Sunteți pe pagina 1din 33

CDB 3062

CHEMICAL ENGINEERING LAB III

SEPT 2018 SEMESTER

PROJECT WORK

NAME ID
ONG JIA YEE 22565
SUSANNE DORIS 22570
SHARIFAH NASIBAH KHAIRANI BINTI SYED 22595
AHMAD FUADDIN
AIMI FARZANA BINTI YAZID 22596
FOO PEI LING 22599

Submitted to :
DR NASSER MOHAMED RAMLI

Date of Submission :
12th NOVEMBER 2018
Table of Contents
1.0 SUMMARY ..................................................................................................................... 3
2.0 INTRODUCTION & THEORY ...................................................................................... 4
3.0 PROCEDURES ............................................................................................................... 9
4.0 CALCULATION ........................................................................................................... 10
4.1 Determination of First Order Plus Dead Time (FOPDT) Transfer Function .......................... 10
4.2 Calculation of Controller Tuning Parameters ......................................................................... 11
4.3 Calculation of Overshoot, Settling Time and Decay Ratio ..................................................... 13
5.0 RESULT ........................................................................................................................ 15
5.1 Proportional (P) only Controller ............................................................................................. 15
5.2 Proportional-Integral (PI) Controller ...................................................................................... 17
5.3 Proportional-Integral-Derivative (PID) Controller ................................................................. 19
6.0 DISCUSSION ................................................................................................................ 23
6.1 Comparison of Proportional (P) only Controller .................................................................... 23
6.2 Comparison of Proportional-Integral (PI) Controller ............................................................. 25
6.3 Comparison of Proportional-Integral-Derivative (PID) Controller ........................................ 26
6.4 Comparison of P, PI and PID Controller for Set Point Change .............................................. 27
6.5 Comparison of P, PI and PID Controllers for Disturbance Change ........................................ 28
7.0 CONCLUSION .............................................................................................................. 29
8.0 REFERENCES .............................................................................................................. 30
9.0 APPENDIX.................................................................................................................... 31
9.1 Tabulation of Data .................................................................................................................. 31
1.0 SUMMARY
A system can be defined as a collection of subsystems which direct or control an input signal
to produce the desired output condition. There are two types of system which are open loop
system and closed loop system. An open loop system which is also known as non-feedback
system, is a type of continuous control system in which the output has no influence or effect
on the control action of the input signal. A closed-loop system which is also known as a
feedback control system, is a type of control system which uses the concept of an open loop
system as its forward path but has one or more feedback loops or paths between its output and
its input.
In this project, open-loop system is used to study the effect of step change and disturbance
change to determine the best tuning method for different types of controller. The objective of
the project is to apply various tuning methods to different kinds of changes which are set-point
change and disturbance change that will occur in the processes that have different kinds of
controllers which are Proportional (P) Controller, Proportional-Integral (PI) Controller and
Proportional-Integral-Derivative (PID) Controller, to determine the optimum tuning for each
process, which is the one with fast settling-time, less error, and less overshoot. There are seven
processes all together.
To achieve the objectives that were proposed, each one of the processes is run with MATLAB
software and its feature Simulink. The results are analyzed, and MATLAB tuning method were
discussed. There were six situations in total and they are described as follows:
1. Set-point change for a proportional (P) controller.
2. Disturbance change for a proportional (P) controller.
3. Set-point change for proportional and integral (PI) controller.
4. Disturbance change for proportional and integral (PI) controller.
5. Set-point change for proportional-integral-derivative (PID)controller.
6. Disturbance change for proportional-integral-derivative (PID) controller.
The tuning methods used were: Quarter Decay Ratio Response, Integral Absolute Error (IAE),
Integral Time-weighted Absolute Error (ITAE), Cohen-Coon, and Dahlin Synthesis. Some
applications may require using only one or two actions to provide the appropriate system
control. This is achieved by setting the other parameters to zero. To choose the best tuning
method, the integral of absolute error (IAE), settling time and overshoot of the controller are
observed.
2.0 INTRODUCTION & THEORY
Open loop controller is also known as a non-feedback controller. The control action from the
controller is independent of the process output, which is the process variable that is being
controlled. It does not use feedback to determine if its output has achieved the desired goal of
the input command or process set point. There are a large number of open loop controls, such
as on/off switching of valves, machinery, lights, motors or heaters, where the control result is
known to be approximately sufficient under normal conditions without the need for feedback.
The advantage of using open loop control in these cases is the reduction in component count
and complexity. However, an open-loop system cannot correct any errors that it makes or
correct for outside disturbances, and cannot engage in machine learning.
Fundamentally, there are two types of control loop which is known as open loop (feed forward)
control, and closed loop (feedback) control. In open loop control, the control action from the
controller is independent of the process output or known as controlled process variable. A good
example of this is a central heating boiler controlled only by a timer, so that heat is applied for
a constant time, regardless of the temperature of the building. The control action is the
switching on/off of the boiler, but the controlled variable should be the building temperature,
but is not as this is open-loop control of the boiler, which does not give closed-loop control of
the temperature.
In closed loop control, the control action from the controller is dependent on the process output.
In the case of the boiler analogy this would include a thermostat to monitor the building
temperature, and thereby feedback a signal to ensure the controller maintains the building at
the temperature set on the thermostat. A closed loop controller therefore has a feedback loop
which ensures the controller exerts a control action to give a process output the same as the
"reference input" or “set point". For this reason, closed loop controllers are also called feedback
controllers.
In feedback control, the objective is to reduce the error signal to zero. Error signal is defined
by
𝑒(𝑡) = 𝑦𝑆𝑃(𝑡) − 𝑦𝑚(𝑡)
Where 𝑦𝑆𝑃 = set-point
𝑦𝑚 = measured value of the controlled variable (or equivalent signal from
transmitter)
Thus, different types of controller which are proportional (P) controller, proportional-integral
(PI) controller and proportional-integral-derivative (PID) controller are introduced.

Proportional (P) Controller


When a process has a small capacity, it usually responds quickly to upsets. Therefore, precise
and continuous regulation of the manipulated variable is needed. Proportional control attempts
to stabilize the system and avoid fluctuations by responding to the magnitude as well as the
direction of the error. The relationship between the output and the width of the measurement
span is called the proportional band.
For proportional control, the controller output is proportional to the error signal
𝑝(𝑡) = 𝑝̅+ 𝐾𝑐𝑒(𝑡)
Where,
(𝑡) = controller output
𝑝̅= bias value (adjustable, manual reset)
𝐾𝑐 = controller gain (dimensionless, adjustable, tuning)
Proportional-Integral (PI) Controller
To optimize a control system, the introduction of integral controller is important as it acts as
an offset eliminator. Integral control is seldom used alone, instead it is usually paired with
proportional control to form PI controllers. PI controller is generally used on processes where
no amount of offset can be tolerated.
Integral control is used in conjunction with proportional control as the proportional integral
(PI) controller
(𝑡) = 𝑝̅+ 𝐾𝑐 [(𝑡) + 1 𝜏1 ∫ 𝑒(𝑡 ′ )𝑑𝑡 ′ 1 0
Where
1 𝜏1 = repeats per minute or reset rate

The disadvantages of PI controller is that it produces oscillatory response, reset windup when
a sustained error occurs and the integral term becomes very large and the controller output
eventually saturates. Proportional-Integral (PI) controller does not have offset and has better
dynamic response than reset alone and has possibilities exist for instability due to lag
introduced.

Proportional-Integral-Derivative (PID) Controller


PID controller on the other hand employs three general control modes namely, proportional,
integrative and differential modes. In other words, PID controller considers all aspect of
process errors, namely direction, magnitude, duration and rate of change. PID control can be
advantageous on many processes. According to Zhuang et. al (1993), processes that benefit
most from PID control are those with rapid and large disturbances for which derivative action
can respond to the rapidity of the changes whereas the integral action can respond to the
duration of them. This shows that all three modes of the controller complement each other.
When the proportional mode produces offset, the integrative control mode will eliminate the
error whereas the derivative control action will ensure that the system is stable, non-oscillatory
and possesses a shorter settling time (Bhilai, 2014).

Quarter Decay Ratio Response


This method was proposed by Ziegler and Nichols in 1942, and it consists of two steps: Step
1. Determination of the dynamic characteristics, or personality, of the control loop. Step 2.
Estimation of the controller tuning parameters that produce a desired response for the dynamic
characteristics determined in the first step-in other words, matching the personality of the
controller to that of the other elements in the loop. In this method the dynamic characteristics
of the process are represented by the ultimate gain of a proportional controller and the ultimate
period of oscillation of the loop. The parameters can be determined by the direct substitution
method if the transfer functions of all the components of the loop are known quantitatively.
But because this is not usually the case, we must often experimentally determine the ultimate
gain and period from the actual process by the following procedure:

1. Switch off the integral and derivative modes of the feedback controller so as to have a
proportional controller. In some controllers, the integral mode cannot be switched off but
can be de-tuned by setting the integral time to its maximum value or, equivalently, the
integral rate to its minimum value.

2. With the controller in automatic (i.e., the loop closed), increase the proportional gain (or
reduce the proportional band) until the loop oscillates with constant amplitude. Record
the value of the gain that produces sustained oscillations as Kcu, the ultimate gain. This
step is carried out in discrete gain increments, bumping the system by applying a small
change in set point at each gain setting. To prevent the loop from going unstable, smaller
increments in gain are made as the ultimate gain is approached

Integral of Absolute Error (IAE), Integral of Time-Weighted Absolute Error (ITAE) and
Integral of Squared Error (ISE)

All the measures require a fixed experiment to be performed on the system and the integrals
are evaluated over a fixed time period. ISE integrates the square of the error over time. ISE
will penalise large errors more than smaller ones because the square of a large error will be
much bigger. Control systems specified to minimise ISE will tend to eliminate large errors
quickly, but will tolerate small errors persisting for a long period of time. Often this leads to
fast responses, but with considerable, low amplitude, oscillation.

IAE integrates the absolute error over time. It doesn't add weight to any of the errors in a
systems response. It tends to produce slower response than ISE optimal systems, but usually
with less sustained oscillation.
ITAE integrates the absolute error multiplied by the time over time. What this does is to weight
errors which exist after a long time much more heavily than those at the start of the response.
ITAE tuning produces systems which settle much more quickly than the other two tuning
methods. The downside of this is that ITAE tuning also produces systems with sluggish initial
response which is necessary to avoid sustained oscillation.

Ho et. al (1995) developed tuning formula for minimum error based on first order plus dead
time transfer. These formulas will show the same response as quarter decay ratio except that
the integral time depends more on the effective process time constant and less on the process
dead time.

Dahlin Synthesis
Dahlin synthesis or also known as lambda tuning was developed by Dahlin and Higham in
1968. It is used extensively in industry in its digital form, specifically in paper machine control
system. The process is assumed to be first order transfer function. The tuning parameter
determines the pole location. The smaller the lambda the better the performance the controller,
but the closer it takes to the instability region

Controller Type Proportional Integral Time Derivative Time


Gain
I D
Kc

Proportional integral 𝜏 𝜏

derivative, PID K (to + ) 𝜏 2

Cohen-coon
The Cohen-Coon tuning rules are suited to a wider variety of processes than the Ziegler-
Nichols tuning rules. The Ziegler-Nichols rules work well only on processes where the dead
time is less than half the length of the time constant. The Cohen-Coon tuning rules work well
on processes where the dead time is less than two times the length of the time constant.
Controller Type Proportional Gain, Integral Time I Derivative Time D
Kc
Proportional only, P - -
1 𝜏 𝑡𝑜
( ) (1 + )
𝐾 𝑡𝑜 3𝜏

Proportional 1 𝜏 9 𝑡𝑜 30 + 3(𝑡𝑜 ⁄𝜏 ) -
( )( + ) 𝑡𝑜
integral, PI 𝐾 𝑡𝑜 10 12𝜏 22 + 3(𝑡𝑜 ⁄𝜏 )
Proportional integral 1 𝜏 4 𝑡𝑜 32 + 6(𝑡𝑜 ⁄𝜏 ) 4
( )( + ) 𝑡𝑜 𝑡𝑜
derivative, PID 𝐾 𝑡𝑜 3 4𝜏 13 + 8(𝑡𝑜 ⁄𝜏 ) 11 + 2(𝑡𝑜 ⁄𝜏 )
3.0 PROCEDURES
1. First Order Plus Dead Time (FOPDT) method is used to determine constants. The
controller is set to manual mode or open loop response.
2. From the Manual Control, a suitable step test parameters (step time and step size) is
performed to calculate the value of process steady state gain (K), dead time (t0) and
time constant (τ) as shown in the figure below. The detailed calculations are showed in
the next chapter.

Figure 2.1: The constants obtained from the open loop response

3. After determining the process steady state gain (K), dead time (t0) and time constant
(τ), the proportional gain (KC), integral time (τi) and derivative time (τD) can be
calculated by using the tuning formulas attached in the appendix. The detailed
calculations are showed in the next chapter.

4. After calculating the tuning parameters, Proportional (P), Integral (I) and Derivative
(D) are determined for P controller, PI controller and PID controller as follows:

i. For P controller, only proportional gain, Kp =Kc


ii. For PI controller, proportional gain, Kc and integral gain, KI = Kc/ τi
iii. For PID controller, proportional gain, Kc and integral gain, KI = Kc/ τi and
derivative gain, KD = KPτD

5. The calculated P, I and D values are inserted into the automatic mode page and the step
change in set point and disturbance is specified.
6. The response graphs are obtained for five tuning methods. The response graphs of PV
are analyzed and interpreted using different parameters. The needed parameters are
overshoot, settling time and decay ratio. The detailed calculations for the parameters
are showed in the next chapter.
7. The best tuning method among P controller, PI controller and PID controller for set
point and disturbance is selected depends on the parameters obtained from the graphs.
8. The ultimate best controller for set point change and disturbance change is selected.
4.0 CALCULATION
4.1 Determination of First Order Plus Dead Time (FOPDT) Transfer Function
An open loop response is simulated by using Microsoft Excel to determine the transfer function
of First Order Plus Dead Time (FOPDT). A step change of 30 is introduced to the system at a
step time of 10th minute. To form the transfer function of FOPDT model, the values of process
gain, 𝐾, time constant, 𝜏, and dead time, 𝑡0 must be determined from the graph as shown below.

3625

1600

16.5

Figure 4.1.1: Process reaction curve for open loop response

1. Dead Time, 𝒕𝟎
Dead time, t0 is the delay from when a controller output (CO) signal is issued until when
the measured process variable (PV) first begins to respond. From the graph in Figure
3.1.1, we can observe that the response starts at 15th minute which is 5 minutes after the
change is introduced to the system which is at 10th minute. Thus, the dead time, 𝑡0 of
the system is calculated as the following:
𝑡0 = 15 𝑚𝑖𝑛𝑠 − 10 𝑚𝑖𝑛𝑠 = 5 𝑚𝑖𝑛𝑠
2. Process Gain, 𝑲
According to Smith and Corripio (1985), gain, 𝐾 is “the steady-state change in output
divided by the sustained change in input”. To determine the gain of the process, the
value of initial steady-state output which is 1600 and the value of final steady-state
output which approaches 3625 as indicated by blue-coloured arrow are obtained from
the graph in Figure 3.1.1. Since a step change of 30 is introduced into the process, hence
the gain of the process is calculated as the following:
∆𝑝𝑣
𝐾=
∆𝑚𝑣
3625 − 1600
𝐾= = 67.5
30

3. Time Constant, 𝝉
Time constant is a variable which describes the speed with which the measured process
variable responds to changes in the output. To be more specific, time constant is the
time required for the process variable to reach 63.2% of its total and final change.

𝑦 = 0.632(∆𝑝𝑣)
𝑦 = 0.632 × 3625 = 2291

At 63.2% of its final change, extrapolation is done from y-axis to x-axis to determine
the time required to reach 63.2% of its final change as indicated by the orange-coloured
arrow which is 16.5 minutes. Hence, the time constant of the process is calculated as
the following:
𝜏 = 16.5 𝑚𝑖𝑛 − 15 𝑚𝑖𝑛 = 1.5 𝑚𝑖𝑛

4. Transfer Function
Transfer function is a function which describes the characteristic of the process. The
modified standard transfer function of FOPDT is as the following:

𝐾𝑒 −𝑡0 𝑠
𝐺(𝑠) =
𝜏𝑠 + 1
Therefore, the transfer function of the process is shown as following:
67.5𝑒 −5𝑠
𝐺(𝑠) =
1.5𝑠 + 1

4.2 Calculation of Controller Tuning Parameters


After the values of process gain, 𝐾, time constant, 𝜏, and dead time, 𝑡0 are determined, the
values are substituted into the tuning formulas which are shown in 8.0 Appendix to calculate
the values of proportional gain, 𝐾𝐶 , integral time, 𝜏𝐼 and derivative time, 𝜏𝐷 . The following
table shows the sample calculation of proportional gain, 𝐾𝐶 , integral time, 𝜏𝐼 and derivative
time, 𝜏𝐷 by using the tuning formula of Quarter Decay Ratio Response:
Table 4.2.1: Sample calculation of proportional gain, 𝐾𝐶 , integral time, 𝜏𝐼 and derivative
time, 𝜏𝐷 for Quarter Decay Ratio Response Tuning
Controller Type Proportional Gain Integral Time Derivative Time
𝑲𝑪 𝝉𝑰 𝝉𝑫
−1
Proportional 1 𝑡0 - -
( )
only, P 𝐾 𝜏
1 5 −1
= ( )
67.5 1.5
= 0.00444
Proportional 0.9 𝑡0 −1 3.33𝑡0 -
( )
integral, PI 𝐾 𝜏 = 3.33(5)
−1
0.9 5 = 16.650
= ( )
67.5 1.5
= 0.00400
Proportional 1.2 𝑡0 −1 2.0𝑡0 1
( ) 𝑡
integral 𝐾 𝜏 = 2.0(5) 2 0
derivative, PID 1.2 5 −1 = 10.000 1
= ( ) = (5)
67.5 1.5 2
= 0.00533 = 2.500

After the proportional gain, KC, integral time, 𝜏I and derivative time, 𝜏D are calculated, the
proportional gain, 𝐾𝐶 , integral gain, 𝐾𝐼 and derivative gain, 𝐾𝐷 for proportional (P) controller,
proportional integral (PI) controller and proportional integral derivative (PID) controller are
calculated by the following formulas:
Proportional gain: 𝐾𝑐 = 𝐾𝑐
𝐾𝑐
Integral gain: 𝐾𝐼 = 𝜏𝐼
Derivative gain: 𝐾𝐷 = 𝐾𝑐 × 𝜏𝐷
The following table shows the sample calculation of proportional gain, 𝐾𝐶 , integral gain, 𝐾𝐼
and derivative gain, 𝐾𝐷 by using the tuning formula of Quarter Decay Ratio Response:

Table 4.2.2: Sample calculation of proportional gain, 𝐾𝐶 , integral gain, 𝐾𝐼 and derivative
gain, 𝐾𝐷 for Quarter Decay Ratio Response Tuning
Controller Type Proportional Gain Integral Gain Derivative Gain
𝑲𝑪 𝑲𝑰 𝑲𝑫
Proportional 0.00444 - -
only, P

Proportional 0.00400 0.00400 -


integral, PI 16.650
= 0.000240
Proportional 0.00533 0.00533 0.00533 × 2.500
integral 10.000 = 0.0133
derivative, PID = 0.000533

After the proportional gain, 𝐾𝐶 , integral gain, 𝐾𝐼 and derivative gain, 𝐾𝐷 are calculated, the
values are substituted into the controller parameters in Microsoft Excel for both set point
change and disturbance to observe the error, overshoot, decay ratio and settling time of the
process for different type of controller using different type of tuning formula.

4.3 Calculation of Overshoot, Settling Time and Decay Ratio


The following shows the sample calculation of overshoot, settling time and decay ratio of the
PID controller for set point change using Dahlin Synthesis tuning formula.

a
c
b

b
b

t1 t2
Figure 4.3.1: Graph of process variable against time of PID controller for set point change
using Dahlin Synthesis setting
1. Overshoot
𝑎
𝑂𝑣𝑒𝑟𝑠ℎ𝑜𝑜𝑡 =
𝑏
According to the graph in Figure 3.3.1,
𝑎 = 1641 − 1630 = 11
𝑏 = 1630 − 1600 = 30
Hence, the overshoot of PID controller for set point change using Dahlin Synthesis
setting is calculated as following:
11
𝑂𝑣𝑒𝑟𝑠ℎ𝑜𝑜𝑡 = = 0.367
30
2. Settling Time
𝑆𝑒𝑡𝑡𝑙𝑖𝑛𝑔 𝑡𝑖𝑚𝑒 = 𝑡2 − 𝑡1
According to the graph in Figure 3.3.1,
𝑡1 = 10 𝑚𝑖𝑛𝑠
𝑡2 = 54.60 𝑚𝑖𝑛𝑠
Hence, the settling time of PID controller for set point change using Dahlin Synthesis
setting is calculated as following:
𝑆𝑒𝑡𝑡𝑙𝑖𝑛𝑔 𝑡𝑖𝑚𝑒 = 54.60 𝑚𝑖𝑛𝑠 − 10 𝑚𝑖𝑛𝑠 = 46.40 𝑚𝑖𝑛𝑠
3. Decay Ratio
𝑐
𝐷𝑒𝑐𝑎𝑦 𝑟𝑎𝑡𝑖𝑜 =
𝑎
According to the graph in Figure 3.3.1,
𝑎 = 1641 − 1630 = 11
𝑐 = 1632.5 − 1630 = 2.5
Hence, the decay ratio of PID controller for set point change using Dahlin Synthesis
setting is calculated as following:
2.5
𝐷𝑒𝑐𝑎𝑦 𝑟𝑎𝑡𝑖𝑜 = = 0.227
11
5.0 RESULT
In this project, five tuning methods are introduced, analysed and compared for first order plus
dead time (FOPDT) model. The tuning methods are quarter decay ratio response, integral of
absolute error (IAE), integral of time-weighted absolute error (ITAE), Dahlin synthesis and
Cohen Coon. The following shows the graphs of process variable against time towards both
set-point change and disturbance change for different type of controllers which is P controller,
PI controller and PID controller using different tuning methods.

5.1 Proportional (P) only Controller


Proportional (P) only controller is a controller that gives output which is proportional to current
error. It compares desired or set point with actual value or feedback process value.
The following equation shows the working principle of Proportional (P) only controller:

𝑝(𝑡) = 𝑝̅ + 𝐾𝑐 𝑒(𝑡)
where 𝑝(𝑡) = controller output
𝑝̅ = bias value
𝐾𝑐 = controller gain
𝑒(𝑡) = error signal

There are two type of tuning methods for P controller towards set point change and disturbance
change which are:
1. Quarter Decay Ratio Response
2. Cohen Coon

The following shows the graphs of process variable against time of P controller towards set
point change using tuning methods of Quarter Decay Ratio Response and Cohen Coon.

(a) (b)
Figure 5.1.1: Graph of process variable against time of P controller towards set point
change using tuning methods of (a) Quarter Decay Ratio Response and (b) Cohen Coon
Based on the graph in Figure 5.1.1, the process response before achieving the new steady-state
of P controller for both Quarter Decay Ratio Response and Cohen Coon tuning methods is
oscillatory. However, the process response before achieving the new-steady state for Cohen-
Coon tuning method is more oscillatory compared to that of Quarter Decay Ratio Response as
Cohen-Coon method has more peak compared to that of Quarter Decay Ratio Response which
only has only one peak. Furthermore, we can also observe that response for both Quarter Decay
Ratio Response and Cohen-Coon has offset. Nevertheless, Cohen-Coon has a smaller offset
compared to that of Quarter Decay Ratio Response as the new steady-state of Cohen-Coon is
closer to the set point compared to that of Quarter Decay Ratio Response which the new steady-
state of Cohen-Coon is further from the set point. Other than that, Quarter Decay Ratio
Response reaches the new steady-state faster than that of Cohen-Coon where Quarter Decay
Ratio Response takes shorter time to settle compared to that of Cohen-Coon.

The following shows the graphs of process variable against time of P controller towards
disturbance change using tuning methods of Quarter Decay Ratio Response and Cohen Coon.

(a) (b)
Figure 5.1.2: Graph of process variable against time of P controller towards disturbance
change using tuning methods of (a) Quarter Decay Ratio Response and (b) Cohen Coon

Based on the graph in Figure 5.1.2, the process response before achieving the new steady-state
of P controller for both Quarter Decay Ratio Response and Cohen Coon tuning methods is
oscillatory. In addition, the first peak of Quarter Decay Ratio Response has the same height as
that of Cohen-Coon. However, the process response before achieving the new-steady state for
Cohen-Coon tuning method is more oscillatory compared to that of Quarter Decay Ratio
Response as Cohen-Coon method has more peak compared to that of Quarter Decay Ratio
Response which only has only one peak. Furthermore, we can also observe that response for
both Quarter Decay Ratio Response and Cohen-Coon has offset. Nevertheless, similar to the
set point change, Cohen-Coon has a smaller offset compared to that of Quarter Decay Ratio
Response as the new steady-state of Cohen-Coon is closer to the set point compared to that of
Quarter Decay Ratio Response. Other than that, Quarter Decay Ratio Response reaches the
new steady-state faster than that of Cohen-Coon where Quarter Decay Ratio Response takes
shorter time to settle compared to that of Cohen-Coon. In conclusion, the process response of
P controller towards both set point change and disturbance change are quite similar.
5.2 Proportional-Integral (PI) Controller
Proportional-Integral (PI) controller is a controller which uses integral action in conjunction
with proportional band to remove offset.
The following equation shows the working principle of PI controller:

𝑡
1
𝑝(𝑡) = 𝑝̅ + 𝐾𝑐 [𝑒(𝑡) + ∫ 𝑒(𝑡 ′ )𝑑𝑡 ′
𝜏𝐼
0
where 𝑝(𝑡) = controller output
𝑝̅ = bias value
𝐾𝑐 = controller gain
𝑒(𝑡) = error signal
𝜏𝐼 = integral or reset time

There are four type of tuning methods for PI controller towards set point change and
disturbance change which are:
1. Quarter Decay Ratio Response
2. Integral of Absolute Error (IAE)
3. Integral of Time-weighted Absolute Error (ITAE)
4. Cohen Coon

The following shows the graphs of process variable against time of PI controller towards set
point change using tuning methods of Quarter Decay Ratio Response, IAE, ITAE and Cohen
Coon.

(a) (b)
(c) (d)
Figure 5.2.1: Graph of process variable against time of PI controller towards set point
change using tuning methods of (a) Quarter Decay Ratio Response, (b) IAE, (c) ITAE and (d)
Cohen Coon

According to the graph in Figure 5.2.1 (a), the response of Quarter Decay Ratio Response
towards set point change shows that the trend is increasing which is closer to the set point, but
the time is insufficient to observe the new steady state. According to the graph in Figure 5.2.1
(b), the response of IAE towards set point change shows that the trend is decreasing which the
response is moving away from the set point and the time is insufficient to observe new steady
state too. Thus, we can say that both responses of Quarter Decay Ratio Response and IAE have
offset. According to the graph in Figure 5.2.1 (c) and (d), ITAE and Cohen Coon do not have
offset as the new steady state of both ITAE and Cohen Coon is able to reach the set point.
However, the response of ITAE is smoother and reaches the set point at a shorter time
compared to that of Cohen Coon which the response is oscillatory and takes longer time to
reach the set point.

The following shows the graphs of process variable against time of PI controller towards
disturbance change using tuning methods of Quarter Decay Ratio Response, IAE, ITAE and
Cohen Coon.

(a) (b)
(c) (d)
Figure 5.2.2: Graph of process variable against time of PI controller towards disturbance
change using tuning methods of (a) Quarter Decay Ratio Response, (b) IAE, (c) ITAE and (d)
Cohen Coon

Based on the graph in Figure 5.2.2 (a), the response of Quarter Decay Ratio Response towards
disturbance change decreases and approaches the initial steady state but the time is insufficient
to observe the new steady state for Quarter Decay Ratio Response. Based on the graph in
Figure 5.2.2 (b), the response of IAE towards disturbance change increases and moves away
from the initial steady state. This means that the IAE setting is not suitable for PI controller.
Furthermore, the time is insufficient to observe the new steady state for IAE. Despite, we can
still conclude that both Quarter Decay Ratio Response and IAE have offset. Based on the graph
in Figure 5.2.2 (c) and (d), ITAE and Cohen Coon do not have offset as the new steady state
of both ITAE and Cohen Coon is able to return to its initial steady-state. Nevertheless, the
response of ITAE is smoother and its final steady-state reaches the initial steady state at a
shorter time compared to that of Cohen Coon which is more oscillatory, and its final steady
state takes longer time to reach the initial steady state. In conclusion, the response of IAE
towards both set point change and disturbance change is the most unstable among the four
tuning methods as its response moves away from the set point.

5.3 Proportional-Integral-Derivative (PID) Controller


Proportional-Integral-Derivative (PID) controller is a controller which consists of proportional
band, integral action and derivative action to anticipate where the process is heading by looking
at the time rate of change of the error.
The following equation shows the working principle of PID controller:
𝑡
1 𝑑𝑒
𝑝(𝑡) = 𝑝̅ + 𝐾𝑐 [𝑒(𝑡) + ∫ 𝑒(𝑡 ′ )𝑑𝑡 ′ + 𝜏𝐷
𝜏𝐼 𝑑𝑡
0
where 𝑝(𝑡) = controller output
𝑝̅ = bias value
𝐾𝑐 = controller gain
𝑒(𝑡) = error signal
𝜏𝐼 = integral or reset time
𝜏𝐷 = derivative time
There are five type of tuning methods for PID controller towards set point change and
disturbance change which are:
1. Quarter Decay Ratio Response
2. Integral of Absolute Error (IAE)
3. Integral of Time-weighted Absolute Error (ITAE)
4. Dahlin Synthesis
5. Cohen Coon

The following shows the graphs of process variable against time of PID controller towards set
point change using tuning methods of Quarter Decay Ratio Response, IAE, ITAE, Dahlin
Synthesis and Cohen Coon.

(a) (b)

(c) (d)
(e)
Figure 5.3.1: Graph of process variable against time of PID controller towards set point
change using tuning methods of (a) Quarter Decay Ratio Response, (b) IAE, (c) ITAE, (d)
Dahlin Synthesis and (e) Cohen Coon

According to the graph in Figure 5.3.1 (a), (b) and (c), the response of Quarter Decay Ratio
Response, IAE and ITAE has similar trend and their response is less oscillatory compared to
that of Dahlin Synthesis and Cohen Coon. However, IAE takes the shortest time to achieve the
set point followed by ITAE and Quarter Decay Ratio Response. Based on the graph in Figure
5.3.1 (d) and (e), the response of both Dahlin Synthesis and Cohen Coon is more oscillatory.
We can also observe that Dahlin Synthesis has a higher overshoot compared to that of Cohen
Coon. Furthermore, all the tuning methods for PID controller do not have offset as their final
steady states are able to achieve the set point. Hence, we can conclude that PID controller is
able to be stabilized.

The following shows the graphs of process variable against time of PID controller towards
disturbance change using tuning methods of Quarter Decay Ratio Response, IAE, ITAE,
Dahlin Synthesis and Cohen Coon.

(a) (b)
(c) (d)

(e)
Figure 5.3.2: Graph of process variable against time of PID controller towards disturbance
change using tuning methods of (a) Quarter Decay Ratio Response, (b) IAE, (c) ITAE, (d)
Dahlin Synthesis and (e) Cohen Coon

According to the graph in Figure 4.3.2, we can observe that the response of all tuning methods
is able to return to their initial steady state even when disturbance is introduced to the system.
Furthermore, we can also see that the beginning of the response is very oscillatory for all tuning
methods. The response of IAE and ITAE have similar trend, while the response of Dahlin
Synthesis and Cohen Coon have similar trend. In conclusion, we can conclude that the final
steady states of PID controller for all tuning methods are able to return to its original steady
states even.

More specifically, the comparison of integral absolute of error, settling time and overshoot of
different type of controllers should be done instead of comparison of the graphs as they are
more accurate compared to observing the graphs.
6.0 DISCUSSION
To choose the best tuning method, the integral of absolute error (IAE), settling time and
overshoot of the controller are observed. A step change of 30 is introduced at a step time of 10
minute. The optimum controller will have less error, fast settling time and less overshoot. To
choose the best tuning method, we will use ranking method where the best is rated with the
highest number and the worst is rated with the lowest number.

According to Smith and Corripio (1985), error is “a function of time for the duration of the
response”. The integral of absolute error (IAE) tends to produce slower response than integral
of squared error (ISE) but usually with less sustained oscillation. It does not put weight to any
of the errors in a system response. Therefore, the smaller the IAE, the better the tuning method
for the controller.

According to Prabakar and Li (2015), overshoot is “the amount by which the response signal
that exceeds the final value”. It can be expressed in fraction or in percentage. However, we
must assume that the first peak occurs around half a cycle from the application of the step
change. Hence, the lesser the overshoot, the better the tuning method for the controller.

According to Zhong (2006), settling time is “the time it takes for the response to achieve its
final steady state”. The typical band limits are ±5 %, ± 3% and ± 1% of the total change.
Thus, the faster the settling time, the better the tuning method for the controller.

6.1 Comparison of Proportional (P) only Controller


Proportional (P) only controller is mainly used to stabilize the process where it takes immediate
corrective action as soon as an error is detected.

Proportional only (P) Controller – Set Point Change

The following table shows the comparison of data of P controller for Quarter Decay Ratio Response
and Cohen Coon towards set point change.

Table 6.1.1: Comparison of error, settling time, overshoot and decay ratio of P controller for
set point change using different tuning formulas

Quarter Decay Ratio Cohen Coon


Response
Integral of Absolute 3037.3 2436.0
Error
Settling Time (min) 20.11 29.20
Overshoot 0.286 0.604
Decay Ratio - 0.307

According to Table 6.1.1, the best tuning method of Proportional (P) only controller towards
set point change is Quarter Decay Ratio Response because it has a faster settling time, a lesser
overshoot and it does not have decay ratio as the response only shows one peak compared to
that of Cohen Coon. However, Quarter Decay Ratio Response has a higher error than that of
Cohen Coon as it has a higher offset where the distance between the new steady state and the
set point is longer compared to that of Cohen Coon.

Proportional only (P) Controller – Disturbance Change

The following table shows the comparison of data of P controller for Quarter Decay Ratio Response
and Cohen Coon towards disturbance change.

Table 6.1.2: Comparison of settling time, overshoot and decay ratio of P controller for
disturbance change using different tuning formulas

Quarter Decay Ratio Cohen Coon


Response
Settling Time (min) 19.35 26.41
Overshoot 0.304 0.650
Decay Ratio 0.071 0.288

According to Table 6.1.2, the best tuning method of Proportional (P) only controller towards
disturbance change is Quarter Decay Ratio Response because it has a shorter settling time,
lesser overshoot and lesser decay ratio compared to that of Cohen Coon.

In conclusion, Quarter Decay Ratio Response is the best tuning method of Proportional (P)
only controller for both set point change and disturbance change.
6.2 Comparison of Proportional-Integral (PI) Controller
Proportional-Integral (PI) controller is mainly used to eliminate the offset or steady-state error
of the process.

Proportional-Integral (PI) Controller – Set Point Change

The following table shows the comparison of data of PI controller for Quarter Decay Ratio Response,
IAE, ITAE and Cohen Coon towards set point change.

Table 6.2.1: Comparison of error, settling time, overshoot and decay ratio of PI controller
for set point change using different tuning formulas

Quarter IAE ITAE Cohen Coon


Decay Ratio
Response
Integral of Absolute 1544.3 5166.5 482.0 530.5
Error
Settling Time (min) - - 39.43 50.67
Overshoot - - - -
Decay Ratio - - - -

According to Table 6.2.1, ITAE has the smallest error among the four tuning methods and
followed by Cohen Coon, Quarter Decay Ratio Response and lastly IAE. Furthermore, ITAE
also has the fastest settling time compared to that of Cohen Coon. The settling time of Quarter
Decay Ratio Response and IAE could not be determined as the observation time is insufficient.
In addition, the response of IAE diverges from the set point which results in undetermined
settling time. Other than that, the all four tuning methods do not have overshoot and decay
ratio. Therefore, we can conclude that ITAE is the best tuning method for PI controller towards
set point change as it has the smallest error, the fastest settling time and it does not have
overshoot and decay ratio.

Proportional-Integral (PI) Controller – Disturbance Change

The following table shows the comparison of data of PI controller for Quarter Decay Ratio Response,
IAE, ITAE and Cohen Coon towards disturbance change.

Table 6.2.2: Comparison of settling time, overshoot and decay ratio of PI controller for
disturbance change using different tuning formulas

Quarter IAE ITAE Cohen Coon


Decay Ratio
Response
Settling Time (min) - - 40.00 52.50
Overshoot - - - -
Decay Ratio - - - -
Similar to that of set point change, the settling time of Quarter Decay Ratio Response and IAE
could not be determined because the observation time is insufficient and the response of IAE
also diverges from the set point which results in unable to determine settling time. However,
ITAE has a faster settling time compared to that of Cohen Coon. Furthermore, all four tuning
methods for disturbance change do not have overshoot and decay ratio which is similar to set
point change. Hence, we can conclude that ITAE is the best tuning method for PI controller
towards disturbance change as it has the fastest settling time, does not have overshoot and
decay ratio. In conclusion, ITAE is the best tuning method of Proportional-Integral (PI)
controller for both set point change and disturbance change.

6.3 Comparison of Proportional-Integral-Derivative (PID) Controller


Proportional-Integral-Derivative (PID) controller is mainly used to stabilize the controlled
process by the derivative action, reduce overshoot and decrease oscillation around set point.

Proportional-Integral-Derivative (PID) Controller – Set Point Change

The following table shows the comparison of data of PID controller for Quarter Decay Ratio Response,
IAE, ITAE, Dahlin Synthesis and Cohen Coon towards set point change.

Table 6.3.1: Comparison of error, settling time, overshoot and decay ratio of PID controller
for set point change using different tuning formulas

Quarter IAE ITAE Dahlin Cohen


Decay Synthesis Coon
Ratio
Response
Integral of 817.1 383.1 427.3 394.5 378.7
Absolute Error
Settling Time 88.30 33.26 39.43 44.60 54.61
(min)
Overshoot - - - 0.367 0.167
Decay Ratio - - - 0.227 0.648

According to Table 6.3.1, the integral of absolute error of Cohen Coon is the smallest, followed
by IAE, Dahlin Synthesis, ITAE and lastly Quarter Decay Ratio Response. Other than that, the
settling time of IAE is the fastest, followed by ITAE, Dahlin Synthesis, Cohen Coon and lastly
Quarter Decay Ratio Response. Quarter Decay Ratio Response, IAE and ITAE do not have
overshoot and decay ratio. However, Dahlin Synthesis and Cohen Coon have overshoot, but
Dahlin Synthesis has higher overshoot than that of Cohen Coon. Dahlin Synthesis and Cohen
Coon also have decay ratio, but Cohen Coon has higher decay ratio than that of Dahlin
Synthesis. Thus, we can conclude that IAE is the best tuning method of PID controller for set
point change as it has moderate error, the lowest settling time and it does not have overshoot
and decay ratio.
Proportional-Integral-Derivative (PID) Controller – Disturbance Change

The following table shows the comparison of data of PID controller for Quarter Decay Ratio Response,
IAE, ITAE, Dahlin Synthesis and Cohen Coon towards disturbance change.

Table 6.3.2: Comparison of settling time, overshoot and decay ratio of PID controller for
disturbance change using different tuning formulas

Quarter IAE ITAE Dahlin Cohen


Decay Synthesis Coon
Ratio
Response
Settling Time 84.62 29.67 42.72 59.52 55.78
(min)
Overshoot - - - - -
Decay Ratio 0.537 0.337 0.374 0.159 0.310

According to Table 6.3.2, the settling time of IAE is the fastest, followed by ITAE, Cohen
Coon, Dahlin Synthesis and lastly Quarter Decay Ratio Response. In terms of decay ratio,
Cohen Coo has the lowest value, followed by Cohen Coon, IAE, ITAE and Quarter Decay
Ratio Response. Furthermore, all tuning methods do not have overshoot towards disturbance
change for PID controller. Therefore, we can conclude that IAE is the best tuning method of
PID controller for disturbance change because it has the fastest settling time, moderate decay
ratio and it does not have overshoot.

6.4 Comparison of P, PI and PID Controller for Set Point Change


The following table shows the comparison of data of the best tuning methods of P, PI and PID
controllers for set point change.

Table 6.4.1: Comparison of error, settling time, overshoot and decay ratio of P, PI and PID
controller for set point change

Type of Controller P Controller PI Controller PID Controller


Tuning Method Quarter Decay ITAE IAE
Ratio Response
Integral of 3037.3 482.0 383.1
Absolute Error
Settling Time (min) 20.11 39.43 33.26
Overshoot 0.286 - -
Decay Ratio - - -
According to Table 6.4.1, we can see that the integral of absolute error of PID controller using
IAE tuning method is the smallest, followed by PI controller using ITAE and lastly P controller
using Quarter Decay Ratio Response. Furthermore, we can also observe that the settling time
of P controller using Quarter Decay Ratio Response is the fastest, followed by PID controller
using IAE and lastly PI controller using ITAE. Other than that, we can see that PI controller
using ITAE and PID controller using IAE do not have overshoot while P controller using
Quarter Decay Ratio Response has the overshoot of 0.286. In addition, all P, PI and PID
controllers do not have decay ratio. In conclusion, we can conclude that PID controller using
IAE tuning method is the best for set point change as the error is the smallest, the settling time
is moderate, and overshoot and decay ratio do not exist.

6.5 Comparison of P, PI and PID Controllers for Disturbance Change


The following table shows the comparison of data of the best tuning methods of P, PI and
PID controllers for disturbance change.

Table 6.5.1: Comparison of settling time, overshoot and decay ratio of P, PI and PID
controller for disturbance change

Type of Controller P Controller PI Controller PID Controller


Tuning Method Quarter Decay ITAE IAE
Ratio Response
Settling Time (min) 19.35 40.00 29.67
Overshoot 0.304 - -
Decay Ratio 0.071 - 0.337

According to Table 6.5.1, the settling time of P controller using Quarter Decay Ratio Response
is the fastest followed by PID controller using IAE and lastly PI controller using ITAE.
Furthermore, PI controller using ITAE and PID controller using IAE do not have overshoot
while P controller using Quarter Decay Ratio Response has an overshoot of 0.304. Other than
that, decay ratio of PID controller using IAE is higher than that of P controller using Quarter
Decay Ratio Response while PI controller using ITAE does not have decay ratio. In conclusion,
we can summarize that PI Controller using ITAE tuning method is the best for disturbance
change as it does not have overshoot and decay ratio, yet it has the slowest settling time.
7.0 CONCLUSION
In conclusion, the experiment is conducted by testing different type of tuning formula with
different type of controllers. The tuning methods are quarter decay ratio response, integral of
absolute error (IAE), integral of time-weighted absolute error (ITAE), Dahlin synthesis and
Cohen Coon. Each of the methods has their own efficiency in removing the error, settling time
and preventing overshooting. In addition, different type of controllers which are P controller,
PI controller and PID controller were used in each of the methods respectively to determine
which controller is the best choice for the respective tuning methods. Analysis was done based
on the graphs plotted using the spreadsheets.
The best tuning for each P, PI and PID controller from five methods is determined where
Quarter Decay Ratio Response is the best tuning method for P. Also, Minimum Error Integral
Tuning ITAE Formulas works best for PI controller and Integral of Absolute Error (IAE) is the
optimum choice of tuning for PIC controller.
Besides that, the optimum choice of controller for each tuning method is also found out from
the analysis of the response graphs. For setpoint change, PID controller using IAE tuning
method is the best for set point change as the error is the smallest, the settling time is moderate,
and overshoot and decay ratio do not exist. For disturbance change, the optimum choice is PI
Controller using ITAE tuning method as it has the slowest settling time, does not have
overshoot and decay ratio.
8.0 REFERENCES
Bhilai, S. (2014). Comparative study of P, PI and PID controller for speed control of VSI-fed
induction motor.
Zhuang, M., & Atherton, D. P. (1993, May). Automatic tuning of optimum PID controllers.
In IEE Proceedings D-Control Theory and Applications (Vol. 140, No. 3, pp. 216-224). IET.
Ho, W. K., Hang, C. C., & Cao, L. S. (1995). Tuning of PID controllers based on gain and
phase margin specifications. Automatica, 31(3), 497-502.
Prabakar, K., & Li, F. (2015). Proportional integral controller gain tuning using real time digital
simulation models and multi-objective optimization based co-simulation. IFAC-
PapersOnLine, 48(30), 473-478.

Smith, C. A., & Corripio, A. B. (1985). Principles and practice of automatic process control
(Vol. 2). New York: Wiley.

Zhong, J. (2006). PID controller tuning: A short tutorial. Mechanical Engineering, Purdue
University, 1-10.
9.0 APPENDIX
9.1 Tabulation of Data
Table 9.1.1: Tuning Values for Quarter Decay Ratio Response

Controller Type Proportional Integral Time Derivative Time


Gain I D
Kc
Proportional only, P 0.00444 - -
Proportional integral, PI 0.00400 16.650 -
Proportional integral 0.00533 10.000 2.500
derivative, PID

Table 9.1.2: Tuning Values for Minimum Error Integral Tuning IAE

Controller Type Proportional Integral Time Derivative Time


Gain I D
Kc
Proportional integral, PI 0.00398 -26.471 -
Proportional integral 0.00565 4.891 1.569
derivative, PID

Table 9.1.3: Tuning Values for Minimum Error Integral Tuning ITAE

Controller Type Proportional Integral Time Derivative Time


Gain I D
Kc
Proportional integral, PI 0.00288 3.125 -
Proportional integral 0.00511 4.902 1.414
derivative, PID

Table 9.1.4: Tuning Values for Dahlin Synthesis

Controller Type Proportional Integral Time Derivative Time


Gain I D
Kc
Proportional integral 0.00342 1.500 2.500
derivative, PID
Table 9.1.5: Tuning Values for Cohen Coon

Controller Type Proportional Integral Time Derivative Time


Gain I D
Kc
Proportional only, P 0.00938 - -
Proportional integral, PI 0.00523 6.250 -
Proportional integral 0.00963 6.555 1.132
derivative, PID

Table 9.1.6: Value of Controller Parameters for Quarter Decay Ratio Response

Controller Type Proportional Integral Gain Derivative Gain


Gain KI KD
Kc
Proportional only, P 0.00444 - -
Proportional integral, PI 0.00400 0.000240 -
Proportional integral 0.00533 0.000533 0.0133
derivative, PID

Table 9.1.7: Value of Controller Parameters for Minimum Error Integral Tuning IAE

Controller Type Proportional Integral Gain Derivative Gain


Gain KI KD
Kc
Proportional integral, PI 0.00398 -0.000150 -
Proportional integral 0.00565 0.00116 0.00886
derivative, PID

Table 9.1.8: Value of Controller Parameters for Minimum Error Integral Tuning ITAE

Controller Type Proportional Integral Gain Derivative Gain


Gain KI KD
Kc
Proportional integral, PI 0.00288 0.000922 -
Proportional integral 0.00511 0.00104 0.00723
derivative, PID
Table 9.1.9: Value of Controller Parameters for Dahlin Synthesis

Controller Type Proportional Integral Gain Derivative Gain


Gain KI KD
Kc
Proportional integral 0.00342 0.00228 0.00850
derivative, PID

Table 9.1.10: Value of Controller Parameters for Cohen Coon

Controller Type Proportional Integral Gain Derivative Gain


Gain KI KD
Kc
Proportional only, P 0.00938 - -
Proportional integral, PI 0.00523 0.000837 -
Proportional integral 0.00963 0.00147 0.0109
derivative, PID

S-ar putea să vă placă și