Experiment 13: Process Control 1 (Armfield PCT 9
& 10)
James Forsyth
H00128857
Group 8
Performed: 16th February 2015
Submitted: 16th March 2015
Supervisor: Professor G Markx
Contents
SynopsisPage 3
Introduction..Page 3
Theory...Page 3
Experimental TechniquePage 4
ResultsPage 7
DiscussionPage 12
Conclusion..Page 13
References..Page 13
Appendices.Page 13
Synopsis
2
The purpose of this experiment was to investigate different control systems. It
involved looking at both on-off and linear control systems and within that
Proportional and Proportional-Integral. The results we gained were valid and
accurate and gave a god indication of all the different systems looked at and
what occasions they are most appropriate and the optimum setting for sing
them.
Introduction
The aim of this experiment was to compare the performance of a number
of different control systems. The experiment involved comparing an on-off
controller against a linear feedback controller. We then looked at
comparing different types of linear control systems, a proportional
controller and a proportional-integral controller. We also determined the
optimum value for the controller gain, Kc, and investigated how the
response of a proportional control system depends on the value of Kc.
Theory
In this investigation we looked at two different types of control systems,
on-off control and linear feedback control. On-off is the simpler of the two
as the system just literally turns its self on and off constantly in order to
achieve a relatively stable value called the set point. This system however
has its faults in that you rarely get the exact value that you desire, the
system just fluctuates between limits and the controller is constantly
being turned on and off. For our experiment a tank of water with a set
height was used and the flow of water in was constantly turned on and off
in order to maintain the same liquid level. On-off systems are used so
commonly because of their simplicity and are cheaper.
The other control system we looked at was linear feedback control. A
linear control system works by receiving a measurement from the
sensor/transmitter and comparing it with a pre aligned set point. The
difference or error (e(t)) between the two is then calculated and the
controller output is calculated from this error. There are different ways of
doing this depending what set up on the system is used. The two main
controller used are proportional (P) and proportional integral (PI)
controllers. The equation used for calculating the output signal from the
error is the following:
m ( t )=m+
Kc e (t )
Where
m ( t ) is the output signal of the controller,
controller output signal if the error is zero,
e(t )
is the error. K c
the form
PB=
100
.
Kc
Kc
m
is the bias or
is the controller gain and
is also known as the proportional band which takes
The equipment we used gave vales of 1/5 of the
proportional band, so this had to be converted when calculating
Kc
values. The reason that PI controllers are more commonly used is that the
P controllers even at steady state do not reach the exact set point which
leads to an offset. The value of this offset is dependent on the controller
gain. PI controllers are used to get rid of this offset and are therefore more
accurate, however the equation used to describe the controller output is
more complicated:
m ( t )=m+K
c e (t )+
Kc
e ( t ) dt
I
The integration part of this equation removes the offset since the error.
Another controller type that could be used is a proportional integral
derivative or PID controller. This type of controller is used if there is a need
to predict the error and therefore do something to prevent it. This types
can only be used in specific cases though, for example if the system
makes a lot of noise this will affect the fluctuations of the error difference
over the time difference.
Experimental Technique
The equipment used for this experiment was called an Armfeild Process
Module for level and flow control and was set up in the following manner:
Figure 1: Sketch of the PCT9 Process Module for flow and level control
As visible form the diagram the system consists of 2 tanks, T1 and T2. The water
level in T2 is what would be controlled for the first part of the experiment. T2
contained a water level sensor and an overflow. The unit consisted of 5 manually
controlled valves and 3 solenoid valves. A rotameter controlled the flowrate of
water which filled T2. Valves V3, V4, SOL2 and SOL3 all controlled the flowrate of
water coming out T2.
The control module used for this experiment was called The Armfield PCT10. The
diagram below shows the front of the module:
Digital Process Controller
Voltmeter
Ammeter
Power Supply
4-20 mA Manual Output
Positioner
Switched Output for On-OffMotor
control
Signal Conditioning Channels - (CH1) (CH2)
5
Figure 2: The front of PCT10
The pump was connected to the 240V connection on the side of PCT10
and the system tested to ensure that T2 filled when the pump was on. The
rotameter was tested to ensure it was functioning correctly by adjusting
valve 2. SOL1 was closed by connecting it to the 24V outlet in the section
marked motor positioner on PCT10. The motor positioner and manual
output were connected. This stopped the flow into the tank. SOL2 and 3
were also tested in this manner. The valves opened when the manual
output knob is twisted to open which applied the power.
On-Off Control:
The first part of the experiment involved looking at on-off control. T2 was
emptied and SOL, 3 and V4 were all closed. The on-off controller switch
was placed in the tank. The level switch was connected to connections A
and C in the On-Off control section on PCT10. The pump was switch on
and V2 was fully opened to allow the tank to fill up. The level switch was
set to a height of 80mm which is the height that we want the system to
maintain. Once the water reached this level SO1 started turning on and off
continuously in order to keep the water at this level. This was then
repeated opening SOL2 and the time intervals between the flow turning
off and on recorded. This was then repeated again with SOL3 open and
SOL2 closed and then again opening both SOL2 and 3. The time between
SOL1 switching on and off was recorded each time.
Linear Feedback Control:
The level sensor first had to be calibrated. The sensor in T2 was again
connected to PCT10 and the rest of the system was set up as before. The
tank was filled to 130mm and then V2 was closed so there was no flow out
or in the tank, it was set at 130mm. The reading on the voltmeter was
adjusted using a small screwdriver to read exactly 1V. The tank was then
emptied to 30mm at which point the voltmeter was again adjusted to read
exactly 0V. The tank was then filled to 130mm again and then emptied in
10mm intervals with a voltmeter reading being taken at each interval.
Next the flow sensor was calibrated. The sensor was connected to PCT10
and was connected to the voltmeter. With the motorized valve fully open,
V2 was adjusted to give a flow of 3000cm3/min. At this flowrate the
voltmeter was adjusted using a small screwdriver to give a reading of 1V.
V2 was then closed to give a flowrate of 0 and the voltmeter adjusted to
read 0V. The flow was then turned back on and set to 3000cm3/min. The
flow was adjusted using V2 and voltage reading were taken at 500
cm3/min intervals down to 0.
The system was then set up for linear feedback control. All valves were
closed apart from V2 and V4 was opened to a 30o angle. The motorized
valve was fully opened to give a flowrate of 3000cm3/min. The level
sensor was again connected to PCT10 so both sensors were connected.
The table below shows the controller settings for level control with a
proportional-integral controller:
Parameter
Proportional Band
Cod
e
Prop
Settin
g
10
Integral Time
Derivative Time
Cycle Time
Output Limit
Set Point Limit
Linear Range
Units
1/20ths of Controlled Variable
Span; i.e. actual proportional
band = 5xProp
Minutes
Seconds
Seconds
%
%
For 4-20mA = 0-100%
Int
0.2
dEr
0
CY-t 20
Pr-L 100
SP-L 100
CS-058
1
Direct/ Reverse Action
CS-r-Reverse Acting Controller
2
- Denotes Don't care values
The system was connected to the computer to record the data.
The set point was set to 20% and the data logging system was started.
The set point was then changed consecutively to 40%, 60%, 80% and
100% whilst continually recording. The datalogging was stopped after it
reached 100%.
Next the ability of a proportional controller to deal with a disturbance was
investigated. The set point was set to 50% and V4 was opened to 30o. The
controller was set up in the following manner:
Parameter
Cod
e
Prop
Settin
g
20
Integral Time
Int
Derivative Time
dEr
Cycle Time
CY-t
Output Limit
Pr-L
- Set Point Limit
SP-L
Linear Range
CS-1
Direct/ Reverse Action
CS-2
Denotes Don't care values
0
0
20
100
100
-058
-r--
Proportional Band
Units
1/20ths of Controlled Variable
Span; i.e. actual proportional
band = 5xProp
Minutes
Seconds
Seconds
%
%
For 4-20mA = 0-100%
Reverse Acting Controller
The settings used made the controller operate in manual mode so there
was no control. SOL2 was opened to create a disturbance and the
7
datalogging was turned on and gathered information until the system
reached a steady state.
This exact procedure was repeated repeated with the controller switched
to normal mode. This was then repeated again a further 6 times, each
time altering the Proportional Band to 10, 4, 2, 1, 0.5 and 0 respectively.
The datalogger was reset and recorded the data individually for each Prop
value.
The exact procedure was then repeated again to test the ability of a
proportional-integral controller to deal with the same disturbance. The set
up was identical however and Integral time value of 0.2 was set, as
opposed to 0 for proportional. The same values of PB were used: 20, 10, 4,
2, 1, 0.5 and 0.
Results
On-Off control:
SOL2 Open and SOL3 closed - Average time on = 0.53 seconds = 0.009
min
Average time off = 1.53 seconds =
0.026 min
Flowrate SOL2 =
t on
cm3
cm3
3000
=772
min t off t on
min
SOL3 Open and SOL2 closed - Average time on = 1.66 seconds = 0.028
min
Average time off = 1.006 seconds =
0.017 min
Flowrate SOL3 =
cm3
1868
min
Both Open - Average time on = 40.15 seconds = 0.669 min
Average time off = 0.65 seconds = 0.011 min
Flowrate SOL3 =
2952
cm3
min
The flowrate when both were open is almost exactly equal to that of the
flowrate in.
Linear Feedback Controller:
Table 1
Calibrating Level Sensor
Height (mm)
130
120
110
100
90
80
70
60
50
40
30
Voltage (V)
1.003
0.88
0.779
0.68
0.576
0.475
0.376
0.277
0.179
0.082
0
Figure 1
Voltage vs Height
140
R = 1
120
100
80
Height (mm)
60
40
20
0
0.2
0.4
0.6
Voltage (V)
0.8
1.2
This figure shows how the controller was successfully calibrated between
130mm and 30mm.
Table 2
Calibrating Flow Sensor
3
Flowrate (cm /min)
Voltage (V)
500
0.01
1000
0.24
1500
0.394
2000
0.593
2500
0.785
3000
Figure 2
10
Florate vs Voltage
1.2
1
R = 1
0.8
Voltage (V) 0.6
0.4
0.2
0
500
1000
1500
2000
2500
3000
3500
Flowrate (cm3/min)
Figure 2 shows the flow sensor was calibrated correctly. There is a very
small reading at 500 cm3/min since at such a low flowrate the sensor
struggles to pick anything up, hence the tiny increase from 0-500, from
500 upward it increases rapidly in a linear fashion.
11
Changing The Setpoint
120
100
80
%T / %C
60
40
20
0
50
100
150
200
250
300
350
400
450
Time (sec)
Changi
ng the Setpoint:
Figure 3
Where Series 1 = Liquid Height
Series 2 = Output Voltage
Series 3 = Output Flowrate
Figure 3 shows how when the set point is changed the output voltage and
output flowrate almost change at the same rate in order to reach the new
set point, with the output flowrate slightly slower as shown by the curved
line up until 100% rather than the distinctive flat almost 90o angle shown
by the output voltage.
Effect of a Disturbance on Proportional Controller- without control:
Figure 4
Proportional Controller - Without Control
60
50
40
% Height 30
20
10
0
20
40
60
80
Time
12
100
120
140
Where Series 1 = Liquid Height
Series 2 = Output Voltage
Series 3 = Output Flowrate
Figure 4 shows how when the disturbance occurred with the system on
manual the controller does not respond and therefore the tank will just
keep on emptying. This is clearly visible by the liquid level dropping
rapidly and no change at all in the output voltage or output flowrate.
Proportional Controller:
Figures 5-11, which can be found in the appendices, show the graphs for
how a Proportional Controller with different Proportional Bands of 20, 10,
4, 2, 1, 0.5 and 0 cope with a disturbance. For each figure:
Series 1 = Liquid Height
Series 2 = Output Voltage
Series 3 = Output Flowrate
From these graphs it is clear that a PB of 2 is the most accurate as it gets
closest to the set value, has the quickest response and has a lot less
fluctuations that 1, 0.5 and 0. These results will be analysed in more detail
in the discussion.
Proportional-Integral Controller:
Figures 12-17, which can be found in the appendices, show the graphs for
how a Proportional-Integral Controller with different Proportional Bands of
10, 4, 2, 1, 0.5 and 0 can cope with a disturbance. For each figure:
Series 1 = Liquid Height
Series 2 = Output Voltage
Series 3 = Output Flowrate
The oscillations at lower values of Kc for the flowrate were less when using
a PI controller.
13
Discussion
The first part of the experiment involved using an on-off control system.
The flowrates of the water flowing through SOL2 and 3 were calculated by
timing the length of time the flow was turned on and off. The flow turned
both on and off rapidly when SOL3 was closed and SOL2 was left open
with it being on for 0.53 seconds and 1.53 off. This resulted in a flowrate
of 772 cm3/min flowing out of SOL3. Next SOL2 was closed and SOL3
opened. This resulted in both being in and off for roughly the same time,
about 1 second each. This resulted in a higher flowrate of 1868 cm3/min.
When both were open it was on for 40 seconds and off for just 0.65. This
results in large fluctuations for the actual liquid level which is a
disadvantage of on-off control.
The next part of the experiment was to change to a linear controller. The
level and flow sensors were successfully calibrated with a very slight error
caused by only doing the calibration once. This would be a possible
improvement to accuracy should the experiment be repeated.
The changing of the set point was then looked at by setting the initial level
to 20% of the height, 50mm, and starting the datalogger. The set point
was changed up to intervals of 40, 60, 80, and 100%. The results can be
seen from figure 3, when the set point was changed you can clearly see
the sharp increase form the voltage output which was quickly followed by
the output flowrate, which causes the level to increase to the new set
point.
Figure 4 shows how if the control system is just set to manual then when a
disturbance is created the system will not do anything to adjust. There is
no signal output and therefore no flow so the tank will just slowly drain till
it reaches a steady state much lower than the set point.
Figures 5-11 show the graphs for how a Proportional Controller with
different Proportional Bands of 20, 10, 4, 2, 1, 0.5 and 0 cope with a
disturbance. These values for PB can be converted to Kc using the
100
PB=
K c equation, giving 1, 2, 5, 10, 20 and 40, (each PB value was
multiplied by 5). As these values of Kc increased the liquid height got
closer to the set point after the disturbance and was quicker up until a
value of Kc = 10. After this point the voltage output and therefore output
flowrate fluctuated massively resulting in the liquid height fluctuating
more also. So a PB value of 2 which equals a Kc value of 10 was the
optimum one to use when using a Proportional Controller.
Figures 12-17 show the graphs for how a Proportional-Integral Controller
with different Proportional Bands of 10, 4, 2, 1, 0.5 and 0 can cope with a
disturbance. These results were very similar to the proportional controller
14
ones. The only difference being a slightly smaller offset which means for
more accurate results. This can be seen by comparing the graphs of the
same PB, which show that theres a smaller dip after the disturbance.
Conclusion
The first aim of this experiment was to look at an on-off control system.
The results we got showed how although its a simple system it has its
disadvantages since the liquid level fluctuated around the set point
constantly.
The second aim of the experiment was to compare Proportional Control
against Proportional-Integral Control and look at the effect of varying Kc
values. This was done successfully and it can be clearly see that for
proportional control a value of PB of 2 or Kc value of 10 was the optimum
one for level control in this system since it gave the least offset from the
50% set point, had the least fluctuations compared to the lower PB values
and had the quickest response time compared to the higher PB values.
Comparing Proportional to Proportional Integral in terms of Kc values both
were pretty similar as seen by the graphs.
References
Process Control with the Armfeild PCT9 and 10: Laboratory Notes
Appendices
Figure 5
Proportional Controller - PB of 20
70
60
50
40
% Height
30
20
10
0
20
40
60
80
Time (sec)
15
100
120
Figure 6
Proportional Controller - PB of 10
70
60
50
40
% Height
30
20
10
0
10
20
30
40
50
60
70
80
90
Time (sec)
Figure 7
Proportional Controller - PB of 4
70
60
50
40
% Height
30
20
10
0
10
20
30
40
50
60
Time (sec)
Figure 8
16
70
80
90
Proportional Controller - PB of 2
70
60
50
% Height
40
30
20
10
0
20
40
60
80
100
120
140
Time (sec)
Figure 9
Proportional Controller - PB of 1
80
70
60
50
% Height
40
30
20
10
0
20
40
60
80
Time (sec)
17
100
120
Figure 10
Proportional Controller - PB of 0.5
120
100
80
% Height
60
40
20
0
10
20
30
40
50
60
Time (sec)
Proportional Controller - PB of 0
120
100
80
% Height
60
40
20
0
10
20
30
40
50
60
70
Time (sec)
Figure 11
18
Figure 12
PI Controller - PB of 10
80
70
60
50
% Height
40
30
20
10
0
10
20
30
40
50
60
Time (sec)
Figure 13
19
70
80
90
PI Controller - PB of 4
70
60
50
40
% Height
30
20
10
0
10
20
30
40
50
60
70
80
Axis Title
PI Controller - PB of 2
70
60
50
40
% Height
30
20
10
0
10
20
30
40
50
60
70
80
Time (sec)
Figure 14
PI Controller - PI of 1
100
80
60
% Height
40
20
0
10
20
30
40
50
60
70
80
Time (sec)
Figure 15
20
Figure 16
PI Controller - PB of 0.5
120
100
80
% Height
60
40
20
0
10
20
30
40
50
60
Time (sec)
PI Controller - PB of 0
120
100
80
% Height
60
40
20
0
10
15
20
25
30
35
40
45
50
Time (sec)
Figure 17
21
