MECE 3350U

Control Systems

Lecture 9
Dominant Poles and Zeros

1/41
MECE 3350 - C. Rossa 1 / 41 Lecture 9
Videos in this lecture

Lecture: [Link]

Exercise 40: [Link]

Exercise 41: [Link]

Exercise 42: [Link]

Exercise 43: [Link]

Exercise 44: [Link]

2/41
MECE 3350 - C. Rossa 2 / 41 Lecture 9
Outline of Lecture 9

By the end of today’s lecture you should be able to

• Understand the concept of dominant poles

• Recognize the influence zeros on the transient repose

• Simplify a transfer function to lower orders

3/41
MECE 3350 - C. Rossa 3 / 41 Lecture 9
Applications

The roll control autopilot of an aircraft has the following structure:

How can we calculate the k that yields an overshoot of less than 2%?

4/41
MECE 3350 - C. Rossa 4 / 41 Lecture 9
Applications

A ventricular assist device is a mechanical pump used to support heart function

and blood flow in people with weak or failing hearts.

The model of the heart and pump system results in a third order transfer
function. How can we analyse the transient response of the system?

5/41
MECE 3350 - C. Rossa 5 / 41 Lecture 9
First order system

Consider the response of a first order system to an unit step input:

1 1
 
X (s) =
s +a s
Using partial fraction expansion:

1/a 1/a
X (s) = −
s s +a
The inverse transform yields
1
x (t) = (1 − e −at )
a
The transfer function has one pole located at s = −a.

→ How does the magnitude of s = −a influence the transient response?

6/41
MECE 3350 - C. Rossa 6 / 41 Lecture 9
The effect of an additional pole
Let us now examine the step response of
p 1 1 1
   
X (s) = = .
(s + 1)(s + p) s (s + 1)( p1 s + 1) s
Partial fraction expansion gives:
p 1
x (t) = 1 − e −t + e −pt
p−1 p−1
1 3
y

0 -2
0 time [sec] 15 0 15

Conclusion: If p >> 1, the term 1/(p − 1)e −pt is negligibly small as t → ∞.

7/41
MECE 3350 - C. Rossa 7 / 41 Lecture 9
The effect of an additional pole

If the magnitude of the real part of a pole is at least 5 to 10 times that of a

dominant pole, then the pole may be regarded as insignificant.

8/41
MECE 3350 - C. Rossa 8 / 41 Lecture 9
Second order systems with an additional pole
Consider the 3rd order function
1
T (s) = .
(s 2 + 2ζωn s + 1)(γs + 1)
Real part of the poles are: −1/γ and −ζωn . Thus, if
 
1
  ≥ 10|ζωn | (1)
γ 
The response can be approximated by
1
Ta (s) = .
s 2 + 2ζωn s + 1
Take ωn = 1, and ζ = 0.45: gives two poles at s = −0.45 ± 0.89i.
Example 1: γ = 1.00 → Adds a pole to s = −1
Example 2: γ = 0.22 → Adds a pole to s = −4.5.
Example 3: γ = 0.10 → Adds a pole to s = −10.
9/41
MECE 3350 - C. Rossa 9 / 41 Lecture 9
Second order systems with an additional pole
Original 3rd order function:
1
T (s) = .
(s 2 + 2ζωn s + 1)(γs + 1)
2nd order approximation:
1
Ta (s) = .
s 2 + 2ζωn s + 1
1.5
y

0
0 Time (seconds) 15 0 Time (seconds) 15 0 Time (seconds) 15

10/41
MECE 3350 - C. Rossa 10 / 41 Lecture 9
Additional zeros

Consider the transfer function with an additional zero s = −z:

n ω2
Y (s) (s + z)
= 2 z (2)
R(s) s + 2ζωn s + ωn2
If z >> ζωn , the zero will have minimal effect on the step response.

The unit step response of the above equation is:

ω2
Y (s) ωn2 z
n
s
= 2 2
+ 2 (3)
R(s) s + 2ζωn s + ωn s + 2ζωn s + ωn2
If x (t) is the inverse of the first term, then the time response is

1 d
 
y (t) = x (t) + x (t) (4)
z dt
Conclusion: The additional zero speeds us transients, making rises and falls
sharper.

11/41
MECE 3350 - C. Rossa 11 / 41 Lecture 9
Additional zeros

n ω2
Y (s) (s + z)
= 2 z (5)
R(s) s + 2ζωn s + ωn2
Consider: ωn = 1, ζ = 0.45, z = 0.7, 1, 10

2
y

0
0 time (seconds) 15

12/41
MECE 3350 - C. Rossa 12 / 41 Lecture 9
Simplification to a lower order
A more precise approach: Match the frequency response.

Consider the high order system:

am s m + am−1 s m−1 + . . . + a1 s + 1
GH (s) = K (6)
bn s n + bn−1 s n−1 + . . . + b1 s + 1
with m ≥ n, which is to be mapped to a lower order system

cp s p + cp−1 s p−1 + . . . + c1 s + 1
GL (s) = K (7)
dg s g + dg−1 s g−1 + . . . + d1 s + 1
such that p ≤ g ≤ n.
The c and d coefficients of the approximate solution GL are obtained via

dk
Mk = M(s) (8)
ds k

dk
∆k = ∆(s) (9)
ds k
13/41
MECE 3350 - C. Rossa 13 / 41 Lecture 9
Simplification

Let us define
2q
X (−1)k+q M k (0)M 2q−k (0)
M2q = (10)
k!(2q − k)!
k=0

2q
X (−1)k+q ∆k (0)∆2q−k (0)
∆2q = (11)
k!(2q − k)!
k=0

So that the c and d coefficient are obtained by equating

M2q = ∆2q (12)

for q = 1, 2... and up to the number required to solve for the unknowns.

14/41
MECE 3350 - C. Rossa 14 / 41 Lecture 9
Location of poles

ωn2
H(s) =
s 2 + 2ζωn s + ωn2

The poles are

p
s = ζωn ± jωn 1 − ζ2
s = −σ ± jωd
p
where σ = ζωn , and ωd = ωn 1 − ζ2

→ Poles are located at a radius ωn

→ The angle to the imaginary axis is θ = sin−1 ζ

15/41
MECE 3350 - C. Rossa 15 / 41 Lecture 9
Exercise 40

A closed-loop control system has a transfer function T (s) as follows

Y (s) 2500
T (s) = = .
R(s) (s + 50)(s 2 + 10s + 50)

Plot the time response to an unit step input when:

→ (a) The actual T (s) is used (use Matlab)

→ (b) Using the dominant complex poles
→ (c) Compare the results

16/41
MECE 3350 - C. Rossa 16 / 41 Lecture 9
Exercise 40 - continued
(a) The actual function is

Y (s) 2500
T (s) = = .
R(s) (s + 50)(s 2 + 10s + 50)
(b) The approximate transfer function is

17/41
MECE 3350 - C. Rossa 17 / 41 Lecture 9
Exercise 41

A closed-loop control system transfer function as two dominant complex

conjugate poles. Sketch the region in the left-hand s-plane where the complex
poles should be located to meet the given specifications:

→ (a) 0.6 ≤ ζ ≤ 0.8, ωn ≤ 10

→ (b) 0.5 ≤ ζ ≤ 0.707, ωn ≥ 10
→ (c) ζ ≥ 0.5, 5 ≤ ωn ≤ 10
→ (d) ζ ≤ 0.707, 5 ≤ ωn ≤ 10
→ (e) ζ ≥ 0.6, ωn ≤ 6

18/41
MECE 3350 - C. Rossa 18 / 41 Lecture 9
Exercise 41 - continued
→ (a) 0.6 ≤ ζ ≤ 0.8, ωn ≤ 10

19/41
MECE 3350 - C. Rossa 19 / 41 Lecture 9
Exercise 41 - continued
→ (b) 0.5 ≤ ζ ≤ 0.707, ωn ≥ 10

20/41
MECE 3350 - C. Rossa 20 / 41 Lecture 9
Exercise 41 - continued
→ (c) ζ ≥ 0.5, 5 ≤ ωn ≤ 10

21/41
MECE 3350 - C. Rossa 21 / 41 Lecture 9
Exercise 41 - continued
→ (d) ζ ≤ 0.707, 5 ≤ ωn ≤ 10

22/41
MECE 3350 - C. Rossa 22 / 41 Lecture 9
Exercise 41 - continued
→ (e) ζ ≥ 0.6, ωn ≤ 6

23/41
MECE 3350 - C. Rossa 23 / 41 Lecture 9
Exercise 42

A closed-loop transfer function is

Y (s) 108(s + 3)
T (s) = = .
R(s) (s + 9)(s 2 + 8s + 36)

→ (a) Determine the steady state error for a unit step input.

→ (b) Assume that the complex poles dominate and determine the percent
overshoot an setting time.

→ (c) Plot the actual system response and compare it with (b)

24/41
MECE 3350 - C. Rossa 24 / 41 Lecture 9
Exercise 42 - continued
(a) Steady-state error for r (t) = 1.

Y (s) 108(s + 3)
T (s) = = .
R(s) (s + 9)(s 2 + 8s + 36)

25/41
MECE 3350 - C. Rossa 25 / 41 Lecture 9
Exercise 42 - continued
(b) Overshoot and settling time considering the dominant poles.

Y (s) 108(s + 3)
T (s) = = .
R(s) (s + 9)(s 2 + 8s + 36)

26/41
MECE 3350 - C. Rossa 26 / 41 Lecture 9
Exercise 42 - continued
(c) Overshoot and settling time considering the dominant poles.

Y (s) 108(s + 3)
T (s) = = .
R(s) (s + 9)(s 2 + 8s + 36)

27/41
MECE 3350 - C. Rossa 27 / 41 Lecture 9
Exercise 42 - continued
T = tf([108 324],[1 17 108 324]);
step(T); stepinfo(T)
H = tf([108/9 324/9],[1 8 36]);
step(H); stepinfo(H)

2
y

0
0 time (seconds) 2

28/41
MECE 3350 - C. Rossa 28 / 41 Lecture 9
Exercise 43

Consider the following closed loop system

Where τ can take the values τ = 0, 0.05, 0.1 or 0.5. For r (t) = 1:

→ (a) Record the percent overshoot, rise time, and settling time as τ varies.
→ (b) Describe the effects of varying τ .
→ (c) Compare the location of the zero with that of the closed-loop poles.

29/41
MECE 3350 - C. Rossa 29 / 41 Lecture 9
Exercise 43 - continued
The closed loop transfer function

30/41
MECE 3350 - C. Rossa 30 / 41 Lecture 9
Exercise 43 - continued

5440(τ s + 1)
T (s) =
s 3 + 28s 2 + (432 + 5440τ )s + 5440
Matlab commands:
H = tf([5440*t 5400],[1 28 432+5440*t 5440]);
infostep(H)
damp(H)

τ Tr Ts P.O. zero pole

0.05

0.1

0.5

31/41
MECE 3350 - C. Rossa 31 / 41 Lecture 9
Exercise 43 - continued
t = 0;
H1 = tf([5440*t 5400],[1 28 432+5440*t 5440]);
step(H1);

1.5

1
y

0.5

0
0 time (seconds) 1.5

32/41
MECE 3350 - C. Rossa 32 / 41 Lecture 9
Exercise 44

The roll control of an aircraft is shown. The goal is to select a suitable K so

that the response to a step command r (t) = A will provide a fast response with
an overshoot of less than 20%.

Steps for designing the controller:

→ (a) Determine the closed-loop transfer function
→ (b) Determine the poles for K = 0.7, 3, and 6;
→ (c) Using the concept of dominant poles find the expected overshoot
→ (d) Plot the actual response with Matlab and compare it with (c)

33/41
MECE 3350 - C. Rossa 33 / 41 Lecture 9
Exercise 44 - continued
(a) The closed-loop transfer function

34/41
MECE 3350 - C. Rossa 34 / 41 Lecture 9
Exercise 44 - continued
(b) Finding the poles

12k 12k
T (s) = = 3 (13)
s(s + 3)(s + 7) + 12k s + 10s 2 + 21s + 12k

35/41
MECE 3350 - C. Rossa 35 / 41 Lecture 9
Exercise 44 - continued
(c) Overshoot considering the dominant poles (k =0.7, 3, and 6).

12k 12k
T (s) = = 3 (14)
s(s + 3)(s + 7) + 12k s + 10s 2 + 21s + 12k

36/41
MECE 3350 - C. Rossa 36 / 41 Lecture 9
Exercise 44 - continued

(d) Step-unit response using Matlab

37/41
MECE 3350 - C. Rossa 37 / 41 Lecture 9
Exercise 44 - continued
(c) Overshoot considering the dominant poles (k =0.7, 3, and 6).

12k 12k
T (s) = = 3 (15)
s(s + 3)(s + 7) + 12k s + 10s 2 + 21s + 12k

38/41
MECE 3350 - C. Rossa 38 / 41 Lecture 9
Skills check 25 - From 2018 midterm examination

True or false (1 mark)? The complex poles of P(s) will dominate its response
to a unit step input. Justify your answer (4 marks).

150
P(s) = (16)
(s 2 + 3s + 10)(s + 15)

For answer see last slide 39/41

MECE 3350 - C. Rossa 39 / 41 Lecture 9
Skills check 26 - From 2018 midterm examination

Based on the given unit step response h(t) of the function H(s) shown below,
what temporal signal best describes the unit response of G(s)?

10 10(s + 1)
H(s) = , G(s) = 2
s 2 + 2s + 10 s + 2s + 10

(a) g1 (t)
(b) g2 (t)
(c) g3 (t)
(d) g(t) ≈ h(t)
(e) None of the above

40/41
MECE 3350 - C. Rossa 40 / 41 Lecture 9
Next class...
• Stability
• Solution to skills check: 25 - True, 26 - (b) (why not (a)?).

41/41
MECE 3350 - C. Rossa 41 / 41 Lecture 9