Second intermediate examination APC FS20

Thierry Prud’homme
[email protected]

Exercises Serie: 120 Points Topics: Applied Control and Automation

Exercises Points Results

ODE Diagram - Steady State 35
Controller development 53
Process modelling 32
Total: 120

Second Int. Exam. APC FS20 - Page 1 of 4

[Exercise 1] ODE Diagram - Steady State . . . . . . . . . . . . . . . . . . . . . . . 35 Points

A process can be modeled with the following differential equation:

ÿ(t) + 5ẏ(t) + 6y(t) = 4u(t) − 2v(t)

with y(t) the process output, u(t) the control signal (the process input that can be
manipulated) and v(t) a disturbance.
(a) (7 Points) Draw, on paper, the ODE diagram of this process.
(b) (6 Points) If we assume that the signals y(t), u(t) and v(t) are constants and equal
to ȳ, ū and v̄ respectively, what is the value taken by ȳ if ū is equal to 5 and v̄ is
equal to 4? Justify your answer.
(c) (6 Points) Laplace-Transform the differential equation and isolate Y (s) (Laplace-
Transformation of y(t)) on the left-hand side.
(d) (6 Points) A P Controller is envisioned for that process. Extend your ODE Diagram
with this controller.
(e) (4 Points) Compute the open-loop transfer function of the closed-loop controlled
process.
(f) (6 Points) Derive the reference tracking closed-loop transfer function.

Second Int. Exam. APC FS20 - Page 2 of 4

Solution:
A process can be modeled with the following differential equation:

ÿ(t) + 5ẏ(t) + 6y(t) = 4u(t) − 2v(t)

with y(t) the process output, u(t) the control signal (the process input that can be
manipulated) and v(t) a disturbance.

1. The highest derivative hat to be isolated on the left-hand side of the equation:

ÿ(t) = −5ẏ(t) − 6y(t) + 4u(t) − 2v(t)

The ODE Diagram can be seen in figure 1.

Figure 1: ODE Diagram

2. Signals are constant in steady-state. It means that their derivatives are equal to 0.
The differential equation can be simplified:
4ū − 2v̄
ȳ =
6
4×5−2×4
=
6
= 2

3. The differential equation is linear and can be Laplace-transformed. The following

transformations have to be performed:

y(t) → Y (s)
ẏ(t) → sY (s)
ÿ(t) → s2 Y (s)

It follows:

s2 Y (s) = −5sY (s) − 6Y (s) + 4U (s) − 2V (s)

s2 + 5s + 6) Y (s) = 4U (s) − 2V (s)

4 −2
Y (s) = 2 U (s) + 2 V (s)
s + 5s + 6 s + 5s + 6
4. The ODE-Diagram with the P controller can be seen in figure 2

Second Int. Exam. APC FS20 - Page 3 of 4

 Figure 2: ODE Diagram

5. The open-loop transfer function L(s) can be computed as follows:

L(s) = C(s)G(s)

where C(s) the transfer function of the controller is and G(s) the transfer function
Y (s)
U (s)
. The transfer function of the P controller is:

U (s)
C(s) = = Kp
E(s)

The transfer function G(s) is:

Y (s) 4
G(s) = = 2
U (s) s + 5s + 6

It follows:
4Kp
L(s) =
s2 + 5s + 6
6. The reference tracking closed-loop transfer function can be derived as follows:

Y (s) L(s)
Tr (s) = =
R(s) 1 + L(s)
4Kp
s2 +5s+6
= 4Kp
1 + s2 +5s+6
4Kp
=
s2 + 5s + 6 + 4Kp

Second Int. Exam. APC FS20 - Page 4 of 4

[Exercise 2] Controller development ............................ 53 Points

The block diagram of a closed-loop controlled process can be seen in figure 3.

Disturbance

Tracking error Control signal

Reference signal + Process output
+
System
+
-

Figure 3: Block diagram of a closed-loop controlled process

A single disturbance V (s) can be seen acting directly on the process output.
The process can be modelled by the following transfer function:

Ks
G(s) =
1 + T1 s

with:

Ks = 10
T1 = 0.25

(a) (4 Points) Determine the orders of the numerator an denominator of G(s). What
kind of transfer function is it?
(b) (5 Points) This process (uncontrolled) is being excited by a step signal which jumps
from 1 to 5 at time 2 (s). Sketch the evolution of the process output on paper for
this excitation.
(c) (5 Points) A P controller is introduced. Give the transfer function K(s) of this
controller.
Ym (s)
(d) (5 Points) Derive the reference tracking closed loop transfer function Yf (s)
with Kp
undefined (do not replace Kp by a value).
Ym (s)
(e) (5 Points) Derive the disturbance rejection closed loop transfer function V (s)
with
Kp undefined (do not replace Kp by a value).
(f) (5 Points) Compute the steady-stater error e(∞) in percent for a constant reference.
Express this error as a function of Kp .
(g) (5 Points) How large should Kp be chosen to get e(∞) smaller than < 1% (again
for a constant reference) ?
(h) (6 Points) Is the closed loop controlled process stable for this value of Kp ? (a process
is stable if the roots of the denominator of its transfer function are all negatives)

Second Int. Exam. APC FS20 - Page 5 of 4

(i) (3 Points) Now, the following controller is introduced:
1+s
K(s) = Kp
s
What kind of controller is it (P, PI, PID or PD)?
(j) (5 Points) Derive the reference closed loop transfer function YYmf (s)
(s)
and express it as
a function of Kp . Compute the steady state error after a reference step change.
(k) (5 Points) A PI controller is now being envisioned. Ti is chosen equal to T1 and
Kp (controller gain) equal to Ks . Derive the reference closed loop transfer function
Ym (s)
Yf (s)
for this controller. How does the closed-loop controlled system behaved?

Second Int. Exam. APC FS20 - Page 6 of 4

Solution:

1. The order of the denominator is zero and the order of the numerator is 1. The
process behaves like a first-order lag.
2. The process output for this step excitation can be seen in figure 4

Figure 4: ODE Diagram

3. The transfer function of the P controller is:

U (s)
C(s) = = Kp
E(s)

4. The open-loop transfer function L(s) is:

K
L(s) = C(s)G(s) = Kp
1 + T1 s

The reference tracking closed loop transfer function can be derived as follows:
Ym (s) L(s)
Tr (s) = =
Yf (s) 1 + L(s)
Ks Kp
1+T1 s
= Ks Kp
1 + 1+T 1s

Ks Kp
=
T1 s + 1 + Ks Kp

Second Int. Exam. APC FS20 - Page 7 of 4

5. The disturbance rejection closed loop transfer function can be derived as follows:
Ym (s) 1
=
V (s) 1 + L(s)
1
= Ks Kp
1 + 1+T 1s

1 + T1 s
=
T1 s + 1 + +Ks Kp

6. The final value of e(t) can be computed as follows (is we assume that the final value
of the reference signal is 1):

e(t → ∞) = Tr (s = 0)
1
=
1 + L(0)
1
=
1 + Ks Kp

expressed in percent:
100
e(t → ∞)% =
1 + Ks K p
7. The following inequality has to be satisfied:

e(t → ∞)% < 1

100
< 1
1 + K s Kp
Ks Kp > 99
99
Kp >
Ks
Kp > 9.9

8. The reference tracking closed loop transfer function has already be computed:
Ks Kp
Tr (s) =
T1 s + 1 + Ks Kp

The poles determine whether the closed-loop controlled process is stable or not.
The poles are the solutions to:

D(s) = T1 s + 1 + Ks Kp = 0
It follows that there is a unique solution s1 :
1 + Ks K p
s1 = −
T1
= −400

The unique pole is real and negative. The closed-loop controlled process is stable.

Second Int. Exam. APC FS20 - Page 8 of 4

 9. It is a PI controller.
10. The open-loop transfer function L(s) is:
1 + s Ks
L(s) = C(s)G(s) = Kp
s 1 + T1 s
The reference tracking closed loop transfer function can be derived as follows:
Ym (s) L(s)
Tr (s) = =
Yf (s) 1 + L(s)
Ym (s) Kp 1+s Ks
s 1+T1 s
=
Yf (s) 1 + Kp 1+s Ks
s 1+T1 s
Kp Ks (1 + s)
=
s(1 + T1 s) + Kp Ks (1 + s)
Kp Ks (1 + s)
= 2
T1 s + (1 + Kp K)s + Kp Ks

The final value of e(t) can be computed as follows (is we assume that the final value
of the reference signal is 1):

e(t → ∞) = Tr (s = 0)
1
=
1 + L(0)
= 0

There is no steady-state error with the PI controller.

11. The transfer function of a PI controller is:
1 + Ti s
C(s) = Kp
Ti s
If Ti is chosen equal to T1 , the open-loop transfer function is:
1 + T1 s Ks
L(s) = C(s)G(s) = Kp
T1 s 1 + T1 s
Kp Ks
=
T1 s
The reference tracking closed-loop transfer function can then be derived:
L(s)
Tr (s) =
1 + L(s)
Kp Ks
T1 s
=
1 + KTp1Ks s
Kp Ks
=
T1 s + Kp Ks
1
= T1
Kp Ks
s+1

Second Int. Exam. APC FS20 - Page 9 of 4

We recognize a first-order lag transfer function with a DC gain equal to 1 and a
time constant τf that is equal to:

T1
τf =
Kp Ks

Second Int. Exam. APC FS20 - Page 10 of 4

[Exercise 3] Process modelling .................................. 32 Points

An experiment has been carried out for a process P in order to find the right modell and
identify its parameters.
The result of this experiment can be seen in the figure 5.

Figure 5: Process step answer

The y-axis for the input signal is on the left-hand side and the y-axis for the output
signal is on the right-hand side. The input signal jumps from 0 to 4 at time t = 1 (s).
(a) (11 Points) Which transfer function mathematically explains the dynamic behavior
of this process (first-oder, second-order, etc.)?
(b) (11 Points) Compute the parameters of this transfer function?
(c) (10 Points) Propose a controller for this process using Ziegler-Nichols
Solution:

1. A first order lag with delay can be recognized. The corresponding transfer function
is:

K
G(s) = e−Tt s
τs + 1

Second Int. Exam. APC FS20 - Page 11 of 4

2. The DC gain K can be computed as follows:
∆y
K =
∆u
10 − 0
=
4−0
= 2.5

The delay Tt can easily be read from the step answer:

Tt = 0.5

The 63 % rule can be used to compute the time constant τ .

0.63 × ∆y = 6.3
The process output reaches the value 6.3 at approximately t = 1.8 (s), the time
constant is:
τ = 1.8 − 1.5 = 0.3 (s)
The process transfer function is:

2.5
G(s) = e−0.5s
0.3s + 1
3. The Ziegler-Nichols Table proposes:

Controller Kp Ti Td
P a - -
PI 0.9a 3L -
PID 1.2a 2L 0.5L
with :

1 τ
a = = 0.24
K Tt
L = Tt = 0.5

If we decide to go for a PID controller, the parameter values would be:

Kp = 1.2a = 0.288
Ti = 2L = 1.0
Td = 0.5L = 0.25

Second Int. Exam. APC FS20 - Page 12 of 4