You are on page 1of 17

CDB 3062

CHEMICAL ENGINEERING LABORATORY III


MAY 2016

EXPERIMENT : 9 – LEVEL FLOW CONTROL

GROUP : 17

GROUP MEMBERS : ROSLINA AIFAR BINTI ROSLI 19414


MOHD NOOR HADEEF BIN NOOR AZLAN 19431
KOI ZI KANG 18868
MUHAMMAD AMIRUL MUKMIN BIN 19030
NUR AMALINA BINTI DAWI CAHYONO 19092
LAB INSTRUCTOR : Ms. Mansore

DATE OF EXPERIMENT : 30th MAY 2016


Table of Contents
1.0 Summary ........................................................................................................................................... 1

2.0 Theory ............................................................................................................................................... 2

3.0 Objectives ......................................................................................................................................... 6

4.0 Procedures ........................................................................................................................................ 6

5.0 Results ............................................................................................................................................... 8

6.0 Discussions ...................................................................................................................................... 12

7.0 Conclusion ....................................................................................................................................... 14

8.0 References………………………………………………………………………………………………………………………………...15
1.0 SUMMARY

The level flow trainer is an interesting system that can help us to understand various flow
mechanism of water. The objectives of this experiment are to study and illustrate the level control
and the effect of different control mechanism to the flow of water.

During our experiment, we only managed to conduct test disturbance using Proportional
Only Control (P) and Proportional-Integral and Derivative Control (PID). . For Proportional Only
Control (P), this type of control always has steady-state error and maximum overshoot as well as
long settling time. Meanwhile, the Proportional Integral and Derivative Control (PID) has no steady
state error and smaller overshoot compared to other type of control. 3 types of disturbance tests
were carried out for PID control with 2 different settings. Those tests include set point set change,
load step test and load pass disturbance test.

From the result that we obtained, the lower the set time value and the lower the
proportional band of the system, the higher the efficiency of the response of the system to the set
point step change and load step test but not to load pass disturbance test. Besides, we found that
the PID controller is very useful in reducing the error and number of oscillation which eventually
producing such a very smooth graph. However, the PID controller is not being able to detect a short
and sudden disturbance therefore the graph could not show the differences clearly. In addition,
when we compared the graph of P control mode and PID control mode, we can see that the graph
with P control mode shows a very minimal change for a long period of time and it does not eliminate
steady state error. In contrast, the PID control mode illustrated rapid response and produced
smooth graph which means that PID controller produces a more satisfactory performance.

1|Page
2.0 THEORY

There are several types of process control mode as listed below:


 On-Off Control
 Proportional Control
 Integral Control
 Derivative Control
 Proportional-Integral Control
 Proportional-Integral-Derivative Control

2.1 On-Off Control

In this type of control experiment, the following steps are need to be followed to ensure that the
water level in the tank are closed to the setpoint level.

i. The level in the tank is measured and compared with the setpoint level.
ii. If the actual level is below the setpoint level by a specified amount then the outlet control
valve is shut.
iii. If the actual level is above the setpoint level by a specified amount then the outlet control

valve is opened.

The specified amount is known as the deadzone and defined as a percentage of the total height of
the tank in the experiment. The graphs below show that if the deadzone is small then the level is
close to its setpoint and the control valve is being opened and closed continuously. On the other
hand, if the deadzone is large then there is not as much opening and closing of the valve and the
level is not very close to the setpoint.

Small Deadzone
Large Deadzone

2|Page
2.2 Proportional (P) Controller

To start with, a proportional (P) controller is engaged by toggling Control Mode from Manual to
Automatic. The control signal, u is computed as

𝑢(𝑡) = 𝐾(𝑟(𝑡) − 𝑦(𝑡))

where r is the reference and y the measurement. In this case this means that the pump voltage
becomes proportional to the control error 𝑒 = 𝑟 − 𝑦. The constant K is the gain of the controller.

In P controller, the steps which describes the action of the controller is:

i. The level in the tank is measured and compared to the setpoint.


ii. The resulting compensatory action of the controller is proportional to the error between the
above two values.

The variable which determines the proportionality is known as the proportional band (PB). The
relationship between the adjustment and error is:

100 𝑅𝑎𝑛𝑔𝑒 𝑜𝑓 𝑀𝑒𝑎𝑠𝑢𝑟𝑒𝑑 𝑉𝑎𝑟𝑖𝑎𝑏𝑙𝑒


𝐴𝑑𝑗𝑢𝑠𝑡𝑚𝑒𝑛𝑡 = ( × 𝐸𝑟𝑟𝑜𝑟 × ) + 𝑉𝑎𝑙𝑢𝑒 𝑎𝑡 𝑁𝑜 𝐸𝑟𝑟𝑜𝑟
𝑃𝐵 𝑅𝑎𝑛𝑔𝑒 𝑜𝑓 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑 𝑉𝑎𝑟𝑖𝑎𝑏𝑙𝑒

 Small PB will show small offset and results in more oscillatory response as the adjustment
will be sensitive with the changes of error.
 Large PB will results in large offset but a smoother response as the adjustment is not so
sensitive.

Small Proportional Band Large Proportional Band

3|Page
The value at no error always stays constant. If there are changes to the conditions, either a setpoint
change or other disturbance, then there must be a corresponding change in the adjustment.
However, for any adjustment there must be an error. This leads to one of the disadvantages of using
proportional control. At the new steady state there has to be an error or offset.

2.3 Integral (I) Control

Integral control is also known as reset or floating control. For integral action, the controller output
depends on the integral of the error signal over time,
1 t
p(t)  p 
I  e(t )dt 
0

where 𝜏i is an adjustable parameter and referred to as the integral time constant or reset time, has
units of time. The advantage of Integral controller is that it eliminates the offset. For the process
being controlled to be at steady state, the controller output p must be constant so that the
manipulated variable is also constant. The control action by the integral controller is very little until
the error signal has persisted for some time. On the other hand, proportional controller takes
immediate corrective action as soon as an error is detected.

2.4 Derivative (D) Control

Derivative control is also known ad rate action or anticipatory control in which it anticipates the
future error by considering its rate of change. For ideal derivative action,

de(t )
p (t )  p   D
dt

where  D is the derivative time, and has units of time. As long as the error is constant de/dt = 0, the
controller output is equal to p . However, derivative action is never used alone. It is always used in
conjunction with P or PI control in which the derivative control action tends to stabilize the
controlled process.

2.5 Proportional-Integral (PI) Control

Proportional Integral controller is the controller that works by summing the current controller error
and the integral of all previous errors.

1 1
𝑢(𝑡) = 𝐾(𝑒(𝑡) + ∫ 𝑒(𝜏) 𝑑𝜏)
𝑇𝑖 0

4|Page
where e is the control error, i.e., 𝑒 = 𝑟 − 𝑦. The pump voltage is now given as the sum of two
terms. The first one consists of a constant K times the control error and is called the P part (cf. P
controller). The second term is a constant K/Ti times the integral of the control error. This part of the
sum is hence referred to as the I part (integral part) and changes as long as the reference and output
differ. It could simply be thought of as a way of continuously adjusting the value at no error so as to
eliminate the offset.

The new variable introduced to define the amount of integral action is known as the integral reset
time of the controller.

 Smaller reset time results in more accurate response as the adjustment is less sensitive to
the change of error.
 Larger reset time results in less accurate response as the adjustment is more sensitive to the
changes of error.

Small Reset Time Large Reset Time

Thus, the advantage of PI controller is that it can eliminate the offset. The disadvantages of PI
controller is that it give rise to a higher maximum deviation, a longer response time and a longer
period of oscillation than with P controller. This type of control action is therefore used where the
above can be tolerated and offset is undesirable.

2.6 Proportional-Integral-Derivative (PID) Control

To achieve good control performance, additional insight into the system might sometimes be
needed. For example, the derivative of the control error gives an estimate of future error values. By
letting the control signal depend also on the error derivative, one obtains a control law where the
control signal increases when the error is growing and vice versa. This can be used to counteract

5|Page
growing control errors in a faster manner and give smoother control action in the vicinity of the
reference. If we expand the control law to incorporate derivative action, we obtain the PID
controller with control law
1 1 𝑑𝑒(𝑡)
𝑢(𝑡) = 𝐾(𝑒(𝑡) + ∫ 𝑒(𝜏) 𝑑𝜏 + 𝑇𝑑 )
𝑇𝑖 0 𝑑𝑡
𝑑𝑒
The control signal now consists of a P part, and I part and a D part. The D part is given by (𝐾𝑇𝑑 𝑑𝑡
).

The constant Td is the derivative time of the controller. It can be interpreted as the prediction
horizon for a linear error estimator.
Thus, the advantage of PID controller is that it shows a rapid response with no offset. However, PID
controller is the most complex and difficult to tune, and it will be the best controller if it is properly
tuned.

3.0 OBJECTIVES

 To study the response of level control process.


 To study the effect of different control mechanism.

Problem Statement: How will a process respond towards different controller settings in a PID
control mode and how different will the process respond in a P control mode?

4.0 PROCEDURES

1) It was made sure that we wore proper safety attire including the safety helmet before the
experiment was begun.
2) A start-up checklist was done and preliminary exercises was conducted with the GA.
3) Tank 32 was filled with water. The tank was toped up with water whenever necessary.
4) In the beginning, Pump 33 was left to manual over-ride mode.
5) The inflow pump was fully opened while. For the outflow pump, its manual suction valve was
fully opened but its manual parallel discharge valves was shut fully.
6) The two manual discharge valves of outflow pump was shut off.
7) The main power supply was switched on.
8) Initially, the set point (SV1) of LIC31 was set to 400mm. The manual flow control was adjusted
to prevent overflow of the system. The experiment was begun once the reading reached a
steady state and stable value close to SV1.

6|Page
9) For part f (i), set point step test was conducted. In manual mode, the MV was set to 75% and
the first trial where PB1= 30%, TI1=25 sec and TD1= 0 was set on the manual setting board.
10) The graph sheet was adjusted and a label was written to indicate the beginning of first trial.
11) The auto mode was pressed on. The process was stopped by pressing manual mode button
mode when the level stops fluctuating and reached the set point SV1.
12) The process was repeated twice with an increase of 20% to the original SV and a decrease of
20% to the original SV subsequently. The sheet was labelled every time a new trial was begun.
13) In the next part of the experiment f (ii), a load step test was conducted. Manual mode was
switched on. The set point SV1 was maintained at 400mm while the MV was increased by 10%
of the original MV. The sheet was labelled again when the auto mode was on. The process was
stopped by switching on manual mode when the response reached steady state. The process
was repeated with a decrease to the MV by 10%.
14) For part f (iii), the pump bypass manual valve B32 was opened fully for 5 seconds before it was
closed again and the graph sheet was labelled once the valve was opened.
15) Step 9 to step 14 was repeated with second trial PID1 settings in which PB1 = 10%, TI1 = 15 sec
and TD1 = 0 sec.
16) For part h, after manual mode was switched on, Proportional (P) control mode was tested by
setting PB1 = 10%, TI1 = 9999 sec (max) and TD1 = 0 sec. The graph sheet was labelled once
the auto mode was on. The process stopped in the middle of the process due to a very long
period with minimal changes in the response.

7|Page
5.0 RESULTS

Graph 5.1: Initial Startup of the System till Steady State Condition

Graph 5.2: Set Point Step Increase (Trial I)

Graph 5.3: Set Point Step Increase (Trial II)

8|Page
Graph 5.4: Set Point Step Decrease (Trial 1)

Graph 5.5: Set Point Step Decrease (Trial II)

Graph 5.6: Load Step Increase (Trial I)

Graph 5.7: Load Step Increase (Trial II)

9|Page
Graph 5.8: Load Step Decrease (Trial I)

Graph 5.9: Load Step Decrease (Trial II)

Graph 5.10: Load Pass Disturbance Test (Trial I)

Graph 5.11: Load Pass Disturbance Test (Trial II)

10 | P a g e
Graph 5.12: Proportional (P) Control Mode

11 | P a g e
6.0 DISCUSSIONS

For this level flow trainer experiment, we only carried out test disturbance for the objective
of studying level control. The test disturbance was tested using two types of control: P and PID only.
Disturbance is the process transient behavior when a disturbance enters which is called as
regulatory control or load change and the control system should be able to return each controlled
variable back to its set point.

For the PID test disturbance, we carried out two trials with different setting values for PB1
and TI1 whereby for the first trial, PB1 = 30%, TI1 = 25secs, TD1 = 0 sec whereas for second trial, PB1
= 10%, TI1 = 15 sec and TD1 = 0 sec. For each trial, 3 types of tests were carried out: set point step
test, load step test and load pass disturbance test. The results for each respective test were
compared between both the trials.

For part f (i) which is set point step test, the step increase and decrease of the set point
which was initially at 400mm is 20% (±80mm). Based on the graphs 5.2, 5.3, 5.4 and 5.5, it is distinct
that the response of the set point step increase and decrease for trial 2 is sharper than that of trial 1
only that the response of set point step decrease for trial 2 takes a longer time to reach its steady
state probably due to an extreme drop of the level flow. This shows that trial 2 is more efficient in
terms of the response because the set time of trial 2 is lesser than trial 1 which means trial 2 is more
responsive to the step change. Also, the proportional band of trial 2 is lower meaning that the gain
for the process is higher which results in a higher amplitude and rapid response. So from f(i) we can
say that trial 2 provides a more rapid response than trial 1 but both trials show a smooth response
with little oscillations to the set point change.

For part f (ii) which is load step test, we step changed the MV quickly by ±10%. This
stimulates a load step test at the inflow. A load step test is provides information about the stability
of the response. As from graphs 5.6, 5.7, 5.8 and 5.9 we can see that the step increase and decrease
of trial 2 is sharper than that of trial 1 and the amplitude of trial 2’s response is larger than trial 1’s
response. This is because the set time of trial 2 is lesser than that of trial 1 which provides trial 2 a
more responsive condition to step change. Also, the lesser proportional band provides trial 2 a larger
gain which results in a higher amplitude. Therefore, trial 2 is more responsive than trial 1 in terms of
load step test but overall both trials give a smooth response with little oscillation.

For part f (iii) which is a load pass disturbance test, the pump bypass manual valve B32 was

12 | P a g e
shut fully for 5 sec. From both graphs 5.10 and 5.11, there is no significant response from both
graphs towards this disturbance. This means that the disturbance is not impactful enough to cause
any significant response and the PID controller settings are not responsive enough to detect such a
short and sudden disturbance.

For part h in which we switch the control into only Proportional (P) control mode by setting
PB1 = 10%, TI1 = 9999 sec (max) and TD1 = 0 sec. As we can see from graph 5.12, the response of the
control is very slow and there’s very minimal change for a long period. As for comparison, PID
control yields a more rapid and smooth response because PID control is a combination of three
modes: proportional (fast reaction on change of the controller input), integral (increase in control
signal to lead error towards zero) and derivative (suitable action inside control error area to
eliminate oscillations). The PID controller makes a control loop respond faster with less overshoot
and most popular method of control by a great margin. In contrast, P controller only accounts for
reducing error signal to zero and it can only stabilize 1st order unstable process. P controller cannot
eliminate offset or steady state error after a set point change or a sustained disturbance. P controller
will also increase the maximum overshoot of the system which is undesirable.

Overall, test disturbance using P and PID controller provides information about the
smoothness and responsiveness of a controller as well as the setting of the controller itself in order
to select the best controller settings for an optimal and satisfactory performance of the process.

ERRORS & RECOMMENDATIONS

1) Error: The graph sheet was constantly printing at an incorrect angle. This makes it hard for
us to determine the proper angle or slope of the graph.
Recommendation: Improve the printer’s design by adding holders for it to print out straight.

2) Error: We had to guess and mark the starting and the end of each disturbance point
ourselves, which gives us an inaccurate point of beginning.
Recommendation: Design a button that we can press in order to tell the machine to mark the
point or automatically mark the point when we have reset the MV or SV1 value.

3) Error: The water level constantly fluctuates which makes it harder to determine if it has
reached the desired set point or not.
Recommendation: Wait for the water level to stabilize before stopping the experiment.

13 | P a g e
7.0 CONCLUSION

In this experiment, test disturbance is the only test we carried out to study level control. The
types of controller used for the test are P and PID only. For PID controller, we had two trials with
setting values of PB1 = 30%, TI1 = 25 sec, TD1 = 0 sec for the first trial and PB1 = 10%, TI1=15 sec, TD1
= 0 sec for the second trial. Set point step test, load step test and load pass disturbance test are the
tests that were carried out for both trials.

For set point step test, the set point will set to increase and decrease by 20% from the default
value of 400mm (±80mm). The graphs produced by this test shows that the graph of the second trial
depict a sharper change in values compared to the first trial. We can conclude that the second trial is
much more responsive than that of the first trial, but with similar performances in terms of oscillation
frequency towards the set point change.

For load step test, the value of MV is increased and decreased subsequently by 10% to
stimulate a load step test at the inflow. The graphs produced by this test also depicts that the second
trial results are much more responsive than that of the first trial so we can conclude that the second
trial settings produce a more efficient and satisfactory performance.

For load pass disturbance test, the pump bypass manual valve B32 was shut fully for 5 sec.
The graphs produced by this test depicts that there is no significant response of disturbance produced
by both graphs. It can be concluded that the PID controller settings are not calibrated well to detect
such a short and sudden disturbance.

For P controller, the settings are such that PB1 = 10%, TI1 = 9999 sec (max) and TD1 = 0 sec.
The response produced from the test is very slow in terms of period of time and there were only
minimal changes occurring within long period of time. This is because P controller could only reduce
error signal to zero and only stabilize 1st order unstable process and could not eliminate offset or
steady state error after a set point change or a sustained disturbance. Moreover, P controller has the
undesirable tendencies to increase the maximum overshoot of the system.

In conclusion, the disturbance tests show that P and PID controller with different control
settings do produce different response which allow us to study and identify the best settings to
achieve an optimal and satisfactory performance of the process. All in all, the objective to study level
control is achieved.

14 | P a g e
8.0 REFERENCES

Electrical Engineering and Technology. Retrieved from http://www.electrical4u.com/types-of-


controllers-proportional-integral-derivative-controllers/

Liptak, B. G. (1995). Process Control: Instrument Engineers’ Handbook 3rd Edition.

Rao, K. S. & Mishra, R. (2014). Comparative study of P, PI and PID controller for speed control of VSI-
fed induction motor. International Journal of Engineering Development and Research, 2 (2),
2740 - 2744.

Seborg, D. E., Edgar, T. F., Mellichamp, D. A. & Doyle, F. J. (2004). Process Dynamics and Control (3rd
ed.). United States of America: John Wiley & Sons, Inc.

15 | P a g e

You might also like