Objective:

The objective of this experiment is:

➢ To control the level of water in the process vessel using an on/off controller to automatically start
and stop the feed pump as required to maintain the required level in the process vessel (inflow control).

➢ To determine the variations in water level due to the dead band inherent in an on/off controller.

➢ To change the level inside the process vessel by changing the set point on the on/off controller and to
determine the effect of disturbances to the process.

➢ To investigate the effect of different ranges of inflow and outflow to / from the process vessel.

➢ To control the level of water in the process vessel using a proportional controller to automatically
vary the speed of the feed pump (inflow control).

➢ To determine the response of a level process when using a P only controller to vary the pump speed.

➢ To determine the response of a level process when using a P+I controller to vary the pump speed.

➢ To change the level inside the process vessel by changing the set point on the P or P+I controller and
to determine the effect of disturbances to the process.

➢ To investigate the effect of different ranges of inflow and outflow to / from the process vessel.

The PCT50 Level Control Process consists of a combined sump tank / level vessel assembly together with
an electrical connection box, mounted on a common base plate. The process is free standing, without
the need for a permanent water supply or drain connection, and suitable for locating on a work bench
alongside a PC. The process incorporates an electronic interface with sensor conditioning, drive circuits
and integrated USB connection. This means that the user only needs to fill the process with water
(where appropriate), connect the mains adaptor and connect the supplied USB cable to a suitable PC for
the equipment to be operational. The use of transparent materials gives clear visibility of the process in
operation and the location of control components such as sensors and control elements. The facility to
vary the operating characteristics to allow optimization of the PID settings to suit the characteristics and
to demonstrate the differing responses of both optimum and non-optimum controller configuration.
Supplied with software and USB lead allowing control and data acquisition using a PC. Disturbances
(step changes) can be applied to the process, remotely from the PC, to test the stability of the system
with the controller settings under investigation. The magnitude of the disturbance can be changed in
fixed steps for repeatability or varied continuously to achieve a specific response in the system. Manual
operation via the PC will allow open loop testing of each process to determine the system response
prior to assigning appropriate values for the P, I and D terms, cycle time etc in a closed loop
configuration. In addition to control of the process and initiation of step changes, the software supplied
will also allow recording of the responses, graphing etc. when testing any of the control loops. The
electrical interface, mounted alongside the process, incorporates additional front panel connectors that
allow the input and output signals to be connected to an alternative PID controller, Programmable Logic
Controller or similar control equipment if required by the end user. The electrical supply is provided by a
universal 24V DC in-line adaptor with interchangeable leads to suit the local electrical supply.PCT50 is a
highly visible and easy to understand water level control process. It comprises two clear acrylic tanks; a
process tank mounted above a sump tank. Water is pumped up to the process tank and drains back to
the sump tank via two valves, one manually variable and the other switched by software. Each valve can
also be fitted with different-sized external orifices to change their characteristics. This flexible
arrangement allows a wide range of control scenarios to be set up, including two fundamentally
different processes, ie:

Level control by proportionally varying pump speed (Inflow control)

Level control by time-proportioned opening of a solenoid valve (Outflow control)

When using inflow control, repeatable disturbances can be implemented using the solenoid valve, and
varied by using different orifice sizes. When using outflow control the pump speed can be stepped to
provide a full range of repeatable disturbances. These techniques allow direct comparison of different
controller settings.
Theory

1) On/off control

An on/off controller is a simple and effective way of controlling many processes but does have
disadvantages because its output can only be on or off. In the case of the level process in this exercise
the pump is started and stopped.

An on/off controller incorporates a dead band to avoid rapid switching of the controlled variable when
at the setpoint i.e. in this case the level must rise above the setpoint by a fixed amount before the pump
switches off and the level must fall below the setpoint by a fixed amount before the pump starts again.

Note: On PCT50 an On/off controller with fixed dead band is created by setting the Proportional Band to
0 % in the PID controller. In a commercial On/off controller the dead band can be varied to suit the
process. This allows the choice of less frequent switching but larger variations in the process variable or
closer control of the process variable but more frequent switching with attendant wear etc. of the
components.

2) Proportional term (P)

The Proportional Band setting on a process controller makes a change to the output (pump speed on
PCT50) that is proportional to the current error value (the difference between the measured level and
the set point on the controller). The proportional response can be adjusted by multiplying the error by a
constant Kp, called the Proportional Gain. This is related to the Proportional Band setting on the
controller as follows:

Proportional Gain (Kp) = 100 % / Proportional Band %

i.e. 100 % PB means unity gain (change in controller output = error at input)

and 50 % PB means a gain of 2 (change in controller output = 2x error at input)

A low setting of the Proportional Band (large gain) results in a large change in the output for a given
change in the error. If the Proportional Band is too low, the system can become unstable. In contrast, a
large setting of the Proportional Band (low gain) results in a small output response to a large input error,
and a less responsive or less sensitive controller. If the Proportional Band is too high, the control action
may be too small when responding to system disturbances resulting in slow response and offsets of the
resulting process variable from the set point.

A Proportional-only controller will not always settle at the set point, but may retain a steady-state
offset. Offset can be reduced in Proportional-only control by reducing the Proportional Band setting.
However, if the PB is set too small then hunting or oscillating will occur. The offset can be minimized by
adding a bias to the set point (setting the set point above or below the required value to compensate for
the offset) but this technique is only appropriate if the system characteristics are known and fixed. A
better solution is to remove the offset by adding Integral action to the controller (P+I) as described
below.

3) Integral term (I)

The contribution from the integral term is proportional to magnitude and duration of the error. The
Integral term in a PID controller is the sum of the instantaneous error over time and gives the
accumulated offset that should have been corrected previously. The resulting controller output is the
sum of the contribution from the Integral term and the contribution from the P term.

When the I term is correctly adjusted any residual offset in the process variable due to the P term will be
gradually reduced by the Integral term until the offset is eliminated. If the time setting of the I term is
too long then correction to any offset will be very slow.

However, since the integral term responds to accumulated errors from the past, it can cause the present
value to overshoot the set-point value or to make the process completely unstable if the time setting of
the I term is too short. If this occurs the I term makes adjustments to the controller output faster than
the process can respond. I.e. the I term winds up the controller output so that the process overshoots
considerably, hence the term Integral Wind-up or Integral saturation.

Careful selection of the I term in combination with the P term will give efficient response to changes in
the system.

Apparatus

Refer to the equipment diagram for details about the PCT50 Level Process, Figures (1 and 2), the front
view shows the layout of the PCT50 and the circuit diagrams show the process in operation with inflow
to the process vessel (controlled by varying pump speed) or outflow from the process vessel (controlled
by opening and closing a solenoid valve on a time cycle). These alternatives are described in detail later.

PCT50 is a level control process using water as the working fluid for safety and convenience in use.

Water stored in the lower sump tank is transferred to the upper process vessel via a submerged,
variable speed centrifugal pump (10). A quick release connector (9) allows the flexible pump outlet tube
to be disconnected to aid priming of the pump after filling the sump tank with water. A vertical inlet
arrangement in the process vessel allows visualization of the water entering the vessel, independent of
water level and an integral non-return valve (8) prevents water from draining back into the sump vessel
when the pump speed is reduced or stopped. An inline ball valve (CV1) above the quick release
connector allows the flow of water entering the process vessel to be varied to suit a particular
demonstration.

The water level inside the process vessel is measured using an electronic pressure sensor (2) mounted
through the floor of the vessel. One side of the pressure sensor is connected to the inside of the process
vessel and the other is open to atmosphere, allowing the pressure in the process vessel to be measured
relative to atmosphere. This sensor therefore measures the level of water inside the process vessel. The
level is indicated (1) on the scale on the side of process vessel.

Water drains from the process vessel back into the lower sump tank via two outlets in the floor of the
process vessel. Flow through the main outlet is continuous. Flow through the second outlet can be
started and stopped by a solenoid valve (SOL) under remote control. Both outlets incorporate a hand
operated ball valves (CV2 and CV3) that allows the flow of water to be varied continuously to suit a
particular demonstration. Both outlets also incorporate an interchangeable orifice (3 and 5) that allows
the flow to be fixed at predetermined values, varying only with the size of the orifice and the head in the
process vessel. The orifice size is changed by unscrewing the plastic fitting containing the orifice and
replacing it with the required alternative. The fitting incorporates an ‘O’ ring seal and only requires hand
tightening. The alternative sizes of orifice (12) are stored in threaded holes at the front of the base plate
when not in use.

An overflow (6) in the process vessel returns water to the sump tank, preventing the process vessel from
flooding in use.

The pressure sensor measuring the level, the centrifugal pump and the solenoid valve are connected to
an electrical interface (4) that incorporates the necessary signal conditioning, allowing the process to be
operated directly from a PC using a single USB port. Alternatively the process can be operated using a
controller such as PCT54 (PID Controller) or PCT55 (Programmable Logic Controller) connected to the
electrical interface.

The computer software supplied with PCT50 allows control of the level process and data logging of the
responses using a PC. Alternatively the software allows data logging only while operating the process
remotely using a PID controller or PLC.

When filled with water, the PCT50 is self contained requiring only a mains electrical supply to the in-line
DC converter and connection to either a PC via the USB port or to a process controller such as PCT54
(PID Controller) or PCT55 (PLC with PID controller).

The unit is drained using the ball valve (11) located at the rear

.
Experimental procedure

a. On-off control

1. Choose the PID box on the mimic diagram, set the Proportional Band, Integral Time and Derivation
time to 0 then set the Set Point to 75 mm (default at start up). Click Apply to enter the changes to the
settings as shown in Figure (3).

2. Choose the Automatic Mode of Operation.

3. Select the icon to begin data logging.

4. The pump will switch on, the level will gradually rise and water will drain from the vessel via outlet
valve CV2. The level will continue to rise until it reaches the level set on the controller I.e. the setpoint of
75 mm. When the level rises above the setpoint by a fixed amount (the dead band) the pump will switch
off and the level will fall as water drains via outlet valve CV2. When the level falls below the setpoint by
a fixed amount (the dead band) the pump will switch on again and the cycle will continue.

5. In the PID controller adjust the set point to 50 mm, click Apply then observe the pump remains off
until the level falls to the new set point then switch on and off as before to maintain the new level.

6. Adjust the set point to 1000 mm and observe the response of the process.

Effect of Disturbances

1. Open the solenoid valve by clicking the button on the mimic diagram (1 indicates that the valve is
open). Observe that the level falls faster due to the additional outflow via the solenoid valve then the
controller switches on the pump to restore the original level.

2. Close the solenoid valve by clicking the button on the mimic diagram (0 indicates that the valve is
closed). Observe that the process returns to the original characteristics.
Effect of differences in Inflow Rate and Outflow rate

1. Close inlet valve CV1 so that the level only just rises when the pump is running. Allow the level to rise
and fall through several cycles then open the solenoid valve to disturb the process. If the combined
outflow exceeds the inflow then the level will continue to fall with the pump operating and the required
level cannot be maintained. Close the solenoid valve again.

2. Open inlet valve CV1 so that the level rises rapidly when the pump is running. Allow the level to rise
and fall through several cycles then open the solenoid valve to disturb the process.

3. Choose the icon to finish data logging.

4. If time permits repeat the exercise with different sizes of orifice fitted to the outlet valve CV2 and
solenoid valve, E.g. 2 mm at CV2 and 4 mm at the solenoid valve (smaller outflow with larger
disturbance). The filling characteristics with different settings of CV1 could also be investigated.

5. Create a new results sheet for each run by selecting the icon in the tool bar of the software.

b. P only controller

1. Choose the PID box on the mimic diagram, set the Proportional Band P to 200%, the Integral Time I to
0 and the Derivation time D to 0 then set the Set Point to 75 mm (default at start up). Click Apply to
enter the changes to the settings.

2. Choose the Automatic Mode of Operation.

3. Select the icon to begin data logging.

4. The pump will start, the level will gradually rise and water will drain from the vessel via outlet valve
CV2. The level will continue to rise until it reaches a steady level but this will not correspond with the
setpoint of 75 mm on the controller because of the offset inherent in a P only controller.

5. Open the solenoid valve by clicking the button on the mimic diagram (1 indicates that the valve is
open) to disturb the process. Observe that the level falls slightly, the pump speeds up slightly due to the
increased error but a large offset remains. When the level has settled close the solenoid valve and allow
the level to settle.

6. Adjust the set point to 100 mm then click apply and observe the response to a requested increase in
water level. When the level has settled, return the set point to 75 mm and observe the response.

7. Adjust the P term to 100 % in the PID controller then click Apply. Observe that the pump speeds up
and the offset reduces slightly. Open then close the solenoid valve as before to disturb the process and
observe the response, the stability of the level and any reduction in offset.

8. Adjust the P term to 50 % and repeat the above steps.

9. Adjust the P term to 20 % and repeat the above steps.

10. Adjust the P term to 10 % and repeat the above steps

11. Adjust the P term to 5 % and repeat the above steps

12. Adjust the P term to 2 % and repeat the above steps.

13. Choose the icon to finish data logging.

c. P + I controller

1. Create a new results sheet for this run by selecting the icon in the tool bar of the software.

2. Choose the PID box on the mimic diagram, set the Proportional Band P to the value selected above,
the Integral Time I to 0 seconds and the Derivation time D to 0 then set the Set Point to 75 mm. Click
Apply.

3. Select the icon to begin data logging.

4. Choose the Automatic mode of operation and allow the water level to stabilise.

5. Adjust the I term to 100 seconds in the PID controller then click Apply. Observe that the pump speed
varies to reduce the offset. When the level is settled at the set point open the solenoid valve to disturb
the process. Observe that the level falls slightly, the pump speed varies and the level gradually returns
to the set point. When the level has settled close the solenoid valve and allow the level to settle.

6. Adjust the set point to 100 mm then click Apply and observe the response to a requested increase in
water level. When the level has settled return the set point to 75 mm and observe the response.

7. Adjust the I term to 50 seconds in the PID controller then click Apply. Observe that the level returns to
the set point faster than before. When the level is settled at the set point open then close the solenoid
valve as before to disturb the process.

8. Adjust the set point to 100 mm then click Apply and observe the response to a requested increase in
water level. When the level has settled return the set point to 75 mm and observe the response.

9. Continue reducing the Integral time setting and repeating the disturbance and set point changes until
the level becomes unstable.
Results:

a. On-off control

1. Choose the icon (or View \ Graph) to display a graph of the responses obtained. Observe the rising
and falling water level about each of the set points as the pump starts and stops to maintain the
required level. Observe the change in the rate of filling when the solenoid valve is opened to disturb the
process and when the inflow is reduced by partly closing valve CV1.

2. From the graph or table of results (icon or View\Table) determine the variation in level about the set
point to determine the dead band of the controller. In a commercial controller the dead band would
usually be adjustable to suit the process but too small a variation would lead to wear of components
switching rapidly on and off.

3. From the graph observe that the rate of filling when the pump is on is not necessarily the same as the
rate of emptying when the pump is stopped.

b. P only controller

1. Choose the icon (or View \ Graph) to display a graph of the responses obtained.

2. Choose Format \ Graph Data and Plot Level L1 and Set point on 1 axis, and Controller Output on the
second axis.

3. Observe the changes in water level and the stability of the responses for each change in P setting with
a disturbance or change in set point.

4. Choose a setting for P that gives stable control with a reasonably small offset that can be used in the
next part of the exercise introducing Integral action.

c. P + I controller

1. Choose the icon (or View \ Graph) to display a graph of the responses obtained.

2. Choose Format \ Graph Data and Plot Level L1 and Set point on 1 axis, and Controller output on the
second axis.

3. Observe the changes in water level and the stability of the responses for each change in I setting with
a disturbance or change in set point.

4. Choose a setting for I that gives stable control and reduces any offset swiftly without causing the
system to become unstable.

5. Since the effect of Integral Action is related to the Proportional Band setting (reduced PB gives greater
effect for any setting of I) the above exercise should be repeated at a completely different value of PB to
determine the necessary change in I for effective control of the level and to demonstrate the
relationship between P and I.

6. If time permits repeat the exercise with different sizes of orifice fitted to the outlet valve CV2 and
solenoid valve, E.g. 2 mm at CV2 and 4 mm at the solenoid valve (smaller outflow with larger
disturbance). The filling characteristics with different settings of CV1 could also be investigated e.g.
partly closing valve CV1 to reduce the flow or increasing the flow by opening the valve.
set-p-75
100
80
60
40
20
0
00:00 00:17 00:35 00:52 01:09 01:26 01:44 02:01 02:18
-20

set-p-75

set-p-50
70

60

50

40

30

20

10

0
0 20 40 60 80 100 120 140
-10

set-p-50

set-p-1000
160
140
120
100
80
60
40
20
0
00:00 00:17 00:35 00:52 01:09 01:26 01:44 02:01 02:18
-20

set-p-1000
 set.75.p2.i15
100

80

60

40

20

0
0 20 40 60 80 100 120 140
-20

set.75.p2.i15

set,75,p2.i10
90
80
70
60
50
40
30
20
10
0
-10 0 20 40 60 80 100 120 140

set,75,p2.i10

set-p2-75
90
80
70
60
50
40
30
20
10
0
-10 0 20 40 60 80 100 120 140

set-p2-75
 set,75,p5
80

70

60

50

40

30

20

10

0
00:00 00:17 00:35 00:52 01:09 01:26 01:44 02:01 02:18
-10

set,75,p5

set75
60

50

40

30

20

10

0
00:00 00:17 00:35 00:52 01:09 01:26 01:44 02:01 02:18
-10

set75
 set,75,p2,
90
80
70
60
50
40
30
20
10
0
-1000:00 00:17 00:35 00:52 01:09 01:26 01:44 02:01 02:18

set,75,p2,

These graphs were obtained by microsoft excel and they contain a plot which consists of time vs flow
level.

Discussion:

1) No Control

Try Running the experiment with no control :

• What do you think will be the effect of altering the control valve position on the level in the
tank?
• What do you think will be the effect of altering the flowrate to the system?
• What are the implications of altering the flowrate to the tank on the required control valve size
and range of control?

➢ You should have noticed that opening the valve tended to reduce the level in the tank to a new
steady state and closing the valve brought the level to a higher steady state. Taken to the
extremes, as the valve is closed, the tank will overfill and as the valve is opened, the tank will
empty. Clearly, both of these possibilities are unacceptable indicating the need for some sort of
control.
➢ Likewise, increasing the flow (for a set valve position) brings the tank level to a new higher
steady state and decreasing the inlet flow reduces the level to a new steady state. It should be
noted that as the flowsheet shows to overfill line, the tank could overfill at high flows.
Decreasing the flow could, in the extreme, drain the tank. This is as a result of the mass balance.
(ie like filling or emptying a bath!)
➢ Increasing or decreasing the flow to the tank will require a larger or smaller valve respectively
(assuming no control) and the range of control is limited by the size of this valve.

2) On/Off Control

Try running the experiment with On/Off control

• When do you think this sort of control could prove useful and when (and why) would you
consider it impractical?
• What is the effect of altering the dead zone on the on/off controller?
• What is the effect of altering the flowrate to the tank?

➢ Since the valve position swings fully closed and fully shut very frequently, the valve would soon
wear out. This is obviously undesirable where valve reliability is important. However, it can be
used in situations like domestic water systems (see the hot water tank experiment) where tight
control of a temperature is not necessary.
➢ You should have noticed that the valve is either wide open or fully closed. The process is
controlled within a band or level range and the width of this band (and how often the valve
either opens or closes) is determined by the deadzone. The longer the deadzone, the wider the
bandwidth, but the further away the process is allowed to get from the setpoint. Also, the more
time there is between the valve swinging position to the opposite state.
➢ Another interesting observation can be seen at long deadzone values. The variation in the level
is assymmetric (this is seen by the fact the level is not a "sawtooth" around the setpoint)
indicating the this process is in fact non-linear.
➢ It should be noted that this sort of control cannot be used for tight control around a setpoint
and is not frequently used in the chemical industry.

3) Proportional Only Control

• Now try running the experiment with Proportional Only control

• What is the effect of changing the proportional band of the controller?
• Under what circumstances does the process oscillate?
• Does the process return to its setpoint?
• What would you recommend as being the best setting for the proportional controller?

You should have noticed two main features of proportional only control

➢ Increasing the proportional band decreased the amount of oscillation in the process
variable (ie increased the "damping" in the system)
➢ As proportional band was increased, so the process was controlled at a point
increasingly offset from the original setpoint.
 The effect of increasing proportional band is summarised in more detail below :

➢ At very low proportional bands the level in the tank tended to oscillate greatly. In response to
this, the valve position increasingly oscillates and would tend towards on-off control when the
valve oscillates between the fully open and fully closed position.
➢ At low proportional bands, the process was prone to oscillate before returning to a steady state
and this oscillation in process variable tended to be away (offset) from the setpoint.
➢ At very high proportional bands, the process was quickly returned to a steady state and with
very little oscillation but at a point significantly different from the original setpoint. The actual
significance of the proportional band can be understood in relation to the range of the
measurement sensor (here the level sensor) and the range of operation of the control valve. A
proportional band of 100 indicates that the control valve will move over 100% of its range for a
100% measurement sensing range.

Proportional Integral Control

You should have noticed that using proportional only control has a number of disadvantages. Using
your chosen optimal value of proportional band found from the previous experiment (or altering it if
you think you got it wrong!), try running the experiment with the full Proportional and Integral (P+I)
control and think about the following questions :

• What problems associated with Proportional control are eliminated by using P+I control?
• What is the effect of changing the Proportional Band now?
• What is the effect of changing the Integral (reset) time?
• What would you recommend as being the best settings for the PI controller?

➢ At low values of integral action (reset) time and low proportional bands, the system
is prone to significant oscillation. This large amount of oscillation means that the
control tends towards on/off control as the valve swings from fully open to fully
closed.
➢ As the proportional band is increased the oscillation in the system becomes less and
the process is successfully returned to the setpoint (recall with proportional only
control, there was an "offset"). However, the process is away from the setpoint for a
relatively long period of time.
➢ As the integral action (reset) time is increased, so the system is returned to the
setpoint more quickly. However, at high proportional bands and high integral action
(reset) times. the system again takes a long time to return (and in fact it may not) to
the setpoint as the system is "overdamped"
➢ The main advantage of P+I control is that offset is eliminated. The combination of
correct proportional bands and a correct integral action (reset) time produces a
quick response to a disturbance in the process that returns the process to the
setpoint without offset.
References:
1. R. N. Bateson, Introduction to Control System Technology, Prentice-Hall, Upper
Saddle River, NJ, 1999, 6th Ed., pp. 304-307.
2. F. G. SHINSKEY, process control system , McGraw-Hill Book Co- ,1995