Section 3 Stability PDF
Section 3 Stability PDF
3:
STABILITY
MAE 4421 – Control of Aerospace
& Mechanical Systems
2 Introduction
K. Webb MAE 4421
Stability
3
Consider the following 2nd‐order systems
and
has two real poles:
and
has a complex‐conjugate pair of poles:
,
The step response of each system is:
K. Webb MAE 4421
Stability
4
Both step responses are a superposition of:
Natural response (transient)
Driven or forced response (steady‐state)
Natural Response Driven Response
1.5 2.5 1
cos 2 sin 2 1
In both cases, the natural response decays to zero
as
K. Webb MAE 4421
Stability
5
Both step responses are characteristic of stable systems
K. Webb MAE 4421
Stability
6
Now, consider the following similar‐looking systems:
and
has two real poles
and
has a complex‐conjugate pair of poles
,
The step responses of these systems are:
K. Webb MAE 4421
Stability
7
Again, step responses consist of a natural response
component and a driven component
Natural Response Driven Response
1.5 2.5 1
cos 2 sin 2 1
Now, as , the natural responses do not decay
to zero
They blow up – why?
Exponential terms are positive
K. Webb MAE 4421
Stability
8
Step responses characteristic of unstable systems
K. Webb MAE 4421
Stability
9
Why are the exponential terms positive?
Determined by the system poles
For the over‐damped system, the poles are
and
And, the step response is
For the under‐damped system, the poles are
,
The step response is
K. Webb MAE 4421
Stability and System Poles
10
Sign of the exponentials determined by , the real
part of the system poles
If
Pole is in the left half‐plane (LHP)
Natural response as
System is stable
If
Pole is in the right half‐plane (RHP)
Natural response as
System is unstable
K. Webb MAE 4421
Purely‐Imaginary Poles
11
Two purely‐imaginary poles
,
K. Webb MAE 4421
Marginal Stability
12
Step response for this undamped system is
Natural Response Driven Response
Natural response neither decays to zero, nor grows
without bound
Oscillates indefinitely
System is marginally stable
K. Webb MAE 4421
Marginal Stability
13
Step response is characteristic of a marginally‐stable
system
K. Webb MAE 4421
Repeated Imaginary Poles
14
We’ll look at one more interesting case before presenting a
formal definition for stability
Consider the following system
16 16
8 16 4
Repeated poles on the imaginary axis
, 2 and , 2
The step response for this system is
Natural Response Driven Response
cos 2 ⋅ sin 2 1
K. Webb MAE 4421
Repeated Imaginary Poles
15
Multiplying time factor causes the natural response
to grow without bound
An unstable system
Results from repeated poles
Multiple identical poles on the imaginary axis
implies an unstable system
K. Webb MAE 4421
Repeated Imaginary Poles
16
Step response shows that the system is unstable
K. Webb MAE 4421
17 Definitions of Stability
K. Webb MAE 4421
Definitions of Stability – Natural Response
18
We know that system response is the sum of a natural
response and a driven response
Can define the categories of stability based on the
natural response:
Stable
A system is stable if its natural response → 0 as → ∞
Unstable
A system is unstable if its natural response → ∞ as → ∞
Marginally Stable
A system is marginally stable if its natural response neither
decays nor grows, but remains constant or oscillates
K. Webb MAE 4421
BIBO Stability
19
Alternatively, we can define stability based on the
total response
Bounded‐input, bounded‐output (BIBO) stability
Stable
A system is stable if every bounded input yields a
bounded output
Unstable
A system is unstable if any bounded input yields an
unbounded output
K. Webb MAE 4421
Closed‐Loop Poles and Stability
20
Stable
A stable system has all of its closed‐loop poles in the
left‐half plane
Unstable
An unstable system has at least one pole in the right
half‐plane and/or repeated poles on the imaginary axis
Marginally Stable
A marginally‐stable system has non‐repeated poles on
the imaginary axis and (possibly) poles in the left half‐
plane
K. Webb MAE 4421
21 Determining System Stability
K. Webb MAE 4421
Determining Stability
22
Stability determined by pole locations
Poles determined by the characteristic polynomial, Δ
Factoring the characteristic polynomial will always tell
us if a system is stable or not
Easily done with a computer or calculator
Would like to be able to detect RHP poles without a
computer
Form of Δ may indicate RHP poles directly, or
Routh‐Hurwitz Criterion
K. Webb MAE 4421
Stability from Coefficients
23
A stable system has all poles in the LHP
⋯
Poles:
For all LHP poles, 0, ∀
Result is that all coefficients of Δ are positive
If any coefficient of Δ is negative, there is at least one
RHP pole, and the system is unstable
If any coefficient of Δ is zero, the system is unstable or, at
best, marginally stable
If all coefficients of Δ are positive, the system may be
stable or may be unstable
K. Webb MAE 4421
Routh‐Hurwitz Criterion
24
Need a method to detect RHP poles if all
coefficients of are positive:
Routh‐Hurwitz criterion
General procedure:
1. Generate a Routh table using the characteristic
polynomial of the closed‐loop system
2. Apply the Routh‐Hurwitz criterion to interpret the
table and determine the number (not locations) of RHP
poles
K. Webb MAE 4421
Routh‐Hurwitz – Utility
25
Routh‐Hurwitz was very useful for determining stability
in the days before computers
Factoring polynomials by hand is difficult
Still useful for design, e.g.:
6 8
Stable for some range of gain, , but unstable beyond
that range
Routh‐Hurwitz allows us to determine that range
K. Webb MAE 4421
Routh Table
26
Consider a 4th‐order closed‐loop transfer function:
Routh table has one row for each power of in
First row contains coefficients of even powers of (odd if
the order of Δ is odd)
Second row contains coefficients of odd (even) powers of
Fill in zeros if needed – if even order
K. Webb MAE 4421
Routh Table
27
Remaining table entries calculated using entries from
two preceding rows as follows:
0
0
0 0
0
0 0
0 0
0 0
0 0
0 0 0
0 0
K. Webb MAE 4421
Routh Table – Example
28
Consider the following feedback system
The closed‐loop transfer function is
5000
20 124 5240
The first two rows of the Routh table are
1 124
20 1 5240 262
Note that we can simplify by scaling an entire row by any factor
K. Webb MAE 4421
Routh Table – Example
29
Calculate the remaining table entries:
1 124
20 1 5240 262
1 124 1 0
1 262 138 1 0 0
1 1
1 262 1 0
138 0 262 138 0 0
138 1
How do we interpret this table?
Routh‐Hurwitz criterion
K. Webb MAE 4421
Routh‐Hurwitz Criterion
30
Routh‐Hurwitz Criterion
The number of poles in the RHP is equal to the number
of sign changes in the first column of the Routh table
Apply this criterion to our example:
1 124
1 262
138 0
262 0
Two sign changes in the first column indicate two
RHP poles system is unstable
K. Webb MAE 4421
Routh‐Hurwitz – Stability Requirements
31
Consider the same system, where controller gain is left
as a parameter
Closed‐loop transfer function:
100
20 124 240 100
Plant itself is stable
Presumably there is some range of gain, , for which the
closed‐loop system is also stable
Use Routh‐Hurwitz to determine this range
K. Webb MAE 4421
Routh‐Hurwitz – Stability Requirements
32
100
20 124 240 100
Create the Routh table
1 124
20 1 240 100 12 5
1 124 1 0
1 12 5 112 5 1 0 0
1 1
1 12 5 1 0
112 5 0 12 5 138 0 0
112 5 1
K. Webb MAE 4421
Routh‐Hurwitz – Stability Requirements
33
1 124
1 12 5
112 5 0
12 5 0
Since 0, only the third element in the first column can
be negative
Stable for
112 5 0
22.4
Unstable (two RHP poles) for
112 5 0
22.4
K. Webb MAE 4421
Routh Table – Special Cases
34
Two special cases can arise when creating a Routh
table:
1. A zero in only the first column of a row
Divide‐by‐zero problem when forming the next row
2. An entire row of zeros
Indicates the presence of pairs of poles that are mirrored
about the imaginary axis
We’ll next look at methods for dealing with each of
these scenarios
K. Webb MAE 4421
Routh Table – Zero in the First Column
35
If a zero appears in the first column
1. Replace the zero with
2. Complete the Routh table as usual
3. Take the limit as → 0
4. Evaluate the sign of the first‐column entries
For example:
10
3 2 6 6 9
First two rows in the Routh table:
1 2 6
3 1 6 2 9 3
K. Webb MAE 4421
First‐Column Zero – Example
36
1 2 6
1 2 3
1 2 1 6 1 0
1 2 0 1 3 3 1 0 0
1 1 1
Replace the first‐column zero with and proceed as usual
1 2 1 3 1 0
3 2 3 0 0
3 0
3 0 0
2 3 2 3 2 3
3 3 0 0
3 0 0
2 3 2 3 2 3 2 3
Continuing on the next page …
K. Webb MAE 4421
First‐Column Zero – Example
37
1 2 6
1 2 3
3 0
2 3
3 0
3
3 0 0
2 3
2 3 2 3 2 3
3 0 0
3 3 3
3 0 3 0 3 0
2 3 2 3 2 3
3 0 0
3 3 3
3 3 3
2 3 2 3 2 3
Next, take the limit as → 0
K. Webb MAE 4421
First‐Column Zero – Example
38
Taking the limit as → 0 and looking at the first column:
1 1
1 1
0
2 3
lim
→
∞
3
3 0
2 3
3 3
Two sign changes
Two RHP poles
System is unstable
K. Webb MAE 4421
Routh Table – Row of Zeros
39
A whole row of zeros indicates the presence of pairs of
poles that are mirrored about the imaginary axis:
At best, the system is marginally stable
Use a Routh table to determine if it is unstable
K. Webb MAE 4421
Routh Table – Row of Zeros
40
If an entire row of zeros appears in a Routh table
1. Create an auxiliary polynomial from the row above
the row of zeros, skipping every other power of
2. Differentiate the auxiliary polynomial w.r.t.
3. Replace the zero row with the coefficients of the
resulting polynomial
4. Complete the Routh table as usual
5. Evaluate the sign of the first‐column entries
K. Webb MAE 4421
Row of Zeros – Example
41
Consider the following system
1
5 11 23 28 12
The first few rows of the Routh table:
1 11 28
5 23 12
1 11 1 28 1 0
5 23 6.4 1 5 12 25.6 4 5 0 0
5 5 5
5 23 5 12 5 0
1 4 3 1 1 0 12 4 1 0 0
1 1 1
Continuing on the next page …
K. Webb MAE 4421
Row of Zeros – Example
42
1 11 28
5 23 12
1 4 0
1 4 0
1 4 1 4 5 0
1 4 0 1 4 0 1 0 0
1 1 1
A row of zeros has appeared
Create an auxiliary polynomial from the row
4
Differentiate
Replacing the row with the coefficients of
1 11 28
5 23 12
1 4 0
1 4 0
0 2 0 0
1 4 1 0 1 0
2 0 4 2 0 0 2 0 0
2 2 2
No sign changes, so RHP poles, but
Row of zeros indicates that system is marginally stable
K. Webb MAE 4421
Stability Evaluation – Summary
44
If coefficients of have different signs
System is unstable
If some coefficients of are zero
System is, at best, marginally stable
If all coefficients have the same sign
System may be stable or unstable
Generate a Routh table and apply Routh‐Hurwitz criterion
Replace any zero first‐column entries with and let take the
limit as → 0
Replace a row of zeros with coefficients from the derivative
of the auxiliary polynomial
If no RHP poles are detected, the system is marginally stable
K. Webb MAE 4421