Indian Institute of Technology Bombay

Project Report
2023-2024

AE 308
Control Theory

By: Lagnesh Mahapatra

Roll No.: 210010038
Course Instructor: Prof. Arnab Maity
1 Introduction
So my aim is to design a PD controller for the system that I will describe below
in detail to meet the desired specifications. PD controller gives a fairly good
enough performance, for much accurate tracking purposes we can use PID

1.1 System Description

The open loop transfer function of the system/plant is given by

K(s + 6)
(s + 3)(s + 4)(s + 7)(s + 9)

considering unity feedback.

1.2 Unit Step Response, Steady State Errors

So first of all, let’s take K = 1 and study the steady-state errors and the system
properties. K is just a gain factor so we can adjust later on to meet our desired
requirements.
The unit response of the system considering K = 1 is given as follows:

Figure 1: Unit Step Response of the System (K = 1)

As it is a type zero system, the value of Kp is finite and equal to

1
Kp = lim G(s) =
s→0 126

The output C(s) is given by R(s)·G(s)

1+G(s) . So, as the input R(s) is a step input,
and applying the final value theorem, the steady-state error is given by:

2
  
R(s)
Steady State Error = lim s
s→0 1 + G(s)
1
=
1 + Kp (1)
126
=
127
= 0.9921259842519
So, it’s huge, and the system is unable to track it properly. As we will
increase the gain further, the steady-state error will reduce. We need an Integral
controller to completely reduce the steady-state error to zero. We can also see
from the graph that the steady-state error is significant.

1.3 Stability Margins: Bode plot

The Bode plot of the uncompensated system with K = 1 is given as follows:

Figure 2: Bode Plot of the Uncompensated System (K = 1)

Our system has pretty good margins, gain margin of 62.4 dB, and a phase
margin of infinite, so it has good stability margins.

1.4 Analysis using Root locus

The Root locus plot of the system with K = 1 is given as follows:
So, as there are 4 finite poles and 1 finite zero and 3 infinite zeros, the root
locus will consist of three asymptotes as we can see. The gain of the system can
be increased nearly by 1310 before the dominant poles jump into the right half
of the complex plane. The breakaway point is close to −3.53.

3
 Figure 3: Root Locus Plot of the System (K = 1)

2 Controller Design
The following specifications need to be met:
1. Peak Overshoot to a unit step input should be less than 15%

2. Settling Time to a unit step input should be less than 1 sec

So I have decided to go with a PD controller because D will help to reduce
the overshoot and the settling time, and the P will help to reduce the steady-
state error with minimal impact on settling time. However, it increases the
overshoot, but we can choose the gains wisely to tackle this.
Let’s convert the given specifications to desired second order poles. The
damping ratio (ζ) is given by the percent overshoot as follows:
 
ln %OS
100
ζ = −r   = 0.51693 (2)
2 2 %OS
π + ln 100

If the damping ratio is more than this, the percent overshoot will reduce
further; our damping ratio has to be better than this.
Settling time should be less than 1 sec, so:
4
<1 (3)
σ
So, the magnitude of σ has to be greater than 4. In simple words, the real
part of the dominant second order poles should be less than -4. So now we want
the root locus to pass through these poles so that we get the desired transient
response. For that we need to apply the magnitude and phase condition of Root
locus method
Now, let’s choose the real part of the dominant pole to be -6 and the damping
ratio to be 0.6. I have taken slightly higher numbers because we will be almost
sure that we will get the desired performance, if we take close to the values we

4
got earlier, then chances are there we may get close results but not better. We
can calculate the imaginary part as follows:
σ = ζωn (4)
6 = 0.6ωn (5)
ωn = 10 rad/s (6)
p p
Imaginary part = −ωn 1 − ζ 2 = −10 × 1 − 0.62 = −8 (7)
So we got the desired poles −6 + 8j and −6 − 8j. The root locus has to pass
through these.
Assuming the transfer function of the PI controller is K1 (s + a), the open
loop transfer function of the resulting compensated system becomes
K2 (s + a)(s + 6)
G(s) =
(s + 4)(s + 3)(s + 7)(s + 9)
where K2 = 75.28 is the cascaded gain, K2 = KK1 .
Applying the root locus conditions, we choose the poles to be −6 + 8j and
−6 − 8j. This leads to a = 5.03 and K2 = 75.28.
The desired closed-loop performance includes a settling time close to 0.67
seconds and a damping ratio close to 0.6 at the gain of 75.28 on the root locus
plot. Now lets check with simulation results.

3 Simulation Results
Below is the step response:

Figure 4: Step Response with SS value and maximum value shown on graph

The steady state value is close to 0.751 and the maximum value is 0.825, so
percent overshoot is easily calculated as 9.854 percent. The settling time is also
well below 1 second as shown in the figures below. The margins are also pretty
good, with a gain margin of 15 and phase margin of 79.5 degrees. From the
Root Locus plot we observe that the system doesn’t jump into the Right Half
plane for any values of Gain, which is a great thing. The Root Locus has been
drawn by taking the 75.3 factor into account. So our controller is ready and all
the specifications have been implemented successfully.

5
 Figure 5: Step Response with settling time shown on the graph

Figure 6: Bode Plot with margins of the System for (K2 = 75.28)

Figure 7: Root Locus Plot of the Compensated System (K2 = 75.28)

4 Conclusion
So, we have succesfully developed a PD controller to control the system and
meet our specifications. We can also try out PID to acheive the same task.