1 2 3 4 5 7
6 UNIVERSITI TEKNOLOGI MARA
8
FAKULTI KE KEJURUTERAAN KI KIMIA
9
CHEMICAL PROCESS CONTROL 10
11 12 1 13 14 2 15 & 16 17 5 18 6 19 7 20 8 21 9 22 23 24 25 26 27 28 29 30 31 &2 33 34 35 ----&6 &7 &8 39 40 41 42 43
NAM NAME STU,ENT I, ,ATE SU/MIT SEMESTER PROGRAMME CO,E GROUP ASSIGNMENT SU/MIT TO
(CPE562)
% SHEH MUHAMMA, AFNAN /IN SEH HANAFI % 201&210&82 % 75122015 %5 % EH221 % EH2215A % CONTROL LOOP SIMULATION % SIR MOH, AI-A, AHMA,
Re!"#
%$C'e#e *+%
Re'e#e *+%
-------------------------------
-----------------------------
(SIR MOH, AI-A, AHMA, ) )
(
,!.e%
,!.e%
44
CHAPTER 1 : INTRODUCTION
45
History of PID controller
46
PID also known as proportional–integral–derivative controller is a control
feedback mechanism. In early years, PID controller is used as automatic ship steering.It was implemented as a mechanical device such a lever , spring and a mass and were often energied by compressed air. !he first PID controller was developed by "lmer #perry in $%$$ and theoretical analysis first introduced by &ussian 'merican engineer (icolas )inorsky, *)inorsky $%++. !he goal is stability, not general control, which simplified the problem significantly. Proportional control provides stability against small disturbances while derivative term was added to improve stability and control. In modern years, PID controllers in industry are implemented in programmable logic controllers *P-s and applied in industrial ovens, plastics in/ection machinery, hot stamping machines . It used the the implementation of the PID algorithm. 40 48
PID controller theory an e!"ation
4%
1
12
+
τ i ( s) s
+ τ ( ) d s
G c ( PID ) = K c ¿
1$
3here
K c
is the PID control gain,
τ i ( s)
is the integral gain,
τ d ( s )
is the derivative
gain 1+ 5#
Pro$ortional Action
14
Proportional *P control has a function in determining the magnitude of the
difference between the set point and the process variable which is indicated as error. !hen this proportional control will applies appropriate proportional changes to the control variable to eliminate error. )any control systems will, in fact, work uite well with only Proportional control due to it fast response time and its ability to minimie fluctuation. 5owever, it contains large offset. It is an instantaneous response to the control error for improving the response of a stable system. ontrastly, it cannot control an unstable system by itself. !herefore when the freuencies leaving the system , the gain is the same with a nonero steadystate error. 55
Inte%ral Action
16
Integral *I control usually e7amines the offset of set point and the process
variable over time and corrects it when and if necessary. !his integral control has small offset and always return to steady state but it leads to slow response time. Integral action drives the steadystate error towards 2 but slows the response since the error must accumulate before a significant response is output from the controler. #ince an integrator introduces a system pole at the origin, an integrator can be detrimental to loop stability. 8nly controllers with integrators can windup where, through actuator saturation, the loop is unable to comply with the control command and the error builds until the situation is corrected. 10 58
Deri&ati&e Action
1%
Derivative *D control, monitored the rate of change of the process variable and
conseuently makes
changes
to
the
output
variable
to
provide
unusual
changes.
3hen there is a 9process upset9, meaning, when the process variable or the set point uickly changes the PID controller has to uickly change the output to get the process variable back eual to the set point. 8nce the PID controller has the process variable eual to the set point, a good PID controller will not vary the output. !hus, there are two responses occur such as fast
response *fast change in output when there is a 9process upset9, but slow response *steady output.
62 '1
Controller %ain
6+
!he proportional gain *: c determines the ratio of output response to the error
signal. ;or instance, if the error term has a magnitude of $2, a proportional gain of 1 would produce a proportional response of 12. In general, increasing the proportional gain will increase the speed of the control system response. 5owever, if the proportional gain is too large, the process variable will begin to oscillate. If : c is increased further, the oscillations will become larger and the system will become unstable and may even oscillate out of control. '#
Deati(e
64
Deadtime is a delay between when a process variable changes, and when that
change can be observed. ;or instance, if a temperature sensor is placed far away from a cold water fluid inlet valve, it will not measure a change in temperature immediately if the valve is opened or closed. Deadtime can also be caused by a system or output actuator that is slow to respond to the control command, for instance, a valve that is slow to open or close. ' common source of deadtime in chemical plants is the delay caused by the flow of fluid through pipes. '5
Effect of increasin% an ecreasin% &al"e of P)I *D to+ar $rocess res$onse
66
3hen parameters of an e7isting controller have to be tuned, there will be a
problem in the identification of PID controller. ontroller structure has to be determined since manufacturers do not provide data on controller structure whether serial or parallel. )anual tuning of controller parameters had to be done if they are changed with time. 8ther than that, manual tuning of controller parameters also had to be done when change in process parameters occurred. )anual parameter tuning can be done using trial and error and if rules shown in the table below< 67 71
P!"!e.e" Increasing K
68
S3ee 4
Re$34$e 72 Increases
69 73
S.!*.+
70
A"!+
Deteriorate
74
Improves
75
79
Increasing Ki Increasing K d
76 80
Decreases increases
77 81
Deteriorate Improves
78 82
Improves No effect
=> =4 =1
,ettlin% ti(e < !he time at which the P? reaches @ 1A of the total change in the
=6 =0
O&ershoot
process variable *BP?. < )ost notably associated with Ponly controllers, is the difference fromthe
#P to == =%
Decay ratio
where the P? settles out at a steady state value. <
!he sie of the second peak above the new steady state divided by thesie of the %2
first peak above the same steady state level
-1
%+
8b/ective of this study is to determine the effect of PIDCs parameters
to the process controllability. !o study the effect of controller gain, effect of integral time, effect of derivative time and effect of deadtime on the control loop process. %> %4 %1 %6 %0 %= %% $22
$2$ $2+ $2> $24 1.5
CHAPTER / : 0ETHODOO23
106
LA/ 1% Ee. 4 C4."4e" G! .4 P"4e$$ C4."4!*.+
107
P"4e"e
$
8pen matlab software then new model is opened by selecting file button.
+
!hen, untitled window will appear.
> lick button simulink library browser, then drag clock, to workspace, constant, PID controller, $2= order.
transfer fcn , sum, scope and display. 'rrange and connected all simulink in the right
1 s s
+
+ $2 s
4
Process transfer function is set as
1
PID controllerEs parameter was setup as P$2.21, I$2.2$, D$2
6
#et simulation parameters to 622
0
&un the simulation
=
Plot P? vs time
$2%
, process set point$
FFplot*time,P?
%
&un a second set of PIDEs value P+2.$, I+2.2$, D+2
$2
Plot the second process response
$$2
FFfigure*+,plot*time,P?
$$
&un a third set of PIDEs value P>2.+, I>2.2$, D>2
$+
Plot the third process response
$$$
FFfigure*>,plot*time,P?
$>
?iew all the figure in figure palette.
$4
ombine response of figure*+ and figure*> into figure*$
$1
&ename the 7a7is as time and ya7is as P? and every figure as PID$, PID+, and PID>.
$6
#how the #P at $.
112 113
114
115
Figure 1 : PFD FOR EFFECT OF CONTROLLER GAIN
116
117
LA/ 2% Ee. 4 I.e:"! G! .4 P"4e$$ C4."4!*.+
118
P"4e"e
$
8pen mat lab software then new model is opened by selecting file button.
+
!hen, untitled window will appear.
$$% >. lick button simulink library browser, then drag clock, to workspace, constant, PID controller, transfer fcn, sum, scope and display. 'rrange and connected all simulink in the right order. 1 s s
$+2
4. Process transfer function is set as
+
+ $2 s
, process set point$
$+$
1. PID controllerEs parameter was setup as P$2.21,I$2.2$,D$2
$++
6. #et simulation parameters to 622
$+>
0. &un the simulation
$+4
=. Plot P? vs time
$+1
FFplot*time,P?
$+6
%. &un a second set of PIDEs value P+2.21 I+2.2+ D+2
$+0
$2. Plot the second process response
$+=
FFfigure*+,plot*time,P?
$+%
$$. &un a third set of PIDEs value P>2.21 I> 2.24, D>2
$>2
$+. Plot the third process response
$>$
FFfigure*>,plot*time,P?
$>+
$>. ?iew the figure in figure palette
$>>
$4. ombine response of figure *+ and figure *> into figure *$
$>4
$1. &ename the 7a7is as time and ya7is as P? and every figure as PID$, PID+, PID>.
$>1
$6. #how the #P at $.
$>6
137
1#8
Figure 2 : PFD for integral gain
$>% $42
11 12 1& 1 15 16
LA/ &% Ee. 4 ,e";.;e .e .4 P"4e$$ C4."4!*.+
17
P"4e"e
$4=
$. 8pen )at lab software then new model is opened by selecting file button. $4%
!hen, untitled window will appear.
$12 >. lick button simulink library browser, then drag clock, to workspace, constant, PID controller, transfer fcn , sum, scope and display. 'rrange and connected all simulink in the right order. 1 s s
+
+ $2 s
$1$
4. Process transfer function is set as
, process set point$
$1+
1. PID controllerEs parameter was setup as P$2.21,I$2.2$,D$2
$1>
6. #et simulation parameters to 622
$14
0. &un the simulation
$11
=. Plot P? vs time
$16
FFplot*time,P?
$10
%. &un a second set of PIDEs value P+2.21 I+,2.2$ D++
$1=
$2. Plot the second process response
$1%
FFfigure*+,plot*time,P?
$62
$$. &un a third set of PIDEs value P>2.21 I>2.2$, D>4
$6$
$+. Plot the third process response
$6+
FFfigure*>,plot*time,P?
$6>
$>. ?iew the figure in figure palette.
$64
$4. ombine response of figure *+ and figure *> into figure*$
$61
$1. &ename the 7a7is as time and ya7is as P? and every figure as PID$, PID+, PID>.
$66
$6. #how the #P at $.
$60 $6=
169 170
171
172
Figure 3 : PFD for eri!iti!e ti"e
17& 17
LA/ % Ee. 4 e!.e .4 P"4e$$ C4."4!*.+
175
P"4e"e
$06
$. 8pen mat lab software then new model is opened by selecting file button.
$00
+. !hen, untitled window will appear.
$0= >. lick button simulink library browser, then drag clock, to workspace, constant, PID controller, transfer fcn , variable time delay , sum, scope and display. 'rrange and connected all simulink in the right order.
1 s s
$0% 4. Process transfer function is set as and set $=2
+
+ $2 s
, process set point$. 'dd Gtransport delayH
!ime Delay to 1.
$=$
1. PID controllerEs parameter was setup as P$2.+, I$2.2$,D$2
$=+
6. #et simulation parameters to 622
$=>
0. &un the simulation
$=4
=. Plot P? versus time
$=1
FFplot*time,P?
$=6
%. &un a second set of !ime delay 0
$=0 $== $=%
$2. Plot the second process response FFfigure*+,plot *time,P?
$%2
$$. &un a third set of !ime delay %
$%$
$+. Plot the third process response
$%+
FFfigure*>,plot *time,P?
$%>
$>. ?iew the figure in figure palette.
$%4
$4. ombine response of figure *+ and figure *> into figure *$
$%1
$1. &ename the 7a7is as time and ya7is as P? and every figure as PID$, PID+, PID>.
$%6
$6. #how the #P at $.
197
198
1##
Figure $ : PFD for eati"e
200 201 202 20&
CHAPTER & % RESULT AN, ,ISCUSSION
20
LA/ 1% Ee. 4 C4."4e" G! .4 P"4e$$ C4."4!*.+
205
Re$.
206
1 . 8 PI D1 1 . 6 PI D2
1 . 4 PI D3 1 . 2
SP 1 V P
0 . 8
0 . 6
0 . 4
0 . 2
0 0
1 0 0
2 0 0
3 0 0
4 0 0
5 0 0
6 0 0
t i me
2%7
Figure 5 : Co"&ination of 3 gra'( )ontroller gain
/.8
DI,,CU,,ION
+2%
In the figure above shows > different graph plotted in order to observe the oscillations of
each graph plotted. !he > different values of Proportional *P are considered which are 2.21, 2.$, and 2.+. ased on the graph, it can be concluded that the high proportional value will lead the system to become unstable and oscillate. !he proportionality is given by controller gain. ;or a given change in time, the amount of output process value *P? will be determined by the controller gain. It is the best controller gain if the peak of the graph reaches the set point. ;rom the graph obtained, figure > has the best controller gain since the peak point of the graph is nearest to the set point *#P$. !hatCs why this condition will contribute to better processes.
210 211
212
LA/ 2% Ee. 4 I.e:"! G! .4 P"4e$$ C4."4!*.+
21&
Re$.
214
1 . 8 PI D3 PI D2 1 . 6 PI D1
1 . 4
1 . 2
SP 1 V P
0 . 8
0 . 6
0 . 4
0 . 2
0 0
1 0 0
2 0 0
3 0 0
4 0 0
5 0 0
T I ME
215
Figure * : Co"&ination of 3 gra'( Integral ti"e
216 /1
DI,,CU,,ION
+$=
;or second e7periment is to find the effect of integral time. !he larger value of integral
time, the more oscillates of the graph obtained. ased on observation of the graph, there are more oscillations for integral time, I2.24. !hus, the integration will take part until the area under the curve becomes ero. If there is decreasing in I, it will result in instability system. ;rom the graph, it can be concluded that increasing too much I will contribute the present value to overshoot the set point value. ;igure 6 has a better process since the peak point reaches nearest to
the set point. #o that, we can conclude that the increasing value of I will lead the graph to more oscillations.
219
LA/ &% Ee. 4 ,e";!.;e Te .4 P"4e$$ C4."4!*.+
220 221
1 . 8
1 . 6
PI D1 PI D2
1 . 4 PI D3 1 . 2
SP
1 V P
0 . 8
0 . 6
0 . 4
0 . 2
0 0
1 0 0
2 0 0
3 0 0
4 0 0
5 0 0
6 0 0
T I ME
///
Figure 7 : Co"&ination of 3 gra'( Deri!ati!e ti"e
++> //4
DI,,CU,,ION:
++1
;rom the the graph obtained, it can be concluded that the larger values of derivative will
decrease the overshoot. esides that, this change will lead to instability since it will slow down
transient response. In fact, derivative control is used to reduce the magnitude of the overshoot produced. Derivatives term is also used in slow processes such as processes with long time constant. ++6 227 228 229 230
2&1
LA/ % Ee. O ,e!.e .4 P"4e$$ C4."4!*.+
2&2
Re$.
233
2 . 5
PI D3 2
PI D2 PI D1 1 . 5
V P
SP 1
0 . 5
0 0
1 0 0
2 0 0
3 0 0
4 0 0
5 0 0
T i me
234 235
Figure + : Co"&ination of 3 Gra'(, for Di-erent eati"e
6 0 0
/#'
DI,,CU,,ION
+>0
ased on the graphs, it can be concluded that the increasing in !ime Delay will produce
more oscillations on the graph. !he calculation is starting at the dead time icon. !he more time delay, the instability of the system also increases. !his is due to the long stopped reaction time. ;or time delay 1, there is not much oscillation occur. 3hen we increase the time delay to 0, there is small oscillation occur. 238 239 240 241 242
/4#
CHAPTER 4 : CONCU,ION AND RECO00ENDATION
+44 +41
!he performance of each of the three types of controllers varies due to the differing
components of controller euation. In Ponly control, the only ad/ustable tuning parameter is : as the proportional term is the only term in the corresponding controller euation. !he advantage of Ponly control is that there is only one tuning parameter to ad/ust and therefore the best tuning values are obtained rather uickly.!the disadvantage to Ponly control is that it permits offset. !o minimie offset, : may be increased, however this results in greater oscillatory behavior. +46
!he advantage to PI control is that it eliminates the offset present in Ponly control
by minimiing the integrated area of error over time. !o assess the effect changing the two tuning parameters has on a PI controller performance, both : and JI were halved and doubled. In this process, using these tuning parameters actually resulted in increased magnitude of oscillations over time and thus an unstable system. "ither lowering JI, or increasing : from the initial value resulted in a greater peak overshoot, larger settling time and larger decay ratio. +40
In PID control all three terms are utilied. !he function of the derivative term is to
determine the rate of change of the error *slope thus influence the controller output. ' rapidly changing error will have a larger derivative and therefore a larger effect on controller output. !he derivative term will therefore work to decrease the oscillatory behavior in the process variable. !o
assess the effect of changing derivative time, a comparison of the tuning parameter JD was made for the PID controller by halving and doubling the initial value. +4=
Increasing the derivative time results in less oscillatory behavior of the process
variable however there is also an increased noise in the controller output. Increasing JD also increase rise time, settling time, and decreases peak overshoot +4% +12 +1$ +1+ +1> /54
RECO00ENDATION /55
In choosing the KbestC performing controller it must be noted that best
performance is sub/ective, meaning that some processes may desire a P? response with no overshoot, others may be able to tolerate overshoot and prefer faster rise times. ;or a process that desires fast rise time with the minimal amount of oscillatory behavior and overshoot it would be suggested to use a moderate to moderately aggressive PI controller. 256 257
/58
RERENCE,
1. 'bdul 'i Ishak L Maliawati 'bdullah. *+2$4. PID !N(I(O Fundamental Concepts +1%
and Application. NI!) Press.
2. 5. ischoff, D.5offmann, ".?.!eri. *$%%0. Process ontrol #ystem, Control of 260
Temperature, Flow and Filling Level. ;esto Didactic Omb5 L o.
>. asso, hristophe *+2$+. 9Designing ontrol -oops for -inear and #witching Power +6$
#upplies< ' !utorial Ouide9. 'rtech 5ouse, I#( %0=$62=201100
+6+ 4. lanke, ).Q :innaert, ).Q -une, R.Q #taroswiecki, ). *+226, iagnosis and +6> +64 +61 266
Fault!Tolerant Control *+nd ed., #pringer
