You are on page 1of 72

Design Via Root Locus

Chapter 9

(c) 2019 Farrokh Sharifi 1


Question of the Day
• Does control theory apply to politics?

• How to tune the government for improved


transient response and steady-state behavior?

(c) 2019 Farrokh Sharifi 2


Introduction: Example
• P, I, and D control
Self-driving car
https://www.youtube.com/watch?v=4Y7zG48uHRo

(c) 2019 Farrokh Sharifi 3


Introduction: Objectives (section 9.1)
• Improving the steady state error by designing cascade
compensators
• Improving the transient response by designing cascade
compensators
• Improving both the transient response and the steady state-
error by designing a cascade compensator
• Improving the transient response by designing feedback
compensators
• Implementing the compensators physically

(c) 2019 Farrokh Sharifi 4


Introduction: Design via Root Locus
• What do we mean by
design via root locus?
• What are our objectives?
 Improving Transient response
 Improving steady state error
Introduction:
Compensation Configurations
• There are two main configurations for compensators covered
in this chapter:
 Cascade compensation, Fig. a
 Feedback Compensation, Fig. b

(c) 2019 Farrokh Sharifi 6


Introduction: Compensator types
• Compensators may be divided into to two different types:
 Active Compensators ( Ideal compensators )
 Used to create pure integration or differentiation. For example active amplifiers
with power sources maybe used.
 Pros: They can reduce the steady state error to zero.
 Cons: They are expensive.
 Passive Compensators
 Used to approximate differentiation and integration.
 Pros: They are cheap and they do not require additional power sources
 Cons: They cannot reduce steady state error to zero.
Conclusion:
Factors such as cost, weight, desired performance, transfer function and
interface determine the type

(c) 2019 Farrokh Sharifi 7


Note from Previous Chapters
• What happens to steady-state error if we increase the
type of the system of forward path?

(c) 2019 Farrokh Sharifi 8


Improving Steady-State Error via
Cascade Compensation (section 9.2)P I
K ( s  a) Ka
Technique 1: ideal integral compensator. Gc  K
s s
Placing a pole at the origin of the open loop system increases the
system type by one improved steady state error
By proportional-plus-integral (PI) controller, we mean the ideal
integral compensator. This means feeding the error (proportional) +
the integral of the error forward to the plant. We require an active
controller for this.
( s  zc )
Gc 
Technique 2: Lag compensator. ( s  pc )
(no pure integrator). We use a passive controller and place the pole
close to zero instead of exactly on zero. The name of this comes
from it frequency characteristics.

9
(c) 2019 Farrokh Sharifi
Improving Steady-State Error via
Cascade Compensation - PI
K ( s  a) 1
Gc   K  Ka (gain + integrator)
s s
Question 1: what is the steady state error of a type zero system
for a step input? What is the steady state error of the system if we
add a pole at zero and therefore increase the type to one?

Question 2: When we add a pole at zero to the system to change


its type, the transient response also changes due to the angular
contribution of the pole. How can we maintain the desired
transient response while aiming to reduce the steady state error
simultaneously?

(c) 2019 Farrokh Sharifi 10


Improving Steady-State Error via
Cascade Compensation -PI
• Step 1: Assume the
desired transient response
of the original system is
achieved at point A on the
system root locus.

º
11
Improving Steady-State Error via
Cascade Compensation - PI
• Step 2: However, the
steady state error must be
reduced, so we add a pole
at zero. The new root
locus will be:

12
Improving Steady-State Error via
Cascade Compensation - PI
• Step 3: To make the root locus go
through point A, we have to
change the transfer function once
again. We add a zero close to the
pole so that the angular
contribution of the pole is
cancelled out and hence the root
locus again passes the desired
point A. This is the Ideal integral
compensator.
• Conclusion: while the system
type has increased (better steady-
state error), the transient response
has not changed (same point on s-
plane). º

(c) 2019 Farrokh Sharifi


Improving Steady-State Error via
Cascade Compensation - PI
• Example 9.1: System in figure (a), is operating with a
damping ratio of 0.174, We shall see what happens when we
add the PI controller shown in figure b for a step input.

14
Improving Steady-State Error via
Cascade Compensation - PI
Example: (cont.)
• The uncompensated system
has the root locus shown on
the right. Type 0 system.
• Intersection with   0.174 (  100.02 )
gives dominant closed loop
poles at 0.694  j3.926 & K  164.6
• Using K = 164.6, the third
pole is found at
-11.61. For type 0 system
with unit step input we
have

164.6
K P  lim KG(s)   8.23
s 0 20
(c) 2019 Farrokh Sharifi 15
Improving Steady-State Error via
Cascade Compensation - PI
Example: (cont.)
• We add the compensator and the root locus becomes:
dominant poles: -0.678+/- j3.837, K = 158.2
third pole: -11.5
New root locus
fourth pole: -0.0902

• Note that both systems’


closed loop poles are almost the same.
Gains also almost the same.
Fourth pole of compensated system is cancelled by
the zero  transient response the same. Steady-state
error improved.
(c) 2019 Farrokh Sharifi 16
Improving Steady-State Error via
Cascade Compensation - PI
Example: (cont.)
• Almost the same transient response but improved steady-
state error.

17
Improving Steady-State Error via
Cascade Compensation - PI
• Implementing PI controller:

18
Improving Steady-State Error via Cascade
Compensation - Lag Compensation
( s  zc )  zc   pc  0 
zc
1
Gc 
( s  pc ) pc
Close to zero
Let’s see improvement. Assume a type one system:
K ( s  z1 )( s  z2 ) ( z  zn ) ( s  zc )
G , Gc 
s( s  p1 )( s  p2 ) ( s  pn ) ( s  pc )
Kz1 z2 zn
Before compensation: KV 0  lim s 0 sG ( s ) 
p1 p2 pn

Kz1 z2 z n zc z
After compensation: KVN  lim s 0 sGc ( s)G ( s)  .  KV 0 c
p1 p2 pn pc pc
Increasing velocity constant
(c) 2019 Farrokh Sharifi 19
Improving Steady-State Error via Cascade
Compensation - Lag Compensation
• Increasing error constants will improve steady state error.

(c) 2017 Farrokh Sharifi 20


Improving Steady-State Error via Cascade
Compensation - Lag Compensation
• The effect on the transient response? The dominant pole P is at
the same location in the figure below (new pole and zero angles
cancel each other) so the transient response will not change.
• The effect on the required gain, K? The length of the new vectors
(of new pole and zero) are approximately equal and cancel each
other so the gain will not change much.

21
Improving Steady-State Error via Cascade
Compensation - Lag Compensation
• What improvement can we expect in the steady state error?

1. As it can be seen Kv can be improved by the ratio of the


compensator zero and compensator pole. This must be a large
ratio ( for example 10).
2. For the transient response to stay unchanged we must have the
pole and zero of the compensator close to each other.

How can we achieve 1 and 2? The pole and zero must be close to
the origin!  Choose  zc   pc  0 close to 0.
22
Improving Steady-State Error via Cascade
Compensation - Lag Compensation
• Example 9.2: Design a lag compensator for the system (Example
9.1) shown below to improve the steady state error by a factor of 10
when the system is operating with a damping ratio of 0.174.

164.6
K p  lim G ( s )   8.23
s 0 1  2  10
 Type 0 system

(c) 2019 Farrokh Sharifi


Improving Steady-State Error via Cascade
Compensation - Lag Compensation
Example: (cont.) increase by factor of 10

Arbitrarily selecting :

24
Improving Steady-State Error via Cascade
Compensation - Lag Compensation
Example: (cont.)
We may now look at the root locus of the compensated system
and compare it with the uncompensated:

(c) 2019 Farrokh Sharifi 25


Improving Steady-State Error via Cascade
Compensation - Lag Compensation
Example: (cont.)
The table below summarizes the results:

(c) 2019 Farrokh Sharifi 26


Improving Steady-State Error via Cascade
Compensation - Lag Compensation
Example: (cont.)
The step response of the compensated and uncompensated
systems are shown below:

27
Improving the Transient Response via
Cascade Compensation (section 9.3)
We already fixed the steady state error without effecting the
transient response. Now we want to improve the transient
response itself. Main objectives are:
• Improving settling time
• Improving over shoot
Technique 1: ideal derivative compensator (PD).
This requires active components.
Gc  K (s  zc )  Ks  Kzc
Technique 2: Lead compensator. ( s  zc )
Gc   pc   zc  0
This approximates differentiation ( s  pc )
z
with a passive network. ( c  1)
pc

(c) 2019 Farrokh Sharifi 28


Improving the Transient Response via
Cascade Compensation -PD

Note:
The transient response is selected by choosing appropriate
closed loop pole location. If the desired location is on the root
locus the gain has to be adjusted. If not, the root locus has to be
changed by the introducing a compensator so that the new root
locus contains the desired pole locations. Poles and zeros can
be added to the forward path to do that.
One simple way is to add a zero to the forward path:
Gc  s  zc
(differentiator + pure gain)
(c) 2019 Farrokh Sharifi 29
Improving the Transient Response via
Cascade Compensation -PD
Lets look at an example, assume we add a zero with the transfer
function:
Movie, showing the effect of changing gain:
Improving the Transient Response via
Cascade Compensation -PD
Movie showing the effect of compensator zero position:
Improving the Transient Response via
Cascade Compensation -PD
Conclusion:
• Transient responses unattainable by gain adjustment can be
obtained by augmenting the system’s poles and zeros with an
ideal derivative compensator
• Ideal derivative compensator (PD) can speed up the response
of the system while maintaining the overshoot ( same damping
ratio). Why?
• Settling time improves
• Peak time improves
• Steady state error may improve

(c) 2019 Farrokh Sharifi 32


Improving the Transient Response via
Cascade Compensation -PD
Same damping ratio: same percent overshoot
Example:

Larger imaginary More negative real


part: smaller peak part: shorter settling
time time

Uncompensated Compensated Compensated


(added zero) (added zero)

Remember for second-order systems:


Imaginary
real part
part
Improving the Transient Response via
Cascade Compensation -PD
Steps of designing an ideal derivative compensators to meet a
transient response specification:
• Step 1: Determine a desired point for the pole location that
yields a desired transient response
• Step 2: Evaluate the sum of angles from the open-loop poles
and zeros to the design point in step 1.
• Step 3. Calculate the required angular contribution of the
compensator by calculating the difference of 180 and the
calculated angles in step 2.
• Step 4. Based on the required angular contribution calculate the
position of the zero.

(c) 2019 Farrokh Sharifi 34


Improving the Transient Response via
Cascade Compensation -PD
Example: In the system shown below, design a PD compensator to yield 16%
overshoot with a threefold reduction in settling time.

%OS /100  16 
Step 1(a))   0.504 
  arccos(0.504)  59.74
Step 1(b) Find closed loop pole of uncompensated system. Search along the 180  59.74  120.26
line of damping ratio where Sum of angles from poles and zeros become
odd multiple of 180.
Step 1(c)

Step 1(d) Since 2nd-order approximation is used, search for the third pole to
the left of -6 where the same K = 43.35 is obtained. °
3rd pole = -7.59 > 5 x (-1.205), OK!

(c) 2019 Farrokh Sharifi


Improving the Transient Response via
Cascade Compensation -PD
Example (cont.):
Step 1(e) Based on step 1, we must find the desired second-order pole. Three fold improvement is
settling time means that the desired settling time should be: 1  3.32  1.107
3
Therefore, the real part of the pole is:
4 4
Ts  
n 
This is shown here where the imaginary part is
obtained from:

120.26
Improving the Transient Response via
Cascade Compensation -PD
Example (cont.):
Step 2. We have to find the some of the angles from the poles to the desired pole location:

120.26  86.35  68.8  275.6

Step 3. Based on step 3, we calculate the difference


of 180 to 275.6:
 zc  275.6  (2k  1)180
k  0   zc  455.6 or 95.6
6.19
Or simply arctan( )  86.35
4  3.613
275.6  180  95.6

6.19
arctan( )  68.8
6  3.613 120.26
Improving the Transient Response via
Cascade Compensation -PD
Example (cont.):
Step 4. The geometry is shown for this angular contribution in the root
locus is shown below and we must now solve for sigma.
95.6
  3.006  Gc  s  3.006

This results in the following root locus:


σ

120.26
Improving the Transient Response via
Cascade Compensation -PD
• Implementing PD controller

The transfer function of the controller is:

Problems of PD controllers are:


• Sensitivity to noise
• Requires active components
Solution is to use a lead compensator to overcome the
disadvantages of ideal differentiation. 39
Improving the Transient Response via
Cascade Compensation - Lead Compensation
A lead compensator, approximates a single zero or ideal
differentiation with a zero and pole such that the angular
contribution is positive. This is possible when the pole is
further away from the imaginary axis compared to the zero:

40
Improving the Transient Response via
Cascade Compensation - Lead Compensation
The advantages of passive lead networks are:
1. No addition power supply is required as we use passive
elements
2. Noise due to differentiation is reduced.

The disadvantage of passive lead network is that :


• Addition of pole does not reduce the number of branches
of the root locus that cross the imaginary axis. The
addition of a single zero, of the PD controller, reduces
the number of branches that cross into the right half-
plane.

(c) 2019 Farrokh Sharifi 41


Improving the Transient Response via
Cascade Compensation - Lead Compensation
Steps of designing a lead compensator:
• Step 1: Determine a desired point for the pole location
that yields a desired transient response
• Step 2: Evaluate the sum of angles from the open-loop
poles and zeros to the design point in step 1.
• Step 3. Calculate the required angular contribution of the
compensator by calculating the difference of 180 and the
calculated angles in step 2.
• Step 4. An infinite number of poles and zeros can satisfy
this angular contribution. Choose arbitrarily either a pole
or zero and based on the required angular contribution
find the location of the remaining compensator zero or
pole. 42
Improving Steady State Error and
Transient Response (section 9.4)
We may use the combination of derivative and integration to
improve the transient and steady-state error independently.

If we design an active PD controller followed by an active


PI controller, the resulting compensator is called a
proportional-plus-integral-plus-derivative(PID) controller.

If we first design a passive lead compensator and then


design a passive lag compensator, the resulting compensator
is called a lag-lead compensator.

(c) 2019 Farrokh Sharifi 43


Improving Steady State Error and Transient
Response - PID Controller Design
A PID controller transfer function is provided below:

The corresponding block diagram is as is shown below:

44
Improving Steady State Error and Transient
Response - PID Controller Design
The steps in general for designing a PID controller are as
follows:
• Step 1: Evaluate the performance of the uncompensated
system to determine how much improvement in transient
response is required.
• Step 2: Design the PD controller to meet the transient
response specifications. This includes the zero location
and the loop gain
• Step 3. Simulate the system to make sure all requirements
are met.
• Step 4. Redesign if the simulation shows requirements
have not been met.
(c) 2019 Farrokh Sharifi 45
Improving Steady State Error and Transient
Response - PID Controller Design
• Step 5: Design the PI controller to yield the required
steady-state error.
• Step 6: Determine the gains K1,K2, K3.
• Step 7. Simulate the system to make sure all requirements
are met.
• Step 8. Redesign if the simulation shows requirements
have not been met.

(c) 2019 Farrokh Sharifi 46


Improving Steady State Error and Transient
Response - PID Controller Design
Example: Design a PID controller so that the system can operate with a
peak time that is two-thirds that of the uncompensated system at
20% overshoot and with zero steady-state error for a unit step input.

Step 1

%OS /100  20    0.456    arccos(0.456)  62.87 


angle with positive real-axis=180 -62.87  117.13

search along the line of 117.13 deg for dominant poles 117.13

p1,2  5.415  j10.57


 (distances to poles from p1 )
K  121.5
 (distances to zeros from p1 )
Improving Steady State Error and Transient
Response - PID Controller Design
Example (cont.)
Step 1: search the region between -8 and -10 for the third pole with the same
K value: p3 = ₋ 8.169 which is close to zero at -8  2-nd order approx. valid

= 4 / 5.415 = 0.739 sec

= 3.14 / 10.57 = 0.297 sec

Step 2  
Tp   
n 1   2 d
 
imaginary part of compensated system dominant poles d    15.87
Tp (2 / 3)(0.297)
d
real part of compensated system dominant poles  =  8.13
tan117.13
(c) 2019 Farrokh Sharifi 48
Improving Steady State Error and Transient
Response - PID Controller Design
Example (cont.)
Step 2 : sum of angles from uncompensated system poles and zeros to
the desired compensated system pole = -198.37 deg
 zc  198.37  (2k  1)180  k  0   zc  378.37 or 18.37
15.87
tan(18.37 )   zc  55.92  GPD  ( s  55.92)
zc  8.13
RL for PD-compensated system

117.13

(c) 2017 Farrokh Sharifi 49


Improving Steady State Error and Transient
Response - PID Controller Design
Example (cont.)
Steps 3 and 4 :Simulate and see both peak time and steady-state errors
have been improved.

Steady-state error is not


Zero yet

(c) 2019 Farrokh Sharifi 50


Improving Steady State Error and Transient
Response - PID Controller Design
Example (cont.)
Step 5 design PI controller to reduce steady-state error to zero.
Place zero of PI close to origin. s  0.5
GPI ( s) 
s
Sketch the Root Locus for the whole system of PID compensated system
Search along the damping ratio line to find
dominant poles and gain there
p1,2  7.516 j14.67
K  4.6

51
Improving Steady State Error and Transient
Response - PID Controller Design
Example (cont.)
Step 6 s  0.5 4.6( s  55.92)( s  0.5)
GPID ( s )  K .( s  55.92) 
s s
K1 K2
K ( s 2
 s  )
4.6( s  56.42s  27.96)
2 3
K3 K3
 
s s
 K1  259.5, K 2  128.6, K 3  4.6
Steps 7 and 8

52
Improving Steady State Error and Transient
Response - PID Controller Design
Example (cont.)

(c) 2019 Farrokh Sharifi 53


Improving Steady State Error and Transient
Response - PID Controller Design
Tutorial- Consider a temperature control for a chemical process. The
uncompensated system is operating with a rise time approximately
the same as a second-order with a peak-time of 16 sec and 5%
overshoot. There is also a considerable steady-state error. Design a
PID controller so that the compensated system will have a rise time
approximately equivalent to a second-order system with a peak-time
of 8 sec and 5% overshoot and zero steady-state error.

54
Improving Steady State Error and Transient
Response - PID Controller Design
Step 1:

2.88 102
K P  lim s 0 GH   0.575
0.4  0.5  0.163 1.537
1
e()   0.6349
1  KP

(c) 2019 Farrokh Sharifi 55


Improving Steady State Error and Transient
Response - PID Controller Design
Step 2:

0.393 -0.374+j0.393
 p 1.537  arctan  18.67
1.537  0.374
0.393
 p 0.5  arctan  72.22
0.5  0.374
0.393
 p 0.4  arctan  86.2
0.4  0.374
0.393 -1.537 -0.5 -0.4 -0.163
 p 0.163  180  arctan  118.23
0.374  0.163
 p 1.537   p 0.5   p 0.4   p 1.537  295.3
Improving Steady State Error and Transient
Response - PID Controller Design
Step 2:

 (lengths to poles)
GPD  ( s  0.19) 0.35K  
 (lengths to zeros including new zero)
0.3932  (1.537  0.374) 2  0.3932  (0.5  0.374) 2 
0.3932  (0.4  0.374) 2  0.3932  (0.374  0.163) 2
 0.2
0.393  (0.374  0.19)
2 2

57
Improving Steady State Error and Transient
Response - PID Controller Design
Steps 3, 4: Simulations. OK.

Step 5:

Step 6:

(c) 2019 Farrokh Sharifi 58


Improving Steady State Error and Transient
Response - PID Controller Design

(c) 2019 Farrokh Sharifi 59


Improving Steady State Error and Transient
Response - PID Controller Design

• Ball and Beam:


http://www.youtube.com/watch?v=heAegVuji7Q

• Helicopter control
http://www.youtube.com/watch?v=y1hjVZq1VNY

(c) 2019 Farrokh Sharifi 60


Improving Steady State Error and Transient
Response -Lead-Lag Compensator Design
If instead of an ideal derivative and an ideal integral we use
a lead and lag network we have a lead-lag compensator. The
steps for the design are as follows:
• Step 1: Evaluate the performance of the uncompensated
system to determine how much improvement in transient
response is required.
• Step 2: Design the lead compensator to meet the transient
response specifications. The design includes the zero
location, pole location and open loop gain.
• Step 3. Simulate the system to make sure all requirements
are met.
• Step 4. Redesign if the simulation shows requirements
have not been met. 61
Improving Steady State Error and Transient
Response -Lead-Lag Compensator Design
• Step 5: Evaluate the steady state error performance for
the lead-compensated system to determine how much
more improvement in the steady-state error is required.
• Step 6: Design the lag compensator to yield the required
steady state error.
• Step 7. Simulate the system to make sure all requirements
are met.
• Step 8. Redesign if the simulation shows requirements
have not been met.

(c) 2019 Farrokh Sharifi 62


Improving Steady State Error and Transient
Response -Lead-Lag Compensator Design
• Example: Design a lag-lead compensator for the system
shown below to operate with a 20% overshoot and a twofold
reduction in settling time. Further the compensated system
will exhibit a tenfold improvement in steady-state error for a
ramp input.

Solution:
Step1: First we evaluate the performance of the
uncompensated system.
OS  20%    0.456
(c) 2019 Farrokh Sharifi 63
Improving Steady State Error and Transient
Response -Lead-Lag Compensator Design
Example (cont.):

Step 2: We begin the lead compensator design. For twofold


reduction of the settling time, the real part of the dominant
poles must be increased by a factor of two:

64
Improving Steady State Error and Transient
Response -Lead-Lag Compensator Design
Example (cont.):
Step 2: The imaginary part of the design point is therefore:

We arbitrarily choose the lead zero coincident with the pole


at -6. Why? The sum of the angles to the design point from
the uncompensated system’s poles and zeros and the
compensator zero is -164.65. Therefore -180+164.65=-15.35
65
is the angular contribution of the pole.
Improving Steady State Error and Transient
Response -Lead-Lag Compensator Design
Example (cont.):
The pole position pc is found to be -29.1 from the equation
below:

The lead-compensated system is shown below. The gain at


the design point is 1977. The open loop lead-compensated
system will be
1977
GLC ( s ) 
s ( s  10)( s  29.1)

66
Improving Steady State Error and Transient
Response -Lead-Lag Compensator Design
Example (cont.):
Step 3 and 4: We check the design with simulation and
confirm.

(c) 2019 Farrokh Sharifi 67


Improving Steady State Error and Transient
Response -Lead-Lag Compensator Design
Example (cont.):
Step 5: We design the lag compensator. The task is to
improve the static error constant for a ramp by tenfold. The
original systems static error constant is
192.1
uncompensated G ( s)   KV  lim s 0 sG( s)  3.201
s( s  6)( s  10)
1977
lead  compensated GLC ( s)   KV  LC  lim s 0 sGLC ( s)  6.794
s( s  10)( s  29.1)
1 3.201 1
e(  )   eLC ()  e(  )  e()
KV 6.794 2.122

With total 10 fold improvement, the portion left to lag compensation will
be 10/2.122=4.713.
( s  0.04713)
Step 6. We arbitrarily choose pole at 0.01. Then G lag ( s ) 
( s  0.01)
(c) 2019 Farrokh Sharifi 68
Improving Steady State Error and Transient
Response -Lead-Lag Compensator Design
Example (cont.): Open loop lead-lag compensated system will be

K ( s  0.04713)
G LLC ( s ) 
s ( s  10)( s  29)( s  0.01)

Valid 2nd-order system

69
Improving Steady State Error and Transient
Response -Lead-Lag Compensator Design
K ( s  0.04713)
G LLC ( s ) 
Example (cont.): s ( s  10)( s  29)( s  0.01)

Step 7- Simulate

(c) 2019 Farrokh Sharifi 70


Improving Steady State Error and Transient
Response -Lead-Lag Compensator Design
K ( s  0.04713)
G LLC ( s ) 
Example (cont.): s ( s  10)( s  29)( s  0.01)

Step 7- Simulate

(c) 2019 Farrokh Sharifi 71


Summary
• PI and Lag Compensation Methods for improving steady-
state error were introduced.
• Algorithms for designing PI and Lag Compensation methods
were given.
• PD and Lead Compensation Methods for improving transient
response were given.
• Algorithms for designing PD and Lag Compensation
methods were provided.
• PID and Lead-Lag Compensation methods for improving
both steady-state error and transient response were
introduced, and the Algorithms for their design were
provided.

(c) 2019 Farrokh Sharifi 72

You might also like