sectional areas A1 and A2. The
flow rate through the throttle valve is given by: q = ∆h /R, where ∆h is the head
difference across the valves in meters and R is the throttle valve resistance.
Write the equations describing the system operation and derive the following
transfer functions:
H1 Q H1
G1 ( s ) , G 2 (s) 2 and G 3 (s)
Q2 Q3 Q3
40
M Galal RABIE, Automatic Control for Mechanical Engineers
10. The given hydraulic system includes two rigid cylindrical tanks with areas A1
and A2. Write the equations describing the system operation and derive the
transfer function G(s) = Q(s) /Q1(s).
11. Derive the transfer function G(s) = Y(s)/X(s) for the given proportional plus
integral control system.
41
M Galal RABIE, Automatic Control for Mechanical Engineers
42
M Galal RABIE, Automatic Control for Mechanical Engineers
4 BLOCK
DIAGRAMS
4.1 INTRODUCTION
When dealing with the dynamic behavior of a system, one starts usually by the
derivation of the equations and the transfer functions describing the behavior of its
components. These transfer functions are needed for the analysis of the dynamic
behavior of the components. Therefore, when carrying out the integrated system
analysis, it is necessary to derive the transfer function of the overall system. This
derivation becomes easier when representing the system by a block diagram.
The following symbols are used as basic structural elements of the block diagrams.
Branching Points are indicated by heavy point with single incoming signal and one
or more outgoing signals. All of the out-going lines carry the same signal, Fig.4.1.
43
M Galal RABIE, Automatic Control for Mechanical Engineers
Summation Point is represented by a small circle with more than one in-going signal
and only one out-going signal. The in-going signal sign is indicated by (+) or (-) signs.
The value of the out-going signal equals the sum of the in-going signals, each with its
assigned sign, Fig.4.3.
The system transfer function can be deduced from its block diagram by extracting
and manipulating the equations relating the system variables. But, in the case of
complicated block diagrams, it is might be necessary to simplify the diagram by
applying the rules of block diagram algebra.
Figure 4.4 shows the block diagram of a typical system with negative feedback. The
system transfer function can be deduced from the block diagram as follows:
Y GE , F HY and E X F (4.1)
Or
Y G( X HY ) (4.2)
Then
Y G
(4.3)
X 1 GH
44
M Galal RABIE, Automatic Control for Mechanical Engineers
It is important to give the following definitions for the closed loop (feedback) system.
Characteristic equation
1 GH 0 (4.7)
y( t ) y i ( t ) (4.8)
Y G
G(s) (4.9)
X 1 GH
45
M Galal RABIE, Automatic Control for Mechanical Engineers
Example 4.1 shown is the block diagram of a feedback system. Find its transfer
function, G=Y/X.
Y 200
E s(s 10)
200
Y s(s 10) 200 0 .5
G(s) 2
X 2 200 s 10s 400 s 2
0. 5
1 s 1
s(s 10)
20 20
Equivalent
Type
Y G1G 2 X
2. Parallel System
Y (G1 G 2 )X
46
M Galal RABIE, Automatic Control for Mechanical Engineers
Type Equivalent
Equation Block Diagram
Block Diagram
3. Loop Reduction
Y G( X HY )
Or
G
Y X
1 GH
4. Displacement of
Z G( X Y )
Summation Point
Z GX Y
5. Displacement of Feedback
point
Y GX
6. Re-arrangement of
Summation Points
Z W XY
47
M Galal RABIE, Automatic Control for Mechanical Engineers
Type Equivalent
Equation Block Diagram
Block Diagram
feedback and summation points
7. Modification of position of
Z XY
Example 4.2 Apply the rules of block diagram algebra to deduce the transfer function
of the system described by the given block diagram.
48
M Galal RABIE, Automatic Control for Mechanical Engineers
Example 4.3 Find the system transfer function, G(s)=Y(s)/X(s), by successive block
diagram reduction.
49
M Galal RABIE, Automatic Control for Mechanical Engineers
4.5 EXERCISE
50
M Galal RABIE, Automatic Control for Mechanical Engineers
3. Find the transfer function, G(s) o (s) / i (s) , of the system described by the
following block diagram.
5. Derive the transfer function of the system described by the following block
diagram; G(s) o (s) .
i
51
M Galal RABIE, Automatic Control for Mechanical Engineers
7. Redraw the following block diagram to obtain the transfer function G(s)=Z/X.
52
M Galal RABIE, Automatic Control for Mechanical Engineers
5 SIGNAL
FLOW GRAPH
5.1 INTRODUCTION
Signal flow graphs are alternatives to block diagrams. They are drawn either using
the system of equations or based on the block diagrams. The signal flow graph
consists of nodes, which represent the system variables and directed line segments
which carry the relation-ship between the variables. Originally, the signal flow graph
was introduced by S. J. MASON (1955), for the cause and effect representation of
linear systems, described by algebraic equations.
The signal flow graphs consist of line segments (called Branches) and junction points
(called nodes). The, nodes represent the variables, summation points and branching
points. They are connected together by the line segments. The branches have
associated branch gain and direction. The signal is transmitted through a branch in
the direction of the arrow. The conventions for signal flow graphs are summarized in
Table 5.1.
Element Convention
1. Variable X
2. Transfer Function
Y
G(s) Or Y GX
X
3. Summation Point
Z G1X G 2 Y
53
M Galal RABIE, Automatic Control for Mechanical Engineers
4. Parallel Connection
Y G1 X G 2 X
( G1 G 2 ) X
5. Series Connection
Z G2 Y
Y G1 X
Z G1G 2 X
6. Loop
7. Self Loop
Y G1 X G 2 Y
G1
Y X
1 G2
Example 5.1 Figure 5.1 shows a signal flow graph of a system described by five
state variables, X1, X 2 , X 3 , X 4 and X 5 . This graph was constructed for a system
described by the following equations, where ti is a transfer function.
X 2 t 12 X1 t 32 X 3
X 3 t 23 X 2 t 43 X 4
X 4 t 34 X 3 t 24 X 2 t 44 X 4
X 5 t 25 X 2 t 45 X 4
54
M Galal RABIE, Automatic Control for Mechanical Engineers
The transfer function of the system can be obtained from the signal flow graph, in a
relatively simple way, by applying Mason's formula. For this purpose, it is important to
give the following definitions, applied to the signal flow graph, Fig.5.1.
Input node (Source): is a node which has only outgoing branches, X1.
Output node (Sink): is a node, which has only ingoing branches X5.
Forward path: is a path from input node to output node along which no node is
encountered more than once.
Feedback loop: is a path, which originates and terminates at the same node, along
which no node is encountered more than once.
Path gain: is the product of all gains (transfer functions) along the forward path. The
following are the path gains of signal flow graph shown in Fig.5.1.
M1 t 12 t 23 t 34 t 45
M2 t12 t 24 t 45
M3 t12 t 25
Loop gain: is the product of the gains of branches forming the loop. The following
are the loop gains of signal flow graph shown in Fig.5.1.
1 t 23 t 32 2 t 34 t 43
3 t 44 4 t 24 t 43 t 32
55
M Galal RABIE, Automatic Control for Mechanical Engineers
Given a signal flow graph, the task of solving for input-output relations by algebraic
manipulation may be quite tedious. Fortunately, there exists a general gain formula
available, known as the Mason's formula. It enables to deduce the system transfer
function on the basis of the signal flow graph. This general formula is:
Mi i
k
G(s) i 1 (5.1)
and
∆i = ∆ for all loops that do not touch the ith forward path (5.4)
Example 5.2 Find the transfer function of the system represented by the signal flow
graph of Fig.5.1.
M1 t 12 t 23 t 34 t 45
M2 t12 t 24 t 45
M3 t12 t 25
56
M Galal RABIE, Automatic Control for Mechanical Engineers
2. Loops Gains
1 t 23 t 32
2 t 34 t 43
3 t 44
4 t 24 t 43 t 32
3. ∆ and ∆i
1 (1 2 3 4 ) 1 3
1 ( t 23 t 32 t 34 t 43 t 44 t 24 t 43 t 32 ) t 23 t 32 t 44
1 1, 2 1 and 3 1 ( 2 3 ) 1 t 34 t 43 t 44
4. Transfer Function
M11 M2 2 M3 3
G(s)
t 12 t 23 t 34 t 45 t 12 t 24 t 45 t 12 t 25 (1 t 34 t 43 t 44 )
G(s)
1 t 23 t 32 t 34 t 43 t 44 t 24 t 43 t 32 t 23 t 32 t 44
Example 5.3 Find the transfer function of the feedback system shown in Fig.5.2.
M1 G , 1 GH
1 ( GH) 1 GH , 1 1
M1 1 G
G(s)
1 GH
57
M Galal RABIE, Automatic Control for Mechanical Engineers
Example 5.4 Derive a mathematical model for the following system, draw its signal
flow graph and derive its transfer function by applying Mason’s formula.
1. Mathematical Model
q k( x y ) Q k( X Y )
d2 y
qp m Q P m s 2 Y; P Q ms 2 Y
dt 2
dy P
p fr P fr s Y; Y
dt fr s
3. Transfer Function
k k ms
M1 , 1 and 2
fr s fr s fr
k ms k ms 2
1 1 , 1 1
fr s f r fr s
k
1
Y M11 fr s k
G(s)
X k ms 2
ms fr s k
2
1
fr s
58
M Galal RABIE, Automatic Control for Mechanical Engineers
Example 5.5 Draw the signal flow graph of the system represented by the given
block diagram and find its transfer function (apply the conversion guide lines on page
61).
M1 G1G 2 G 3 M2 G1G 4
1 G1G 2 G 3 2 H1G1G2
3 H2 G 2 G 3 4 H2 G 4
5 G1G 4
1 1 2 3 4 5 1 1 2 1
G1G 2 G 3 G1G 4
G(s)
1 G1G 2 G 3 H1G1G 2 H2 G 2 G 3 H2 G 4 G1G 4
59
M Galal RABIE, Automatic Control for Mechanical Engineers
5.4 EXERCISE
1. Draw the signal flow graph for the systems described by the following block
diagrams and find the closed loop transfer function using Mason's formula.
a)
b)
c)
2. Find the closed-loop transfer function Vo/Vi of the system described by the
following signal flow graph.
60
M Galal RABIE, Automatic Control for Mechanical Engineers
3. Draw a block diagram for the system represented by the following signal flow
graph and find its transfer function.
4. Derive the expression relating the output velocity ωo to the load torque TL and
input signal ωi for the following signal flow graph, given:
90 2.5 1
G1 G2 G 3 25 G 4 3 .8 G5
1 0.06 s 1 0.19 s 1355 s
G 6 0 .4 G 7 0 . 4 H1 0.001 H 2 4
NB Guide lines for the conversion of Block diagrams to Signal flow graphs
61
M Galal RABIE, Automatic Control for Mechanical Engineers
62
M Galal RABIE, Automatic Control for Mechanical Engineers
6 TIME DOMAIN
ANALYSIS
6.1 INTRODUCTION
In the transient conditions, the variables describing the dynamic systems change with
time. It is of interest to evaluate how do these variables vary with time. In the analysis
problem, a reference input signal is applied to the system and the variation of the
state parameters with time is observed and analyzed (Time Domain Analysis). The
step, ramp and impulse signals are usually used as input signals for the time domain
analysis. The step input is most widely used and represents the most possible severe
excitations to which the system may be subjected.
Step Function
0 for t 0
f (t) (6.1)
k 2 for t 0
k2
F(s) (6.2)
s
f (t) k 2 t (6.3)
k2
F(s) (6.4)
s2
63
M Galal RABIE, Automatic Control for Mechanical Engineers
Impulse Function
f ( t ) k 2 ( t ) (6.5)
F(s) k 2 (6.6)
The response of the dynamic system to the input signal is divided into two parts:
transient part and steady state part. The transient part decays with time. It becomes
zero after certain time. The steady state part of response is the part which remains
constant after the transient part has decayed. Given the transfer function, the
transient response can be calculated according to the following procedure.
y( t ) L -1 Y(s) (6.7)
Example 6.1 Find the response of the system described by the following transfer
function to a step input of magnitude k2.
Y( s ) 5
G(s)
X(s) (s 1)(s 2)(s 3)
The input x(t) is step function of magnitude k2, its Laplace transform is:
k2
X(s)
s
5 k2 5 2 .5 2 .5 5
and Y(s) G(s)X(s) k2
s(s 1)(s 2)(s 3) 6s s 1 s 2 6(s 3)
5 5
Then: y( t ) k 2 2.5e t 2.5e 2 t e 3 t .
6 6
Figure 6.4 shows the step response, evaluated analytically, where the steady state
value of y is y ss 5k 2 / 6 .
64
M Galal RABIE, Automatic Control for Mechanical Engineers
Herein, the responses of the basic elements to step, ramp and impulse input signals
are evaluated.
k1
G(s) (6.8)
s
k2
X(s) (6.9)
s
k
Y(s) G(s) X(s) (6.10)
s2
k=k1k2 (6.11)
y( t ) k t (6.12)
65
M Galal RABIE, Automatic Control for Mechanical Engineers
k2
X(s) (6.13)
s2
k
Y(s) G(s) X(s) (6.14)
s3
k=k1k2
y( t ) 0.5 k t 2 (6.15)
X(s) k 2 (6.16)
k
Y(s) G(s) X(s) , k=k1k2 and y( t ) k (6.17)
s
The first order element has a transfer function of the following form.
k1
G(s) (6.18)
Ts 1
k2
X(s) , a step input of magnitude k2 (6.19)
s
k 1k 2 k 1 T 1 1
Y(s) G(s) X(s) k( ) k( ) (6.20)
s(Ts 1) s(Ts 1) s Ts 1 s s 1/ T
In the case of unity gain system, the unit step response is given by:
y( t ) 1 e t / T (6.22)
66
M Galal RABIE, Automatic Control for Mechanical Engineers
The step response of a first order element is plotted in Fig.6.7. The study of this
figure shows that:
i) y(T)=0.632 yss
ii) A tangent drawn at a point on the response curve intersects the yss horizontal
line T seconds later.
iii) For t 4T, y ss y( t ) 0.02 y ss
iv) If the transfer function gain is k1 and the applied step magnitude is k2 then the
steady state output is k = k1 k2.
k2
x( t ) k 2 t , X(s) (6.23)
s2
k 1 T T
Y(s) G(s)X(s) k 2 , k=k1k2 (6.24)
s (Ts 1)
2
s s s 1/ T
y( t ) k t T(1 e t / T ) (6.25)
The response of a first order element to ramp input is plotted in Fig.6.8. The transient
part of response (kTe-t/T) decays with time. The steady state response is k( t T ).
67
M Galal RABIE, Automatic Control for Mechanical Engineers
x( t ) k 2 ( t ) , X(s) k 2 (6.26)
k 1k 2 k/T
Y( s ) , where k = k1 k2 (6.27)
Ts 1 s 1/ T
k t / T
y( t ) e (6.28)
T
68
M Galal RABIE, Automatic Control for Mechanical Engineers
The response decays exponentially to a final zero steady state value after a sudden
rise to (k/T) at t=0, Fig.6.9.
The following is the general form of transfer function of a second order element:
k 1 n2 k1
G(s) 2
s 2 n s n s
2 2
2s (6.29)
1
n n
2
For a second order system of gain k1, subjected to step input of magnitude k2, the
step response can be calculated as follows:
k2
X(s) (6.30)
s
k 1k 2 n2 B A1 A2
Y(s) G(s) X(s) k (6.31)
s(s 2 n s n ) s s p1 s p 2
2 2
n2
B Lim s0 2 1
2 (6.32)
s 2 n s n
Assignment: prove that the constants A 1 and A 2 and the roots of the characteristic
equation p1 and p 2 are as given by the following expressions:
1 1
A 1 1 , A 2 1 (6.33)
2 2 1 2 2 1
p1 n 2 1 ,
p 2 n 2 1 (6.34)
69
M Galal RABIE, Automatic Control for Mechanical Engineers
y( t ) k 1 A 1e p1t A 2 e p2t (6.35)
Assignment: Prove that y(t) is tangent to the horizontal axis at the origin.
Four distinct types of response are possible according to the types of roots,
p1 and p 2 as follows:
1. >1 Gives two real negative unequal roots of the characteristic equation.
The constants A 1 and A 2 are real. The transient part of response
decays exponentially. There are no response overshoot and the
response is said to be over-damped.
2. =1 Gives two equal real negative roots. The transient part of response
decays exponentially. There are no response overshoot and the
response is said to be critically damped.
3. 1> >0 Gives a pair of complex conjugate roots. The response is under-
damped; with overshoot and damped transient oscillations.
4. =0 Gives a pair of complex conjugate roots of zero real part and the
response is oscillatory.
The over-damped system is characterized by >1, the roots p1 and p 2 are real
negative and unequal while the coefficients A 1 and A 2 are real. The step response of
an over-damped second order system, given by Eq. 6.35, is plotted in Fig.6.10.
70
M Galal RABIE, Automatic Control for Mechanical Engineers
y( t ) k 1 A 1e p1t A 2 e p2t (6.36)
The increase in the damping ratio results in an increase of the duration of the
transient response; the system response slows down.
An expression for the step response of critically damped second order, =1, is
deduced as follows:
k2
X(s) (6.37)
s
k 1k 2 n2 k n2 1 1 n
Y(s) G(s) X(s) k 2 (6.38)
s(s 2 2n s n2 ) s(s n ) 2 s s n ( s n )
y( t ) k 1 e nt n te nt k 1 (1 n t )e nt (6.39)
The step response of a critically damped system is shown in Fig.6.11. The same
figure carries the response of an over-damped element.
Fig.6.11 Step response of critically damped and over-damped second order systems
In the case of under damped second order system, 1> >0, the characteristic
71
M Galal RABIE, Automatic Control for Mechanical Engineers
equation has two complex conjugate roots. An expression for the step response, y(t),
is deduced as follows:
1 i
A 1 1 and A 1 1 i (6.40)
2 1 2 2
2 1 2
p1 n i 1 2 and p 2
n i 1 2 (6.41)
y( t ) k 1 A 1e p1t A 2 e p2 t (6.42)
i
n n 1 2 t i 1 2 t
n n
y( t ) k 1 A 1e
A 2e
(6.43)
1 i i
k 1 e nt n 1 2 t
e in 1 2 t e in 1 2 t
e in 1 2 t
e (6.45)
2 2 1 2
cos
1 i
2
1
e e i and sin ei e i
2i
(6.46)
y( t ) k 1 e n t cos n 1 2 t sin n 1 2 t (6.47)
1 2
k 1
e nt
1 2
1 2
cos n 1 2 t sin n 1 2 t
(6.48)
k 1
e nt
1 2
sin( ) cos n 1 2 t cos( ) sin n 1 2 t
(6.49)
Or:
y( t ) k 1
e n t
1 2
sin n 1 2 t where cos 1( )
(6.50)
Figure 6.12 shows the different components constituting the step response y(t) while
72
M Galal RABIE, Automatic Control for Mechanical Engineers
Fig.6.13 shows the step response of second order systems with different damping
ratios. These figures show that the response of under damped system, 1> >0,
overshoots during the transient period. The maximum overshoot σ is one of the
parameters characterizing the response of the under damped system.
e nt
(E) sin n 1 2 t
1 2
73
M Galal RABIE, Automatic Control for Mechanical Engineers
y( t ) k 1
e n t
1 2
sin n 1 2 t
(6.51)
dy
dt
k n
e n t
1 2
sin n 1 2 t n e n t cos n 1 2 t
(6.52)
n
e nt
1 2
sin n 1 2 t ne n t cos n 1 2 t 0 (6.54)
tan n 1 t
2
1 2
tan( ) (6.55)
Then
n 1 2 t n where n 0,1, 2, 3,.. (6.56)
n
t (6.57)
n 1 2
t (6.58)
n 1 2
and
y max k k e 1 2 (6.59)
ln()2
y y ss y max k
max e 1 2
and (6.60)
y ss k ln()2 2
100e 1 2
%
(6.61)
74
M Galal RABIE, Automatic Control for Mechanical Engineers
The transient response oscillates by a frequency different than the natural frequency
of the system. The frequency of the transient oscillation of response is called
damped natural frequency, given by the following expression, see Eq. 6.50.
d n 1 2 (6.62)
The duration of one cycle of transient oscillations, also called the period, is given by:
2 2
(6.63)
d n 1 2
k2
X(s) (6.64)
s
k n2 1 s
Y(s) G(s) X(s) k 2 ;
2
where k k 1k 2 (6.65)
s(s n )
2 2
s s n
In the case of zero damping, the response is oscillatory. The output y(t) oscillates
harmonically around the steady state value yss as shown in Fig.6.14.
75
M Galal RABIE, Automatic Control for Mechanical Engineers
Figure 6.15 shows the transient response of second order system to unit step input
for different values of damping ratio.
k2
x( t ) k 2 t, and X(s) (6.67)
s2
k n2 B B A1 A2
Y(s) G(s) X(s) k 21 2 ; k k 1k 2 (6.68)
s (s 2n s n )
2 2 2
s s s p 1 s p 2
2 2 1 2 2 1
A1 and A2 (6.70)
n 2 2 1 n 2 2 1
n n
2
y( t ) k t A 1e p1t A 2 e p2 t (6.71)
n
76
M Galal RABIE, Automatic Control for Mechanical Engineers
Where:
p1 n n 2 1 and p 2 n n 2 1 (6.72)
1. For >1 The response is over damped. In the steady state, the transient
part of response decays and y( t ) k( t 2 / n )
2. For =1 The response is critically damped.
3. For 0< <1 The response is of damped oscillatory character with overshoot
4. For 0 For = 0, the response is oscillatory and for < 0, the system
is unstable, the amplitude of the transient oscillation increase
with time.
k n2
Y(s) G(s) X(s) (6.74)
s 2 2 n s n2
y( t ) k A 1e p1t A 2 e p2 t (6.75)
Where:
p1 n n 2 1 and p 2 n n 2 1 (6.76)
77
M Galal RABIE, Automatic Control for Mechanical Engineers
n
A 1 A 2 (6.77)
2 2 1
The response of a second order element to unit impulse input is plotted in Fig.6.17,
for different values of damping ratio. This figure shows that:
Moreover, for = 0, the response is oscillatory and for < 0, the system is unstable,
as the amplitude of the transient oscillation increases with time.
Y(s) P(s)
G(s) (6.78)
X(s) Q(s)
where P(s) and Q(s) are polynomials in the Laplace operator (s).
The order N of polynomial Q(s) should be greater than or equal to that of P(s). The
polynomial Q(s) has N roots; p1, p2 and pN. Therefore, for a step input, X(s)=k2/s, the
system response y(t) can be calculated by the following expression:
78
M Galal RABIE, Automatic Control for Mechanical Engineers
y( t ) k 1 A 1e p1t A 2 e p2 t ... A N e pNt (6.79)
The transient part of response consists of the summation of terms of the form A i e pit .
The contribution which each term makes towards the overall response depends on
the magnitude and sign of Ai and on the position of the poles in the complex, R-I,
plane.
k 1 n2
G(s) (6.80)
s 2 2 n s n2
p n 2 1 (6.81)
The effect of location of roots, in the R-I plane, on the step response of a second
order element, is illustrated by Fig.6.18. The transfer function has either real roots or
pairs of complex conjugate roots. The responses, plotted in Fig.6.18, are calculated
for unity gain second order system and unit step input function. This figure shows
that:
1. If the two roots are located at the origin, the system is simply a double
integrating member. For bounded input, step input for example, the output
increases continuously with time. Then the system is unstable due to the
unbounded output.
2. For roots located on the left hand side of the real (horizontal) axis, the two
roots are real negative. The step response has the form:
y( t ) k 1 A 1e p1t A 2 e p2 t . Therefore, if the roots are located on the horizontal
axis (negative real roots) the step response is damped and non-oscillatory.
The displacement of the roots to the left decreases the settling time.
3. If the roots are located at the vertical axis, zero damping, ζ = 0, the system is
oscillatory. It has two complex conjugate roots with zero real part. The step
response is of the form y( t ) k1 cos( n t ) . The displacement of the root
upwards increases the frequency of oscillations.
79
M Galal RABIE, Automatic Control for Mechanical Engineers
4. If the damping ratio is in the range 1 > ζ > 0, the two roots are complex
conjugate with negative real part. They are located to the left of the vertical axis.
The roots are; p n i 1 2 , where the real part and imaginary parts are:
R n and I n 1 2
80
M Galal RABIE, Automatic Control for Mechanical Engineers
1
Figure 6.19 shows the location of the root in the R-I plane, where cos ( ) . In
this case the step response has an overshoot and damped transient oscillations.
The farthest poles from the imaginary axis will have contribution to the transient
response, which decay most rapidly. The system response is influenced mostly
by the poles closest to the imaginary axis, called the dominant poles. Generally,
the poles located 5 times far to the left from the vertical axis relative to the
dominant roots can be neglected.
The increase of the magnitude of imaginary part of root increases the frequency
of transient oscillations.
5. The presence of any root with positive real part, located to the right of vertical
axis, increases the magnitude of system output without limit and the system is
unstable. The instability means that the system output is unbounded for bounded
input.
A first order system can be completely described by specifying the time constant and
gain of transfer function. A second order system can be clearly described by
specifying its gain, damping ratio and natural frequency. For higher order system, the
algebraic equation is not very helpful since the form of the response is not readily
apparent and the plot of the response is not satisfactory since a numerical
description is required for analysis. Generally, with reference to Fig.6.20, the
following parameters are used to describe the step response of a system:
81
M Galal RABIE, Automatic Control for Mechanical Engineers
y max y ss y max y ss
100% (6.82)
y ss y ss
1 2 (6.83)
100e %
Delay time is the time required for the response to reach 50% of the final steady
state value
Rise time is the time required for the response to rise from 5% to 95% (or from 10%
to 90%) of the step size. This definition avoids the practical difficulty of having to
determine the exact start of the transient response.
Settling time is the time required for the response to reach an end state within 5% of
final, steady state, value. For second order system, the settling time is:
3 0.5 ln(1 2 ) 3
ts Or t s (6.84)
n n
82
M Galal RABIE, Automatic Control for Mechanical Engineers
Equation 6.84 can be used only to estimate the settling time for an under-damped
second order element. It cannot be used to calculate the damping ratio or the natural
frequency. As for the first order element, the following expression can be deduced
systematically.
ts 3 T (6.85)
Steady state error is the steady state deviation between the actual response y(t)
and the required steady state value.
Period is the duration of one complete cycle of the transient response oscillations
The best response is that which gives the minimum value of performance index. For
a second order element, the optimum (best) response is obtained at = 0.7 for the
IAE and ITAE. On the other hand, using the IES, the best response is obtained at
= 0.5. In practice, a damping factor of 0.7 is preferred.
83
M Galal RABIE, Automatic Control for Mechanical Engineers
the goodness of fit and then to adjust the model and its parameters values to
minimize the chosen error index.
If the step response apparently rises exponentially to the new steady state, it is likely
that the system is of first order. The time constant can be found by direct
measurement on the step response plot, as shown in Fig.6.7.
k1 k / k 2 (6.89)
1. In the case of step input of magnitude k2, if the steady state value is k then the
gain of transfer function is given by: k1 = k /k2
2. Find the period of one oscillation cycle, from the plotted response and calculate
the damped natural frequency, d 2 / rad/s.
3. Find the overshoot ratio σ from the plot then calculate the damping ratio.
e
1 2
ln()2 (6.90)
Or
ln()2 2
84
M Galal RABIE, Automatic Control for Mechanical Engineers
6.5 EXERCISE
1. Derive an expression for and plot, the transient response of the systems
described by the following transfer functions to step input of magnitude 4.
0.45 0 .1 1
G(s) , G(s) and G(s)
s 0. 5 2s 0.5 3s 1
2. Estimate the transfer function of the system whose step response is given in the
following figure. Find also, from the graph, the settling time.
3. Estimate the transfer function of the system whose step response is given in the
following figure. Find also, from the graph, the settling time, rise time and delay
time.
85
M Galal RABIE, Automatic Control for Mechanical Engineers
o 240
( s)
i 15s 60s 240
2
Plot the transient response of the system to the input angular displacement ( t )
where i ( t ) 0 at t 0 and i ( t ) / 3 at t 0
5. The following is the characteristic equation of a second order system.
s 2 0 .6 s 9 0
6. The following are two block diagrams of two control systems. Find the natural
frequency, damping ratio, damped natural frequency, settling time and
maximum percentage overshoot for each of them.
7. Find an expression for the response of the system having the following transfer
function to a unit step input.
s(3 5s)
G(s)
(s 1)(s 2)(s 3)
8. The following figures show the responses of four different systems to step
inputs of different magnitudes.
a) Find the settling time, rise time, delay time, period of one oscillation cycle
and maximum percentage overshoot.
b) Write the system transfer function.
c) Calculate the transfer function gain, damping ratio and natural frequency of
each system.
86
M Galal RABIE, Automatic Control for Mechanical Engineers
(A)
(B)
(C)
87
M Galal RABIE, Automatic Control for Mechanical Engineers
(D)
9. Find the value of constant K for the critical damping of the system described by
the following block diagram.
10. The step response of an under damped second order element is given by the
following expression.
y 1
e n t
1 2
sin n 1 2 t
a) Derive an expression for the maximum percentage overshoot.
1
G(s)
s 15s 400
2
88
M Galal RABIE, Automatic Control for Mechanical Engineers
128L
where R = line resistance=
D 4
D 2L
C= line capacitance=
4B
4L
I= line inertia=
D 2
a) If the line is open at its extremity, pL= 0, the relation between QL(s) and Po(s)
will be given by:
QL 1
( s)
Po Is R
PL 1
Po ICs RCs 1
2
Calculate the parameters of the transfer functions and plot, in scale, the step
response of the line in the two cases.
89
M Galal RABIE, Automatic Control for Mechanical Engineers
90
M Galal RABIE, Automatic Control for Mechanical Engineers
7 FREQUENCY
RESPONSE
7.1 INTRODUCTION
Input signal
x( t ) x( t ) sin( t ) (7.1)
91
M Galal RABIE, Automatic Control for Mechanical Engineers
AMPLITUDE RATIO is the ratio of the amplitude of the output sinusoidal wave to that
of the input in the steady state conditions (also called the magnitude or gain).
y( t )
Gain (7.3)
x( t )
PHASE SHIFT is the shift in phase ( ) between the output sinusoidal wave relative
to the input (also called the phase).
FREQUENCY RESPONSE Is the variation of amplitude ratio and phase with the
frequency of the input signal.
92
M Galal RABIE, Automatic Control for Mechanical Engineers
The following is the general form of the transfer function of a linear system:
A A
X(s) (7.5)
s
2 2
(s i)(s i)
The first two terms of y(t) describe the steady state response, while the other terms
give the transient part of the response. For a system to be stable, the poles p1, p2,...,
pn should be real negative or complex of negative real part. In this case, the terms of
the transient response decay as the time increases. Therefore, after all transients
have died out, the steady state part of response is:
AG(s) A
A 1 lim si G(i) (7.10)
s i 2i
AG(s) A
A 2 lim si G( i) (7.11)
s i 2i
93
M Galal RABIE, Automatic Control for Mechanical Engineers
i
G ( i ) G ( i ) e (7.13)
where
G(i) R 2 G(i) I2 G(i) (7.14)
IG(i)
tan 1 G(i) (7.15)
RG(i)
A A
y( t ) G(i) e i e it G(i) e i e it (7.16)
2i 2i
y( t )
A
2i
G(i) e i( t ) e i( t ) (7.17)
y( t )
The magnitude ratio = g G(i) and the phase shift G(i)
x( t )
Then, for linear system, the frequency response is independent of the magnitude of
input signal. The gain (magnitude ratio) and phase (phase shift), can be calculated
using the transfer function G(s) according to the following procedure:
1. Find G(i) by replacing (s) by (i) in the expression of G(s).
2. Separate the real and imaginary parts of G(i) by multiplying both the
numerator and denominator by the conjugate of the denominator.
3. Find the expressions for phase and gain.
4. Substitute for to find the gain and phase corresponding to different values
of the frequency of input signal.
Example 7.1 Find the expressions for the gain and phase of the first order element.
1
G(s)
Ts 1
1 1 iT 1 T
G(i) i
1 iT 1 iT 1 T
2 2
1 2T 2
1 2 T 2 1
Gain G(i) 2 2 2
2 2 2
(1 T ) (1 T ) 1 2 T 2
94
M Galal RABIE, Automatic Control for Mechanical Engineers
The Nyquist diagram is a plot showing the variation of the magnitude and phase in
polar coordinates. The curve drawn is the locus of the termini of the system output
vectors. In this locus, numerical values of ω should be indicated against the points.
The following expressions for gain and phase of first order element were deduced in
example 7.1.
1
RG(i) (7.19)
1 2 T 2
T
IG(i) (7.20)
1 2 T 2
1
Gain (7.21)
1 2 T 2
1
I2 R 2 R (7.23)
1 2 T 2
I2 R2 R 0 (7.24)
2 2
1 1
I2 R (7.25)
2 2
The frequency response of a first order element is plotted in the polar plot of Fig.7.3.
The plot is a semi-circle with center (0.5,0) and radius 0.5. For some input signal
frequencies, the corresponding points on the plot are marked by determining the
corresponding phase; Phase tan 1(T ) .
95
M Galal RABIE, Automatic Control for Mechanical Engineers
For 0 , the phase = 0 and for 1/ T , the phase = - 45o. The further increase in
the frequency increases the phase lag up to a maximum value of 90o. Thus, the polar
plot of a first order element occupies only the first quadrant of the R-I plane.
If the transfer function gain is different from unity, (k1=2, for example), the scale of
both of I and R axes is multiplied by k1. The frequency response plot starts at the
point (k1, 0) for 0 .
The expressions for the gain and phase of second order element are deduced as
follows:
n2
G(s) (7.26)
s 2 2 n s n2
1 / n
2
2 / n
G(i) i
1 / 4
n
2 2 2 2
/ n2 1 / 4
n
2 2 2 2
/ n2
(7.28)
1
Gain G(i)
1 / 4
n
2 2 2 2
/ n2
(7.29)
96
M Galal RABIE, Automatic Control for Mechanical Engineers
2 / n 2 / n
Phase tan 1 2
tan 1 (7.30)
1 / n 1 / n 2
The study of the expressions for gain and phase of second order element shows that:
The Nyquist plot of a second order element is shown by Fig.7.4. It occupies the first
two quadrants of the R-I Plane.
If the transfer function gain is different from unity, (k1 for example), the scale of both
of I and R axes is multiplied by k1. The frequency response plot starts at the point
(k1, 0) for 0 .
1
G(s) (7.31)
s
97
M Galal RABIE, Automatic Control for Mechanical Engineers
1 1 1
G(i) i ; I &R 0 (7.32)
i
1 1 1
G(i) i ; I &R 0 (7.32)
i
1
Gain (7.33)
The whole plot lies on the vertical (imaginary) axis. The phase lag is -90o for all
frequencies. The magnitude is inversely proportional to the input signal frequency .
In the case of higher order elements, the frequency response can be calculated from
the transfer function according to the procedure discussed in sec. 7.2. However, if
the transfer function appears in factorized form, the phase and gain are obtained
more easily by thinking of the system as a number of elements connected in series,
Fig.7.6. The overall phase for any given input signal frequency is obtained by adding
the individual phase components and the overall system gain is obtained by
multiplying together the individual gains.
If the polar plots of the individual elements are given (From experimental results for
example), the polar plot of the system can be constructed graphically.
98
M Galal RABIE, Automatic Control for Mechanical Engineers
Example 7.2 Draw the Nyquist plot of the system described by the following transfer
function:
25 1 25
G(s)
( 4s 1)(s 2 10s 25) 4s 1 s 2 10s 25
Given the Nyquist plot of the first and second order elements, the frequency
response of the system described by G(s) can be calculated point by point as shown
by Fig.7.7.
For certain frequency, find the phases φ1 & φ2 and gains g1 &g2 of the second order
and first order elements respectively. Then find the resultant gain (g = g1 x g2) and
resultant phase (φ = φ1 + φ2). Repeat this process point by point to construct the
resultant plot. The resulting polar plot of a third order element occupies three
quadrants. Generally, the polar plot of a system of order n occupies the first n
quadrants of the plane.
Fig.7.7 Graphical plotting of the Nyquist diagram of a third order element, given the
plots of its basic elements
7.4.1 Introduction
The Bode diagram is a plot of the frequency response consisting of two different
plots; magnitude plot and phase plot, against the frequency of input signal, Fig.7.8.
99
M Galal RABIE, Automatic Control for Mechanical Engineers
Figure 7.6 shows a system consisting of elements connected in series. The Bode
diagram of the system can be obtained if the Bode plots of its components are
known, by graphical addition of phase and gain plots.
100
M Galal RABIE, Automatic Control for Mechanical Engineers
Therefore, the Bode diagrams of the majority of systems are easily constructed if the
Bode plots of these typical elements are known.
The constant term (also called proportional term or gain term) presents a constant
gain and zero phase as shown in Fig.7.9.
1
G(s) (7.47)
s
1 1 1
G(i) i , I & R=0 (7.48)
i
1
G(i) (7.49)
I
Phase tan 1 90 o (7.51)
R
The Bode plot of an integrating member, G(s)=1/s, is shown in Fig.7.10. The gain plot
is a straight line inclined by -20 dB/decade, (one decade is the distance on the
horizontal axis along which the frequency is increased by 10 times). The plot is a
straight line of slope – 20 dB/decade passing by the point (1,0). It presents a
constant phase of -90 degrees.
Assignment: Derive the expressions for the gain and phase of integrating and
differentiating members of different powers and discuss their plots given in Fig.7.10.
101
M Galal RABIE, Automatic Control for Mechanical Engineers
1
G(s) (7.52)
Ts 1
1 1 iT 1 T
G(i) i (7.53)
1 iT 1 iT 1 T
2 2
1 2T 2
1 2 T 2 1
Gain G(i) 2 2 2
2 2 2
(7.54)
(1 T ) (1 T ) 1 2 T 2
102
M Galal RABIE, Automatic Control for Mechanical Engineers
1
Gain 20 log 20 log 1 2 T 2 (7.55)
2 2
1 T
The gain and phase plots of the first order elements are shown in Fig.7.11. These
plots can be simplified by asymptotic approximation lines as follows:
103
M Galal RABIE, Automatic Control for Mechanical Engineers
The gain plot is a straight line inclined by -20 dB/decade, intersecting the
horizontal line at ω=1/T.
These two lines intersect at a corner frequency (Break point) c 1/ T . The real plot
rounds off the junction. The maximum difference is 3 dB at ω=ωc, as shown in
Fig.7.11. At a frequency of ω = 0.5/T and ω = 2/T, the difference between the
straight-line approximation and real plot is 1 dB. Therefore, for simplicity, the straight-
line approximation can be used.
A linear approximation may be also plotted for the phase as shown in Fig.7.11. But,
the real phase plot is frequently used.
Assignment: Prove that the gain and phase of the simple lead element are given by
the following expressions and plot the Bode diagram; real plot and straight line
approximation. Show that the plots are mirror image of the simple phase-lag plots
around the zero gain and zero phase lines
n2
G(s) 2 (7.60)
s 2 n s n2
1
G(i)
1 / 4
n
2 2 2 2
/ n2
(7.61)
Gain 20 log 1 / n 4
2 2 2 2
/ n2 dB (7.62)
104
M Galal RABIE, Automatic Control for Mechanical Engineers
2 / n
Phase tan 1 (7.63)
1 / n 2
The Bode plots of a second order element are shown in Fig.7.12, for different values
of damping ratio. A straight line approximation of the gain is also plotted.
105
M Galal RABIE, Automatic Control for Mechanical Engineers
These two straight lines intersect at n ; this is called the corner frequency
(Break point). At this frequency the phase = -90o and the gain depends only on the
damping ratio.
G(s) s 2 2 n s n2 / n2 (7.64)
Assignment: Prove that the gain and phase of the quadratic lead element are given
by the following expressions and plot the Bode diagram; real plot and straight line
approximation.
Gain 20 log 1 / n 2 4
2 2 2
/ n2 dB (7.65)
2 / n
Phase tan 1 (7.66)
1 / n 2
The magnitude and phase have the same numerical values as those of second order
element, but have opposite sign. They are represented on the Bode plot by families
of curves which are mirror image of those of the second order element, with respect
to the zero gain and zero phase lines.
Example 7.3 Plot the Bode diagram of the system described by the following transfer
function:
(5 0.5s)(2 0.05s)
G(s)
(1 0.25s)(1 0.004s)(1 0.004s)
1 1
G(s) 10 (1 0.1s) (1 0.025s)
(1 0.25s) (1 0.004s) 2
106
M Galal RABIE, Automatic Control for Mechanical Engineers
The Bode plot of the system can be drawn by plotting the frequency response of the
individual elements, by straight line approximation and finding out the gain plot of the
system by graphical summation as shown in Fig.7.13. The real gain plot of the
system and the straight-line approximation are shown in Fig.7.14. The point by point
plot is directly calculated by using the gain expression deduced from the transfer
function.
Example 7.4 Find the transfer function of the subsystem whose straight-line
approximation of the Bode gain plot is given in Fig.7.15.
107
M Galal RABIE, Automatic Control for Mechanical Engineers
108
M Galal RABIE, Automatic Control for Mechanical Engineers
The Nichol's chart enables to find, in a simple way, the frequency response of the
closed loop, if the open loop response is known. The open loop frequency response
is plotted on the horizontal and vertical coordinates; the horizontal is for the phase in
degrees while the vertical is for the gain in decibels. The values of frequency of the
input signal (ω) should be indicated on the plot in a way similar to the polar (Nyquist)
plot. The gain of the closed loop response can be read off directly from the
intersection of the open loop frequency response plots with the contours encircling
the point (-180o, 0 dB). The phase of the closed loop response is found at the
intersection with the contours radiated from the point (-180o, 0 dB). The Nichol's chart
is of special interest for the analysis of system stability.
7.6 EXERCISE
1 1
(a) G(s) (b) G(s)
1 2s s (1 2s)
1 1
(c) G(s) (d) G(s)
(s s 1)
2
s (s s 1)
2
1 1
(e) G(s) (f) G(s)
(1 2s)(s 2 s 1) s(1 0.02s)(1 0.01s)
2.Draw the Bode plot; magnitude only, for the following systems using the straight-
line approximations.
100s 1
(a) G(s) (b) G(s)
s 10 s s 1
2
2 2
(c) G(s) (d) G(s)
(0.5s 1)(0.125s 1) (s 1)(s 2)(s 2 3s 16)
s
(e) G(s) (s 10) 2
100
3.Plot the Bode and Nyquist diagrams for the compensated open loop of the
system described by the following block diagram, assume three different values
for k<20.
109
M Galal RABIE, Automatic Control for Mechanical Engineers
4.Draw the magnitude and phase Bode plots for the system (c) of problem 1.
6.Find the transfer functions for the systems having the following magnitude Bode
plots.
a)
b)
110
M Galal RABIE, Automatic Control for Mechanical Engineers
c)
111
M Galal RABIE, Automatic Control for Mechanical Engineers
112
M Galal RABIE, Automatic Control for Mechanical Engineers
8 FEEDBACK SYSTEM
ACCURACY AND STABILITY
8.1 INTRODUCTION
In the case of open loop systems, the system output is not compared with the
required value (which is determined by the input signal). The external disturbances,
or deterioration with increasing age, may cause changes of the system parameters
and its transfer function. Therefore the actual output may deviate far from the
required one. Introducing a feedback path allows the actual system output to be
compared with the input signal (required output), Fig.8.1. An error signal is generated
and forms the actuating signal. An element, of transfer function H(s), is inserted in
the feedback path. It represents the dynamic characteristics of the transducer which
measures the output signal and converts it to the same type as the reference input
signal. The main reason for introducing the feedback is to minimize the error between
the actual and desired system outputs.
The magnitude of the steady state error is a measure of the system accuracy.
Closing the loop should reduce or eliminate the steady state error. But the system
can be very oscillatory or even unstable.
The steady state error is a measure of the system accuracy. It is the value to which
the error signal tends as the transient disturbances die out.
Considering the system shown in Fig.8.1, the error signal is given as follows:
113
M Galal RABIE, Automatic Control for Mechanical Engineers
X(s)
E(s) (8.2)
1 G(s) H(s)
The steady state error ess is the limiting value of e(t);
Equation 8.4 shows that the steady state error depends on the compensated open
loop transfer function, G(s)H(s) and input signal X(s).
k2
X(s) (8.5)
s
s X(s) k2
e ss Lim Lim (8.6)
s 0 1 G(s) H(s) s0 1 G(s) H(s)
k2
e ss (8.7)
1 Lim G(s)H(s)
s 0
Or
k2
e ss , where k p Lim G(s)H(s) (8.8)
1 kp s0
It is evident that, in the case of a step input, the steady state error becomes zero only
if k p . This condition is satisfied if G(s)H(s) includes an integrating element (a
factor sn in the denominator, n 0 ).
If the compensated open loop transfer function contains no integral term, the system
response has a nonzero steady state error ess.
It is more convenient to define the relative steady state error rather than the absolute
error. The relative steady state error is defined as:
114
M Galal RABIE, Automatic Control for Mechanical Engineers
The position error coefficient equals the gain of the compensated open loop transfer
function, k p Lim G(s)H(s) . Generally, greater open loop gain results in smaller
s0
steady state error.
Example 8.1 Find the required position error coefficient kp if the maximum allowable
relative steady state error e ss 1% .
1
ess 0.01 or 0.01 then k p 99
1 kp
Example 8.2 The following system, Fig.8.2, is excited by a step input of magnitude 5,
calculate the steady state error then find the required gain of the open loop if the
maximum allowable steady state error is less than 2% of the required output.
Fig.8.2
60
G(s)H(s) , k2 5
(s 2)(s 10)
k p Lim G(s)H(s) 3
s0
1 1
e ss 0.25 25%
1 k p 1 3
115
M Galal RABIE, Automatic Control for Mechanical Engineers
e ss
e ss or e ss k 2 e ss 5 0.25 1.25
k2
1
ess 0.02 or 0.02 then k p 49
1 kp
Therefore, the gain of the open loop should be increased K times by introducing a
proportional element (where; K 49 / 3 ). Figure 8.3 shows the system after adding a
gain within the recommended range.
Fig.8.3
k
X(s) (8.12)
s2
s X(s) k
e ss Lim Lim (8.13)
s 0 1 G( s) H( s) s 0 s sG( s) H( s)
k
e ss
Lim s G(s)H(s) (8.14)
s 0
Or
k
e ss , where k v Lim s G(s)H(s) (8.15)
kv s0
116
M Galal RABIE, Automatic Control for Mechanical Engineers
In the case of ramp input, the steady state error becomes zero, only if k v . This
condition is satisfied if, G(s)H(s), includes a factor sn in the denominator, where
n 2.
A system is said to be stable if its response to any bounded input is also bounded.
For a system to be useful, it should be stable. It has been shown (sec. 6.2.5), that the
transient response of a dynamic system is highly affected by the location of the roots
of the characteristic equation in the R-I plane. For a system to be stable, the roots
should take place in the left half of the s-plane, or, in other words, the roots should be
real negative or complex with negative real part and none of them is of zero real part.
Therefore, a decision about the system stability can be reached if the locations of the
roots of the characteristic equation of the closed loop in the R-I plane are known.
The roots of the first and second order characteristic equations are easily found.
However, in the case of higher order characteristic equation, finding out the roots is
too complicated. Therefore, around 1880, Routh and Hurwitz developed
independently a method for determining whether any root of the characteristic
equation has positive real part.
A necessary but not sufficient condition for the polynomial roots to have non zero
negative real parts is that all of the coefficients of the characteristic equation should
have the same sign and that none of them is zero. If this condition is satisfied, then
the necessary and sufficient condition of stability is that the Hurwitz determinants of
the polynomial must be positive, where the determinants are given by:
a1 a3 a5 a7
a1 a3 a5
a1 a3 a a2 a4 a6
D1 a 1 , D 2 , D3 a0 a2 a4 , D4 0 , etc. (8.17)
a0 a2 0 a1 a3 a5
0 a1 a3
0 0 a2 a4
117
M Galal RABIE, Automatic Control for Mechanical Engineers
n a0 a2 a4 a6 …
n-1 a1 a3 a5 a7 …
n-2 b1 b2 b3 b4 …
n-3 c1 c2 c3 …
n-4 d1 d2 …
… … …
0 …
The first two rows are formed by the coefficients of the characteristic equation, while
the subsequent rows are calculated as follows:
b1c 2 b1c 3
d1 b 2 d2 b 3 ,… (8.20)
c1 c1
The values of the elements of the second column of this array (ao, a1, b1, c1, d1, ..)
should be positive. Every change of sign means that there exists a root with positive
real part.
Example 8.3 Apply the Routh-Hurwitz stability criterion to check the stability of the
system having the following characteristic equation:
All of the coefficients are positive and none of them is zero. The Routh array is
constructed as follows:
4 1 1 2
3 2 4 0
2 -1 2 0
1 8 0
0 2
118
M Galal RABIE, Automatic Control for Mechanical Engineers
There are two sign changes in the second column, which indicates that the
characteristic equation has either two real positive roots or a pair of complex-
conjugate roots with positive real part. Actually, this characteristic equation has four
roots, two real roots and one pair of complex conjugate roots.
p1 0.515589 , p 2 2.1877 ,
When constructing the Routh array, two kinds of difficulties can occur because of
existence of zeros in the second column of the array:
A zero appears in the second column
A complete row of zero elements appear
If a zero appears in the first column, the division by zero should be avoided as
follows. Replace s by 1/σ in the characteristic equation and apply the Routh-Hurwitz
criterion to the new equation in σ. If the polynomial in σ has neither real positive roots
nor complex roots with positive real parts then, the polynomial in s also has neither
real positive roots nor complex roots with positive real parts.
5 1 4 2
4 1 4 1
3 0 1
2
1
0
σ 5 +2 σ4 +4 σ3 +4 σ2 + σ +1 = 0
5 1 4 1
4 2 4 1
3 2 0.5
2 3.5 1
1 -0.07
0 1
119
M Galal RABIE, Automatic Control for Mechanical Engineers
There are two changes in sign in the first column then the characteristic equation has
two roots of positive real parts. Then as σ has a positive value, s also will have a
positive value, since s=1/ σ.
If a complete row of zero elements appears, the equation having the row just above
that one is to be differentiated. The coefficients of the resulting equation replace the
zero values of the row and the construction of the array can be continued.
3 1 16
2 10 160
Write the equation
10s2+160 d / ds
20s+0
1 0 20 0 0 Re place 20 0
0 160
6 1 5 6 1
5 3 12 9
4 1 3 1
3 3 6
2 1 1
1 3
0 1
There exist no sign changes in the second column then the system having this
characteristic equation is stable.
Nyquist stability criterion judges the stability of closed loop system on the basis of its
compensated open loop frequency response. It decides if the system is stable or not
and determines the degree of stability.
The Routh-Hurwitz criterion determines if the system is stable or unstable, but it does
not give any information about the degree of its stability. The system may be stable,
but if the roots of its characteristic equation are too close to the vertical axis of the R-I
120
M Galal RABIE, Automatic Control for Mechanical Engineers
plane, the transient oscillations may take too long time to decay, Fig.6.18. On the
other hand, the polar plot (Nyquist diagram) of the compensated open loop transfer
function, G(iω)H(iω), enables to determine whether any of the roots of the
characteristic equation of the closed loop system have positive real part, without
actually evaluating them. This plot shows also the degree of system stability and how
to improve it. The experimental results of the frequency response evaluation can be
directly used without any need to determine the system transfer function.
The Nyquist stability criterion states that if the compensated open loop of a
system is stable then the system with closed loop is also stable provided that
the open loop polar plot does not enclose the critical point (-1,0).
Figure 8.4 shows the case of an unstable system whose open loop polar plot
encircles the point (-1, 0) and a stable system whose open loop polar plot does not
encircle this critical point.
When the system is excited by a harmonic signal, the phase shift varies with the
frequency of input signal. The polar plot of the compensated open loop intersects
with the negative part of the real, horizontal, axis at certain input frequency, called
the phase cross-over frequency. At this frequency, the phase lag is 180o and the gain
is g, Fig.8.4. The Nyquist stability criterion can be explained as follows.
If the reference input to the closed loop is a sine wave x( t ) A sin(t ) , then the
signal returning to the error detector will have a different amplitude and phase,
Fig.8.5(a). At the phase cross-over frequency, the phase lag is 180o and the
magnitude ratio is (g); f(t)=ge(t). Then, in the steady state, the returning signal is
inverted and added to reference input, Fig. 8.5(b) and the steady state error is given
as follows:
e( t ) x( t ) g e( t ) (8.21)
121
M Galal RABIE, Automatic Control for Mechanical Engineers
x( t ) A
e( t ) sin(t ) (8.22)
1 g 1 g
There are three different stability cases, depending on the magnitude ratio at the
phase cross-over frequency:
The amplitude of the returning signal is less than that of the input signal, g 1.
The amplitude of the error signal, converges to a constant value of A /(1 g) .
Consequently, the system output converges to a steady state bounded value
and the system is stable, Fig.8.5(c).
The magnitude ratio at the phase cross-over point is equal to one ( g 1 ). The
open loop polar plot intersects the horizontal axis at the critical point (-1, 0).
The amplitude of the error signal tends to infinity and the system becomes
unstable. Moreover, if the input signal is removed, the error signal and system
output will continue to oscillate.
The amplitude of the returning signal is greater than one, g 1 . The open loop
polar plot intersects the horizontal axis at a point (-g, 0). The locus encloses
the point (-1,0). The amplitude of error signal will build up continuously and the
system is unstable, Fig.8.5(d). Moreover, if the input signal is removed, the
error signal continues to increase or grow up (for g 1).
Fig.8.5(c) Steady state input, feedback Fig.8.5(d) Steady state input, feedback
and error signals for feedback system, and error signals for feedback system,
excited by harmonic input at the phase excited by harmonic input at the phase
cross-over point, g 1 cross-over point, g 1
122
M Galal RABIE, Automatic Control for Mechanical Engineers
The magnitude ratio g depends on the gain k of the compensated open loop. The
effect of open loop gain on the system response is illustrated by example 8.7.
Example 8.7 Plot the open loop Nyquist diagram, the closed loop response to unit
step input and the closed loop Bode diagram of the following system, for open loop
gain k=1, 5, 8 & 10. Find the relative steady state error for each case.
Apply Routh-Hurwitz stability criterion to find the limiting value of the open loop gain,
for system stability and discuss the obtained results.
Fig.8.6
s 3 3 s 2 3s 1 k 0
3 1 3
2 3 1+k
1 k
1 3 0
3
0 1+k
1 k
The system is stable if 3 >0 and 1+k>0
3
Or 8 > k > -1
123
M Galal RABIE, Automatic Control for Mechanical Engineers
1
The relative steady state error is e ss , Where k p Lim G(s)H(s)
1 k p s0
k kp e ss e ss %
1 1 0.5 50
5 5 0.1667 16.67
8 8 0.1111 11.11
10 10 0.0909 9.09
The closed loop gain is k/(1+k). Figure 8.7 shows the Nyquist plot of the open loop
for different values of open loop gain k. The step response of closed loop to unit step
input is plotted in Fig.8.8 and its Bode plot is shown in Fig.8.9.
Fig.8.8 Response of the closed loop to unit step input for different values of open
loop gain k
The study of Figs. 8.7, 8.8 and 8.9 shows the following:
1. For k=1, the open loop Nyquist plot intersects the real axis at (-0.125,0); far
enough from the critical point. The transient response is stable with considerable
steady state error and slight overshoot. The closed loop Bode plot shows low
magnitude ratio at resonance frequency.
2. For k=5, the open loop Nyquist plot intersects the real axis at (-0.625,0); nearer to
the critical point. The system is still stable since the locus does not enclose the
critical point. The transient response presents considerable transient damped
oscillations, overshoot and smaller steady state error. The Bode plot shows higher
124
M Galal RABIE, Automatic Control for Mechanical Engineers
magnitude ratio at resonance and the slope of the phase curve becomes steeper.
It is evident that this system is less stable than the preceding case.
3. For k = 8, the open loop Nyquist plot passes through the critical point. The step
response is oscillatory. The gain plot of Bode diagram shows that the closed loop
magnitude ratio tends to infinity at resonance, which indicates clearly the system
instability.
4. For k = 10, the open loop Nyquist plot encloses the critical point and the system is
unstable. The step response becomes unbounded. The closed loop Bode plot
shows finite resonance gain and the phase becomes positive for frequencies
greater than the resonance frequency.
Fig.8.9 Bode diagram of the closed loop for different values of open loop gain.
Generally, it is not sufficient to have a stable system, but the system should have
certain degree of stability, relative stability. In the time domain, the relative
stability can be measured by the maximum percentage overshoot, settling time and
damping ratio. In the frequency domain, the relative stability can be evaluated, on the
basis of the Nyquist stability criterion, by means of the compensated open loop polar
plot; G(iω)H(iω). The closeness of the compensated open loop polar plot to the
125
M Galal RABIE, Automatic Control for Mechanical Engineers
critical point gives an indication of the degree of stability. The gain margin and
phase margin are measures of the degree stability, Fig.8.10. These margins are
defined as follows:
Fig.8.10 Definition of the gain margin and phase margin in the polar coordinates
Phase cross-over point is a point at which the Nyquist plot of G(i)H(i) intersects
the negative real axis where G(icp )H(icp ) 180 o .
Phase cross-over frequency, cp is the frequency at the phase crossover point.
Gain cross-over point is a point on the compensated open loop polar plot at which
the magnitude of G(i)H(i) is unity, G(i)H(i) 1.
Gain cross-over frequency, cg is the frequency at the gain crossover point.
Gain margin (GM) is the allowable increase in the open loop gain in decibels (dB)
before the closed loop system becomes unstable.
1 1
Gain M arg in (GM) (8.23)
g G(icp ) H(icp )
Or
126
M Galal RABIE, Automatic Control for Mechanical Engineers
Example 8.8 For the system described by the following block diagrams:
(a) Find the values of constant k for a steady state error e ss = 0.1 and for
e ss = 0.05.
(b) Derive the expressions for the open loop gain and phase.
(c) Calculate the phase cross-over frequency and find the value of constant k if the
gain margin is 10 dB
(d) Calculate the gain cross-over frequency and phase margin for the value of k
calculated in (c)
(e) Give your comment on the calculation results.
k
(a) k p Lim G(s)H(s) Lim k
s0 s 0 s 1 3
1 e ss
kp
e ss
127
M Galal RABIE, Automatic Control for Mechanical Engineers
The increase in gain reduces the steady state error. For e ss = 0.05, the open loop
gain k should be: k =19
k
Gain G(i)
1
(b) 3
2
k k
Gain g
1 2
cp
3
8
Or g 1/ 10 , Then k 8 / 10 2.53
For gain margin GM=10 dB, the open loop gain should be k=2.53
k
g G(icg ) 1
1 2
cg
3
3 tan1(cg ) 128.36o
128
M Galal RABIE, Automatic Control for Mechanical Engineers
(e) To achieve the required stability level, the open loop gain should be k 2.53 .
Meanwhile, to achieve the required precision, the open loop gain should be
k 19 . The two demands are in contradiction. They cannot be satisfied
simultaneously without implementation of a convenient controller.
Fig.8.11 Gain and phase margins on the open loop Bode plot
The Bode plot of the open loop frequency response G(i)H(i) is a powerful tool for
the stability analysis, Fig.8.11. It enables to find the phase and gain margins by direct
readings on the plot. The phase margin is determined as follows:
1. Find the gain crossover point on the gain plot; the point on the plot
corresponding to gain = 0 dB. Then find the gain crossover frequency cg .
2. Find the corresponding phase, on the phase diagram G(icg )H(icg ) .
3. The phase margin can be directly obtained from the phase diagram as shown in
Fig.8.11.
129
M Galal RABIE, Automatic Control for Mechanical Engineers
The gain margin can be found from the Bode diagram as follows:
1. Find the phase crossover point on the phase diagram; the point at which the
phase equals –180o. Then find the phase crossover frequency cp .
2. Find the corresponding gain on the gain plot.
3. The gain margin can be directly found from the gain plot as shown in Fig.8.11.
Gain M arg in (GM) 20 log G(icp ) H(icp )
When plotting the compensated open loop frequency response on Nichol's chart, the
gain and phase margins can be directly obtained as shown in Fig.8.12.
The phase margin is given by the horizontal distance from the origin (-180o, 0) to the
intersection point of the open loop plot with the horizontal (zero dB) axis.
The gain margin is given by the vertical distance from the origin (-180o, 0) to the
intersection point of the open loop plot with the vertical (-180o) axis.
130
M Galal RABIE, Automatic Control for Mechanical Engineers
8.4 EXERCISE
60
G(s)H(s)
(s 2)(s 10)
d 2 o
t m 0.004 e k Nm
dt 2
e i o
d 2 o d o
J f tm
dt 2 dt
2000
G(s)H(s)
(s 2)(s 10)
131
M Galal RABIE, Automatic Control for Mechanical Engineers
4. For the system represented by the following block diagram, find the
positional error coefficient, the steady state error ess and the steady state
value of y(t) if a step input of magnitude k2 is applied.
(a) 0.1s 3 s 2 k 0
(b) 0.1 s 2 1.1 s k 1 0
(c) 0.1 s 3 s 2 s k 0
(d) T1T2 s 3 (T1 T2 ) s 2 s k 0
(e) s 4 2 s 3 6 s 2 8 s k 0
(f) s 4 20k s 3 5 s 2 (10 k) s 15 0.
6. The open loop transfer function of control system with unity feed back is
given by:
k(s 1)
G(s)
s(1 Ts)(1 2s)
Find the closed loop transfer function then apply the Routh-Herwitz stability
to find the conditions of system stability.
7. Applying the Routh-Hurwitz criterion to calculate the values of A for which
the control system shown in the following figure is stable.
132
M Galal RABIE, Automatic Control for Mechanical Engineers
8. Plot the Bode diagrams for the system described by the given compensated
open loop transfer function then determine the phase margin and gain
margin.
5
GH(s)
s(1 0.6s)(1 0.1s)
4k(1 0.01s)
G(s)
s(1 0.1s)(1 0.4 s)
10. The following is the open loop transfer function of a unity feedback system.
k
G(s)
s(1 0.1s)(1 0.01s)
133
M Galal RABIE, Automatic Control for Mechanical Engineers
12. Given the frequency plot of the compensated open loop of a control system.
Find the phase and gain cross-over frequencies, the gain and phase
margins and the gain of the closed loop transfer function if H(s)=1.
13. The Bode diagram of the compensated open loop of a unity feedback
control system is given in the following figure. Deduce the forward path
transfer function for the system and sketch the phase plot of the frequency
response. Hence comment on the stability and the transient and steady
state performance of the system when operating in the closed-loop mode.
134
M Galal RABIE, Automatic Control for Mechanical Engineers
14. Sketch the Nyquist diagrams and check the stability of systems with the
following open loop transfer functions.
1 1
(a) G(s)H(s) (b) G(s)H(s)
s(s 1) s 5 (s 1)
15. Plot the open loop frequency response of the following system and find the
gain margin and phase margin.
10
G(s)H(s)
s(1 0.2s)(1 0.02s)
16. The following figure shows the asymptotic gain Bode plot of a third order
system having a transfer function of the form:
k
G(s)
s(1 T1s)(1 T2 s)
135
M Galal RABIE, Automatic Control for Mechanical Engineers
136
M Galal RABIE, Automatic Control for Mechanical Engineers
9 ROOT LOCUS
ANALYSIS
9.1 INTRODUCTION
The discussions of the transient response of a linear system, chapter 6, showed that
the system stability depends on the location of the roots of the characteristic equation
in the R-I plane. The root locus is a plot of the location of roots of the characteristic
equation of the closed loop system.
This process is illustrated by plotting the root locus of the system described by
Fig.9.1.
The roots location in the R-I plane is plotted in Fig.9.2 for different values of K in the
range 0 K . This plot is called root locus. The response of the closed loop
system to a unit step input is given in Fig.9.3. The study of Figs 9.2 and 9.3 shows
that:
The closed loop is always stable for all values of K.
137
M Galal RABIE, Automatic Control for Mechanical Engineers
The complete root loci are symmetrical with respect to the real axis of the R-I plane
and the number of branches of the root loci is equal to the order of the characteristic
equation. The construction and analysis of root loci gives good support to the system
analysis as well as to the synthesis and design of closed loop systems. There are
several available computer programs for roots evaluation and for the plot of the root
loci such as the MATLAB and CODAS. This section presents several examples of
138
M Galal RABIE, Automatic Control for Mechanical Engineers
closed loop control systems with different types of controllers. Their root loci are
plotted and analyzed. The block diagrams of these systems are of the form given in
Fig.9.4.
The closed loop transfer function and characteristic equation are given by the
following expressions:
KG c (s)Gp (s)
G(s) (9.1)
1 KG c (s)Gp (s)
It is important for a control engineer to build up knowledge that enables him to relate
root position to the transient behavior of the system. Moreover, it is necessary to
know the effect of altering the position of poles and zeros or introducing new poles or
zeros on a root locus plot and hence on the system dynamic behavior.
Generally, the addition of a pole (located in the left half of the R-I plane) to the
transfer function pushes the root loci towards the right half-plane and reduces the
system relative stability. This is illustrated by example 9.1 and 9.2. Example 9.1
shows the effect of introducing a pole at the origin and example 9.2 shows the effect
of introducing a pole at s = -1. The effect of adding a pole on the transient response
is shown in Figs. 9.5 to 9.10. The study of these figures shows that the addition of a
pole slows down the response and reduces the system stability or even makes it
unstable.
139
M Galal RABIE, Automatic Control for Mechanical Engineers
1
Example 9.1 G c (Simple integrator, introducing a pole at s=0)
s
140
M Galal RABIE, Automatic Control for Mechanical Engineers
1
Example 9.2 G(s) (Simple lag; introducing a pole at s= -1)
s 1
141
M Galal RABIE, Automatic Control for Mechanical Engineers
142
M Galal RABIE, Automatic Control for Mechanical Engineers
s 2 2
Example 9.4 G c 1 (Proportional plus integral controller, PI)
s s
143
M Galal RABIE, Automatic Control for Mechanical Engineers
The effect of adding a zero to the open loop function is illustrated by example 9.4,
where a zero (at s 2 ) is added. Generally, the addition of a left-half plane zeros to
the open loop transfer function has the effect of moving and bending the root loci
toward the left-half of R-I plane, Figs. 9.11 thru 9.13, which tends to make the system
more stable and of faster response.
9.3. EXERCISE
1. Plot the root locus of the following system and sketch the response of the closed
loop to unit step input for K=1 & K=100.
2. Find the transfer function of the systems whose closed loop root loci are plotted in
the following figures. Sketch the closed loop response for three different values of
gain K.
144
M Galal RABIE, Automatic Control for Mechanical Engineers
3. Plot the root locus for the given system and find the location of roots and value of k
for a damping ratio ζ = 0.5
145
M Galal RABIE, Automatic Control for Mechanical Engineers
146
M Galal RABIE, Automatic Control for Mechanical Engineers
10 COMPENSATION OF
CONTROL SYSTEMS
10.1 INTRODUCTION
Example 10.1 Find the open loop gain K needed to produce a steady state position
error < 5%. Find the value required to give a gain margin of 6 dB and a phase margin
of 27o.
Fig.10.1
When applying a step input of magnitude k2, the steady state error ess is given as
follows.
k2 1
e ss or e ss
1 k p 1 kp
147
M Galal RABIE, Automatic Control for Mechanical Engineers
K
k p Lim K
s 0 ( s 1) 3
1
for ess 0.05, 0.05 and K 19
1 kp
The value of K needed to satisfy the relative stability requirements is K=4; for a
gain margin = 6 dB and a phase margin = 27o. Naturally, decreasing the value of K
below this value improves the relative stability. Then, in order to satisfy the stability
requirements, K should be smaller than or equal 4 ( K 4 ).
The requirements of the steady state error are satisfied if K>19. But the requirements
of the relative stability are satisfied only if K<4. Then the two requirements are in
contradiction. Therefore, the addition of compensator is required to reshape the open
loop frequency response so that the low frequency gain is high enough, in addition,
the plot avoids the critical point (-1,0) insuring the required gain and phase margins.
The Nyquist plot can be reshaped to satisfy these requirements by one of the
following methods:
148
M Galal RABIE, Automatic Control for Mechanical Engineers
1. Start with gain K=19 (to satisfy the precision requirement) and introduce a phase
lead at high frequencies in order to attain the required phase and gain margins
(phase lead compensation)
2. Start with gain K=4 (Satisfying the stability requirement) and introduce a phase lag
at low frequencies to meet the steady state error requirements (phase lag
compensation).
3. Start with a gain between the two limits; 4<K<19, then introduce phase lag at low
frequencies and phase lead at high frequencies (lag-lead compensator).
Series Compensation
The compensated open loop transfer function is given by: G c (s) G p (s)
G1G 2
The compensated open loop transfer function is given by:
1 G 2Gc
Formally, the parallel compensator can be always selected to obtain the same results
as in series compensation.
149
M Galal RABIE, Automatic Control for Mechanical Engineers
The compensating unit is included in a supplementary feed forward loop; along which
the disturbance signal is introduced in the system.
The transfer function of the phase lead compensator is of the following form:
1 1 aTs
Gc ; a >1 (10.1)
a 1 Ts
The gain can be kept unity by connecting a proportional element (gain = a), in series
with the compensator. The compensator transfer function becomes:
1 aTs
Gc (10.2)
1 Ts
The frequency response of a typical phase lead element is given in Fig.10.6. This
figure shows that the phase lead element presents a phase lead within certain range
of frequency associated with permanent gain increase at high frequency range. The
phase lead can be calculated as follows.
T(a 1)
tan 1 (10.3)
1 aT 2 2
1 1 1
logm log log (10.4)
2 aT T
150
M Galal RABIE, Automatic Control for Mechanical Engineers
Or
1
m (10.5)
aT
a 1 a 1
max (m ) tan 1 sin 1 (10.6)
2 a a 1
Fig.10.6 Bode plot of a phase lead element, T=0.02 sec & a=25
For a phase lead element having T =0.02 sec and a =25, the frequency response
parameters are calculated as follows.
1 1
Corner frequencies 1 2 (1/ sec) and 2 50 (1/ sec)
aT T
25 1 1 1
m 10 (1/ sec), max sin 1 67.38 o and 42.7
o
25 1 T aT
Example 10.2 Plot the Nyquist and Bode plots of the open loop of the following
system with and without compensation and discuss the effect of introducing a phase
lead compensator having the following transfer function:
0 .5 s 1
Gc
0.02s 1
151
M Galal RABIE, Automatic Control for Mechanical Engineers
Fig.10.7
(1) K=4, Stable with GM=6 dB and PM=27o (2) K=19, Unstable without compensator
(3) K=19, Stable with phase lead compensator, GM=15 dB and PM=22o
Fig.10.8 Nyquist plot of the non-compensated open loop system and of the system
with series connected phase lead compensator.
Fig.10.9 Bode plot of the non-compensated open loop system and system with
series connected phase lead compensator.
152
M Galal RABIE, Automatic Control for Mechanical Engineers
The open loop frequency response plots of the system with K=19, with and without
compensation are given in Figs. 10.8 and 10.9. The study of these figures shows the
following:
The phase lag element has a transfer function of the following form:
1 Ts
Gc ; a >1 (10.7.)
1 aTs
The Bode plot of a typical phase lag element is shown in Fig.10.11. This phase lag
element presents a phase lag associated with gain reduction (negative gain in dB). It
permits the gain of the original system to be increased without affecting the stability.
The time constant T can be chosen such that the phase margin of the non-
compensated system is not much changed.
The phase lag compensator is used when the phase margin is acceptable but the
gain is too low to satisfy the steady state error requirements.
153
M Galal RABIE, Automatic Control for Mechanical Engineers
Example 10.3 Plot the Bode diagram of the open loop of the following system with
and without compensation and discuss the effect of introducing a phase lag
compensator having the following transfer function:
4s 1
Gc
150 s 1
Fig.10.12
The open loop frequency plots of the system, for K= 100, with and without
compensation are given in Fig.10.13. The study of this figure shows the following:
154
M Galal RABIE, Automatic Control for Mechanical Engineers
Fig.10.13 Bode plot of the non-compensated open loop system and system with
series connected phase lag compensator
1 bT2 s 1 aT1s
Gc , where a >1, b<1 and T1<T2 (10.8)
1 T2 s 1 T1s
Fig.10.14 Bode plot of a lag lead element, T1 =0.01 sec, T2 = 30.5 sec, a=14 and
b=0.0667
155
M Galal RABIE, Automatic Control for Mechanical Engineers
The Bode plot of the frequency response of a typical lag-lead element is shown in
Fig.10.14. The phase lag-lead compensator combines the advantages of the phase
lead and phase lag compensators.
Example 10.4 Plot the Bode plot of the open loop of the following system with and
without compensation and discuss the effect of introducing a lag-lead compensator
having the following transfer function:
1 2s 1 0 .4 s
Gc
1 30.6s 1 0.01s
Fig.10.15
Fig.10.16 Bode plot of the non-compensated system and system with series
connected lag lead compensator.
156
M Galal RABIE, Automatic Control for Mechanical Engineers
The open loop Bode plots of the system, for K= 100, with and without compensation
are given in Fig.10.16. The study of this figure shows the following:
The proportional integral derivative controller (PID) is the most common form of
feedback. It was an essential element of early governors. Today, more than 95% of
the control loops are of PID or PI type. The PID controllers are found in all areas
where control is used. They have survived many changes in technology, from
mechanics and pneumatics to microprocessors via electronic tubes, transistors and
integrated circuits. The microprocessor has had a dramatic influence on the PID
controller. Practically all the PID controllers made today are based on
microprocessors. This has given opportunities to provide additional features like
automatic tuning, gain scheduling and continuous or online adaptation.
1 t de( t )
u( t ) K e( t )
Ti 0
e( )d Td
dt
(10.9)
The error signal e(t) is the difference between the instantaneous values of the input
signal, x(t) and the feedback signal f(t); as illustrated by Fig.10.17.
e( t ) x( t ) f ( t ) (10.10)
de( t )
Derivative-term; proportional to the derivative of the error, D( t ) KTd .
dt
157
M Galal RABIE, Automatic Control for Mechanical Engineers
The controller parameters are the proportional gain K, the integral time Ti and the
derivative time Td. The most well-known methods for estimating and tuning the PID
parameters are those developed by Ziegler and Nichols. They have had a major
influence on the practice of the PID control for more than half a century. The
estimation methods are based on characterization of process dynamics by a few
parameters and simple equations for the controller parameters. The PID controllers
can be designed, according to Ziegler-Nichols rule as explained by the following
example. The process of design of a PID controller is illustrated by replacing the
proportional controller, Fig.10.18, by the PID controller, Fig.10.17.
The Implementation of the PID controller is carried out according to the following four
steps:
Connect the system as shown in Fig.10.18 then apply a step input (with gain
K=1, for example).
Calculate the step response then change the proportional gain K until
continuous oscillations are observed. The resulting step responses are shown
in Fig.10.19. For proportional gain K = 1, the calculated step response shows
that the system response has considerable steady state error ( e SS 50% ).
158
M Galal RABIE, Automatic Control for Mechanical Engineers
Fig.10.19 Step response of the closed loop system for open loop gain K=1 and for
the limiting gain KL=8
P-Controller
D(s) K 0.5 K L (10.11)
PI-Controller
1
D(s) K 1 (10.12)
Ti s
PID-Controller
1
D(s) K 1 Td s (10.13)
Ti s
The first estimate of the PID controller parameters is calculated by applying Ziegler
and Nichols rule, table 10.1. The calculation results are given in table 10.2.
159
M Galal RABIE, Automatic Control for Mechanical Engineers
Table 10.1 Summery of formulae used to calculate a first estimate of the P, PI and
PID controllers, according to Ziegler-Nichols rules.
Table 10.2 First estimate and tuned parameters of the P, PI and PID controllers
according Ziegler-Nichols rules.
Connect the controllers as shown in Fig.10.20 and calculate the step response. The
resulting response is shown in Fig.10.21. This figure shows that the P-controller
response has great steady state error. The PI and PID controllers stabilized the
system, with no steady state error. The controller setting according to Ziegler-
Nichols rule improved the closed loop system stability and precision. But a final
tuning of the controller parameters must be done iteratively until satisfactory
response is obtained.
160
M Galal RABIE, Automatic Control for Mechanical Engineers
Fig.10.21 Step response of the closed loop system, with P, PI and PID controllers.
The tuning of the gain of the proportional controller does not improve the system
behavior, due to the contradiction between the stability and precision requirements.
The tuning of the PI and PID controllers improved radically the system stability.
Figure 10.22 shows that the responses of the system with PI and PID controllers
converge rapidly to the required steady state value, without steady state error.
Fig.10.22 Step response of the closed loop system, equipped with P, PI and PID
controllers, with tuned parameters.
161
M Galal RABIE, Automatic Control for Mechanical Engineers
10.6 EXERCISE
1. Consider the system described by the following open loop transfer function:
K
G(s)H(s)
s(s 1)(1 0.125s)
2. Consider the system described by the following open loop transfer function:
K
G(s)H(s)
s(0.5s 1) 2
(a) Plot the Bode diagrams for K=1 and K=10 and discuss the system stability.
(b) A series phase lag compensator having the following transfer function is
introduced:
1 10 s
G c ( s)
1 100 s
Plot the Bode plot of the compensated system and discuss the results for
K=1 and K=10.
3. Consider the system described by the following open loop transfer function:
K
G(s)H(s)
s(0.5s 1) 2
4. Consider the system described by the following open loop transfer function:
K
G(s)H(s)
s(0.2 s 1)(0.3 s 1)(0.5 s 1)
162
M Galal RABIE, Automatic Control for Mechanical Engineers
REFERENCES
2009 M Galal Rabie, Fluid Power Engineering, McGraw-Hill, NY, May 18, 2009.
1998 Richard C. Dorf and Robert H. Bishop, Modern Control Systems, Addison-
Wesley, CA, 1998.
1998 Ogata K., System Dynamics, Third Edition, Prentice Hall International Inc,
NJ, 1998.
1997 Golten J. and Verwer A., Control System Analysis and Design, McGraw
Hill, 1997.
1997 Ogata K., Modern Control Engineering, Third Edition, Printce Hall Inc., NJ,
USA, 1997.
1991 Kue B.C., Automatic Control Systems, Printce Hall of India, New Delhi,
1991.
1985 Ogata K., Modern Control Engineering, Second Edition, Printce Hall of
India, New Delhi, 1985.
1979 Emanuel P. and Leff E., Introduction to Feedback Control Systems,
McGraw Hill, 1979.
1978 Scwarzenbach J. and Gill K.F., System Modeling and Control, JW, NY,
1978.
1974 Auslander D.M., Takahashi Y. and Rabins M.J., Introducing Systems and
Control, McGraw Hill, 1974.
1961 Francis H. Raven, Automatic Control Engineering, Mc Graw Hill, 1961.
163
M Galal RABIE, Automatic Control for Mechanical Engineers
164
M Galal RABIE, Automatic Control for Mechanical Engineers
INDEX
165
M Galal RABIE, Automatic Control for Mechanical Engineers
Nyquist plot97, 99, 120, 123, 124, 125, Routh-hurwitz117, 118, 119, 120, 123,
126, 148, 152 132, 133
Nyquist stability criterion, 120 Schematic, 8
Open loop, 3 Second order element, 37, 69
Open loop control, 4 Self-loop, 54
Order of the system, 30 Series compensation, 149
Output node, 55 Series connection, 54
Over-damped, 70, 71 Settling time, 82
Parallel connection, 54 Signal flow graph, 10, 53, 58, 59, 61
Parallel, feedback, compensation, 149 Simple lead element, 104
Partial fraction, 20 Sink, 55
Path gain, 55 Source, 12, 55
Performance index, 83 Speed control, 7
Period, 73, 75, 83, 84, 86, 88, 159 Stability, 113, 117
Phase cross-over frequency, 126 Stability criterion, 117, 118, 120, 121,
Phase cross-over point, 126 123, 125, 132
Phase lag compensator, 153 State space, 10
Phase lead compensator, 38, 150, 151, Static system, 2
152, 156 Steady state error, 113, 114, 115, 116,
Phase margin, 126, 127, 128, 129, 130, 117, 121, 123, 124, 127, 128, 131,
133, 135, 147, 148, 153, 154, 157 132, 133, 138, 147, 148, 149, 153,
Phase shift, 91, 92, 94, 121, 150 156, 158, 160, 161
PID, 157, 158, 159, 160, 161, 162 Steady state error, 83
Polar plot, 95, 98 Steady state error with ramp input, 116
Polar plot for first order element, 95 Steady state error with step input, 114
Polar plot of higher order elements, 98 Step function, 63
Polar plot of integrating member, 97 Step response of second order element,
Polar plot of second order element, 96 69, 75
Position error coefficient, 114, 148 Step response testing, 83
Positive feedback, 45 Summation point, 44, 47, 53
Properties of Laplace transform, 18 Superposition, 2, 45, 51
Proportional element, 31 System analysis, 12
Proportional integral derivative controller, System control, 3
157 System definition, 1
Quadratic lead element, 106 System inputs, 2
Ramp function, 63 System outputs, 2
Regulation,3 System representation, 8
Relative stability, 125, 139, 147, 148 Third and higher order systems
Resistance-capacitance network, 36 Time domain, 17, 18, 82, 125
Response of second order element to Transfer function, 9, 29
input impulse, 77 Transfer matrix, 30, 81
Response of second order element to Transient response characteristics, 81
ramp input, 76 Under-damped, 70, 71, 73, 83
Rise time, 82 Velocity error coefficient, 116
Root locus, 137, 139, 144, 145 Ziegler and Nichols, 159, 159
Routh-Hurwitz stability criterion, 117 Ziegler-Nichols rule, 158, 160
Routh array, 118, 119
166