You are on page 1of 29

MONASH

University Malaysia

CHE 3162 Process Control Laboratory

CHE 3162 Process Control


Laboratory
Liquid Flow Control
Name

Mah Wei-Jun
How ZhongXing
Yang Ge Hoa
Ang Lin Yang














Student ID
24640093
24554758
24686727
24133027

MONASH University Malaysia

CHE 3162 Process Control Laboratory

Table of Contents
1.0 Summary ................................................................................................................................................................ 4
2.0 Introduction .......................................................................................................................................................... 6
3.0 Objectives .............................................................................................................................................................. 7
4.0 Experimental Procedure ................................................................................................................................. 7
4.1 Diagram of Apparatus ................................................................................................................................ 7
4.2 Description of Apparatus ........................................................................................................................... 8
4.3 Methodology .................................................................................................................................................... 8
4.3.1 Experimental Start-Up ........................................................................................................................... 8
4.3.2 Closed Loop Flow Control .................................................................................................................. 8
4.3.2.1 Proportional Controller ...................................................................................................................... 8
4.3.2.2 Proportional-Integral (PI) Controller ............................................................................................ 9
4.3.2.3 Proportional-Derivative (PD) Controller ..................................................................................... 9
4.3.2.4 Proportional-Integral-Derivative (PID) Controller .................................................................. 9
4.3.2.5 Step Test Closed Loop Tuning Method ....................................................................................... 9
4.3.2.6 Experimental Shut-Down .............................................................................................................. 10
5.0 Results and Analysis ................................................................................................................................... 11
5.1.0 Closed Loop Flow Control .................................................................................................................. 11
5.1.1 Proportional Controller .................................................................................................................. 11
5.1.2 Proportional-Integral (PI) Controller ...................................................................................... 13
5.1.3 Proportional-Derivative (PD) Controller ................................................................................ 15
5.1.4 Proportional-Integral-Derivative (PID) Controller ........................................................... 17
5.2 Step Test Closed Loop Tuning Method ............................................................................................ 19
5.3 Calculation Step .......................................................................................................................................... 20
6.0 Discussion ........................................................................................................................................................... 23
6.1 Define proportional gain, integral gain and derivative gain ................................................... 23
6.2 The effects of proportional gain applied on the proportional-only control system and
the effects to response of the system. ......................................................................................................... 23
6.3 Discuss the effect of integral gain applied on the proportional-integral control system.
Relate the effect to the response of the system (e.g. rise time, overshoot percentage,
settling time and steady state error). ........................................................................................................ 24
6.4 The effects of derivative gain applied on the proportional-derivate control system and
the effects to response of the system. ......................................................................................................... 25

MONASH University Malaysia

CHE 3162 Process Control Laboratory


6.5 Discuss the system response curve of PID controller tuned and compare it with the
system response curve obtained from proportional controller, proportional-integral
controller and proportional-derivative controller. ............................................................................. 26
6.6 Discuss whether the step test closed loop tuning method is suitable for all types of
process control. ................................................................................................................................................... 27
7.0 Conclusion .......................................................................................................................................................... 28
8.0 Nomenclature .................................................................................................................................................... 29
9.0 Reference ............................................................................................................................................................ 29

MONASH University Malaysia

CHE 3162 Process Control Laboratory

1.0 Summary
Throughout the conduction of this experiment, a detailed application of how the
closed-loop proportional-only control (P control), closed-loop proportional-derivative
control (PD control), closed-loop proportional-integral control (PI control) and a closedloop proportional-integral-derivative control (PID control) are used in the regulations of
maintaining the liquid level at a fixed level inside a reactor or a tank is seen. Basically, a
proportional control will provides a fast response for adjustments made on the
manipulated variable, however it will always have an offset whereas for the proportionalintegral (PI) control it is able to remove the offset thus making the magnitude of error to
be zero. As for the proportional-integral-derivative (PID) control, it was proven to be the
best device among the other controllers as it can prevent large value overshoots by
removing offset values. Besides that, it also reduces the settling time if optimum values
of proportional, integral and derivative gains are entered to the controller.
Four different combinations of control system were performed in these
experiments, which are primarily made up from proportional (P), integral (I) and
derivative (D). These controllers are then used for different combinations of P, PI, PD
and PID. Among the combinations made, the combination that was able to provide the
least or shortest time to reach the steady state was chosen and used for our tuning on the
step test closed-loop system.
For the proportional (P) control system experiment, the Ki and Kd values were set
to zero while Kp is being varied. Results showed that when the Kp value was too small,
the offset of the system will be large and when the Kp was set to a large value, the system
will become unstable as it starts oscillating. However, a lower Kp value will require a
shorter time to achieve steady state compared to a higher Kp value. For the proportionalintegral (PI) control system, Kp was set to 5, Kd was set to zero while Ki was being
altered. In the tuning process, it was concluded that when Ki = 0.01, the response curve
was the best compared to the response for other values of Ki.
As for the proportional-derivative (PD) control system, values of Kp and Ki were
set to 5 and zero respectively with Kd being tuned to different values. Based on results
attained, it was concluded that the system was unable to achieve the set point value and
the system was starting to become less stable as oscillations were increasing when Kd was
changed from a smaller value to a bigger value. After series of observations and result
analysis that were made, values of Ki and Kd that gave the best response graph in the PI
and PD experiments was chosen for the proportional-integral-derivative (PID) control
system experiment. With these optimum values, the best response curve for the PID
control can be accomplished.
Lastly, series of calculations were carried out using optimum values of Kp, Ki and
Kd that was determined previously for the PID control system. The process gain, Kp was
calculated to be 0.06, time constant, Tp to be 0.0167 minutes while the dead time, P was
determined to be 0.0333 minutes by considering calculations based on 80% and 20%
valve openings as these 2 openings gave the smoothest flow in step change loop tuning.

MONASH University Malaysia

CHE 3162 Process Control Laboratory


To conclude this experiment, the PID control system gave the best response curve
as the overshoot was smaller, time required to achieve steady state was shorter and a
stable system without oscillations. However, values of Kp, Ki and Kd have to be altered
based on the process as different set points will have their own optimum Kp, Ki and Kd
values.

MONASH University Malaysia

CHE 3162 Process Control Laboratory

2.0 Introduction
Generally, every chemical processes will have some disturbances along its operation.
These processes usually deals with hazardous and dangerous chemicals and thus, process
control are necessary to ensure the safety of operation and also to maintain the quality of
the desired product. The simplest forms of both automatic and manual control rely on
adjusting a manipulated variable (MV) in order to compensate for observed undesired
variations of the process variable (PV) or output.
Basic elements that make up a process control system in industries consist of a controller,
process, measuring element and control element. A closed loop control exists when all
these components are interlinked and information can be passed around the loop.
Currently, the Proportional-Integral-Derivative (PID) algorithm is the most sort-after
algorithm used in industry. Some of the most common processes that require the usage of
PID are fluid flow monitoring, heating and cooling systems, flow control and temperature
control. Prior to using the PID control, we need to define a set point and a process
variable. The set point is the desired value of the controlled parameter while the process
variable refers to the system parameter that needs to be controlled such as the pressure,
temperature and flow rate. A PID controller determines a controller output value, such as
the heater power or valve position and applies the controller output value to the system,
which in turn drives the process variable towards the set point value.
In this experiment, the main function of the process control unit (LS-33 139 Basic Flow
Control Unit) is to control the water flow rate in the system. Firstly, a pump will deliver
water from a storage tank through a piping system. Next, the flow rate will be measured
by the flow sensor and the value will be fed back to the control valve to regulate the
incoming flow rate.
All the elements and units in this experiment represents a scaled down process model of a
common industrial process. Experiments were conducted on various closed loop control
systems as listed below.
(i)
Proportional Controller a type of linear feedback control system; controller
output signal is proportional to the error input signal.
(ii)
Proportional-Integral Controller two-term controller; integral action
eliminates offset and makes the control system less stable.
(iii)
Proportional-Derivative Controller two-term controller; derivative section
responds to the rate of change of the process error.
(iv)
Proportional Integral-Derivative Controller a generic control loop feedback
mechanism; a PID controller calculates an error value as the difference
between a measured process variable

MONASH University Malaysia

CHE 3162 Process Control Laboratory

3.0 Objectives
The objective of this experiment was to accomplish and develop understanding in the
basic flow control process through operation of the process control unit, which controls
the water flow rate in the system. In addition, responses from different PID settings in the
closed loop flow control are studied and interpreted. On top of that, the step test closed
loop tuning method in the flow control process is being demonstrated and analysed.

4.0 Experimental Procedure


4.1 Diagram of Apparatus

Water
Storage Tank

Figure 1: LS-33 139 Basic Flow Control Unit

MONASH University Malaysia

CHE 3162 Process Control Laboratory

4.2 Description of Apparatus


The equipment shown above is the process control unit that is designed to control the
flow rate of water in the system. The pump located in water storage tank is used to
transfer water through a piping system. The water flow rate is then measured by a flow
sensor located along the pipe. This is followed by the transmission of the value which
will be fed back to the control valve to adjust the incoming flow rate in order to achieve
the required set point value.

4.3 Methodology
4.3.1 Experimental Start-Up
Before the experiment began, all the equipment was checked to ensure there are in good
and safe condition. Firstly, the process control unit was placed on a level surface. Also,
all the pipes and fittings were checked to ensure it is well-connected to prevent any
leakage of water. Next, the computer was turned on. Then lastly, it was followed by the
starting up of the LS-33 139 Basic Flow Control DAQ Software in the computer. Before
the main power supply and the apparatus power supply (on the control panel of the
equipment) was switched on, the water level in the storage tank was observed to be
approximately of the tank.
4.3.2 Closed Loop Flow Control
4.3.2.1 Proportional Controller
*The following procedures were carried out through the DAQ software on the
computer.*
First, the initial set point was set to 4LPM in the DAQ software. Then, the experiment
was continued by setting KP = 1, KI = 0 and KD = 0. To initiate the software, the
Run/Execute button was clicked. In order to record the data obtained, The Logging
Started button was clicked. To allow the traceability of results in future, the time the
experiment started was recorded. The submersible water pump was switched on. Then,
the system was allowed to run for few minutes and the response of the system from the
graph was observed. Next, the Data Analysis button was clicked to access all recorded
data and the response graph. Print Screen button was used to record the system response
graph. The experiment was again repeated with equal increment in the Proportional Gain
KP (4, 9, 13 and 17) and the system response to different Proportional Controller was
observed. Each time the KP value was altered, the time was recorded. The proportional
gain was increased until the response starts to oscillate or to a point where the offset
between the set value and process value was unable to be further decreased. Then, the
data and response graph were recorded. The water pump was turned off before pressing
the Stop button at the end of this experiment.

MONASH University Malaysia

CHE 3162 Process Control Laboratory


4.3.2.2 Proportional-Integral (PI) Controller
*The following procedures were carried out through the DAQ software on the
computer.*
The second part of the experiment was continued by setting the desired set value flow
rate to 4 LPM. Next, the proportional gain was changed to the critical gain (KP = 5) and
KI was altered to 0.01. The software was then started by pressing the Run/Execute
button. Subsequently, the Logging Started button was clicked on to record the data.
The time the experiment started was recorded. Then the submersible water pump was
switched on. After allowing the system response to stabilize, the data was recorded and
the Print Screen button was again used to save the generated response graph. The
submersible water pump was turned off and the flow output was allowed to drop till 0
LPM. The experiment was repeated by varying the Integral gain, KI to values of 0.001,
0.01, 0.05, 1 and 10.
4.3.2.3 Proportional-Derivative (PD) Controller
*The following procedures were carried out through the DAQ software on the
computer.*
The third part of the experiment was started by setting the desired flow rate set value to 4
LPM. The proportional gain was set to the critical gain (KP = 5), while KI = 0 and KD =
0.01. The program was allowed to run and the response was observed. Before the data
could be recorded, the system response was allowed to stabilize. Subsequently, the data
was recorded and the Print Screen button was clicked to obtain the generated response
graph. Then the submersible water pump was turned off and the flow output was allowed
to drop till 0LPM. The experiment was repeated with increasing derivative gain, KD to
values of 0.02, 0.2, and 2.
4.3.2.4 Proportional-Integral-Derivative (PID) Controller
*The following procedures were carried out through the DAQ software on the
computer.*
By comparing the system response graphs generated previously, the KI and KD values
were altered to obtain the best system response by retuning the PID controller. Several
trials were carried out and the different system response graphs were compared
interpreted with each other to conclude the best combination of PID controller.
4.3.2.5 Step Test Closed Loop Tuning Method
*The following procedures were carried out through the DAQ software on the
computer.*
The initial set point was set to 4LPM in the DAQ software. The KP, KI and KD values
were determined to be 5, 0.01 and 0.01 respectively as these values provide the best
system response graph obtained previously. The software was again started by pressing
the Run/Execute button. After that, the submersible water pump was switched on and
the system was allowed to stabilize. The Open Loop button was clicked and the sliding
bar was adjusted to the maximum (100%). At the same time, the Logging Started
button was clicked on. The water was allowed to flow for 20 seconds for stabilization of
the system to be achieved. This was followed by adjusting the sliding bar to

MONASH University Malaysia

CHE 3162 Process Control Laboratory


approximately 90%. The steps stated were repeated with equal decreasing valve opening
(approximately decrement of 10%) until the valve was fully closed. Then, the Logging
Started button was clicked to stop recording the data. Next, from the system response
graph obtained, two different valve opening percentage which gave the smoothest flow
were selected. Subsequently, the Logging Started button was clicked followed by the
Open Loop button. The sliding bar was adjusted to the higher valve opening percentage
value of the two and the system was allowed to run for 20 seconds. Then, the sliding bar
was changed to the lower valve opening percentage value and again the water was
allowed to flow for 20 seconds. The data shown was recorded and analyzed carefully.
Lastly, calculations were performed according to the lab manual to determine the P, I and
D values for the closed loop control.
4.3.2.6 Experimental Shut-Down
The water pump was switched off, followed by the pressing of STOP button at the end of
the experiment. The DAQ software was closed and the control panel power was turned
off. All the .jpg files were transferred to USB drive. Lastly, the computer and the
process control unit was shut down.























10

MONASH University Malaysia

CHE 3162 Process Control Laboratory

5.0 Results and Analysis


5.1.0 Closed Loop Flow Control


5.1.1 Proportional Controller
Set point value = 4LPM
The response graphs for proportional flow control are shown below.

Figure 2: Proportional Controller response graph for Kp = 1, KI = 0, KD = 0

Figure 3: Proportional Controller response graph for Kp = 5, KI = 0, KD = 0




11

MONASH University Malaysia

CHE 3162 Process Control Laboratory

Figure 4: Proportional Controller response graph for Kp = 9, KI = 0, KD = 0


Figure 5: Proportional Controller response graph for Kp = 13, KI = 0, KD = 0


Figure 6: Proportional Controller response graph for Kp = 17, KI = 0, KD = 0





12

MONASH University Malaysia

CHE 3162 Process Control Laboratory


5.1.2 Proportional-Integral (PI) Controller
Set point value = 4LPM
The response graphs for PI flow control are shown below.

Figure 7: Proportional-Integral Controller response graph for Kp = 5, KI = 0.001, KD = 0


Figure 8: Proportional-Integral Controller response graph for Kp = 5, KI = 0.01, KD = 0




13

MONASH University Malaysia

CHE 3162 Process Control Laboratory

Figure 9: Proportional-Integral Controller response graph for Kp = 5, KI = 0.05, KD = 0


Figure 10: Proportional-Integral Controller response graph for Kp = 5, KI = 1, KD = 0


Figure 11: Proportional-Integral Controller response graph for Kp = 5, KI = 10, KD = 0





14

MONASH University Malaysia

CHE 3162 Process Control Laboratory


5.1.3 Proportional-Derivative (PD) Controller
Set point value = 4LPM
The response graphs for PD flow control are shown below.

Figure 12: Proportional-Derivative Controller response graph for Kp = 5, KI = 0, KD = 0.01


Figure 13: Proportional-Derivative Controller response graph for Kp = 5, KI = 0, KD = 0.02


15

MONASH University Malaysia

CHE 3162 Process Control Laboratory

Figure 14: Proportional-Derivative Controller response graph for Kp = 5, KI = 0, KD = 0.2


Figure 15: Proportional-Derivative Controller response graph for Kp = 5, KI = 0, KD = 2




















16

MONASH University Malaysia

CHE 3162 Process Control Laboratory


5.1.4 Proportional-Integral-Derivative (PID) Controller
Set point value = 4LPM
After concluding findings from previous controller set up experiments, the controller is
then retune by adjusting values of Kp, KI, and KD to determine the best PID value for the
best system response.
The response graphs for PID controller are shown below.

Figure 16: PID Controller response graph for Kp = 5, KI = 0.01, KD = 0.005

Figure 17: PID Controller response graph for Kp = 5, KI = 0.01, KD = 0.01



17

MONASH University Malaysia

CHE 3162 Process Control Laboratory

Figure 18: PID Controller response graph for Kp = 5, KI = 0.01, KD = 0.05


Figure 19: PID Controller response graph for Kp = 5, KI = 0.05, KD = 0.01

Figure 20: PID Controller response graph for Kp = 5, KI = 0.10, KD = 0.01

18

MONASH University Malaysia

CHE 3162 Process Control Laboratory

5.2 Step Test Closed Loop Tuning Method


Set point value = 4LPM

Figure 21: Step change response graph for 10% valve opening decrement from 100%

Figure 22: Response graph of 2 valve openings for smoothest flow (80% and 20%)











19

MONASH University Malaysia


CHE 3162 Process Control Laboratory

5.3 Calculation Step


1) Process Gain, Kp

,
! =

,
! =

!"#$% !"!#!$%

!"#$% !"!#!$%

Values for flow rate at the respective valve opening percentage are obtained from Figure 22.
! =

2.8 6.4

20.1754 80.2632

= .
where PV is the flow rate in LPM and CO is the valve opening percentage

2) Time Constant, TP
a) = !"#$% !!"#"$%
= 2.8 6.4
= .
b) Initial steady state = 6.4 LPM
+ 0.63 = 6.4 + 0.633.6
= .
c) From Figure 22, the time when PV passes through the initial steady state + 0.63(PV)
is 03:42:36 PM.
d) From Figure 22, the time when PV starts a first clear response to the step change in the CO
is 03:42:35 PM.

e) ! =

!"#$ !"#$ !"#$ ! !!"#$ !"#$ !"#$ !


!"

= .

20

MONASH University Malaysia


CHE 3162 Process Control Laboratory


3) Dead Time, P
a) From Figure 22, the time when PV starts a first clear response to the step change in the CO
is 03:42:35 PM.
b) From Figure 22, the time when CO was stepped from its original value to its new value
is 03:42:37 PM
c) ! =

!"#$ !" !"#$ ! !!"#$ !" !"#$ !


!"

= .

4) P, I and D value for Closed Loop Control
TC is the larger of (0.1TP) or (0.8P)
Since TP = 0.0167 and P = 0.0333
a) 0.1TP = 1.6710-3
b) 0.8P = 0.0267
Therefore, TC = 0.0267
PI Controller
For PI controller, the KC value is calculated using the equation below,
! =

!
! ! + !

0.0167

0.06 0.0333 + 0.0267

= .
Ti value for PI controller is the same as TP, thus
= = .

21

MONASH University Malaysia


CHE 3162 Process Control Laboratory


Ideal PID
For the ideal PID, KC value is calculated from the equation below where,
! =

! + 0.5!

! ! + 0.5!

0.0167 + 0.50.0333

0.06 0.0267 + 0.50.0333

= .
Ti is calculated using,
! = ! + 0.5!
= 0.0167 + 0.50.0333
= .

Td is then calculated using the equation shown below,
! =

! !

2! + !

0.01670.0333

20.0167 + 0.0333

= . !
The values calculated for both PI and Ideal PID are then tabulated in Table 1 as shown below.
Table 1: P, I and D values for Closed Loop Control

KC

Ti

Td

PI

4.639

0.0167

Ideal
PID

12.82

0.03335

8.33710-3

22

MONASH University Malaysia

CHE 3162 Process Control Laboratory

6.0 Discussion
6.1 Define proportional gain, integral gain and derivative gain
Proportional controller is a type of closed-loop feedback control system in which the
adjustment to the manipulated variable is proportional to the error signal; increasing the
manipulated variable will cause an increase in the error. The corrections for a
proportional control are based on present errors. The proportional gain, Kp is a type of
control in which adjusting the output will require the constant value to be multiplied with
the error. The effect of using a proportional gain is that the rise time can be reduced
significantly. Adding to that, the steady state error can be reduced but it cannot be
eliminated completely. In order to eliminate the effect of disturbances completely, an
integral controller is introduced in which the manipulated variables can be altered until
the magnitude of error is eliminated completely for a step-like input. Corrections for an
integral control are based on the sum or integral of past errors. The integral gain, Ki has
the ability to eliminate the steady-state error; however, the downside of it is that the
transient response will behave much slower as compared to the proportional controller.
The derivative gain, Kd is a variable that allows the controller to reduce the output
whenever the process value (PV) overshoot the set point (SP). Some advantages of using
a derivative gain would be that the stability of the system can be improved, overshoot can
be reduced and also the transient response can be improved.

6.2 The effects of proportional gain applied on the proportional-only control


system and the effects to response of the system.
In a practical industrial system, disturbances are often present in the system as there are
unavoidable. These disturbances are undesirable as they present errors within the system
and thus, needs to be eliminated with a processing control system. For the first part of the
experiment, the simplest controlling system which is the proportional control (P) system
is used to maintain the water level in the tank by regulating the amount of water flow rate
being pumped into the system, in which it is set to a desired value of 4 litres per minute
(LPM). Based on theoretical explanations, the steady state error of a typical proportional
control system is inversely proportional to the proportional gain of the control system, KP. This means that when the value of the proportional gain is increased to a larger value,
the steady state error of the system will be reduced.
Based on analysis and observations on the response graph of different KP values from KP
= 1 to KP = 17, it can be seen that when KP value is increased with Ki and Kd values being
fixed, settling time taken for the proportional control system to reach steady state is
longer but with a smaller steady state error. Besides that, when the KP is increased from a
smaller value to a larger value, it is very obvious that the overshoot (difference between
the first peak value to the steady state value) increased dramatically. The rise time of the
system varies with increasing proportional gain as there are no obvious trends for the rise
time. For example, in Figure 2 where KP = 1 and Figure 6 where KP = 17, the rise time
for the both system based on the graphs are having slight changes only.

23

MONASH University Malaysia

CHE 3162 Process Control Laboratory


Aside from that, it is also observed that when the ! value is increased, the response of
the system starts to oscillate instead of approaching the steady state value. This
oscillation of the response indicates that the system is unstable. From all the proportional
control response graphs above, the best proportional control system response graph was
Figure 5 with a KP value of 13. This is because the output flow rate was 2.5 LPM, which
is closest to the set point of 4 LPM compared to the other lower proportional gain values
used, which contain a greater offset value. However, even though using a proportional
gain of 13 reduces the offset value, the settling time will be increased. So, there are exact
proportional gain value that can give the best graph for this system that could reach the
steady state with the shortest time possible as the value of the gain increases, the control
system graph will eventually oscillating and become unstable. Hence, from the analysis
made, the proportional control system is expected to always have an offset value and will
not reach the set point even though different values of KP are being used.

6.3 Discuss the effect of integral gain applied on the proportional-integral


control system. Relate the effect to the response of the system (e.g. rise time,
overshoot percentage, settling time and steady state error).
For the proportional integral (PI) controller experiment, the proportional gain was fixed
at Kp = 5, while the integral gain was varied accordingly with values of Ki = 0.01, 0.001,
0.05, 1, 10 respectively. Based on figure 7 to figure 11, it was observed that the value of
the integral gain would result in a different control system. As what we can observe from
the graph, when the value of Ki increases, the response graph will take a longer time to
reach the steady state that means it is having a greater settling time.
Clearly, it can be seen from the graph that when the integral gain is the lowest, the
overshoot value will be the highest. When Ki = 0.001, the overshoot value is the highest
and oscillations were observed in the graph. As for the settling time to converge to steady
state, it was found that the value increases as the integral gain is increased. However, if
the integral gain value were too small, the system would not achieve steady state and will
continue to oscillate as can be seen when Ki = 0.001. Besides that, unlike proportionalonly controller, the steady state errors are completely eliminated in this type of control
system as the integral gain is increased. As we can see from figure 8, the system has
started to achieve steady state at the step point, even when the Ki is only increased to
0.01. However, it was found that at Ki value of greater than 1, the steady state is reached
at PV value greater than the set point.
In this proportional integral system, the output of the controller is always proportional to
the amount of time the error is present. Next, the respond of the integral controller is
quite slow at first, but in a long period of time, the offsets can be eliminated completely.
However, integral action tends to produce some undesired oscillations to the system. This
explained well on why the time required for settling to the set-point value is longer as
compared to the proportional-only controller mode.

24

MONASH University Malaysia

CHE 3162 Process Control Laboratory


Based on the 5 data obtained from the experiment, it can be concluded that the best
proportional integral controller would be the one with Kp value of 5 and Ki value of 0.01.
This is because there is no oscillation produced and the system achieved steady state at
the set point value at the shortest period of time.

6.4 The effects of derivative gain applied on the proportional-derivate control


system and the effects to response of the system.
From the experiment conducted previously, the proportional gain was set to 5 (! = 5)
and integral gain was kept at zero (! = 0) with varying value of derivative gain, ! for
the proportional-derivative control system experiment.
In the experiment, Kd value was being varied from 0.01 to 2. Based the output of the
graphs from the Figure 12 to Figure 15, the control system will vary according to the
value of the derivative gain of the system. For the derivative control system, when the
value of the derivative gain is increased, the instability of the system increases as it starts
to oscillate.
For this derivative mode, it is clear that when the derivative gain is low, the overshoot
value is not as high as compared to the response for a higher value of ! which has a
higher overshoot value and oscillations in the response. Besides that, by varying the value
of the derivative gain, the offset of the system does not change which means the
derivative gain will not influence the steady state error but instead will cause the output
to oscillate in a high frequency mode when the KD value is increased to a bigger value. In
addition, the rise time for the lower value of derivative gain is shorter compared to the
rise time of the higher value of derivative gain which can be seen from Figure 12 and
Figure 15. From Figure 15, due to its large derivative gain value of 2, the system is
oscillating and a longer rise time is required compared to Figure 12 which has a
derivative gain of 0.01.
By comparing Figure 12 and Figure 15, the stability of the system is very obvious that
when KD is low, the system can still be considered a stable system whereas for a larger
value of KD, the system will become unstable and oscillates. Thus it can be said that the
settling time of the response for a lower KD is shorter compared to the response for a
higher KD value due to the instability of the system as KD is increased.

25

MONASH University Malaysia

CHE 3162 Process Control Laboratory

6.5 Discuss the system response curve of PID controller tuned and compare it
with the system response curve obtained from proportional controller,
proportional-integral controller and proportional-derivative controller.
Based on the PID controller experiment conducted, the Kp value chosen was kept
constant at 5 throughout the experiment. The experiment was conducted with the Ki and
Kd values that gave the best system response curve under PI and PD controllers. Ki value
selected was 0.01 and Kd value chosen was 0.005. From the system response curve,
overshoot is observed and it takes 15 seconds settling time to reach steady state. By
keeping Ki constant and increasing Kd value to 0.01, initially overshoot is again observed,
but at a relatively lower offset value compared to the previous graph. It also takes a
shorter settling time of 8 seconds to achieve steady state condition. Next, by increasing
the Kd value to 0.05 with Kp and Ki being kept constant, oscillation is observed and steady
state cannot be achieved. Therefore with Kd value being set at optimum value of 0.01, the
best system response curve among the three can be obtained.
Then, the experiment is proceeded by keeping Kd value constant at 0.01 while increasing
the Ki value to 0.05 and 0.10 respectively. Hence the results obtained from this section
can be compared with the second result obtained from the previous paragraph. By setting
Ki value to be 0.05, overshoot can be seen at a similar peak; however it takes 52 seconds
for the system to remove the offset value to achieve steady state system. Subsequently,
the Ki value was altered to 0.1, in which again an overshoot can be obtained but it takes
100 seconds to reach steady state, which is relatively longer compared to the previous
experiments. From this section, with the Ki value being set at 0.01, the system response
curve observed is the most desired among the three different sets of values.
For the comparisons of system response curve of PID controller with other controllers,
the best system response curve from each section is chosen for further analysis and
comparison. By comparing PID controller to P controller, the system response curve of
PID controller is much desired as the system response curve of P controller does not
achieve the desired set point value of 4LPM. As for PI controller, its best system
response curve obtained has an overshoot of 7.2, which is higher than that of PID
controller. It also has a higher settling time of 15 seconds compared to 8 seconds as
required by a PID controller. Last but not least, for a PD controller, it is impossible to
obtain a system that can provide the desired flow rate of 4LPM and reaching the steady
state on the same set-up of gain values. Therefore, it can be deduced that a PID controller
proves to be the best device among the others controllers as it can prevent large value
overshoot by removing offset values. Besides that, it also reduces the settling time if
optimum values of proportional, integral and derivative gains are entered to the
controller.

26

MONASH University Malaysia

CHE 3162 Process Control Laboratory

6.6 Discuss whether the step test closed loop tuning method is suitable for all
types of process control.
For the step test closed loop tuning method we have used in our experiment, it is actually
a type of method known as Zeigler-Nichols tuning method. It is a type of trial and error
looping repeating technique used in the control system. After conducting the step test
closed loop tuning method experiment by following the procedures as mentioned above,
the gain and period could be computed. This method is an effective way that can be used
in the control system to obtain the gain and period, however it is not a general tuning
method that can be used for all types of process control. Furthermore, a guaranteed
success rate with this method in all processing control system should not be expected.
The reasons that the step test closed loop tuning method is not suitable for all types of
process control is because this method requires the entire process in the system to be
brought to the extreme ends of the instability state in order to determine the gain. In
practical applications, industries are always trying to reduce the operational time since
the operating costs are not cheap. When large adjustments or changes on the gain are
made, the instability of the process increases dramatically as well. Thus, it is not suitable
for applications in real industrial plants due to the oscillations that will occur when the
gain is changed. These permanent oscillations in the system to determine the gain and
period will put the operations of the plant at risk. Furthermore, unit operations such as
distillation columns, furnaces, separators, reactors and such will be dangerous if the
system approaches unstable state. As the system approaches unsteady state, it would be a
very alarming proposition for the plant personnel.
This Ziegler-Nichols tuning method is not suitable for all types of processes as the
recommended settings are not suitable for systems that are lag-dominated, which
basically means the control output has a slower respond due do delay present in the
system. This method can only be used on processes where the lag-phase of the output
exceeds 180 degrees at high frequencies. A clear example would be that this method does
not work on a simple second order process. Lastly, this method is time consuming and is
a monetarily expensive technique. The trial and error method could take few hours for the
loop testing procedure. In real life industry scenarios, this method consumes a lot of time
and would be a very inefficient and wasteful method. This is so because in the industrial
world, minutes or seconds could mean thousands or millions of dollars. However,
Zeigler-Nichols method is useful for fine and specialty chemical industries, where
products require high quality and purity instead of mass volume produced.

27

MONASH University Malaysia

CHE 3162 Process Control Laboratory

7.0 Conclusion
Based on the experiments that were conducted, the overall characteristics of each
different control systems were analysed and investigated. This is done by changing
different values of gains for different control systems and analysing the response and
study effect after changing certain gains.
For the proportional control system, Ki and Kd values were fixed to zero while Kp
was being varied. The smallest Kp value gave the most stable response system and when
the Kp value was increased, the system starts becoming unstable as oscillations are
present. Also, offsets are always present in the response for all values of Kp. For
proportional-derivative (PD) controller system, the Kp and Ki was set to 5 and zero
respectively while Kd was set initially to 0.01 and then being varied. For the PD control,
the system was unable to achieve the set point thus causing offsets. Moreover, a larger
derivative gain will cause the system to oscillate and end up being unstable. Thus, since
the response when Kd = 0.01 was the best, this value chosen as the optimum value to be
used in the PID control system tuning process. Not to forget, a value of 0.01 was chosen
as the optimum value for Ki since it was found out that the response when Ki = 0.01 gave
the shortest settling time without any oscillations in the proportional-integral (PI) control
although the initial Ki value was set at 0.001.
Subsequently, after the P, PI and PD experiments were conducted, the
optimum gain values that is to be used for the PID control system was concluded to be
Kp, = 5, Ki = 0.01 and Kd = 0.01. These values were chosen as the optimum values as they
are able to provide the shortest settling time and with an zero offset value. In addition, a
series of calculations was performed in the step test closed loop tuning method and the 2
valve openings that gave the smoothest flow were determined to be 80% and 20%. With
consideration of values for the 2 valve openings chosen, the process gain, Kp, time
constant, Tp, and dead time, P were calculated to be 0.06, 0.0167 minutes and 0.0333
minutes respectively.
As a whole to conclude this experiment, PID was concluded to be the best system
that can be used for a closed-loop flow controlling system, followed by the PI controlling
system and PD controlling system. The proportional-only controlling system is not
suitable as a stand-alone control system because the offset cannot be eliminated with this
control system. Furthermore, another downside would be that the system will start
becoming unstable if the proportional gain was set at a certain large value. Similarly, the
PD controlling system will become unstable as well if a high value of derivative gain is
used. On the other hand, PI controlling system can achieve stability and reach the set
point value as it is able to eliminate the offset. However, if the Ki value is too small, it
will tend to oscillate and become unstable. When the Ki value is too big, the settling time
of the system will also increase as shown in Figure 10 and the output response will reach
to a value that is above the set point value and the settling time required would be very
long as shown in Figure 11.

28

MONASH University Malaysia

CHE 3162 Process Control Laboratory

8.0 Nomenclature

Symbol

Meaning

Kp
Ki
Kd

Proportional Gain
Integral Gain
Derivative Gain
Proportional Integral Derivative
Control
Proportional Integral Control
Proportional Control
Proportional Derivative Control
Dead Time
Time Constant
Manipulated Variable
Process Value
Critical Time
Set Point

PID control
PI control
P Control
PD control
P
Tp
MV
PV
TC
SP

9.0 Reference
Moodle. 2014. Monash University CHEMICAL ENGINEERING LABORATORY
MANUAL CHE3162 Process Control. [ONLINE] Available
at:http://moodle.vle.monash.edu/pluginfile.php/2528762/mod_resource/content/3/
Sunway%20Lab%20Manual%202014.pdf. [Accessed 24 August 14].
Seborg, Edgar and Mellichamp, Process Dynamics and Control 2nd edition,
Wiley, 2004. [Accessed 24 August 14].

29