Stability of Feedback Systems

1
 Stability of Feedback Systems
• Most industrial processes are stable without feedback control.
Thus, they are said to be open-loop stable or self-regulating.
• An open-loop stable process will return to the original steady
state after a transient disturbance (one that is not sustained)
occurs.
• By contrast there are a few processes, such as exothermic
chemical reactors, that can be open-loop unstable.

Definition of Stability: An unconstrained linear system is

said to be stable if the output response is bounded for all
bounded inputs (BIBO). Otherwise it is said to be
unstable.

2
 Stability of Feedback Systems
Example 11.4
Consider the general feedback control system shown in Fig. 11.8
with the following transfer functions:
1
Gc = Kc Gv = (11-73)
2s + 1
1 1
G p = Gd = Gm = (11-74)
5s + 1 s +1

The closed-loop TF due to a unit step in the setpoint

K mGcGv G p K c (s + 1) 1
Y = =
1 + GcGv G pG m 10s 3 + 17s 2 + 8s + 1 + K c s
Show that the closed-loop system produces unstable responses if
controller gain Kc is too large.
For what values of Kc, I, D the closed-loop will be stable? 3
 Stability of Feedback Systems
As a starting point for the stability analysis, consider the block
diagram in Fig. 11.8. Using block diagram algebra that was
developed earlier, we obtain
K mGcGvG p Gd
Y= Ysp + D (11-80)
1 + GOL 1 + GOL
where GOL is the open-loop transfer function, GOL = GcGvGpGm.
For the moment consider set-point changes only, in which case
Eq. 11-80 reduces to the closed-loop transfer function,
Y K mGcGvG p
= (11-81)
Ysp 1 + GOL
This equation is a rational function that can be factored into poles
(pi) and zeroes (zi) as:
This form of TF is called the zero-pole (zp) form

Y (s − z 1 )(s − z 2 ) (s − z m )
=K ' , where n  m (11 − 82)
Y sp (s − p1 )(s − p 2 ) (s − p n )
4
 Stability of Feedback Systems

Assuming a step in Ysp and solving the above equation we

get:
y (t) = A0 + A1e p1t + A 2e p 2t + + A n e p nt

• If any one of the poles in the previous equation is positive,

then y(t) will be unbounded (unstable).
• If any pole is complex number with a positive real part,
the system will also be unstable.
• However, if all poles have negative real part, the system
will be stable.

5
Comparing Eqs. 11-81 and 11-82 indicates that the poles are also
the roots of the following equation, which is referred to as the
characteristic equation of the closed-loop system:
1 + GOL = 0 (11-83)
General Stability Criterion. The feedback control system in Fig.
11.8 is stable if and only if all roots of the characteristic
equation ( poles of TF ) are negative or have negative real
parts. Otherwise, the system is unstable. Imaginary part

Figure 11.25 Stability regions in the

complex plane for roots of the Stable Unstable
Real part
Region 0 Region
characteristic equation.

6
Contributions of (location of roots of) characteristic equation roots to
closed-loop response

7
Example 11.8
Consider a process, Gp = 0.2/(-s + 1), and thus is open-loop unstable.
If Gv = Gm = 1, determine whether a proportional controller can
stabilize the closed-loop system.

Solution
The characteristic equation for this system is
s − 0.2Kc − 1 = 0 (11-92)
Which has the single root, s = 1 + 0.2Kc. Thus, the stability
requirement is that :
p  0 → 0.2K c + 1  0 → K c  −5
This example illustrates the important fact that feedback control
can be used to stabilize a process that is not stable without
control. 8
 Example
Determine the dynamic behavior (i.e., check the poles of
the overall TF) of a PI controller [Kc = 1, τI = 1]
applied to a 2nd order process [Kp = 1, τp = 5, ζ = 2].
  1 + 1
+   
Assume Gv(s)c= 1, G m(s) = 1. D 
C( s ) K E( s ) s
I s
 
Gc( s )
C( s ) 
Kc  1 +
1
+ D s  PI controller
E( s ) I s
 
Y ( s) Kp
G p ( s) = = 2 2
U ( s)  p s + 2 p s + 1
Characteristic Equation:
GOL+1 = 0

Gc(s) Gv(s) Gp(s) Gm(s) + 1 = 0

Imaginary Axis
K   1 + 1   1  
Kp
 c  1 + 1 0
 I s
     p2 s 2 + 2  p  s + ( 1) 
 
Substitute values for every parameter…
Real Axis

1  1 + 1   1  1 1 + 1 0 25s 3 20s 2

     2s 1 0
  1 s  2
 25 s + 20 s + ( 1) 
Dominant root is
Check Poles of CL TF (roots of CE) closest to origin
s = -0.764 Its negative,
Solve for values of s (poles)… hence, the loop is
In Matlab:
s = -0.018 ± 0.228 i stable but oscillatory
roots([25 20 2 1]) Real part is Imaginary part,
negative, so so oscillatory
stable
Routh Stability Criterion
The Routh stability criterion is based on a characteristic equation
that has the form a s n + a s n−1 + + a s + a = 0 (11-93)
n n−1 1 0

Routh array: where:

Row an −1an −2 − an an −3
b1 =
an −1
1 an an-2 an-4 an −1an −4 − an an −5
b2 =
2 an-1 an-3 an-5 an −1

3 b1 b2 b3
4 c1 c2 and:

b1an −3 − an −1b 2
c1 =
n + 1 z1 b1
b1an −5 − an −1b3
c2 =
b1
11
Routh Stability Criterion:
A necessary and sufficient condition for all roots of the
characteristic equation in Eq. 11-93 to have negative real parts is
that all of the elements in the left column of the Routh array are
positive. A necessary condition for stability is that all of the
coefficients in the characteristic equation must be positive.

Example 11.9
Determine the stability of a system that has the characteristic
equation: s 4 + 5s3 + 3s 2 + 1 = 0 (11-98)
Solution
Because the s term is missing, its coefficient is zero. Thus, the
system is unstable.
For a stable system, all coeff. of the characteristic equation must be positive
12
Example ..

To test the stability of a system having a characteristic equation

F(s) = s 3 + 6 s 2 + 12 s + 8 = 0
The Routh Array is constructed as follows..

Column 1 (pivot column) includes no changes of sign and therefore the

roots of the characteristic equation have only real parts and the system
is stable.

13
Example 11.10
Find the values of controller gain Kc that make the feedback
control system of Ex. 11.4 stable. G = K G =
1
(11-73)
c c v
2s + 1
1 1
Solution G p = Gd = Gm = (11-74)
5s + 1 s +1
From Eq. 11-76, the characteristic equation is

10s3 + 17 s 2 + 8s + 1 + Kc = 0 (11-99)

All coefficients are positive provided that 1 + Kc > 0 or Kc < -1.

The Routh array is
17  8 − 10(1 + K c )
10 8 b1 = 0
17
17 1 + Kc
K c  12.6
b1 b2
c1
14
To have a stable system, each element in the left column of the
Routh array must be positive. Element b1 will be positive if
Kc < 7.41/0.588 = 12.6. Similarly, c1 will be positive if Kc > -1.
Thus, we conclude that the system will be stable if

−1  Kc  12.6 (11-100)

This example illustrates that we can use the Routh array

to determine the stability limits of controller parameters

15
Characteristic Equ. with a Delay Element
If there is a delay element (e-θs) in the transfer function, we
can not apply the Routh test directly. The time delay can be
approximated using Pade approximation:
−s 1 − 2 s
1/1 Pade approximation: G1 (s ) = e 
1 + 2 s
1 − 2 s + 12s
2 2

2/2 Pade approximation: G 2 (s ) = e −s 

1 + 2 s + 12s
2 2

See example 11.11

16
 Direct Substitution Method
• The imaginary axis divides the complex plane into stable and
unstable regions for the roots of characteristic equation, as
indicated in Fig. 11.26.
• On the imaginary axis, the real part of s is zero, and thus we can
write s=iw. Substituting s=iw into the characteristic equation
allows us to find a stability limit such as the maximum value of
Kc.
• As the gain Kc is increased, the roots of the characteristic
equation cross the imaginary axis when Kc = Kcm.

17
Example 11.12
Use the direct substitution method to determine Kcm for the system
with the characteristic equation given by Eq. 11-99.

Solution 10s3 + 17 s 2 + 8s + 1 + Kc = 0 (11-99)

Substitute s = iω and Kc = Kcm into Eq. 11-99:

−10iω3 − 17ω2 + 8iω + 1 + Kcm = 0

or (11-105)
(1 + Kcm − 17ω2 ) + i (8ω − 10ω3 ) = 0
Both parts should be zero

18
Equation 11-105 is satisfied if both the real and imaginary parts
are identically zero:

1 + Kcm − 17ω2 = 0 (11-106a)

(
8ω − 10ω3 = ω 8 − 10ω2 = 0) (11-106b)
Therefore,
ω2 = 0.8  ω = 0.894 (11-107)
and from (11-106a),
Kcm = 12.6

We conclude that the stability limit for controller gain is 12.6. At

this value, continuous oscillation with a frequency of 0.894
rad/min occurs. The corresponding period of oscillation will be
P=2π/0.894=7.03 min.
19
 Root Locus Diagrams
• Root locus analysis is a graphical method for examining how
the roots of a system change with variation of a certain system
parameter, commonly a gain within a feedback system.

• This is a technique used to determine stability of a control the

system.

• The root locus plots the poles of the closed loop transfer
function as a function of a gain parameter.

20
Root Locus Diagrams
Example 11.13
Consider a feedback control system that has the open-loop
transfer function,
4 Kc
GOL ( s ) = (11-108)
( s + 1)( s + 2 )( s + 3)
Plot the root locus diagram for 0  Kc  20.

Solution
The characteristic equation is 1 + GOL = 0 or

( s + 1)( s + 2)( s + 3) + 4Kc = 0 (11-109)

21
• The root locus diagram in Fig. 11.27 shows how the three roots
of this characteristic equation vary with Kc.
• When Kc = 0, the roots of the CE are merely the poles of the
open-loop transfer function, -1, -2, and -3.
Kc p1 p2 p3
0 -1 -2 -3
0.1 -1.42+ 0.093i -1.42- 0.093i -3.159
1 -1.1+ 1.192i -1.102- 1.192i -3.796
5 -0.58+ 2.244i -0.581- 2.244i -4.837
15 0+ 3.317i 0- 3.317i -6.000
20 0.193+ 3.665i 0.193- 3.665i -6.386
p1
p3

p2

Figure 11.27
Root locus diagram for third-order system. (*) denotes an open-loop pole. Dots denote locations of the
closed-loop poles for different values of Kc. Arrows indicate change of pole locations as Kc increases.

22
Example:
If we have only a proportional controller (i.e., one design parameter) and
real negative open-loop poles. The closed-loop characteristic equation is
given as::
 4 
1 + K c   = 0
 ( s + 3) ( s + 2 ) ( s + 1) 

We can easily use MATLAB to find that the ultimate gain is roughly 15. The
statements to use are:
s=tf('s')
G=(4/((s+1)*(s+2)*(s+3)))
k=0:0.1:100; % We have to use our own gain vector in this example
rlocus(G,k) % because the MATLAB default plot does not cross the Im axis
rlocfind(G)

23
Example:

Kcm=15

After entering the rlocfind() command, MATLAB will prompt us to click a point
on the root locus plot. In this problem, we select the intersection between the
root locus and the imaginary axis for the ultimate gain.

24
Routh Test using Matlab

K=1:2000; % specify the range of K

for m=1:length(K) % number of K values to test
denH=[1 18 77 K(m)]; % specify the denominator of H(s)
poles=roots(denH); % evaluate poles
r=real(poles); % vector containing real values
if max(r)>=0 % test poles for real parts
poles % display poles with real part positive
K=K(m) % display the corresponding value of K
break % stop loop if rhp poles are found
end % end if
end % end for
poles =
-18.0025
0.0012 + 8.7775i
Results 0.0012 - 8.7775i

K= 1387
25
 Nonlinear
process model

No Apply stability tests

Linearize? for nonlinear
systems

Yes

Obtain transfer Figure 11.29. Flowchart

function model
for performing a stability
analysis.
Model contains
No
time delays?

Yes

Approximate Apply Routh

e-s? stability criterion

No

Substitute s = jw or
apply frequency
stability criterion

26
Assignment 4

From your textbook , solve the following problems:

P-11.4
P-11.10
P-11.15
P-11.16

27