Lecture 2

Introduction to Control Strategies and

Modelling for Control

Process Control
Prof. Kannan M. Moudgalya

IIT Bombay
Tuesday, 24 July 2012

1/39 Process Control Modelling for Control

Outline

1. Recall: different types of variables

2. Feed forward control
3. Feedback control
4. Scilab code
These will be illustrated with room
temperature control and the mixer problem

2/39 Process Control Modelling for Control

1. Recall: different types of variables

3/39 Process Control Modelling for Control

Recall the Mixing Process
I Mixing of two
streams
Control I Variable stream
Mixture Valve
A, B has composition
(x1 ) varying
x1 , w1
I i.e. x1 is
Pure A disturbance
x2 = 1 I Want output
w2 =?
composition
constant
x, w I Control streams
flow can be
changed
4/39 Process Control Modelling for Control
Variables in Mixing Process

Feed composition x1 is a
Control
Mixture Valve
A, B 1. Controlled variable
x1 , w1 2. Manipulated
Pure A variable
x2 = 1
w2 =? 3. Disturbance variable
Answer:
x, w Disturbance variable

5/39 Process Control Modelling for Control

Variables in Mixing Process

Outlet composition x is a
Control
Mixture Valve
A, B 1. Controlled variable
x1 , w1 2. Manipulated
Pure A variable
x2 = 1
w2 =? 3. Disturbance variable
Answer:
x, w Controlled variable

6/39 Process Control Modelling for Control

What is the Control Effort?
Control
Mixture Valve
A, B
x1 , w1
Pure A
x2 = 1
w2 =?

x, w

Flow rate of pure stream, i.e. w2

It is also called as the Manipulated Variable

7/39 Process Control Modelling for Control

2. Feed Forward Control

8/39 Process Control Modelling for Control

Mass Balance of the Mixing Process
I Overflow: w1 + w2 = w
Control
Mixture Valve
A, B I w1 x1 + w2 x2 = wx
x1 , w1
Pure A
I = (w1 + w2 )x
x2 = 1
w2 =? I Find w2 with x2 = 1
x x1
x, w I w2 = w 1
1x
I Want x to be at x:
Find an expression
x x1
for w2 . I w 2 = w1
1x
9/39 Process Control Modelling for Control
How to Use the Mass Balance?
x x1
I w 2 = w1
1x
Control
Mixture Valve
A, B I Suggest a strategy
x1 , w1 to implement
Pure A
x2 = 1
w2 =?

x, w

10/39 Process Control Modelling for Control

Feed Forward Control of Mixing Process

AT AC
Control
Mixture Valve
A, B
x1 , w1
x x1
Pure A I w 2 = w1
x2 = 1 1x
w2 =?

x, w

11/39 Process Control Modelling for Control

Feed forward control

I Model a system EXACTLY

I Correct for any deviation IMMEDIATELY
I Known as feed forward control
I Explain with room temperature control

12/39 Process Control Modelling for Control

Salient features of feed forward control
I Disturbance:
change in x1
AT AC I Disturbance is
Control measured and its
Mixture Valve
A, B effect quantified
x1 , w1 I Immediate correction
Pure A can be made
x2 = 1
w2 =? I Exact model is
required
x, w I Only the disturbance
variable needs to be
measured
13/39 Process Control Modelling for Control
Shortcomings in FF Applied to Mixing
I What if w1 also
varies?
AT AC I What if some
Control phenomena are not
Mixture Valve
A, B modelled - leaks?
x1 , w1 I What if x changes
Pure A because of some
x2 = 1
w2 =? mistakes that we
made?
x, w

14/39 Process Control Modelling for Control

Shortcomings in Feed Forward Control -
General

I Exact model is required

I As the controlled variable is not
measured, will not know even if there are
deviations
I Effects of other disturbances not
considered
I Need an exact model to quantify every
disturbance

15/39 Process Control Modelling for Control

Feed forward: good

Feed forward control is good because

1. Only one disturbance variable needs to be
measured
2. Need only one model
3. Required model is easy to build
4. Fast correction against changes in
disturbance measured and whose effect is
known
Answer: 4, Fast correction

16/39 Process Control Modelling for Control

Feed forward: bad
1. Controlled variable is not measured
2. Effects of other variables not considered
3. Exact model is required for every variable
4. Very slow in responding
Feed forward control is bad because
A. 1 and 2
B. 2 and 3
C. 1, 2 and 3
D. 1 and 4
Answer: C: 1, 2 and 3
17/39 Process Control Modelling for Control
Assignment for discussion

Apply the feed forward control strategy to the

room temperature control problem

18/39 Process Control Modelling for Control

3. Feedback Control

19/39 Process Control Modelling for Control

How do we Overcome the Shortcomings?
- General

I Measure the controlled variable directly

I Compensate for any deviations

20/39 Process Control Modelling for Control

Can we Visualise this Control Strategy?
What is meant by
I Measure the controlled variable directly &

I Compensate for any deviations?

Control
Mixture Valve
A, B
x1 , w1
Pure A
x2 = 1
w2 =?

x, w
21/39 Process Control Modelling for Control
Direct Regulation of Controlled Variable
Control
Mixture Valve
A, B AC
x1 , w1
Pure A
x2 = 1
w2 =?
AT

x, w
I w2 (t) = w2,old + K(x x(t))
I w2,old is nominal operating value
I x is the desired value
I K to be decided
I Is K positive or negative?
22/39 Process Control Modelling for Control
Salient features of feedback control -
General

I Disturbance affects the system before

corrective action
I Feedback control requires measurement
of the controlled (=output) variable
I Useful when exact knowledge is not
available
I Popular control strategy

23/39 Process Control Modelling for Control

Salient features Applied to Mixing
Control
Mixture Valve
A, B AC
x1 , w1
Pure A
x2 = 1
w2 =?
AT

x, w
I Disturbance affects the system before
corrective action
I Feedback control requires measurement
of the controlled (=output) variable
I Useful when exact knowledge is not
available
24/39 Process Control Modelling for Control
Assignment for discussion

Apply the feedback control strategy to the

room temperature control problem

25/39 Process Control Modelling for Control

Feed Forward + Feedback

I In reality, one uses a combination of feed

forward and feedback controllers

26/39 Process Control Modelling for Control

4. Modelling and Scilab Solution

27/39 Process Control Modelling for Control

Modelling

I Modelling of transients in mixing process

I Scilab solution

28/39 Process Control Modelling for Control

Modelling of Transients in Mixing Process
Control
Mixture Valve
A, B
x1 , w1

x2 6= 1
w2 =?

x, w
dx
V = w1 x1 + w2 x2 (w1 + w2 )x
dt
dx
V + (w1 + w2 )x = w1 x1 + w2 x2
dt
dx
Divide by w1 + w2 : +x=C
dt
Integrating Factor = et/
29/39 Process Control Modelling for Control
Solution to the Mixing Process
dx x C
et/ + et/ = et/
dt
d  t/  C
e x = et/
Z dt 
 Z C
d et/ x = et/ dt

t/ t/
e x=e C+K
Use IC: at t = 0, x = x(0):
K = x(0) C
h i
t/ t/
x=e x(0) + C 1 e
30/39 Process Control Modelling for Control
Solution through Integrating Factor

h i
t/ t/
x=e x(0) + C 1 e
V
=
w1 + w 2
w1 x1 + w2 x2
C=
w1 + w2
x(0) = C(0)

31/39 Process Control Modelling for Control

Example: w1 from 500 to 400

I Initial conditions:

V = 2 m3
= 900 kg/m3
w1 (0) = 500 kg/min
w2 (0) = 200 kg/min
x1 (0) = 0.4
x2 (0) = 0.75
I w1 changes to 400 kg/min suddenly

32/39 Process Control Modelling for Control

Scilab code 01-ex2p1a.sce
1 V = 2;
2 rho = 900;
3 w10 = 5 0 0 ;
4 w20 = 2 0 0 ;
5 x10 = 0 . 4 ; x20 = 0 . 7 5 ;
6 x0 = ( w10 x10+w20 x20 ) / ( w10+w20 ) ;
7 //
8 // A f t e r the sudden change
9 w1 = w10 ; w2 = 1 0 0 ;
10 x1 = 0 . 6 ; x2 = x20 ;
11 t a u = V r h o / ( w1+w2 ) ;
12 C = ( w1 x1+w2 x2 ) / ( w1+w2 ) ;
13 t = log sp ac e ( 3 ,1.39 ,100) ;
14 x=e x p ( t / t a u ) x0 + C(1 e x p ( t / t a u ) ) ;
15 p l o t 2 d ( t , x ) , x t i t l e ( , t , x )
16 s e t ( gca ( ) , g r i d , [ 1 1 ] )
17 a 33/39 x l Control
= g c a ( ) ; a . Process a b e l .Modelling
f o n fort Control
s i z e =4;
Plot of x vs. t for 01-ex2p1b.sce

0.518

0.516

0.514

0.512

0.510

0.508

0.506

0.504

0.502

0.500
0 5 10 15 20 25

34/39 Process Control Modelling for Control

Example:
I Initial conditions:

V = 2 m3
= 900 kg/m3
w1 (0) = 500 kg/min
w2 (0) = 200 kg/min
x1 (0) = 0.4
x2 (0) = 0.75
I w2 changes to 100 kg/min suddenly
I Simultaneously, x1 changes to 0.6
suddenly
35/39 Process Control Modelling for Control
Scilab code 01-ex2p1b.sce
1 V = 2;
2 rho = 900;
3 w10 = 5 0 0 ;
4 w20 = 2 0 0 ;
5 x10 = 0 . 4 ; x20 = 0 . 7 5 ;
6 x0 = ( w10 x10+w20 x20 ) / ( w10+w20 ) ;
7 %
8 % A f t e r the sudden change
9 w1 = w10 ; w2 = 1 0 0 ;
10 x1 = 0 . 6 ; x2 = x20 ;
11 t a u = V r h o / ( w1+w2 ) ;
12 C = ( w1 x1+w2 x2 ) / ( w1+w2 ) ;
13 t = log sp ac e ( 3 ,1.39 ,100) ;
14 x=e x p ( t / t a u ) x0 + C(1 e x p ( t / t a u ) ) ;
15 plot2d ( t , x)
16 s e t ( gca ( ) , g r i d , [ 1 1 ] )
36/39 Process Control Modelling for Control
Plot of x vs. t for 01-ex2p1b.sce

0.60

0.55

0.50
0 5 10 15 20 25

37/39 Process Control Modelling for Control

What we learnt today

I Feed forward and feedback control

I Concept
I Salient features
I Pros and cons
I Applied to mixing process
I Generalised
I Dynamic simulation of mixing
I Mathematical formulation
I Scilab code

38/39 Process Control Modelling for Control

Thanks

I kannan@iitb.ac.in

39/39 Process Control Modelling for Control