Lecture 6

Modelling Second Order Systems

and Examples

Process Control
Prof. Kannan M. Moudgalya

IIT Bombay
Wednesday, 31 July 2013

1/39 Process Control Second Order Models and Response

Outline

1. Poles, zeros, response for different pole

locations
2. Solution to second order systems
3. Integrating/capacitive systems

2/39 Process Control Second Order Models and Response

1. Poles, zeros, response for different pole
locations

3/39 Process Control Second Order Models and Response

Transfer function terminologies

I Let the transfer function be G(s) = N(s)/D(s)

where N(s) and D(s) are polynomials in s
I The roots of N(s) = 0 are called zeros of G(s)
I The roots of D(s) = 0 are called poles of G(s)

4/39 Process Control Second Order Models and Response

Transfer Function
The ai in the following transfer function are known
as
(s − a1 ) · · · (s − am )
G(s) =
(s − b1 ) · · · (s − bn )

1. Poles
2. Zeros
3. Gain
4. Time constants
Answer: 2. Zeros
5/39 Process Control Second Order Models and Response
Transfer Function
The bj in the following transfer function are known
as
(s − a1 ) · · · (s − am )
G(s) =
(s − b1 ) · · · (s − bn )

1. Poles
2. Zeros
3. Gain
4. Time constants
Answer: 1. Poles
6/39 Process Control Second Order Models and Response
Poles and zeros of two transfer functions

I Let the transfer function be G(s) = N(s)/D(s)

where N(s) and D(s) are polynomials in s
I The poles of 1 + KG(s) are
1. same as the poles of G(s)
2. same as the zeros of G(s)
3. No relation between the poles/zeros of G(s) and
1 + KG(s)
Answer: 1. same as the poles of G(s)

7/39 Process Control Second Order Models and Response

Poles and zeros of two transfer functions
I Let the transfer function be G(s) = N(s)/D(s)
where N(s) and D(s) are polynomials in s
I Let the closed loop transfer function be T(s):
KG(s)
T(s) =
1 + KG(s)
I The poles of T(s) are
1. Poles of G(s)
2. Zeros of G(s)
3. Zeros of D(s) + KN(s)
4. Poles of D(s) + KN(s)
Answer: 3. Zeros of D(s) + KN(s)
8/39 Process Control Second Order Models and Response
Partial fraction of second order system
K M A B C
I Y(s) = = + +
s2 + 2ζωn s + ωn2 s s+a s+b s

Given that y(t) is the output of a real life plant,

and a is complex,
1. There is no relationship between a and b
2. b is real
3. a = −b
4. a and b are complex conjugates
Answer: 4, i.e. a and b are complex conjugates
If a and b are not complex conjugates,
y(t) will be imaginary, not realistic for a real plant!
9/39 Process Control Second Order Models and Response
Partial fraction of second order system
K M A B C
I Y(s) = = + +
s2 + 2ζωn s + ωn2 s s+a s+b s

Given that y(t) is the output of a real life plant,

and A is complex,
1. There is no relationship between A and B
2. B is real
3. A = −B
4. A and B are complex conjugates
Answer: 4, i.e. A and B are complex conjugates
If A and B are not complex conjugates,
y(t) will be imaginary, not realistic for a real plant!
10/39 Process Control Second Order Models and Response
Response for different pole locations
d
c

a b

c
d
Poles are indicated by crosses
11/39 Process Control Second Order Models and Response
Step response for pole at a

a
I For the pole at a
place indicated by a,
the response is of the
2
form e−α t
I The exponential part
will decay, reaching a
constant value
12/39 Process Control Second Order Models and Response
Scilab code to plot step response of a
negative pole, a.sce

s = %s
G a = 1 / ( s +3)
t = 1:0.1:5;
y a = csim ( ’ step ’ , t , G a ) ;
plot2d ( t , y a )

13/39 Process Control Second Order Models and Response

Response for pole at b

b
I For the pole at a
place indicated by b,
the response is of the
2
form eβ t
I The exponential part
will grow unbounded

14/39 Process Control Second Order Models and Response

Scilab code to plot step response of a
positive pole, b.sce

s = %s
G b = 1 / ( s −2)
t = 0:0.01:6;
y b = csim ( ’ step ’ , t , G b ) ;
plot2d ( t , y b )

15/39 Process Control Second Order Models and Response

Response for pole at c

I For the poles at

places indicated by c,
c
the response is of the
2
form eγ t ×(sinusoidal
terms)
I There will be growing
oscillations
16/39 Process Control Second Order Models and Response
Scilab code to plot step response of
complex conjugate poles in right half
plane, c.sce

s = %s
G c = 1 / ( s −(1+2∗%i ) ) / ( s −(1−2∗%i ) )
t = 0:0.01:10.15;
y c = csim ( ’ step ’ , t , G c ) ;
plot2d ( t , y c )

17/39 Process Control Second Order Models and Response

Response for poles at d

d
I For the poles at
places indicated by d,
the response is of the
form
2
e−δ t ×(sinusoidal
terms)
I There will be
d
decaying oscillations

18/39 Process Control Second Order Models and Response

Scilab code to plot step response of
complex conjugate poles in left half plane,
d.sce

s = %s
G d = 1 / ( s −(−1+3∗%i ) ) / ( s −(−1−3∗%i ) )
t = 0:0.01:6;
y d = csim ( ’ step ’ , t , G d ) ;
plot2d ( t , y d )

19/39 Process Control Second Order Models and Response

2. Second order system

20/39 Process Control Second Order Models and Response

Second order system
K
I Y(s) = G(s)U(s), G(s) =
τ 2 s2 + 2ζτ s + 1
I τ is the time constant
I ζ is the damping coefficient
I K is the steady state gain
K M
I Step response: Y(s) =
τ 2 s2 + 2ζτ s + 1 s
A B C
I = + +
s+a s+b s
I Is there a relation between a and b, A and B?
I For ζ < 1, a = b∗ , A = B∗
21/39 Process Control Second Order Models and Response
Step response of a second order system

K M A B C
I Y(s) = = + +
τ 2 s2 + 2ζτ s + 1 s s+a s+b s
2 2
I a and b are roots of τ s + 2ζτ s + 1 = 0
p
−2ζτ ± 4ζ 2 τ 2 − 4τ 2
I a, b =
2τ 2
ζ 1p 2
I =− ± ζ −1
τ τ
I ζ ≥ 1, a, b are real
I ζ < 1, complex, a = b∗ , A = B∗

22/39 Process Control Second Order Models and Response

Step response for ζ > 1
K M
Y(s) =
τ 2 s2 + 2ζτ s + 1 s

I Known as an overdamped system (ζ > 1)

I Its solution is given by

n
y(t) = KM 1 − e(−ζt/τ ) ∗
" p ! p !#
ζ2 − 1 ζ ζ2 − 1
cosh t +p sinh t }
τ ζ2 − 1 τ

23/39 Process Control Second Order Models and Response

Step response for ζ = 1

K M
Y(s) =
τ 2 s2 + 2ζτ s + 1 s

I Known as critically damped system (ζ = 1)

I Its solution is given by
   
t
y(t) = KM 1 − 1 + e−t/τ
τ

24/39 Process Control Second Order Models and Response

Step response for ζ < 1

K M
Y(s) =
τ 2 s2 + 2ζτ s + 1 s

I Known as underdamped system (ζ < 1)

I Its solution is given by

n
y(t) = KM 1 − e(−ζt/τ ) ∗
" p ! p !#
1−ζ 2 ζ 1−ζ 2
cos t +p sin t }
τ 1 − ζ2 τ

25/39 Process Control Second Order Models and Response

Another form of II order system

K
G(s) =
τ 2 s2 + 2ζτ s + 1
will also be written as
ωn2
G(s) = 2
s + 2ζωn s + ωn2

I K is taken as 1
I ωn : natural frequency, ζ: damping factor.
I For different ζ, get underdamped (ζ < 1),
critically damped (ζ = 1) and
overdamped (ζ > 1) systems
26/39 Process Control Second Order Models and Response
You should be able to derive all the
previous expressions for y(t)

27/39 Process Control Second Order Models and Response

Solution to underdamped system
Underdamped second order system:

ωn2
G(s) =
s2 + 2ζωn s + ωn2
p
For ζ < 1, rts of den. = −ζωn ± jωn 1 − ζ 2
The step response is,

y(t) = 1
" p #
e−ζωn t p 1 − ζ2
−p sin ωn 1 − ζ 2 t + tan−1
1 − ζ2 ζ

28/39 Process Control Second Order Models and Response

Step response of underdamped system

Mp
y(t)

tr tp ts
29/39 Process Control Second Order Models and Response
Scilab code step-II-order.sce
exec ( ’ second . s c i ’ ) ;
s = %s
omegan = 1 ; t = 1 : 0 . 1 : 1 5 ;
y = s e c o n d ( omegan , 0 . 1 , t ) ;
y = s e c o n d ( omegan , 0 . 5 , t ) ;
y = s e c o n d ( omegan , 1 , t ) ;
y = s e c o n d ( omegan , 2 , t ) ;
f u n c t i o n y = s e c o n d ( omegan , z e t a , t )
G = omegan ˆ 2 / ( s ˆ2+2∗ z e t a ∗ omegan ∗ s+
omegan ˆ 2 )
y = c s i m ( ’ s t e p ’ , t , G) ;
plot2d ( t , y)
endfunction
30/39 Process Control Second Order Models and Response
3. Integrating/capacitive systems

32/39 Process Control Second Order Models and Response

Integrating Process
Qi (t)

dh
I A = Qi − Q
dt
h(t) I In deviational variables,
Q d∆h
I A = ∆Qi (t) − ∆Q(t)
dt
I Taking Laplace transform,
As∆h(s) = ∆Qi (s) − ∆Q(s)
I ∆h(s) =
1 1
∆Qi (s) − ∆Q(s)
As As
33/39 Process Control Second Order Models and Response
Integrating Process
Qi (t)
In case of a step distur-
bance in feed flow, tank
will
h(t) 1. overflow
Q
2. run dry
3. overflow or run dry
4. stabilise at some
suitable value
Answer: 3: overflow or
run dry

34/39 Process Control Second Order Models and Response

Ramp input to a first order system
K a
Y(s) = U(s), U(s) = ↔ at
τs + 1 s2
K a A B C
I Y(s) = = + +
τ s + 1 s2 τs + 1 s s2
I Multiply by τ s + 1, let s = −1/τ , A = Kaτ 2
I Multiply by s2 :
Ka A
I = s2 + Bs + C
τs + 1 τs + 1
I By letting s = 0, C = Ka
I Differentiating blue equation w.r.t. s
I and letting s = 0, B = −Kaτ
35/39 Process Control Second Order Models and Response
Final value of ramp input to a first order
system
A B C
I Y(s) = + +
τs + 1 s s2
Kaτ 2 Kaτ Ka
I Y(s) = − + 2
τs + 1 s s
I Inverting
y(t) = Kaτ e−t/τ − 1 + Kat


I

I y(t = ∞) = Ka(t − τ )
I Let K = 1 and plot u(t) and y(t) in the same
plot

36/39 Process Control Second Order Models and Response

Rectangular Pulse

M
I Give the following input, u(t): t = 0 t = t1
I u(t) = 1(t) × M − 1(t − t1 ) × M
I 1(t) denotes unit step input
I u(t) = [1(t) − 1(t − t1 )] × M
 
1 1 −t1 s
I U(s) = − e M
s s
M
I U(s) = [1 − e−t1 s ]
s

37/39 Process Control Second Order Models and Response

What we learnt today

I Poles, zeros, response of systems for different

pole locations
I Solution to second order systems
I Integrating or capacitive systems

38/39 Process Control Second Order Models and Response

 Thank you

39/39 Process Control Second Order Models and Response