UNIVERSITI KUALA LUMPUR

MALAYSIAN INSTITUTE OF CHEMICAL AND BIO-ENGINEERING

TECHNOLOGY

NKB 31102
CHEMICAL ENGINEERING LABORATORY 3

LECTURER :

DR RAPIDAH BINTI OTHMAN

TITLE :

STATEMENT 3
AIR FLOW CONTROL IN A CLOSED LOOP TUNING

PREPARED BY :

MUHAMMAD IKHWAN BIN SHAMSURI 5522011525

AKIF NABIL BIN AMIZAR 55220115390
ASFA ASHMIRA BINTI AZIZAN 55220115270
RAIHANA BINTI MOHD NOH 55220115055

Page of
LABORATORY 3 : STATEMENT 3
AIR FLOW CONTROL IN A CLOSED LOOP TUNING

1.0 INTRODUCTION/ THEORY results in a loop that overshoots its

set point after a disturbance or set
The air flow process control training point change. It response in general
system (Model AF922) was use air to is somewhat oscillatory, the loop is
simulate vapor phase flow process. marginally robust and withstand only
Vapor volumetric flow can measured small changes process conditions
either by the differential pressure (Copeland, 2008).
across orifice or by a variable area
flow meter. The compensation for The PID value from tuning procedure
pressure and temperature variation can obtained:
using “Perfect Gas Law” can be
made to the flow measurement by  PB corresponds to
the differential pressure uniform oscillation = PB
measurement.Proportional integral * ultimate (correspond
derivative controller (PID controller) ultimate gain, Kcu)
is a control loop feedback  Distance between two
mechanism that used in industrial adjacent peaks = Tn*
control systems and many other ultimate (ultimate
applications require a continuously period, Pu)
modulated control. Each tuning rule Tn* can be calculated based on Eq.
is analyzed in the closed-loop (A.1) in the appendices
characteristics (Ieeexplore.ieee.org.
2018). PID controllers are probably  The approximate PI
the most commonly used controller controller settings
structures in industry (Gain, n.d.). using Ziegler Nichols
formula based on Eq.
Moreover, there might some
(A.2) in the
challenges to control and
appendices.
instrumentation engineers in the
aspect of tuning of the gains required 2.0 Methods
for good transient performance.
Several prescriptive rules used in This experiment was carried out by
PID tuning. However PID tuning is a using Model AF922 Gas Flow
difficult problem because it need Process Control Training System
satisfy complex criteria within the Equipment.
limitations of PID control (Finn First, we need to identify all
Haugen, 2010). Various methods for major components in the plant which
loop tuning, and more sophisticated are the airtank (T90), the Orifice
techniques are the subject of patents. meter (FE91), the differential
Tuning rules were designed by pressure transmitter, with a square
Ziegler-Nichols is at 1 over 4 root function (FT91), the absolute
amplitude decay response. This

Page of
pressure transmitter (PT91), the RTD VF and close the valve MV901 at the
temperature sensor (TE91), the pipeline PLI . Make sure the valves
temperature transmitter (TIT911), the MV900A and MV900D are open and
ON/OFF controller (FIC90), the flow valves MV900B and MV900C are
controller (PID) (FIC91), FCV90 shut. Place the panel controller
Pneumatic ON/OFF control valve FIC91 in the Manual (M) mode and
with Solenoid valve, TSV90, the make sure the control valve
control valve with FCY91 positioner PP is connected. Then,
current-to-air (I/P) converter (FCV91) open fully the control valve FCV91 by
and pneumatic positioner, PP. k. setting the MV of FIC91 to -6.3%.
FPTR91 Recorder. Shut fully the manual valve MV900A
and set AR900 to the pressure
Next, the start-up procedures begin indicated at AR900.
with switch the "PANEL,
SCADA/DDC" selector switch at the Then, adjust the manual valve
front of the cubicle to MV900A till the air flow rate is about
"PANEL,SCADA" position. Then, 50 kg/hr . Step (10), (11) and (12)
switch on the main power supply at have set the process air supply
the front of the cubicle. If any pressure and flow rate. After that,
annunciator gets activated, press the check that vent valve and bottom
ACKNOWLEDGE button to silence drain valve of vessel T90 are shut.
the buzzer. After that, set the Set MV = 106.3% at FIC91 to fully
instrument air supply regulator to the shut the control valve FCV91. Next,
pressure indicated at the air check control valve FCV91 is fully
regulators at the pneumatic control shut and FI911 reads the air
valves FCV90 and FCV91. Cleanup volumetric flow rate as zero. Also
if there is any condensed water in the check the mass flow rate Fm at
air regulators. Then, trace the two FIC91 is zero. Now set the MV =
pipelines, PLI and PLII as follows: 50% at FIC91 and check control
valve FCV91 is 50% opened. State
Pipeline I (PLI): the variables that are recorded in
AR900-MV900A-T90-MV900D-PLI-F each channel of the recorder
I911-TE911-FE91/FT91-FCV91-FCV FPTR91. Red pen is for mass flow
1-VF rate (kg/hr). The green pen for
Pipeline II (PLII): volumetric flow rate (m3/hr) and blue
AR900-MV900A-T90-MV900D-PLII- pen for temperature (oC).
FCV90-PG902-V For control of the air flow system, we
After that, open fully the pipelines PLI begin with open the control valve
and PLII at their various manual FCV91 fully with MV = -6.3%. Adjust
valves. Close fully the manual its setpoint, SV = 25 kg/hr. Next,
by-pass valve around the control access the PID parameters of FIC91
valve FCV91 and the ON/OFF and set the first PID trial values
control valve FCV90. Open the vent which are PB equal to 200%, TI

Page of
equal to 6 s and TD equal to 0 ii) Distance between two adjacent
s .Start the recorder chart drive by peaks = Tn* ultimate (ultimate period,
pressing “RCD” push button at Pu) Tn* can be calculated from this
FPTR91. Switch FIC91 to Auto (A) formula:
mode and watch the recorder
response until the air flow (red pen) Tn *(sec s) = (measurement(mm)×
is fairly steady. Then, propose a 3600) / chartspeed (500mm hr)
pulse disturbance by quickly opening iii) The approximate PI controller
and shutting fully the by-pass valve settings by using Ziegler Nichols
around control valve FCV91. formula are: PB = 2 x PB*, TI =
Observe the response till it becomes Tn*/1.2
fairly steady.
To test this new PI controller settings,
PID controller tuning begin with set set this value into FIC91 (in Auto
FIC91 in manual mode with set point mode). Perform the test with a pulse
SV = 25 kg/hr. Adjust the PID values disturbance. Then, increase PB
to which are PB are about 100%, TI value to damp the oscillation and fine
equal to 9999 s which is maximum tune the PB and TI values to get as
and TD equal to 0 s which is close to quarter amplitude damping
minimum. Next, adjust MV in FIC91 response.
until the PV value closes to the set
point SV value which is 25 kg/hr. Lastly, for shutdown procedures.
Transfer FIC91 to Auto (A) mode. First, stop the recorder chart by
Then, propose a set point pulse pressing “RCD” press button at
disturbance as set increase SV to 28 recorder FPTR91 and then switch
kg/hr and hold the value for a few FIC91 to Manual mode with MV = 0%.
seconds. Then, bring back the SV to Next, switch off the main power
25 kg/hr. Observe the mass flow rate supply and shut off the process air
(red pen) response. Repeat the supply at AR900. Finally shut off the
closed loop tuning method as above instrument air supply.
with a decreasing PB value until
3.0 Results
uniform oscillation (constant
amplitude) is observed. Next, The first part on the procedure, the
examine the chart response for the
data was set at Air Flow controller
uniform oscillation.
was :
New PID value from tuning
procedures can be obtained as MV = -6.3% SV = 25 kg/hr
follows:
PB = 200% TI = 6s
i) PB corresponds to uniform
oscillation = PB* ultimate TD= 0s
(corresponds to ultimate gain, Kcu)

Page of
The graph obtain based on data
above

The red and green line was increase

drastically and slightly increase then
decrease, after that the graph
constant in a straight line until it 4.0 Discussion

increase again in the amplitude of The experiment was conducted by

the graph. After the the bypass valve using Flow Ratio Plant Control
was closed, the graph changed back (WF922) but specifically the model
to steady state. AF922 to identify and understand the
concept of air flow process control.
By using the first trial data with SV =
The air flow process control training
28kg/h and 25 kg/h, the graph that
system (Model AF922) uses air to
acquired was :
simulate a gas or vapor phase flow
process. Gas/vapor volumetric flow
can be measured either by the
differential pressure across orifice or
by a variable area flow meter.
For Zeigler-Nicholas PID tuning :
Compensation for pressure and
temperature variation using Perfect
Gas Law can be made to the flow
measurement by the differential
pressure measurement.
The second part of the procedure
was the PID Controller Tuning. The
The aim of this experiment is to
PID Flow Controller was set to :
identify the important components of
SV = 25 kg, TI = 9999s (Maximum)
the air pressure control system and
PB = 100%, TD = 0s (Minimum)
to mark them in the P&I Diagram.
Next, to carry out the start-up

Page of
procedures systematically. Then to
study gas volumetric flow rate Once the values have been adjusted,
measurement using orifice plate. the chart drive was started by
Furthermore to study gas mass flow pressing “RCD” pushbutton at
rate (Fm) control using PID controller. FPTR91. The FIC91 was switch to
Last but not least, to design a PID Auto (A) mode and the recorder
controller using Ziegler Nichols response (i.e. air flow response) was
closed loop method. watched until the air flow (red pen) is
fairly steady. A pulse disturbance
The experiment was begun by
was introduced by quickly opening
performing all the start-up
and shutting fully the by-pass valve
procedures step-by-step to ensure
around control valve FCV91. The
everything ran smoothly. The MV
response was observed till it
was then set to be 106.3% and then
becomes fairly steady.
50%. The graph has 3 colored lines
and each color has its own The graph was observed for a while
representation; Red = Mass Flow until the red line which represents
Rate (kg/hr), Green = Volumetric Mass Flow Rate (kg/hr) becomes
Flow Rate (mᵌ/hr), and Blue = fairly steady (straight line). The red
Temperature (˚C). line had more responses compared
to the green line, indicating that the
Based on the task given, our first
Mass Flow Rate is more sensitive
closed loop tuning was done (in
compared to the Volumetric Flow
Manual Mode) using the parameters
Rate.
as follows:
Continuing to next task, we were
asked to evaluate the performance of
PB 200%
the air flow controller through the use
TI 6s
of PID feedback control configuration
TD 0s
using the tuning parameter. This
SV 25kg/hr
experiment applied Ziegler-Nichols
Closed Loop method which is one of
the most common methods used to
tune loops. The SV was set to be

Page of
25kg/hr and the PID values as was between the peak to the axis was
set as follows: estimated to be 2mm each. Once
constant amplitude was obtained, we
PB 100%
could then calculate Tn using this
TI 9999 secs (max)
formula:
TD 0 sec (min)

m sec e
The TI was set to the maximum ‸㈲ 㔸吠‸ ‸m橔 thh
count of 9999 secs solely to defeat hh
㈠㈲吠橔 㰍‸‸吾 e
the variability/function of the TI itself. ㈠吠
m thhe hh
m 쳌
Next, the MV was adjusted to an
extent where the value of PV reads
Next, in order to get TI, formula as
almost closes to SV (25kg/h). The
below was followed:
value of MV recorded was closes to
PV, this is easy to achieve where the
a m 쳌
pulse disturbance valve was closed
a 쳌 쳌
tightly. This made the reading of PV
a
to be stable and easy to approach
the reading value of SV. After that,
Final set point value we would want
the set point pulse disturbance was
introduced where the SV was set to to obtain was PB. Thus, the equation
used was as follows:
28 kg/hr and the response on the
graph was observed for a few
seconds before it was set back to 香 m‸쳌e 香 吾e

25kg/hr. This was done to observe 香 m‸쳌 hh

the fluctuation of the red line 香 m‸쳌 hhh

response.
Based on the calculation performed,

When the set point pulse is we had now obtained the new PID

introduced, the set point PB was values of PB = 100%, TI = 12 sec

decreased to 50% until uniform and TD = 0 sec (min value). In the

oscillation (constant amplitude) was graph, oscillations were observed to

formed. On the graph, the distance exist and later the PB was decreased

Page of
to 96% until the oscillation becomes sensitive in noise and disturbance.
damped.The disadvantage of using For steady state error and transient
response can be controlled easily in
Ziegler-Nichols’ closed loop method
closed-loop system by simple
is that the Experiment can be time adjustment of amplification in loop. It
consuming. Then, it can venture into produce their tuning rules by
stimulating great number of different
unstable regions while testing the P
processes and correlate by the
controller, which could cause the controller parameters with features of
system to become out of control. It the step response.
does not hold for I, D and PD Based on the statement 3 given, the
controllers. [Woolf, 2007] first closed loop tuning was done
manually using the PB: 200 %, TI: 6
s, TD: 0 s and SV: 25 kg/hr. At a set
It is also important to remember the point PB was decreased to 50% until
impact of the measurement noise on uniform oscillation (constant
the control signal. The more amplitude) was formed. On the graph,
the distance between the peak to the
aggressive controller fast control the axis was estimated to be 2mm each.
more sensitive is the control signal to Once constant amplitude was
the measurement noise. Again, to obtained Tn was calculated. Tn value
obtained is 14.4 s, TI=12. We had
descrease this sensitivity, the
now obtained the new PID values of
controller gain can be descreased. PB = 100%, TI = 12 sec and TD = 0
sec (min value). In graph, oscillations
were observed to exist and later the
Towards the end of the experiment,
PB was decreased to 96% until the
when readings were obtained, the oscillation becomes damped. The
shut-down procedures were followed graph show Mass Flow Rate is more
step-by-step after the RCD button sensitive compared to the Volumetric
Flow Rate. Moreover, the chart can
was pressed once again to stop be conclude that the process is
recording. Results were analysed stable and successful. Regarding to
and recorded. the performance of the air flow
controller using PID feedback control
configuration using the tuning
5.0 Conclusions parameter, the are some errors
problems had occurred which was
In the conclusion, for this experiment the pulse disturbance valve was not
is P&I Diagram controller we used to closed tightly which can caused the
tune by using Ziegler Nichols in a reading of PV to be slightly unstable
Close-Loop Method. Closed-loop is and difficult to approach the reading
more accurate than open-loop in less value of SV. For PB Value affect the

Page of
stability of the oscillations in the SV value which was 25 kg/hr. A set
graph. Based in the theory, if value of point pulse disturbance was induced
proportional band (PB) increases, by set the SV to 28 kg/hr and was
noise will appeared and caused the held for a few seconds and then we
damped oscillations to produce very turned back the SV to 25 kg/hr. The
rapid. Hence, the Ziegler Nichols Closed Loop Tuning Method was
Closed Loop Method was proved. repeat as above with a lower PB
value until uniform oscillation shown
6.0 Appendices in the chart.
Eq. (A.1) : 8.0 Keywords
m sec e
 Set value (SV)
‸㈲ 㔸吠‸ ‸m橔 e thh
hh  Measure value (MV)
㈠㈲吠橔 㰍‸‸吾 e  Proportional Band % (PB)
㈠吠
 Integral Time (TI)
Eq. (A.2) :  Derivative Time (TD)
香 香 m 쳌
9.0 References
7.0 Abstract
[1] Ieeexplore.ieee.org. (2018).
This experiment will determine the
Various PID controller tuning for air
important units of the air flow control
temperature oven system - IEEE
system, PID controller using Ziegler
Conference Publication. [online]
Nichols closed loop method design
Available at:
and the performance of the air flow
http://ieeexplore.ieee.org/document/
controller system. We tested the
7047441/?reload=true [Accessed 19
performance by tuned the system as
Mar. 2018].
PB = 200% , TI = 6s, TD = 0, SV = 25
kg/hr and MV = -6.3%. Then, the PID [2] The Good Gain method for PI(D)
values was adjusted as PB = 100%, Controller Tuning, Finn Haugen
TI = 9999s (Maximum) and TD = 0s TechTeach
(Minimum). The MV in PID Flow http://techteach.no/publications/articl
Controller (FIC91) was adjusted until es/good_gain_method/good_gain_m
the PV value closes to the set point ethod.pdf, 2010 (Accessed 26.02.18).

Page of
Page of