Anda di halaman 1dari 5

MM2DTC

CONTROL LABORATORY [20] mark scheme


O.J. Bakker and W. Sun
May 13, 2009
Contents
1 Summary 1
2 System Modelling 1
3 Results 2
3.1 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
3.2 Theoretical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
4 Discussion 4
1 Summary
The objective of this laboratory exercise is to investigate the eects of proportional and inte-
gral control actions on the transient response and steady state error of a liquid level control
system. [2/7]
If we consider test 1 to be the normal situation, then if:
The proportional gain is increased, the steady state error is reduced [2/7]
The value of the resistance of the restrictor is increased (i.e. the valve is closed for a
bit), the steady state error is reduced. [2/7]
If integral action is added there is no steady state error [2/7]
If we apply more integration action, there will be more damping and the oscillation
behaviour becomes slower, i.e. the eigenfrequency decreases. [2/7]
Conclusions:
1. All these ndings are in agreement with the derivations from the theoretical / analytical
model. [2/7]
2. The results are in close agreement with the analytical model. [2/7]
Total: 2 marks.
1
2 System Modelling
From the mass balance (the continuity equation) we learn, that what comes in the tank must
go out or stays in the tank:
Q
i
= Q
o
+A
dH
dt
[1]
For the linearised ow over the restrictor holds according to the electrical analogon:
Q
o
=
H
R
[1]
R
Q
0
(I)
H (V)
R
Q
0
(I)
H (V)
Figure 1: The electrical analogon for linearised ow, with the electrical symbols between
brackets
After the Laplace transform (with assumed zero conditions!) and some rearrangement we
obtain the transfer function relating the water level H and inow Q
i
:
H(s)
Q
i
(s)
=
R
1 +ARs
. [1] (1)
The block diagram is already given in the lab sheets. Its explanation is as follows:
We can set the desired value for the water level H
r
on the computer. The reference voltage
V
REF
for the motor of the pump also changes with this reference height.
The control system measures the actual water level H in the water tank and compares this
with the desired value for the water level H
r
. Note that the computer translates the desired
H
r
of 150 mm into a voltage of 6 V, also the transducer which measures the water level gives
an output in Volts!
The control system subtracts the two voltages from each other. The dierence is called the
error voltage V
E
. This error voltage is then amplied by the controller, which acts as some
kind of dynamic, frequency dependent amplier. This controlled voltage is related to the
error voltage as follows: V
con
= G
C
(s)V
E
. The amplied voltage - let us now call this the
controlled voltage V
con
- which can be a positive or negative voltage is then added to reference
voltage V
REF
which supplies the electricity to the pump. The relation between the output of
the pump Q
i
(!) is linear related to the voltage that is put on pump: Q
i
= (V
ref
+V
con
)K
m
The relation between the input in the tank Q
i
and the actual water level was already derived
in equation (1). The block diagram can thus be drawn on basis of the following blocks: [2].
Or: Error voltage [1], and controller transfer function [1]; or similar.
Total: 5 marks, grand total Summary + Modelling = 2+5=7.
2
3 Results
3.1 Experimental Results
A set graph of the experiment should be provided by the module conveyor. The students
should provide two graphs concerning the water level in m or mm. Possibly they can make a
remark about the sensor reading. Alternatively they can provide a graph of the error voltage.
Graphs should be clear. Quantities and units should be displayed at both axes. Deduce a
half mark for each fault you spot. [1
1
2
] 2.
Some students will take all the individual sample points as seconds. The time units can easily
be found in the data le from the computer. Deduce a half mark for this for each spotted
error. Total marks: 3.
3.2 Theoretical Results
For water level H holds:
H =
_
[H
R
H] KK
P
_
1 +
1
T
I
s
_
+V
REF
_
K
m
R
1 +ARs
Rearranging with Maple learns us:
H =
(H
R
KK
p
T
I
s +H
R
KK
p
+V
REF
T
I
s) K
m
R
ART
I
s
2
+ (1 +KK
p
K
m
R) T
I
s +KK
p
K
m
R
[1
1
2
]. (2)
The term in the denominator is clearly a second order function: s
2
+ 2
n
s +
2
n
. [1/2]Give
only [1/2] for students that show the whole second order function, mentioning s
2
is the
highest order does not show that the student understands what (s)he is doing. Obviously the
eigenfrequency
n
can be dened as:
H
R
[m]
K
[V] + V
E
[V]
G
c
V
REF
[V]
+
+
K
m
ARs 1
R
+
V
con
[V]
H [m]
V
supply
[V]
K
Q
i
[m
3
/s]
-
sensor reading
computer conversion
controller pump tank
H
R
[m]
K
[V] + V
E
[V]
G
c
V
REF
[V]
+
+
K
m
ARs 1
R
+ ARs 1
R
+
V
con
[V]
H [m]
V
supply
[V]
K
Q
i
[m
3
/s]
-
sensor reading
computer conversion
controller pump tank
Figure 2: The block diagram with the larger blocks that describe the behaviour of the
controller, the pump and the tank
3

n
=
_
KK
p
K
m
AT
I
. [1/2]
Damping becomes then clearly:
=
1 +KK
p
K
m
R
2R

T
I
AKK
p
K
m
[1/2].
Thus:

n

_
1
T
I
;
_
T
I
.
Hence zero marks when students show:

T
I
T
I
;
T
I

T
I
.
In this case, they have not understand that damping and oscillation time (eigenfrequency)
depend (inverse) proportional to the integration time constant T
I
.
V
E
can be dened as follows:
V
E
= K (H
R
H) (3)
Then the substitution of equation (2) into (3) yields us the following expression for V
E
:
V
E
=
K ([1 +ARs]H
R
K
m
RV
REF
)
1 +ARs +KK
p
K
m
R
_
1 +
1
T
I
s
_ (4)
For application of V
REF
and H
R
as steps we will use the following expressions:
V
REF
=

V
REF
s
; H
R
=

H
R
s
When we do not apply integral control, the value for T
I
goes to , so the term in the
expression for the controller disappears. Applying the nal value theorem gives in this case
for the steady state error:
v
E,ss
=
K
_

H
R
K
m
R

V
REF
_
1 +KK
p
K
m
R
[1].
Only in this form it can be seen that for K
p
, v
E,ss
= 0 and for R

H

V
ref
K
m
, v
E,ss
= 0.
With lHopitals theorem the same can be deduced for R . Do not give marks in case students
fail to bring everything in this simplied fraction.
And if we apply some integral action, after the applying the nal value theorem, we obtain
the following steady state error:
v
E,ss
= 0 [1].
Marks: 3+2=5, grand total marks: 3+5=8.
4
4 Discussion
Proportional control
From experiments
1. Change in the tap setting changes the steady state error. [1]
2. Increase in of the control eort K
p
decreases the steady state error. [1/2]
These observations are backed up by the expression for the steady state error for the propor-
tional control only. If the control eort would be , then the steady state error would be
zero. And for a certain tap setting (probably almost closed, deducted from experiment) the
steady state error would also be eliminated. [1/2]Or similar concerning the tap setting.
PI control
1. There is no steady state error, this is according to the theoretical model. [1]
2. If T
I
increases, the eigenfrequency decreases, agrees with theory [1]
3. and the damping increases, also in agreement with theory. [1]
4. note that the rise time t
r
= 1.8/
n
, thus if the frequency increases the rising time
decreases, i.e. the system becomes stier.
marks: 2+3=5; total marks for report: 7+8+5=20.
5

Anda mungkin juga menyukai