You are on page 1of 19

Department of Chemical Engineering

CHME 426 Chemical Engineering Laboratory III


Title of the Experiment: Numerical Experiment on Process Control
Submitted by: Group (1) Section: L51
Date of experiment: 11-6-2015
Date of submission: 15-6-2015
Item

Report presentation
Abstract
Introduction and Theory
Experimental Procedure
Results and Discussion
Conclusion and
recommendation
Citation and References
Total

Grades
Spelling,
grammar,
05
sentence structure
Appearance and formatting
05
10
10
05
Raw Data
Sample Calculation
Data Processing (Tables and graph)
Discussions
10
05
100

Scor
e

05
10
10
25

Table of conten

Abstract.........................................................................................................................................4
Objectives....................................................................................................................................5
Introduction [1]..........................................................................................................................5
Procedure and report requirement [1]..............................................................................6
Results and Discussions.........................................................................................................6
Conclusion and Recommendation....................................................................................17
References.................................................................................................................................18

List of Figures
Figure 1: a representative Simulink model [1]................................................................................6
Figure 2: transfer function in Simulink model [1]...........................................................................6
Figure 3: The response of the system with a continuous oscillation...............................................7
Figure 9: Changing the denominator coefficient from 12 to 7......................................................12
Figure 10: the process performance due to changing the kc and I

in theory.........................13

Figure 12:changing I

for kc/2.................................................................................................14

Figure 11: changing I

for kc...................................................................................................14

Figure 13:changing I

for kc/3.................................................................................................15

Figure 14: The PID controller for a disturbed system...................................................................16

List of Tables
Table 1: Controller setting based on the continuous cycling method..............................................7
Table 2: Result Ziegler-Nichols tuning relation...............................................................................8
Table 3: Result Tyreus-Iuben tuning relation...................................................................................8

Abstract
In this report, a numerical experiment on process control has been studied.
First of all, different types of controllers such as: Proportional, Proportional
Integral and Proportional Integral Derivative controllers were generated by
using two most widely methods which are ZN method and TL. It was found
that the method which has a better performance was selected to proceed on
the process. Moreover, the effect of changing the process parameters such
as dead time and gain of the response were studied and it can be
summarized that as the gain is increasing the oscillation becomes more
significant. In addition, the dead time affect the response and as it is
increasing the response becomes more sluggish. Finally, the effect of having
disturbances in the process was observed to affect the speed of the
response.

Objectives
The objective of this experiment is to be familiar with MATLAB Simulink
model used in process control, as well as designing the PID control
parameter using different tuning method such as Z-N and T-L methods. It
also aims to analyze the effect of each PID parameters such as Kc, Ti and Td
on control and discuss with respect to set point and disturbance.

Introduction [1]
SIMULINK is a graphical programming block diagram

for multi-domain

simulation and Model-Based Design. It supports simulation, automatic code


generation and it is part of MATLAB. SIMULINK has a graphical editor,
customizable block libraries, and solvers that used for modeling and
simulating dynamic systems.
SIMULINK adds a new class of windows which is block diagram windows.
These windows contain the models, and these models can be edited. Any
physical process control system can be simulated though MATLAB SIMULINK.

Figure 1: a representative Simulink model [1]

A transfer function (Gc) corresponding to a physical model is entered as


shown here.Gc model constants is the characteristic of the process.

Figure 2: transfer function in Simulink model [1]

Procedure and report requirement [1]


In this experiment SIMULINK in MATLAB is used (figure 1).First starting with P controller and
varying value of Kc and plotting it to get continuous oscillation then determining value of
ultimate gain and ultimate period. After that, runnig the Simulink model for P, PI and PID at the
values of previous step by Z-N and T-L methods. Next step is to change the denominator

coefficient from 10 to 5 in Gs (In1out1). Then, running the model using the best PID parameter
and plotting the response. Again, running the code for the coefficient value of 7 and plotting
responses for values of coefficient 10, 7 and 5. Step after is disturbance block, which value 1
must be entered in the block, otherwise there is no disturbance.

Results and Discussions


The value of kc changed until continuous oscillation was obtained as shown below and the actual
gain was found to be3.26 and with an ultimate period of 16.

Figure 3: The response of the system with a continuous oscillation

For determining the PID parameters for the Z-N method and the T-L equations from the process
control books were provided in following table:

Table 1: Controller setting based on the continuous cycling method

For Ziegler-Nichols method

Table 2: Result Ziegler-Nichols tuning relation

P
PI
PID

Kc

1.63
1.47
1.96

0
13.33
8

0
0
2

Tyreus-Luyben method

Table 3: Result Tyreus-Iuben tuning relation

P
PI
PID

Kc

0
1.011
1.47

0
35.2
18.4

0
2.54

ZieglerNicholsmethod

Figure 4: (P, PI and PID) for Z-N metho

TyreusLuybenmethod

Figure 5: (PI and PID) for T-L method

Figure (4) &(5) shows that the difference ofusing P,PI,and PID controllers forboth
methods.Ingeneral,proportionalcontrollercanstabilizeonlyfirstorderunstableprocess,which
canbeappliedforthiscase.Thedynamicofthiscontrollerisrelatedtothecontrollergain,
whichindicatessteadystateerror,thespeedofthedynamics,andthemagnitudeoftheamplitude.
AsitisclearfromthefiguresthesteadystateofthePcontrollerreachesatagainvalueof0.5,
whichissmallerthanPIandPIDcases.Moreover,proportionalintegralcontrollerismostwidely
adaptedinindustrialapplicationssinceithassimpleinstructoranditiseasytobedesigned.The
useofPIcontrollerwilleliminatetheoffsetofPcontroller.Howeveritwillaffectthespeedof
theresponseandoverallstabilityofthesystem,whichleadstolowerprocessstability.Incaseof
PIDcontroller,asitshownithasahigheroscillation.Thederivativecontrollerincreasesthe
stabilityofthesystemandtherefore,itcanbeconcludethatthebestcontrolleristhecombined
PID.
In figure (5) T-L method is shown and it can be notices that the PI and PID parameters
performance have lower oscillation but the response is very slow in order to reach steady since
the integral time value is quit large.
1. Comparisonofthetwomethods

Figure 6: (Z-N) compare to (T-L)

In comparison of the two methods, Tyreus-Luyben method has the same procedure as Z-N
method but the final controllers are different. Also it has only P and PID compared to the other
one. Tyreus-Luyben is more conservative than the Z-N method and it gives better performance
with smaller value of integral time. However, sluggish performance will take place when the
integral time is large.

Figure 7: The best PID control between two methods

Tohavethebesttunethekc needtobeincreaseandthe I needtobedecreasebecauseit


giveslessoscillationbutslowerresponse,andthatishowthebestPIDtunewascreatedasit
shows in figure (7) it is plot between the two methods to show faster response with less
oscillation.

Figure 4: Changing the denominator coefficient from 12 to 7

Figure(9) shows the result obtained by changing the dominator coefficient of Gs from 12-7. It is
noticed that the process becomes unstable and has high oscillation. Since figure(9) shows the
effect of having Gs value of 7, it can be predicted that changing the Gs to a value of 5 or less will
have a negative effect on the process stability and oscillation.

Effect of process parameters

Theeffectofchangingintegraltimeparameterhasbeenstudiedthroughsimulatingthe
processatdifferentintegraltime. Ithasnoticedfromfigure(11)thatastheintegraltimeis
increasing,theoscillationreducesandthishasapositiveeffectontheprocessstability.The
I I
integraltimevaluewasselectedtobe 2 , 3 I /4 andtheprocessresponseofthesedifferent
valuesisplottedinfigure(11)whichalsoimplementsthatthegreaterthevaluesofintegraltime,
theslowerresponse.Moreover,theeffectofchangingkcvaluewasplottedforallintegraltime
valuesanditwasnoticedthatdecreasingthevalueofthekcwoulddecreasetheoscillationwhile
increasingitwillleadtomoreoffset.
In conclusion, the experimental trends obtained are similar to the theoretical trend that gotten
from process control book and it can be shown in figure (10).

Figure 5: the process performance due to changing the kc and

in theory

Figure 6: changing
Figure 7:changing

I
for kc/2

for kc

Figure 8:changing

for kc/3

Effect of disturbance
Inordertostudyandpredicttheeffectofthedisturbanceontheprocessresponse,the
systemwassettobeindisturbedmood.Figure(12)showstheobtainedresultanditisclearthat
bothPIandPIDcontrollershasthealmostthesameresponse.Ontheotherhand,itwasnoticed
thatPcontrollerhasafasterresponsecomparedtoothersanditreachsteadystatefaster.

Figure 9: The
for a
system

PID controller
disturbed

Conclusion
To conclude, the objective of the experiment was to be familiar with MATLAB
Simulink model used in process control, as well as designing the PID control
parameter by two different tuning methods Z-N, and T-L. It was found that
the best controller is PID in both methods but in comparison the PI and PID
parameters performance had lower oscillation but the response was very
slow in order to reach steady. Moreover, the effect of each PID parameters
such as Kc, Ti and Td on control was studied with respect to a set point. The
effect of changing integral time parameter studied through simulating the
process at different integral time. It was noticed that as the integral time is
increasing, the oscillation reduces and this has a positive effect on the
process stability. Finally, in order to study and analyze the influence of the
disturbance on the process response, the system was set to be in disturbed
mood. It was clear that both PI and PID controllers have almost the same
response. Instead, it was observed that P controller has a faster response
compared to others and it reach steady state faster.

Recommendation
It is recommended to use MATLAb software and generate more plots for
wider values of the process parameters in order to be able to precisely
determine its effect on the process dynamic, stability and response

References
[1] Chemical Engineering Lab IIIManual- Qatar University -Spring2014