Chapter 5 Solutions

5.1 Prove that for a system with two real poles and a real zero For the transfer function

sa
G( s)  , a  p1  p2 or p1  p2  a
( s  p1 )( s  p2 )

the breakaway point is at a distance from the zero.

For the transfer function

sa
G( s)  , a  p1  p2 or p1  p2  a
( s  p1 )( s  p2 )

The breakaway and break-in points are obtained by minimizing the gain

(  p1 )(  p2 )
K
 a
A necessary condition for a minimum or maximum is that the first derivative with respect to  be
zero. The first derivative is

dK

 (  p1 )  (  p2 ) (  a)  (  p1 )(  p2 )
d    a 2
2 2  ( p1  p2  2 a)  a( p1  p2 )   2  ( p1  p2 )  p1 p2

   a 2
 2  2 a  a( p1  p2 )  p1 p2

   a 2
Equating to zero and solving for  gives

 b  a  a 2  a( p1  p2 )  p1 p2

 a  ( a  p1 )( a  p2 )

For a < p1 < p2 or p1 < p2 <a, there are two real solutions for b . One solution corresponds to a
breakaway point since it lies between the two poles and will yield a positive gain K. The second is
a break-in point that corresponds to a negative gain. For a outside the specified range, the root
locus is entirely on the real axis with no breakaway or break-in points. The breakaway point is
where K is at a maximum and the break-in point is where K is at a minimum.

5.2 Use the result of Problem 5.1 to draw the root locus of the system The transfer function

K ( s  4)
KG ( s) 
s( s  2 )

1
 The transfer function

K ( s  4)
KG ( s) 
s( s  2 )

has a breakaway point at

 b  4  ( 4  0)( 4  2)

 11716
.

The second solution obtained using the results of Problem 5.1 is not between the two open-loop
poles (b =-6.8284) and corresponds to the break-in point associated with a negative gain.

The root locus off the real line is a circle centered at the zero (-4) with radius (8) 1/2
The root locus is shown in Figure 5 -1.

1
Imag Axis

-1

-2

-3

-4
-10 -8 -6 -4 -2 0 2
Real Axis
Figure 5-1 Root locus for Problem 5.2.

5.3 Sketch the root locus of the following systems:

K K ( s + 2)
( i ) KG ( s)  (ii) KG ( s) =
s( s  2 )( s  5) s( s + 3)( s + 5)
K K ( s + 2)
( i ) KG ( s)  (ii) KG ( s) =
s( s  2 )( s  5) s( s + 3)( s + 5)

2
 6

Imag Axis 0

-2

-4

-6
-8 -6 -4 -2 0 2
Real Axis

Figure 5-2 Root locus for Problem 5.3 (i)

1
Imag Axis

-1

-2

-3

-4

-5
-8 -6 -4 -2 0 2
Real Axis

Figure 5-3 Root locus for Problem 5.3 (ii)

5.4 Consider the system in 5.3(ii) with a required steady-state error of 20%, and an adjustable PI
controller zero location. Show that the corresponding closed-loop characteristic equation is given
by

sa 1
1 K  0
 s  ( s  3)( s  5)

Next, rewrite the equation as

1  K f G f ( s)  0

3
where Kf = K, K z is constant and Gf(s) is a function of s, and examine the effect of shifting the
zero on the closed-loop poles.

a) Design the system for a dominant second order pair with a damping ratio of 0.5. What is n
for this design?
K ( s + a)
For the transfer function KG ( s ) =
s ( s + 3)(s + 5)

For the required steady-state error of 20%, Kv = 100/20 = 5

Lim Ka
Hence K v  s  0 sG ( s )  5
15

i.e. K a = 5  15 = 75

The closed-loop characteristic equation with unity feedback and cascade PD control is

Ks  75
1  KG ( s) C( s)  1  0
s( s  3)( s  5)

Multiply by the denominator of the loop gain then divide by all terms that do not include K to
obtain

Ks
1 0
s  8s  15s  75
3 2

where K a =75 is fixed. To examine the effect of shifting the zero on the closed-loop poles, we
obtain the root locus for the new transfer function using the MATLAB command rlocus and obtain
the desired design suing the cursor.

For a dominant second order pair with a damping ratio of 0.5, w n =5.04 rad/s, K = 25.3,
a =75/25.3 = 2.96.

Root Locus
10

7 System: g
Gain: 25.3
6 Pole: -2.52 + 4.37i
Damping: 0.5
Imaginary Axis

Overshoot (%): 16.3 5

5 Frequency (rad/sec): 5.04
0.5
4

-1
-8 -7 -6 -5 -4 -3 -2 -1 0
Real Axis

Figure 5-4 Part (a) Root locus for K a = 75.

4
b) Obtain the time response using a CAD program. How does the time response compare with that
of a second order systems with the same n and z as the dominant pair? Give reasons for the
differences.

The time response of the system is shown below together with the response of a second order
systems with the same n and z as the dominant pair. For the second order system, the percentage
overshoot is given by

1  2
PO  100 e  / %  16.3%

Step Response

0.8
Amplitude

0.6

0.4

0.2

0
0 0.5 1 1.5 2 2.5
Time (sec)

Figure 5-5 Time response for K z = 75 and a zero at 2.96.

Our system behaves almost exactly like the second order underdamped pair because it has a zero at
2.96 very close to a third pole at 3. So it has PO = 16.3 %. The system has time to first peak
is
 
Tp    0.72s
d 4.37

c) Discuss briefly the tradeoff between error, speed of response and relative stability in this problem.

The root locus of the system of Figure 5 -6 for a zero fixed at 2.96 shows that increasing the gain
of the system to improve its steady-state error would result in a decrease in the damping ratio.
Increasing the gain would speed up the system but would decrease its relative stability and the PO
will be unacceptably high. For example, if the gain is increased to 40 we obtain the time response
of Figure 5 -7 with a PO of 25.4%. Thus the choice of 20% steady-state error due to ramp is a
compromise that yield an acceptable transient response.

5
 Root Locus
8 System: g
Gain: 40
Pole: -2.52 + 5.81i
7 Damping: 0.399
Overshoot (%): 25.5
Frequency (rad/sec): 6.33
6

Imaginary Axis
4

-1
-4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0
Real Axis

Figure 5-6 Root locus for K = 40 with a zero at 2. 7.

Step Response
1.4

1.2 System: gc
Peak amplitude: 1.25
Overshoot (%): 25.4
At time (sec): 0.547
1

0.8
Amplitude

0.6

0.4

0.2

0
0 0.5 1 1.5 2 2.5
Time (sec)

Figure 5-7 Time response for K = 40 with a zero at 2.7.

5.5 Prove equations (5.13) and (5.14) and justify the design Procedures 5.1, 5.2.

By the angle condition, the sum of the angles of all the blocks in the loop must be 180 at any
closed-loop pole. If the compensator angle is c and the loop gain (before compensation) is L(s), the
angle conditon gives

 C  180    L s cl  (5.13)

Equation (5.14) follows from the geometry of Figure 5 -8

6
 d
a   n (5.14)
tan  C 
j
scl

d

c =  (scl + a) 

 n
a

Figure 5-8 Zero location for a PD controller.

5.6 Show that a PI feedback controller is undesirable since it results in a differentiator in the forward
path. Discuss the step response of the closed-loop system.

Let the feedforward transfer function be G(s) and the feedback controller transfer function be

sa
C ( s)  K
s
The closed-loop transfer function of the system is

G(s)
T (s) 
1  C ( s )G ( s )
G(s)

sa
1 K G(s)
s
sG ( s )

s  K  s  a  G(s)

The corresponding steady-state error is

e ( )  Lim
s 0
1  T (s) sR(s)
 sG ( s) 
 Lim
s 0 
1 sR ( s )
 s  K  s  a  G ( s) 
 LimsR( s)
s 0

Thus the system has a large error due to step and cannot follow a ramp or a parabolic.

7
5.7 Design a controller for the transfer function

1
G(s) 
( s  1)( s  5)

to obtain (i) zero steady-state error due to step, (ii) a settling time of less than 2 s, and (iii) an
undamped natural frequency of 5 rad/s. Obtain the response due to a unit step and find the
percentage overshoot, the time to the first peak and steady-state error percent due to a ramp input.
1
The plant transfer function is G ( s) 
( s  1)( s  5)

For zero steady-state error due to step, a PI controller is needed. The simplest design is by
cancellation of the pole at 1 with the controller zero. This leaves the loop gain

1
L( s)  C( s) G ( s) 
s( s  5)

The corresponding root locus includes a vertical line at 2.5 i.e. for sufficiently high gain we have
an underdamped system with  n = 2.5 and Ts = 4/2.5 < 2. For  = 0.5, n = 2.5/0.5 = 5 rad/s.
From the symmetry of the root locus, the gain K at the desired closed-loop pole is equal to the
square of the distance from the location to the origin i.e. K = 52 = 25. The same results can be
obtained using a CAD tool or analytically, since the compensated system is simple.

The design yields a second order system and the remaining design criteria can be computed
analytically as follows

1  2
PO  100e  %  16.3%
 
Tp    0.726 s
d 5 1  (.5) 2
25 100
Kv  sC( s)G ( s) s  0   5  e( )%   20%
s  5 s 0 Kv

Root Locus
5

4 System: l
Gain: 24.9
Pole: -2.5 + 4.32i
Damping: 0.501
Overshoot (%): 16.2
3 Frequency (rad/sec): 4.99
Imaginary Axis

-1
-5 -4 -3 -2 -1 0 1
Real Axis

Figure 5-9 Root locus for problem 5.7

5.8 Repeat Problem 5.7 with a required settling time less than 0.5 s and an undamped natural frequency
of 10 rad/s.

8
 1
The plant transfer function is G ( s) 
( s  1)( s  5)

For zero steady-state error due to step, a PI controller is needed but the transient response
specifications can not be met with pole-zero cancellation. For proportional control, the root locus
includes a vertical line at 3 which yields a settling time of 4/3 > 0.5 s. Hence, PI control cannot
meet the design specifications and PID control is needed.

We begin the design by selecting a PD controller using a CAD tool. We choose  =0.8 and n = 10
rad/s which give a settling time of 0.5 s. We obtain the values K= 10 and z = 9.5. Next, we add a
PI controller with a zero at  n/10 = 0.8. The PID controller is now of the form
( s  0.8)( s  9.5)
C ( s)  K
s
Using a CAD tool, we obtain a gain K = 11.8 for  = 0.8. The corresponding time response is good
with PO of 6%, a time to first peak of 0.27 s, and a settling time of about 0.44 s.

Root Locus
7

6 System: l
Gain: 11.8
Pole: -8.49 + 6.37i
5 Damping: 0.8
Overshoot (%): 1.52
Frequency (rad/sec): 10.6
4
Imaginary Axis

-1
-20 -18 -16 -14 -12 -10 -8 -6 -4 -2 0 2
Real Axis

Figure 5-10 Root locus for the PID controlled system of Problem 5.8

9
 Step Response
1.4

System: untitled1
Peak amplitude: 1.06
1.2 Overshoot (%): 6.06
At time (sec): 0.274 System: untitled1
Settling Time (sec): 0.441
1

0.8
Amplitude

0.6

0.4

0.2

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Time (sec)

Figure 5-11 Time response for the PID controlled system of Problem 5.8

5.9 Consider the oven temperature control system of Example 3.5 with transfer function

K
G( s) 
s  3s  1
2

a) Design a proportional controller for the system to obtain a percentage overshoot less than 5 %

The percentage overshoot specification yields

ln(.05)
  2
 .69
ln(.05)   2

We select a damping ratio of 0.7. We can solve the problem analytically since the system is second
order. It is more convenient to use MATLAB to obtain the gain value. We obtain the root locus plot
of Figure 5 -13 for the system and observe that for a gain of about 3.5 we have the desired damping
ratio.

10
 Root Locus
2
System: g
Gain: 3.55
Pole: -1.5 + 1.52i
Damping: 0.704
Overshoot (%): 4.46
Frequency (rad/sec): 2.13
1.5

1
Imaginary Axis

0.5

-0.5

-1
-3 -2.5 -2 -1.5 -1 -0.5 0 0.5
Real Axis

Figure 5-12 Root locus for the oven temperature control system.

b) Design a controller for the system to reduce the steady-state error due to step to zero without
significant deterioration in the transient response.

We need a PI controller to reduce the steady-state error to zero. The proportional control design yields a
pole with real part equal to 1.5. The PI controller zero is at one tenth this value, which is approximately
0.1 and the controller transfer function is

s  0. 1
C (s)  K
s

For a gain of 3.5, we have a dominant pair of poles with about the same damping ratio as for proportional
control. However, the step response for the system is sluggish and the settling time is very large. We
increase the gain to K = 7 and obtain the step response of Figure 5 -14. Although the 2% settling time is
large (over 17s), the 5 % settling time is about half this value and the response is acceptable with an
overshoot less than 4 %.

11
 Step Response
1.4

System: gcl
1.2 Peak amplitude: 1.04
Overshoot (%): 3.73
At time (sec): 1.46 System: gcl
Settling Time (sec): 17.6
1

0.8
Amplitude

0.6

0.4

0.2

0
0 5 10 15 20 25 30 35 40 45
Time (sec)

Figure 5-13 Step response for the system with PI control and a gain of 7.

5.10 For the inertial system governed by the differential equation

  
Design a feedback controller to stabilize the system and reduce the percentage overshoot below
10% with a settling time of less that 4 s.

The transfer function of the system is given by

1
G (s) 
s2

We use PD control to stabilize the system. The feedback control does not result in a closed-loop
zero. The closed-loop system is second-order with the damping ratio constraint

ln(.1)
   .591
2
ln(.1)  2

We choose  = 0.6 and calculate the settling time constraint

4 4
n    1.667
Ts  4  0 .6

12
 We choose n = 2 rad/s. The corresponding desired closed-loop pole is at

s cl  1.2  j1.6

The angle of the compensator is

 C    L s cl   1.287 rad / s

The compensator zero is calculate from

d
a   n  1.6667
tan  C 

The closed-loop transfer function with additional precompensator gain is

4
G(s) 
s  2.4 s  4
2

For this system, analytical design is also simple since the closed-loop chacacteristic equation is

s 2  K  s  a   s 2  2 n s   n2  s 2  2.4s  4

The step response of the system shown in Figure 5 -14 meets the design specifications.

Figure 5-14 Step response for the compensated system of Problem 5.9

Computer Exercises
5.11 Consider the oven temperature control system of Example 3.5 with transfer function

13
 K
G (s) 
s  3s  10
2

a) Obtain the step response of the system with the a PD cascade controller with gain 80 and a zero at
5.
System: gc
Time (sec): 0.236
Step Response
Amplitude: 0.984
1
System: gc
Final Value: 0.889
0.9
System: gc
Settling Time (sec): 0.549
0.8

0.7

0.6
Amplitude

0.5

0.4

0.3

0.2

0.1

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time (sec)

Figure 5-15 Step response of cascade compensated system.

b) Obtain the step response of the system with the a PD feedback controller with a zero at 5 and
unity gain and a forward gain of 80.

14
 System: gf
Step Response
Settling Time (sec): 0.617
0.9
System: gf
0.8 Final Value: 0.889

0.7

0.6

0.5
Amplitude

0.4

0.3

0.2

0.1

0
0 0.2 0.4 0.6 0.8 1 1.2
Time (sec)

Figure 5-16 Step response of feedback compensated system.

c) Why is the root locus identical for both systems?

For both configurations, the loop gain is given by

s5
L( s)  C ( s)G ( s )  16
s  3s  10
2

Hence, the root locus is identical for the two configurations.

d) Why are the time responses different although the systems have the same loop gains?

The time responses are different because the cascade configuration has a closed loop zero at 5 while the
feedback configuration has no closed-loop zero.

e) Complete a comparison table using the responses of (a) and (b) including the percentage
overshoot, the time to first peak, the settling time, and the steady-state error. Comment on the
results and explain the reason for the differences in the response.

The step responses give the values shown in Table P5.1. The percentage overshoot for cascade
compensation is calculated using the equation

y peak  y final 0.984  0.889

%OS   100%   100%  10.7%
y final 0.889

The system with cascade compensation has a faster response due to the closed-lop zero at 5 but the
response is more oscillatory. The two systems have the same steady-state error of 11%.

Table P5.1 Summary of simulation results for cascade and feedback compensation.

15
Configuration %OS Tp Ts e()%
Cascade 10.7% 0.236 0.549 11%
Feedback No overshoot Not defined 0.617 11%

5.12 Use SIMULINK to examine a practical implementation of the cascade controller of Problem 5.11.
The compensator transfer function includes a pole since PD control is only approximately
realizable. The controller transfer is of the form

0.2 s  1
C ( s )  80
0.02 s  1

(a) Simulate the system with a step reference input both with and without a saturation block
with saturation limits 5 between the controller and plant. Export the output to MATLAB
for plotting (you can use a Scope block and select “Save data to workspace”).

0.2s+1 1
80
Step 0.02 s+1 s 2 +3s+10
Subtract Gain Transfer Fcn Saturation Transfer Fcn 1 Sat y

0.2s+1 1
80
Step 1 0.02 s+1 s 2 +3s+10
Subtract 1 Gain 1 y
Transfer Fcn 2 Transfer Fcn 3

Figure 5.17 Simulation diagram for the cascade control system with and without saturation.

(b) Plot the output of the system with and without saturation together and comment on the
difference between the two step responses.

16
 1
No saturation

0.9

0.8

0.7
Saturation

0.6
y

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
time

Figure 5.18 Step response of the cascade control system with and without saturation.

The saturation block reduces the output of the controller resulting in a slower response with a smaller
peak and a much smaller steady-state value.

5.13 Consider the system

1
G(s) 
( s  1) 4

and apply the Ziegler-Nichols procedure to design a PID controller. Obtain the response due to a
unit step input as well as a unit step disturbance signal.

The application of an open-loop unit step input gives the response shown in Figure 5 -19. It can be
seen that K=1 and L=1.3. Since the value of +L is the time interval between the application of the
step input and the time when the process output reaches 63.2% of its final value, we have  =3.
The Ziegler-Nichols rules given in Table 5.1 provide the following PID parameters: Kp=2.77,
Ti=2.6, and Td=0.65. The resulting process output when a step is applied to the set-point signal at
time t=0 and a load disturbance signal is applied at time t=35 is plotted in Figure 5 -20.

Conversely, if we compute the sum  +L as the time interval between the application of the step
input and the intersection of the tangent line with the straight representing the final steady-state
value of the process output, we obtain  +L=6, and therefore  =3. Thus, we can determine the
following PID parameters: Kp=4.34, Ti=2.6, and Td=0.65. The resulting process output when a step
is applied to the set-point signal at time t=0 and to the load disturbance signal at time t=60 is
shown in Figure 5 -21. It appears that in this case the output is more oscillatory than in the
previous case. This actually occurs in general because the estimate of the dominant time constant

17
is usually higher when estimated using the intersection of the tangent line with the steady-state
value of the process output. This implies that the proportional gain estimate is higher than the
estimate obtained by considering the time when the process output reaches 63.2% of its final
value.

0.9

0.8

0.7

0.6
process output

0.5

0.4

0.3

0.2

0.1

0
0 2 4 6 8 10 12 14
time

Figure 5-19 Open-loop step response

1.5

1
process output

0.5

0
0 10 20 30 40 50 60 70
time

Figure 5-20 Set-point and load disturbance step response for Kp=2.77, Ti=2.6, and Td=0.65

18
 1.8

1.6

1.4

1.2

process output
1

0.8

0.6

0.4

0.2

0
0 20 40 60 80 100 120
time

Figure 5-21 Set-point and load disturbance step response for Kp=4.34, Ti=2.6, and Td=0.65

5.14 Write a computer program that implements the estimation of a first-order-plus-dead-time transfer
function with the tangent method and then determines the PID parameters using the Ziegler-Nichols
formula. Apply the program to the system

1
G ( s) 
( s  1) 8

and simulate the response of the control system when a set-point step change and a load
disturbance step are applied. Discuss the choice of the time constant value based on the results.

The following Matlab function that implements the Ziegler-Nichols procedure:

function [Kp,Ti,Td]=zn(Gs)

[y,t]=step(Gs); % step response

tfinal=1.5*t(end); % refinement to reduce numerical problems
t=0:tfinal/1000:tfinal;
[y,t]=step(Gs);

% tangent method
sampling_time=t(2)-t(1);
dy=diff(y);
[diff_max,max_pos]=max(dy);
t_max=t(max_pos);
dy_max=diff_max/sampling_time;
L=t_max-y(max_pos)/dy_max;
Ttop=t_max+(y(end)-y(max_pos))/dy_max;
T=Ttop-L; % the time constant is determined by considering
% the intersection of the tangent line with the
% straight line representing the final
% steady-state value of the process output
% T=t(find(y>0.632*y(end),1))-L;

19
% use the second form of T to determine the time constant
% as the time interval between the application of the
% step input and the time when the process output attains
% the 63.2% of its final value

K=y(end);

% Ziegler-Nichols tuning
Kp=1.2*T/K/L;
Ti=2*L;
Td=0.5*L;

The open-loop step response with the tangent is plotted in Figure 5-22.

0.9

0.8

0.7

0.6
process output

0.5

0.4

0.3

0.2

0.1

0
0 2 4 6 8 10 12 14 16 18 20
time

Figure 5-22Figure Open-loop step response for Problem 5.11

If we run the MATLAB function with the given process transfer function, two options are
available for calculating the time constant. The first is by considering +L as the time interval
between the application of the step input and the intersection of the tangent line with the steady-
state output level. We obtain K=1, =6.7 and L=4.3. Thus, we have Kp=1.87, Ti=8.6, and Td=2.15.
The set-point step response is shown in Figure 5-23, while the load disturbance response is shown
in Figure 5-24Figure 5-23. Both responses are highly oscillatory. This is not surprising as the
system is of high-order and the approximation with a first-order-plus-dead-time model is not
accurate.

20
 Step Response
1.6

1.4

1.2

Amplitude
0.8

0.6

0.4

0.2

0
0 100 200 300 400 500 600
Time (sec)

Figure 5-23 Set-point step response for Problem 5.11 with =6.7

Step Response
0.6

0.5

0.4

0.3
Amplitude

0.2

0.1

-0.1

-0.2
0 100 200 300 400 500 600
Time (sec)

Figure 5-24 Load disturbance step response for Problem 5.11 with =6.7

The second approach is to take the value of +L as the time interval between the application of the
step input and the time when the output attains 63.2% of its final value. We obtain K=1, =4.2 and
L=4.3. This yields Kp=1.23, Ti=8.6, and Td = 2.15. The resulting set-point step response is shown in
Figure 5-25, while the load disturbance response is shown in Figure 5-26. Note that in the second
case the time constant is smaller than the first and the process output is far less oscillatory.

21
 Step Response

0.8

Amplitude
0.6

0.4

0.2

0
0 10 20 30 40 50 60 70
Time (sec)

Figure 5-25 Set-point step response for Problem 5.11 with =4.2

Step Response
0.7

0.6

0.5

0.4
Amplitude

0.3

0.2

0.1

-0.1
0 10 20 30 40 50 60
Time (sec)

Figure 5-26 Load disturbance step response for Problem 5.11 with =4.2

5.15 Apply the script of Problem 5.14 to the system

1
G(s) 
( s  1) 2

and simulate the response of the control system when a set-point step change and a load
disturbance step are applied. Compare the result obtained with those of Problem 5.14.

The application of the script of Problem 5.14, where we take the value of +L as the time
interval between the application of the step input and the time when the output attains 63.2% of
its final value, yields K=1, =1.94 and L=0.28. This yields Kp=8.30, Ti=0.56, and Td = 0.14. The
resulting set-point step response is shown in Figure 5-27, while the load disturbance response is
shown in Figure 5-28. It can be seen that, while the response is still oscillatory, as it is typical of
the Zeigler-Nichols tuning rules, the settling time is much less than that obtained in Problem
5.14. In fact, being the process of second order instead of eighth order, the approximating first
order model has a smaller time constant and a smaller ratio L/T. Similar considerations holds for

22
 the case where +L is considered as the time interval between the application of the step input
and the intersection of the tangent line with the steady-state output level.

Step Response
1.5

1
Amplitude

0.5

0
0 2 4 6 8 10 12
Time (sec)

Figure 5-27 Set-point step response for Problem 5.15

Step Response
0.1

0.08

0.06

0.04
Amplitude

0.02

-0.02

-0.04
0 2 4 6 8 10 12
Time (sec)

Figure 5-28 Load disturbance step response for Problem 5.15.

5.16 Use the Ziegler-Nichols closed-loop method to design a PID controller for the system

based on Table 5.2. Obtain the response due to a unit step input together with a unit step
disturbance signal. Compare the results with those of Problem 5.13.

23
The critical gain and the critical frequency of the system can be easily computed by using
the MATLAB commands

s=tf('s');
Gs=1/(s+1)^4;
allmargin(Gs)
ans =
GainMargin: 4.000000000000012
GMFrequency: 1.000000000000001
PhaseMargin: -180
PMFrequency: 0
DelayMargin: Inf
DMFrequency: 0
Stable: 1

Thus, we have the critical gain Kc=4 and the period of oscillations Tc=6.28. The Ziegler-
Nichols rules given in Table 5.2 provide the following PID parameters: Kp=2.4, Ti=3.14,
and Td=0.785. The response of the system due to a unit step reference input at t=0 together
with a unit step disturbance at t=35 is plotted in Figure 5 -209. The figure shows that the
closed-loop rules provide a less oscillatory response than that of the open-loop rules. This
is not surprising if one compares the PID parameters obtained in the two cases (see Table
P.1). The closed-loop approach results in a smaller proportional gain and higher integral
and derivative time constants, which all contribute to a less aggressive response.

Table P.1- PID controller parameters for Ziegler-Nichols design

Parameter Open-loop Closed-loop
Kp 2.77 1.87
Ti 2.6 8.6
Td 0.65 2.15

24
 1.4

1.2

process output
0.8

0.6

0.4

0.2

0
0 10 20 30 40 50 60 70
time

Figure 5-299 Output response of the system of Problem 5.16 for a step reference input
together with a step load disturbance with Kp=2.4, Ti=3.14, and Td=0.785

25