Lab Report Tuning CPE622
Lab Report Tuning CPE622
NAME STUDENT NO
NUR FATIHAH BINTI ZULKEFLI 2016595727
AMNI BINTI SAUFIAN 2016718381
MUHAMMAD AFIQ BIN JOHARI 2016537543
GROUP : EH2206A
EXPERIMENT : PID TUNING
DATE PERFORMED : 6th SEPT 2019 – 6th NOV 2019
SEMESTER 6
PROGRAMME / CODE : CHEMICAL ENGINEERING / EH220
SUBMIT TO : MADAM SYAZANA MOHAMAD PAUDZI
Date: Date:
TABLE OF CONTENTS
1.0 ABSTRACT ................................................................................................................ 4
2|Page
6.1.2 DISCUSSION...................................................................................................... 24
6.2.1 DISCUSSION...................................................................................................... 26
7.1.2 DISCUSSION...................................................................................................... 31
8.1.2 DISCUSSION...................................................................................................... 35
9.1.2 DISCUSSION...................................................................................................... 39
10.0 DISCUSSION.............................................................................................................. 41
13.0 REFERENCES............................................................................................................ 43
3|Page
1.0 ABSTRACT
Process stability of a PID control loop depends upon the proportional, integral and derivative
constants used. Using the conventional tangent method and the proper tuning rule, the
optimum P, I and D can be estimated. With this optimum P, I and D set into the controller, an
optimum response is normally achieved. The variation in these control algorithms of PID
controllers only affect the shape and size, but not the characteristics, of the process response
curve. These characteristics of PID controller are the tendency to produce overshoot,
undershoot, off-set and oscillation in the system response. The selection of P, I, and D values
is very crucial. They determine whether the process is oscillatory, stable or unstable.
4|Page
2.0 INTRODUCTION
PID controllers are found in a wide range of applications for industrial process control.
Approximately 95% of the closed loop operations of industrial automation sector use PID
controllers. PID stands for Proportional-Integral-Derivative. These three controllers are
combined in such a way that it produces a control signal. (elprocus, 2019). As a remark’s
controller, it offers the manage output at favored levels. Before microprocessors have been
invented, PID controller was implemented via the analog digital components. But nowadays
all PID controllers additionally have the built in PID controller instrument. Due to the ability
and reliability of the PID controllers, those are traditionally used in manner manage
application.
5|Page
Due to limitation of p-controller wherever there continually exists an offset between the
method variable and point. I-controller is required, that provides necessary action to
eliminate the steady state error worth reaches to zero. It holds the worth to final
management device at that error becomes zero. Integral management decreases it output
once negative error takes place. It limits the speed of response and affects stability of the
system. Speed of the response is exaggerated by decreasing integral gain ki.
In the above figure, because the gain of the I-controller decrease, steady state error
additionally goes on decreasing. For many of the cases, PI controller is employed notably
wherever high-speed response does not needed. While mistreatment the PI controller, I-
controller output is prescribed to somewhat vary to beat the integral finally end up conditions
wherever integral output goes on increasing even at zero error state. Thanks to the
nonlinearity within the plant.
I-controller doesn’t have the aptitude to predict the long run behavior of error. Thus, it reacts
unremarkably once the point is modified-controller overcomes this drawback by anticipating
future behavior of the error. Its output depends on rate of modification of error with relevance
time, increased by spinoff constant. It offers the starter motor for the output thereby
increasing system response
6|Page
Figure 2: PID Controller
In the on top of figure response of D controller is a lot of, compared to PI controller and
additionally sinking time of output is cut. It improves the steadiness of system by
compensating part lag caused by I-controller. Increasing the by-product gain will increase
speed of response.
Before the working of PID controller takes place, it must be tuned to suit with dynamics of
the process to be controlled. Designers give the default values for P, I and D terms and
these values couldn’t give the desired performance and sometimes leads to instability and
slow control performances. Different types of tuning methods are developed to tune the PID
controllers and require much attention from the operator to select best values of
proportional, integral and derivative gains. (elprocus, 2019)
7|Page
There are different types of tuning methods, typically used tune the PID controllers and it
requires much attention from the operator to select best values of proportional, integral and
derivative gains where before the working of PID controller takes place, it must be tuned to
suit with dynamics of the process to be controlled. It because sometimes when the values
of P, I and D was set up, these values couldn’t give the desired performance and sometimes
leads to instability and slow control performances. There many types of tuning methods of
PID controller where firstly is process reaction curve technique where it is an open loop
tuning technique. It produces response when a step input is applied to the system because
initially, some control output should be applied manually and the response curve need to
be recorded. Then, the slope, dead time, rise time of the curve and finally these values will
be substitute into the P, I and D equations to get the gain values of PID terms. Then, Ziegler-
Nichols method proposed closed loop methods for tuning the PID controllers. Those are
continuous cyclin method and damped oscillation method where the proportional gain is
increased till system oscillates at constant amplitude. (elprocus, 2019)
In real process, the PID application was applied. Digital PID control system was one of the
examples. The PID algorithm, now widely used in industrial process control that has been
recognized and employed for nearly a century where it is originally in pneumatic controllers.
Electronics is the first used to model PID controls in control system design with analog
computes in the 1940s and 1950s and became increasingly involved in actual process control
loops where first as analog controllers and later as digital controllers.
Air conditioning or refrigeration system are one of the examples that might considering use
the PID loop. This system is basically maintaining the temperature in a narrow range by using
continuous monitoring and control which as opposed to thermostatic on-off control.
Figure 4 shows a basic block diagram of a control system that regulates temperature by
8|Page
continuously adjusting the fan speed, increasing or decreasing the airflow from a low
temperature source. The system is required to maintain the room temperature as close as
possible to the user-selected or setpoint value where to do this, the system must accurately
measure the room temperature and adjust the fan speed to compensate. A precision current
source in the system shown in Figure 4 drives a current through a resistive temperature sensor
which a thermistor or RTD are in series with a reference resistor and adjusted to represent the
desired temperature. he analog-to-digital converter (ADC) digitizes the difference between the
reference voltage and the thermistor voltage as a measure of the temperature error. An 8-bit
microcontroller is used to process the ADC results, and to implement the PID controller. The
microcontroller adjusts the fan speed, driving it via the digital-to-analog converter (DAC).
External program memory and RAM are required to operate the 8-bit microcontroller and
execute the program. The rate at which the fans run would be directly related to the
temperature difference from the setpoint if the proportional control (P) on its own were used.
Adding an integral term (PI) resulting in the fan speed rising or falling with the ambient
temperature. It will adjust the room temperature to compensate for errors due to the ambient
rise in temperature to compensate for errors due to the ambient rise in temperature through
the day and then the temperature falls in the evening. The integral term thus removes the
offset, but if the integral gain is too high, oscillation about the setpoint can be introduced where
the oscillation is inherent in temperature control system by employing on-off thermostats. This
oscillatory tendency can be greatly reduced by adding in a derivative term (PID). The
derivative term responds to the rate of change of the error from the setpoint. It helps the system
rapidly correct for sudden changes due to a door or window being opened momentarily.
(Neary, 2004)
PID controller manipulates the process variables like pressure, speed, temperature, flow, etc.
and some of the applications use PID controllers in cascade networks where two or more
PID’s are used to achieve control. According to the proportional, integral and derivative control
calculations in algorithm, the controller produces combined response or controlled output
which is applied to plant control devices. All control applications don’t need all the three control
elements. Combinations like PI and PD controls are very often used in practical applications.
(elprocus, 2019)
9|Page
3.0 OBJECTIVES
1) To perform open loop test using DCS Delta A-V Emerson and DCS FOXBORO.
2) To perform closed loop test using DCS Delta A-V Emerson and DCS FOXBORO.
4) To calculate the PB and I value using Ziegler Nichol’s and Cohen-Coon’s method.
5) To find the Td, Tc and RR value using tangent method, reformulated tangent
method and numerical tangent method
6) To calculate either the response curve is linear, slightly non-linear or roughly non-linear.
7) To perform the correct method in order to do open loop test and closed loop test.
10 | P a g e
4.0 THEORY
4.1 LINEARITY
Linearity test aims to determine the relationship between independent variables and the
dependent variable is linear or non-linear. The linearity test is a requirement in the correlation
and linear regression analysis. Thus, a good research in the regression model there should
be a linear relationship between the free variable and dependent variable. There are two
steps of decision-making process in the linearity test where the first one is if the value
significant deviation from linearity is more than 0.05, then the relationship between the
independent variables are linearly dependent. Secondly, if the value of significant deviation
from linearity is less than 0.05, then the relationship between independent variables with the
dependent is non-linear. (Tests, 2015)
Process linearity is a process that looks at the relationships between elements of a process
and the output. Basically, linearity is the property of mathematical relationship or function
which means it can only be graphically represented as a straight line. A linear function where
the function in a graph lies on the straight line and which can be described by giving the
slope and y-intercept of that line which means their graph pass right through the origin (0,0).
Linearity error or also can be defined as non-linearity is the deviation of the sensor output
curve from a specified straight lie over a desired pressure range. The linearity error value is
normally specified as a percentage of the specified pressure range. If a sensor is only used
over half the specified range and you are able to set the maximum value to be used then the
linearity error is calculated from this value, which of course is going to provide improved
accuracy over that specified by the manufacturer. There are two common ways of specifying
the linearity error which is best fit straight line and terminal base linearity. Usually, the linearity
error is the least significant of all errors in any given pressure sensor and commonly in the
region of 0.1 to 0.2 or better. (Applied Measurement Limited, 2019)
11 | P a g e
Figure 5: Best Fit Straight Line for Linearity
The error of best fit straight line is specified as the maximum deviation (+/-x) % of span of
the output value from the straight line.
Terminal base linearity or also known as end-point linearity was determined by drawing a
straight line between the end data points on the output curve and the data point was chosen
to achieve the maximum length of the perpendicular line. (Applied Measurement Limited,
2019)
12 | P a g e
4.2 OPEN LOOP TEST
An open loop control system acts completely on the basis of input and the output has no effect
on the control action. Then, it works on fixed operation conditions and there are no
disturbances. Open loop control system is mostly stable and no effect on gain. The structure
of open loop control system is rather easy to construct and implemented where it makes the
overall cost of these systems is low. (Nasir, 2018)
Open loop system tends to be simple and inexpensive as they do not provide feedback from
the machine movement to the controller where in other words, open loop systems only act
solely on the basis of the input and di not use feedback from the output to self-correct while
the test is running. Therefore, the test procedure entered into an open-loop controller may
vary due to external disturbance such as noise which is without the operator noticed it.
Unfortunately, the data acquired from universal testing machine with an open loop controller
may be imprecise. It is because when the running systems with an open loop controller want
to keep the speed constant which is might be one of the problems as there is no direct
feedback to monitor and regulate control. The speed of the machine might change during a
test due to various reasons, and the open loop system does not have a feedback control to let
the controller know of these changes. The test will run at a modified speed without the machine
or the operator noticing. (Yalcin, 2017) The open loop response test is very important where
it is enabling the plant or refinery personnel to define the process by determining the process
dead time (Td), the process time constant (Tc) and response rate (RR). New controller settings
can be calculated and applied by using these values and it is important parameter which
resulting on improvements for loop performance, lead to better overall control of the process
and enhance the productivity. Basically, Tc is the lag time or the time needed for the process
to be constant where more specifically it represents the time needed for the PV to reach 63.2%
of its total while RR is the rate of respond and the faster the better. Dead time (td) is the delay
from the controller output (CO) signal is issued until when the first measured process variable
(PV) begin to respond. (Nor Rahman, Musdafa kamal, Mat Yuna, & Md Tahir, 2019)
13 | P a g e
4.2.1 TANGENT METHOD
The tangent method starts with an open loop test and it is done by putting the conntroller
in manual mode. The response curve is then analyzed for the process deadtime (Td) and
the response rate (RR) by drawing the tangent line to the steepest point of the response
14 | P a g e
4.2.2 REFORMULATED TANGENT METHOD
Reformulated tangent method are basically same as tangent method but the difference
between these two methods is reformulated tangent method use angle to determine the RR
where;
tan 𝜃 𝑎
𝑅𝑅 = ×
∆𝑀𝑉 𝑏
While dead time (Td) will be calculated as same as tangent method;
𝑇𝑑 = 𝑇𝑑(𝑙𝑒𝑛𝑔𝑡ℎ) × 𝑏
Then, there are three ways that can be used to calculate the time constant which are:
1. Tangent method
𝑇𝑐 = 𝑇𝑐 (𝑙𝑒𝑛𝑔𝑡ℎ) × 𝑏
2. Two-point method
𝑇𝑐 = 1.5 × (𝑇63.2 − 𝑇28.3 ) × 𝑏
3. Tangent and a point method
𝑇𝑐 = (𝑇63.2 − 𝑇𝑑) × 𝑏
(Ishak A. A., 2018)
15 | P a g e
4.2.3 NUMERICAL METHOD
Numerical method is applied when we used Foxboro unit. Only a set of data is needed,
a graph is not necessary. There are three numerical technique that we use:
1. Tangent method
PV − PV
RR = 1 −1
2 ∆h ∆MV
PV − PV
1 𝑖
Td = t1 − 2 ∆h( )
𝑃𝑉 − 𝑃𝑉
1 −1
PV − PV
𝑓 𝑖
TC = 2 Δh( )
𝑃𝑉 − 𝑃𝑉
1 −1
3. Two-point method
𝑇𝑐 = 1.5 × (𝑇63.2 − 𝑇28.3 )
Close loop test use PID setting in the controller, PID value of the process control that can be
estimated by using tuning rule where it is based on their settling criteria which is Ziegler Nichols
and Cohen-Coon.
16 | P a g e
Tuning rule by Ziegler Nichols Tuning rule by Cohen-Coon
PB (%) I (s) PB (%) I (s)
𝜇
100 1 + 11
111.1 𝑅𝑅 𝑇𝑑 3.33 𝑇𝑑 𝜇 𝑅𝑅 𝑇𝑑 3.33
11𝜇
𝑇𝑑
1+ 1+
11 5
Table 1: Tuning Rule
If the respond is not stable, the PID controller can be tune to improve loop performance and
lead to better overall control of the process and enhanced productivity. Some of tuning theory
are when the value of PB is decreases, the process response becomes faster. Then, when
the value of I is decreases, the process response becomes faster. Lastly, when the value of
D is increases, the process response becomes faster. This is because MV is inversely
proportional to PB and I, meanwhile directly proportional to D. Concurrently, MV is also directly
proportional to process response (PV). (Ishak A. A., 2018)
17 | P a g e
5.0 METHODOLOGY
2) The faceplate was opened by double clicked at the Controller example as below,
PIC92
4) The stabilized the process, the manual or auto mode was selected.
5) The data of manipulated variables was recorded for the initial value.
8) For self-regulating system: Once steady state reached, the result was recorded by
printed the response curve.
18 | P a g e
9) For non-self-regulating: Once the response curve can be calculated, that response
curved been printed.
10) Tangent and Reformulated tangent method was used to determine the time delay,
Response rate and time constant.
11) PI controller setting was calculated using Ziegler Nichol’s and Cohen- Coon’s
method.
2) Details icon and Faceplate was clicked to set the controller setting
4) Fine tuning has been done to reduce the oscillation and overshoot.
3) Then, the manual mode was changed to automatic mode after 3 seconds the MV
value was changed.
2) The set point was changed higher 10% from the current operating process value.
19 | P a g e
5.2 DCS FOXBORO
5.2.1 OPEN LOOP TEST
• FIC31
• LIC31
5) The auto mode then was changed into manual mode to perform the open loop test
6) The manipulated variables were changed in range 5-20% of the initial manipulated
variables.
7) For self-regulating system: Once steady state has been reached, the result was
recorded by printed the response curve.
8) For non-self-regulating: Once the response curve can be calculated, that response
curved was printed.
9) The data was collected from AIM Historian Data Display at desktop.
10) Numerical method has been chosen to calculate the Response rate, time delay, and
time constant.
11) PI controller setting were calculated using Ziegler Nichol’s and Cohen- Coon’s
method.
20 | P a g e
5.2.3 LOAD DISTURBANCE TEST
3) Then, the manual mode was changed to automatic mode after 3 seconds the MV
value was changed.
2) The set point was changed higher 10% from the current operating process value.
• Flow = 6 m3/h
• Level = 1000 mm
5.3.2 FOXBORO
• Flow = 3 m3/h
21 | P a g e
6.0 PRROCESS LINEARITY LAB
22 | P a g e
6.1.1 SAMPLE CALCULATION FOR PV (%)
𝑚3 𝑃𝑉(%) = 𝑃𝑉(𝑚𝑚) × 𝑎
0.5 1
𝑎=( ℎ × 100%) × %
𝑝𝑟𝑜𝑐𝑒𝑠𝑠 𝑠𝑝𝑎𝑛 15𝑚𝑚 𝑃𝑉(%) = 56 𝑚𝑚 × 0.5556( )
𝑚𝑚
𝑃𝑉(%) = 31.14%
𝑚3
0.5
𝑎=( ℎ × 100%) × 1
𝑚3 15𝑚𝑚
6
ℎ
%
𝑎 = 0.5556 ( )
𝑚𝑚
0 0 0
10 56 31.14
20 58 32.24
30 69 38.34
40 77 42.78
50 85 47.23
60 99 55.0
70 105 58.34
80 119 66.12
90 128 71.12
Table 2: Process Steady (PV) Values for Each Manipulated Variable (MV)
23 | P a g e
Process Response (PV %) versus Manipulated Variable (MV %)
90
80 y = 0.6262x + 15.719
70
60
PV (%)
50
40
30
20
10
0
0 20 40 60 80 100 120
MV (%)
6.1.2 DISCUSSION
A graph of PV versus MV was plotted to determine the linearity of the process. Generally,
we have to start from 0% of MV with 3 m3/h of set point but in this experiment.
Unfortunately, this experiment was started with 10% of MV instead. Based on the figure
11 it shows that this process is slightly nonlinear against the manipulated variable as the
process response are slightly move away from the linear threndline of the process. This
may be happened because the MV was started from 0% so the system was not
stabilized fully. Then, a mistake was happened as the data printed out was not the data
of the current reading but it is data from previous reading. It seems not a good data but
it does not affect the data reading at all. An enhanced nonlinear PID (EN-PID) controller
that exhibits the improved performance than the conventional linear fixed-gain PID
controller is proposed in this paper, by incorporating a sector-bounded nonlinear gain in
cascade with a conventional PID control architecture. (Su, Sun, & Duan, 2005)
24 | P a g e
6.2. FLOW CONTROL (FIC 31) - FOXBORO
An experiment was carried out with constant of 2,9 m3/h set point. It was set in a manual
mode to a flow controller FIC 31 by using Foxboro system. The MV of this process was
changed consistently until it reaches the limit of 100% with 10% differences. From the
experiment, the data below was obtained.
MV (100%) PV (m3/s)
10 0
20 0
30 0.5638
40 1.0030
50 1.4203
60 1.9075
70 2.2294
80 2.4263
90 2.4608
100 2.4608
Table 3: Process Steady (PV) Values for Each Manipulated Variable (MV)
25 | P a g e
PV (m3/s) vs MV (%)
3.5
2.5
2
PV (m3/s)
1.5
0.5
0
0 20 40 60 80 100 120
-0.5
MV (%)
6.2.1 DISCUSSION
Based on the data obtained, the graph of PV against MV was plotted. From the graph, it can
conclude that this process is also slightly non-linear. Since the PV is directly proportional to
MV, so the slightly nonlinear are because of the relationship between its dependent and
independent variables. To demonstrate that the extended linearization methodology can
address a large class of process control problems, attention will be focused on the design of
nonlinear proportional-integral-derivative (PID) controllers based on a parameterized version
of the well-known Ziegler-Nichols tuning specifications. In recent years, control schemes have
been proposed that involve nonlinear gain weightings of various types. The extended
linearization approach differs significantly from these methods in that the nonlinear
characteristics of the plant, as reflected in the set point and linearization families, are explicitly
used to specify the nonlinear gains. (W, 1987)
26 | P a g e
7.0 TANGENT METHOD LAB
Figure 14: An Open Loop Process Response when the MV was changed from 50% to 60%
27 | P a g e
7.1.1 CALCULATION
∆𝑥 = 11 𝑚𝑚
∆𝑦 = 27 𝑚𝑚
𝑀𝑉𝑖 = 50%
𝑀𝑉𝑓 = 60%
𝑚3
0.5
𝑎=[ ℎ × 100%] × 1
𝑚3 14𝑚𝑚
6
ℎ
%
𝑎 = 0.5952
𝑚𝑚
15𝑠 𝑠
𝑏= = 0.2308
65𝑚𝑚 𝑚𝑚
Time Constant, Tc:
𝑇𝑐 = 𝑇𝑐(𝑙𝑒𝑛𝑔𝑡ℎ) × 𝑏
𝑠
𝑇𝑐 = 11𝑚𝑚 × 0.2308
𝑚𝑚
𝑇𝑐 = 2.5388 𝑠
Dead Time, Td:
𝑇𝑑 = 𝑇𝑑(𝑙𝑒𝑛𝑔𝑡ℎ) × 𝑏
𝑠
𝑇𝑐 = 1𝑚𝑚 × 0.2308
𝑚𝑚
𝑇𝑑 = 0.2308 𝑠
Response Rate, RR:
∆𝑃 ∆𝑦
( ∆𝑡 ) (∆𝑥 ) 𝑎
𝑅𝑅 = = ( )
∆𝑀𝑉 ∆𝑀𝑉 𝑏
27
(11) 0.5952
𝑅𝑅 = ( )
10 0.2308
0.6330
𝑅𝑅 =
𝑠
28 | P a g e
Since this process was a tangent method, so it will obey the tuning rule by Cohen-Coon.
Where; Controller mode= Proportional – Integral (PI)
𝑇𝑑
𝜇=
𝑇𝑐
0.2308
𝜇=
2.5388
𝜇 = 0.09091
100
𝑃𝐵 = 𝜇 𝑅𝑅 𝑇𝑑
1+
11
100
𝑃𝐵 = (0.6330)(0.2308)
0.09091
1 + ( 11 )
𝑃𝐵 = 14.49%
𝜇
1 + 11
𝐼 = 3.33 [ ]𝑇
11𝜇 𝑑
1+
5
0.09091
1 + 11
𝐼 = 3.33 [ ] (0.2308)
(11)(0.09091)
1+
5
𝐼 = 0.6988 𝑠
In Emerson system, Gain, Kc was used instead of PB. So, Kc was calculated.
100
𝐾𝑐 =
𝑃𝐵
100
𝐾𝑐 =
14.49
𝐾𝑐 = 6.9013
29 | P a g e
Then, the Kc and I values were used in closed loop test. After that, for open loop test, the
load disturbance and set point test were performed.
Figure 15: Process Response Test for Closed Loop, Disturbance Test and Setpoint Test
Table 4- Summary of Proportional – Integral (PI) Mode for the Respective Lab
30 | P a g e
7.1.2 DISCUSSION
In tangent method, the process system was used is Emerson. The value of Response Rate
(RR), Date Time (Td) and Time Constant (Tc) were determined by using tangent as shown
in calculation section above, Then, the value of Proportional Band (PB) and Integral Time(I)
were calculated using the Tuning Rules by Cohen-Coon where the PI controller mode were
chosen for the process due to PI controllers have two adjustment parameters to adjust. The
integral action allows the PI controllers to eliminate the offset where it is a major weakness
of a P-only controller, Therefore PI controllers provide a balance of complexity and capacity
since the PI controller does not have the ability to predict the future systems errors, so it
cannot decrease the rise time and eliminate oscillations. Rather than that, derivative value
was determined in this experiment since the process response is fast as the manipulated
variable is change from one point to another point. (Sivaranjith, 2018) Other than that,
controller gain, Kc were used instead of PB value in the closed loop test.
Closed loop test, disturbance load test and setpoint test were performed and was showed
in Figure 15. So, from the Figure 15, can be seen that the graph was oscillated enormously
as the calculated value of Kc and I was inserted into the system where the calculated values
are still far from their optimum condition. Thus, a fine tuning was applied to correct the error
and to ensure the system were stabilized enough so there is no oscillation occur until it
reaches to the set point value. Fine tuning was done by reducing the value of Kc by dividing
the original calculated value which is 6.9013 with 4 and resulting with the new value of Kc
which is 1.7253. Then, the original calculated value of integral time (I), 0.6988 seconds
were multiplied with 4 giving the new value I with 2.7952 seconds. The increasing I and
decreasing Kc makes MV action slower where it results in a slower process response and
the new process response shifts to the right. This is because MV is directly proportional to
Kc and PV. (Ishak A. A., 2018) The result show that the oscillation starts to slow down a bit
and oscillated not rapidly as before. Then, fine tuning was performed again since the
process are still not stable where the value of Kc was divided again with 4. Same goes to
the interval time, I where it will be multiplied with 4 again. This resulting with Kc value of
0.4313 while I value is 11.1808. At this point, it gets a bit slower with a short peak and as it
reaches the setpoint, it gets calm and constant line as it has stabilized. Thus, the closed
loop test for this process ended.
31 | P a g e
Then, this experiment was proceeded with the load disturbance test. Firstly, the controller
was set into the manual mode and the current value MV was changed by adding up 10%.
In this process, the value of MV was changed from 50% to 60%. After that, the manual
mode was changed to automatic mode again after 3 seconds the MV value was changed.
By using the tuning rule of Cohen-Coon, it shows that there overshoot or undershoot value
where the graph is immediately straight increase in value to the set point before it stable
again and there is no oscillation occur. This shows that for disturbance load test, the result
was positive.
Lastly, the setpoint test was performed where the setpoint was changed with adding up
10% of the operating process value which is 0.6 m3/h. The value was changed from 3.0
m3/h to 3.6 m3/h. From the result, it shows that there is a slightly oscillation occur for the
process to reach the new set point value. However, after a second, the process calmly
reaching to the new setpoint. This is because all control loop has a tendency to oscillate
because of the built-in-timing constant of the control system components and the
dynamically changing variables. (Controls, 2019)
32 | P a g e
8.0 REFORMULATED TANGENT METHOD
8.1 LEVEL CONTROL (LIC 11)- EMERSON
An open loop process response when a change I manipulated variable was made from
24.5% to 34.5%. The process dead time (Td), response rate (RR) and time constant (Tc)
is determined from the figure---- using reformulated tangent method.
100
𝑎= = 6.25%/𝑚𝑚
16
1𝑠
𝑏= = 0.07143𝑠/𝑚𝑚
14𝑚𝑚
𝜃 = 15°
MVi=24.5%
MVf=34.5%
∆𝑀𝑉 = 𝑀𝑉𝑓 − 𝑀𝑉𝑖
=34.5-24.5
= 10%
33 | P a g e
8.1.1 SAMPLE CALCULATION
34 | P a g e
After
Tuning
Setpoint
Test
Before Load
Tuning Disturbance
Test
8.1.2 DISCUSSION
From the calculation part, the original PB and I value for the experiment which uses
reformulated tangent method are 11.5491% and 17.6294s, respectively. These values
are calculated using the reformulated tangent method and Zeigler-Nicholas method. First
of all, the value of P is converted into Kc by using formula, thus Kc=8.6251. Then we ran
closed loop test and performance test to make sure that these values are suitable for the
process. It turns out that these values are slightly less accurate, thus we proceed with the
fine-tuning process for LIC11. From the result, we find that these values are quite not
accurate and precise since it produces oscillation and over damping.
35 | P a g e
9.0 NUMERICAL METHOD LAB
36 | P a g e
9.1.1 SAMPLE CALCULATION
Change Unit of PV from mm/h to %
m3
𝑃𝑉 ( )
𝑃𝑉(%) = h 𝑥 100%
m3
𝑃𝑟𝑜𝑐𝑒𝑠𝑠 𝑠𝑝𝑎𝑛 ( )
h
At t = 1,
m3 m3
𝑃𝑉 ( ) 0.981621 ( )
h h
𝑃𝑉(%) = m3 𝑥 100% = m3 𝑥 100% = 32.72%
𝑃𝑟𝑜𝑐𝑒𝑠𝑠 𝑠𝑝𝑎𝑛 ( ) 3( )
h h
Response Rate, RR
PV − PV
RR = 1 −1
2 ∆h ∆MV
At t = 2,
PV − PV 42.44%−32.72%
1 −1
RR = = 𝑥 100% = 0.486/𝑠
2 ∆h ∆MV 2 (1)(10%)
All the data for RR is determined for each respective time. RRmax is determined from the
highest value of RR presence in the calculated data of RRs.
0.486 60 𝑠
RRmax = ( ) = 29.16/min
𝑠 1 𝑚𝑖𝑛
Dead Time, Td
PV − PV
1 𝑖
Td = t1 − 2 ∆h( )
𝑃𝑉 − 𝑃𝑉
1 −1
Td = 1s
For Foxboro system, time unit must be converted to min. Therefore,
1𝑚𝑖𝑛
Td = 1s ( 60𝑠
) = 0.0167 min
37 | P a g e
1𝑚𝑖𝑛
TC = 2.9938s ( ) = 0.0499 𝑚𝑖𝑛
60𝑠
𝜇 𝑇𝑑 0.0167
1+ 1+ 1+
11 11𝑇𝑐 11(0.0499)
I (Time) = 3.33 Td[ 11𝜇 ] = 3.33Td [ 11𝑇𝑑 ] = 3.33(0.0167min) [ 11(0.0167) ] = 0.5935 min
1+ 1+ 1+
5 5𝑇𝑐 5(0.0499)
The PB and I value is then used for closed loop test proceeding with load disturbance test
and set point test for an open loop process response. The figure shows the result of the
three test.
Before tuning
process After tuning
process
Load disturbance
test Set point test
Figure 18 - Process Response for Closed Loop Test, Load Disturbance Test and Set Point
Test.
38 | P a g e
9.1.2 DISCUSSION
Firstly, there is an open loop test. This test must be done manually as the modified
variables are adjusted as a 10% rise. The process can also be handled by the person in
manual mode, which ensures that the process follows the person's instructions. The test is
completed as the answer to the process reaches a stable state. The answer of the process
must follow any changes.
Response rate, Dead Time and Time constant can actually be measured using tangent
and reformulated tangent method as well as numerical method using graphical method. The
obvious difference between these two methods is the graphical approach that uses a graph to
calculate RR, Td, and TC, but the numerical method requires a data table to calculate RR, T d,
and TC. Apart from that, numerical methods should also be used.
Cohen Coon method is chosen to calculate the value or PB in percent and I in min for
the tuning check. There are many forms of tuning method but Cohen Coon is being used for
this experiment. The PB, I, and D meaning is considered dependent on process forms. PB is
only considered for processes that are not stable while PB is called standard, flow, and others
more for processes included. PB, I, D.
39 | P a g e
Figure 19 shows the result for the numerical system data tuning procedure. The value
of PB is 47.26%, while the value of I is 0.5935 min. When these two values are key in, the first
graph shows no oscillation. This implies that the value for this method is the optimal value.
Load disturbance monitoring was done. The graph shows that the response to the process is
easy to answer very quickly during the study. This indicates that the optimal value is right and
there is no need to recalibrate. Set point test logically to see how quickly the process response
will respond to the set point change and again it shows that the time taken for the process
response shows that the value of PB and I for this process is the optimum value.
40 | P a g e
10.0 DISCUSSION
Reformulated
Tangent LIC11 11.5941 8.6251 5.5941 - 2.1563 17.6294
Method
Numerical
Method FIC31 47.26 - 35.61 47.26 - 35.61
Table 7- Summary of Proportional – Integral (PI) Mode for the Respective Lab
The time unit used for closed loop monitoring, load disturbance testing and set point
testing is in minute, minute for FOXBORO system. Nonetheless, the data tabulated for Integral
Time is considered in second, s in order to see the relation between the data meaning for all
the laboratory involved. Table above indicates the percentage of error occurring before
optimizing the process control loop.
41 | P a g e
A decrease in the Controller gain, Kc value, resulting in a slower controlled variable,
MV action. Because with Proportional Band (PB), Kc is inversely proportional, the PB value i
ncreases as Kc is reduced. As PV is directly proportional to MV, this in effect slows down t
he process response. The response curve of the method will shift to the right. As a result, the
process response is aggressively and enormously oscillating to be calmer and smaller. The
rise in I reduces the operation of the controller as I am inversely proportional to MV. As MV is
directly proportional to PV, a slower action of MV will result in a slower process response, PV.
A slower response to the process will shift the response curve to the right. This results in the
reduction of the number of oscillations and the stable response of the cycle to the set point.
Compared with the Tangent method and the Reformulated Tangent method, numerical
method yields the most accurate data according to the tuning laws.
11.0 CONCLUSION
FOXBORO and EMERSON flow control loop demonstrate both a somewhat nonlinear
process response relationship as well as a manipulated component. To order to confirm the
optimal condition for the process control loop, closed loop test, load disturbance test and set
point test can be conducted. A decrease in the gain of the controller, the Kc value, slows down
the process response, PV, and shifts the process response curve right. An increase in the
value of PB, slows down the response of the process, PV, and shifts the response curve to
the right. A rise in I will lead to a slower process response, while PV will change the process
response curve to the right. Similar to the Tangent method and the Reformulated Tangent
method, numerical method yields the most accurate data that corresponds to the tuning law.
12.0 RECOMMENDATIONS
There are few recommendations that can be made during the experiment to ensure that the
outcome is accurate. The first suggestion that can be made is to ensure that the tangent line
is straight and that it fits the curve to which the calculation error. For tangent and
reformulated tangent process, this can be achieved. In addition, the numerical approach is
prone to the positions taken by the decimal places. The number of decimal places throughout
the experiment must be constant. We are taking 4 decimal places in the calculation of this
experiment but the value that was inserted into the system was in 2 decimal point. Other than
that, to ensure that the PB and I value is the optimal value that does not result in a circulating
graph, the result that is the response rate curve needs to be expanded where for all results,
the RR value must be estimated to ensure that the overall RR for the experiment can be
42 | P a g e
determined. In order for the experiment to evaluate the process to be linear or not, the phase
size or changes made to the manipulated variables must be constant in the range 5-20 percent
because the outcome depends on the changes made.
13.0 REFERENCES
controlguru. (2019). Controller Tuning Using Closed. Retrieved from Controller Tuning Using
Set Point Driven Data: https://controlguru.com/controller-tuning-using-set-point-
driven-data/
elprocus. (2019). The Working Principle of a PID Controller for Beginners. Retrieved from
https://www.elprocus.com/the-working-of-a-pid-controller/
Ishak, A., & Hussain, M. (2000). REFORMULATION OF THE TANGENT METHOD FOR PID
CONTROLLER TUNING. Kuala Lumpur: Faculty of Engineering, Universiti Malaya.
Nasir. (2018). Difference Between Open Loop & Closed Loop Systems. Retrieved from
Electrical Equipment: http://engineering.electrical-equipment.org/panel-
building/difference-between-open-loop-closed-loop-systems.html
Neary, E. (2004, January). AnalogDialogue. Retrieved from Mixed-Signal Control Circuits Use
Microcontroller for Flexibility in Implementing PID Algorithms:
https://www.analog.com/en/analog-dialogue/articles/mixed-signal-control-circuits-pid-
algorithms.html#
Nor Rahman, M., Musdafa kamal, N., Mat Yuna, N., & Md Tahir, N. (2019). PROCESS
CONTROL AND INSTRUMENTATION. Shah Alam: FKK, UiTM Shah Alam.
opticontrols. (2011, March 7). PID Controllers Explained. Retrieved from Manual Control:
https://blog.opticontrols.com/archives/344
43 | P a g e
RaghavendraRao. (2017, October 12). Process Stability and Capability. Retrieved from
Benchmark6ix sigma: https://www.benchmarksixsigma.com/forum/topic/34898-
process-stability-process-capability/
Shinskey, F. G. (2011, May 09). Meditating on Disturbance Dynamics. Retrieved from Loop
Control: https://www.controlglobal.com/articles/2011/meditating-on-disturbance-
dynamics/
Sivaranjith. (2018, April 17). AutomationForum.Co. Retrieved from Continuous control mode-
P, PD, PI & PID control modes: https://automationforum.co/continuous-control-mode-
p-pd-pi-pid-control-modes/
Station, C. (2019). How to Perform a Step Test. Retrieved from Control Station:
https://controlstation.com/perform-step-test/
Su, Y., Sun, D., & Duan, B. (2005). Design of an enhanced nonlinear PID controller.
Mechatronics, 15.
Tests, S. (2015). SPSS Tests. Retrieved from Step By Step to Test Linearity Using SPSS:
https://www.spsstests.com/2015/03/step-by-step-to-test-linearity-using.html
Yalcin, D. (2017, December 14). Open-Loop vs Closed-Loop Systems in the Materials Testing
Industry. Retrieved from Materials Testing System Manufaturer:
https://www.admet.com/open-loop-vs-closed-loop-systems-materials-testing-industry/
44 | P a g e