LAPLACE TRANSFROM

AND TRANSIENT
ANALYSIS
UNIT THREE
AUTOMATIC CONTROL I
DR (MRS) E. A. ADJEI
LAPLACE TRANSFROM TECHNQUES
• The Laplace transform of a function f(t) is denoted by . The inverse
Laplace transform of f(s) is denoted by L-1[F(s)]. Of f(t).
• Laplace transform method is used extensively to facilitate and
systematize the solution of ordinary constant – coefficient equations.
It is used to analyze Linear – Time – Invariant systems. LTI systems are
equations in which none of the terms depends explicitly on the
independent variable, time.
Definition of Laplace Transform
• Laplace transform function is given as

L f (t )   f (t )e  st
dt  F ( s ) where L f (t ) 
0
• Evaluation of the integral results in a function F(s) that has ‘s’ as the
parameter. This parameter, ‘s’ is a complex quantity of the form 0 +jw.
Since the limits of integration are zero and infinity, it is immaterial
what value f(t) has for negative or zero time. There are limitations on
the function f(t) that are Laplace-transformable. Basically, the
requirement is that at the Laplace integral converge, which means
that this integral has a definite functional value.
Laplace Transform
• The function f(t) must be piecewise continuous in a finite time
interval
0  t1  t  t 2
• A function is piecewise continuous in a finite interval if that interval
can be divided into finite number of subintervals, over each of which
the function in continuous and at the ends of each of which f(t)
possesses finite right and left hand limits.
Laplace Transform

• A function f(t) is of exponential order if there exists a constant ‘a’ such

that the product  at
e f (t )
is bounded for all values of t greater than some finite value T.
This imposes the restriction that σ, the real part of s, must be greater
than a lower bound σa for which the product  a t
e f (t )
is of exponential order.
• A linear differential equation whose constant co-efficient s and finite
number of terms is Laplace transformable if the driving function is
Laplace-transformable.
Laplace Transform Theorems
• Several theorems are useful in applying the Laplace transform are
presented below:
• Theorem I: Linearity: If ‘a’ is a constant or is independent of ‘s’ and ‘t’,
Laf (t )  a L f (t )  aF ( s )
if f(t) is transformable, then

• Theorem II: Superposition; If f1(t) and f2(t)are both Laplace-

transformable, then the principle of superposition applies:

L f 1 (t )  f 2 (t )  L f 1 (t )  L f 2 (t )  F1 ( s)  F2 ( s)
 Laplace Transform Theorems
• Theorem III: Real Differentiation; If the Laplace transform of f(t) is F(s)
and if the first derivative of f(t) with respect to time, Df(t), is
transformable, then
LDf (t )  sF ( s )  f (0)
• The term f(0) is the value of the function f(t) evaluated at the origin t = 0.
• The transform of the second derivative D2f(t) is:

 2
 2
L D f (t )  s F ( s )  sf (0)  Df (0)
• Where Df(0) is the value of the limit of the derivative of f(t) at the origin
t = 0.
 Laplace Transform Theorems
• The transform of the nth derivative Dnf(t) is

 n
 n
L D f (t )  s F ( s)  s n 1 n2
f (0)  s Df (0)......sD n2
f ( 0)  D n 1
f (0)
• Note that the transform includes the initial conditions, whereas in the
classical method of solution, the initial conditions are introduced
separately to evaluate the co-efficients of the solution of the
differential equation.
 Laplace Transform Theorems
• Theorem IV: Real-Integration
• If f(t) is of exponential order, then the Laplace transform of
exists and is given by
 f (t )dt

 
L  f (t )dt 
F ( s ) f 1 (0)
s

s
where F ( s )  L f (t ) and f 1 (0)   f (t )dt , evaluated at t  0
 Example
• Find
.. the. solution x(t) of the differential. equation
x  3 x  2 x  0, x(0)  a, x(0)  b ......eqn 1
where ‘a’ and ‘b’ are constants.
• Solution:
• By writing the Laplace transform of X(t) or , Lx(t ) we
X (obtain
s)
.
L  x   sX ( s )  x(0) .......eqn 2
 
 ..  2
.
L  x   s X ( s )  sx (0)  x(0) .......eqn 3
 
• Substituting eqn (3) and eqn (2) into eqn (1), we have
 2 .

 s X ( s )  sx (0)  x(0)   3sX ( s )  a   2 x( s )  0 ......eqn 5
Example
Laplace Transform Theorem
• Theorem V: Complex differential:
• If f(t) is a Laplace transformable, then except at poles of F(s).
d
Ltf (t )  F ( s )
ds .

• Where F (s ) = L f (t ). This is known as the complex differential

theorem. Also
2
2

L t f (t ) 
d
2
F ( s)
ds
EULER THEROEM (POWER SERIES)
EULER THEROEM (POWER SERIES)
EULER THEROEM (POWER SERIES)
Example 1
Example 2
Example 3
Example 4
INVERSE LAPLACE TRANSFORM
PARTIAL FRACTION EXPANSION METHOD FOR FINDING INVERSE LAPLACE
TRANSFORM
• For problems in control systems analysis, F(s), the Laplace transform
of f(t), frequently occurs in the form F(s) = where A(s) and B(s) are
polynomials in ‘s’. In the expansion of F(s) = into a partial fraction
form, it is important the highest power of ‘s’ in A(s) be greater than
the highest power of ‘s’ in B(s). If such is not the case, the numerator
B(s) must be divided by the denominator A(s) in order to produce a
polynomial in ‘s’ whose numerator is lower than the denominator.
Partial Fraction Expansion Method for
Finding Inverse Laplace Transform
3 2
• Example s  5s  9 s  7
G s  
( s  1)( s  2)

• Here, since the degree of the numerator polynomial is higher than

that of the denominator polynomial, we must divide the numerator
by the denominator. We have
2 1
G s   
S 1 S  2
 Partial Fraction Expansion When f(s)
Involves Distinct Poles Only
• Consider F(s) written in the factored form
B( s ) K ( s  z1 )( s  z 2 )...( s  z m )
F s    for m  n
A( s ) ( s  P1 )( s  P2 )....( s  Pn )
• Where P1,P21,……Pn and z1, z2,…z3 are either real or complex quantities,
but for each complex Pi or zi there will occur the complex conjugate
of Pi or zi, respectively. If F(s) involves distinct poles only, then it can
be expanded into a sum of simple partial fractions as follows:
B( s) a1 a2 an
• F s     ....
A( s ) s  p1 s  p2 s  pn
Partial Fraction Expansion Method for
Finding Inverse Laplace Transform
s3
• E.g. Find the Laplace transform of
F s  
( s  1)( s  2)
• The partial fraction expansion of F(s) is s3 a1 a2
F s    
 s3   s3  ( s  1)( s  2) s  1 s  2
a1  ( s  1)    2
 ( s  1)( s  2)  s  1  ( s  2)  s  1
 s3   s3 
a 2   ( s  2)     1
 ( s  1)( s  2)  s  2  ( s  1)  s  2

• Thus  2  1   1 
f t   L F ( s )  L 
1 1
 L  2 e t
 e 2t
for t  0
 s  1  s  2 
Partial Fraction Expansion When f(s)
Involves Multiple Poles
• Example: s 2  2s  3
F ( s)  3
......1
( s  1)
The partial fraction expansion of this F(s) involves three terms.
B( s) b1 b2 b3
F (s)    Whereb 2
 1,b2,b3 3 are
.....2determined as
A( s ) ( s  1) ( s  1) ( s  1)
• Multiply both sides of the equation by (s+1)3, we have
 3 B( s)  2
 ( s  1)   b ( s  1)  b2 ( s  1)  b3 ...3
A( s ) 
1 letting s = -1

 3 B( s) 
( s  1) A( s )   b3 ....4
  s  1
Partial Fraction Expansion When f(s)
Involves Multiple Poles
• Differentiating both sides of the equation with respect to ‘s’ yields
d  3 B( s) 
 ( s  1)   2b1 ( s  1)  b2 ...5 and letting s  1
ds  A( s ) 
d  3 B( s) 
 ( s  1)   b2
ds  A( s )  s  1
• By differentiating both sides of equation 5 with respect to ‘s’, the
result is
2
d  3 B( s) 
2 ( s  1) A( s )   2b1 ...6
ds  
• From the proceeding analysis it can be seen that the values b1, b2,b3
are found systematically as
 3 B(s)  2
b3  ( s  1)   ( s  2s  3) s  1  2
 A( s )  s  1
d  3 B(s) 
b2  ( s  1)
ds   
d
A( s )  s  1 ds
( s 2

 2 s  3) s  1  0 
1 d2   2
b2 
2! ds 2  ( s  1) 3 B ( s )
 
1 d
A( s )  s  1 2! ds

2
( s 2

 2s  3) s  1  1

• We obtain
 1  1  0  1  1 
f t   L F ( s )  L 
1 1
  L  2 
 L  3
 e t
 t 2 t
e
 s  1  ( s  1)   ( s  1) 
 (1  t 2 )e t , for t  0
SOLVED TUTORIAL PROBLEMS
• Using Laplace transform, solve the following equations with zero
initial conditions.

2
d y dy
2
 3  2 y  5
dx dx
Solution d2y 
L 2 
 s 2
Y s   S y ( o )  y 1
(o )
d x 
 dy 
L 3   3 sY ( s )  y o 
 dx 
L 2 y  2 Y ( s )
5
L 5 
s
s2 Y(s) – s y(0) – y1(0) + Y(s) – 3 y(0) + 2 Y(s) = 5/ s
 2
 Y ( s ) s  3s  2   5
s
5 5
s
2

 Y ( s )   5  3s  2  
s s 2  3s  2  
 Solution
5
 Y `( s )   Partial fraction Decomposition
s s  2 s  1
5 A B C
   
s s  2 s  1 s s  2 s 1
5 A ( s  2) ( s  1)  B( s) (s  1)  c( s) (s  2)
 
s ( s  2) s  1 s (s  2) (s  1)

 5  A ( s  2) ( s  1)  B( s) ( s  1)  c( s) ( s  2)
 Solution
• Using the Heaviside cover up method;
• For Residue B, s = -2

 5  B(2)  2  1
 5  A (0  2) (0  1)
5  ( 2) ( A) (1)
 5 (2) ( B) (1)  5  B 5  2A
2
• For Residue C, s = 1  A  5
2
 5  c  1  1  2
5   c (1)  c   5
5 5
5 2  2 5
    Y ( s)
s (5  2) (5  1) s s  2 s  1
 Solution
• Taking the Laplace Inverse; L  1  Y ( s )  y ( x)
 52  5
L 1  
s 2

 52  5
L 1   e 2 x
2
s  2
L  1  5 
   6 e x
 s  1

• Solution for the differential equation is:

y ( x)  5  5 e 2 x
 5e x
2 2
Example 2
3 2
d y d y dy
3
 10 2   5 y  8 u (t )
dt dt dt
• As initial conditions are zero, taking Laplace transform on both sides
and assuming u(t) = x
• s3 Y(s) + 10s2 Y(s) + 20s Y(s) + 5 Y(s) = 8X(s)
• On rearranging, 8 X ( s)
Y (s)  3 2
( s  10s  20 s  5)
2
d y (t ) dy (t )
 10  2 y (t )  20 x (t  4)
dt dt
 Example 2
 d y (t ) 
2
2
L 2   s Y ( s )  s y ( 0)  y ( 0)
 dt 
 dy (t ) 
L 10   10 S Y ( s )  y (0)
 dt 

L 2 y (t )  2 Y ( s)
20 x 80 x
L 20 xt  80 x  2 
s s
20 x 80 x
 s Y ( s )  s y (o)  y (o)  10sY ( s )  y (o)  Y ( s )  2 
2 1

s s
 Example 2
1
 Given y (o)  0
y (o )  0
2 20 x 80 x
 s Y ( s)  10 s Y ( s )  2 Y ( s )  s y (o)  y (o)  10 y (o)  2 
s s

 2 20 x
 Y ( s ) 5  10s  2  2 
s

80 x
s
20 x  (80 x) s
 Y ( s)  s2

s 2  10s  2 
20 x  (80 x) s
 Y (s)  2 2

s 5  10 s  2 
Example 3
• Calculate the current (t) in the integro differential equation.

di (t ) t  4t
 6  o i ( x) dx  5i (t )  10e
dt
• Assume initial conditions to be zero.
Example 3 - Solution
• Taking Laplace transform of both sides of equation,

  6  
s I ( s )  I (o )  5 I ( s )  I ( s ) 
 I o 

10
s s s4
• Substituting initial conditions to be zero i.e. I(o+) = o
 6 10
I ( s) s  5   
 s s4

10s 10 s
I(s) = 
 
s  4 s  5s  6 s  2 s  3 s  4
2
Solution
• The equation above contain single order poles. Using partial fraction
method;  I ( s )  A1  A2  A3
s2 s  3 s 4
10 s
 A1  s  2  I ( s ) s   2    10
s  3 s  4 s  2

10 s
A2  s  3 I ( s) s  3   30
s  2 s  4 s  3
10 s
A3  s  4 I ( s ) s   4  s   4   20
s  2 s  3
Substituting these partial factor coefficients and taking inverse Laplace
transform;
; (t )  L 1
I10  10 e 2 t
 2e 4 t

Example 3
• Find the inverse Laplace transform of

Answer
Exercises
• Given the function f(t), find F(s) • In the following expressions, find
• f(t)= 17 f(t) for the given F(s) functions:
•

•)
 TRANSIENT ANALYSIS

• Transient analysis means analyzing a system in unsteady state. If the

variables involved in defining the state of a system does not vary with
respect to time, then the system is said to be in steady state. If not, it
is in unsteady state
• A hot coffee put in air. (transient heat transfer)
• Just after switching on a capacitive circuit. (transient voltage-current
characteristic in capacitor)
• An iron piece exposed to carbonaceous environment during surface
hardening process. (transient mass transfer)
TIME RESPONSE OF CONTROL
SYSTEMS
• We can analyze the response of control systems in both the time
domain and the frequency domain. If the output of a control system
for an input varies with respect with time, then it is called the time
response of the control system. The time response consists of two
parts:
• Transient response and
• Steady state response
Control Test Inputs
Examples of standardized control inputs are:
Step input, ramp function, impulse function, parabolic function, etc.
Step Function
• For example, by switching a voltage in an electrical circuit or a sudden
force which is retained in a mechanical network. If it is assumed that
the step is of unit magnitude and is applied at t=0, the unit step input

is a primary test signal can be described by U (t )  10:: tt00
where the input magnitude is shown to be unity after time t=0. The
Laplace transform of the ideal unit step is given by U(s) =1/s
U(t)

t
Unit Set Input Signal
Impulse function
• This is a simple derivation of the step input. It therefore has a Laplace
transform of U(s) = 1 when applied as a control signal. In its strictest
form, an impulse function last for a time interval δt as δt tends to 0,
and is of infinite magnitude.
• As the practical application of such a signal poses obvious problems,
it is usual to approximate the impulse with a signal of unit magnitude
which lasts for a very short time period δt.
Ramp Input
• If the integral is taken of the step input, this results in a function
which increases linearly with respect to time from a zero value at time
t = 0. The ramp function which can also be regarded as velocity
function is shown in terms of a unit magnitude ramp in fig2.
• The unit ramp input can be written as U (t )  U(s) 
t : t  0=1/s2
0: t  0

U(t) Unit Ramp Input Signal

t
Parabolic input
• By integrating the ramp function, hence obtaining an acceleration
function, a square law signal known as the parabolic function is
achieved. This is shown in terms of a unit parabolic signal in figure.
 t 2: t 0
• The unit parabolic input can be written as 2
U (t )   0: t 0


• Where the division by 2 is included in order
to think of the function as a step acceleration
term with respect to the original step input.
This can also be considered as U(s) =1/s3
• By means of the Laplace transform.
 Parabolic input
• It is apparent that by combining the step, ramp and parabolic inputs to
t2
form one single input signal, for all t>0, this can be written as U (t )  1  t 
• It is quite possible, however, to consider further integrated terms in order 2
2 p
t t
to arrive at an overall input, U (t )  1  t 
2 although
  
p! a more general
expression is obtained by considering each separate term to have a
magnitude equal to a constant di, resulting in a power series or
polynomial function t 2for t>0; U(t)
t p = 0 for t<0.
U (t )  d 0  d1t  d 2    d p
2 p!
• This polynomial function can be written more concisely as p
ti
U (t )   d i ; t  0
where di are scalar constant values. The final standard test i 0 i!
signal introduced here is the sinusoidal function which can be written in
unit magnitude form as
•
U (t )   cos wt ;t  0
0 ; t 0
TIME DOMAIN PERFORMANCE
CRITERIA
• TIME RESPONSE
• The initial part of the response is most often described by means of
two terms, firstly the delay time, which is the time taken for the
output step response to reach 50% of it’s final value. Secondly, the
rise time which is the time taken for the output to rise from 10% to
90% of its final value.
• At a later stage in the time response, we have the settling time. This is
defined as the time taken for the step response to reach and stay
within a specified percentage of its final value. A commonly
encountered figure is 5%.
TIME RESPONSE
• Maximum overshoot: it is the maximum difference between the
transient and steady-state step response values that occurs after the
response has first passed its final value. It is frequently written as a
percentage of the steady-state output response value,
• Maximum overshoot = (maximum overshoot/ steady-state value) X
100%
• A single input-output linear dynamic system transfer function G(s)
relates input, U(s), to output, Y(s) as : Y(s) = G(s)U(s)
• To control output Y(s), adjust U(s) by means of a controller K(s), so
that
• Y(s) = K(s)G(s)U(s) and if K(s)G(s)=1 , the output Y(t) will follow
reference input r(t).
STEP RESPONSE
• A step response is a system’s response to a step function. The response function
is also called the unit step response. When measuring a system’s step response,
a signal is applied at the system input that jumps from zero to 1 at the time t0.
The input signal remains unchanged for the duration of the measurement.
• The step response of multi-input systems is the collection of step responses for
each input channel. Step response and transfer function are interchangeable via
Laplace transform
• Performance indices
• Peak time
• Settling time
• Rise time
• Steady state error
• Percentage overshoot
• Example: step, Ramp and Parabolic inputs
Step Response
• Applying step input to a first order system transfer function, we have
K
G( s) 
s  a0
where K is the system gain, then the system output is obtained from
Y(s) = G(s)U(s) to be K 1
Y (s) 
s  a0 s
• in which 1/s is the unit step Laplace transform. The output then is
K K
a0 a0
Y (s)   or int erms of its inverse transform
s s  a0
K K  a0 t
y (t )   e a0  0
a0 a0
SECOND ORDER SYSTEMS
• General second order system wn2
G ( s)  2
s  2wn s  wn2
• 1. Under-damped case (0<ξ<1): in this case C(s)/R(s) can be written
for general second order system
wn2 s  wn  wn  2  1 but wd  wn  2  1 and   wn
G ( s)  2
s  2wn s  wn2 where wd  damped natural frequency
• Second order system. wn  undamped natural frequency
• Damped systems
• Under damped
• Critically damped
• Over damped
Effects of Damping ratio
• if ξ<0, then the poles will be right-half-plane (have positive real parts),
and the system will be unstable.
• If ξ=0, so s=±wnj, the system will be marginally stable or neutrally
stable, e.g 0±5j
• If 0<ξ<0.707, the system response will be oscillatory and under
damped with very small overshoot.
• If 0.707<ξ<1, the system response will be non-oscillatory and under-
damped. The poles are in the left half of the s-plane
• If ξ = 1, the system response will be critically damped (no overshoot)
There are two real and positive poles.
• If ξ>1, the system response will be over-damped with two real positive
poles.
Example Y s  9
 2
U s  s  6 s  9
2
wn
• Second order system = 2 2
s  2wn s  wn
2
9 wn
 2  2 2
s  6s  9 s  2wn s  wn
2 1
 wn  9  wn   9  3rads
6 6
• Also2w : 6     1
2 wn 23
Hence the system is critically damped
 • For a critical damped system, the response is as follow
1im   wt   1   2 
y t  1  e 
sin wn 1   2 t  tan 1   
 1  1   2
    
    

-1  1  
2 
   we get :
1im  e wt 
Putting tan y (t )  1  sin (W 1   2 t   
     1  1  2 
 

 y (t ) 
1in 
 1
1 
e wnt
1 2
   

sin wn 1   2 t cos   cos ( w 1   2 t  Sin 


2
2 1  
Since, sin   1   , cos    and tan  


 y (t ) 
1im 
1 
 1 
e wnt
1   2
  
sin Wn 1   2 t   cos wn 1   2 t   
1  2




Now 1im
 1

sin (Wn 
1   2 t )   Wn 1   2

And
1im
 1
cos wn 
1   2t   1

Hence y (t ) 
1im 
1 
  1
e wnt
12
wn 2

1   t  1. 1   2






1im  e wnt 
 1  1   wnt  1
2

 1  1   2

1im 
1 
 1 
ewnt
1   2

wn 1   2 t  1. 1   2  






 y (t )  1  e  wnt
1  Wnt 
1. Rise time for a critically damped system, tr = t90% - t10%

Steady state response y   = 1 – e  wn   1  wn   
 y    1  y t10%   0.1 (1)  0.1

 y t10%   0.9 (1)  0.9

 y t10%   0.1  1  e t10 %
1  3t10% 
 0.9  e 3t1 0%
1  3t10% 
0.9 3t1 0%
  e taking natural log of both sides
1  3t10%
ii. Approximate methods
 0.9  can be used to solve
 In     3t10%
1  3t10%  this equation
y(t10%) = 0.1772705361s
Similarly,

 0.9  1  e 3t 90% 1  3t 90%   t 90%  1.29657339 s

 0.1  e 3t90% 1  3t 90% 
 Rise time tr  1.29657339 s  0.1772705361s
 0. 1  3t 90%
In    In e
1  3t90%   tr  1.119302854 s
tr  1.12 s
 0. 1 
In     3 t 90% Check with Matlab tr = 1.12s exact!
1  3 t 90% 
 (iii) Peak time, tp occurs at y(t)max  3e 3tp  3e 3t 3tp   3e 3tp  0
y(t)max occurs when

dy (t )  
 9e 3tp tp   0  
 e 3tp tp   0
 0
dt 3tp
3t 3  e  0  tp  0
 y (t )  1 e  3e tp
dy (t )
  (3e 3tp ) (1  3tp )  (3) (e 3tp )  0
dt  tp  0 or tp   never occurs 
• Setting time, ts2% occurs at y(ts2%) = 0.98

 y ts 2%   1  e 3ts2%
1  3ts 2%   0.98
 ts 2%  1.944640567 s
• Similarly ts5% occurs at y(ts5%) = 0.95
 y ts 5%   1  e 3ts5% (1  3ts 5% )  0.95
•  ts 5%  1.581288173s

• Check Matlab ts2% = 1.94s exact!

• Maximum overshoot ymax occurs at y(tp)
 ymax  1  e 3tp 1  Wntp 
• If tp = 0  y max  1  e 1  o 
o

•  11  0
•  % Overshoot  100%  y max  100%  0  0%
• If tp = ∞
 ymax  1  e  1  3 () 

1
 1   1     1  0  1  0
e

 % Overshoot  100%  0  0%
Stability Analysis
• Stability deals with asymptotic behaviour of a system for all inputs as
time approaches infinity. Absolute stability deals with qualitative
nature (yes/no). Relative stability deals with how close a system is to
being unstable, (how stable).
• Linear system is stable if all the poles of the closed loop transfer
function, y(s)/r(s), lie the left half of the s – plane. This is a necessary
and sufficient condition for the system to be stable.
• If the system has some roots with real parts equal to zero, but none
with positive real parts, the system is said to be marginally stable.
Example S2 + 1 = 0. the roots are ±j, since these roots have no roots
with positive real part but only imaginary values, the system is
marginally stable.
Routh Stability Criterion
• The Routh criterion is a method for determining continuous system
stability, for systems with an nth-order characteristic equation of the
form; a s n  a s n 1  a n2
s  a s  a  0 1
n n 1 n2 1 0
The criterion is applied using a Routh table defined as follows
Sn an an-2 an-4 -------
Sn-1 an-1 an-3 an-5 --------
  b1 b2 b3 ---------
S1
c1 c2 --------
S0
d1
Routh Criterion
• Where an , an-1, ……, a0 are the co-efficients of the characteristic
equation and a n 1 a n  2  a n a n 3 a n 1 a n  4  a n a n 5
b1  , b2 
a n 1 a n 1
b1 a n 3  a n 1b2 b1 a n 5  a n 1b3
c1  , c2 
b1 b1
• This method provides a way of determining the stability of a closed
loop system from the co-efficient of the closed loop characteristic
equation without computing its roots.
• The Routh Criterion: all roots of the characteristic equation have
negative real parts if and only if the elements of the first column of
the Routh table have the same sign. Otherwise, the number of roots
with positive real parts is equal to the number of changes of sign.
 •Example 2: S3 + 3S2 +3S + 1+K = 0.

S3 1 3 0 0
S2 3 1+K 0
S1 (8-K)/3 0
S0 1+K

For no sign changes in the first column, it is necessary that the conditions 8-K>0,
• 1 + K>0 be satisfied. Thus the characteristic equation has roots with
negative real parts if -1<K<8, the simultaneous solution of these two
inequalities.
• A row of zeroes for the S1 row of the Routh table indicates the at the
polynomial has a pair of roots which satisfy the auxiliary equation
formed as follows: AS2 +B= 0. Where A and B are the first and second
elements of the S2 row.
• To continue the table, the zeroes in the S1 row are replaced with the
co-efficients of the derivative of the auxiliary equation. The derivative
of the auxiliary equation is 2AS + 0 = 0. The co-efficients 2A and 0 are
then entered into the S1 row.
•Example 3: S4+ 6S3 + 11S2 +6S + 15 = 0.

S4 1 11 15 0 0
S3 6 6 0
S2 10 15 0
S1 -3 0
S0 15

This system is unstable

 •Consider the polynomial equation S3 + 6S2 +11S + 6 = 0. The Routh
array is
1 11 0
S3 6 6 0
S2 10 0
S1 6
S0

This system is stable

Try: S3 + 4S2 +8S +12 = 0, S4 + S3 - S -1 = 0, S4+ S3 + S2 +11S + 10
= 0.
Special Cases
• There are three (3) special cases that have to be considered.
• 1. There is a zero in the first column, but other elements in the row are non-
zero.
• 2. All elements in the row are zero.
• 3. All elements in the row are zero and there are repeated roots on the
imaginary axis.

• CASE 1: Zero only in the first column.

• The way to deal with this occurrence is to replace the zero with a very small
positive number, ε, and to complete the array.
 • For example: S5+ 2S4 + 2S3 + 4S2 +11S + 10 = 0. Then the array is

S5 1 2 11 0 0 0
S4 2 4 10 0 0
S3 ε 6 0 0
S2 (4ε-12)/ε (10ε)/ε 0
S1 (-72/ε-10ε)/(-12/ε)
S0 10

NB: the 4 is ignored since a very small value compared with the
12. There are two sign changes as ε tends to zero.
CASE 2: Zeroes in a row
• If the whole row is zeroes (even if the element has just one element),
then there are poles that are symmetric about the origin of the s –
plane. Example: (S+1) (S-1). To solve it, the auxiliary polynomial U(s),
which immediately precedes the zero entry in the Routh array, must
be used. U(s) always has degree that is even, and indicates the
number of symmetric poles pairs. For example: S3 + 2S2 +4S + K = 0.
S3 1 4 0
S2 2 K 0
S1 (8-K)/2 0
S0 K

For a stable system, 0<K<8. If K=8, then the whole of the S1 row
is zero, so there are symmetric poles about the origin.
• Auxiliary polynomial, U(s) = 2S2 + K = 0. Factorizing U(s) will give the
two symmetric poles,
K
s  2 j
2
• The situation with the entire row of zero elements can be remedied
by the following procedure:
• Form the auxiliary equation U(s)=0 by using the co-efficients of the
preceding row of zeroes.
• Take the derivative of the auxiliary equation with respect to s, that is
dU(s)/ds = 0
• Replace the row of zeroes with the co-efficients of dU(s)/ds = 0
• Continue the construction of the Routh array in the usual manner.
 Example: S5+ 4S4 + 8S3 + 8S2 +7S + 4 = 0.
The Routh array is
S5 1 8 7 0 0 0
S4 4 8 4 0 0
S3 6 6 0 0
S2 4 4 0
S1 8 0
S0 4

We get row of zeroes in the S1 row, set U(s) =4 S4 + 4 = 0 which gives us the root s±j = 0
Calculate dU(s)/ds = 8S
There are no sign changes in the first column, so the system has no right-half-plane poles
 CASE 3: Repeated roots on the imaginary axis
• The Routh – Hurwitz criterion indicates that a system has poles in the right
half of the s-plane. A system needs poles in the left half plane for stability. A
system with repeated poles on the imaginary axis results in an unstable
system. Example: S5+ S4 + 2S3 + 2S2 +S + 1 = 0. The Routh array is
S5 1 2 1 0 0 0
S4 1 2 1 0 0
S3 4 4 0 0
S2 1 1 0
S1 ε
S0 1
U(s) = S4 + 2S2 + 1 = 0 (S2 + 1) = 0 S = ±j dU(s)/ds = 4S3 + 4S
The system is unstable since there are repeated roots on the imaginary axis.
But we know from the auxiliary polynomial that we have 2 poles on the
imaginary axis, so the system is marginally stable.
Matlab Tutorials For System Stability
Analysis
• Introduction
• In control engineering, in order to analyze and design control systems,
a mathematical model of the actual plant is required. This model
approximates the behaviour of the plant to some extent. By
engineering requirement, a number of assumptions have to be made
about the plant in order to obtain the model.
• The nature and validity of the assumptions must be bear in mind
when applying control theory to real engineering systems. Analysis
and design taking into account the amount of difference between the
actual plant and model is a more advanced topic, studied within an
area known as robust control.
 Matlab Tutorials For System Stability Analysis
• One of the most common assumptions is that the plant is linear. This is valid
for most plants within a certain limited operating range. The second
assumption is that the plant dynamics are exactly known and are time
invariant.
• Again, this is valid to some extent, but the fact that the plant dynamics and
system disturbances are not exactly known, and their characteristics change
over time is what motivates the use of feedback.
• Indeed, if the plant and disturbances were exactly known, there would be
no need for feedback control; all control could be open-loop.
• However, for the sake of designing the feedback controller, this assumption
is made. Another common assumption is that the plant is second order.
Matlab Tutorials For System Stability
Analysis
• This assumption can be made for a great many simple control systems
(such as servo systems). This is because the higher order dynamics are
generally of high frequency, and are thus beyond the bandwidth of
the system.
• In the first exercise, the step response of second order liner systems
will be investigated. One of the ways in which actual (stable) plants
can be identified and modelled is by applying a step signal to the
actuator input and observing the response.
• Another is by means of applying sine waves over a range of
frequencies and observing the gain and phase change of the output.
Linear second-order dynamic system response
• A standard form of a second-order dynamic system, G(s) is

• Where wn is the natural frequency and ζ is the damping ratio.

• In this exercise, you will investigate the step response of the above
system in an open-loop configuration for a variety of natural
frequencies and damping ratios. The open loop configuration is
shown in Figure 1. The system input is R(s)and the output is C(s).
Step Response
• In this section with a constant natural frequency , the effect of different
damping ratios
• = 0.1, 0.3, 0.5, 0.707, 0.85, 1.0, 3.0, 6.0, on the step response is investigated.
The natural frequency you will use in this section is = 5.
• Type

• Now enter into the workspace the first value of by typing

• zt = 0.1
• To check which variables have been defined in the workspace, type
• who
Step Response
• Now define the system transfer function G(s) as the ratio of the
numerator and denominator polynomials, respectively by typing
• G=tf (wn^2, [1, 2*zt*wn, wn^2])
• Note that in MATLAB, ^ denotes “raise to the power of”. Also note
that in MATLAB, the coefficients of a polynomial in ‘s’ are defined in a
row vector of the coefficients of the polynomial in descending powers
of s. Record the transfer function.
• You can check the natural frequency, damping ratio and system
poles/eigenvalues by typing damp(G)
Step Response
• Alternatively, you can obtain just the system poles/eigenvalues by typing
• pole(G)
• Record the values of the poles.
• To get more information on these or any other MATLAB functions, just type
• help damp
• help pole
• To plot the step response of G(s), type step(G)
• Left click the mouse on the line of the plot locus to give the amplitude and time.
Hence measure the overshoot Mp and the peak time tp. Record the values.
• Now repeat the process for the next value of.
Step Response
• Record the values of the poles. Now plot the step response on the
same axes as the previous plot.
• To do this type
• hold on
• step(G)
• Measure the overshoot Mp and the peak time tp. Record the values.
Remember that tp cannot be measured if there is no overshoot.
• Repeat the process for all the values of zt.
• Question 1
• Produce a table of the results including the transfer functions, poles,
overshoots and peak times. Include the print out of the step responses.
Comment on the responses and the results.
•  Plot Mp against and tp against . This can be done in MATLAB; type
• help plot to see how.
• Question 2
• Include the plots and explain the curves.
• Question 3
• What is the significance of the damping ratio = 0.707?
• Now clear the plot window by typing
• clc
• Note
• When is small, the response is damped oscillatory, and the output
taking some time to settle. This is because the poles have a large
imaginary component compared to the real component.
• The step response simulation shows that as increases, the response
becomes less oscillatory, until = 0.707 where there is no more
resonance.
• When reaches the critical value of unity, there is no more overshoot,
and the response tends asymptotically towards 1.
• As increase further, the response increasingly resembles a first order
response, with increasing rise and settling times. This is because the
low frequency pole (closest to imaginary axis) becomes dominant.