Process Dynamics

7/17/2019 Process Dynamics

http://slidepdf.com/reader/full/process-dynamics-568edfb2a1f5a 1/66

Fundamentals of processdynamics and control

Process Dynamics



Training Course on:Sustainable Industrial Development: Process Simulation, Analysis, Optimization and Control Bangkok, 8-12 July, 200

Motivating example: level control

If the outlet ow is simply set equal to the

inlet ow, the tank may overow or run

empty (because of ow measurement

errors)

Flow in

Flow out

The inlet fo comes!rom an u"stream

"rocess, an# may

change ith time

The le$el in the tankmust %e ke"t

constant in s"ite o!

these changes




LTLC

SP

Flow in

Flow out

Introducing a level controller

The le$el controller &'C(looks at the le$el

(monitoring)

)! the le$el starts to

increase, the 'C sen#sa signal to the out"ut

$al$e to $ary the out"ut

fo (change)

This is the essence of feedback control




Feedback control

)t is the most im"ortant an# i#ely use#

control strategy

)t is a closed-loop control strategy

Block diagram

process

transmitter

controller

disturbance

comparator manipulated

variable

controlled

variable

* – error set-point

ysp y




ack to level control

LT LC

SP

Flow in

Flow out

desired value(set-point)

transmitter

controller

controlledvariable

(measurement)

manipulated

variable

disturbance

process




More on control !argon

Input variables : in#e"en#ently stimulate the

system+ they can in#uce change in the internal

con#itions o! the "rocess manipulated &or control( $aria%les u; m → at our #is"osal

disturbance $aria%les d → e cannot #o anything on them

utput variables : measurements y , %y hich one

o%tains in!ormation a%out the internal state o! the

system &eg tem"erature, le$el, $iscosity, re!racti$e in#e(

!tates : minimum set " o! $aria%les essentials !or

com"letely #escri%ing the internal con#ition o! a

"rocess &eg com"osition, hol#u", enthal"y(




"rocess dynamics

.i$en a #ynamic mo#el o! the "rocess, it

in$estigates the "rocess res"onse to $arious

in"ut changes

To elements are necessary:

/ a dynamic model o! the "rocess

/ a knon forcing function

time

u &t (

0

A

!tep input

0

u &t (

0

A

#ulse input

time0 b




"rocess models: #hich$

e ill consi#er only to classes o! #ynamic

"rocess mo#els state-space mo#els

input-output mo#els

!tate$space models can %e #eri$e# #irectly !rom the

general conser$ation euation:

%ccumulation & (Inlet – utlet) '

(eneration – onsumption)

They are ritten in terms o! #ierential euations

relating "rocess states to time ⇒ They occur in the

“time #omain”




"rocess models : #hich$"rocess models : #hich$ (cont (cont’’d)d)

Input$output models com"letely #isregar# the

"rocess states They only gi$e a relationshi" %eteen

"rocess in"uts an# "rocess outputs ⇒ They occur in

the “'a"lace #omain”

)(

);;;(

d

)(d

xy

xx

h

t d u f

t

t

=

=

!tate$space model Input$output model

)()()( sU sG sY =

)( sG)( sU )( sY states

output

G &s( is calle# transfer

function o! the "rocess




%inear systems

)n the time #omain, a linear system is

mo#ele# %y a linear #ierential euation

3or eam"le, a linear, nth-or#er system is:

)(d

d

d

d

d

d011

1

1 t ub yat

ya

t

ya

t

ya

n

n

nn

n

n =++++ −

−−

4ur assum"tions: – the coe5cients o! the

#ierential euationsare constant

– the out"ut y is eualto the state x

*ote The 'a"lace-#omain re"resentation is "ossi%le only !or

linear &or lineari6e#( systems

e ill assume that the "rocess %eha$ior in the

$icinity o! the stea#y state is linear




First-order systems

P is the "rocess steady state gain &it can %e 70 or

τP is the "rocess time constant &it is alays 70(

)(dd t u K yt y

P P =+τ

& Di$i#ing %y a0 (

)(1

)( sU s K sY P

P

+τ=

1)(

+τ=

s

K sG

P

P

+ime$domain model

aplace$domain

model

Trans!er !unction o! a 9rst-or#er system:




&esponse of 'rst-order systems

e only consi#er the res"onse to a step

!orcing !unction o! am"litu#e A

−= τ

− P

t

P e AK t y 1)(

The time-#omain

res"onse is:

)t takes ∼ ;< time

constants !or the

"rocess to reach

the ne stea#y

state

00 A

, tie

0.632 AK P

τP AK P

,




etermining the process gain

=n o"en-loo" test can %e "er!orme# starting

!rom the re!erence stea#y state: ste" the in"ut to the "rocess

recor# the time "ro9le o! the measure# out"ut

until a ne stea#y state is a""roache# check i! this "ro9le resem%les

i! so, calculate P as:

)1()( / P t

P e AK t y τ−−=

state steadyref new

ref ssnew ss

P uu

y y

K

∆∆

=−−

= )input(

)output(,, The gain is a

#imensional9gure

The process gain can be determined

from steady state information only




etermining the time constant

3rom the same o"en-loo" test:

#etermine τP gra"hically &note: it has the #imension o! time(

ou need dynamic

information to

determine the

process time

constant

Determining the$alues o! P an# τP

!rom "rocess #ata is

knon as process

identi!cation0 time

0.632 AK P

τ

AK P

o

u t p u t , y




*n alternative approach

>tate the i#enti9cation task as an o"timi6ation

"ro%lem: "iven a !rst-order model, !nd t#e P and τ P values t#at allo$

t#e model to best-!t t#e experimental data

?ou ill nee# a com"uter "ackage to "er!orm the

9tting &eg Control >tation T@, @atla% T@(

)t is %etter to ste" u" an# #on the mani"ulate#

in"ut se$eral times to ca"ture the “true” #ynamic

%eha$ior o! the "rocess

Ae$er trust on the “ra” 9tting results only =lays

u#ge the results %y su"erim"osing the 9tte# cur$e to

the "rocess one




45

50

55

45

50

55

0 500 1000 1500

3itting a 9rst-or#er mo#el to "lant #ataProcess: hite line @o#el: yello line@o#el: 3irst 4r#er 3ile Aame: 9tD34tt

>>E: F288.ain &G( H 1<1, Time Constant &T1( H 1IJI

P

r o c e s s

V

a r

i a b l e

M

a n i p u l a t e d

V

a r i a b l e

Time

*n alternative approach (cont’d)Eam"le using Control >tation T@

r e s u l t s o f - t t i n g




+xtension to nonlinear systems

>trictly s"eaking, the gain an# time constantare in#e"en#ent o! the o"erating stea#y state

%or linear systems only

)! a true &ie nonlinear( system is %eing

consi#ere#, the ecitation seuence must %esuch that the "rocess is not mo$e# too !ar

aay !rom the nominal stea#y state

statesteadyany

linear ,u

y K P

∆

∆=

statesteadynominal

nonlinear ,u

y K P ∂

∂=

linear nonlinear




Further remarks

“Slo$” and “%ast ” processes The time

nee#e# to

a""roach the

ne stea#ystate increases

ith increasing

τP

*ote

For all "’s, the

output starts to

change

immediately after

the input has beenchanged

0 50 100 150 200

0.0

0.2

0.4

0.6

0.8

1.0

1

510

50

100

P

y / K

P A !

time units




"ure time-delay systems

L

v

/ @any real systems #o not

react immediately to

ecitation &as 9rst or#er

systems instea# #o(

/ The time nee#e# to

“trans"ort” a fui# "ro"ertychange !rom the inlet to

the outlet is:

Plu" &o$

Incompressible&uid

v

L

P =θ: dead time

or time delay

Eam"les: trans"ortation lags &eg #ue to "i"e length, to

recycle, …(+ measurement lags &eg gaschromatogra"hs(




"ure time-delay systems (cont’d)

The "rocess out"ut is sim"ly

shi!te# %y θP units in time

ith res"ect to the in"ut

Models

'ime domain :

θ≥θ−θ<

= P P

P

t t x

t t y

,)(

,0)(

s P e sU

sY θ−=)(

)(

time

y &t (

0

.ecorded output

θP

u &t (

time0

0

%pplied input (aplace domain :




F"T systems

The #ynamic

%eha$ior o!

many realsystems can

%e

a""roimate#

as )irst Order

Plus Dead

'ime &34PDT(

0

θ

time

first-order response




Modeling a F"T system

The %eha$ior o! a "ure time-#elay system is sim"ly

su"erim"ose# to that o! a 9rst-or#er system

1)(

+τ=

θ−

s

e K sG

P

s

P P

)()(d

)(d P P P t u K t y

t

t yθ−=+τ 'ime domain

(aplace domain

=""roimating a real system as a 34PDT linear

system is etremely im"ortant !or controller #esign

an# tuning




.econd-order systems

Time-#omain re"resentation: )(d

d

d

d012

2

2 t bu yat

y

at

y

a =++

)(d

d2

d

d2

22 t Ku y

t

y

t

y=+ζτ+τ

'a"lace-#omain re"resentation:

12)(

)(22 +ζτ+τ= s s

K

sU

sY H "rocess gain

τ H natural "erio#

ζ H #am"ingcoe5cient

)1)(1()(

)(

21 +τ+τ=

s s

K

sU

sY




/nderdamped systems

024681012141610.00.20.40.60.81.01.21.4..0.60.4=.

t /τ

Open-loopresponse to

a input step

disturbance




verdamped systems

Open-loop

response toa input step

disturbance024681012141610.00.20.40.60.81.01.21.4..

1.01.52.0=3.0

t /τ




+0ect of the damping coe1cient

The $alue o! ζ com"letely #etermines the #egree o!

oscillation in a "rocess res"onse a!ter a "ertur%ation

ζ 7 1 : o$er#am"e#, sluggish res"onse

0 ζ 1 : un#er#am"e#, oscillating res"onse &the #am"ing is attenuate# as ζ #ecreases(

ζ 0 : unstable system

&the oscillation am"litu#e gros in#e9nitely(

The importance of 2nd order




The importance of 2nd-ordersystems

Control systems are o!ten #esigne# so that the

controlled &ie, close#-loo"( "rocess res"on#s as an

un#er#am"e# secon#-or#er system

051015202530.00.20.40.60.81.0..

desiredvalue

timeunits




Inverse-response systems

There is an

initial in$ersion

in the res"onse:

the "rocess starts

mo$ing a$ay

!rom its ultimate$alue

The "rocess

out"ut

eventually

hea#s in the

#irection o! the

9nal stea#y state

input variable

time




Inverse-response systems

(cont’d)

)n$erse res"onse is the net result o! toi*i* opposing #ynamic mo#es o! ii*ii* di/erent

magnitudes, o"erating on iii*iii* di/erent di/erent

time scalestime scales

the !aster mo#e has a small magnitu#e an# is

res"onsi%le !or the initial, “rong ay” res"onse

the sloer mo#e has a larger magnitu#e an# is

res"onsi%le !or the long-term, #ominant res"onse




+xample process: drum boiler

In the long run, the le$el is e"ecte# to increase, %ecause e

ha$e increase# the !ee# material ithout changing the heat su""ly

But immediately after the col# ater has %een increase#, a #ro"

in the #rum liui# tem"erature is o%ser$e#, hich causes the

%u%%les to colla"se an# the o%ser$e# le$el to re#uce

Disturbance :

ste" increase in the

col# !ee#ater

forate

Output :le$el in the %oiler

Col# !ee#ater

>team

Kot me#ium



Fundamentals of processdynamics and control

Process Control




Feedback control

)()()( t yt yt e sp −= y

sp

H set "oint &target $alue(

y H measure# $alue

The "rocess in!ormation & y ( is !e# bac+ to the

controller The o%ecti$e is to re#uce the error si"nal to

6ero, here the error is #e9ne# as:

process

transmitter

controller

disturbance

comparator manipulated

variable

controlled

variable

* – error set-point

ysp y




The typical control problems

&egulatory control – the task is to counteract the eect o! eternal

#istur%ances in or#er to maintain the out"ut at

its constant set-"oint &disturbance reection(

.ervo control

– the o%ecti$e is to cause the out"ut to track the

changing set-"oint

In both cases, one or more variables are

manipulated by the control system




Material balance control 3 4

%i5uid holdup control

&le$el control(

LTLC

SP

Flow in

Flow out

/ )! the le$el # ten#s to

increase, the error

&#sp – #( #ecreases

/ The controller sen#s a

signal to the control

$al$e actuator/ The fo out is

increase#

/ The le$el in the tank

#ecreases

Material balance control 3 4




Material balance control 3 4(cont’d)

The controller’s o% is to en!orce the total mass

%alance aroun# the tank, in or#er to ha$e neither

accumulation nor #e"letion o! liui# matter insi#e the

tank

rate of mass out rate of mass inset %y the controller unknon to the controller

The e5uality is enforced by the controller

regardless of the value of the level set$point

The task of a process control




The task of a process controlsystem

@onitoring certain $aria%les that in#icate "rocess

con#itions at any time &measurements(

Making rational decisions regarding 6hat

corrective action is needed (current state vs7

desired state)

)n#ucing changes in the a""ro"riate "rocess

$aria%les to im"ro$e "rocess con#itions &$al$es to

mani"ulate(

once more

%ccording to what rationale does a %ccording to what rationale does a

feedback control system work0feedback control system work0




n-o0 control: the simplest one

The control $aria%le is mani"ulate# accor#ing to:

<≥

=0if ,

0if ,)(

min

max

eu

eut u

The 'nal control element is either

completely open8maximum, or

completely closed8minimum

#ea#

%an#utput

input

4A

433 time

i#ely use# as

thermostat in

#omestic heating

systems,re!rigerators, …+

also in noncritical

in#ustrial a""lic’ns

&some le$el an#

heating loo"s(




.ummary for on-o0 control

*dvantages sim"le L easy to #esign

ine"ensi$e

easily acce"te# among o"erators

"itfalls not eecti$e !or “goo#” set-"oint control &the

controlle# $aria%le cycles(

"ro#uce ear on the 9nal control element &it can%e attenuate# %y a large #ea# %an#, at the e"ense o! a

loss o! "er!ormance(




"roportional (") controllers


)()( 0 t e K ut u C +=u0 is the controller bias

C is the controller gain

The controller gain can %e a#uste# &“tune#”( to

make the mani"ulate# $aria%le changes as

sensiti$e as #esire# to the #e$iations %eteen set-

"oint an# controlle# $aria%le

The sign o! C can %e chosen to make the

controller out"ut u increase or #ecrease as the

error increases




"-only controllers

)()( 0 t e K ut u C +=

const0 == uu : at the nominal stea#y state

The %ias u0 is the $alue o! the controller out"ut

hich, in manual mo#e, causes the measure#

"rocess $aria%le to maintain stea#y state at the#esign le$el o! o"eration

Me &t (H0N hen the "rocess #istur%ances are at

their e"ecte# $alues

The %ias $alue is assigne# at the controller #esignle$el, an# remains !xed once the controller is "ut

in automatic




9 %8h

LTLC

SP

Flow in

Flow out

4 %8h

disturbance

Aominal o"eration:

u must %e I0 'Oh ⇒ if

e ; then u;< %8h< %8h

)()( 0 t e K ut u C +=

9 %8h

LTLC

SP

Flow in

Flow out

2 %8h

disturbance= %8h

)! the #istur%ance

changes to 20 'Oh,

the stea#y state is

maintaine# only i!

uH<0 'Oh ⇒ since

u;< %8h, the error

must be

"-only controllers (cont’d)




"-only controllers (cont’d)

The mani"ulate# in"ut u must change to guarantee

that the "rocess stays at stea#y state, ie

)()( 0 t e K ut u C +=

1hat if the disturbance changes during the process0

0uu ≠

= stea#y state error e ≠ 0 must %e en!orce# %y the P-

only controller to kee" the "rocess at stea#y state:

% #$only controller cannot remove o/$set

00.. )( ut e K uu C s s ≠+=




no control

K C "0!

o##$set

set$point

increasin K C

c o n t r o

l l e d

% a r i a b l e

time

"erformance of "-only controllers

.esponse to a disturbance step c#an"e

/ hate$er the $alue o!

C, the oset is

re#uce# ith res"ect

to o"en-loo"o"eration

/ )ncreasing C :

the oset is re#uce#

the system may

oscillatethe "rocess res"onse

is s"ee#e# u"

/ =lthough the o"en-

loo" res"onse may %e

1st or#er, the close#-




.ummary for "-only control

*dvantages conce"tually sim"le

easy to tune &a single "arameter is nee#e#, C +

the %ias is #etermine# !rom stea#y state

in!ormation(

"itfalls cannot remo$e o-set &o-set is en%orced by the

controlle#(




"I controllers

τ++= ∫

t

I

C t t et e K ut u0

0 d)(1)()(

u0 is the controller bias

C is the controller gain

τI is the integral time

&also calle# reset time(

PHPro"ortional , )H)ntegral

integral action contri%ution

The P controller cannot remo$e o-set %ecause the

only ay to change the controller %ias #uring non-

nominal o"erations is to cause e ≠ 0

The rationale %ehin# a P) controller is to set the

“actual” %ias #ierent !rom u0 , thus letting the

error %e 6ero





"I controllers (cont’d)

τ++= ∫

t

I

C t t et e K ut u0

0 d)(1

)()(

Aote that until e ≠ 0, the mani"ulate# in"ut kee"s on

changing %ecause o! the "resence o! the integralterm

The change in u &t ( ill sto" only hen e H 0

+he integral action can eliminate o/$set




"erformance of "I controllers

.esponse to a disturbance step c#an"e: e/ect o% G C

The o0set is

eliminated

)ncreasing C :

the "rocess res"onseis s"ee#e# u"

the system may

oscillate%2+I*

3or large

values of

the

controller

gain, the

closed$loop

response

may be

!i"ed

increasin K C

open$loop

K C "0!

set point

c o n t r

o l l e d

% a r i a b l e

time





"erformance of "I controllers(cont’d)

.esponse to a disturbance step c#an"e: e/ect o% τ)

)ncreasing τ) :

oscillations are

#am"ene#

the "rocess res"onse

is ma#e more sluggish

%2+I*

3or small

values of

the integral

time, the

closed$loop

response

may be

unstable 4

K C !i"ed

increasin #

set point

c o

n t r o l l e d

% a r i a b l e

time




.ummary for "I control

*dvantages steady state o/-set can be eliminatedstead y state o/-set can be eliminated

the "rocess res"onse can %e consi#era%ly

s"ee#e# u" ith res"ect to o"en-loo"

"itfalls tuning is har#er &to "arameters must %e

s"eci9e#, C an# τI( the "rocess res"onse %ecomes oscillatory+ %a#

tuning may e$en lea# to insta%ility

the integral action may “saturate”




"I controllers

τ+

τ++= ∫ t

t et t et e K ut u D

t

I

C d

)(dd)(

1)()(

0

0

τD is calle# derivative time

i ) )! the error i! increasing $ery ra"i#ly, a large#e$iation !rom the set"oint may arise in a short

time

ii ) >luggish "rocesses ten# to cycle

PHPro"ortional , )H)ntegral , DHDeri$ati$e

#eri$ati$e action contri%ution

The rationale %ehin# #eri$ati$e action is to

antici"ate the !uture %eha$ior o! the error signal %y

consi#ering its rate o% c#an"e






.esponse to a disturbance step c#an"e

)ncreasing τD :the oscillations cause#

%y the integral

action are

#am"ene#the "rocess res"onse

is s"ee#e# u"

%2+I*

*oisy

measurement

s may disrupt

the controller

performance

4

no deri%ati%e action

τ& " 0

increasin$

set-point

c o n t r o l l e d

% a r i a b

l e

time




e6are measurement noise >

The #eri$ati$e action reuires #eri$ation o! the

out"ut measurement y ith res"ect to time:

t

y y

t

e sp

d

)d(

d

d −=

)! the measure#

out"ut is noisy, its

time #eri$ati$emay %e large, an#

this causes the

mani"ulate#

$aria%le to %e

su%ect to a%ru"t

changes ⇒

Attenuate or

suppress t#e

derivative action

%100&

%50&

0

'50&

'100&

controlled variable

manipulated variable

time

time




.ummary for "I control

.0/ <τθ P P

*dvantages oscillations can %e #am"ene# ith res"ect to P) control

"itfalls tuning is har#er than P) &three "arameters must %e

s"eci9e#, C , τ) an# τD( the #eri$ati$e action may am"li!y measurement noise ⇒

"otential ear on the 9nal control element

/se of derivative action

a$oi# using the D action hen the controlle# $aria%le hasa noisy measure or hen the "rocess is not sluggish &

(

?ontroller selection




recommendations

hen stea#y state osets can %e tolerate#, use a

P-only controller &many liui# le$el loo"s are on P control(

hen oset cannot %e tolerate#, use a P) controller&a large "ro"ortion o! !ee#%ack loo"s in a ty"ical "lant are

un#er P) control(

hen it is im"ortant to com"ensate !or some

natural sluggishness in the system, an# the

"rocess signal are relati$ely noise-!ree, use a P)D

controller

" f




0 5 10 15 20 25 300.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

t s

t p

t r

0.'5

1.05

P

c

b

a

n o r m a l i (

e d c o n t r o l l

e d v a r i a b l e

time units

"erformance assessment

t r rise time

t p time to !rst pea+

t s settlin" time

a Ob overs#oot

c Oa decay ratio

P period o% oscillation

* “good”

decay ratio

is 48@

(“5uarter

amplitude”

decay)

&set-"oint tracking "ro%lem(




"erformance indexes

∫

+∞

= 0d)(!"# t t e : inte"ral o% t#e absolute value o% error

/ The controller’s tuning

"arameters & C + τ) + "ossi%ly

τD( are chosen such that )=E

is minimi6e#

/ >emi-em"irical !ormulae can

%e #eri$e# %ase# on a 34PDT

o"en-loo" i#enti9cation

/ '#e optimal controller ’ s

settin"s %or load disturbance

reection are di/erent %rom

t#ose %or set-point trac+in"

I*+ corresponds to the

shaded area

set-point

c o n t r o l l e d v a r

i a b l e

time




Tuning guidelines

3it a 34PDT mo#el to the "rocess #ata o%taine# %y

ste" &or "ulse( changes in the mani"ulate#

$aria%le the "rocess must %egin at the nominal stea#y state

the sam"ling rate shoul# %e at least ten times !aster than

the "rocess time constant the measure# $aria%le shoul# %e !orce# to mo$e at least

ten times !rom the noise %an#

Determine initial $alues !or C , τI &an# "ossi%ly τD (

!rom suggeste# correlations

Ae$er ever trust %lin#ly on these settings =lays

re9ne the tuning on-9el#

T i l ti f "I t l




Tuning correlations for "I control

&%ase# on 34PDT o"en-loo" i#enti9cation(

5 ? I

IM? for balanced setpoint tracking and

disturbance re!ection)( P P P

P

K τ+θτ

P τ*ote

τC is the larger

o! &0,1τP (

an# &0,8θP (

minimum IT*+ for

set point tracking

( ) $1%.0/

&%.0 −τθ P P

P K ( ) $2$.0

/1%.00'.1 P P

P

τθ−

τ

minimum IT*+ fordisturbance re!ection

( ) $.0/

&$.0 −τθ P P

P K ( ) %&0.0/%.0 P P

P

τθτ

Controller tuning can %e "er!orme# automatically using

the “Design Tools” mo#ule o! Control Station™

∫

+∞

= 0d)(!*"# t t et : inte"ral o% t#e time-$ei"#ted absolute

value o% error

* disadvantage of feedbackl




gcontrol

)n con$entional !ee#%ack control the correcti$e

action !or #istur%ances #oes not %egin until a%ter the controlle# $aria%le #e$iates !rom the set "oint

stack gas

cold oil

hot oil

fuel gas

TC

TT

If either the cold

oil Ao6 rate or the

cold oiltemperature

change, the

controller may do

a good !ob in

keeping the hot oiltemperature at

the setpoint

#hat if the pressure of the fuel gas changes$

? d t l 3 4




stack gas

cold oil

hot oil

fuel gas

TC

TT

PC

PT

?ascade control 3 4

T6o control loops are nested 6ithin eachother: the master controller and the slave

controller the out"ut signal o! the master &"rimary( controller ser$es

as the set "oint o! the sla$e &secon#ary( controller

The performancecan be improved

because the fuel

control valve 6ill be

ad!usted as soon as

the change in supply pressure is detected

sla$e loo"

master loo"

set point




?ascade control 3 2

)eed in

Products out

Coolin

* ater in

Coolin

* ater out

TC

TT

/ The TC may reect

satis!actorily

#istur%ances such as

reactant !ee# ' an#com"osition

/ )! the ' o! the cooling

ater increases, it

sloly increases the

reactor ' / The TC action may %e

#elaye# %y #ynamic lags

in the acket an# in the

reactor




?ascade control 3 2?ascade control 3 2 (cont(cont’’d)d)

)eed in

Products out

Coolin

* ater in

Coolin

* ater out

TC

TC

TT

TT

set point

master loo"

sla$e loo"/ The "er!ormance can %e

im"ro$e# %ecause the

cooling ater rate ill

%e a#uste# as soon as a

change in the ac+ettem"erature is #etecte#

/ This kee"s the heat

remo$al rate at a

constant le$el, an# the

reactor tem"erature is

less aecte# %y the

unknon #istur%ance

Tuning a cascade loop




Tuning a cascade loop

1 Begin ith %oth the master an# the sla$e controllers

in manual2 Tune the sla$e &inner( loo" !or set-"oint tracking 9rst

&the tuning gui#elines "resente# %e!ore can %e

use#(

( Close the sla$e loo", an# a#ust the tuning on line toensure goo# "er!ormance

4 (eavin" t#e inner loop closed, tune the master loo"

!or #istur%ance reection &the tuning gui#elines

"resente# %e!ore can %e use#(

5 Close the master loo", an# a#ust the tuning on line

to ensure goo# "er!ormance% #$only controller is often su6cient for the slave loop

. d t l




.ummary on cascade control

)t is use# to im"ro$e the #ynamic res"onse o! the

"rocess to loa# #istur%ances

)t is "articularly use!ul hen the #istur%ances are

associate# ith the mani"ulate# $aria%le or hen

the 9nal control element ehi%its nonlinear %eha$ior

The #istur%ances to %e reecte# must %e $it#in the

inner loo"

The inner loo" must res"on# muc# more 0uic+ly

than the outer loo"

'$o controllers must %e tune#

ontrol !tation™




ontrol !tation ™

)t is a so!tare !or "rocess control, analysis,

tuning an# training

De$elo"e# %y Pro! Doug Coo"er at the

Chem Eng De"t &ni$ o! Connecticut,

>torrs, CT, >=(

)n!ormation on the so!tare at the !olloing

)nternet site:

http://www.engr.uconn.edu/control/

/seful references



/seful references

>e%org, D E, T 3 E#gar an# D = @ellicham" &18(

Process Dynamics and Control, John iley L >ons, Ae ?ork&>=(

4gunnaike, B = an# K Qay &1( Process Dynamics,

1odelin" and Control, 4!or# ni$ersity Press, Ae ?ork

&>=(

@arlin, T E &2000( Process Control: Desi"nin" Processes and

Control Systems %or Dynamic Per%ormance, @c-.ra-Kill, Ae

?ork &>=(

Qiggs, J B &1( C#emical Process Control, 3erret

Pu%lishing, 'u%%ock &>=(

Process Dynamics

Documents

process simulation

process states

process outputs

process response

process inputs

upstream process

process dynamicsgiven

enthalpytraining course