Lecture 5

Classical Control Overview III

Dr. Radhakant Padhi
Asst. Professor
Dept. of Aerospace Engineering
Indian Institute of Science - Bangalore
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
2
A Fundament al Pr obl em i n
Cont r ol Syst ems
Poles of open loop transfer function are easy to find and they
do not change with gain variation either.
Poles of closed loop transfer function, which dictate stability
characteristics, are more difficult to find and change with gain
( ) ( )
( )
( )
1 ( ) ( )
( ) ( )
( ) ( )
( ) ( )
( ) ( )
( )
( ) ( ) ( ) ( )
G H
G H
G H
G H G H
open loop transfer function KG s H s
closed loop transfer function
KG s
T s
KG s H s
let
N s N s
G s and H s
D s D s
KN s D s
T s
D s D s KN s N s
=
=
+
= =
=
+
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
3
Ex ampl e
3 2
( 1) ( 3)
Let ( ) , ( )
( 2) ( 4)
( 1)( 3)
Then ( ) ( )
( 2)( 4)
Poles of open loopTransfer Function ( ) ( ) (0, 2, 4)
Now Closed Loop Transfer Function ( )
( 1)( 4)
( )
(6 ) (8 4 ) 3
P
s s
G s H s
s s s
K s s
KG s H s
s s s
KG s H s are
T s
K s s
T s
s K s K s K
+ +
= =
+ +
+ +
=
+ +

+ +
=
+ + + + +
oles of ( ) arenot immediately known; they depends on as well.
Systemstability and transient responsedepends on poles of ( )
Root Locus gives vivid pictureof thepoles of ( ) varies
T s K
T s
T s as K

ADVANCED CONTROL SYSTEM DESIGN

Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
4
Vec t or Repr esent at i on of Compl ex
Number s and Compl ex Func t i ons
We can conclude that (s + a) is a complex number and can be represented by a
vector drawn from the zero of the function (- a) to the point s.
s j = +
( )
( ) F s s a a j = + = + +
( )
( ) ( ) zero F s zero s a a = + =
( )
5 2
7
s j
s
+
+
Regular representation:
Alternate representation:
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
5
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
Magni t ude and Angl e of Compl ex
Func t i ons
1
1
1
1
( )
Let ( ) , ,
( )
The magnitude of ( ) at any point s is
( )
( )
The angle , , of ( ) at any point s is
m
i
i
n
j
j
m
i
i
n
j
i
s z
F s m number of zeros n number of poles
s p
M F s
s z
zero lengths
M
pole lengths
s p
F s
zero

=
=
=
=
+
=
+
+
= =
+
=

1 1
( ) ( )
m n
i j
i j
angles pole angles s z s p
= =
= + +

ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
6
Ex ampl e
0
0
0
0 0 0
0
( 1)
( )
( 2)
Vector originating at 1: 20 116.6
Vector originating at 0: 5 126.9
Vector originating at 2: 17 104.0
of ( )
20
= 116.6 126.9 104.0
5 17
=0.217 114.3
s
F s
s s
M F s
+
=
+

Root Locus Analysis

Dr. Radhakant Padhi
Asst. Professor
Dept. of Aerospace Engineering
Indian Institute of Science - Bangalore
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
8
Fac t s
Introduced by W. R. Evans (1948)
Graphical representation of the closed loop poles in
s-plane as the system parameter (typically controller
gain) is varied.
Gives a graphic representation of a systems stability
characteristics
Contains both qualitative and quantitative information
Holds good for higher order systems
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
9
A Mot i vat i ng Ex ampl e
Pole location as a function of gain for
the system.
Ref: N. S. Nise:
Control Systems Engineering,
4
th
Ed., Wiley, 2004
over damped
under damped
Note: Typically K > 0 for negative feedback systems
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
10
A Mot i vat i ng Ex ampl e
Real parts are same:
same settling time
As gain increases,
damping ratio decreases
%age overshoot
increases
Poles are always negative:
system is always stable
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
11
Pr oper t i es of Root Loc us
The closed loop Transfer Function
( )
( )
1 ( ) ( )
Characteristics equation: 1 ( ) ( ) 0
Closed loop poles are solution of characteristics equation.
However, 1 ( ) ( ) 0 is a complex quantity. Henc
KG s
T s
KG s H s
KG s H s
KG s H s
=
+
+ =
+ =
0
0
e, it can be expressed as
( ) ( ) 1 1 (2 1)180 0, 1, 2, 3,...........
The abovecondition(Evan's condition)can bewritten as
( ) ( ) 1 (Magnitudecriterion)
( ) ( ) (2 1)180 (Anglecriterion)
KG s H s k k
KG s H s
KG s H s k
= = + =
=
= +
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
12
Ex ampl e
Example: Let us consider
a system as in the Figure
and consider two points:
2
2
: 2
2
P j

+



1
: 2 3 P j +
0
If the angle of a complex number is an odd multiple of 180 for an open loop
transfer function, then it is a pole of the closed loop system with
1 1

( ) ( ) ( ) ( )
K
G s H s G s H s
= =
Fact :
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
13
Ex ampl e
1 1 2 3 4
0 0
0
2
For : 56.31 71.57 90 108.43
= 70.55 (2 1)180
Therefore 2 3is not a point on root locus
For : 180
2
Hence 2 is on root locus for somevalueof .
2
Moreov
i
i
P
k
j
P
j K

= + = +
+
+
=

+

3 4
1 2
2
(1.22)
2
er, 0.33
(2.12) (1.22)
L L
K
L L

= = =

ADVANCED CONTROL SYSTEM DESIGN

Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
14
Summar y
Given the poles and zeros of the open loop transfer function
( ) ( ), a point in the - plane is on the root locus for a
particular value of gain , if the anglesof the zeros minus the
angles of the po
KG s H s s
K
0
les add up to (2 1)180.
Furthermore, the gain at that point canbe found by dividing
the product of the lengths of the poles by the product of the
lengths of the zeros.
k
K
+
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
15
Sk et c hi ng t he Root Loc us:
Basi c Fi ve Rul es
1. Number of Branches: The number of branches of the root locus
equals the number of closed loop poles (since each pole should
move as the gain varies).
2. Symmetricity: A root locus is always symmetric about the real
axis, since complex poles must always appear in conjugate pairs.
3. Real-axis segments: On the real axis, for K > 0, the root locus
exists to the left of an odd number of real-axis finite open-loop
poles and zeros.
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
16
Sk et c hi ng t he Root Loc us:
Basi c Fi ve Rul es
4. Starting and ending points: The root locus begins at the finite
and infinite poles of G(s)H(s) and ends at the finite and infinite
zeros of G(s)H(s)
5. Behavior at Infinity: The root locus approaches straight lines
as asymptotes as the locus approaches infinity
a
a
The equation of asymptotes is given by the real-axis intercept
and angle as follows:
. .
(2 1)
. .
a
a
finite poles finite zeros
No of finite poles No of finite zeros
k
No of finite poles No of finite zeros
where k

+
=

0, 1, 2, 3,.....
. and the angle is given in radians wrt positive extension of real axis
=
For additional rules, refer to:
N. S. Nise: Control Systems
Engineering, 4
th
Ed., Wiley, 2004.
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
17
Ex ampl e
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
18
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
Cont r ol Tuni ng vi a Gai n
Adj ust ment usi ng Root Loc us
Sample root locus:
A. Possible design point via
gain adjustment
B. Desired design point that
cannot be met via simple
gain adjustment (needs
dynamic compensators)
PID Control Design
Dr. Radhakant Padhi
Asst. Professor
Dept. of Aerospace Engineering
Indian Institute of Science - Bangalore
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
20
I nt r oduc t i on
The PID controller
involves three
components:
Proportional feedback
Integral feedback
Derivative feedback
By tuning the three
components (gains), a
suitable control action is
generated that leads to
desirable closed loop
response of the output.
Ref : en.wikipedia.org/wiki/PID_controller
( )
u t
( )
U s
( )
E s
Ref: N. S. Nise: Control Systems Engineering, 4
th
Ed., Wiley, 2004.
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
21
Phi l osophy of PI D Desi gn
The proportional component determines the reaction to the
current value of the output error. It serves as a all pass
block.
The integral component determines the reaction based on
the integral (sum) of recent errors. In a way, it accounts
for the history of the error and serves as a low passblock.
The derivative component determines the reaction based
on the rate of change of the error. In a way, it accounts for
the future value of the error and serves as a high pass
block.
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
22
PI D Cont r ol l er


The final form of the PID algorithm is:
( )= ( ) ( ) ( )
( )
/
( )
The tuning parameters are:
Proportional gain,
Integral gain, /
Derivative gain,
t
p i d
o
p i d
p
i p i
d p d
d
u t K e t K e d K e t
dt
U s
K K s K s
E s
K
K K T
K K T
+ +
= + +
=
=

ADVANCED CONTROL SYSTEM DESIGN

Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
23
Ef f ec t of Pr opor t i onal Ter m
Proportional Term:
If the gain K
p
is low, then control action may be too
small when responding to system disturbances. Hence,
it need not lead to desirable performance.
If the proportional gain K
p
is too high, the system can
become unstable. It may also lead to noise amplification
( )
: Proportional term of output
: Proportional gain (a tuning parameter)
out p
out
p
P K e t
P
K
=
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
24
Ef f ec t of I nt egr al Ter m
Integral term :
0
( )
: Integral term of output
: Integral gain, a tuning parameter
The integral term (when added to the proportional term)
accelerates the movement of the process towards setpoint
and e
t
out i
out
i
I K e d
I
K
=

liminates the residual steady-state error that occurs
with a proportional only controller.
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
25
Ef f ec t of I nt egr al Ter m
2 2
i
It can beseen that adding an integral termto a pureproportional
termincreases thegain bya factor of
1 1
1+ 1 1, for all .
andsimultaneously increases thephase-lagsince
1
1+
i
j T T
j T

= + >

Destabilizing effects of the Integral term:

1
i
1
tan 0for all .
Because of this, both thegain margin (GM) and phasemargin (PM)
arereduced, and theclosed-loopsystembecomes moreoscillatory
and potentially unstable.
i
T

= <


Ref : Li, Y. , Ang, K.H. and Chong, G.C.Y. (2006)
PID Control System Analysis and Design.
IEEE Control Systems Magazine 26(1):pp. 32-41.
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
26
Ef f ec t of I nt egr al Ter m
Integrator Windup:
If the actuator that realizes control action has
saturated and if it is neglected, this causes low
frequency oscillations and leads to instability.
Remedies:
Automatic Resetting, Explicit Anti-windup.
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
27
Ef f ec t of Der i vat i ve Ter m
Derivative term :
( )
: Derivative gain (a tuning parameter)
The derivative term speeds up the transient behaviour. In general, it has
negligible effect on the steady state performance (for step inputs,
out d
d
d
D K e t
dt
K
=

the
effect on steady state response is zero).
Differentiation of a signal amplifies noise. Hence, this term in
the controller is highly sensitive to noise in the error term! Because
of this, the

derivative compensation should be used with care.

ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
28
Ef f ec t of Der i vat i ve Ter m
( ) [ ]
1
2
Adding a derivativetermto pureproportional termreduces the
phaselag by
1 tan 0, / 2 for all
1
which tends to increasethePhase Margin.
In themeantime, however, thegain increases by a factor of
1 1
d
d
d
T
j T
j T T

+ =


+ = +
2
1, forall .
which decreases the Gain Margin.
Hencetheoverall stability may beimproved or degraded.
: Low pass filter, Set-point filter, Pre-filter etc.
d
>
Additional requirement
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
29
Ef f ec t of I nc r easi ng Gai ns on
Out put Per f or manc e
Parameter Rise time Overshoot Settling
time
Error at
equilibrium
K
p
Decreases Increases Small
change
Decreases
K
i
Decreases Increases Increases Eliminated
K
d
Indefinite
(can either
decrease or
increase)
Decreases Decreases No Effect
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
30
Tuni ng of PI D Desi gn
Offline method. Good for
first-order processes
Results in good tuning in
general
Cohen-Coon
Method Advantages Disadvantages
Manual Tuning No math required. Largely
trial-and-error approach
(based on general
observations)
Requires experienced
personnel
ZieglerNichols Proven online method. Very aggressive tuning,
May upset some inherent
advantages of the process
Software Tools Online or offline methods. Some cost and training of
personnel is involved
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
31
Zi egl er Ni c hol s Met hod f or Gai n
Tuni ng
The K
i
and K
d
gains are first set to zero. The P gain is increased
until it reaches the critical gain, K
c
, at which the output of the loop
starts to oscillate. Next, K
c
and the oscillation period P
c
are used to
set the gains as per the following rule.
Control Type K
p
K
i
K
d
P 0.50K
c
- -
PI 0.45K
c
1.2K
p
/ P
c
-
PID 0.60K
c
2K
p
/ P
c
K
p
P
c
/ 8
Note: The constants used may vary depending on the application.
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
32
Li mi t at i ons of PI D c ont r ol
It is a SISO design approach and hence can
effectively handle only such system
System should behave in a fairly linear manner.
Hence, it is valid in close proximity of an operating
point (about which the linearized system is valid)
Does not take into account the limitations of the
actuators
Techniques to overcome:
Gain scheduling, Cascading controllers, Filters in loop etc.
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
33
ADVANCED CONTROL SYSTEM DESIGN
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
34