Chap02b.advanced process control.pdf
-
Upload
anthony-patterson -
Category
Documents
-
view
215 -
download
0
Transcript of Chap02b.advanced process control.pdf
-
8/14/2019 Chap02b.advanced process control.pdf
1/15
Advance Topics in Process Control
CHE 522 1
CHAPTER 2
SISO Linear
System Analysis
4. Analysis of first order systems
A First order system is that whose output is modeled by a first orderdifferential equation.
(1) first order lag (self-regulating)
)(10 tbuyadt
dya =+
)(11
0 tu
a
by
dt
dy
a
a=+
)(tukydt
dyp=+
where1
0
a
a= ;
1a
bkp =
Taking Laplace transform with all variables are in deviation form we get:
)(1
)( sus
ksy
p
+=
The term, (time constant) and kp(static gain) characterize the first ordersystem.
Chemical Engineering Department
King Saud University, 2003
-
8/14/2019 Chap02b.advanced process control.pdf
2/15
Advance Topics in Process Control
CHE 522 2
Time constant: characterize the speed of response of a first-order system.
It is a measure of the time necessary for a process to adjust to a change in
the input.
Static gain: characterize the sensitivity of the output to the input signal.
Dynamic response:
The dynamic response for a step change in uof magnitude of a: u(s)=a/s
)1()( / = tp eakty
The transient response for step change is shown in Figure (2).
0 T
u(t)
y(t)
ime
a
kpa
Figure 1: Response of a first-order lag system to step change in the input
Dynamic characteristics:
For a step change in the input, the process reaches a new steady state. The ultimate value of the output is kpa, which can be found from the
last Equation by setting t
The time constant, can be found be setting t = in the last equationwhich gives y = 0.632 kpa. Then from Figure2, the time needed for y
to reach 0.632 kpais . The smaller the value of , the steeper is the initial response of the
output.
The larger static gain of a process, the larger steady state value of itsoutput for the same input change.
Chemical Engineering Department
King Saud University, 2003
-
8/14/2019 Chap02b.advanced process control.pdf
3/15
Advance Topics in Process Control
CHE 522 3
0 Time
y(t)
kp1
a kp1
kp2< kp1
kp2
a kp2
0 T
y(t)
ime
kpa
1
2
1< 2
Figure 2: Effect of static gain, time constant on the response of first-order lag system
(2) Pure capacitive system (integrator)
)(0 tbudt
dya =
Example: Liquid storage tank with fixed outlet flow
FtFthA i = )()(&
F
As
sF
As
sh i1
)(1
)( = ; kp1/A
)()( sus
ksy
p=
Dynamic response:
For a step change in uof magnitude ofa: u(s) = a/s
2)( s
ak
sy
p
=
which have the following time response:
atkty p=)(
The transient response looks like this:
Chemical Engineering Department
King Saud University, 2003
-
8/14/2019 Chap02b.advanced process control.pdf
4/15
Advance Topics in Process Control
CHE 522 4
0 T
u(t)
ime
y(t)
kp
a
Figure 3: Response of an integrator first-order system to a step change
Apparently, the ultimate value for the output is not achievable, yas t.
A pure capacitive process causes serious control problem because itcan not balance itself.
For small change in the input, the output grows continuously. Thisattribute knows as non-self-regulating (integrator) process.
5. Analysis of second order system
A second order system is that whose output is modeled by a second orderdifferential equation.
)(212
2
0 tbuyadt
dya
dt
yda =++
)(22
1
2
2
2
0 tua
by
dt
dy
a
a
dt
yd
a
a=++
)(22
22 tuky
dt
dy
dt
ydp=++
)(12
)(22
suss
ksy
p
++=
= natural period of oscillation (characteristic time constant)
Chemical Engineering Department
King Saud University, 2003
-
8/14/2019 Chap02b.advanced process control.pdf
5/15
Advance Topics in Process Control
CHE 522 5
= damping ratiokp = static gain
Sources of second order dynamics in the chemical industries come from a
series of first-order systems, or a processing system with a controller.
Dynamic response:For input step change of magnitude of a, u(s) = a/swe have:
))((
/
)12()(
21
2
22 pspss
ak
sss
aksy
pp
++=
where+
=
121p
=
121p
Case A: (over-damped) when > 1, we have two distinct and real poles
+
= )1sinh(
1)1cosh(1)( 2
2
2/ tteakty tp
Case B: (critically-damped) when = 1, we have two equal poles
+= /11)( tp e
takty
Case C: (under-damped) when < 1, we have two complex conjugatepoles
+
= )sin(
1
11)(
/
2wteakty
tp
The dynamic response for all cases is shown in Figure (4)
Chemical Engineering Department
King Saud University, 2003
-
8/14/2019 Chap02b.advanced process control.pdf
6/15
Advance Topics in Process Control
CHE 522 6
0 T
u(t)
ime
y(t)
1
Figure 4: The response of second order system to step change
The over-damped response is sluggish and resembles a little theresponse of a first-order system. IT becomes more sluggish with
larger values of . The critically damped response is faster than the over-damped one. The under-damped response is faster than the others, but oscillates.
The oscillatory behavior becomes pronounced with smaller values .
Characteristics of under-damped system
Overshoot, A/B
=
21exp/BA
Decay ratio, C/A
=
21
2exp/AC
Period of oscillation
==
21frequencyradian
Chemical Engineering Department
King Saud University, 2003
-
8/14/2019 Chap02b.advanced process control.pdf
7/15
Advance Topics in Process Control
CHE 522 7
21
2
== noscillatioofperiodT
Rise time, tr: is the time the output takes to first reach the new steadystate value Process time, tp: is the time required for the output to reach its first
peak.
Settling time, ts: is the time required for the output to reach within5% of the new steady state value.
y
A
B
C
T
Timetr tp ts Figure 5: characteristics of second-order under-damped response
Chemical Engineering Department
King Saud University, 2003
-
8/14/2019 Chap02b.advanced process control.pdf
8/15
Advance Topics in Process Control
CHE 522 8
6. Delay time
The delay time known as transportation time is basically the time required
for a material to move a specific distant. However, time delay is an inherentproperty of any chemical process.
sesG =)(
t
y(t)
delay time
u(t)
Sources:
The use of chromatography to measure composition of liquids orgases is another source of dead time.
Flow in a long pipe. Time lag produced from staged processes.
Therefore, a first-order system with dead-time is:
1)(
1 +
=
s
KesG
s
Therefore, a second-order system with dead-time is:
12)(
2 ++=
ss
KesG
s
Chemical Engineering Department
King Saud University, 2003
-
8/14/2019 Chap02b.advanced process control.pdf
9/15
Advance Topics in Process Control
CHE 522 9
7. Approximation of higher order systems
Systems with order higher than one can be represented by:
==
+==
n
in
ii
i
s
KsGsG
1
1
1)(
)()(
It can be approximated by low order transfer function with dead time as
follows:
1)(
1 +=
s
KesG
s
where 1is the dominant time constant and the dead time is:
==
n
ii
2
8. Transfer function and Block diagram
Transfer function relates the process input to the process output in a linear
mathematical function. Therefore,
Transfer Function: is the Laplace transform of the output,y(s) divided bythe Laplace transform of the process input, u(s), with all initial conditions
are zero (deviation form)
)()()(
susysG =
Example: For the storage tank we have:
)(1)(
)(sG
s
K
sF
shp+
= ; with =AR, K = 1/R
Example: For the heated tank we have:
Chemical Engineering Department
King Saud University, 2003
-
8/14/2019 Chap02b.advanced process control.pdf
10/15
Advance Topics in Process Control
CHE 522 10
)(1)(
)(sG
s
K
sQ
sTp+
= ; with = V/F, K = 1/FCp
Example: For the CSTR we have:
)(1)(
)(sG
s
K
sC
sCp
Ai
A +
= ; with = V/(F+Vk), K =F/(F+Vk)
8.1 Properties of Transfer Function
Order: The order of the system is the highest derivative of the output
variable in the defining differential equation. For Transfer Function, it isthe highest power ofsin the denominator.
Pole: is the root of the denominator of the transfer function, i.e. the root ofthe characteristic polynomial. It directly determines:
The stability of the system (positive poles) The potential of periodic transient (imaginary poles)
Zero: is the root of the numerator of the transfer function. It determines aninverse response (positive zero).
Casuality: A physical system is causalwhen the order of the denominatoris greater than the numerator, and when the transfer function goes to 0 as s
, the system is hence strictly proper. If the transfer function contains
es or the order of numerator is higher than the denominator, then the
system is non-casual or not realizable because the current values of thesystem depends on the future values of the variables.
Steady state gain: is the steady state value of the transfer function, is
evaluated by settings= 0 in the stable transfer function.
8.2 Effect of poles and zeros
The poles and zeros of a transfer function affect the dynamic of a process.
Consider a particular transfer function:
Chemical Engineering Department
King Saud University, 2003
-
8/14/2019 Chap02b.advanced process control.pdf
11/15
Advance Topics in Process Control
CHE 522 11
)12)(1()(
221 +++
=ssss
KsG
The poles, i.e. the roots of the characteristic equationare:
01=s
12
1
=s
2
2
2
3
1
+
= js
2
2
2
4
1
= js
The poles can be represented in the complex plane as follows:
Real axis
imaginaryaxis
XX
X
X
0
1
1
2
2
21
2
21
+
Figure 6: roots of the characteristic equation of G(s)
Complex poles indicate the response will contain sine and cosinemodes, i.e. will exhibit oscillation.
Negative poles will result in a stable decaying response.Chemical Engineering Department
King Saud University, 2003
-
8/14/2019 Chap02b.advanced process control.pdf
12/15
Advance Topics in Process Control
CHE 522 12
Positive poles indicate that the response will have unstable mode.
The transfer function can be written as follows:
)())((
)())(()(
21
21
n
m
n
m
pspsps
zszszs
a
bsG
=
L
L
Positive zero leads to inverse response. Zero-pole cancellation occurs when a zero has exactly the same
numerical value as a pole.
A zero can exert a profound effect on the coefficient of responsemode.
8.3 Block diagrams
Block diagram is a graphical representation of transfer functions and their
interactions.
Block diagram assist the engineer in determining the quantitative aspects of
dynamic performance and in understanding the qualitative features of the
system.
y(s)= G(s)u(s)
G(s)u(s) y(s)
y(s)=A+By(s)
B
+
+
A
y(s)=G1(s)G2(s)u(s)
Chemical Engineering Department
King Saud University, 2003
-
8/14/2019 Chap02b.advanced process control.pdf
13/15
Advance Topics in Process Control
CHE 522 13
G1(s)
u(s) y(s)G
2(s)
y(s)=G1(s)A G2(s)BG
1(s)
y(s)A
B
+
_
G2(s)
y(s) = G1(s)(AB)
y(s)A
B
+
_
G1(s)
8.4 Input-output relation
Direct action: an increase in the input lead to an increase in the output.
I/Iu y
Reverse action: an increase in the input lead to a decrease in the output.
I/Du y
Chemical Engineering Department
King Saud University, 2003
-
8/14/2019 Chap02b.advanced process control.pdf
14/15
Advance Topics in Process Control
CHE 522 14
Fi
F
h
Fi
F
h
A series of I/I blocks always result in I/I action:
I/Iu yI/I
An odd series of I/D blocks results in I/D action:
I/Iu yI/D
An Even series of I/D blocks results in I/I action:
I/Du yI/D
Chemical Engineering Department
King Saud University, 2003
-
8/14/2019 Chap02b.advanced process control.pdf
15/15
Advance Topics in Process Control
CHE 522 15
It is important that the overall system has an I/D action (negative
feedback), therefore, the controller action must be adjusted such that the
overall action of the closed-loop be I/D or reveres action.
Alternatively:
vekk pc +=
Chemical Engineering Department
King Saud University, 2003